Effect of the Partial Equilibrium State at the Zirconia Surface in a Cr/Cr₂O₃-Type Zirconia Oxygen Sensor on Continuous Measurement of the Oxygen Content in Molten Steel

Abstract

Zirconia oxygen sensors are used in the steel-refining process to instantly measure the oxygen concentration in molten steel. It is desirable to develop sensors that are capable of continuous and long-time measurements, but such measurements are difficult with conventional sensors. In this study, the factors that hinder long-time measurement by a single zirconia oxygen sensor with a mixture of Cr and Cr₂O₃ as the reference electrode were investigated. Continuous measurements of a molten iron containing carbon and aluminum for deoxidization were carried out with the sensor. The electromotive force between the reference electrode (positive) and sample electrode (negative) of the sensor decreased from a positive value to zero with time. This was because reduction of chromium oxide at the surface of the zirconia solid electrolyte was promoted owing to oxygen diffusion from the inside to the outside of the zirconia tube and the equilibrium in the reference electrode was disturbed during the measurements. In addition, the reference electrode at the surface partially and temporarily melted owing to diffusion of the components from the electrolyte to the reference electrode. Additionally, a test to extend the lifetime of the sensor was performed with application of a direct electrical current to the sensor. The electromotive force with application of a current was maintained at a higher value than that without application of a current owing to promotion of re-oxidation at the zirconia surface. However, it was found that a high current can cause over-oxidation and dissolution of the reference electrode at the surface.

1. Introduction

In the smelting process, refining process, and manufacturing process of practical metal materials, such as iron and copper alloys, it is important to measure the composition of the molten metal during each process because the composition of the metal controls the physical and chemical properties. In the steel-making process, the oxygen potential in the steel greatly changes by the reaction with the added ingredients in each stage of refinement.^1,2,3,4) Thus, the oxygen concentration is an essential factor that indicates the degree of the adjusting components in steel. To optimize the oxygen concentration in molten steel during the processes, the oxygen concentration needs to be instantly measured with high accuracy.

Zirconia oxygen sensors are typically used to measure the oxygen concentration in molten metals because they can directly measure the oxygen concentration without sampling. In the steel-making process, it is possible to measure the oxygen concentration or the concentrations of other elements in the molten steel. However, these sensors can only be used for a short period of time otherwise the process takes 30 min or more, and many sensors are used as consumable sensors. The capability for continuous use of the zirconia oxygen sensor in the actual steel-making process has been proposed in several review papers.^5,6,7,8) The use of a single zirconia oxygen sensor with a Cr/Cr₂O₃ reference electrode is limited to approximately 10 s, and there is no zirconia oxygen sensor capable of continuous measurement of the processing time of refined steel. That is, sensors with a long lifetime need to be developed. Long-time measurement would ensure the product quality, the process efficiency, and energy saving.

It is necessary to explore the phenomena during measurement by zirconia sensors toward realization of continuous measurement of the oxygen concentration in molten steel. Many factors have been assumed to be the reasons for inaccurate measurement by zirconia oxygen sensors, such as the instability of the reference electrode, corrosion of the solid electrolyte to the slag, and polarization in the solid electrolyte between the sample and reference electrodes,^9,10) and tests to remove these inhibitors have been carried out for long-time measurement. As one inhibiting factor, we found that a metallic molybdenum layer forms on the zirconia surface in a Mo/MoO₂-type zirconia oxygen sensor by reduction of gaseous Mo oxide molecules adsorbed on the zirconia surface to metallic Mo, which causes a decrease of the electromotive force (EMF) as the measurement value.¹¹⁾ This metallic molybdenum phase is in a different equilibrium state from the initial state in the reference electrode. In addition, by application of a direct-current (DC) voltage, the metallic molybdenum layer can be removed owing to re-oxidation of molybdenum and the EMF recovers to an adequate value. According to these results, it is considered that the partial equilibrium state at the zirconia surface has a large influence on operation of the zirconia sensor.

To develop a method based on a zirconia oxygen sensor for continuous measurement of the oxygen concentration in molten steel for up to 30 min, in the present study, we attempted to clarify the main factor that disturbs the measurement by a zirconia oxygen sensor with a Cr/Cr₂O₃ reference electrode. In addition, the behavior of the zirconia sensor with an applied DC to satisfy the higher electric potential at the reference electrode than in the molten steel was observed.

2. Experimental Method

2.1. Structure of the Zirconia Oxygen Sensor

A zirconia oxygen sensor is an electrochemical sensor based on a zirconia solid electrolyte stabilized by dopants for measurement of the oxygen concentration in high-temperature molten metals, slags, or atmospheres. Zirconia solid electrolytes are generally doped with MgO, CaO, and Y₂O₃ to stabilize zirconia in a crystal structure of the tetragonal or cubic system over a wide temperature range, and they have high oxygen-ion conductivity owing to the oxygen vacancies in the structure produced by the dopants. Construction of the solid-electrolyte oxygen cell is different for different practical conditions.¹²⁾ The zirconia oxygen sensor used in this study was composed of two electrodes. The first electrode was a reference electrode composed of a mixture of a metal and a metal oxide to adjust the oxygen potential, and the metal wire was inserted into a single-sealed tube of zirconia solid electrolyte. The second electrode was a sample electrode composed of a metal wire immersed in the sample, such as molten metal. When the oxygen concentration in molten steel is measured, both of the electrodes are dipped in the molten steel and an EMF (E) is generated between the electrodes in the measurement. According to Wagner theory, the current flow of charge carriers in an electrolyte is regarded as the amount of electricity carried by the gradient of the electrochemical potential of the carrier. Assuming that the charge carriers are oxygen ions, electrons, and holes, the EMF generated between electrolytes can be expressed by the following relationships:¹³⁾   

$$\begin{array}{c}E=\frac{1}{4F}\int t_{\mathrm{i}\mathrm{o}\mathrm{n}}\mathrm{d}{\mu }_{{\mathrm{O}}_2};\end{array}$$(1)

