You are on page 1of 7

Sensors and Actuators B 93 (2003) 338–344

Hydrogen sensing using titania nanotubes


Oomman K. Varghese, Dawei Gong, Maggie Paulose, Keat G. Ong, Craig A. Grimes*
Department of Electrical Engineering and Materials Research Institute, 217 Materials Research Laboratory,
The Pennsylvania State University, University Park, PA 16802, USA

Abstract

Titanium dioxide nanotubes, made by anodization, are highly sensitive to hydrogen; for example, cycling between nitrogen atmosphere and
1000 ppm hydrogen a variation in measured resistance of 103 is seen for 46 nm diameter nanotubes at 290 8C. The hydrogen sensors are
completely reversible and have response times of approximately 150 s. Field emission scanning electron microscopy and Glancing angle X-
ray diffraction (GAXRD) are used to study the surface morphology and crystal structure of the nanotubes.
# 2003 Elsevier Science B.V. All rights reserved.

Keywords: Titania; Nanotube; Hydrogen; Nanoporous; Gas sensor

1. Introduction Pt/TiO2 by hydrogen [22–24]. Elevated temperature hydro-


gen sensors examine the electrical resistance with hydrogen
We recently reported [1] the fabrication of self organized concentration; for example, Birkefeld et al. [25] observed
titania nanotube arrays using an anodization technique. that the resistance of anatase phase of titania varies in
Although the as prepared nanotubes are amorphous, they presence of carbon monoxide and hydrogen at temperatures
crystallize on annealing at elevated temperatures and are above 500 8C, but on doping with 10% alumina it becomes
structurally stable to at least 600 8C. This stability of selective to hydrogen.
structure, which is one of the essential criteria of a gas
sensing material, prompted us to study the gas sensing
behavior of these nanotubes to technologically important 2. Experimental
gases, such as oxygen, carbon monoxide, ammonia, carbon
dioxide and hydrogen. Titania nanotubes [1] were grown from titanium foil
Titania has earned much attention for its oxygen sensing (99.5% pure from Alfa Aesar, Ward Hill, MA, USA) of
capabilities [2–7]. Furthermore with proper manipulation of thickness 0.25 mm. The anodization was performed in an
the microstructure, crystalline phase and/or addition of electrolyte medium of 0.5% hydrofluoric acid (J. T. Baker-
proper impurities or surface functionalization titania can Phillipsburg, NJ, USA) in water, using a platinum foil
also be used as a reducing gas sensor [8–19]. The interaction cathode. Well-defined nanotube arrays were grown using
of a gas with a metal oxide semiconductor is primarily a anodizing potentials ranging from 12 to 20 V. Nanotube
surface phenomenon, therefore nanoporous metal oxides length increases with anodization time, reaching a length of
[14,15,20,21] offer the advantage of providing large sensing 400 nm in approximately 20 min, and then remains constant.
surface areas. For the present study the samples were anodized for 25 min.
Hydrogen sensing is needed for industrial process control, The samples were then annealed at 500 8C in a pure oxygen
combustion control, and in medical applications with hydro- ambient for 6 h, with a heating and cooling rate of 1 8C/min.
gen indicating certain types of bacterial infection. In this A field emission scanning electron microscope (FESEM)
work we report on the hydrogen sensing properties of titania from JEOL (model JSM6300), Peabody, MA, USA was used
nanotubes made via anodization [1]. Room temperature to study the surface morphology of the nanotubes. A glan-
metal oxide hydrogen sensors are generally based on cing angle X-ray diffractometer (GAXRD) from Philips
Schottky barrier modulation of devices like Pd/TiO2 or (model X’pert MRD PRO), The Netherlands was used to
determine the crystalline phase.
*
Corresponding author. The electrode geometry of the titania nanotube sensors is
E-mail address: cgrimes@engr.psu.edu (C.A. Grimes). shown in Fig. 1a. The sensor consists of a base titanium

