Professional Documents
Culture Documents
Abstract
Perovskite-type LnMO3 (Ln = La, Pr, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm, Yb, Lu; M = Cr, Fe) oxides were prepared by the polyol mediated
synthesis. Thick films of the prepared materials were electrically characterised using high throughput impedance spectroscopy (HT-IS). All samples
showed p-type semiconductivity. Ethanol and propylene were used to test the gas sensing behaviour in a temperature range from 200 to 500 ◦ C. A
correlation between the (Ln–O) binding energy and the sensing properties is observed.
© 2006 Elsevier B.V. All rights reserved.
Keywords: LnFeO3 ; LnCrO3 ; Polyol method; High throughput impedance spectroscopy; Gas sensor
0925-4005/$ – see front matter © 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.snb.2006.11.022
182 M. Siemons, U. Simon / Sensors and Actuators B 126 (2007) 181–186
Table 1
Summary of materials prepared (DS stands for Debye–Scherrer)
Ln-precursor Compound DS (nm) SEM (nm)a Compound DS (nm) SEM (nm)a
Fig. 1. Left: Scheme of the HT-IS setup. For reasons of clearness only two measuring lines are displayed. Right: Layout of the multielectrode array with photograph
of an already deposited sample plate (adapted from Ref. [23]).
M. Siemons, U. Simon / Sensors and Actuators B 126 (2007) 181–186 183
ter of 5 mm (as shown in Fig. 1 on the right hand side) [4,23]. The performed. One hundred Hertz have been chosen to stay in any
array allows automated pipetting robot assisted sample prepa- case in the quasi-dc plateau, i.e. the frequency independent part
ration and coating, respectively. Time consuming process steps at low frequencies of the impedance spectra. While each mea-
like thermal annealing or sample conditioning can be proceeded surement took 1.5 s, the time interval between the measurements
in parallel. For the thick film deposition on these electrodes, was 4 s.
the powder materials were dispersed by mixing in a mortar with
an aqueous polyethylene imine (0.5 wt%) solution. The resulting 3. Results and discussion
suspensions had a concentration of 0.08 mol/L. The suspensions
were deposited into wells formed by a Teflon mask on top of 3.1. Characterisation of the gas sensing properties of
the microelectrode substrates by use of an Eppendorf pipette or LnMO3
a laboratory roboting system (Lissy, Zinsser Analytic GmbH).
Each well had an O-ring for sealing against the substrate. The Fig. 2 shows the plot of real part of the complex impedance
substrate was first dried for 48 h at room temperature in air. After versus the imaginary part (Argand plot) representative for
that it was held for 6 h at 700 ◦ C in air to remove the organic TbFeO3 in synthetic air, propylene and ethanol at 300 ◦ C. Auto-
residues. To compensate possible gradients in gas concentration mated data fitting allows describing the impedance spectra with
and temperature as well as single failures due to contact defects the impedance function of a circuit equivalent, consisting of a
over the substrate two to three statistically selected positions resistor and a capacitor in parallel (see Fig. 2). Further informa-
were equipped with the same material composition. Sample tion is given in Ref. [25]. The use of fit functions reduces the
plates (coated arrays) were inserted into the measuring head. complex impedance function describing the electrical properties
Fig. 1 (left) shows the scheme of the HT-IS setup. The leads to of each material to one value for the resistance R and the capac-
each IDC are electrically connected via Al2 O3 covered platinum itance C, respectively. While the capacitance can be assigned to
wires in the measuring head. Constant contact pressure is assured the geometric capacitance of the IDC, the resistance values under
using spring loaded contact tips which connect the IDCs via high the respective measuring conditions are taken for the determi-
frequency capable relay matrices (multiplexers) to the measur- nation of the material sensitivity. The absolute sensitivity S of a
ing instruments. A set of gas flow controllers is used to compose material to a test gas is defined by:
the different test gases. In addition, the test gases can be humidi- ⎧
⎪ Rgas
fied by bubbling the carrier gas through a water reservoir at room ⎪
⎪
⎨ Rair R >R
temperature. The test gases are led to a quartz glass bell covering S= gas air
all sample positions on a plate. For further technical details see ⎪
⎪ R
⎩ R
air
⎪
Ref. [5]. gas Rair >Rgas
2.3. Electrical characterisation where Rgas is the resistance obtained from the fit under test gas,
while Rair is the reference resistance value for a measurement
Electrical measurements were performed using an impedance under synthetic air.
analyser (Agilent 4192 A) in a frequency range from 10 to
106 Hz (15 measuring points per frequency decade) and an
amplitude of 100 mV (rms). The measurements were taken out
in a temperature range between 500 and 200 ◦ C in 25◦ -steps.
