For nearly a century, solid-state potentiometric sensors that
employ oxide electrodes have been extensively studied and applied for sensing toxic gases1. Transition metal oxide has been
widely used as a sensing electrode for various reasons, including
its flexibility of production, simplicity of use, its ability to catalyze
a large number of gases, and its thermal and chemical stability2-5.
Noble metal oxides are widely used as additives or sensitizers to
improve the sensing performance of functional materials6. Few
reports consider the unique properties of palladium oxide (PdO)
and its performance in gas sensing as a functional element7, 8.
In our previous work8, PdO′s performance as a sensing electrode was explored based on a YSZ (8% (x, molar fraction) Yttriastabilized Zirconia) electrolyte, which was measured via a potentiometry method to detect carbon monoxide (CO). Attractive
phenomena were observed. Changes in the electric potential of the
sensor in a CO environment are independent of the mobile oxygen
in the electrolyte and the thickness of the PdO electrode. A hypothesis is proposed that a potentiometric sensor based on a PdO
electrode works in capacitive mode during a sensing process.
The working mechanism of a potentiometric sensor with a PdObased electrode is our focus in this paper. In order to verify the
hypothesis, several aspects of the working principle were studied
by experimental methods. First, PdO was doped by Ni, La, and
Mg. The doping is capable of shrinking the particle size of PdO
and creating defects in PdO. The influence on the sensing performance of the sensor was discussed. Second, zirconia was replaced by other materials to test the impact of the electrolyte on
the sensing performance. Third, the source of the sensing signal
was located by measuring the impedance and resistance. The
working model of the sensors with PdO-based electrodes is discussed.
Pd (C2H3O2)2, Ni (NO3)2∙6H2O, La (NO3)3∙6H2O and Mg (NO3)2∙
6H2O were obtained from Sinopharm Chemical Reagent Co. Ltd., China. Y-type zeolite (Na-Y, Si/Al molar ratio=2.5) was pruchased from Alfa Aesar, China. The above chemical reagents were
of analytical grade and used without further purification.
Three types of dopants are fabricated based on PdO. Elements
used for doping are transition metal (Ni2+), alkaline earth metal
(Mg2+), and rare earth metal (La3+). The molar ratios of the palladium and doped elements are 100 : 5, 100 : 10, and 100 : 15, which are denoted by M-PdO-5, M-PdO-10, and M-PdO-15, respectively (where M represents the doped elements, i.e., Ni, Mg, La). The preparation is described in Ref.8: a certain quality of
Pd (C2H3O2)2 is dissolved in alcohol by energetic stirring. A corresponding quality of doped element in nitrate is weighed and
added to the palladium acetate solution. Hydrolysis is achieved by
adding deionized water at 90 ℃ for 2 h. The obtained solution is
centrifuged to precipitate the resulting material, and then dried in
a box oven at 130 ℃ for 12 h. The resulting sample is calcined at 700 ℃ for 2 h in air.
X-ray diffraction (XRD, Bruker D8 Focus, Germany) patterns
are obtained with Cu Kα radiation (0.1541 nm). Scanning electron
microscopy (SEM, Hitachi S-4800, Japan) is performed to obtain
images of the samples. The stoichiometric ratio, as well as the
quality percentage of the doping element in the samples, is analyzed with an energy dispersive spectrometer (EDS, QUANTAX
400, Germany). The impedance spectroscopy is measured with an
Agilent 4284A (USA) at 100 mV in the range of 25 Hz-1 MHz.
The electrical resistance of the sensors was measured by a computer-controlled Agilent multimeter (Agilent 34405A, USA).
YSZ and alumina chips (8% (molar fraction) Y2O3 doped zirconia), fabricated by the Shanghai Institute of Ceramics (Chinese
Academy of Sciences, China), are used as the electrolyte in the
sensor. The YSZ and alumina discs measure 7 mm × 7 mm × 0.5
mm and 7 mm × 7 mm × 0.8 mm, respectively. For the reference
electrode, commercial Pt paste (TR27905, Japan Tanaka Precious
Metal Industries Ltd.) is applied to one side of the YSZ by screen
printing, and then dried at 130 ℃ for 20 min. The Pt reference
electrode, which is considered an ideal reversible electrode9, is
sintered at a temperature of 1200 ℃ for 2 h. The prepared PdObased samples and α-terpineol are mixed to obtain the sensing
electrode paste, which is screen printed on the other side of the
YSZ and sintered at 700 ℃ for 2 h. Finally, both electrodes are
attached with Pt lead wires. Sensing electrodes with areas of
3 mm × 3 mm (9 mm2), 5 mm × 5 mm (25 mm2), and 7 mm × 7
mm (49 mm2) are created by screen printing. If there are no special
instructions, the size of the sensing electrode is 5 mm × 5 mm. The
structure and image of the sensors are shown in Fig. 1.
