A Double Layer Sensing Electrode “BaTi_(1-X)Rh_xO₃/Al-Doped TiO₂” for NO₂ Detection above 600 °C

Abstract

NO₂ emission is mostly related to combustion processes, where gas temperatures exceed far beyond 500 °C. The detection of NO₂ in combustion and exhaust gases at elevated temperatures requires sensors with high NO₂ selectivity. The thermodynamic equilibrium for NO₂/NO ≥ 500 °C lies on the NO side. High temperature stability of TiO₂ makes it a promising material for elevated temperature towards CO, H₂, and NO₂. The doping of TiO₂ with Al³⁺ (Al:TiO₂) increases the sensitivity and selectivity of sensors to NO₂ and results in a relatively low cross-sensitivity towards CO. The results indicate that NO₂ exposure results in a resistance decrease of the sensors with the single Al:TiO₂ layers at 600 °C, with a resistance increase at 800 °C. This alteration in the sensor response in the temperature range of 600 °C and 800 °C may be due to the mentioned thermodynamic equilibrium changes between NO and NO₂. This work investigates the NO₂-sensing behavior of duplex layers consisting of Al:TiO₂ and BaTi_(1-x)Rh_xO₃ catalysts in the temperature range of 600 °C and 900 °C. Al:TiO₂ layers were deposited by reactive magnetron sputtering on interdigitated sensor platforms, while a catalytic layer, which was synthesized by wet chemistry in the form of BaTi_(1-x)Rh_xO₃ powders, were screen-printed as thick layers on the Al:TiO₂-layers. The use of Rh-incorporated BaTiO₃ perovskite (BaTi_(1-x)Rh_xO₃) as a catalytic filter stabilizes the sensor response of Al-doped TiO₂ layers yielding more reliable sensor signal throughout the temperature range.

1. Introduction

As compared with other emission pollutants, the amount of NO_x generated by combustion processes is very high in the atmosphere and the increase of NO_x and derivatives damage the upper atmospheric ozone layer which provides a protective shield acting as a filter to UV radiation coming from the sun. Moreover, NO_x may react with atmospheric water resulting in the formation of acid rain [1].

In case of complete combustion, the reactant burns in oxygen, producing a limited number of products. When a hydrocarbon burns in pure oxygen, the stoichiometric reaction will only yield carbon dioxide and water as well as nitrogen oxides (NO_x). Incomplete fuel oxidation yields nitrogen oxides (NO_x), mainly unburned hydrocarbons (HC), carbon monoxide (CO), CO₂, and water [2,3]:

$$C_n{H}_n+air\stackrel{combustionsystem}{\to }N{O}_x+CO+HCs+H_{2}O+C{O}_2$$

(1)

Nitrogen oxides need to be converted over noble metal-containing catalysts to harmless compounds such as N₂, H₂O, and CO₂. Despite the presence of a catalyst, for instance a three-way-catalytic converter in automotive exhausts, the conversion is not always completed to fully eliminate NO and/or NO₂. The resulting emission still contains toxic greenhouse gases.

NO_x concentrations in diesel exhausts vary typically between 50 ppm up to 1000 ppm [4]. If concentrations are given in mass units, NO_x is usually expressed as NO₂ equivalent [4]. Catalytic converters reduce exhaust gas emission below a level foreseen by regulations in order to protect human life [5,6].

Although modern vehicles are getting cleaner due to improved engines and better emission controls, this holds only when engine and converter are in proper condition. When an engine is not at maximum efficiency, consequently performance is lost, fuel is wasted, and air pollutant emissions increase. Typically, in order to adjust the fuel-efficiency, excess air (or oxygen) is added in diesel engines during combustion. Additionally, emissions will increase drastically when defects occur in a catalytic converter for instance by engine misfire or due to the driving-related issues. These may remain unnoticed if the vehicles have no integrated emission control system. Onboard diagnostic (OBD), which is in vehicles since 1996, can detect emission problems of a car or truck. OBD systems are designed to alert the driver when a component in the engine management or exhaust system begins to deteriorate or malfunction. Early detection of minor problems can often prevent more costly damage to components, such as the catalytic converter or even the engine itself. Systems such as OBD rely on multiple gas sensors controlling the exhaust condition and emissions [7].

