Research on Application Characteristics of Zirconia-Based High-Temperature NOx Sensors

Abstract

The zirconia solid electrolyte SOFC (solid oxide fuel cell) has the characteristics of oxygen ion conduction function, high-temperature resistance, thermoelectric coupling effect, etc. A NOx sensor based on zirconia solid electrolyte has common characteristics and problems with the SOFC in principle and application. The research objective of this paper is to solve the application problems of smart NOx sensors in diesel vehicles or gasoline vehicles. Improvements in the application performance of the NOx sensor can help the NOx emissions of gasoline vehicles or diesel vehicles better meet the requirements of emission regulations. The smart NOx sensor is a regulatory sensor required by vehicles for China’s Phase VI Vehicle Exhaust Emission Regulations or Euro Phase VI Vehicle Exhaust Emission Regulations. The smart NOx sensor is a key sensor device for improving fuel efficiency and reducing pollution. Moreover, its measurement performance includes dynamic immunity to interference, response speed, and measurement accuracy, which are key factors affecting vehicle emissions. This paper focuses on the impact of the physical structure, electrode characteristics, and control strategies of the sensor on its performance during the application. An excellent sensor structure, electrode structure, and control strategy are given based on application analysis and experimental testing. The results show that the application performance of this smart NOx sensor meets the requirements of exhaust aftertreatment systems.

1. Introduction

Fossil fuel combustion is still the main energy conversion and utilization mode in China, which has low conversion efficiency and serious emission pollution. In order to achieve energy conservation and environmental protection, it is necessary to transform to a clean, low-carbon, safe, and efficient energy structure and improve the efficiency of energy conversion. At same time, it is necessary to reduce the environmental pollution caused by various products such as CO, CO₂, and NO in the combustion process. SOFCs have the advantages of high energy efficiency and low emissions because they can directly convert chemical energy from fossil fuels, biomass fuels, or other hydrocarbon fuels into electrical energy [1,2]. The core principle is the use of a special battery induction principle. The solid zirconia has the characteristics of oxygen ion conduction at high temperatures of 600~800 °C [3]. When the oxygen concentration at the electrodes on both sides of the solid electrolyte layer is different, the oxygen is catalytically ionized. A redox reaction occurs on the electrode surface [4]. At the same time, when a certain strength of voltage is applied to both ends of the electrode, the oxygen ion penetration process occurs in the electrolyte layer. Then, a stable current signal is formed on the external circuit. This process presents an electrolytic cell effect. SOFC power generation is based on the electrical characteristics of solid zirconia matrix cells [5]. A physical diffusion barrier layer or diffusion gap is set on the gas diffusion direction, which can effectively limit the diffusion speed of the gas. Thus, the oxygen at both ends of the solid electrolyte electrode is limited, and the oxygen ion rate formed by the catalysis is limited. Finally, the current formed in the external electrical loop reaches the limit saturation state. Zirconia solid electrolyte is the material basis for SOFC and NOx sensors, and the core principles of both are the same [6,7].

With the continuous escalation of vehicle exhaust emission standards, the internal combustion engine industry has been unable to meet the requirements of energy saving and emission reduction at the same time. In order to save energy, it is necessary to improve fuel utilization, but the nitrogen oxide is generated under high temperature and oxygen-enriched conditions [8,9]. A higher-accuracy NOx and oxygen measurement is required to meet control and OBD requirements for SCR (selective catalytic reduction) systems of the diesel aftertreatment system [10,11,12]. Currently, people pay more attention to the pollution of NOx to the environment and the harm of NOx to human health. The NOx concentration detection and emission control technology has become the most advanced and popular research field in the world.

