The oxygen sensor has been used to control combustion process, especially in the automobile engineering. Various types of oxygen sensors such as electrochemical type($ZrO_2$), limiting-current type and resistive type have been developed. Among them, $SrTiO_3$, which is a nonstoichiometric oxide with perovskite structure, is an oxygen sensor that operates on the principle of the semiconductor. In the case of pure $SrTiO_3$, it has large diffusion coefficient originating from coupled motion of ionic charge carriers and electronic charge carriers(ambipolar diffusion phenomena). So, it was reported that one possible way to realize the measurement and control of fast combustion processes, especially in automobile engineering is the use of resistive oxygen sensors based on thin or thick films of titanates. But in this case, the sensor can be only applicable at higher temperature than at the one of electrochemical type oxygen sensors($ZrO_2$). But pure $SrTiO_3$, this material shows p-n transition type conductivity with decreasing oxygen partial pressure. $SrTiO_3$ is p-type semiconductor in the lean-burn region, but the p-type conductivity changes into n-type near $PO_2=10^{-2}$ Pa. Therefore this results in the unfavorable resistivity characteristics for oxygen sensor. The most useful material for detecting the stoichiometric point in automobiles is one which stands for an n-type or p-type conductivity over whole range(no p ↔ n transition). N-type dopants like La can extend the n-type semiconductivity region in a oxygen partial pressure higher than $10^{-2}$ Pa.
In this study, the obejectives for oxygen sensors (test condition is 0.1Hz switching rate between λ=0.95 and λ=1.05 in the range of 300-700℃) prepared by La-doped $SrTiO_3$ thick film are as follows.
1) The effects of La addition on the oxygen sensors in the case of type A(Pt electrode and thick film sensors was sintered simultaneously at 1300℃ for 1 hour) sensors. : to find optimum composition, the contents of La was varied from 2 to 10 mol%.
2) To lower the operation temperature than the one of $ZrO_2$ sensor and examine the microsturctural effect on the transition behavior of lambda sensor, type B(10mol% $La_2O_3$-doped $SrTiO_3$(Sr/Ti ratio=0.9)) sensors was developed. The sintering condition was controlled from 1275 to 1325℃. And Pt electrode sintered at 1050℃ for 10 minutes.
3) To fasten the response time and examine the effect of catalyst on the response behavior, Pt and Au particles were distributed on the top surface of thick film sensors by DC magnetron sputtering.
The lattice parameter increases linearly with increasing $La_2O_3$ content within 10 mol%. In X-ray diffration analysis, it may be considered that it is difficult to determine the well-defined solubility limit of $La_2O_3$ but it may be exist near 10 mol%. XPS spectra shows that regardless of La addition, the energy separation of two peaks in Sr3d and Ti2p are almost constant as ca. 1.6eV, 5,7eV respectively. This means that Sr and Ti species existed in the form of SrO and $TiO_2$. But binding energy of each element decrease with La addition.
Type A sensors shows that with increasing the La addition, the difference of the sensor resistance between rich- and lean-burn condition was increased and sensitivity was increased. But response time was very late not enough to show fully-satureated voltage(or resistance) at 0.1 Hz switching rate between λ=0.95 and λ=1.05. And Nonideal response, that is nonsymmetric response between rich to lean transition and vice versa was very serious.
Type B Sensors : The sensor sintered at 1250℃ for 1hour has very small grain size less than 0.1 m and not reacted enough to form definite boundaries between grains to distingsh each grain. Many open pores exist in the site of organic verhicles which were vaporized at relatively low temperature. With increased temperaure, neck formation between grains occured gradually and grains formed definite grain boundaries. The grain size of the sensor sintered at 1325℃ for 4 hours is about 3 m and surface sturcture is appeared very densely except for a few large pores. In the case of the sensors which sintered at relatively low temperature, the response time of the sensors operated at 608℃ is shorter than 400 msec and increases longer about 500 msec at 366℃. Moreover the voltage switching curve shows step-wise form, that is nonideal behaviors(differences of response time transition from lean to rich and vice versa.) could not observed. And the response time did not depend on the sintering temperature. But increasing the sintering temperature, the response time of the sensors increases gradually with decreaing the operation temperature, especially on low operating temperature. So the sensor sintered at 1300℃, for 1 hour shows sluggish response only at 366℃ and the one sintered at 1325℃ for 1 hour shows slower response at 433℃ as well as 366℃. The sensor which was sintered at high temperature and had dense microstructure and large grain shows very sluggish response time regardless of operation temperature. There is a 2 to 3 orders of magnitude change in resistance between rich and lean within the temperature range of interest. The sensitivity, $S=R_{lean}/R_{rich}$, of the sensors sintered at 1275 and 1300℃ for 1 hour is $10^3 -10^4$ and less dependent on the operating temperature. Lean resistance at high temperature, extrapolated to 850℃ from measured values, does not overlap the rich resistance at low temperature. Therefore, there is a clear indication whether the exhaust gas is rich or lean of stoichiometry from 366 to 850℃. But the sensors sintered at 1325℃ shows relatively low sensitivity about $10^2-10^3$. Moreover there is resistance ambiguity with temperature and difficult to applicate as an lambda sensor. So to overcome these problems, it seemed to be important to decrease the activation energy of the sensor in lean condition. And the sintering condition and the micirostructure of the sensor shoud be optimized.
Pt and Au particles smaller than 0.1㎛ on the top surface of the sensor was distributed very homogeneously by DC magnetron sputtering. The response time was decreased about than 250 msec for Pt and 100 msec for Au. But the nonideal behavior that is the difference between rich→lean transition time and lean→rich transition time could not overcome regardless of Pt and Au.