0925-4005/03/$ – see front matter # 2003 Elsevier Science B.V. All rights reserved.
doi:10.1016/S0925-4005(03)00222-3
O.K. Varghese et al. / Sensors and Actuators B 93 (2003) 338–344 339

Fig. 1. Schematic representation of (a) the electrode geometry, and (b) the experimental apparatus.

metal foil with a nanotube array grown on the top. An under test was heated to the desired temperature. The test
insulating barrier layer separates the nanotubes from the gases examined, oxygen, carbon dioxide, ammonia, carbon
conducting titanium foil. Preliminary studies using evapo- monoxide or hydrogen, were mixed in appropriate ratios
rated gold films as well as silver paste as interdigital with nitrogen to create the necessary test gas ambient.
electrodes showed that these materials diffuse into the titania
nanotubes at elevated temperatures resulting in sensor con-
tamination. Hence a pressure contact was used to electrically 3. Results and discussion
contact the nanotubes, with two spring-loaded parallel
10 mm by 2 mm platinum contact pads (100 mm thickness). The surface morphology of nanotube arrays prepared
A schematic diagram of the experimental set up used for using an anodization potential of 20 V and annealed at
the gas sensing studies is shown in Fig. 1b. The test chamber 500 8C for 6 h in a pure oxygen ambient is shown in
consists of a 1.3 l quartz tube, with stainless steel end caps, Fig. 2a and b. It can be seen from these images that the
placed inside a furnace (Thermolyne, USA model 21100 nanotubes are uniform over the surface. The nanotubes are
tubular furnace). The electrical contact was formed by approximately 400 nm in length and have a barrier layer [1]
attaching the platinum pads to the ends of a spring-loaded thickness of 50 nm. For the nanotubes fabricated using
‘U’ shaped quartz rod. Gas flow through the test chamber 20 V anodization the average pore diameter, as determined
was controlled via a computer-controlled mass flow con- from FESEM images, is 76 nm (standard deviation 15 nm),
troller (MKS instruments, Austin, TX, USA). The electrical with a wall thickness of 27 nm (standard deviation 6 nm).
resistance of the titania sensors were measured using a The sample anodized at 12 V was found to have an average
computer-controlled digital multimeter (Keithley, USA pore diameter of 46 nm (standard deviation 8 nm) with a
model 2000). Prior to data collection the test chamber wall thickness 17 nm (standard deviation 2 nm). The por-
was initially evacuated using a mechanical pump, where- osities of the 20 and 12 V samples were calculated as 45 and
upon nitrogen (99.999% pure) was passed while the sensor 61%, respectively.
340 O.K. Varghese et al. / Sensors and Actuators B 93 (2003) 338–344

Glancing angle X-ray diffraction patterns of a 20 V


sample annealed at 500 8C for 6 h in oxygen ambient is
shown in Fig. 3. It can be seen that both anatase and rutile
phases of titania are present in the sample. A detailed study
[26] of these structures using high resolution transmission
electron microscope (HRTEM) showed that the anatase
crystallites were concentrated on the walls of the nanotubes
and rutile on the barrier layer.
Nanotubes annealed in a pure oxygen ambient were found
to be stable (intact) to temperatures of approximately
580 8C. Above this temperature protrusions were seen com-
ing out through the nanotubes, an effect which spread with
increasing temperature. These protrusions, which are due to
oxidation of the titanium substrate, collapse the nanotubes.
Fig. 4 shows the response of the 20 V sample as a function
of ambient temperature, as it is switched from a nitrogen
environment to one containing 1000 ppm hydrogen, and
then back to nitrogen. The plot was made using (Rg/R0)1
versus time where R0 is the base resistance of the sensor, i.e.
the sensor resistance before introducing the test gas, and Rg
the measured resistance in the presence of test gas. The
sensor shows increasing hydrogen sensitivity with tempera-
ture, with a three order of magnitude change in resistance at
temperatures above 300 8C. At all the temperatures the
original resistance it recovered without hysteresis.
The sensitivity S is defined by the formula
R0  Rgs