Generally, metal oxide semiconducting (MOx materials are employed for the detection of various oxidizing and reducing gases due their wide applicability and with simple and cheap fabrication [8]. Despite having many benefits, limited working temperature range and poor selectivity are two main drawbacks highlighted in the literature [9,10,11]. At lower temperatures (below 150 °C) and at temperatures above 400 °C, the sensing properties of the semiconducting materials are reduced drastically due to decreased conductivity and/or thermal instabilities [10,11,12,13].

A large amount of efforts are being spent by current research and development to address such issues in order to broaden the application spectrum of semiconducting oxides in high-tech industries. The most widely utilized semiconducting oxides are SnO₂, WO₃, ZnO, and TiO₂ [10,11,12,13,14,15,16,17,18,19]. Among those, TiO₂-based thin films and layers are reported to be more stable at higher temperatures and the stability can be further improved by bulk doping and by surface decorating with suitable cations [10,13,15,16,17,18,19,20]. In a previous study [10], we demonstrated that Al-doped TiO₂ sensor layers can detect NO₂ up to 800 °C in the concentration range of 50 ppm to 200 ppm, but the sensor response alters from the n-type at 600 °C to the p-type at 800 °C. Studying the reasons causing this change in sensor response, we demonstrated that this correlates well with the NO₂/NO thermodynamic equilibrium change occurring at temperatures above 600 °C.

Typically, the MOx-containing resistive sensors yield signals in opposite directions toward NO and NO₂. Thus, such changes in the high-temperature environmental condition are detrimental for diagnostic sensors and impair their long-term reliability. A common trend of improving the chemical stability and selectivity of this type of sensors is the use of a catalytic layer that acts as a catalyst, a sequestration barrier or a physical or chemical filter for undesired gases [21,22]. Hence, in order to maintain a reasonable sensitivity, better chemical stability and reliable sensor response with MOx sensor layers at elevated temperatures, the use of an additional catalytic filter or layer may be necessary and useful. Previously, it was reported that mesoporous and noble metal-containing catalytic filters can improve the sensing ability of semiconductors [21,22]. The use of SnO₂/CeO₂ duplex layers is shown to yield improved CO-response when the sensor is exposed to heavily-reducing gas mixtures (i.e., methane and/or other hydrocarbons) [23]. Hübner et al. [24] demonstrated later that a BaTi_1-xRh_xO₃ catalytic layer on the surface of SnO₂ sensor layer can eliminate the sensitivity towards hydrogen when exposed to high humidity CO + H₂ gas mixture while maintaining the selectivity towards CO. Our recent report indicates that La-based complex perovskite (LaFeCoPdO₃) layers deposited on sputtered TiO₂ sensor layers enable the hold of their sensing ability despite aging in a harsh exhaust environment at 850 °C [25].

The current context follows this type of an approach where a BaTi_0.95Rh_0.05O₃ perovskite layer is combined with the Al-doped TiO₂ (Al:TiO₂) sensing layer in order to achieve a sensor array that can be integrated in catalytic converter monoliths of vehicles and, thus, allowing reliable monitoring of NO₂ emission in exhaust environments at temperatures exceeding 600 °C. Such sensor arrays will be placed within the catalytic converter of a vehicle and, therefore, must detect NO₂ with long-term selectively over a threshold concentration by withstanding these harsh conditions (i.e., thermal aging caused by high gas flow rates and elevated temperatures). An adaptive and self-regenerative catalyst such as this can provide a thermal protection for the MOx-based sensing layer, as well as enabling a reliable sensor response toward NO₂ at temperatures as high as 900 °C.

