Sei sulla pagina 1di 6

Home Search Collections Journals About Contact us My IOPscience

Unprecedented ultra-high hydrogen gas sensitivity in undoped titania nanotubes

This article has been downloaded from IOPscience. Please scroll down to see the full text article.

2006 Nanotechnology 17 398

(http://iopscience.iop.org/0957-4484/17/2/009)

View the table of contents for this issue, or go to the journal homepage for more

Download details:
IP Address: 161.6.94.245
The article was downloaded on 24/05/2013 at 07:47

Please note that terms and conditions apply.


INSTITUTE OF PHYSICS PUBLISHING NANOTECHNOLOGY
Nanotechnology 17 (2006) 398–402 doi:10.1088/0957-4484/17/2/009

Unprecedented ultra-high hydrogen gas


sensitivity in undoped titania nanotubes
Maggie Paulose1 , Oomman K Varghese2 ,
Gopal K Mor2 , Craig A Grimes2 and Keat G Ong1
1
SentechBiomed Corporation, 200 Innovation Boulevard, State College, PA 16803, USA
2
Department of Electrical Engineering, Department of Materials Science and Engineering,
The Pennsylvania State University, University Park, PA 16802, USA

E-mail: kgong@sentechbiomed.com

Received 9 September 2005, in final form 10 November 2005


Published 14 December 2005
Online at stacks.iop.org/Nano/17/398
Abstract
A highly ordered array of micron-length undoped titania nanotubes exhibits
an unprecedented variation in electrical resistance of about 8.7 orders of
magnitude (50 000 000 000%), at room temperature, when exposed to
alternating atmospheres of nitrogen containing 1000 ppm hydrogen and air.
This represents the largest known change in electrical properties of any
material, to any gas, at any temperature. The nanotube arrays were
fabricated using anodic oxidation of titanium foil in a pH 4.0 electrolyte
containing potassium fluoride, sodium hydrogen sulfate monohydrate and
sodium citrate tribasic dihydrate. The dramatic change in resistance is
believed to be due to the highly active surface states on the nanoscale walls
of the tubes, high surface area of the nanotube architecture, and the
well-ordered geometry allowing for hydrogen-sensitive tube-to-tube
electrical connections.
(Some figures in this article are in colour only in the electronic version)

1. Introduction titania nanotube array fabricated by anodization of titanium


we have observed an unprecedented, phenomenal change in
Metal oxide semiconductors are known to vary their electrical electrical resistance in response to low levels of hydrogen
resistance with respect to the gas ambient and hence are at room temperature. Carbon monoxide and methane,
widely used for chemical sensing [1, 2]. However, elevated both strong reducing gases, do not significantly impact
temperatures, typically several hundred degrees Celsius, are the resistance of the nanotubes. Our results demonstrate
needed for these materials to obtain any notable gas responses. the ability of the nanodimensional structures to possess
Furthermore, in pure form these materials do not have the extraordinary properties that micro/macroscale structures have
ability to selectively respond to a particular gas from a never exhibited.
mixture and hence chemical modification, such as doping Earlier we reported on the fabrication of titania nanotube
or surface functionalization, must be used [2–5]. For arrays having lengths up to 400 nm and pore diameters
example, to improve the sensitivity and selectivity of metal from 22 to 90 nm using anodic oxidation of titanium thick-
oxides to hydrogen at room temperature it is common to use film foils and thin films in hydrofluoric acid containing
palladium as a dopant or thin surface coating [6, 7]. With electrolytes [18, 19]. The structure consisted of adjacently
the advent of nanotechnology, nanostructures of functional located parallel nanotubes separated from the un-anodized
metal oxides such as SnO2 , ZnO, TiO2 , WO3 , Ga2 O3 and titanium foil by a thin oxide layer (barrier layer). Nanotube
In2 O3 have been developed for environmental sensing, and arrays of length 200 nm and pore diameter 22 nm coated with
it has become evident that nanoscale architectural features a thin palladium layer showed a resistance variation of about
have the ability to yield superior and often unexpected five orders in magnitude on exposure to 1000 ppm (parts per
gas-dependent electrical behaviour [8–17]. With a unique million) hydrogen at room temperature [20]. However, device
architecture comprising a highly ordered, high aspect ratio performance was limited, and made problematic, by the very