Y-type zeolite is a microporous, crystalline aluminosilicate that
has mobile sodium cations inside the framework10. The zeolite
pellets are prepared by pressing 200 mg of zeolite powder under
5 t of uniaxial force, and then sintered at 700 ℃ for 2 h. The
processes involved in the preparation of the sensor with zeolite are
the same as those previously described, with the exception of the
sintering temperature for the Pt reference electrode, which is
700 ℃. The framework of zeolite will collapse at a temperature
higher than 800 ℃10.
The sensor is placed in a hermetic quartz tube and heated under
different temperatures, which are controlled by a temperature
controller and heater in a tubular furnace. The flow rate of different gases is controlled independently by using computercontrolled electronic mass flow controllers (D07-19B, Seven Star
Electronics, China). During the test, the total flow rate is maintained at 100 mL∙min-1, and the sensing and reference electrodes
are exposed to the same atmosphere. The electric potential (V)
between the sensing and reference electrodes is recorded by using
a multifunction data acquisition card (USB-6221, National Instruments, USA). The response of the potentiometric sensor is
measured as the difference in potential (∆V) between the sensors in the background gas (N2 and 5% (φ, volume fraction) O2 are
usually used) and in the targeted gas. The positive terminal is
always connected to the PdO-based electrode, and the negative
terminal is connected to the Pt electrode.
Since PdO will decompose to metallic Pd at approximately
800 ℃11, PdO-based dopants are prepared at 700 ℃. XRD patterns of pure and doped PdO are shown in Fig. 2 and Fig.S1 (see
Supporting Information). All diffraction peaks coincide with the
corresponding peaks of PdO given in the standard data file
(JCPDS File No. 43-1024). No peaks corresponding to the oxidic
dopant are observed, even for the 15% (x, molar fraction) doped
samples. This may occur because of the small mass percentage of
dopant in the samples. The peaks become broader and weaker
when Ni is doped. Peak positions and full width at half maximum
(FWHM) are used to determine the cell parameters and crystallite
sizes for all palladium oxide12. The lattice parameters for all
samples are calculated from the XRD peak positions in Fig. 2 and
are listed in Table 1. The cell volume has an initial increase during
the incorporation of Ni, and then decreases with the increasing
Typical SEM images of pure and Ni-doped PdO are shown in
Fig. 3(a-d). The grain becomes smaller as the level of doping
increases. The average particle sizes are estimated to be~90, ~72, ~59, and~50 nm for PdO and the dopants Ni-PdO-5, Ni-PdO-10, and Ni-PdO-15, respectively. Agglomerates are clearly observable, and they cause a random spatial distribution. The images of
the surfaces of the YSZ disc and the Na-Y pellet are shown in
Fig. 3(e, f).
To investigate the relationship between the defects of PdO and
the sensing performance of the sensors, an EDS analysis of pure
and Ni-doped PdO is carried out to study the impact of the doping.
The results in Table 2 indicate that the samples are non-stoichiometric, and the atomic ratio of Pd/Ni increases with the
doping level. Moreover, the largest oxygen vacancy is observed
on the surface of Ni-PdO-10.
The role of a dopant is to disrupt the chemical bonds on the
surface of the host oxide to ready the doped oxide for chemical
interaction13, 14. In the case of Ni-doped PdO, according to XRD
patterns and SEM images, the crystals become smaller as the
doping content increases. Comparing the relative ionic radius of
Pd2+ (0.064 nm) and Ni2+ (0.069 nm), substituting Ni into the Pd
positions would increase the unit cell volume. The experimental
trend shows an initial increase in unit cell volume, with a decrease
that follows the doping content, as shown in Table 1. It is also
noticeable that the unit cell volumes of all Ni-doped samples are
larger than the undoped volumes. Combining the atomic ratios on
the surface obtained by EDS in Table 2, the ratio of Ni increases
with the doping level. When the host oxide suffers heavy doping, the dopant tends to be segregated and forms a layer of oxide
clusters that heal the oxygen vacancies on the surface13. By contrast, a lower concentration of dopant will solid-dissolve into the
crystal of PdO and expand its unit cell volume14-16. At the same
time, a solid solution will create more defects by crystal distortion17, 18.
Various sensing electrodes made by PdO-based dopants on YSZ
are tested in CO environments. The sensing responses for 100 × 10-6 (volume fraction) CO are given in Fig. 4. A general improvement is observed for the sensors with Ni-and La-doped PdO
electrodes in the range of 400-550 ℃. The best working temperature is in the range of 450 to 500 ℃. At temperatures above
500 ℃, the sensors′ responses decay. The doped electrodes, in
which Pd and foreign elements are present in a molar ratio of 100 :
5, have the best performance. In particular, the best CO sensor
uses Ni-doped PdO at 450 ℃. Hereafter, the sensor made of a NiPdO-5 sensing electrode is used to conduct the study in this paper.
