Вы находитесь на странице: 1из 10

RSC Advances

PAPER

Highly stable low temperature alcohol sensor


based on hydrothermally grown tetragonal titania
Cite this: RSC Adv., 2015, 5, 82159
nanorods
B. Bhowmik and P. Bhattacharyya*

Highly stable, low temperature (27–175 C) alcohol (ethanol, methanol and 2-propanol) sensing
performance of a hydrothermally grown tetragonal TiO2 nanorod array, targeting the detection level of
1–100 ppm, is reported in this paper. Synthesized titania nanorods were characterized by X-ray
diffraction, energy dispersive spectroscopy (EDS) and field emission scanning electron microscopy
(FESEM) to investigate the crystallinity, chemical compositions and surface morphology, respectively.
Photoluminescence spectroscopy (PL spectra) was employed to find out the surface oxygen vacancies
and band gap of the material. An alcohol sensing study, in resistive mode, was carried out and the
optimum temperature for alcohol sensing was found to be 75  C with the corresponding maximum
response magnitude of 69%, 75% and 80% towards 100 ppm ethanol, methanol and 2-propanol,
with corresponding response time/recovery time of 13 s/24 s, 11/21 s and 9/18 s, respectively. The
Received 22nd July 2015
Accepted 21st September 2015
sensor also offered efficient detection (response magnitude of 29%, 37% and 40%) of 1 ppm alcohol(s)
with fast response time and recovery time of 12–14 s and 4–5 s, respectively. The minimal cross
DOI: 10.1039/c5ra14518j
sensitivity, with other interfering species like acetone, benzene, xylene and 2-butanone, and appreciable
www.rsc.org/advances long term stability showed the potential of the developed sensor for commercial applications.

desorption kinetics, but non-uniformity in the nanoforms (lack


1. Introduction of repeatability in the nanoforms limits the regular array
One dimensional (1-D) oxide nanoforms with tuneable struc- formation probability) severely constraints the repeatability and
tural, morphological and electronic properties have recently reliability of the sensors.2,4 However, fairly good as well as high
attracted the attention of gas/vapor sensor researchers.1–5 The surface to volume ratio of the nanorods makes them nearest
manipulation of one dimensional (1-D) nanostructures viz., competitor to the nanotubes towards gas/vapor sensing.6–8
nanobelts,2 nanotubes,3 nanowires4 and nanorods5 is However, literature review reveals that gas/vapor sensors based
immensely favourable for gas adsorption owing to their high on TiO2 nanorods have not been extensively investigated.
surface to volume ratio, possible quantum connement and Moreover, these sensors offered appreciable response magni-
one dimensional electron transport eventually resulting in tude ([(Ra  Rg)/Ra]  100 > 90%), but usually at the cost of
improved speed of the device. Various nanoforms of the semi- higher operating temperature (180–500  C).5–10 For example,
conducting metal oxides like ZnO, SnO2 and TiO2 were inves- highly crystalline TiO2 nanorods, doped with 3 wt% Nb offered
tigated.2–5,7,8 Among them, nanoforms of TiO2 (nanotubes, the maximum response magnitude of 64%, 94% and 91%
nanowires and nanorods) received serious attention in recent towards 400 ppm H2, 4000 ppm C2H5OH and 4000 ppm LPG at
years owing to its high chemical and physical stability.3,12,13 In 500  C.10 At the same temperature (500  C), response magnitude
this category, nanotubes offer easy diffusion (and subsequent of 95% towards acetone, employing electrospun TiO2/PVP
adsorption) of the gas molecules into the core of the nano- hybrid nanorods structure, were evident in the work of Bian
structure (as both the inner and outer surfaces are available for et al.9 Matrix assisted pulse laser evaporated (MAPLE) TiO2
gas interaction), but, the complete desorption from such nanorods showed trace level NO2 (30 ppb to 50 ppm) detection
nanostructure takes longer time eventually leading to sluggish capability with corresponding response magnitude of 95% at
recovery characteristics of the device.3 On the other hand, 300  C temperature.8 However, the sluggish response time and
though, nanowire and nanobelts attribute relatively faster recovery time of 120 s and 780 s was the primary disadvantage of
this sensor.8 At similar temperature (300  C), use of Pt doping
on TiO2 nanorods was found to be benecial in making the
Nano Thin Film and Solid State Gas Sensor Devices Laboratory, Department of
ethanol response time and recovery time faster (9–12 s and 10–
Electronics and Telecommunication Engineering, Indian Institute of Engineering
Science and Technology, Shibpur-711103, Howrah, India. E-mail: pb_etc_besu@
15 s, respectively with response magnitude of 95%).7 In the work
yahoo.com; Fax: +91 3326682916; Tel: +91 3326684561 ext. 544 of Chen et al.,6 it was found that, a-MoO3/TiO2 core/shell

This journal is © The Royal Society of Chemistry 2015 RSC Adv., 2015, 5, 82159–82168 | 82159
RSC Advances Paper