2. Experimental Section

2.1. Synthesis of Al-Doped TiO₂ Layers by Sputtering

Al-doped TiO₂ thin layers were deposited via a reactive magnetron radio frequency (RF) sputtering technique from metallic targets using oxygen as a reactive gas and without applying any heating of the substrate or system. Sputtering equipment (from von Ardenne GmbH, Dresden, Germany) with two sputter sources was used to deposit the coatings in an argon + oxygen gas mixture. Al-doped TiO₂ thin layers were deposited on the polished Al₂O₃ sensor platforms, onto which the interdigitated Pt-circuits were previously deposited by screen-printing. For the deposition of Al-doped TiO₂, two metallic targets, i.e., Ti (99.95% pure) and Al (99.99% pure) were used and were placed face to face in a plasma reactor chamber. During the coating process, the substrate holder was rotated at 13 rpm. This rotational movement enables the homogenous distribution of aluminum in the TiO₂ matrix. A power of 80 W at the Al target and 500 W at the Ti target yielded 4.64 at.% Al in TiO₂. The partial pressure of argon and oxygen during the coating was 27.4 and 5.7 sccm, respectively. After 12 h of sputtering process, an Al-doped TiO₂ coating with a thickness of 1.2 µm was achieved. Pt-interdigitated electrodes (IDEs) are composed of two connection tracks having 300 µm-wide interlaces with bands/gaps of 300 µm from each other on alumina substrates (see Figure 1). The employed (IDE) sensor design requires no heater at the backside of the sensor platforms. Simultaneously, we deposited thin undoped and Al-doped TiO₂ layers (ca. 1–1.5 µm) onto the sapphire single crystal substrates in order to enable XRD measurements without any interference from the polycrystalline alumina substrate material.

2.2. Synthesis of the Catalysts

Ba-acetate (Ba(C₂H₃O₂)₂, 98% Chempur), Ti-isopropoxide (Ti{OCH(CH₃)₂}₄, 99.999% Aldrich), and Rh-nitrate (Rh(H₂O)(OH)_3-y(NO₃)_y y=2-3) were employed as starting chemicals to prepare the BaTi_(1-x)Rh_xO₃ perovskite. Barium acetate was dissolved in acetic acid and Ti-isopropoxide was diluted with isopropanol and acetyl-acetone solution. Barium and Ti sols were homogenously mixed and then the Rh-nitrate solution was added to this mixture. The amount of the aqueous Rh-nitrate was calculated to obtain perovskite powders of the following composition BaTi_0.9Rh_0.1O₃. The solvents were slowly evaporated under constant stirring at room temperature. The catalysts were finally calcined in a furnace under static air up to 1000 °C for 5 h with a heating rate of 10 °C min⁻¹ [26].

The catalytic powder was then milled and mixed with some binder for screen printing on the interdigitated electrode sensor platforms and/or on the Al-doped TiO₂ sensing layer, which was previously sputtered on Pt-interdigitated electrodes.

2.3. Sensor Array Layout

Sensor platforms were fabricated by deposition of sensing layer on interdigitated Pt-electrodes (IDE) which were deposited previously on Al₂O₃ substrates as shown in Figure 1, left. The catalytic filter was then brought on the sensor layer by means of screen printing as thick and porous layer (see Figure 1, bottom).

2.4. Characterization Methods

Al-doped TiO₂ sensing layers and BaTi_0.9Rh_0.1O₃ catalytic powder layers are investigated by SEM/EDX for their compositions and by X-ray powder diffraction method (XRD) for the phase condition.

Microstructure and morphology analysis of the sensing electrodes in terms of porosity, grain size, and surface condition were done by means of FE-SEM (Carl Zeiss Microscopy, Ultra 55, Oberkochen, Germany). EDX analysis were carried out by GEMINI (Oxford Inst. GmbH, Wiesbaden, Germany). The XRD diffractograms of the catalysts were obtained in a diffractometer D5000 from SIEMENS with Cu Kα radiation (λ = 1.54178 Å). The reflections from the JCPDS (Joint Committee on Powder Diffraction Standards) database were assigned to the experimental diffractograms with the program EVA from BRUKER AXS. Scanning Electron Microscopic (SEM) analysis of the catalysts was carried out in a Zeiss Ultra 55 microscope equipped with an Energy Dispersive X-ray Spectrometer (EDS) from Oxford Instruments. For high-temperature XRD measurement of BaTi_0.95Rh_0.05O₃ catalyst, the specimens were prepared by packing powder in Ø 25 mm alumina crucible. Data collection time was 30 s per scan. The test powders were initially calcined at 700 °C prior to in situ high-temperature X-ray diffraction (HT-XRD) measurements.