In terms of materials, NOx sensors are mainly divided into two categories: zirconia-based solid electrolyte and metal oxide semiconductor. Metal oxide semiconductor materials are not suitable for on-board complex environment application, due to poor stability and short service life. At present, the only mature and commercialized vehicle NOx sensor is a current type based on the limit current working principle of zirconia solid electrolyte material. This NOx sensor relies on three pumping cells, which can measure not only the NOx concentration but also the air–fuel ratio in the exhaust gas. The core ceramic element of the sensor is prepared by tape casting, screen printing, and high-temperature sintering to form a zirconia solid electrolyte [13]. The NOx sensor structure consists of three measuring chambers and three oxygen pumping cells. The three pumping cells are, respectively, composed of electrodes with two kinds of materials, which are Pt-Au and Pt-Rh alloys. Au is insensitive to NOx, and a Pt-Au alloy electrode is used to pump excess oxygen out of the chamber, but does not affect the decomposition of NOx. Rh has strong reducibility, which is very sensitive to NOx; thus, the Pt-Rh alloy electrode can be used to catalyze the reduction of NOx [13,14]. The decomposed oxygen is pumped out by the corresponding pumping cell; thus, the oxygen concentration can be detected. In applications, NOx sensor ceramic elements need to be matched with an external control and detection circuits to achieve measurement functions. The NOx working process is shown in Figure 1.

The key performance requirements of the NOx sensor include cold start speed, dynamic response speed, and anti-interference ability, which all need to be considered when the NOx sensor is applied on vehicles [12]. However, there is a strong coupling effect between the physical structure of the three chambers, and there is also a strong coupling effect between the three oxygen pumping cells. Thus, the closed-loop control performance of the three oxygen pumping cells is the key to the performance of the above three aspects [15]. In the research of NOx sensors, most of the current research work focuses on structural design, material application, and the electrode electrochemical mechanism model, such as electrode materials [16,17,18], electrolyte properties [15,19,20,21,22], cell voltages [23,24], and mechanism models [19,25,26,27,28,29].

Auckenthaler et al. studied the adsorption, dissociation, and mutual reaction of each gas component on the electrode surface, and then established an oxygen sensor model based on the particle occupancy of the gas adhesion points on the electrode surface [30]. Mitterdorfer et al. studied in-depth the adsorption of gaseous oxygen molecules on the metal electrode surface, the diffusion of adsorbed atomic oxygen on the electrode surface, and the charge transfer reaction of atomic oxygen at the three-phase interface. Furthermore, they constructed the micro-kinetic model of the electrode and electrolyte system [31]. The research of Velle et al. and Okamotoa et al. showed that when the temperature T > 800 °C, the whole reaction was controlled by charge transfer. It was found that the diffusion of atomic oxygen on the electrode surface was considered to be the reaction rate-determining step when 600 °C < T < 750 °C. However, it is believed that the dissociative adsorption of oxygen molecules on the metal electrode dominates while T < 500 °C [32,33]. The activities of Pt-YSZ, Pt-Au-YSZ, and Rh-Pt-Au-YSZ thick-film electrodes in O₂, N₂ and NO, N₂ gas mixtures at high temperatures were investigated by electrochemical impedance spectroscopy (EIS) and linear sweep voltammetry (LSV) by Schmidt-Zhang et al., and the experiments results indicated that the Rh-Pt-Au mixture is a suitable working electrode material for amperometric NOx sensors based on the solid electrolyte (YSZ) [16]. Tadashi et al. studied NOx in the decomposition mechanism on the Pt and Pt-Au electrodes, and the conclusion was reached that an increase in the activation energy for NO dissociation enables the preferential oxygen pumping on the Pt-Au electrode [19]. Brailsford et al. studied the physical and chemical processes involved in the working process of the sensor in detail. A first-principles model of the zirconia oxygen sensor was also established [15]. Yusuke et al. designed a new NOx sensor to achieve high accuracy and a quick light-off time with an additional electrode placed next to the NOx detection electrode to eliminate the impact of residual O₂ and conducted electrons [2]. Hisashi et al. developed a higher-accuracy NOx sensor by optimizing the sensing element temperature control and reducing the leakage current to the NOx pumping electrode [12].

None of the above studies have in-depth analysis and research on the dynamic application of NOx sensors under different engine load conditions in vehicles. For example, during acceleration transients, the oxygen concentration in the exhaust gas changes sharply, which interferes with the NOx sensor steady-state measurements and results in a decrease in response speed and measurement accuracy. At the same time, it is important to improve the accuracy of NOx measurement under the conditions of backduging and extreme working conditions, in which the oxygen concentration is less than 0%.

