1. Introduction
In order to monitor the air-to-fuel ratio and to detect directly the aging status of exhaust gas aftertreatment systems, ceramic oxygen gas sensors have been serialized [
1,
2]. For on-board diagnosis (OBD), hydrocarbon (HC) sensors have been proposed [
3], mainly based on a mixed potential principle [
4,
5]. In another approach, it was proposed to detect HC directly in the exhaust with conductometric sensors of metal oxides such as gallium oxide (Ga
2O
3) [
6] or doped strontium titanate (SrTiO
3) [
7]. Since the resistance of these materials also depends on the oxygen concentration of the exhaust and on the sensor temperature [
8,
9], a two-sensor-setup was introduced [
10,
11], with one sensor being catalytically activated, whereas the other one remained non-activated. The activated sensor measures the oxygen concentration in the equilibrium state. It should show the same temperature dependency as the non-activated one. Hence, it acts as a resistive λ-sensor and is not sensitive to HC anymore. The ratio of both sensor signals is expected to result in a sensor signal that is almost unaffected by temperature and oxygen content of the exhaust. However, in a later paper [
12] it turned out that e.g., for 1% Ta-doped SrTiO
3 this method is not sufficient to compensate the oxygen cross-interference completely. The compensation works only effectively at low oxygen concentrations, resulting in a sensor signal that is insensitive to oxygen. In
Figure 1, the normalized conductance
G (ratio of conductance of non-activated and activated sensor) is plotted. The data are recalculated from
Figure 5 of Ref. [
12]. Obviously, only at low oxygen concentrations an (almost) oxygen independent behavior occurs. At higher oxygen concentrations,
G depends on the oxygen concentration. The reason for this is that the conductivity of the activated material is almost insensitive to oxygen, whereas the conductivity of the non-activated one also depends on the oxygen concentration [
12].
In some older patents, a combination of an oxygen pumping cell with a sensor for combustibles to establish a well-defined oxygen concentration at the conductometric device, even if the oxygen content fluctuates, was suggested [
13]. For the first realizations using a thimble-type oxygen sensor mounted onto a steel cell, a commercially available conductometric SnO
2 sensor was used [
14].
In an initial setup, we prepared a
planar sensor in a multilayer ceramic technology including the challenge to join alumina and zirconia. It was verified that it is possible to establish a constant oxygen concentration at the cavity where the sensor film is located by applying an appropriate pumping voltage to the cell [
15]. Due the pumping voltage, oxygen ions are transported through the YSZ membrane, leading to an adjustable oxygen concentration at the position where the gas sensitive film is located.
The assembly as described in [
15] has two main cavities – a sensor chamber and a pumping chamber, which are connected via a very small diffusion channel. The electrochemical pumping cell presents the ceiling of the pumping chamber with one electrode facing the chamber and the opposed one facing the exhaust gas. The sensor chamber has access to the exhaust gas via a second diffusion channel. Since the highest sensitivity of donor-doped SrTiO
3 occurs at an oxygen concentration of 5% or less, in an initial approach it was demonstrated that one can establish a defined oxygen concentration by the electrochemical cell. For that purpose, a conductometric oxygen sensing film of SrTi
1−xFe
xO
3−δ was applied instead of a hydrocarbon sensitive film. Details of this oxygen sensor material can be found in ref. [
16]. A defined oxygen concentration was set in the ambient gas. At the beginning, no pumping voltage was applied. Then, the pumping voltage was increased stepwise. The pumping current followed immediately the pumping voltage, as well as the oxygen sensor resistance. Using a previously measured correlation between sensor film resistance and oxygen concentration, a relationship between cell pumping current and obtained oxygen concentration at the sensor film can be derived for several ambient oxygen concentrations. In
Figure 2, the oxygen concentration at the film is plotted versus the electrical pumping current. Obviously, it is possible to establish a defined oxygen concentration within the requested range at the position of the sensor film independently of the oxygen partial pressure in the exhaust gas. As typical for rich exhausts, no oxygen but water was present in the gas flow regarding the left part of the graph. Oxygen could, however, be generated by electrolyzing H
2O at the electrode facing the exhaust gas where it is abundant. The required current for obtaining an oxygen concentration at the film is about 15 mA. If 10% O
2 were present in the exhaust gas, one has to pump out oxygen from the sensor chamber. With sample 1, the oxygen concentration could not be reduced as required due to a leak at the joining points near the sensor chamber. Sample 2 had an improved setup, with which oxygen could be pumped out, so that 5% could be established with only 2.5 mA.
However, due to its long diffusion paths, the initial demonstrator setup showed a prohibitively slow sensor kinetics. Furthermore, the hydrocarbons reacted at the walls and since only few hydrocarbons reached the film, the sensitivity of this initial setup was lower than expected.
The assembly of sensor and pumping cell in [
15] has been further developed and is called sensor “platform” in the following. This paper deals with the new design of the platform and the integration of the two-sensor-setup. The catalytic activation of one of the sensor layers is discussed and sensors with Ga
2O
3 and 1% Nb-doped SrTiO
3 as gas sensitive films have been tested. These are the pilot test of the new sensor platform. The main function of the platform, which is the detection of HCs under a defined oxygen atmosphere independent of the oxygen partial pressure in the exhaust gas, has to be investigated in the next step.
4. Conclusions
A new sensor platform has been introduced. Very often high-temperature sintering steps as applied in zirconia multilayer technology (like, e.g., for the planar λ-probes) are impossible. Since many sensor materials cannot be co-fired with YSZ due to their lower sintering temperature, methods of low temperature joining are used, e.g., YSZ and Al2O3 are joined with glass solders. Besides conventional technologies such as tape technology and screen-printing, the laser cutting and laser patterning technology was employed. It provided a great freedom of design.
As an exemplary application, a setup using two resistive sensors for HC detection was successfully integrated into the new platform. Activated and non-activated sensor films were tested successfully. On the sensor platform, it is therefore possible to eliminate the remaining oxygen cross-sensitivity of the HC signal by the quotient of the sensor signals of activated and non-activated sensor layer. In order to improve accuracy and to operate the sensor also in rich atmospheres, the electrochemical pumping cell adjusts the oxygen concentration at the sensitive film. A platform heater could be developed and also tested successfully.
In the next step the two-sensor-setup with an integrated heater has to be joined with the electrochemical pumping cell and the covering in order to present the platform which has been introduced here. Tests of HC detection under a defined oxygen concentration at the sensitive layers have to be conducted.