  

$$\begin{array}{c}{t}_{\mathrm{i}\mathrm{o}\mathrm{n}}=\frac{{\sigma }_{\mathrm{i}\mathrm{o}\mathrm{n}}}{{\sigma }_{\mathrm{i}\mathrm{o}\mathrm{n}}+{\sigma }_{\mathrm{e}}+{\sigma }_{\mathrm{h}}}.\end{array}$$(2)

where t_ion is the oxygen ionic transfer number and σ_ion, σ_e, and σ_h are the conductivities of oxygen ions, electrons, and holes, respectively. The zirconia solid electrolyte has an electron conductivity at low oxygen potentials,^14,15,16) where the measurement is generally carried out, while the hole conductivity is negligible in the solid electrolyte. Therefore, the charge compensation occurs because the ion or electron are transferred in the electrolyte. Schmalzried¹⁷⁾ derived the following equation for when the oxygen concentration in molten steel is measured by a zirconia sensor and t_ion in the solid electrolyte is represented by a function of the oxygen pressure, where the reference electrode is the cathode and the sample electrode is the anode:   

$$\begin{array}{c}E=\frac{RT}{F}ln\left\{\frac{{\left(P_{{\mathrm{O}}_2}^{\mathrm{r}\mathrm{e}\mathrm{f}}\right)}^{\frac{1}{4}}+P_{\text{⊖}}{}^{\frac{1}{4}}}{{\left(P_{{\mathrm{O}}_2}^{\mathrm{s}\mathrm{t}\mathrm{e}\mathrm{e}\mathrm{l}}\right)}^{\frac{1}{4}}+P_{\text{⊖}}{}^{\frac{1}{4}}}\right\}.\end{array}$$(3)

where $P_{{\text{O}}_2}^{\text{steel}}$ is the oxygen pressure in molten steel as the sample electrode and $P_{{\text{O}}_2}^{\text{ref}}$ is defined as the oxygen pressure of the reference electrode corresponding to the equilibrium state in which the metal and metal oxide stably exist. $P_{\text{⊖}}$ is defined as the oxygen pressure where the ionic conductivity is equivalent to the electron conductivity of the solid electrolyte. Here, voltage depression in the molten iron and the metal wire of the sample electrode are negligible owing to the very low electrical resistance of molten iron¹⁸⁾ and extremely low current conduction in the oxygen sensor. Hence, it can be assumed that the measured EMF is approximately expressed by Eq. (3). Measurement by the zirconia oxygen sensor is based on the EMF arising from the oxygen-potential difference between the reference electrode and sample electrode.

For instance, a MgO-doped partially stabilized zirconia (MSZ) solid electrolyte has excellent thermal shock resistance and is widely used in steel-refining processes. The $P_{\text{⊖}}$ values of MgO-doped zirconia with different MgO concentrations have been reported in several review papers.^14,19) As an example, $P_{\text{⊖}}$ of 91 mol% ZrO₂–9 mol% MgO has been reported by Iwase et al.:¹⁴⁾   

$$\begin{array}{c}\mathrm{l}\mathrm{o}\mathrm{g}{P}_{\text{⊖}}/\mathrm{a}\mathrm{t}\mathrm{m}=20.40-6.45\times {10}^4{T}^{-1}/\mathrm{K}(T=1\mathrm{\hspace{0.17em} }273-1\mathrm{\hspace{0.17em} }873\mathrm{\hspace{0.17em} }\mathrm{K})\end{array}$$(4)

It is possible to determine the oxygen pressure in molten steel $P_{{\text{O}}_2}^{\text{steel}}$ from Eq. (3) by measuring the electromotive force E and temperature T. Furthermore, it is possible to calculate the oxygen activity a_O or oxygen concentration in the molten steel [O] equilibrated with $P_{{\text{O}}_2}^{\text{steel}}$ from the standard free-energy change (∆G⁰) of oxygen gas dissolving in the molten iron:²⁰⁾   

$$\begin{array}{c}1/2{\mathrm{O}}_2\left(\mathrm{g}\right)=\left[\mathrm{O}\right] \left(1\mathrm{\hspace{0.17em} }\mathrm{w}\mathrm{t}\%\mathrm{\hspace{0.17em} }\mathrm{i}\mathrm{n}\mathrm{\hspace{0.17em} }\mathrm{F}\mathrm{e}\right);\end{array}$$(5)

  

$$\begin{array}{c}\mathrm{\Delta }{G}^0/\mathrm{J}\cdot \mathrm{m}\mathrm{o}{\mathrm{l}}^{-1}=-117\mathrm{\hspace{0.17em} }300-2.89T/\mathrm{K}.\end{array}$$(6)

As the reference material, a Mo/MoO₂ or a Cr/Cr₂O₃ mixture is generally used. To prevent excess oxygen diffusion through the solid electrolyte, the difference between $P_{{\text{O}}_2}^{\text{ref}}$ and $P_{{\text{O}}_2}^{\text{steel}}$ should be small. Li et al.¹²⁾ reported that as the reference electrode, a Mo/MoO₂ mixture is suitable for [O] > 300 ppm and a Cr/Cr₂O₃ mixture is suitable for [O] < 100 ppm. Thus, if a zirconia oxygen sensor with a Cr/Cr₂O₃ reference electrode is used, deoxidized iron is suitable as the measurement sample. The chemical equilibrium and Gibbs free-energy change of the Cr/Cr₂O₃ reference electrode are described by²¹⁾   

$$\begin{array}{c}4/3\mathrm{C}\mathrm{r}\left(\mathrm{s}\right)+{\mathrm{O}}_2\left(\mathrm{g}\right)=2/3\mathrm{C}{\mathrm{r}}_2{\mathrm{O}}_3\left(\mathrm{s}\right);\end{array}$$(7)