In our previous research work, a comprehensive electrochemical model including concentration overpotential, activation overpotential, and ohmic overpotential was developed for the main oxygen pumping cell of the sensor. A diffusion model was also developed to describe the change in gas concentration. Then, the electrochemical model and the diffusion model were combined through the limit current. One important feature of this combined model is that the model eliminates the intermediate variable (gas concentration) and only shows the relationship between the pumping voltage and the pumping current. In order to validate the model at different gas concentrations, the model results were compared with experimental measurements obtained on a designed sensor test rig. Furthermore, the overpotentials of the cathode and the anode were compared to determine the dominant electrode. The changes in the internal characteristics of the sensor during operation were analyzed, which will provide guidance for the design and control of the sensor [29]. In this paper, the dynamic problems of NOx sensors in the application of the aftertreatment system for China’s phase VI emission regulations are studied. In addition, combining the structure of the sensor and the characteristics of electrode materials, the control process parameters and control algorithms are studied to improve the dynamic measurement accuracy and response speed of NOx on vehicle applications.

2. The Working Principle

As shown in Figure 2, the NOx sensor structure consists of three chambers, three pumping cells, a heater, and five signal electrodes [2]. Figure 3 shows a control logic block diagram. At the working state, the zirconia element is heated to around 780 °C. When the exhaust gas enters the first chamber, the CO, H₂, and unburned hydrocarbon in the exhaust gas are oxidized. Moreover, some of the NO is oxidized to NO₂. In addition, the oxygen in the first chamber is pumped out by the first pumping cell with the application of pump voltage Vp₀, so the oxygen concentration in the first chamber is controlled to a low ppm value. Then, the exhaust gas further diffuses into the second chamber, in which the oxygen is pumped out further by the second pumping cell with the application of voltage Vp₁. The oxygen concentration in the second chamber reaches a very low value without NO to decompose. Finally, the NO in the exhaust gas further diffuses into the third measuring chamber, where the NO is catalytically decomposed by the Rh-Pt electrode to form N₂ and O₂. The limit pumping current Ip₂ of oxygen is formed when the oxygen decomposed from NO is pumped out through the measuring pumping cell, which has a linear relationship with NO. All three oxygen pumping cells have the function of pumping oxygen in both directions [34,35,36]. The

Table 1. Symbol description.

Symbol	Description
P+	Common external electrode of pump cell
P−	Internal electrode of main pump cell
M1	Internal electrode of auxiliary pump cell
M2	Internal electrode of measuring pump cell
Ref	reference electrode
Ip0	Pumping current of main pump cell
Ip1	Pumping current of auxiliary pump cell
Ip2	Pumping current of measuring pump cell
C0, C1, C2	Three chambers
Vp0	Pumping voltage of main pump cell between P+ and P−
Vp1	Pumping voltage of auxiliary pump cell between P+ and M1
Vp2	Pumping voltage of measuring pump cell between P+ and M2
V0	Nernst voltage in the first chamber between Ref and P−
V1	Nernst voltage in the second chamber between Ref and M1
V2	Nernst voltage in the third chamber between Ref and M2
D0, D1, D2	Three diffusion channels

is the definition of the structure symbol of the ceramic element.

The NOx sensor ceramic element is assembled in structural parts to meet the application requirements such as heat dissipation, insulation, sealing, waterproofing, and diversion. The NOx sensor ceramic element needs to match the external drive control to achieve the detection and output of NOx concentration. The external circuit consists of feedback heating control based on a three-wire heater or internal resistance, closed-loop control of pumping cells, and pumping current detection circuits. Three closed control loops of pumping cells are coupled to each other because the chambers and diffusion gaps are structurally coupled. The three pump oxygen cells have the Ip₁, V₁, and V₂ set points, which are the inputs of the control target variables, while the control variables are Vp₀, Vp₁, and Vp₂. The set points of Ip₁, V₁, and V₂ affect the response speed and accuracy of NOx dynamic and static measurements. The Ip₁ set point indicates the diffusion flow of the oxygen from the first chamber to the second chamber, and the value of Ip₁ is associated with the oxygen concentration of the first chamber. The set point of V₁ determines the oxygen concentration in the second chamber under steady-state conditions, which also affects the decomposition state of the NO₂ in the second chamber. The V₂ set point must be greater than the set point of V₁ to ensure that oxygen does not diffuse in reverse, and the set point of V₂ needs to ensure that the measuring pump reaches the limit state to promote NO decomposition on the measuring electrode.