Sensing characterization was carried out in a specially constructed apparatus consisting of a tube furnace with cascade control and a custom-built quartz glass reactor providing a thermocouple directed at the specimen. The composition of the mixed gas was controlled by an eight-channel mass flow controller (MFC-647b from MKS Instruments GmbH). The gas flow rate was adjusted to 400 mL/min and the NO₂ concentrations were varied between 50 and 200 ppm. Mostly argon was used as the carrier gas. In some of our experiments, synthetic air (80% N₂ + 20% O₂) was employed as the carrier gas indicated relevantly. The DC electrical measurements were carried out by using a Keithley 2635A Sourcemeter, with a computer-controlled LabView program. A constant current of 1 × 10⁻⁶ A was applied during all sensor measurements. Steady-state gas response measurements were started under a constant carrier gas flow. These sensors were initially exposed to a constant flow of carrier gas to equilibrate the operating conditions. After a stable baseline was reached, the target gas was introduced into the chamber, continuously flowing with a constant for 40 min at the end of which the NO₂-gas flow was ceased and only carrier gas is let flown through. These two steps were repeated for 50, 100, and 200 ppm NO₂ and/or CO gas concentrations.

3. Results and Discussion

3.1. Al-Doped TiO₂ Sensor Layer

As the XRD investigation of the as-coated and ex situ at 500 °C, 800 °C, and 1000 °C heat-treated Al-doped TiO₂ layers indicate, the layer is weakly crystalline and contains the anatase phase in the as-coated case. The top-view SEM image of the as-coated layer displays densely packed fine columnar morphology of the layer (Figure 2, left). Further heat-treatment increases the anatase crystallinity. At around 800 °C, the first indication of anatase-rutile transformation can be seen (Figure 3). After 3 h of heat-treatment at 1000 °C, full conversion to rutile takes place. The Al-doped TiO₂ layers which were used for sensor testing were calcined at 800 °C and, thus, contained mainly anatase phase and some minor amounts of rutile. The SEM top-view micrograph of the 800 °C heat-treated Al-doped layer indicates the crystallization development yielding some porosity at the grain and column boundaries (Figure 2, right). The sensor tests of Al-doped TiO₂ layers towards NO₂ were published in details elsewhere [10]. The most striking observation with this high-temperature sensor material was the alteration of the sensor signal direction at 800 °C yielding a resistance decrease under NO₂ exposure being opposite to the resistance increase observed at 600 °C (see Figure 4). No significant effect of humidity on the sensor response was observed at these temperatures.

This behavior can be assumed to be due to the conversion of NO₂ to NO because of the above-mentioned and the Figure 5 illustrated thermodynamic equilibrium conditions which may result in the formation of ionized oxygen species (e.g., O²⁻-ion) taking over the control of sensing mechanism through surface adsorption (see Equation (2) and [10]):

$$N{O}_2\left(gas\right)+2e^{-}\to NO\left(gas\right)+O^{2-}\left(ads.\right)$$

(2)

This behavior of the Al-doped TiO₂ layer indicates poor NO₂-sensing capability at temperatures above 600 °C. On the other hand, Al-doped TiO₂ shows negligible sensitivity towards CO concentrations varying from 50 to 200 ppm in the given temperature range. Moreover, these layers are more sensitive to NO₂ when NO₂ and CO are simultaneously present in the test environment, indicating a minimized cross-sensitivity towards CO [9].