0957-4484/06/020398+05$30.00 © 2006 IOP Publishing Ltd Printed in the UK 398


Unprecedented ultra-high hydrogen gas sensitivity in undoped titania nanotubes

Table 1. Summary of the nanotube dimensions under different


a voltage and pH conditions and their maximum resistance variation
to 1000 ppm H2 at room temperature.
10 V (30 nm pore size)

Length Log(S) Log(S)


pH (nm) no Pd with Pd
1.1 350 7.2 7.3
3.0 730 8.0 8.7
4.0 990 8.7 8.3
4.5 1400 8.0 8.3
5.0 2000 – –

b to 630 ◦ C in oxygen for 6 h at a heating/cooling rate of


1 ◦ C min−1 .
For the electrical measurements platinum circular
electrodes (500 µm diameter, 100 nm thick, 1 mm edge to
edge spacing) were sputter coated onto the nanotube-array
samples; increasing the electrode separation increases the
baseline resistance, but does not affect sensitivity. After
fixing the samples to TO-8 steel headers (Electrovac GesmbH,
Klosterneuberg, Austria), gold wire (1 mil) was bonded on
the platinum pads with a Mech-El-907 ultrasonic wire bonder
(Mech-El Industries, Woburn, MA). The behaviour of the
nanotube samples in the presence of different gases was tested
in a sealed Plexiglas chamber of 110 cm 3 volume. Electrical
Figure 1. FESEM images of illustrative nanotube-array samples:
(a) cross-sectional view of nanotubes grown in an electrolyte pH 3 resistance was measured using an electrometer (Keithley,
using 25 V, and (b) top view of a sample fabricated in a pH 1.1 Model 6517A). MKS mass flow controllers were used for
electrolyte using 25 V. regulating the chamber gas (ultra-high purity) flow. Data
collection and instrument control were managed by a personal
computer. Titania, an n-type semiconductor, decreases its
low thickness of the array structure, which provided a low resistance in the presence of reducing gases and hence the gas
resistance path for the current to flow between the platinum- sensitivity S is defined as
film electrodes on the nanotube surface and the underlying
G gas − G air G gas
titanium metal layer. S= ≈ when G air  G gas
Recently we have improved the anodization process to G air G air
fabricate nanotube-array architectures with lengths up to where G air and G gas are, respectively, the conductance of the
6.5 µm, pore diameters from 22 to 110 nm, and wall nanotubes in air and the test gas.
thicknesses from 9 to 34 nm [21, 22]. Our ability to control
the nanotube-array dimensions has enabled our ability to 3. Results and discussion
meaningfully test the gas-sensing properties of the nanotube
arrays. More importantly, our studies on these highly FESEM images (JEOL, Japan, model 6700F) of illustrative
stable nanostructures have revealed their unusual behaviour nanotube-array samples are shown in figure 1. The pore
to hydrogen gas, which can open new avenues in nanoscience diameter depends on the anodization voltage, whereas the
and nanotechnology. nanotube length is a function of both pH and the anodization
voltage. With the increase in voltage at a particular pH value
2. Experimental details the pore size as well as length of the tube increases. For
example, a 10 V anodization in pH 1.1 results in a nanotube
Nanotube arrays were prepared by anodization of 250 µm thick array of 350 nm length and 30 nm pore diameter. A 25 V
titanium foil (99.7% Aldrich, Milwaukee, WI) in an electrolyte anodization in pH 5.0 results in a nanotube array of 6.5 µm
containing 1 M sodium hydrogen sulfate monohydrate, 0.1 M length, the longest nanotube array we have achieved, with a
potassium fluoride (ACS reagent 99% Alfa Aesar, Ward Hill, 110 nm pore diameter. The greatest hydrogen sensitivities are
MA) and 0.2 M sodium citrate tribasic dihydrate (ACS reagent achieved with the 30 nm pore diameter samples fabricated at
99% Aldrich). Electrolyte pH was adjusted by the addition 10 V; consequently, the majority of our discussion is focused
of sodium hydroxide (97% Aldrich). Nanotube lengths varied on these samples. Table 1 summarizes the maximum room
from 380 nm to 6 µm and pore diameters from 30 to 110 nm as temperature hydrogen sensitivity, log(S), of the 30 nm pore
dependent on the electrolyte pH (1.11–5) and the anodization diameter nanotube arrays in response to 1000 ppm hydrogen, as
potential (10–25 V); the anodization was conducted for 17 h a function of length which in turn is determined by electrolyte
for all samples. The as-prepared amorphous samples were pH. We note that in crystallizing the initially amorphous
crystallized by annealing at temperatures ranging from 370 nanotube array samples maximum hydrogen sensitivity was