For the sensing test in Fig. 4, the Ni-and La-doped PdO behaves
better than the undoped version. Despite an increasing response
observed for the sensors with Ni-and La-doped PdO electrodes, different active sites are brought into the samples by La, Ni, and
Mg. According to the classification of dopant-oxide pairs13, La is
a high-valence dopant (HVD), while Mg and Ni are same-valance
dopants (SVD). HVD intends to adsorb O2 from a gas phase at the
vacant sites, and activates it. Thus, a reducing gas can react with
this oxygen and undergo oxidation. In the case of SVD, since
MgO is irreducible and the valence of Ni is flexible, it could be
predicted that the flexible-valance dopant is favorable for promoting the mobility of oxygen vacancies. The sensing signal of
conventional potentiometric sensors can be effectively improved
by decreasing the surface area of the grains of the sensing electrode and increasing the contact area of triple phase boundary
(TPB)19. However, the response was enhanced by creating oxygen
vancancies in the PdO electrode in our study. The oxygen vacancies are the sites for the catalytic oxidation between CO and
adsorbed ionic oxygen6.
An O2-dependent experiment is conducted on the sensor at
450 ℃, as shown in Fig. 5. As can be seen, the increasing percentage of O2 in the environment leads to a rise in the baseline of
the sensor. The average increase is 12 mV when the O2 concentration doubles. In the meantime, the response of the sensor toward
100 × 10-6 (volume fraction) of CO decreases by approximately
4 mV because of the increase in the O2 concentration. O2 does
interfere significantly with the sensing performance.
Sensing electrodes with different areas are tested, and the results for CO detection at 450 ℃ are given in Fig. 6. When the
responses are plotted versus CO concentrations on a logarithmic
scale, the relationship is linear. The responses for all sensors increase as the CO concentration increases. Moreover, the sensitivities of the sensors, which are the slopes of the plots in Fig. 6, are -30.36, -38.21, and -38.77 mV∙decade-1 for sensors with
sensing electrodes having areas of 9, 25, and 49 mm2, respectively.The increasing sensing area leads to a higher response.
Moreover, a potentiometric sensor based on coupling a solid
electrolyte with semiconducting oxides is explained by the mixed
potential theory20, which assumes a thermodynamic potential
independent of the size of the electrode21. However, the potential
is typically proportional to the area of the sensing electrode, as
shown in Fig. 6. These results lead us to think that the sensing
reactions in the sensor with a PdO-based electrode are of the
In order to address the role of electrolyte in gas sensing, a NiPdO-5 electrode was printed on an alumina disc and zeolite pellet, and tested in a CO environment at 450 ℃. The potentiometric
results of the sensor based on the alumina disc are shown in Fig.
S2 (see Supporting Information). No potentiometric response is
observed for the sensor printed on the alumina disc. Potentiometric responses for the sensor based on zeolite are obtained for
100 × 10-6, 200 × 10-6, and 400 × 10-6 (volume fraction) of CO, as shown in Fig. 7(a). Next, an O2 dependence experiment is
conducted. The sensor is tested in 100 × 10-6 (volume fraction) CO
under 5%, 10%, and 15% O2, as shown in Fig. 7(b). A depressive response to CO and a rising baseline with increasing O2 are noticed. Moreover, Ni-PdO-5 electrodes on YSZ and Na-Y have
sensing profiles similar to those from the oxygen-dependent
measurement. Ionic material is essential to generate a potentiometric response to CO. In other words, the changing of the
chemical potential gradient in the electrolyte is the source of the
A solid-state potentiometric sensor for gas sensing is commonly
explained by the mixed potential theory20. When the sensor is
exposed to a targeted species, more than one cathodic or anodic reaction between the targeted species and components of the
electrolyte occurs at the interface of the electrode/electrolyte, which is the so-called TPB. The resulting electrode potential
becomes a mixed potential22. The following pair of reactions (1)
and (2) occur at the interface of the sensing electrode/zirconia for
where Re (g) and Ox (g) represent a targeted gas in its reduced and
oxidized forms, respectively.
To investigate the ionic motion of the potentiometric sensor
during sensing process, impedance is carried out to measure the
ionic conductivity of the sensors under variational CO or O2 environment as shown in Fig. 8. Since palladium oxide is a good
electrical conductor, the measured impedance belongs to the
electrolyte. For the measurements on the sensor with YSZ in Fig. 8
(a, b), the shift of the impedance in different CO concentrations, within experimental error, is negligible.