In order to overcome the NO₂ sensing deficit of the Al-doped TiO₂ layer, a perovskite layer as a catalytic filter is employed. This Rh-incorporating perovskite has a self-regenerative property which results in the segregation of Rh nanoparticles on the reduction and re-incorporation into the lattice upon oxidation [25]. Relying on the NO oxidizing capability of Rh, the effects caused on a sensor due to the high-temperature deviations in NO₂ containing gas environments can be eliminated.

3.2. Catalytic BaTi_0.95Rh_0.05O₃ Powder

The SEM observations of the BaTi_0.95Rh_0.05O₃ catalytic material exhibited that under oxidized conditions (e.g., after calcining at 1000 °C in air), large prismatic grains of perovskite were formed. In turn, on heat-treatment of the oxidized powder, this time under reduced conditions (e.g., 900 °C in 2.5 vol.% H₂), Rh metallic particles segregates at the grain boundaries of the perovskite (Figure 6, left and right, respectively).

In situ high-temperature XRD analysis of pre-calcined powder in the range of RT-1000 °C-RT showed that the increase of temperature results in shifting of all X-ray diffraction peak positions of BaTi_0.95Rh_0.05O₃ to lower Bragg angles (2θ) due to expansion of the lattice (see Figure 7). After heating to 700 °C in air, the BaTi_0.95Rh_0.05O₃ powder yields the BaTiO₃ perovskite, BaTi₂O₅, and BaCO₃ phases.

BaCO₃ disappears at around 800 °C, probably due to the low melting point of the compound. In turn, at this temperature, Ba₂TiO₄ phase appears to form. On heating to 1000 °C, next to the well-crystallized BaTiO₃, the BaTi₂O_5, and Ba₂TiO₄ phases at lower intensity, and a trace TiO₂ phase are detected. At this stage, the presence of some trace amounts of RhO₂ cannot be ruled out even though the matching peak positions were not identified due to the overlapping. After cooling down to RT, the in situ to 1000 °C heated powder contains BaTiO₃, BaTi₂O₅, and the (111) plane of Ba₂TiO₄ phases.

In summary, it can be concluded that heating to 1000 °C and cooling down to RT in air of the BaTi_0.95Rh_0.05O₃ catalytic powder results in full decomposition of BaCO₃ and the formation of BaTi₂O₅ and Ba₂TiO₄ through the irreversible phase conversion. These phases may contain partially-incorporated Rh. No clear identification of the presence of metallic Rh or RhO₂ was possible during in situ X-ray of this in air heat-treated powder.

After reduction treatment in H₂, BaTi_0.95Rh_0.05O₃ perovskite powder displays fine agglomerates which are associated with the presence of metallic Rh (Figure 6, right). The presence of cubic Rh with the 2θ position of 47.8° in the X-ray diffraction pattern of the reduced perovskite powder indicates that the hexagonal perovskite decreased upon reduction in H₂. The sensor tests were carried out with a BaTi_0.95Rh_0.05O₃ powder which was calcined in air at 1000 °C.

In order to understand the contribution of BaTi_0.95Rh_0.05O₃ perovskite to the sensor response, initially the sensor response of this perovskite has been recorded. For that, some interdigital sensor platforms were coated by screen-printing with a thick BaTi_0.95Rh_0.05O₃ perovskite powder. These sensors were tested towards NO₂ and CO in the form of single gas exposures having different concentrations at 600 °C to 900 °C in a dry argon environment (see Figure 8).

Thick layers of BaTi_0.95Rh_0.05O₃ perovskite catalyst yield remarkable response towards NO₂ at the temperatures of 600 °C to 900 °C and in dry argon, but also a high sensor response towards CO (50 to 200 ppm). As the sensor response towards NO₂ and CO improves with the temperature, the CO response alters to the opposite direction at 900 °C (see Figure 8). This behavior of the catalytic layer may be due to the CO-oxidation capability. Our previous unpublished tests carried out with this catalytic perovskite by using 1% CO, 5000 ppm O₂ diluted in He indicated a very low light-off temperature for CO oxidation (below 210 °C). While under less oxygen-containing conditions, i.e., by exposing the perovskite to 1000 ppm CO and 1000 ppm NO in He, it was observed that CO conversion occurs at significantly higher temperatures.