3. Problem Description in Vehicle Application

To meet China’s Phase VI Vehicle Exhaust Emission Regulation, the exhaust gas aftertreatment system needs to install two NOx sensors, namely the pre-NOx sensor and after NOx sensor. As shown in Figure 4, the pre-NOx sensor is installed at the EGR inlet or DPF outlet, while the after-NOx sensor is installed at the outlet of the entire exhaust gas aftertreatment system [2]. Because they are installed in different locations, the functions of these two sensors are also different. The output of the pre-NOx sensor is used as a key input parameter for the execution of the urea injection strategy. The output of the after-NOx sensor is used for real-time monitoring of exhaust emissions and OBD diagnostics. The atmosphere at the installation location of the pre-NOx sensor is very complicated, and the oxygen concentration changes drastically. When the DPF is regenerated or the acceleration displacement changes sharply, there is a large disturbance of Ip₂ current output, which makes the closed-loop control of the pumping cell out of the stable range in the transient process. The magnitude of the interference depends on the severity of the change in oxygen concentration, as shown in Figure 5. On the one hand, the oxygen concentration first increases and then decreases, which results in an increase in the oxygen concentration in the first chamber when the throttle opening increases. Then, the pumping current Ip₁ and the Nernst voltages V₁ and V₂ are out of the stable range with the change in oxygen concentration in the second and third chambers, which result in fluctuations in pumping current Ip₂ corresponding to NOx concentration. Large deviations in NOx measurement affect the accuracy of the urea injection and result in the exhaust emission that does not meet the regulatory requirements. On the other hand, excessive fuel injection leads to insufficient combustion, producing NOx and reducing substances. When the exhaust gas enters the first chamber, those reducing and oxidizing substances react on the catalytic electrodes, resulting in a NOx concentration measurement value lower than the actual value. However, the reduction reaction of NOx under this operating condition can be effectively suppressed when the oxygen concentration is high in the first chamber. In the pump control loop, the value of Ip₁ can be set to a larger value, which means that the oxygen concentration in the first chamber is high. It is also found that the set points of Ip₁, V₁, and V₂ have great impact on the amplitude of transient interference. Therefore, it is necessary to study the influence of the steady-state set points of Ip₁, V₁, and V₂ on the dynamic response speed and measurement accuracy.

4. Theoretical Analysis and Experiments

4.1. Theoretical Analysis

In a steady state, it is assumed that the oxygen concentration of exhaust gas is $c_{O_2,env}$, the first chamber oxygen concentration is $c_{O_2,fc}$, and the second chamber oxygen concentration is $c_{O_2,sc}$. The relationship between the pumping voltages Vp₀, Vp₁ and Nernst voltages V₀, V₁ is as follows.

$$V{p}_0=E_p^0+\frac{R{T}_{cell}}{nF}\ln \frac{P_{O_2,env}}{P_{O_2,fc}}+{\eta }_1,V_{P+,P-}$$

(1)

$$V_0=E_N^0+\frac{R{T}_{cell}}{nF}\ln \frac{P_{O_2,ref}}{P_{O_2,fc}},V_{Ref,P-}$$

(2)

$$V{p}_0=V_0+\frac{R{T}_{cell}}{nF}\ln \frac{P_{O_2,env}}{P_{O_2,ref}}+{\eta }_1,V_{P+,P-}$$

(3)

$$\Delta V=V{p}_0-V_0=\frac{R{T}_{cell}}{nF}In\left(\frac{P_{env}}{P_{ref}}\right)+{\eta }_1$$