399
K G Ong et al

Figure 2. Variation in hydrogen sensitivity S of 30 nm pore Figure 3. Room-temperature resistance variation of a


diameter nanotube-array samples as a function of electrolyte pH. nanotube-array sample prepared in a pH 4.0 electrolyte using 10 V
(30 nm pore diameter), annealed at 480 ◦ C, alternately exposed to
air and 1000 ppm hydrogen in nitrogen.

achieved with a 480 ◦ C annealing. It can be seen from


table 1 that the highest resistance variation, 8.7 orders of
to help hydrogen detection in metal oxide materials, some
magnitude (50 000 000 000%), was obtained for nanotube
nanotube array samples were coated with a 10 nm Pd layer
samples prepared using electrolyte pH 4.0 (length ∼1 µm,
pore diameter ∼30 nm and wall thickness ∼13 nm). To the and their response to hydrogen tested. The Pd layer degraded
best of our knowledge, this is the largest change in electrical the sensitivity of the highly sensitive samples due to a drop in
resistance of any material, to any gas, at any temperature. the base resistance; samples of relatively lower performance
Assuming a 10  measurement resolution, linear extrapolation typically saw a modest sensitivity increase (table 1).
of the sensitivity indicates the ability of the nanotube array It behooves us to consider how hydrogen interacts with the
to detect hydrogen at 20 ppt (parts per trillion). We make titania nanotube architecture, a material which is essentially
note that a test chamber capable of accurately flowing less all surface and no bulk, to achieve such remarkable changes
than, approximately, 10 ppm of any gas is a difficult challenge in electrical resistance. At room temperature there is little
requiring considerable resources. Consequently, we are not reduction of the surface or bulk oxide [23, 24], while the
able to directly measure the low-level sensitivity of the sensor. fast response and recovery without hysteresis (figure 3) rules
The lower limit of our test chamber is approximately 20 ppm, out a significant contribution from diffusion of hydrogen into
at which we see 7.2 orders of magnitude variation in electrical the titania lattice [25]. Oxygen in air may be chemisorbed
resistance to 20 ppm hydrogen. The response of the sensor in the form of O− 2 on the nanotube surface by trapping
is linear to, approximately, 1000 ppm, when it begins to electrons from its conduction band leading to an enhanced
demonstrate a nonlinear response. base resistance [3, 4]. Hydrogen can remove this chemisorbed
Figure 2 shows the variation in hydrogen sensitivity of oxygen thus reducing the resistance [26]. To understand the
the 10 V samples as a function of electrolyte pH; note that role of chemisorbed oxygen, we initially passed air through
the pH 5.0 sample was not tested, as nanotube array formation the test chamber, then nitrogen, then a hydrogen–nitrogen
occurs on only ≈25% of the film. The typical response of a mixture, and finally nitrogen. It was observed that nitrogen
pH 4.0 sample on switching the ambient atmosphere between reduced the 1000 ppm hydrogen sensitivity to approximately
air and 1000 ppm hydrogen in nitrogen is shown in figure 3. six orders of magnitude while only slightly reducing the base
On exposure to hydrogen a rapid reduction in resistance resistance, and greatly extended the needed recovery time.
from several hundred giga-ohms to a few hundred ohms is The high sensitivity in the absence of oxygen indicates that
observed; there is no indication of measurement hysteresis. direct chemisorption of hydrogen on the nanotubes is the
The sensitivity S is not greatly influenced by nanotube length; dominant mechanism leading to the tremendous reduction
for the 10 V samples a sensitivity shift of 7 orders to 8.7 orders in resistance of the titania nanotubes, while the presence
is found with an increase in length from 380 nm to 1 µm. It of oxygen facilitates removal of the chemisorbed hydrogen.
was found that 6 µm long samples (25 V, pH 5.0) showed less Platinum is known to be a catalyst that activates hydrogen
sensitivity, with significantly longer response/recovery times by adsorbing and dissociating hydrogen molecules which
due to the time required for hydrogen to diffuse inside the are then spilled over to the semiconducting material for
long pores. The sensitivity of the nanotube arrays is in stark chemisorption [27, 28]. Replacing the platinum electrodes
contrast to the hydrogen sensitivity of a crystalline TiO2 film with gold reduced the 1000 ppm sensitivity of the pH 4.0
made by thermal oxidation of a Ti film which demonstrates, 10 V nanotube sample to ≈5 orders of magnitude with a
to 1000 ppm hydrogen at room temperature, a resistance considerably slower response time. While it is evident that the
variation of approximately 400%. As palladium is known platinum electrodes facilitate the high sensitivity by providing