nanorods hybrid structure showed 93% response to ethanol at (Sigma Aldrich) was mixed drop wise till the solution turned
only 180  C with response time/recovery time of 40 s/40 s. On transparent. FTO substrate was placed in Teon beaker, con-
the other hand, TiO2/ZnO nanorod hybrid structure offered taining the solution, at an angle of 45 against the wall with
appreciably high response magnitude of 99% towards ethanol conducting surface facing upward. The complete set up was
at only 150  C at the expense of sluggish response time (70 s).5 then placed in an autoclave (70 ml) and transferred to an oven
As reected from the earlier reports, on the TiO2 nanorod with temperature 120  C  2  C for 2 hours. Finally, autoclave
based gas/vapor sensors, extensive efforts have been exerted to was allowed to cool naturally to room temperature. The oxide
maximize the sensor response magnitude (summarized in coated FTO glass was then rinsed in de-ionized water. The dried
Table 2).5–10 To the contrary, other crucial sensor parameters samples were then annealed at 350  C for 3 hours.
like selectivity, stability, operating temperature and response Crystallinity of the sample was analysed using X-ray diffrac-
time/recovery time were relatively less explored arena.5–10 It is tometer (Rigaku MiniFlex II, Cu Ka radiation with l ¼ 1.54 Å at
evident that, existing TiO2 nanorod based sensors, either suffer 2q angle of 20–75 with the scanning rate of 0.02 ). The energy
mostly from higher operating temperature (150–550  C) or their dispersive spectroscopy (EDS) (EDAX, model: EAG AN461) was
response time is relatively sluggish (>60 s) making them inap- carried out to determine the elemental composition in the
propriate for real life applications.5–10 Moreover, such higher sample. The size, thickness and surface morphology of the
operating temperatures of the device not only increase the deposited lm was investigated through eld emission scan-
overall power consumption but also become the prime obstacle ning electron microscopy (ZEISS SIGMA). Band gap and oxygen
to integrate it in a monolithic gas sensor system.11 Therefore vacancies were obtained by photoluminescence spectroscopy
further reduction of operating temperature with appreciable (PL spectra) using Horiba Jobin Yvon, Fluorolog-3 spectrouo-
stability and selectivity, without jeopardizing the response/ rometer (model FL3-22).
recovery kinetics, is a timely demand. The schematic of the two terminal resistive device congu-
In present study, one dimensional TiO2 nanorod array (due ration, employing nanorods as the sensing layer, is shown in
to their numerous benecial features) was employed to develop Fig. 1. Two Pd electrodes having dimensions of 2 mm  1 mm
resistive device for alcohol sensing with an aim to enhance the  50 nm were deposited by e-beam evaporation techniques. The
sensing features compared to present state of art with key focus gap between the two electrodes was maintained at 3 mm. A
on operating temperature, selectivity and stability improve- barrier layer of TiO2 having thickness of 90–110 nm was
ment. In the present endeavour, we report on dramatic formed between FTO substrate and TiO2 nanorods. Such barrier
improvement in the alcohol sensing characteristics, primarily layer was the initial formation of oxide layer on FTO substrate by
the parameters mentioned above, employing tetragonal nano- precipitation of the intense TiO2 gel.6,7,10
rod array of TiO2 as the sensing layer. 1-D tetragonal titania A dynamic gas ow measurement system was used to
nanorods, on FTO glass substrate, were prepared by a low measure the gas sensing properties of the TiO2 nanorod based
temperature (120  C) hydrothermal process. Resistive congu- sensor devices. The details of the gas sensor measurement
ration of the TiO2 nanorod array offered the maximum response system were already described in our earlier publications.12,13
(69%, 75% and 80%) to alcohol (ethanol, methanol and 2- The measurement system consists of a sensing chamber,
propanol) at relatively low temperature (75  C) with appreciably temperature controller, mass ow controllers, Keithley 6487
fast response time and recovery time of (5–30 s) and (4–22 s), picoammeter/programmable voltage source with computer
respectively. Moreover, sensor showed the minimum alcohol interfacing. Two mass ow controllers (MFC) (MKS USA) were
detection level of only 1 ppm with corresponding response used; one to control the ow of carrier gas (air in the present
magnitude of 29%, 37% and 40% towards ethanol, case) through the liquid volatile organic compounds (VOCs)
methanol and 2-propanol, respectively. The minimal cross and other one to control the ow of diluent gas (air in the
sensitivity against other interfering species (acetone, benzene,
xylene and 2-butanone) and excellent long term stability (reli-
ability) of the device paved the path of a promising alcohol
sensor for real life applications mitigating the constraints of the
existing state of art.

2. Experimental
Fluorine doped FTO glass (F:SnO2, Tec 10, Sigma Aldrich) was
used as the substrate for titania nanorod growth. FTO glass was
thoroughly cleaned with acetone, ethanol and de-ionized (18.2
MU cm) water. Titanium tetra(IV) iso-propoxide (TTIP), HCl,
CH3COOH and water were used for the growth of TiO2 nano-
rods. Initially, 25 ml 37% HCl (Merck) was mixed with 40 ml
H2O. Then 5 ml 98.8% CH3COOH (Merck) was added to that Fig. 1 Schematic of the tetragonal TiO2 nanorod array based two
solution. The solution was then stirred vigorously. Under stir- terminal resistive device configurations with requisite dimensions (not
ring condition 0.5 ml titanium tetra(IV) iso-propoxide (TTIP) in scale).

82160 | RSC Adv., 2015, 5, 82159–82168 This journal is © The Royal Society of Chemistry 2015
Paper RSC Advances