(4)

$$V{p}_1=E_{p1}^0+\frac{R{T}_{cell}}{nF}\ln \frac{c_{O_2,env}}{c_{O_2,sc}}+{\eta }_2,V_{P+,M1}$$

(5)

$$V_1=E_{n1}^0+\frac{R{T}_{cell}}{nF}\ln \frac{c_{O_2,ref}}{c_{O_2,sc}},V_{Ref,M1}$$

(6)

$$V{p}_1=V_1+\frac{R{T}_{cell}}{nF}\ln \frac{P_{O_2,env}}{P_{O_2,ref}}+{\eta }_1,V_{P+,M1}$$

(7)

where $P_{O_2,env}$ represents the oxygen partial pressure of the exhaust gas; $P_{O_2,fc}$, $P_{O_2,sc}$, $P_{ref}$, respectively, represent the oxygen partial pressure of the first, second, and reference chambers, and ${\eta }_1, {\eta }_2 \mathrm{represent}$ the overpotential of the electrode.

The relationship between the oxygen diffusion flow and the oxygen partial pressure in the chambers is as follows.

$$\mathrm{N}=\frac{D_1{S}_1}{RT{L}_1}\left(P_{O_2,env}-P_{O_2,fc}\right)$$

(8)

$${\mathrm{N}}_2=\frac{D_2{S}_2}{RT{L}_2}\left(P_{O_2,fc}-P_{O_2,sc}\right)$$

(9)

$${\mathrm{N}}_1=\mathrm{N}-{\mathrm{N}}_2$$

(10)

where N represents the flow of oxygen diffused into the first chamber, ${\mathrm{N}}_1$ represents the oxygen flow pumped out from the first chamber, and ${\mathrm{N}}_2$ represents the oxygen diffused into the second chamber.

According on the relationship between the pumping current and the oxygen flow, there is

$$\begin{array}{cc}I{p}_0=4F{\mathrm{N}}_1=\hfill & \frac{4F{D}_1{S}_1\left(P_{O_2,env}-P_{O_2,fc}\right)}{R{T}_{cell}{L}_1}-\frac{4F{D}_2{S}_2\left(P_{O_2,fc}-P_{O_2,sc}\right)}{R{T}_{cell}{L}_2}\hfill \\ & \hspace{1em}=\frac{4F{D}_1{S}_1\left(P_{O_2,env}-P_{O_2,fc}\right)}{R{T}_{cell}{L}_1}-I{p}_1\hfill \end{array}$$

(11)

$$I{p}_1=4F{\mathrm{N}}_2=\frac{4F{D}_2{S}_2\left(P_{O_2,fc}-P_{O_2,sc}\right)}{R{T}_{cell}{L}_2}$$

(12)

The relationship between concentration and pressure is known as $=\frac{P}{RT}$; therefore

$$I{p}_0=\frac{4F{D}_1{S}_1\left(c_{O_2,env}-c_{O_2,fc}\right)}{L_1}-I{p}_1$$

(13)

$$c_{O_2,fc}=c_{O_2,env}-\frac{L_1(I{p}_0+I{p}_1)}{4F{D}_1{S}_1}$$

(14)

$$I{p}_1=\frac{4F{D}_2{S}_2\left(c_{O_2,fc}-c_{O_2,sc}\right)}{L_2}$$

(15)

Then,

$$P_{O_2,fc}=P_{O_2,ref}{e}^{(E_N^0-V_0)\frac{nF}{R{T}_{cell}}}$$

(16)

$$c_{O_2,fc}=P_{O_2,fc}/R{T}_{cell}$$

(17)

$$P_{O_2,env}-\frac{L_1(I{p}_0+I{p}_1)}{4F{D}_1{S}_1}\ast R{T}_{cell}=P_{O_2,ref}{e}^{(E_N^0-V_0)\frac{nF}{R{T}_{cell}}}$$

(18)

$$P_{O_2,env}-\frac{L_1(I{p}_0+I{p}_1)}{4F{D}_1{S}_1}\ast R{T}_{cell}=P_{O_2,ref}\ast e^{E_N^0}\ast e^{-V_0\ast P_{O_2,ref}\ast \frac{nF}{R{T}_{cell}}}$$

(19)

It is found that the value of Ip₁ is exponentially related to V₀ under steady-state conditions when the external oxygen concentration $c_{O_2,env}$ remains unchanged and Ip₀ remains unchanged in the limiting current state.