The addition of 5% humidity to the argon environment results in an improvement of the NO₂ response of BaTi_0.95Rh_0.05O₃ perovskite by a factor of two at 800 °C (e.g., 0.02 vs. 0.04 for 100 ppm at 800 °C) and somewhat less at 900 °C (see Figure 9). Similarly, CO response in the presence of humidity yields resistance increase at 900 °C, as observed under dry conditions, indicating an adsorption promotion effect of OH^−δ ions (see [14]). The discrepancy, concerning CO response at 900 °C under dry conditions, can be explained with the facts that water vapor or OH-ions at this temperature may either react with CO during adsorption on the catalyst surfaces or the BaTi_0.95Rh_0.05O₃ influences catalytically the adsorption mechanism of CO under oxidizing conditions. All sensor responses with BaTi_0.95Rh_0.05O₃ were p-type at all test temperatures with the exception of CO exposure at 900 °C. Nevertheless, the sensor response towards CO is significant. Thus, considering this fact, it is expected that the BaTi_0.95Rh_0.05O₃ layers as a gas sensing material may display a high CO cross-interference under exposure to CO concentrations higher than 50 ppm at high temperatures.

3.3. NO₂ Response of BaTi_0.9Rh_0.1O₃ Catalyst + Al-Doped TiO₂ Duplex Layers

According to the concept presented above in Section 2.1, duplex sensor layers having catalytic BaTi_0.95Rh_0.05O₃ perovskite and Al-doped TiO₂ are tested for their NO₂-response in the temperature range of 600 °C to 900 °C.

The NO₂ sensor response obtained with these duplex layers display remarkable sensitivity values at 600 °C, 800 °C, and 900 °C despite high gas temperatures (e.g., 0.12 at 600 °C, 0.46 at 800 °C, and 0.92 at 900 °C, upon exposure to 200 ppm), (see Figure 10). Moreover, the change in signal direction which was observed with the Al-doped TiO₂ layer as indicated with the sensor responses given in Figure 4 does not occur when duplex sensing layers are employed.

In order to clarify whether this response at such elevated temperatures is from NO₂ or from NO, the sensor response of this duplex layer towards NO was recorded in dry synthetic air (see Figure 11). Normally as shown with Figure 5, above 600 °C under dry conditions, a temperature-dependent conversion of NO₂ to NO occurs.

The sensor tests of the BaTi_0.95Rh_0.05O₃ perovskite and Al-doped TiO₂ duplex layer carried out toward NO in a dry synthetic air environment at 600 °C exhibited significantly lower sensitivity values (e.g., 0.018 and 0.023 under exposure to 50 ppm and 200 ppm, respectively). Further analyses, which were carried out to observe the effect of catalytic perovskite powder on NO₂/NO conversion at high temperatures, indicated that full NO₂ to NO conversion already occurs at 300 °C if BaTi_0.95Rh_0.05O₃ catalytic powder was exposed to 200 ppm NO₂ in the sensor chamber (see Figure 12). In turn, when the chamber is empty, i.e., without the perovskite catalyst, the NO₂-to-NO conversion develops gradually, starting from 500 °C, and is completed above 800 °C as, at that temperature, no more NO₂ can be measured in the system’s output (see Figure 5). Thus, it is reasonable to assume that the BaTi_0.95Rh_0.05O₃ catalytic top layer conditions the test environment to have NO+O²⁻ at all test temperatures and catalytically oxidizes NO to NO₂ at the interface of the perovskite and the Al-doped TiO₂ layer [10]. Thus it can be concluded that exposure of the sensor to NO₂ at higher temperatures yields a sensor signal which cannot be correlated to NO and to molecular oxygen from the synthetic air, but to NO₂.

Table 1. Sensor properties of various sensor layers in the temperature range of 600–900 °C in dry or in 5% humid argon environment.