Rgs
where R0 is the resistance of the sensor before passing the
gas and Rgs that after passing gas and reaching the saturation
value. The sensitivity of a 20 V sample with temperature, to
1000 ppm hydrogen, is shown in Fig. 5. Sensitivity is seen to
increase with temperature to approximately 380 8C where
the increase in sensitivity with temperature is beginning to
saturate.
The response time, defined as the time needed for the
sensor to reach 90% of the final signal for a given concen-
Fig. 2. The surface morphology of the titania nanotubes after annealing at tration of gas, is plotted against temperature in Fig. 6 (the
500 8C: (a) high, and (b) low magnification images of a 20 V sample, and
time includes that required for the gas to equilibrate inside
(c) a high magnification image of a 12 V sample.

Fig. 3. Glancing angle X-ray diffraction pattern of a 20 V sample (glancing angle ¼ 28) annealed at 500 8C in oxygen ambient. A, R and T represent the
reflections from anatase crystallites, rutile crystallites, and the titanium substrate, respectively.
O.K. Varghese et al. / Sensors and Actuators B 93 (2003) 338–344 341

Fig. 4. Variation of resistance Rg, normalized with respect to baseline resistance R0, of a 20 V sample with time on exposure to 1000 ppm hydrogen at
different temperatures. It may be noted that the inverse of Rg/R0 was used in the plot for representing data in positive y-direction.

the measurement chamber, estimated to be 30 s). The of the 20 V prepared nanotube sensors, kept at 290 8C, is
response time reduces exponentially with temperature. shown in Fig. 7. The behavior of the sensor is consistent,
To check the behavior of the sensor on repeated hydrogen recovering its original resistance after repeated exposure to
exposure, the hydrogen concentration was varied in discrete varying hydrogen gas concentrations. The sensitivity of the
steps of 100 ppm from 0 to 500 ppm while keeping the sensor in this concentration range is plotted in Fig. 8a; there
temperature constant at 290 8C; the chamber was flushed is a linear increase in sensitivity at low concentrations. The
with nitrogen after each exposure to hydrogen. The response sensitivity of the hydrogen sensor over 100 ppm to 4% (the
explosive limit in the presence of oxygen) is shown in
Fig. 8b.
Fig. 9 shows the hydrogen sensitivities, at 290 8C, of
nanotube sensors having a pore diameter of 76 nm, and a
pore diameter of 46 nm. While smaller diameter nanotubes
had greater sensitivity to hydrogen the samples made at
lower anodizing voltages tended to be more brittle, and
harder to mechanically handle without breaking.
The 20 V sample was exposed to oxygen, carbon mon-
oxide, ammonia and carbon dioxide at 290 8C. The sensor
was found to have no detectable variation in resistance on
exposure to carbon dioxide. The sensitivities of the titania

Fig. 5. The sensitivity temperature dependence of a 20 V sample to


1000 ppm hydrogen.

Fig. 7. Resistance of a 20 V sample when exposed to different


concentrations of hydrogen at 290 8C. The nitrogen–hydrogen mixture
Fig. 6. Response time variation of a 20 V sample to temperature. The dots was passed for 1500 s; the chamber was then flushed with nitrogen for
represent measured data. 3000 s before passing the nitrogen–hydrogen mixture again.
342 O.K. Varghese et al. / Sensors and Actuators B 93 (2003) 338–344

Fig. 10. Variation of resistance of a 20 V sample when exposed to


1000 ppm carbon monoxide, 5% ammonia and 1% oxygen at 290 8C.