It is found that Vp₀ and V₀ show a linear relationship from Formula (3) when the external oxygen concentration $c_{O_2,env}$ remains unchanged.

It is found that Vp₀ and Ip₁ show a logarithmic relationship from Formulas (3), (18) and (19) when the external oxygen concentration $c_{O_2,env}$ remains unchanged.

4.2. Test Equipment and Test Methods in Laboratory

The test equipment included NOx synthetic gas equipment, a propane burner stand apparatus, and a diesel pickup truck, as shown in Figure 6, Figure 7 and Figure 8. In the NOx synthetic gas equipment, the N₂, O₂, NO, and gas mixture (H₂, CO, CO₂ in N₂) can be allocated to simulate the changes in oxygen concentration and NOx concentration. The static output characteristics of the NOx sensor were studied on the NOx synthetic gas equipment. A propane burner stand apparatus was used to simulate the gasoline combustion process with propane. Some NO could be added to the exhaust gas. The influence of the oxygen concentration in the first chamber and the activity of the measuring electrode in the first chamber on the measurement accuracy of the NOx sensor were studied on the propane burner stand apparatus. In addition, the driving dynamic characteristics of the NOx sensor were studied on the diesel pickup truck, which had an exhaust aftertreatment system consisting of DOC, DPF, SCR, and ASC. Two NOx sensors were installed on this system. The emission of the diesel pickup truck vehicle complied with the China’s phase VI vehicle emission regulations. Some application characteristics of the NOx sensor were studied and tested on the customer’s engine test bench and revolving drum test bench, as shown in Figure 9.

4.3. Test Results

The NOx sensor was in normal operation, and the oxygen was set to 21% in vol. The V₀ output was recorded with the change in the Ip₁ set point in the pump closed control loop, as shown in Figure 10.

The NOx sensor was in normal operation, and the oxygen was set to 21% in vol. From Figure 11, it can be seen that the output value of V₀ increased with Vp₀, which showed a linear relationship.

As can be seen in Figure 12, the output value of Ip₁ decreased with Vp₀, which showed a logarithmic relationship when the Ip₁ and Vp₀ operated in an open-loop state.

On the one hand, the oxygen concentration in the first chamber depended on the input setpoint of the Vp₀ and Ip₁ closed control loop. Four group tests were conducted on the diesel pickup truck with different Ip₁ set values, and the input of exhaust oxygen concentration and the outputs of V₀, Ip₁, and Ip₂ (corresponds to NOx concentration) were recorded by the data acquisition device, as shown in Figure 13. It was found that the partial pressure of the external oxygen concentration decreased when the throttle opening changed from small to large. The output steady-state value of V₀ was small and the oxygen concentration in the first chamber was in a lean state when Ip₁ was set to a larger value. Then, the output fluctuation amplitudes of Ip₁ and V₀ were relatively small when the external oxygen concentration changed by the same magnitude. Thus, the oxygen concentration of the second chamber could be quickly adjusted to a stable state, which reduced the interference with the oxygen concentration of the third chamber. In conclusion, the changes in external oxygen concentration had little interference with Ip₂ (corresponds to NOx concentration), the process of Ip₂ recovering from dynamic to steady state was smoother, and the accuracy of the NOx measurement and response speed was improved, as shown in Figure 13d.

On the other hand, the oxygen concentration in first chamber also depended on the structural characteristics and electrode characteristics when the Ip₁ set point was identical to each other among several ceramic elements. In order to maintain the oxygen concentration of the first chamber, the Ip₁ input value of the Ip₁ and Vp₀ closed control loop should be set according to the diffusion capacity of the second diffusion channel, which can be seen in

Table 2. Ip₁ set point for different structure sizes of the second diffusion channel.