present case) to achieve the desired concentrations of VOCs in Fig. 2(b) shows the EDS spectra of the TiO2 sample. The
the test chamber. The diluted gas/vapor was then allowed to major elemental composition was found to be Ti and O. The
ow in the sensing chamber. The temperature of the sensing obtained Ti and O weight percentage in the sample were
chamber was maintained by a proportional-integral-derivative calculated to be 46.06% and 47.80%, respectively. The presence
(PID) temperature controller with temperature accuracy of of Sn was found due to the FTO substrate. Non existence of
1  C. Before starting the measurement, sensor device was pre- other peaks in the EDS spectra authenticates the purity of the
heated at 100  C for 1 hour to remove the unwanted adsorbed TiO2 thin lms. Fig. 2(c) shows the FESEM image (top view) of
molecules. Response magnitude was calculated following the the grown TiO2 nanorod array. FESEM image conrms the
eqn (1). formation of densely packed tetragonal nanorod array with
  square top facet (magnied view shown in the inset). The
Ra  Rg
RM ¼  100 (1) lengths of the nanorods were found to be 710–750 nm.
Ra
Lengths of the side arm(s) and diagonal of the tetragonal
where Ra represents the sensor resistance in air and Rg repre- nanorod were 90–94 nm and 120–126 nm, respectively. Such
sents the same in test gas/vapor.12 unique structure offer high specic surface area and higher
concentrations of interaction sites for gas/vapor sensing.
3. Results and discussions Nanorod growth was started by hydrolysis of titanium tetra(IV)
iso-propoxide (TTIP) in the aqueous media of HCl and CH3-
3.1. Structural characterizations COOH. A ligand exchange reaction between TTIP and CH3-
XRD pattern of the annealed TiO2 thin lm containing tetrag- COOH occurred where titanium tetra was replaced by acetoxyl
onal nanorod array is shown in Fig. 2(a). The diffraction peaks group.16,17 The hydrolysis of TTIP in acidic medium produces a
at 30.02 , 39.4 , 43.89 are indexed as (116), (305) and (400) complex polymeric Ti(IV) alkoxides which are saturated in
planes of TiO2 anatase phase. On the other hand, peaks at solution and make a network of –Ti–O– bond together. Such
64.09 and 72.46 correspond to (310) and (311) plane of TiO2 reactions produce a seed layer of TiO2 with a large number of
rutile phase. All the planes in both anatase and in rutile phases heterogeneous nucleation centres on FTO substrate.7,10,16,17 At
are in agreement with the standard JCPDS le no. 21-1272 (ref. elevated temperature, the dissolution of seed layer (by concen-
14) and 86-0148.15 The crystalline size of the TiO2, from the trated HCl) followed by the deposition produces uniform
Scherrer's formula, was found to be 35–37 nm.

Fig. 2 (a) XRD pattern of the hydrothermally grown TiO2 nanorod array (b) EDS spectra of the hydrothermally grown TiO2 nanorod array (c)
FESEM image of the hydrothermally grown TiO2 nanorods (inset shows the magnified view). (d) Room temperature PL spectra of the grown TiO2
nanorods.

This journal is © The Royal Society of Chemistry 2015 RSC Adv., 2015, 5, 82159–82168 | 82161
RSC Advances Paper

nanorod, situated vertically, on FTO substrate. The role of HCl additional advantage and offer opportunity to integrate the
is to grow the crystal TiO2 into one direction by controlling the same with a monolithic gas sensor system.11
dehydration of Ti alkoxides. Further, the minimum lattice Sensor response towards different alcohols for different
mismatch between FTO substrate and anatase/rutile TiO2 plays temperatures was investigated and the results are shown in the
a vital role for uniform nucleation as well as growth of TiO2 Fig. 4(a). It was found that the response magnitude increased up
nanorods.17 to 75  C and then decreased with increase in temperature. Such
Room temperature PL spectra of the grown sample were nonlinear variation in response with temperatures is possibly
carried out for the excitation wavelength of 300 nm and the attributed to the dissimilar adsorption–desorption kinetics at
results are shown in Fig. 2(d). The corresponding PL intensity different temperatures.12 At room temperature (27  C), sensor
were recorded in the wavelength range of 350–550 nm. Mainly showed the response magnitude of 0.8/12%, 1.02/14% and 1.36/
three peaks were observed at 399 nm, 428 nm and 461 nm. The 15% towards 1/100 ppm of ethanol, methanol and 2-propanol,
peak at 399 nm is related to band to band emission i.e. direct respectively. The optimum operating temperature was found to
transition of electron from valence band to conduction band. be only 75  C which is signicantly lower compared to the
From the direct transition, the calculated band gap was found earlier reported TiO2 nanorod based gas/vapor (180–500  C)
to be 3.1 eV. The peaks at 428 nm reveals the band to band sensors.5–10 Such high operating temperatures of those sensors
indirect transition of electron which further proves the states was the prime obstacle to integrate them into the monolithic
are intrinsic rather than surface state.18 The peak at 461 nm gas sensor system.11 At optimum temperature (75  C), the
corresponds to the presence of surface oxygen vacancies (OVs).19 maximum response of the present sensor was found to be 29/
Higher peak intensity at 461 nm than that of the band edge 69%, 37/75% and 40/80% towards 1/100 ppm ethanol,
emission peak at 399 nm conrms the existence of higher methanol and 2-propanol, respectively. Sensor response
amount of OVs than that of Ti vacancies.19 magnitude variations as a function of alcohol concentrations
was also plotted for the temperature of 27  C and 75  C and
shown in Fig. 4(b). As revealed from this gure, at 75  C,
3.2. Sensor study response magnitude increased almost linearly with a steep
Alcohol sensing performance of the TiO2 nanorod array was nature up to 10 ppm and aer that increased in a linear fashion
tested in the temperature range of 27–175  C. The experiment
started at room temperature (27  C) then in the range of
50–175  C, the temperature was increased at a step of 25  C.
Baseline resistance of the sensor device as a function of oper-
ating temperature were investigated in air, 1 ppm and 100 ppm
of alcohols (ethanol, methanol and 2-propanol) and the results
are shown in Fig. 3. In exposure of air, baseline resistance of the
sensor was 6.75 MU and 0.34 MU at room temperature
(27  C) and at 175  C, respectively. On the other hand, with
exposure to 1 ppm/100 ppm alcohols, the baseline resistance
variations were found to be 6.2/5.5 MU (at 27  C) and 0.2/0.05
MU (175  C), respectively. Such a relatively lower baseline
resistance value (<100 MU) of the present sensor is an

Fig. 4 (a) Sensor response magnitude as a function of operating


temperatures for the alcohol concentrations of 1 ppm and 100 ppm.
Fig. 3 Sensor resistance as a function of operating temperatures in air, (b) Sensor response magnitude as a function of alcohol concentrations
ethanol, methanol and 2-propanol; inset shows the magnified view of at 75  C operating temperature; inset shows response magnitude as a
the selected portions. function of concentrations at room temperature.