400
Unprecedented ultra-high hydrogen gas sensitivity in undoped titania nanotubes

Figure 5. The influence of nanotube wall thickness on band bending


due to oxygen chemisorption. (a) When the nanotube wall
half-thickness (t/2) is much greater than the space charge layer, (b)
when the thicknesses are comparable, and (c) when t/2 is less than
Figure 4. Hydrogen sensitivity (1000 ppm) of a 10 V the width of the space charge region, (d) schematic illustration of
nanotube-array sample prepared in a pH 4.0 electrolyte as a function nanotubes, top view, and the tube-to-tube connecting points
of relative humidity. The sensitivity of the sample to carbon corresponding to the case shown in (a).
monoxide and methane in dry ambient is also shown.

is significantly greater than the width of the space charge


spilt-over hydrogen for chemisorption, it is not the only cause region, as shown in figure 5(a), oxygen removal by hydrogen
of observed nanotube response. and subsequent hydrogen chemisorption will have little effect
Hydrogen activation appears to occur on the walls of the on device resistance, and hence high sensitivity cannot be
expected. In contrast, when t/2 is comparable to or less than
undoped nanotubes at highly active surface states provided by
the space charge region the shift in the electrical resistance
nanoscale surface defects. The dissociated hydrogen species
on exposure to hydrogen can be very high (figures 5(b), (c)),
form OH groups with the surface oxygen accompanied by
with a flat-band condition existing when the wall thickness is
electron transfer to the titania conduction band and formation
less than the width of the space charge region. The nanotube
of an electron-rich region within the nanotube walls. To ensure
sample showing the highest sensitivity had a wall thickness of
that the nanotube samples were undoped, samples annealed at
≈13 nm, corresponding to the geometry of figure 5(b). The
480 ◦ C were studied using XPS (Kratos Analytical Axis Ultra).
inter-wall connecting points also appear to play a significant
In addition to the Ti and O peaks the XPS spectra showed a
role in enabling the ultra-high hydrogen sensitivity. The
weak carbon (C 1s) peak. Depth profiling showed that carbon oxygen adsorption and its removal by hydrogen atoms as
was present inside the sample at low concentrations. The well as chemisorption of hydrogen at these constricted points,
carbon peak is believed to come from adventitious carbon or figure 5(d), regulates the current passing from nanotube to
hydroxyl and carboxyl groups. Our studies do not show any nanotube.
effect of C on the hydrogen sensitivity. Figure 4 shows the In summary, our studies on the interaction of titania
response of the nanotubes of 1 µm length and 30 nm pore nanotube arrays with hydrogen reveal an unprecedented gas-
diameter to 1000 ppm hydrogen in humid ambients. In humid dependent shift in electrical resistance. The synergetic effect of
conditions physisorption of water molecules takes place on platinum electrodes, highly active nanoscale surface states that
a layer of initially chemisorbed OH radicals at the nanotube activate oxygen and hydrogen for chemisorption, tremendous
surface which reduces the base resistance of the nanotubes surface area, nanosized walls and inter-tubular connecting
in air [29]. The adsorbed water molecules block the active points is believed responsible for the remarkable behaviour.
sites where hydrogen chemisorption occurs, thus reducing the Extension of the material architecture to other metal oxides
sensitivity. Figure 4 also shows the response of the nanotube should enable dramatically improved broad-spectrum gas-
sample to 1000 ppm carbon monoxide and 1000 ppm methane; sensing materials.
although both gases are strongly reducing, the sensitivity to
these gases is negligible. Acknowledgments
We believe that the crystallized nanoscale walls and inter-
tubular connecting points play critical roles in determining This work was partially supported by the National Institutes of
the remarkable hydrogen sensitivities of the titania nanotube Health grant number 5 R43 HD049233-02, and the National
arrays. The adsorption of oxygen in air takes place on either Science Foundation grant number CTS-0518269. The authors
side of the nanotube walls, creating an electron depletion wish to thank the referees for their careful reading of the
region. The width of the space charge layer L is given by manuscript and helpful suggestions.
L = L D (2eV /kT )1/2 , where L D = (ε0 εkT /2e2 ND )1/2
is the Debye length, eV is the barrier height, kT is the References
thermal energy and ND the ionized donor density. In metal [1] Seiyama T, Kato A, Fujiishi K and Nagatani M 1962 Anal.
oxides the space charge layer extends to a few tens of Chem. 34 1502–3
nanometres [30]. If the nanotube wall half-thickness t/2 [2] Moseley P T 1992 Sensors Actuators B 69 149–56