  

$$\begin{array}{c}\mathrm{\Delta }{G}^0/\mathrm{J}\cdot \mathrm{m}\mathrm{o}{\mathrm{l}}^{-1}=-754\mathrm{\hspace{0.17em} }900+170T/\mathrm{K}(T=298-2\mathrm{\hspace{0.17em} }000\mathrm{\hspace{0.17em} }\mathrm{K}).\end{array}$$(8)

From the above discussion, $P_{{\text{O}}_2}^{\text{ref}}$ is defined as 6.71 × 10⁻¹³ atm at T = 1873 K, and the corresponding oxygen concentration in molten iron equilibrated with $P_{{\text{O}}_2}^{\text{ref}}$ is estimated to be 20 ppm by Eqs. (3), (4), (5), (6), (7), (8) when the Henrian activity coefficient f_O in Eq. (6) is assumed to be unity.

2.2. Cr/Cr₂O₃-type Zirconia Oxygen Sensor

The zirconia oxygen sensors used in this study were made of a mixed powder of Cr with a diameter of 10 μm and Cr₂O₃ with diameter of less than 10 μm as the reference electrode, a single-closure tube (outer diameter (OD) 3 mm × inner diameter (ID) 2 mm × 35 mm) of 8 mol% MSZ as the solid electrolyte, and Mo rods (diameter (φ) 1 mm × 1000 mm for the reference electrode, φ 2 mm × 1000 mm for the sample electrode) as electrode wires to prevent generation of a thermo-EMF between the different metals. The reference material was prepared by mixing Cr metallic powder and Cr oxide powder at a mass ratio of 9:1. The Cr/Cr₂O₃ mixture was filled in a single-closure MSZ tube up to a height of approximately 15 mm, and it was covered with ZrO₂ powder as a filler. A sharpened Mo rod was pushed into the Cr/Cr₂O₃ mixture. The upper part of the single-closure MSZ tube was sealed with an alumina-based ceramic adhesive to prevent penetration of the atmospheric gas during the experiment. In addition, an alumina tube was connected to the upper part of the MSZ tube and fixed to the molybdenum rod inserted in the reference electrode with adhesive to prevent oxidation of the molybdenum during the measurement.

2.3. Continuous EMF Measurements

The molten metal sample was 400 g Fe–C alloy or Fe–C–Al alloy made from electrolytic iron (Atomiron MP, 99.95% purity, Toho Zinc Co., Ltd., Tokyo, Japan, the chemical composition of the impurities in the electrolytic iron is given in Table 1), carbon grains (99.95% purity, Kojundo Chemical Laboratory Co., Ltd., Saitama, Japan), and part of an aluminum wire (99.99% purity, φ1.0 mm). The Fe–C alloy contained approximately 5 mass% carbon, and the Fe–C–Al alloy contained 5 mass% carbon and 0.05 mass% aluminum. We have confirmed the oxygen concentration of the Fe–C alloy adjusted to have a carbon concentration of 0.3 mass% is 20 ppm.¹¹⁾ This approximately corresponds to the equilibrium oxygen potential of the reference electrode, so that the alloys used in this study were more deoxidized with carbon and aluminum than the alloy with 20 ppm oxygen. That is, the oxygen potentials in these alloys are lower than that in the reference electrode, and thus the EMF is a positive value if the reference electrode is the cathode and the sample electrode is the anode.

Table 1. Chemical composition of the impurities in the electrolytic iron sample [ppm].

	O	C	P	S	Si	Mn	Cu	N
Standard	≤ 200	≤ 40	≤ 20	≤ 20	≤ 20	≤ 10	≤ 10	–
Representative	100	20	5	8	< 5	1	1	5

The EMF measurements were performed as follows. An electronic furnace with a SiC heating element and an alumina reaction tube (OD 60 mm × ID 52 mm × 1000 mm) was used. A sealing jacket with a Mo rod as the sample electrode and a thermocouple to measure the temperature in the furnace was attached to the upper part of the reaction tube. A small observation window was prepared in the jacket, which was initially closed by a silicon rubber stopper to seal the system. An alumina Tammann tube (OD 40 mm × ID 34 mm × 150 mm) containing electrolytic iron and carbon grains was installed in the soaking zone of the furnace and a carbon tube (OD 40 mm × ID 34 mm × 50 mm) was placed on the Tammann tube to deoxidize the atmosphere. After placing the materials and carbon tube in the furnace, argon gas (99.999% purity) from the gas inlet of the jacket was flowed in the sealed furnace until the furnace atmosphere was completely replaced with argon gas. The temperature in the furnace was then increased to 1873 K as a set variable. When the EMF measurements were performed for the Fe–C–Al alloy, an aluminum wire wound onto the tip of a carbon rod was immersed in the Fe–C alloy and stirred to adjust the Al concentration to 0.05 mass%.

The continuous EMF measurements were carried out after checking melting of the alloy and the stability of the temperature. First, the zirconia oxygen sensor was inserted in the furnace from the observation window and preheated for 1 min above the molten alloy. The sensor was then immersed in the molten alloy and the EMF produced between the electrodes was measured for over 10 min by a GL220 data logger with insulation resistance of higher than 50 MΩ (Graphtec Co., Ltd., Tokyo, Japan). When the EMF measurement was carried out, the temperature above the molten alloy was measured to be within the range 1873–1913 K. In addition, before and after the continuous measurements, other oxygen sensors made in the same way as the sensor for the continuous measurements were immersed in the molten alloy and the EMF was measured to confirm that the actual oxygen content in the molten alloy was unchanged and to eliminate the influence of molybdenum dissolving in the alloy. At the end of the measurements, the reference electrode of each sensor was removed and cooled in air. The reference electrodes of the sensors after the measurements were cut to observe the cross-section at the inner surface of MSZ or milled for identification by X-ray diffraction (XRD).