82162 | RSC Adv., 2015, 5, 82159–82168 This journal is © The Royal Society of Chemistry 2015
Paper RSC Advances

75  C. Sensor showed appreciable recovery characteristics


within <40 s aer cutting off the alcohol pulse. The baseline
resistance of the sensor was found to remain almost constant
with a deviation of only 0.5–1.7% for the all the cases (ethanol,
methanol and 2-propanol, for all the concentrations). Moreover,
at even 1 ppm ethanol, methanol and 2-propanol, sensor
offered acceptable response magnitude of 29%, 37% and
40%, respectively. Such appreciable response magnitude of
TiO2 nanorods towards trace level of alcohol (1 ppm) at such a
lower temperature (75  C) is a signicant improvement with
reference to the existing state of art.5–10
Fig. 5 Transient characteristics of the TiO2 nanorod based resistive The effect of relative humidity (% RH) on the sensing
device towards ethanol, methanol and 2-propanol at 75  C for the performance of the device was carried out at 75  C with expo-
concentrations of 1, 5, 10, 50 and 100 ppm. sure to 1 ppm and 100 ppm alcohols (ethanol, methanol and 2-
propanol) for humid air background (8% RH, 52% RH and 75%
RH). The maximum base line resistance variation was found to
with relatively lower slope. On the other hand, at room be 1.56%, 2.26% and 2.74% for 1 ppm concentration and
temperature also, sensor showed similar increasing trends in 2.33%, 3.63% and 3.75% for 100 ppm alcohols (ethanol,
response magnitude but the overall response magnitude was methanol and 2-propanol) in 8% RH, 52% RH and 75% RH,
found to be lower than that of the response magnitude at 75  C respectively. Sensor response magnitude was found to be
(shown as inset of Fig. 4(b)). At very lower concentrations (#10 decreased with increase in relative humidity (% RH). The sensor
ppm at present study), the surface coverage's (q) as well as response magnitude (RM%) towards alcohols at dry air and
diffusion of alcohol molecules is very fast which manifest itself different relative levels of humidity is shown in Table 1.
as a rapid increase in response magnitude. However, with Fig. 6(a) and (b) shows the response time and recovery time
increasing concentrations as the adsorption sites are gradually of the sensor device as a function of operating temperatures. As
occupied the slope of the curve decreases. Following the Wol- demonstrated in the Fig. 6(a) and (b), in such temperature range
kenstein adsorption–desorption isotherm, both adsorption and (27–175  C), both response time/recovery time were faster than
desorption of vapor molecules concomitantly occur on the TiO2 the earlier reported ones.5,6 However, such, faster response/
sensing surface when the vapor pulse is ‘ON’.20 To the contrary, recovery characteristics with increase in operating tempera-
when the vapor pulse is ‘OFF’, only desorption process remain ture is attributed to fast adsorption/desorption and diffusion
active. In the present study, the phenomenon (lower slope of kinetics.12,13 Moreover, it is evident from Fig. 6(a) and (b) that
response magnitude aer 10 ppm) suggests that, upon reaching response time became faster at higher concentrations whereas
the maximum surface coverage, the adsorption rate is no longer recovery became slower. Following the Wolkenstein adsorp-
able to compete with the desorption rate. In principle, the tion–desorption isotherm, alcohol surface coverage (q) (occu-
decrease of response rate at high gas concentration could be pancy of the adsorption sites on TiO2 nanorods) at higher
mainly due to the two reasons. On one hand, the adsorption partial pressure (higher ppm level) is faster compared to its low
sites become less and less as well as the sticking probability of pressure counterpart (low ppm level).20 Fast response time of 8
the incoming gas molecule decreases due to the decreased s, 6 s and 5 s were obtained in 100 ppm concentrations of
adsorption energy at higher adsorption density/coverage. On ethanol, methanol and 2-propanol, respectively. On the other
the other hand, the adsorbed gas molecules become important hand, fast recovery time of 5 s, 6 s and 4 s was found in 1 ppm
scattering centres to the carrier in the sensing materials that concentrations of ethanol, methanol and 2-propanol, respec-
lead to the degradation of carrier mobility.21,22 tively. Such fast recovery at lower concentrations possibly
Fig. 5 shows the transient response characteristics of the attributed to the lower scattering phenomena. In general,
sensor with exposure to ethanol, methanol and 2-propanol at adsorbed molecules act as scattering centres for the carriers in

Table 1 Sensor response magnitude (RM%) with exposure to 1 ppm and 100 ppm alcohols (ethanol methanol and 2-propanol) in dry air and
different relative humidity (% RH) background at 75  C

Concentration VOCs RM (%) in dry air RM (%) at 8% RH RM (%) at 52% RH RM (%) at 75% RH

1 ppm Ethanol 29 28.5 26.38 26.2


Methanol 37 36.42 36.16 35.98
Propanol 40 39.37 39.09 38.9
100 ppm Ethanol 69 67.39 67.39 66.9
Methanol 75 73.25 72.75 72.21
Propanol 80 78.16 78.4 77.52