401
K G Ong et al

[3] Madou M J and Morrison S R 1989 Chemical Sensing with [17] Wang Z L 2004 Annu. Rev. Phys. Chem. 55 159–96
Solid State Devices (New York: Academic) [18] Gong D, Grimes C A, Varghese O K, Hu W, Singh R S,
[4] Wang C C, Akbar S A and Madou M J 1998 Chen Z and Dickey E C 2001 J. Mater. Res. 16 3331–4
J. Electroceram. 2 273–8 [19] Mor G K, Varghese O K, Paulose M, Mukherjee N and
[5] Harrison P G and Willett M J 1988 Nature 332 337–9 Grimes C A 2003 J. Mater. Res. 18 2588–93
[6] Christofides C and Mandelis A 1990 J. Appl. Phys. 68 R1–30 [20] Varghese O K, Mor G K, Grimes C A, Paulose M and
[7] Shimizu Y, Kuwano N, Hyodo T and Egashira M 2002 Mukherjee N 2004 J. Nanosci. Nanotechnol. 4 733–7
Sensors Actuators B 83 195–201 [21] Mor G K, Shankar K, Paulose M, Varghese O K and
[8] Varghese O K and Grimes C A 2004 Metal oxide Grimes C A 2005 Nano Lett. 5 191–5
nanostructures as gas sensors Encyclopedia of Nanoscience [22] Varghese O K, Paulose M, Shankar K, Mor G K and
and Nanotechnology vol 5, ed H S Nalwa (Valencia, CA, Grimes C A 2005 J. Nanosci. Nanotechnol. 5 1158–65
USA: American Scientific Publishers) pp 505–21 [23] Eder D and Kramer R 2003 Phys. Chem. Chem. Phys.
[9] Kolmakov A, Klenov D O, Lilach Y, Stemmer S and 5 1314–9
Moskovits M 2005 Nano Lett. 5 667–73 [24] Walton R M, Dwyer D J, Schwand J W and Gland J L 1998
[10] Arnold M S, Avouris P, Pan Z W and Wang Z L 2003 J. Phys.
Appl. Surf. Sci. 125 187–98
Chem. B 107 659–63
[25] Birkefeld L D, Azad A M and Akbar S A 1992 J. Am. Ceram.
[11] Wu N L, Wang S Y and Rusakova I A 1999 Science
Soc. 75 2964–8
285 1375–7
[12] Zhang D H, Liu Z Q, Li C, Tang T, Liu X L, Han S, Lei B and [26] Bates J B, Wang J C and Perkins R A 1979 Phys. Rev. B
Zhou C W 2004 Nano Lett. 4 1919–24 19 4130–9
[13] Comini E, Faglia G, Sberveglieri G, Pan Z W and [27] Roland U and Salzer R 1995 J. Chem. Soc. Faraday Trans.
Wang Z L 2002 Appl. Phys. Lett. 81 1869–71 91 1091–5
[14] Liu J F, Wang X, Peng Q and Li Y D 2005 Adv. Mater. [28] Roland U, Braunschweig T and Roessner F 1997 J. Mol.
17 764–7 Catal. A 127 61–84
[15] Wan Q, Li Q H, Chen Y J, Wang T H, He X L, Li J P and [29] Morimoto T, Nagao M and Tokuda F 1969 J. Phys. Chem.
Lin C L 2004 Appl. Phys. Lett. 84 3654–6 73 243–8
[16] Wang S H, Chou T C and Liu C C 2003 Sensors Actuators B [30] Ogawa H, Nishikawa M and Abe A 1982 J. Appl. Phys.
94 343–51 53 4448–55

402

Potrebbero piacerti anche