But this process would lead to very slow response and


recovery times and complete recovery would be difficult.
Since the sensor completely regains its original resistance
with hydrogen cycling it appears that this is not the domi-
nant mechanism behind high hydrogen sensitivity. Hence,
we believe that the major process behind the interaction
between the nanotubes and hydrogen is the chemisorption
of the dissociated hydrogen on the titania surface [32].
During chemisorption hydrogen acts as a surface state
Fig. 8. The sensitivity variation of a 20 V sample at 290 8C for (a) and a partial charge transfer takes place from hydrogen to
low hydrogen concentrations, and (b) 0.01 to 4% hydrogen concentra- the conduction band of titania. This creates an electron
tions. accumulation layer on the nanotube surface that enhances
its electrical conductance. On removing the hydrogen
nanotubes to the other gases is shown in Fig. 10. The ambient, electron transfer takes place back to hydrogen
sensitivity of the nanotubes to carbon monoxide and ammo- and it desorbs, thus restoring the original resistance of the
nia are negligible compared to that of hydrogen. The resis- nanotubes.
tance of the nanotubes increased in presence of oxygen, and Another factor that may play a role in the hydrogen
did not regain their original electrical conductivity even after sensitivity (and selectivity) is the platinum electrodes. It
several hours in a nitrogen environment. is possible that platinum is acting as a catalyst for interaction
Since the sensor measurements were conducted in atmo- of hydrogen with titania. At elevated temperatures hydrogen
spheres without oxygen, the increase in conductivity cannot dissociation can occur on platinum surfaces. These disso-
be due to hydrogen removing oxygen from the lattice ciated hydrogen atoms may spill [31,35] onto the nanotube
[27–29] or the removal of chemisorbed oxygen [30–32]. surface where they diffuse into the nanotube surface. From
It is likely that the hydrogen molecules get dissociated at the the present study it was not clear how significant a role the
defects on the titania surface. These molecules can diffuse platinum electrodes play.
into the titania lattice, and act as electron donors [25,33,34]. Anatase, the polymorph of titania has been reported to be
of high sensitivity towards reducing gases like hydrogen and
carbon monoxide [13,16,25]. Our nanotube samples contain
anatase phase mainly on the walls and rutile in the barrier
layer. As the diffusing hydrogen atoms go to the interstitial
sites [25,33] and as the c/a ratio of anatase is almost four
times compared to that of rutile, it appears that anatase
lattice accommodates hydrogen easily and hence has a
higher contribution to hydrogen sensitivity.
The effect of chemisorption can be neglected in the
oxygen sensing experiments. As the recovery requires
several hours it appears that the nanotubes contain oxygen
vacancies or titanium interstitial defects in presence of
Fig. 9. A comparison of the variation in resistance of samples having pore nitrogen. On exposing the sensor to oxygen ambient, the
diameters of 46 and 76 nm, vs. time, upon exposure to 1000 ppm of lattice reoxidizes and hence the conductivity of the sensor
hydrogen at 290 8C. decreases. On removing oxygen, the reduction of the lattice
O.K. Varghese et al. / Sensors and Actuators B 93 (2003) 338–344 343