Sensing Layer vs. 200 ppm NO2	Gas Temperature (°C)	Carrier Gas (Argon) Environment	Sensitivity	Reaction Time (min)	Recovery Time (min)
Al:TiO2	600	dry	0.76	3.0	13.0
Al:TiO2	600	5% humidity	0.69	2.5	17.0
BaTiRhO3/Al:TiO2	600	5% humidity	0.92	2.0	4.3
BaTiRhO3	600	dry	0.11	2.2	3.4
Al:TiO2	800	dry	0.50	3.7	22.0
Al:TiO2	800	5% humidity	0.46	3.8	4.2
BaTiRhO3/Al:TiO2	800	5% humidity	0.16	1.6	2.5
BaTiRhO3	800	5% humidity	0.050	3.7	8.3
BaTiRhO3	800	dry	0.052	7.3	7.6
BaTiRhO3	900	5% humidity	0.053	3.4	4.5
BaTiRhO3/Al:TiO2	900	5% humidity	0.061	3.2	4.3

summarizes a selection of sensor properties of single and duplex sensor layers investigated in this study. For comparison purposes, only the responses toward 200 ppm NO₂ are considered. Sensitivities of Al:TiO₂ single layer and BaTiRhO₃/Al:TiO₂ duplex layer are relatively high at 600 °C (0.70 and 0.92, respectively), but the recovery of sensor response with the single Al:TiO₂ layer requires extremely longer times. At higher temperatures, and in a humid environment, this appears to be shortened. The sensor with the single BaTiRhO₃ layer exhibits, in turn, the lowest sensitivities (0.11 and 0.05) at all test temperatures (600–900 °C). No further NO₂-sensing measurement is carried out at 900 °C with the single Al:TiO₂ layer because of the lack of reliability of its sensor signal.

The sensitivity of the sensor with the duplex layer reduces down to 0.16 at 800 °C, and further to 0.06 at 900 °C. Although the sensitivity of the duplex layer sensor (i.e., BaTiRhO₃/Al:TiO₂) is lower than that of the Al:TiO₂ single layer, the reaction and recovery times of the sensor with the duplex layer is partly to half as short than those of the Al:TiO₂ single layer sensor (reaction times = 2.0 min vs. 2.5 min at 600 °C and 1.6 min vs. 3.8 min at 800 °C, recovery times = 4.3 min vs. 17 min at 600 °C and 2.5 min vs. 4.2 min at 800 °C, respectively). The sensitivities and the reaction and recovery times of the BaTiRhO₃/Al:TiO₂ duplex layer sensor remain in the same range at all test temperatures, indicating a maintained reliability with this sensor.

Humidity shows no significant effect on sensitivity, although some influence on reaction and recovery times are observed favoring humidity at 800 °C, and oppositely at 600 °C. Therefore, it is not possible to identify any specific trend regarding the humidity influence. Although it is obvious that the BaTiRhO₃ catalytic layer intervenes in the thermodynamic equilibrium condition of NO₂/NO environment by lowering the conversion temperature significantly, as a single sensor layer, the sensor with the BaTiRhO₃ single layer does not yield good performance for NO₂-sensing.

The literature indicates that perovskites with different formulations give great opportunity to detect gas mixtures by means of sensor arrays. Doping elements either in the lattice of perovskite or on the surface of perovskite could help to improve sensing performance of perovskite oxide and to decrease the sensing temperature [27,28]. La-based, Ce and Pd-doped perovskites are proven to yield good sensing behavior toward CO and CH₄, respectively. No other sensor study is known to date with the BaTiRhO₃ catalytic layer to confirm its behavior. However, Figure 9 displays that the sensor with a single BaTiRhO₃ layer yields greater signals toward CO than NO₂.

The sensors with the BaTiRhO₃/Al:TiO₂ duplex layers, in turn, exhibit a NO₂-selective sensing as exposed to CO + NO₂ mixed gas environments in the temperature range of 600 °C and 900 °C. Although stronger signals are observed by exposure only to CO, it appears that these signals are in the same direction as those toward NO₂ (Figure 13). Usually, the semiconducting sensors yield signals in opposite directions when exposed to oxidizing (e.g., NO₂, O₂) and reducing gases (e.g., CO, CH₄).