This journal is © The Royal Society of Chemistry 2015 RSC Adv., 2015, 5, 82159–82168 | 82163
RSC Advances Paper

by low temperature hydrolysis method was investigated for H2,


C2H5OH and LPG sensing.10 For all the cases (H2, C2H5OH and
LPG), an optimum temperature of 500  C was found with cor-
responding response magnitude of 64.28%, 94.11% and
91.66%, respectively. Fast response time of 3–5 s and sluggish
recovery time of 49–79 s were achieved at 500  C for the H2,
C2H5OH and LPG concentrations of 400–4000 ppm. Similar
high operating temperature (500  C) was also evident in the
work of Bian et al.9 Electrospun TiO2/PVP hybrids nanorods
based sensor showed high response of 95% with fast
response/recovery of 11 s/4 s towards acetone.9 Authors claimed
that the surface catalytic effect of nanostructured TiO2 nanorod
possibly be the reason of such high response magnitude and
fast response time towards acetone. However, high operating
temperature was the main disadvantage of their work.9 The
trace amount of NO2 detection (30 ppb) employing TiO2 nano-
rods synthesized by matrix assisted pulsed laser evaporation
(MAPLE) technique was reported by Caricato et al.8 Their sensor
showed as high as 96.15% response magnitude in exposure of 1
ppm NO2 at an optimum temperature of 300  C. But the slug-
gish response/recovery prole (120 s/780 s) and relatively high
Fig. 6 Sensor (a) response time as a function of temperatures (b) operating temperature are the major constraints the work.8 At
recovery time as a function of temperatures; for the alcohol same temperature (300  C), ethanol sensing performance of Pt
concentrations of 1 ppm and 100 ppm. doped hydrothermally grown TiO2 nanorod based resistive
devices was demonstrated by Epifani and his co-workers.7 The
resistive conguration of their TiO2 nanorod array showed the
the sensing channel materials that lead to the degradation of maximum response of 96% with fast response time of 9–12 s
carrier mobility and hence sluggish recovery characteristics at and recovery time of 10–15 s but at the cost of high operating
higher concentrations level.21 temperature (300  C). Relatively lower operating temperature
A comparative study of the present work with the sensor ethanol sensing performance was reported in the work of Chen
parameters (operating temperature, response magnitude, and et al.6 and Park et al.5 Crystalline a-MoO3/TiO2 core/shell
response time/recovery time) in the earlier reported TiO2 based nanorods synthesized by hydrothermal method showed
gas/vapor sensors are summarized in Table 2. As evident from maximum response of 93.37% at 180  C.6 Authors claim that
Table 1, hydrothermal is the most used technique for TiO2 such high response was due to the effect of formation of het-
nanorod array synthesis. Apart from hydrothermal, electrospun, erojunctions at the interfaces between MoO3 nanoparticles and
hydrolysis and MAPLE technique were also used for nanorod TiO2 nanorods.6 They found that the optimum temperature for
synthesis. However, it is envisaged that, gas/vapor sensing bare a-MoO3 sample was 270  C, whereas bare TiO2 nanorod did
properties of TiO2 nanorod array synthesized by other than not show any response below 180  C. Further, they found that,
hydrothermal technique usually offered high temperature compared to bare MoO3, the inclusion of TiO2 core/shell
sensing.8–10 For example, Nb doped TiO2 nanorod synthesized nanorod with smaller thickness exhibited enhanced response

Table 2 A comparative study of the sensing parameters with earlier reported TiO2 nanorods based gas/vapor sensors with the present work

Material Method Gases/vapors Opt tempb ( C) Detection range (ppm) RMc (%) Res timed (s) Rec timee (s) Ref.

TiO2aNR Hydrothermal Ethanol 75 1–100 69 8 5 This work


Methanol 75 1–100 75 6 6
2-propanol 75 1–100 80 5 4
TiO2/ZnO NR Hydrothermal C2H5OH 150 5–25 99.58 70 30 5
MoO3/TiO2 NR Hydrothermal C2H5OH 180 10–500 93.37 40 40 6
Pt/TiO2 NR Hydrothermal C2H5OH 300 10–1000 96 9–12 10–15 7
TiO2 NR MAPLEf NO2 300 30 ppb to 50 ppm 96.15 120 780 8
TiO2 NR Electrospun CH3COCH3 500 1–1000 95 11 4 9
Nb/TiO2 NR Hydrolysis H2 500 4000 64.28 4 70 10
C2H5OH 500 400 94.11 3 79
LPG 500 4000 91.66 3 45
a
NR ¼ nanorod. b Opt temp ¼ optimum temperature. c
RM ¼ response magnitude. d
Res time ¼ response time. e
Rec time ¼ recovery time.
f
MAPLE ¼ matrix assistance pulse laser evaporation.

82164 | RSC Adv., 2015, 5, 82159–82168 This journal is © The Royal Society of Chemistry 2015
Paper RSC Advances