will not immediately occur hence the sensor requires several [6] S. Hasegawa, Y. Sasaki, S. Matsuhara, Oxygen-sensing factor of
hours to regain its original conductivity. TiO2 doped with metal ions, Sens. Actuators B 13-14 (1993) 509.
[7] A. Rothschild, F. Edelman, Y. Komem, F. Cosandey, Sensing
It should be noted that the conducting titanium foil behavior of TiO2 thing films exposed to air at low temperatures, Sens.
beneath the nanotubes ultimately limits the sensitivity by Actuators B 67 (2000) 282.
reducing the baseline resistance of the sensor. [8] G. Sakai, N.S. Baik, N. Miura, N. Yamazoe, Gas sensing properties
of tin oxide thin films fabricated from hydrothermally treated
nanoparticles dependence of CO and H2 response on film thickness,
Sens. Actuators B 77 (2001) 116.
4. Conclusions [9] V.N. Misra, R.P. Agarwal, Thick film hydrogen sensor, Sens.
Actuators B 21 (1991) 209.
Titania nanotubes prepared using anodization and [10] O.K. Varghese, L.K. Malhotra, G.L. Sharma, High ethanol sensiti-
annealed in an oxygen atmosphere at a temperature of vity in sol–gel derived SnO2 thin films, Sens. Actuators B 55 (1999)
500 8C were found highly sensitive to hydrogen. The nano- 161.
[11] V.A. Chaudhary, I.S. Mulla, K. Vijayamohanan, Comparative studies
tube sensors contain both anatase and rutile phases of titania, of doped and surface modified tin oxide towards hydrogen sensing:
and showed appreciable sensitivity towards hydrogen at synergistic effects of Pd and Ru, Sens. Actuators B 50 (1998) 45.
temperatures as low as 180 8C. The sensitivity increased [12] H. Tang, K. Prasad, R. Sanjines, F. Levy, TiO2 anatase thin films as
drastically with temperature showing a variation of three gas sensors, Sens. Actuators B 26-27 (1995) 71.
orders in magnitude of resistance to 1000 ppm of hydrogen [13] S.A. Akbar, L.B. Younkman, Sensing mechanism of a carbon
monoxide sensor based on anatase titania, J. Electrochem. Soc. 144
at 400 8C. The response time decreased with increasing (1997) 1750.
temperature; at 290 8C full switching of the sensor took [14] Y. Shimizu, N. Kuwano, T. Hyodo, M. Egashira, High H2 sensing
approximately 3 min. Results were highly reproducible with performance of anodically oxidized TiO2 film contacted with Pd,
no indication of hysteresis. Our results showed these sensors Sens. Actuators B 83 (2002) 195.
are capable of monitoring hydrogen levels from 100 ppm to [15] M.C. Carotta, M. Ferroni, D. Gnani, V. Guidi, M. Merli, G.
Martinelli, M.C. Casale, M. Notaro, Nanostructured pure and Nb
4%. At 290 8C nanotubes with smaller pore diameter doped TiO2 as thick film gas sensors for environmental monitoring,
(46 nm) showed higher sensitivity to hydrogen compared Sens. Actuators B 58 (1999) 310.
to larger pore diameter samples (76 nm). The sensors [16] N.O. Savage, S.A. Akbar, P.K. Dutta, Titanium dioxide based high
showed high selectivity to hydrogen compared to carbon temperature carbon monoxide selective sensor, Sens. Actuators B 72
monoxide, ammonia and carbon dioxide. Although the (2001) 239.
[17] E. Comini, G. Faglia, G. Sberveglieri, Y.X. Li, W. Wlodarski, M.K.
sensor was sensitive to high concentrations of oxygen, the Ghantasala, Sensitivity enhancement towards ethanol and methanol
response time was high and the sensor did not completely of TiO2 films doped with Pt and Nb, Sens. Actuators B 64 (2000)
regain the original condition. We believe that the hydrogen 169.
sensitivity of the nanotubes is due to hydrogen chemisorp- [18] I. Hayakawa, Y. Iwamoto, K. Kikuta, S. Hirano, Gas sensing
tion onto the titania surface where they act as electron properties of platinum dispersed-TiO2 thin film derived from
percursor, Sens. Actuators B 62 (2000) 55.
donors. In summary, it was demonstrated that sensors com- [19] N. Savage, B. Chwieroth, A. Ginwalla, B.R. Patton, S.A. Akbar, P.K.
prised of titania nanotubes prepared using anodization can Dutta, Composite n–p semiconducting titanium oxides as gas sensors,
successfully be used as hydrogen sensors. Sens. Actuators B 79 (2002) 17.
[20] C.C. Koch (Ed.) Nanostructured Materials: Processing, Properties
and Applications, Noyes publications, New York, USA, 2002.
[21] H.-M. Lin, C.-H. Keng, C.-Y. Tung, Gas sensing properties of
Acknowledgements nanotcrystalline TiO2, Nanostructured Mater. 9 (1997) 747.
[22] H. Kobayashi, K. Kishimoto, Y. Nakato, Reactions of hydrogen at
Partial support of this work by the National Science the interface of palladium–titanium dioxide schottky diodes as
Foundation through grant ECS-9875104 is gratefully hydrogen sensors studied by work function and electrical character-
acknowledged. istic measurements, Surf. Sci. 306 (1994) 393.
[23] L.A. Harris, A titanium dioxide hydrogen sensor, J. Electrochem.
Soc. Solid State Sci. Technol. 127 (1980) 2657.
[24] K.D. Schierbaum, U. K Kirner, J.F. Geiger, W. Gopel, Schottky
References barrier and conductivity gas sensors based upon Pd/SnO2 and Pt/
TiO2, Sens. Actuators B 4 (1991) 87.
[1] D. Gong, C.A. Grimes, O.K. Varghese, W. Hu, R.S. Singh, Z. Chen, [25] L.D. Birkefeld, A.M. Azad, S.A. Akbar, Carbon monoxide and
E.C. Dickey, Titanium oxide nanotube arrays prepared by anodic hydrogen detection by anatase modification of titanium dioxide, J.
oxidation, J. Mater. Res. 16 (2001) 3331. Am. Ceram. Soc. 75 (1992) 2964.
[2] T. Takeuchi, Oxygen sensors, Sens. Actuators 14 (1988) 109. [26] O.K. Varghese, D. Gong, M. Paulose, C.A. Grimes, E.C. Dickey,
[3] U. Kirner, K.D. Schierbaum, W. Gopel, Low and high temperature Structural stability of titanium oxide nanotube arrays, J. Mater. Res.
TiO2 oxygen sensors, Sens. Actuators B 1 (1990) 103. 18 (2003) 156.
[4] Y. Xu, K. Yao, X. Zhou, Q. Cao, Platinum–titania oxygen sensors [27] R.D. Shannon, Phase transformation studies in TiO2 supporting
and their sensing mechanisms, Sens. Actuators B 13-14 (1993) different defect mechanisms in vacuum-reduced and hydrogen-
492. reduced rutile, J. Appl. Phys. 35 (1964) 3414.
[5] V. Demarne, S. Balkanova, A. Grisel, D. Rosenfeld, F. Levy, [28] K.H. Kim, E.J. Ju, J.S. Choi, Electrical conductivity of hydrogen
Integrated gas sensor for oxygen detection, Sens. Actuators B 13-14 reduced titanium dioxide (rutile), J. Phys. Chem. Solids 45 (1984)
(1993) 497. 1265.
344 O.K. Varghese et al. / Sensors and Actuators B 93 (2003) 338–344