2.4. Intermittent DC Application

As mentioned in Section 2.3, the oxygen partial pressure in the reference electrode could be higher than that of the molten alloy during the experiment. Therefore, oxygen was assumed to move from the reference electrode to the molten alloy during the EMF measurements because the zirconia sensor also has the property as an oxygen pump.^22,23) It is considered that this oxygen transfer causes the phenomenon where Eq. (3) is not satisfied. Li et al.⁹⁾ estimated that the polarization at the reference electrode causes performance degradation of the sensor, and they suggested that the lifetime of the oxygen sensor can be extended when the recharging current flows. Furthermore, it has been suggested that application of a current is effective to remove the thin layer produced at the MSZ surface and to recover the measured value of the EMF.^11,24) However, the phenomena at the Cr/Cr₂O₃ reference electrode during the measurement or recharging have not been clarified. Therefore, to explore the conditions where the EMF can be measured with a recharging current, a direct electrical current was applied between the electrodes to satisfy the higher electric potential at the reference electrode than at the sample electrode (molten alloy) during EMF measurement of the molten Fe–C alloy with 5 mass% carbon at 1873–1913 K, and the cross-sectional microstructure of the zirconia surface was observed after the measurement. In this experiment, it was expected that the current would induce oxygen ions to diffuse from the molten iron to the reference electrode through the zirconia solid electrolyte. The reaction expected to proceed on each interface of the zirconia solid electrolyte during the application is as follows.

Cathode reaction on the interface between the molten Fe–C alloy and MSZ:   

$$\begin{array}{c}\left[\mathrm{O}\right]\left(\mathrm{i}\mathrm{n}\mathrm{\hspace{0.17em} }\mathrm{m}\mathrm{o}\mathrm{l}\mathrm{t}\mathrm{e}\mathrm{n}\mathrm{\hspace{0.17em} }\mathrm{a}\mathrm{l}\mathrm{l}\mathrm{o}\mathrm{y}\right)+2{\mathrm{e}}^{-}\to {\mathrm{O}}^{2-}\left(\mathrm{i}\mathrm{n}\mathrm{\hspace{0.17em} }\mathrm{M}\mathrm{S}\mathrm{Z}\right)\end{array}$$(9)

Anode reaction on the interface between MSZ and the reference electrode:   

$$\begin{array}{c}{\mathrm{O}}^{2-}\left(\mathrm{i}\mathrm{n}\mathrm{\hspace{0.17em} }\mathrm{M}\mathrm{S}\mathrm{Z}\right)\to 1/2{\mathrm{O}}_2\left(\mathrm{g},\mathrm{\hspace{0.17em} }\mathrm{i}\mathrm{n}\mathrm{\hspace{0.17em} }\mathrm{t}\mathrm{h}\mathrm{e}\mathrm{\hspace{0.17em} }\mathrm{r}\mathrm{e}\mathrm{f}\mathrm{e}\mathrm{r}\mathrm{e}\mathrm{n}\mathrm{c}\mathrm{e}\mathrm{\hspace{0.17em} }\mathrm{e}\mathrm{l}\mathrm{e}\mathrm{c}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{d}\mathrm{e}\right)+2{\mathrm{e}}^{-}\end{array}$$(10)

Application of the direct electrical current of 0.8, 1.5, or 2.3 A across the reference and sample electrodes for 4 min and EMF measurement for 1 min were alternately repeated several times. This is to say, the EMF was measured every 5 min and the application was paused when measuring the EMF.

3. Results and Discussion

3.1. Continuous EMF Measurements

The results of the EMF measurements with time for the single oxygen sensor continuously immersed in each molten alloy are shown in Fig. 1, where the reference electrode was the cathode and the sample electrode was the anode. The initial values of the EMF were approximately +60 mV for the Fe–C alloy and approximately +120 mV for the Fe–C–Al alloy. If these are regarded as appropriate values, the oxygen concentrations in the Fe–C and Fe–C–Al alloys calculated from the EMF values by Eqs. (3), (4), (5), (6), (7), (8) are as low as 10 and 2 ppm, respectively. For both of the alloys, the EMF value gradually decreased with time and approached zero. In particular, the decreasing rate of the EMF for the Fe–C–Al alloy was faster than that for the Fe–C alloy. These confirmed that it is difficult to measure the oxygen concentration in molten steel for a long time with a single Cr/Cr₂O₃-type sensor.

Fig. 1.

Results of continuous EMF measurements of the Fe–C alloy with [C] = 5 mass% and the Fe–C–Al alloy with [Al] = 0.05 mass%.

The cross-sectional microstructure of the sensor at the interface between MSZ and the reference electrode after the EMF measurement of the Fe–C alloy is shown in Fig. 2. In the sensor after measurement of the Fe–C alloy, the reference electrode did not contact most of the MSZ inner surface; however, some droplet structures with diameters of 10 to 50 μm existed in that part of the MSZ surface. The results of EDX analysis of each point in the droplet substance shown in Fig. 2 are given in Table 2. Some droplets showed two areas with different contrast, and both areas were mainly composed of metallic chromium. Based on the above results, it is considered that the drops were produced not due to densification of the reference powder. The cross-sectional microstructure of the sensor at the interface between MSZ and the reference electrode after EMF measurement of the Fe–C–Al alloy is shown in Fig. 3. In the sensor after measurement of the Fe–C–Al alloy, the reference electrode was closer to the surface of MSZ compared with the sensor immersed in the Fe–C alloy. In general, Cr/Cr₂O₃ mixed powder greatly shrinks by sintering when heated, and thus the reference electrode must be apart from the MSZ surface. Thus, it is considered that the reference electrode became the liquid phase and filled the MSZ tube during the EMF measurement. In addition, the reference electrode on the MSZ surface separated into two phases. The results of EDX analysis of each point of the reference electrode in Fig. 3 are given in Table 3. The two phases were mainly composed of metallic chromium. The above results indicated that a liquid phase mainly composed of metallic chromium was partially produced at the zirconia surface during the EMF measurement and the chromium material separated into two phases.