at relatively lower temperature. Moderately high response time/ chemical sensor is still a burning issue.5,13 The proper annealing
recovery time prole was the only disadvantage of their sensor of the sample and judicious control of stoichiometry were
devices. A ve times increase in ethanol response was achieved reported to be the effective measures to improve the
by TiO2 nanorods functionalized with ZnO nanoparticles.7 Such stability.13,19 In the present study, the stability study was carried
high response was possibly due to ZnO functionalization which out over a span of four weeks, in air, ethanol, methanol and 2-
offer hetero-interfaces (TiO2 nanorod/ZnO nanoparticle) to the propanol background and the results are shown in Fig. 8.
sensor system. However, their sensor offered sluggish response/ Clearly, TiO2 nanorods offered fairly promising stability over
recovery of 70 s/30 s. As revealed from the literature survey, such a long duration. The resistance dri of 1%, 1.75%,
earlier researchers achieved >90% response magnitude at the 1.93%, and 1.25% were observed in air, 10 ppm of ethanol,
cost of either higher operating temperature or sluggish methanol and 2-propanol, respectively. Such a small dri in
response/recovery prole. As far as the present work is con- baseline resistance in turn ensures the improved repeatability
cerned, the developed titania tetragonal nanorods based sensor of the device also.
offered the most promising result at only 75  C which is the
lowest one ever reported for TiO2 nanorod based gas/vapor
3.3. Alcohol sensing mechanism
sensors. The minimum alcohol detection was as low as 1 ppm
with response magnitude of 29%, 37% and 40% towards Alcohol sensing mechanism of TiO2 tetragonal nanorod array
ethanol, methanol and 2-propanol. Fast response/recovery has been illustrated with the help a Wolkenstein adsorption–
characteristics with appreciably high response magnitude at desorption isotherm by considering (i) electron exchange
relatively low temperature (75  C) makes the present an between adsorbed species and TiO2 nanorod surfaces (Fig. 9(a))
improved alternative compared to the existing ones. and (ii) band bending in the corresponding energy band
The cross sensitivity of the developed alcohol sensor device diagram (Fig. 9(b)). A single tetragonal TiO2 nanorod with
was investigated with the other interfering species like acetone, surface depletion by chemisorbed oxygen ion was considered
2-butanone, benzene and xylene and the results are shown in and shown as Fig. 9(a). Initially, oxygen from ambient at
Fig. 7. The response magnitudes of the sensor towards these elevated temperature gets adsorbed on the TiO2 surfaces in the
interfering species were found to be 15.1/32.52%, 12.4/30.5%, form of O2, O and O2.12,13 The chemisorbed species (O) act
10.2/21.2% and 8.7/20.12% for 1/100 ppm of respective species as the electron trap centres on the semiconductor surfaces and
(acetone, butanone, benzene and xylene, respectively). Acetone cause electron depleted regions on the TiO2 surfaces as depicted
was revealed to the nearest interfering neighbour followed by in Fig. 9(a) and (b). Such accumulation of trapped electrons on
butanone, benzene and xylene. A response magnitude differ- the sensing surface forces the band to bend upward (as shown
ence of 14%, 21.88% and 24.74% in 1 ppm and 36%, in Fig. 9(b)). The details of the oxygen chemisorptions and
43.27% and 46.77% in 100 ppm were observed for ethanol– corresponding band diagram have already been discussed in
acetone, methanol–acetone and 2-propanol–acetone, ref. 12 and 13. Due to the sufficient separations among the
respectively. nanorods, alcohol molecules can diffuse along the nanorod
Long term stability (dri in the base line resistance as well as length and as a result chemisorptions can extend almost up to
resistance variation in alcohol exposure) of the MOX based the bottom of the rods (vertically along Z axis in the Fig. 9(a)).
Due to unavailability of free electrons conductivity of the sensor
device reduces. With exposure to alcohols, the replacement of
oxygen species by alcohols molecules decrease the

Fig. 7 Cross sensitivity study of the TiO2 nanorod array based alcohol
sensor towards different interfering species like acetone, butanone,
benzene and xylene. The response magnitude difference between
alcohols (ethanol, methanol and 2-propanol) with respect to acetone Fig. 8 Long term stability study of the TiO2 nanorod array based
(nearest interfering neighbour) is shown pictorially for both 1 ppm and alcohol sensor device at 75  C in air and in 10 ppm of ethanol,
100 ppm concentrations. methanol and 2-propanol, respectively.

This journal is © The Royal Society of Chemistry 2015 RSC Adv., 2015, 5, 82159–82168 | 82165
RSC Advances Paper

2CH3CH3CHO(ads) + 12O(ads) /
5CO + CO2 + 7H2O + 12e for 2-propanol (11)

A generalized formula for response magnitude of the TiO2


nanorod array can be theoretically represented and correlated
with the present study following the above discussions. A single
TiO2 nanorod is considered here for the sake of simplicity.
Assuming that, initial chemisorptions of oxygen molecules
produce a depletion region of width W and carrier concentra-
tions in nanorod in air and in alcohol are na and ng, (respec-
tively) the rate equation (dn/dt) for the eqn (8) following the
Wolkenstein adsorption–desorption isotherm, can be repre-
sented as;22