If the oxygen ions from the electrolyte participate in the electrochemical reactions, the impedance change in the electrolyte can
be observed. However, the ionic conductivity decreases when
oxygen is enriched in the environment. This occurs because more
O2- enters the electrolyte23. This result agrees well with the electric
potential increase of the potentiometric sensor, as shown in Fig. 5.
No shift of impedance is observed in the CO environment; thus, the number of oxygen ions in the electrolyte and their respective mobility are not disturbed during the electrochemical reactions.
For the sensor with zeolite, the impedance is constant under
different gaseous environments, as shown in Fig. 8(c, d). The results indicate that sodium ions in zeolite will not be influenced by
the gaseous species in the environment.
The mobile ions inside the electrolyte are essential for the
generation of a potentiometric response. Since ionic flows from
the chemical potential gradient are balanced by the electric field
gradient24, the changing resistance across the sensors reflects the
change in the chemical potential gradient under CO sensing
processes. The electrical resistivity across the sensors decreases
for both sensors when CO is introduced (Fig. 9). The resistance of
the sensors is a series of Pt reference electrode, electrolyte, PdObased electrode, and their interfacial barrier. To address the source
of the change in resistances of the sensors, a resistance of Ni-PdO-5 is measured for 100 × 10-6 (volume fraction) CO at 450 ℃ in Fig.
S3 (see Supporting Information). The result indicates that no net
electrons transfer between CO and PdO in the equilibrium state.
Accordingly, the interfacial barrier between the sensing electrode
and electrolyte decreases owing to CO adsorption. Therefore, the
interfacial potential increase between the sensing electrode and
electrolyte is the source of the potentiometric response.
Based on the previous results and discussion, capacitive behaviors were observed for the potentiometric sensor with a PdObased electrode. When PdO is brought into contact with a ionic
conductor, an electrochemical double layer appears25. The potential
drop across the electrochemical double layer can be interpreted as
where kB is the Boltzmann constant, T is the temperature, q is
elementary charge, and ρelectrolyte and ρPdO are the charge densities on
the surface of PdO and the electrolyte, respectively.
The capacitance of the electrochemical double-layer is described by
where A is the area of the electrochemical double layer, d is the
distance between them, L is the perimeter of the PdO electrode, δ
is the edge effect in the capacitor, ε is the permittivity.
The catalytic oxidation of CO over PdO is expressed in equations (5)-(7)26, 27. The reactions could be described as follows27, 28:
oxygen and carbon monoxide from the gas phase co-adsorbed at
the vacant sites on the surface of the PdO, and interacted to form
intermediates (e.g., "carbonate"), which will decompose to form
CO2 and heal oxygen vacancies.
During the catalytic oxidation, the ionic species are redistributed on the surface of PdO, the electrons shift between PdO and
CO and O2, and mobile positive ions from the volume of electrolyte move toward the PdO-based electrode. If the mobile ions are negative, they will move from the surface into the volume of
the electrolyte. During the sensing process, the charge density at
the interface and the distribution of potential inside the electrolyte
are altered. The degree of the alteration is related to the concentration of CO. Therefore, the potential change at the interface is
linearly dependent on the logarithmic concentration of CO according to the equation (3). Moreover, a higher sensitivity of the
sensor with larger area of PdO electrode is observed in the Fig. 6.
According to the equation (4), the capacitance will increase with
the increasing area of PdO electrode because of the edge effect in
the capacitance. A higher capacitance is capable to hold more ionic
species, which are favorable for the reactions at the interficial of
PdO electrode and the electrolyte. Therefore, the charge density
(ρPdO) is increased due to the edge effect, then it will lead to a
higher sensitivity. A simplistic model uses the sensor with Na-Y
as an example, and is shown in Fig. 10.
Potentiometric sensors based on noble metal oxide electrodes
are an emerging field that is still awaiting consolidation and rationalization. A PdO sensing electrode, printed on YSZ and Na-Y, shows a potentiometric response to changes in CO partial pressure. Doping on PdO could result in a fair improvement in the
sensing performance owing to the creation of defects, which
would aid in the improvement of the oxidation reaction. The
potentiometric response of sensors is ascribed to the interfacial
potential between the PdO-based electrode and the electrolyte, e.g., zirconia and zeolite. The most important question is this:
What qualities make PdO different from other metal oxides as a
sensing electrode? The crystal of PdO has a sub-micro size, thus
the PdO electrode has a larger surface area than the oxide electrode used in other literature29, 30. Furthermore, PdO has excellent
catalytic activity; thus, gas molecules will be consumed before
they reach the TPB. Finally, the relatively weak metal oxygen
bonding in PdO could facilitate the transport of oxygen vacancies
and the adsorption of charged species. It is advisable to study PdObased electrodes with other electrolytes to explore more applications of solid-state electrochemical sensors.
Supporting Information: available free of charge via the
internet at http://www.whxb.pku.edu.cn.