Fig. 2.

Cross-sectional microstructure at the MSZ surface exposed to the reference electrode of the Cr/Cr₂O₃-type zirconia sensor immersed in Fe–C alloy with [C] = 5 mass% for 30 min to measure the EMF.

Table 2. Results of EDX analysis of the point of each phase found in the droplet on the zirconia surface of the sensor immersed in Fe–C alloy [at%] (the analysis points are shown in Fig. 2).

	O	Mg	Zr	Mo	Cr
Point 1 (High contrast)	10.15	2.67	0.28	1.22	85.68
Point 2 (Low contrast)	8.36	2.89	0.29	1.05	87.42

Fig. 3.

Cross-sectional microstructure at the MSZ surface exposed to the reference electrode of the Cr/Cr₂O₃-type zirconia sensor immersed in Fe–C–Al alloy with [Al] = 0.05 mass% for 30 min to measure the EMF.

Table 3. Results of EDX analysis of the points of each phase found on the zirconia surface of the sensor immersed in Fe–C–Al alloy [at%] (the analysis points are shown in Fig. 3).

	O	Mg	Zr	Mo	Cr
Point 1 (High contrast)	0.08	5.29	3.35	4.12	87.15
Point 2 (High contrast)	0.08	5.41	3.51	4.59	86.41
Point 3 (Low contrast)	0.07	5.79	2.54	2.82	88.77
Point 4 (Low contrast)	0.08	6.43	2.48	2.62	88.40
Point 5 (Low contrast)	0.07	4.45	2.10	2.52	90.86

The XRD patterns of the milled sample and outer surface of the reference electrode after EMF measurements of the Fe–C alloy compared with those of the reference mixture without heating are shown in Figs. 4(a) and 4(b), respectively. Here, the results of the milled sample showed the average of the whole reference electrode, and the outer surface mainly gave information about the interface between the reference electrode and MSZ. Before the measurement, the XRD pattern of the mixture showed the peaks of Cr and Cr₂O₃ (Fig. 4(a)). The peaks of Cr₂O₃ disappeared during the measurement. In addition, peaks of the spinel structure were observed for the reference electrode after measurement for only 10 min. The peaks of metallic chromium shifted to lower angle with time because of diffusion of molybdenum from the electrode wire into the reference electrode. This indicated that the chromium oxide in the reference electrode was reduced during the EMF measurement. Only metallic chromium existed at the interface between the reference electrode and MSZ (Fig. 4(b)). Thus, it is considered that the reduction reaction of chromium oxide occurred at the surface of MSZ.

Fig. 4.

XRD patterns of (a) the milled sample and (b) outer surface of the reference electrode after measuring the EMFs of the molten Fe–C alloy at different times (●, Cr[bcc]; ◆, Cr₂O₃; △, Cr[hcp]; □, spinel).

3.2. Intermittent DC Application

The results of EMF measurements with time when a direct electrical current was intermittently applied to the zirconia sensor during the EMF measurement compared with those of the continuous measurement without current application shown in Fig. 2 are shown in Fig. 5. When an electrical current was applied, the reference electrode was held at a higher electric potential than the molten alloy. For the EMF measurement during the interval of current application, the EMF rapidly decreased in the initial stage of re-measurement, and the degree of the decrease subsequently switched to slightly decreasing. The experimental points shown in the Fig. 5 indicate the EMF value at the end of each measurement because the measurement values were relatively stable for a few seconds. In the case of application of a current of 0.8 A, the EMF was maintained at approximately +50 mV for 20 min and then decreased with time. Conversely, when a current of above 1.5 A was applied, the EMF increased from the initial value and became stable at approximately +70 mV. In both cases, the EMF with current application was higher than that in the continuous measurement, that is, a recharging current can prevent the decrease of the EMF value measured by the zirconia sensor.

Fig. 5.

Results of intermittent EMF measurement by the Cr/Cr₂O₃-zirconia sensor immersed in molten Fe–C alloy with application of different direct electrical currents compared with those of the continuous measurement.

The cross-sectional microstructures of the interface between MSZ and the reference electrode after the EMF measurements with application of currents of 0.8 and 2.3 A are shown in Figs. 6 and 7, respectively. In the sensor with an applied current of 0.8 A, part of the MSZ surface was in contact with the reference electrode (Fig. 6(a)) and some droplet structures also existed in other part of the MSZ surface. The droplet structure had two phases with different contrast on the surface (Fig. 6(b)). The compositions in these structures close to the MSZ surface are given in Table 4. At the contact area, the phase on the MSZ surface was mainly composed of chromium, magnesium, and oxygen, and the ratio of magnesium far from MSZ was lower than that on the MSZ surface. Magnesium was supplied from the dopants of the zirconia solid electrolyte, and magnesium is considered to dissolve in chromium oxide as a spinel of MgCr₂O₄ in the reference electrode. In contrast, both phases of the droplet structure are regarded to be the liquid phase of chromium for the droplet found in the sensor after the continuous measurement in Fig. 2. Therefore, application of a direct electrical current is considered to partially promote the re-oxidation reaction of the reference materials on the MSZ surface.

Fig. 6.

Cross-sectional microstructure of the zirconia surface exposed to the reference electrode in the Cr/Cr₂O₃-type zirconia sensor after EMF measurement with intermittent application of 0.8 A.