Fig. 9 Schematic illustration of the ethanol molecule adsorption and ng ¼ kalcohol(T)[O]m[A0 –O]mt + na (12)
desorption on a single nanorod surfaces (along with horizontal and
vertical diffusion) assuming depletion region (W) formed in air (b) where kalcohol(T) ¼ reaction rate constant at temperature T, T ¼
corresponding energy band diagram of the nanorod in air and in
absolute temperature, t ¼ reaction time, m ¼ 1 for O ion and
alcohol.
1/2 for O2. The eqn (12) reveals that concentrations of electron
are dynamic and strongly depend on rate constant (k) and
reaction time (t). If we consider that the vapor pulse is ‘ON’ at
concentrations of O ions and releases the electrons back to the time t1 and reach steady state under exposure of alcohols at
TiO2 surface (conduction band). Subsequently, the band time t2, then, a time constant s ¼ (t2  t1) must be considered to
bending is reduced (Fig. 9(b)) and hence higher conductivity of calculate approximate electron concentrations (ng) under steady
the sensor devices is evidenced. The dissociative oxidation of state condition. By putting time constant (s) in eqn (12) we get
alcohols (ethanol, methanol and 2-propanol) on the nanorod
ng ¼ kalcohol(T)[O]m[A0 –O]ms + na (13)
surfaces in presence of O ions can be explained as follows; the
adsorbed alcohols dissociate to form A–O(ads) radicals and
As, resistance (R) is inversely proportional to carrier
adsorbed H+ ions (eqn (2)), where A ¼ CH3CH2O for ethanol,
concentrations (n), the response magnitude (RM) of the present
CH3O for methanol and CH3CH3CH2O for 2-propanol, respec-
nanorod array based sensor in terms of carrier concentrations
tively. Further, A–O(ads) dissociates to form A0 –O (eqn (3)), where
(n) and rate coefficient (kalcohol) can be written as;
A0 ¼ CH3CHO (for ethanol), CH2O (methanol) and CH3CH3CHO
(2-propanol). Finally, A0 –O reacts with O and produces CO, Ra  Rg skalcohol ðTÞ½Om ½Am
RM ¼ ¼ (14)
CO2, H2 and n number of electrons, where n ¼ 4, 4 and 12 for Ra ng
ethanol, methanol and 2-propanol, respectively.6,10,19 The details
of the ethanol, methanol and 2-propanol oxidation on the It is assumed that, negative surface charge on the semi-
nanorod surfaces are described in eqn (2)–(11). conductor surface is due to the chemisorbed oxygen ions only
and every oxygen ion generate one electron on the surface.
A–OH(gas) ¼ A–O(ads) + H(ads) (2)
Therefore, one can nd the number of available O ions by
A–O(ads) ¼ A0 –O + H(ads) (3) calculating number of trapped electrons on the sensing surface.
In chemisorption process, the number of trapped electron
H(ads) + H(ads) / 2H(ads)+ + 2e (4) concentration depends on the degree of surface coverage (q),
distribution of density of surface state (fa) and allowed surface
2H(ads)+ + 2O(ads) / 2OH(ads) (5) states i.e. number of adsorption sites (Nf) and they are related as
per eqn (15);20
2OH(ads) / H2O(ads) + O(ads) (6)
N ¼ O ¼ faqNf (15)
 
O(ads) + e / O(ads) (7)
1
fa ¼     (16)
0 
A –O + 4O(ads) / CO + CO2 + 2H2 + ne 
(8) Ec  Ess eVs
1 þ ga exp  exp 
KB T KB T
CH3CHO(ads) + 4O(ads) / where fa ¼ Fermi Dirac distribution function, ga ¼ degeneracy
CO + CO2 + 2H2O + 4e for ethanol (9) factor, EC ¼ conduction band energy, ESS ¼ surface state energy,
KB ¼ Boltzmann constant, T ¼ absolute temperature, q ¼
2CH3O(ads) + 4O(ads) /
surface coverage ¼ [bP/(1 + bP)], Nf ¼ total surface sites on the
CO + CO2 + 3H2O + 4e for methanol (10)

82166 | RSC Adv., 2015, 5, 82159–82168 This journal is © The Royal Society of Chemistry 2015
Paper RSC Advances

sensing surface. b ¼ adsorption co-efficient, P ¼ gas/vapor range (27–75  C), mobility (m) decreases, carrier concentration
pressure, eVs ¼ barrier height. (n) increases but sticking coefficient (S) decreases but at a very
Therefore, eqn (14) can be written as; slow rate. The resistance Ra and Rg both reduces but the rate of
decrease of Rg is much higher (as sticking coefficient is higher at
Ra  Rg skalcohol ðTÞ½Am fa qNf
RM ¼ ¼ (17) low temperature) than that of Ra. So, response magnitude
Ra ng
[RM% ¼ ((Ra  Rg)/Ra)  100] in the temperature range 27–75  C
is expected to be increased which is in conrmation with the
From eqn (17), it can be seen that the response magnitude is experimental results [Fig. 4(a)]. On the other hand, at relatively
proportional to reaction rate coefficient (kalcohol(T)) and higher temperature range (75–175  C), both the rate of increase
inversely proportional to carrier concentrations. Response of carrier concentration (n) and the rate of decrease of mobility
magnitude is principally governed by the surface coverage (q) (m) is much higher than that in the lower temperature range
and the sticking co-efficient (S) of the target molecules. At a (27–75  C). Consequently, overall base line resistance (Ra) vari-
particular temperature, for homogeneous and at surface, ation is not so signicant in that temperature range (Fig. 3). In
sticking co-efficient (S) is a function of alcohol surface coverage addition, Rg value is found to decrease insignicantly in the
(q) as per eqn (18):23 temperature range (75–175  C) due to the decrease of sticking
co-efficient (S) at much higher rate (compared to low tempera-
S ¼ S0(1  q)r (18) ture zone). Possibly that is why the response magnitude
decreases gradually in that temperature range (75–175  C).
where S0 ¼ sticking coefficient at clean surface and r is a curve
tting constant with a value between 1 and 2.23 The sticking
coefficient decreases with increase in temperature.23 The 4. Conclusions
amount/number of target molecules stick to the sensing surface
is proportional to the sticking coefficient (S) which is also Low temperature (75  C), low ppm (1–100 ppm) alcohol detec-
proportional to the ux (F). Therefore, response magnitude in tion employing one dimensional tetragonal nanorod array of
terms of sticking coefficient can be written as (putting eqn (18) TiO2 has been demonstrated in the present paper. Hydrother-
into eqn (17)); mally grown nanorod array showed both anatase and rutile
"  r # crystallinity with grain size of 35–37 nm. The maximum
S response of 69%, 75% and 80% towards 100 ppm ethanol,
skalcohol ðTÞ½Am fa 1  Nf
Ra  Rg S0 methanol and 2-propanol was achieved at an operating
RM ¼ ¼ (19) temperature of only 75  C which reasonably lower than that of
Ra ng
the earlier reported works. Further, sensor offered the
minimum alcohol detection level of 1 ppm with appreciable
To understand the sensor response characteristics, one response magnitude (29%, 37% and 40% towards ethanol,
needs to thoroughly analyze the parameters like carrier methanol and 2-propanol) with fast response time and recovery
concentration (n), carrier mobility (m) and sticking coefficient time of 12–14 s and 4–5 s, respectively. The sensor also
(S) including the effect of temperature variations as well as offered very promising stability over a span of four weeks. The
variations in gas/vapor exposure. Initially, the effect of gas sensing mechanism of TiO2 nanorod array has also been illus-
exposure on these parameters (n, m and S) and subsequently the trated with the help of Wolkenstein adsorption–desorption
effect of temperature (T) will be considered in this discussion. isotherm and corresponding energy band diagram. Measure-
Conductivity (s) of the sensing layer decreases in air due to the ment of the specic surface area of TiO2 nanorod, using BET
decrease of effective carrier concentrations (n). Upon gas/vapor method, has been considered as a future work for better
exposure, conductivity increases as carrier concentrations (n) in quantitative understanding of the related sensing mechanism.
the semiconductor increases and mobility (m) of the carrier
remains almost unaffected (because of negligible carrier–carrier
scattering). In such situation (when temperature (T) is xed), Acknowledgements
sticking coefficient (S) is the function of surface coverage (q)
only, as per eqn (18). Now, with increase in temperature, the This work was supported by Indian National Science Academy
sticking coefficient decreases gradually but aer a certain (INSA) (No. SP/YSP/81/2013/735). B. Bhowmik thankfully
temperature (75  C in the present study), the rate of such acknowledges COE, TEQIP-II, IIEST, Shibpur for his PhD
decrease is more prominent.23 At higher temperatures, due to fellowship.
thermal vibration, gaseous molecules and TiO2 surface atoms
act as the scattering centres for the new incoming target References
molecules and hence the sticking coefficient is decreased.
Further, carrier concentration (n) is found to increase with 1 V. Galstyan, A. Vomiero, E. Comini, G. Faglia and
increase in temperature whereas carrier mobility (m) decreases G. Sberveglieri, RSC Adv., 2011, 1, 1038.
with increase in temperature (due to dominant lattice scat- 2 W. Zhang, S. Wang, Y. Wang, Z. Zhu, X. Gao, J. Yang and
tering). Therefore, it can be envisaged that, at low temperature H. xin Zhang, RSC Adv., 2015, 5, 2620–2629.
3 Y. Yang, Y. Li and M. Pritzker, RSC Adv., 2014, 4, 35833.