Second Diffusion Channel Structure Size	Ip1 Set Point	Ip1 Limit Current Value in Air
D1: L = 600 μm, W = 2500 μm, H = 16 μm	35 μA	1.45 mA
D1: L = 600 μm, W = 1500 μm, H = 16 μm	25 μA	1.25 mA
D1: L = 600 μm, W = 800 μm, H = 8 μm	15 μA	1.12 mA

. The Ip₁ set value should decrease with the diffusion capacity of the second diffusion channel. The structure diagram of the diffusion channel and the slice diagram of the ceramic element are shown in Figure 14.

At the same time, the activity of the main pump electrode P- also had influence on the oxygen concentration of the first chamber when the Ip₁ set value was the same as each other among several ceramic elements. In the measuring principle, the electrodes of the main pump electrode P− were mainly used to oxidize CO, CH, and NO to CO₂, H₂O, and NO₂. A reasonable oxygen concentration must be set in the first chamber to ensure that CO and CH are fully oxidized and NO is oxidized in small amounts or none. When the oxygen concentration in the first chamber was insufficient, NOx reacted with CO, CH, etc. so that the measured value of NOx was low.

As shown in Figure 15, two NOx sensors were assembled with two ceramic elements based on two kinds of main pump electrode P−. They were installed on the pipe of the propane burner stand apparatus. The NO was added to the exhaust gas of the propane burner stand apparatus, and the concentration of NO in mixed exhaust gas could reach 500 ppm or other values when the exhaust was under a rich state. The measurement values of NOx sensors were different, and the measurement value of the NOx sensor with high activity of the main pump electrode P− was lower than that with low activity of the main pump electrode P−, as shown in

Table 3. The measurement values of the NOx sensor with high- and low-activity P− electrode (sample1 has high-activity P− electrode, sample2 has low-activity P− electrode).

Num.	λ = Air Fuel Ratio/14.7	Ip0 μA	Ip1 μA	Ip2 μA	O2%	NOx ppm
Sample1	λ = 1.000	−32	20	0.772	0.18	473
Sample2		−32	20	0.778	0.19	475
Reference NOx sensor		-	-	-	0.21	501
Sample1	λ = 0.965	−186	20	0.747	−1.34	453
Sample2		−190	20	0.778	−1.35	475
Reference NOx sensor		-	-	-	−1.29	481
Sample1	λ = 0.935	−315	20	0.52	−2.51	294
Sample2		−311	20	0.721	−2.45	435
Reference NOx sensor		-	-	-	−2.44	458
Sample1	λ = 0.912	−445	20	0.324	−3.69	157
Sample2		−438	20	0.647	−3.65	383
Reference NOx sensor		-	-	-	−3.63	389
Sample1	λ = 0.890	−560	20	0.159	−4.69	58
Sample2		−558	20	0.593	−4.63	345
Reference NOx sensor		-	-	-	−4.58	353

.

In the application, it was also found that the NOx measurement value increased with the oxygen concentration in the first chamber when the NO concentration in the environment was fixed. The NO was excessively oxidized to NO₂ after entering the first chamber when the oxygen concentration in the first chamber was high and the activity of pump electrode P− was high. The NO₂ diffused into the second and third chamber, and more oxygen ions formed on the measurement M2 electrode because there were two oxygen atoms in NO₂. Then, the pumping current increased with the oxygen concentration in the environment, as shown in Figure 16. Thus, it was necessary to set an appropriate activity for the pump electrode P−. In addition, the pumping current output correction algorithm should be introduced, such as the two-dimensional linear interpolation correction algorithm based on oxygen concentration and NOx concentration, as shown in Figure 17 and

Table 4. The calibration parameters of the NOx sensor for oxygen concentration dependence.