Fig. 7.

Cross-sectional microstructure of the zirconia surface exposed to the reference electrode in the Cr/Cr₂O₃-type zirconia sensor after EMF measurement with intermittent application of 2.3 A.

Table 4. Results of EDX analysis of the points of each phase found on the zirconia surface of the sensor immersed in Fe–C alloy with a current of 0.8 A [at%] (the analysis points are shown in Fig. 6).

	O	Mg	Zr	Mo	Cr
Point 1 (in Fig. 6(a))	19.92	19.66	0.63	0.46	59.34
Point 2 (in Fig. 6(a))	26.38	17.42	0.74	0.21	55.25
Point 3 (in Fig. 6(a))	17.53	12.22	0.91	0.68	68.67
Point 4 (in Fig. 6(a))	23.13	10.63	0.68	0.45	65.11
Point 5 (in Fig. 6(b))	12.24	3.54	0.89	0.48	82.84
Point 6 (in Fig. 6(b))	10.56	3.44	0.48	0.56	84.97
Point 7 (in Fig. 6(b))	10.46	4.51	0.57	0.58	83.88
Point 8 (in Fig. 6(b))	10.52	5.70	0.50	0.64	82.64
Point 9 (in Fig. 6(b))	8.53	3.30	0.61	0.50	87.06

In the sensor with an applied current of 2.3 A, several types of phases with different contrast were found around the MSZ surface (Fig. 7). The results of EDX analysis of each phase in the reference electrode shown in Fig. 7 are given in Table 5. According to Fig. 7 and Table 5, a phase composed of chromium containing magnesium as a subcomponent was found on the whole MSZ surface. This phase was also composed of the oxide of chromium and magnesium, similar to that observed in the sensor with an applied current of 0.8 A, but the formed oxide phase was more dense. In contrast, the substance separated into two phases slightly away from the MSZ surface. Based on the results of the composition of the structure in Table 5, each phase was regarded to be mainly metallic chromium because the magnesium ratio was small compared with the oxide phase observed on the MSZ surface. These structures even existed in the sensor with application of a current of 1.5 A. Therefore, application of a large current can cause an undesirable reaction for EMF measurement, such as excessive oxidation or diffusion of the components from MSZ.

Table 5. Results of EDX analysis of the points of each phase found on the zirconia surface of the sensor immersed in Fe–C alloy with a current of 2.3 A [at%] (the analysis points are shown in Fig. 7).

	O	Mg	Zr	Mo	Cr
Point 1	0.09	13.22	1.34	0.64	84.71
Point 2	0.15	6.16	14.01	0.40	79.29
Point 3	0.08	5.73	0.86	4.35	88.98
Point 4	0.08	6.07	1.04	6.83	85.98

3.3. Mechanism of the EMF Value Reduction and Change of the Cross-section

Based on the results mentioned in Sections 3.1 and 3.2, the mechanism of the decrease of the EMF and the phenomenon observed on the MSZ surface exposed to the reference electrode with application of a direct current are considered to be as follows.

Initially, when the zirconia sensor was immersed in the molten alloy, Cr and Cr₂O₃ existed as solids in the MSZ tube. In the zirconia solid electrolyte, where there was an oxygen-potential difference between the inside and outside of the electrolyte, oxygen ions flowed from the high oxygen potential side to the low oxygen potential side. In the system of this experiment, oxygen likely diffused from the reference electrode to the sample electrode. Meanwhile, electron reversely flowed against oxygen ion from the sample electrode to the reference electrode because of maintaining electrical neutrality in the electrolyte. At the edges of the zirconia solid electrolyte, the following reactions occurred between the oxygen ion and electron.

Inside edge (exposed to the reference electrode):   

$$\begin{array}{c}1/2{\mathrm{O}}_2\left(\mathrm{g},\mathrm{\hspace{0.17em} }\mathrm{i}\mathrm{n}\mathrm{\hspace{0.17em} }\mathrm{t}\mathrm{h}\mathrm{e}\mathrm{\hspace{0.17em} }\mathrm{r}\mathrm{e}\mathrm{f}\mathrm{e}\mathrm{c}\mathrm{e}\mathrm{n}\mathrm{r}\mathrm{e}\mathrm{\hspace{0.17em} }\mathrm{e}\mathrm{l}\mathrm{e}\mathrm{c}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{d}\mathrm{e}\right)+2{\mathrm{e}}^{-}\to {\mathrm{O}}^{2-}\left(\mathrm{i}\mathrm{n}\mathrm{\hspace{0.17em} }\mathrm{M}\mathrm{S}\mathrm{Z}\right)\end{array}$$(11)

Outside edge (exposed to the sample electrode):   

$$\begin{array}{c}{\mathrm{O}}^{2-}\left(\mathrm{i}\mathrm{n}\mathrm{\hspace{0.17em} }\mathrm{M}\mathrm{S}\mathrm{Z}\right)\to \left[\mathrm{O}\right]\left(\mathrm{i}\mathrm{n}\mathrm{\hspace{0.17em} }\mathrm{m}\mathrm{o}\mathrm{l}\mathrm{t}\mathrm{e}\mathrm{n}\mathrm{\hspace{0.17em} }\mathrm{a}\mathrm{l}\mathrm{l}\mathrm{o}\mathrm{y}\right)+2{\mathrm{e}}^{-}\end{array}$$(12)

Thus, it is considered that oxygen flowed in the zirconia sensor as follows. Initially, oxygen was supplied to the inner wall of the electrolyte from the reference electrode as oxygen gases or chromium oxides. Then oxygen ions flowed through the zirconia solid electrolyte from the inside to outside of the electrolyte, and oxygen was released to the molten iron. However, the vapor pressure of the chromium oxide, such as CrO(g) or CrO₂(g), was low in the equilibrium between Cr and Cr₂O₃. The vapor pressures of CrO(g) and CrO₂(g) at 1800 K were calculated to be 3 × 10⁻⁷ and 8 × 10⁻¹⁰ atm from the equilibrium constant.²³⁾ Therefore, oxygen was mainly removed from the Cr₂O₃ powder of the reference electrode remaining on the zirconia surface. Cr₂O₃ was reduced by oxygen diffusion into the molten alloy at the low oxygen-potential side as oxygen ions passed through the zirconia solid electrolyte. As a result, only metallic chromium remained on the MSZ surface. Thus, the balance between Cr and Cr₂O₃ in Eq. (7) collapsed and the oxygen potential could become lower on the MSZ surface. The EMF value likely decreased with time because the oxygen-potential difference between the inside and outside edges of the zirconia solid electrolyte almost disappeared.