This journal is © The Royal Society of Chemistry 2015 RSC Adv., 2015, 5, 82159–82168 | 82167
RSC Advances Paper

4 J. Sun, P. Sun, D. Zhang, J. Xu, X. Liang, F. Liu and G. Lu, RSC 13 B. Bhowmik, A. Hazra, K. Dutta and P. Bhattacharyya, IEEE
Adv., 2014, 4, 43429. Trans. Device Mater. Reliab., 2014, 14(4), 961.
5 S. Park, S. An, H. Ko, S. Lee, H. W. Kim and C. Lee, Appl. Phys. 14 H. E. Swanson, H. F. McMurdie, M. C. Morris and
A, 2014, 115, 1223. E. H. Evans, Natl. Bur. Stand. Circ., 1969, 25(7), 82.
6 Y.-J. Chen, G. Xiao, T.-S. Wang, F. Zhang, Y. Ma, P. Gao, 15 R. J. Swope, J. R. Smith and A. C. Larson, Am. Mineral., 1995,
C.-L. Zhu, E. Zhang, Z. Xu and Q.-H. Li, Sens. Actuators, B, 80, 448.
2011, 155, 270. 16 C. Cao, C. Hu, X. Wang, S. Wang, Y. Tian and H. Zhang, Sens.
7 M. Epifani, T. Andreu, R. Zamani, J. Arbiol, E. Comini, Actuators, B, 2011, 156, 114.
P. Siciliano, G. Faglia and J. R. Morante, CrystEngComm, 17 B. Liu and E. S. Aydil, J. Am. Chem. Soc., 2009, 131, 3985.
2012, 14, 3882. 18 S. Sreekantan, K. A. Saharudin, Z. Lockman and T. W. Tzu,
8 A. P. Caricato, M. Cesaria, G. Leggieri, A. Luches and Nanotechnology, 2010, 21, 365603.
M. Martino, Appl. Phys. A, 2011, 104, 963. 19 A. Hazra, K. Dutta, B. Bhowmik, P. P. Chattopadhyay and
9 H. Bian, S. Ma, A. Sun, X. Xu, G. Yang, J. Gao, Z. Zhang and P. Bhattacharyya, Appl. Phys. Lett., 2014, 105, 081604.
H. Zhu, Superlattices Microstruct., 2015, 81, 107. 20 A. Rothschild, Y. Komem and N. Ashkenasy, J. Appl. Phys.,
10 S. Singh, H. Kaur, V. N. Singh, K. Jain and T. D. Senguttuvan, 2002, 92(12), 7090.
Sens. Actuators, B, 2012, 171, 899. 21 H. H. Pu, S. H. Rhim, M. G. Josifovska, C. J. Hirschmugl,
11 M. Graf, U. Frey, S. Taschini and A. Hierlemann, Anal. Chem., M. Weinert and J. H. Chen, RSC Adv., 2014, 4, 47481.
2006, 78, 6801. 22 N. Hongsitha, E. Wongrat, T. Kerdcharoen and S. Choopun,
12 B. Bhowmik, K. Dutta, A. Hazra and P. Bhattacharyya, Solid- Sens. Actuators, B, 2010, 144, 67.
State Electron., 2014, 99, 84. 23 M. Johansson, O. Lytken and I. Chorkendorff, Top. Catal.,
2007, 46(1–2), 175–187.

82168 | RSC Adv., 2015, 5, 82159–82168 This journal is © The Royal Society of Chemistry 2015

Вам также может понравиться