		NOx Concentration	0.00 ppm	100.00 ppm	300.00 ppm	800.00 ppm	1500.00 ppm	2000.00 ppm	2900.00 ppm
	Correction Factor
O2 Concentration
0.00%			1.05	1.08	1.11	1.10	1.12	1.13	1.35
1.00%			1.04	1.05	1.09	1.10	1.10	1.11	1.31
3.00%			1.03	1.04	1.06	1.08	1.08	1.09	1.27
5.00%			1.02	1.03	1.05	1.06	1.06	1.07	1.21
7.00%			1.00	1.00	1.04	1.04	1.04	1.05	1.08
12.50%			1.00	0.99	1.00	0.99	0.99	1.00	1.00
20.00%			0.99	0.97	0.94	0.92	0.92	0.92	0.89

. The final NOx measurement value was multiplied by a correction factor. The improvement of the measurement accuracy by the correction algorithm is shown in Figure 18.

4.4. Diesel Engine Bench and Revolving Drum Test

The dynamic test was conducted on the diesel engine bench. The test results showed that the output of the NOx measurement value of the test sample was the same as that of the reference equipment when the input of Ip₁ was set as 25 μA and the activity of the measurement electrode P− was set to low; the test results are shown in Figure 19. Then, the WLTC cycle test was conducted on the revolving drum test bench. The result showed that the NOx emission can meet the GB6b standard of China, which can be seen in Figure 20 and

Table 5. Emission test results.

Test Name	SDPK Test
Test Record Name	WLTC Class3b SDPK Vehicle 20200430 2
Gas composition	CO2	CO	THC	CH4	NMHC	NOX	HC + NOX	NMHC + NOX	N2O
unit	g/km	mg/km	mg/km	mg/km	mg/km	mg/km	mg/km	mg/km	mg/km
Dilute	239.60	154.7	27.7	14.9	18.7	23.7	51.4	42.3339	15.2
Bag	239.46	142.1	26.7	15.3	10.2	22.6	49.2	32.8058	14.8
China GB6b Class 2 _III limit	-	740	80	-	-	50	-	55	3

. At the same time, the dynamic output of the test sample could track the output of the reference equipment under the driving test, and there was no disturbance on the Ip₂ current output and NOx concentration output with an oxygen change of different magnitudes, as shown in Figure 21. The measurement accuracy and response speed of the sensor could meet the requirements of application on a vehicle.

5. Conclusions

This paper analyzes the effects of electrode activity, ceramic element structure size, and input parameters of multi-chamber closed-loop control on the accuracy and response speed characteristics of a NOx sensor in practical applications based on the application analysis and experiment. The following four conclusions can be drawn and verified in the application.

First, when the Ip₁ value is set the same, the oxygen concentration in the first chamber increases as the P− electrode activity decreases. When the P− electrode is set as a low-activity electrode, the oxygen concentration in the first chamber is high, which helps to improve the measurement accuracy of the NOx sensor in rich exhaust gas.

Second, when the Ip₁ value is set the same, the oxygen concentration in the first chamber increases as the diffusion capacity of the second diffusion channel decreases.

The value of the Ip₁ set point in the closed-loop control of Vp₀ and Ip₁ affects the response speed of the dynamic output of the Ip₂ and NOx concentration. Setting Ip₁ too small results in a low oxygen concentration in the first chamber. Then, the anti-interference ability of the Vp₀ and Ip₁ closed-loop control system decreases, resulting in a reduction in the anti-interference ability and measurement accuracy of the NOx sensor. Setting Ip₁ too large increases the oxygen concentration dependence of the NOx measurement and reduces the measurement accuracy of the NOx sensor. The activity of measurement of the P− electrode and the diffusion capacity of the second diffusion channel should be considered when setting an appropriate Ip₁ value for Vp₀ and Ip₁ closed-loop control.

Third, the influence of oxygen concentration dependence needs to be fully considered. A suitable calibration algorithm needs to be adopted to correct the measurement error caused by oxygen concentration dependence when the NOx sensor is applied on a vehicle.

Fourth, the study of the application characteristics of NOx sensors is helpful to the design of high-precision NOx sensors, which is a critical requirement in an exhaust aftertreatment system and is expected to contribute to emission reduction and global environmental protection.