The chromium phase reduced in the measurement was locally observed as the liquid phase on the MSZ surface. The Cr–Cr₂O₃ phase diagram based on the report by Toker et al.²⁶⁾ is shown in Fig. 8. In this study, the experiment was carried out at a temperature lower than 1938 K, which is the eutectic temperature of the Cr–O system.²⁷⁾ This means that the droplet substance cannot be produced from only the reference material. The liquid phase is considered to be produced on the MSZ surface, as mentioned in the above section, and thus it is possible that the elements constituting MSZ, that is, Zr and Mg, cause the melting of the reference material. Here, thermodynamic calculation of the phase diagram in the Cr–Zr–Mg–O₂ system was carried out on the condition that the reference substance (Cr and Cr₂O₃) and MSZ were mixed. For the calculation, FactSage8.1 was used as the software and TDnucl was selected as the thermodynamic database containing the information about each element. The pseudo-phase diagram of Cr–Cr₂O₃ containing 1 mol% MSZ is shown in Fig. 9, in which BCC, spinel, and tetoxide indicate metallic Cr, a solid solution of MgCr₂O₄ or Cr₃O₄, and a solid solution of ZrO₂ and MgO, respectively. As shown in the phase diagram, even if less of the components of MSZ diffused to the reference mixture, the melting temperature in the Cr–O system can be lowered to 1887 K, which is in the range of the temperature in the furnace. For instance, the equilibrium state corresponding to the oxygen potential and diffusion rate of MSZ at 1900 K is shown in Fig. 10. According to Fig. 10, when MSZ diffuses to the reference electrode on the surface in which the equilibrium state between Cr and Cr₂O₃ is satisfied, the reference material can be partially transformed to the liquid phase. As mentioned above, oxygen was lacking at the surface between the reference electrode and MSZ during the EMF measurement, so the oxygen potential in this area decreased with time and the liquid state became an unstable state. Thus, it is considered that the liquid phase temporarily formed for the above reason was continuously deoxidized at the MSZ surface and transformed to metallic chromium, and the trail of the liquid was maintained in a droplet shape.

Fig. 8.

Cr–Cr₂O₃ phase diagram with oxygen isobars reported by Toker et al.²³⁾ The diagram was adapted from a diagram published in Slag Atlas.²⁴⁾

Fig. 9.

Pseudo-phase diagram of Cr–Cr₂O₃ containing 1 mol% MSZ solid electrolyte calculated by FactSage 8.1. (Online version in color.)

Fig. 10.

Phase diagram corresponding to the oxygen potential and diffusion rate of the MSZ solid electrolyte to the Cr–O system at 1900 K calculated by FactSage 8.1. (Online version in color.)

In contrast, it is considered that the reverse reaction is promoted when a direct electrical current is applied so that the electric potential is higher at the reference electrode than in the molten alloy. In this case, current flows to satisfy the above potential difference between the edges of the zirconia solid electrolyte. In the zirconia solid electrolyte, electric charge mainly flows as oxygen ions. The oxygen ions transferred through the zirconia solid electrolyte react with the metallic chromium on the surface, and the phase composed of chromium oxide is produced. Here, the vapor pressure of chromium is relatively high in the reference electrode (the vapor pressure of Cr(g) is 9 × 10⁻⁵ atm at 1800 K in the Cr/Cr₂O₃ equilibrium state),²⁷⁾ so that the oxide phase grows on the whole surface. However, the chromium oxide on the zirconia surface can react and form a solution with magnesium, especially at a high oxygen potential.

4. Conclusions

Using a Cr/Cr₂O₃-type zirconia oxygen sensor as a practical sensor used in the steel-making process, continuous EMF measurement has been performed to reveal the mechanism that hinders continuous long-time measurement of the EMF. When the oxygen potential in the reference electrode is higher than that in the molten iron, continuous EMF measurement with a single sensor is problematic because the EMF decreases from a positive value to 0 mV with time. At that moment, the chromium oxide in the reference electrode is reduced on the inner surface of the zirconia solid electrolyte. Because the chemical equilibrium between chromium and chromium oxide is lost at the surface of the zirconia solid electrolyte, the oxygen-potential difference between both ends of the electrolyte becomes small and the measured EMF value could consequently decrease. In addition, from observation of the cross-sectional microstructure, chromium-type liquid exists on the surface of the zirconia solid electrolyte. The liquid phase is produced owing to the depression of the melting point attributed to diffusion of ZrO₂ and MgO from the electrolyte to the reference electrode.

Application of a direct electrical current to the electrodes of the sensor during measurement intervals was performed to lengthen the measurement time of the Cr/Cr₂O₃-type zirconia sensor. The EMF values were higher than those without application of a direct electrical current, the phase was composed of oxides of chromium and zirconium, and magnesium was observed on the surface of zirconia after application. It is considered that application of a direct electrical current promotes diffusion of oxygen ions to the reference electrode through the zirconia solid electrolyte and a redox reaction occurs on the surface between the reference electrode and zirconia. However, the reaction excessively proceeded if the current value is large. According to the above results, it is possible to restore the EMF value measured by the Cr/Cr₂O₃-type zirconia sensor when the oxygen potential in the reference electrode is higher than that in the sample.