[29] G.C. Mather, F.M.B. Marques, J.R. Frade, Detection mecha- [32] G.B. Raupp, J.A. Dumesic, Adsorption of CO, CO2, H2 and H2O on
nism of TiO2 ceramic H2 sensors, J. Eur. Ceram. Soc. 19 (1999) titania surfaces with different oxidation states, J. Phys. Chem. 89
887. (1985) 5240.
[30] R.M. Walton, D.J. Dwyer, J.W. Schwank, J.L. Gland, Gas sensing [33] J.B. Bates, J.C. Wang, R.A. Perkins, Mechanisms for hydrogen
based on surface oxidation/reduction of platinum–titania thin films. diffusion in TiO2, Phys. Rev. B 19 (1979) 4130.
Part 2. The role of chemisorbed oxygen in film sensitisation, Appl. [34] G.J. Hill, The effect of hydrogen on the electrical properties of rutile,
Surf. Sci. 125 (1998) 199. Br. J. Appl. Phys. 1 (1968) 1151.
[31] M.J. Madou, S.R. Morrison, Chemical Sensing with Solid State [35] U. Roland, T. Braunschweig, F. Roessner, On the nature of split-over
Devices, Academic Press, New York, 1989. hydrogen, J. Mol. Catal. A: Chem. 127 (1997) 61.

You might also like