Amperometric NOx-sensor for combustion exhaust gas control
Prof. Dr. Ir. H. Verweij
Dr. H.J.M. Bouwmeester
Increasing interest in saving fuel has resulted in the development of lean-burn concepts for combustion engines. In these cases an excess of oxygen is present in the gas/fuel mixture. As a result the combustion is more complete leading to a higher energy yield. Unfortunately, the amount of NOx present in the exhaust gas will increase, as discussed in Chapter 1. To meet future emission standards, a new catalyst technology has been developed. Repeated switching between lean-burn and rich-burn mode is required to regenerate the surface of the so-called NOx trap catalysts. This process can be controlled by an amperometric NOx sensor based on stabilised zirconia. In the proposed sensor, two electrodes are placed in sequence in a diffusion chamber. While residual oxygen in the exhaust gas is pumped away at the first electrode, the remaining NOx is sensed amperometrically at the second electrode. A dense oxygen-permeable ceramic membrane thereby shields the first electrode. Whereas oxygen is reduced selectively, the membrane surface needs to be catalytically inert towards reduction of NOx. In this research, a number of studies on different types of mixed conducting materials are conducted. The main focus is on sufficient ionic conductivity combined with negligible catalytic activity towards NOx-reduction.
In Chapter 2, the applicability of a series of perovskites with overall compositions of A0.7A’0.3BOx (where A: Gd, Pr and Y, A’: Sr and Ca, B: Cr, Fe and Mn) was investigated. Ionic conductivity measurements were performed using electron-blocking microelectrodes using a Hebb-Wagner type of electrochemical cell. The ionic conductivity was almost independent of oxygen partial pressure for all compositions measured if the microelectrodes were shielded with a glass encapsulation to eliminate all short-circuiting paths. The absolute value of the ionic conductivity varied between 10-4and 10-6 S/cm. In all cases the measured conductivity was below the threshold value of 2.5·10-4 S/cm, corresponding to a j(O2) of 6.22·10-8 mol/cm2s, required for the targeted application.. The catalytic activity towards NO was studied by means of temperature programmed reaction (TPR). All compositions showed undesired catalytic activity toward reduction of NO.
In Chapter 3 a material with the nominal composition Gd0.7Ca0.3CoOx was studied. After sintering at 1200ºC in air SEM-EDX and XRD measurements revealed that three co-existing phases were present: (Gd0.6Ca0.4)2CoOx, GdCoO3 and CoO. Measurement of the catalytic activity demonstated that the composite Gd0.7Ca0.3CoOx showed no catalytic activity towards NO reduction, as required in the targeted application. The single phase components (Gd0.6Ca0.4)2CoOx and GdCoO3 were found both to be active. At 850°C the permeation rates exhibited by the composite material was close to 5·10-11 mol/cm2s (air/He gradient, thickness membrane 0.89 mm). This is considered too low for real application of this material. Oxygen permeation rates through perovskite-structured GdCoO3-d and (Gd0.6Ca0.4)2CoO4‑d, having the K2NiF4-structure, were found to be below the range of detection (<10-11 mol/cm2s). In general it can be concluded that the three phase material with overal composition Gd0.7Ca0.3CoOx shows interesting catalytic properties but the ionic conductivity is too low for direct application.
Dual phase composite membranes are the subject of investigation presented in Chapters 4-7. The main advantage of these type of materials is that their properties can be tailored to meet the demands imposed by the sensor design.
To improve the ionic conductivity of Gd0.7Ca0.3CoO3-δ (GCC), in Chapter 4, an ionic conductor, Ce0.8Gd0.2O1.9 (CGO), was added to form a mixed conducting composite. In total three compositions with 43, 60 and 75 vol% CGO were studied. After sintering, as a result of solid state reacions, three different phases were found in the composites. The gadolinium content of the CGO phase increased, whereas GCC transformed into a phase with a K2NiF4-structure. As third phase CoO was found. The composites were characterised by having a high electronic conductivity. The oxygen ion conductivity and permeation was measured in a temperature range of 650 to 750°C and 850 to 1000°C, respectively. The extrapolated data showed good agreement. Typical values of the ion conductivity for the composite with 75 vol% CGO are 4.0·10-4 S/cm at 650°C increasing to 4.2·10-2 S/cm at 1000°C. Measurement of the catalytic activity by means of TPR showed that decomposition of NO to oxygen and nitrogen as well as extensive formation of N2O did not occur. It is concluded that these composites are promising candidate membranes for the proposed sensor application.
In Chapter 5, ZrO2/In2O3 composites were evaluated. In total 6 compositions with different InO1.5 contents were prepared using a wet-chemical precipitation technique. SEM-EDX and XRD revealed that after sintering at 1500°C the materials with 30 and 100 mol% InO1.5 were single phase. All other compositions consisted of mixtures thereof. Whereas it is known from literature that pure InO1.5 is a good electron conductor, the structure of the 30 mol% InO1.5 containing material corresponded to the structure found for yttria or calcia stabilised zirconia, which are known to be good ionic conductors. Highest ionic conductivity of 2.5·10-4 S/cm was found for the composite with equal mole fractions of ZrO2 and InO1.5. Composites with an InO1.5 content between 50 and 70 mol% showed appreciable mixed conductivity. Preliminary catalytic experiments showed minimal catalytic reduction of NO over In2O3. Considering the ionic conductivity these composites are interesting candidates for the NOx-sensor. However, it is necessary to collect more data about the catalytic activity. Furthermore, the oxygen permeation must be measured.
Composites in the ZrO2/In2O3/SnO2 system are investigated for their potential use in Chapter 6. In order to determine the optimum component ratio a series of eight compositions have been prepared along the ZrO2-In0.91Sn0.09O1.5+d tie line. The In/Sn ratio has been taken equal to the most used ITO composition (10 w% SnO2 in In2O3). The materials have been prepared analogue to the method used in Chapter 5 for the ZrO2/In2O3 composites. XRD-analysis showed the existence of a new single phase with an overall composition of In3.85Zr2.80Sn0.35O12 (60 mol% ITO in ZrO2) and a rhombohedral structure with cell parameters a = 9.516(4) Å and c = 8.8975(4) Å (hexagonal setting). This new phase was present in all compositions with an ITO content above 30 mol%. The compositions with an ITO content up to 70 mol% were predominantly ionically conducting. Between 80 – 100 mol% the composites were predominantly electronically conducting. Due to the low ionic conductivity measured for In3.85Zr2.80Sn0.35O12 none of the prepared composites is expected to show appreciable mixed conductivity.
In Chapter 7, composites YSZ/Au and CGO/Au were successfully prepared using a sol gel and a citrate complexing method, respectively. Densification was performed using sinterforging. Percolative behaviour of the gold phase is observed at 8 and 20 vol% gold in YSZ and CGO, respectively. The different percolation thresholds are attributed to a different wetting behaviour of gold in both composites. Unfortunately, the ionic conductivity could not be measured due to the low density of the samples after sinterforging.
Finally, in Chapter 8, a summary of the results presented in the preceding chapters is given. For GCC/CGO composites some complementary measurements, which fall outside the scope of the thesis, were presented. Main problem for real application of this material will be the presence of sulphur. Finally, some ideas for further research are presented.