References

• 1)   S.  Matoba and  T.  Kuwana: Tetsu-to-Hagané, 51 (1965), 163 (in Japanese). https://doi.org/10.2355/tetsutohagane1955.51.2_163
• 2)   S.  Matoba and  T.  Kuwana: Tetsu-to-Hagané, 51 (1965), 1114 (in Japanese). https://doi.org/10.2355/tetsutohagane1955.51.6_1114
• 3)   J.  Fukumi,  C.  Taki,  T.  Hatanaka and  H.  Ogura: Tetsu-to-Hagané, 76 (1990), 1956 (in Japanese). https://doi.org/10.2355/tetsutohagane1955.76.11_1956
• 4)   Z.  Xu,  Z.  Zheng,  X.  Gao and  X.  Shen: J. Clean. Prod., 282 (2021), 124529. https://doi.org/10.1016/j.jclepro.2020.124529
• 5)   E. T.  Turkdogan and  R. J.  Fruehan: Can. Metall. Q., 11 (1972), 371. https://doi.org/10.1179/cmq.1972.11.2.371
• 6)   M.  Iwase and  T.  Mori: Trans. Iron Steel Inst. Jpn., 19 (1979), 126. https://doi.org/10.2355/isijinternational1966.19.126
• 7)   K. S.  Goto: Denki Kagaku, 50 (1982), 54. https://doi.org/10.5796/kogyobutsurikagaku.50.54
• 8)   K.  Nagata,  N.  Tsuchiya,  K.  Urata,  M.  Matsuoka and  K. S.  Goto: Tetsu-to-Hagané, 71 (1985), 183 (in Japanese). https://doi.org/10.2355/tetsutohagane1955.71.2_183
• 9)   F.  Li,  Z.  Zhu and  L.  Li: Solid State Ionics, 70–71 (1994), 555. https://doi.org/10.1016/0167-2738(94)90371-9
• 10)   T.  Liu,  X.  Zhang,  L.  Yuan and  J.  Yu: Solid State Ionics, 283 (2015), 91. https://doi.org/10.1016/j.ssi.2015.10.012
• 11)   N.  Saeki,  M.  Suzuki,  M.  Nakamoto,  T.  Tanaka,  K.  Katogi,  H.  Kosaka,  K.  Dozono and  R.  Ishii: ISIJ Int., 61 (2021), 667. https://doi.org/10.2355/isijinternational.ISIJINT-2020-093
• 12)   F.  Li,  Y.  Tang and  L.  Li: Solid State Ionics, 86–88 (1996), 1027. https://doi.org/10.1016/0167-2738(96)00246-9
• 13)   C.  Wagner: Z. Phys. Chem., 21B (1933), 25. https://doi.org/10.1515/zpch-1933-2105
• 14)   M.  Iwase,  E.  Ichise,  M.  Takeuchi and  T.  Yamasaki: Trans. Jpn. Inst. Met., 25 (1984), 43. https://doi.org/10.2320/matertrans1960.25.43
• 15)   T. H.  Etsell and  S. N.  Flengas: J. Electrochem. Soc., 119 (1972), 1. https://doi.org/10.1149/1.2404121
• 16)   J. W.  Patterson: J. Electrochem. Soc., 118 (1971), 1033. https://doi.org/10.1149/1.2408241
• 17)   H.  Schmalzried: Z. Elektrochem., 66 (1962), 572.
• 18)   Y.  Kita,  S.  Ohguchi and  Z.  Morita: Tetsu-to-Hagané, 64 (1978), 711 (in Japanese). https://doi.org/10.2355/tetsutohagane1955.64.6_711
• 19)   M.  Sasabe,  J. Z.  Hua,  C.  Furuta,  H.  Matsushige,  T.  Nagatsuka,  Y.  Kikuchi,  T.  Takaoka and  Y.  Kawai: Tetsu-to-Hagané, 78 (1992), 391 (in Japanese). https://doi.org/10.2355/tetsutohagane1955.78.3_391
• 20)   G. K.  Sigworth and  J. F.  Elliott: Met. Sci., 8 (1974), 298. https://doi.org/10.1179/msc.1974.8.1.298
• 21)   O.  Knacke,  O.  Kubaschewski and  K.  Hesselmann: Thermochemical Properties of Inorganic Substances, 2nd ed., Springer-Verlag, Berlin, (1991), 1.
• 22)   S.  Ozawa,  Y.  Kawanobe,  K.  Kuribayashi and  T.  Nagasawa: Int. J. Microgravity Sci. Appl., 33 (2016), 330214. https://doi.org/10.15011/jasma.33.330214
• 23)   J.  Li and  Y.  Kobayashi: ISIJ Int., 60 (2020), 1135. https://doi.org/10.2355/isijinternational.35.512
• 24)   Y.  Saito and  T.  Maruyama: Bull. Jpn. Inst. Met., 19 (1980), 796. https://doi.org/10.2320/materia1962.19.796
• 25)   A.  Mittal,  G. J.  Albertsson,  G. S.  Gupta,  S.  Seetharaman and  S.  Subramanian: Metall. Mater. Trans. B, 45 (2014), 338. https://doi.org/10.1007/s11663-014-0027-x
• 26)   N. Y.  Toker,  L. S.  Darken and  A.  Muan: Metall. Trans. B, 22 (1991), 225. https://doi.org/10.1007/BF02652487
• 27)   M.  Kowalski,  P. J.  Spencer and  D.  Neuschütz: Slag Atlas, 2nd ed., Verlag Stahleisen GmbH, Düsseldorf, (1995), 21.