Previous studies which used XPS to determine the nature of CO adsorption on the surface of doped LaCoO₃ perovskite formulations as the sensor material have revealed that CO preferentially adsorbs on the lattice oxygen in the form of mono-dentate carbonate on the surface of perovskite-type oxide. Ce (e.g., La_0.9Ce_0.1CoO₃) doping changed not only the total amount of adsorbed CO but also the nature of adsorption phenomena. By adding Ce, in addition to the mono-dentate form, the bi-dentate carbonate form was also appeared on the surface resulting in an increase of the total adsorbed CO [27]. In the case of Pd-doped LaFeO₃, the noble metal gives an active site on the metal oxide to adsorb methane species. Pd is well known for its H dissolving capacity, which could help to create PdH pair in methane adsorption phenomena [28].

In our case, it is likely that under single CO gas exposure in dry argon, CO adsorption rapidly consumes the surface-adsorbed oxygen of the sensing layer forming CO₂ and excess electrons according to Equation (3):

$$C{O}_{\left(g\right)}+O_{ads}^{2-}\to C{O}_2+2e^{-}+V_O^{″}$$

(3)

Hence, it is probable that the sensors with the duplex layer exhibit CO cross-interference. In order to test this, the sensor response of the BaTi_0.95Rh_0.05O₃ catalyst + Al-doped TiO₂ duplex layers was recorded under NO₂ and CO single-gas exposure, followed by NO₂ and CO mixed gas exposure at 800 °C and 900 °C (Figure 13 left and right). These tests displayed that the duplex layer is more sensitive to NO₂ in the presence of CO, while under CO single-gas exposure, relatively high sensor sensitivity is recorded. The resistance decrease obtained on exposure to gas mixtures of 50 ppm NO₂ + 50 ppm CO, 100 ppm NO₂ + 100 ppm CO, and 200 ppm NO₂ + 200 ppm CO equals to that obtained by exposure only to NO₂ at concentrations of 50, 100, and 200 ppm, respectively. In turn, under NO₂ + CO gas mixture exposure, NO₂ is converted to NO + O²⁻ at temperatures above 300 °C through the BaTi_0.95Rh_0.05O₃ catalytic layer and this ionized oxygen may readily react with CO without being in interaction with the sensor surface and is removed as CO₂, while the converted NO reacts with the oxygen adsorbed on the surface of Al-doped TiO₂ sensing layer. Thus, at temperatures as high as 800 °C and 900 °C, the sensor with the BaTiRhO₃/Al:TiO₂-duplex layer gives signals preferentially corresponding to NO₂, despite being exposed to CO + NO₂ gas mixtures at relatively higher concentrations (Figure 13, left and right).

4. Conclusions

Trivalent doping of TiO₂ enables sensor response towards NO₂ at temperatures up to 800 °C. However, above 600 °C, a change in the direction of the sensor signal is observed. This behavior of TiO₂ sensors can be eliminated by coating the sensor layers with a BaTi_0.95Rh_0.05O₃ catalytic filter layer to obtain a reliable sensor response at temperatures as high as 900 °C. The BaTi_0.95Rh_0.05O₃ layer catalytically converts NO₂ to NO already above 300 °C and, thus, conditions the sensor environment by formation of ionized oxygen allowing a sensor response with descent sensitivity. Moreover, CO interference is eliminated by using BaTi_0.95Rh_0.05O₃ as a catalytic filter for sensing NO₂ in the CO + NO₂-mixed gas environments. However, it is not clear if solely the precious metal, Rh, which is present in the BaTi_0.95Rh_0.05O₃ perovskite is responsible for this conditioning or an interaction with the perovskite lattice occurs. In order to clarify this, the tests must be carried out using a catalytic filter made of pure BaTiO₃ perovskite on the Al-doped TiO₂ layer. Conventional TWC catalyst cannot be used for this purpose (i.e., as a catalytic filter) due to the lack of necessary electrical conductivity.