Reference measurements were performed in humid synthetic
air (air), which was also used as carrier gas for ethanol and
propylene. For the gas sensing experiments, the test gases were
mixed with synthetic air to reach a consistent volume flow of
each 100 sccm. Conditioning of the materials was carried out
at 500 ◦ C for 240 min under air and at the following temper-
atures for 90 min. The measuring sequence was air, ethanol
(40 ppm), air, propylene (25 ppm), air. The intermediate mea-
surements in air insure the recovery of the resistance, before
the next test gas is applied. Relative humidity of the gases was
45% at room temperature in order to perform the measure-
ments in a typical humidity range representative for ambient
conditions. For sample conditioning a preliminary gas flow was
applied over 30 min to reach equilibrium. All data achieved are
stored in a database, which has been developed for data mining
and for comparison with complementary measuring methods
[24]. Fig. 2. Argand plot and circuit equivalent exemplarily for TbFeO3 (T = 300 ◦ C).
In order to measure the response time of the individual p-Type semiconductors show an increase of resistance compared to synthetic
samples, measurements at fixed frequency (100 Hz) were air, when reducing gases are applied.
184 M. Siemons, U. Simon / Sensors and Actuators B 126 (2007) 181–186
We describe the response of a material to a test gas by the est sensitivity towards the used hydrocarbons. Measurements for
relative sensitivity S , which is defined as YbFeO3 could only be evaluated above 300 ◦ C. At lower tem-
⎧ perature the resistance of the material exceeds the measurement
⎪ Rgas − Rair
⎪
⎪−
limits (R > 20 M). The highest overall sensitivities for all tem-
⎨ Rgas R >R peratures below 450 ◦ C can be found for SmFeO3 , TmFeO3 and
S = gas air
⎪
⎪ R − R LuFeO3 .
⎪
⎩+
air gas
R Fig. 4 shows the temperature dependence of the sensitivity
air Rair >Rgas
for the LnCrO3 materials. In comparison to LnFeO3 , LnCrO3 in
where values for S range from 1 to −1. Decrease of resistance general shows smaller sensitivities towards hydrocarbons. The
while applying a test gas gives values between 0 and 1, rise of temperature of maximum sensitivity is shifted to lower values.
resistance results correspondingly in values between −1 and 0. LaCrO3 shows the smallest sensitivity towards the applied gases,
Using S instead of the commonly used absolute sensitivity S whereas DyCrO3 appears to be the material with the highest
including algebraic signs turned out to be beneficial for the data sensitivities towards both test gases.
mining process in our high throughput workflow [24]. Further- In both material groups the sensitivity towards ethanol is
more, it allows an immediate differentiation between oxidising much more pronounced than towards propylene. This is presum-
and reducing test gases, respectively. ably caused by the functional OH-group of ethanol, as expected
Fig. 3 shows the temperature behaviour of the sensitivity of from Ref. [26].
the different LnFeO3 materials indicating that the operating tem- For further evaluation we took into account the strength of the
perature affects the properties of all materials. LaFeO3 , which is M–O bond in the metal oxide. The bond strength of the oxygen
the material with the largest lanthanoid radius, shows the small- at the surface near sites is an important factor for the sensing
Fig. 5. Sensitivity S vs. binding energy Ln–O according to Ref. [27] at 300 ◦ C
(without LuFeO3 ). Solid lines are no fit data; they should only serve as a guide
for the eye.
Table 2
Response times (τ 50 /s) of LnFeO3 towards 60 ppm of ethanol at 350 ◦ C
Ln
Fig. 6. Ethanol sensitivity of LnFeO3 (300 ◦ C = TSMax ) and LnCrO3 La Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu
(225 ◦ C = TSMax ) vs. binding energy Ln–O according to Ref. [27]. Solid lines 17 <15 <15 24 26 20 25 19 <15 <15 <15 <15 <15
are no fit data; they should only serve as a guide for the eye.
186 M. Siemons, U. Simon / Sensors and Actuators B 126 (2007) 181–186