A New Design of Oxygen Sensor for Electromotive Force Measurement and Electrochemical Deoxidation by Using Oxygen Pump

Abstract

Electromotive force measurement (EMF method) has developed for many decades. It provides a universal approach to measure quantities such as oxygen partial pressure and activity coefficient of metals. Here we present a new design of oxygen sensor, aiming to avoid complications and inaccuracies which are caused by the effect of extra metallic lead wires. Hereafter, we focus on the development of a cleaner electrochemical deoxidation technology by using the newly developed apparatus. The EMF experiments demonstrate a favorable agreement with previous literature, and the electrochemical deoxidation experiments show remarkable results of oxygen contents reducing. All of these results pinpoint the feasibility of the newly developed apparatus. Based on these positive results, we discuss a possible application of this study in the steelmaking process, illustrating a high potential of a further and ultimate deoxidation by a cheaper and cleaner approach.

1. Introduction

Since 1933, Fischer and Ackermann proposed a new electromotive force measurement (EMF method) to investigate the Gibbs free energy for oxygen dissolution in liquid iron, cobalt, nickel and copper, the concept of the EMF method has been expanded to ranges of metals.^{1,2,3,4,5,6,7,8)} Kiukkola employed the stabilized zirconia solid electrolyte to measure the oxygen activity in liquid Cu–O system and Cu–Fe–O system. Furthermore, he also extended his results and suggested the possibility of investigating the standard free energy of formation of CoO by immersing the Ni/NiO and Co/CoO electromotive sensor simultaneously in the same copper bath.²⁾ Belford and Alcock applied EMF method to measure the solubility of oxygen in molten lead and molten tin by using ZrO₂+14%MgO solid electrolyte and ThO₂+15%Y₂O₃ solid electrolyte respectively.^3,4) Iwase, Miki and Mori employed the oxygen sensor made by the zirconia-based electrolyte in the Ni–O system and investigated the free energy of the dissolved oxygen in molten nickel at various temperatures.⁵⁾ However, ranges of metallic lead wires were dipped into the molten metals in these experiments. Although they reported satisfactory results through the specific oxygen sensor, the immersed part of the extra metallic lead wire was one of the instabilities for the experiment. Furthermore, the extra lead wire would not only raise the difficulty and risk in the practical operation but significantly increase the cost of the manufacturing process. Consequently, the present work presented a newly designed oxygen sensor to solve this problem. The novel idea was applied on the EMF experiments of molten copper and molten nickel respectively to verify its feasibility.

Based on this newly developed apparatus, we further propose a new oxygen pump and investigate a new approach to reduce the oxygen content in molten metals electrochemically.

Traditional deoxidation process always uses some oxophilic metals, named deoxidizer as well in practical manufacturing. These deoxidizers react with dissolved oxygen in the molten metals and form stable oxides as deoxidation by-products. For example, Al and Si are the most common deoxidizers in the steelmaking process, being added in the ladle furnace to produce high-quality steel with an extremely low oxygen concentration of a few ppm order.⁹⁾ Those oxide inclusions will be absorbed in the slag and removed from molten steel so far. Despite the convenience of using the deoxidizer, the demerit of it is also obvious. it is virtually impossible to remove these inclusions completely since very small size inclusions are hard to float in molten steel and takes long time for its satisfactory removal. Remaining oxide inclusions in this way causes internal and surface defects and material degradation.

In recent years, the concept of electrochemical deoxidation is innovated to solve this problem. Several researchers attempted to develop this technology by using similar apparatus mentioned above, i.e., an oxygen sensor with different metallic lead wire, such as chromel (Ni–Cr alloy), Mo, Pt and so on.^{10,11,12,13,14,15,16,17)} Nevertheless, there is no evidence on which we can avoid the oxidation of metallic lead wire at high temperature. The result of low oxygen concentration in molten metals might be caused by both electrochemical and chemical deoxidation in the previous works. In other words, the net efficiency of electrochemical deoxidation is still unclear.

Consequently, the main propose of the present work is to design a new apparatus to avoid the effect of metallic lead wire in both EMF method and deoxidation technology. We start with verifying its feasibility in EMF experiments by using molten copper and molten nickel, respectively. After the EMF experiment, we apply our new apparatus in electrochemical deoxidation experiment to suggest the possibility of a cleaner and easier electrochemical deoxidation technology. At last, we also discuss some optimizations and application ideas for the present work to increase its potential in practice.

2. Experimental Procedure

2.1. Mechanism of EMF Method

The principle of the EMF method is based on the electric potential difference between two electrodes. As shown in Fig. 1, The electromotive force in the open circuit can be expressed by Nernst equation as follows:   

$$E=-\frac{1}{nF}\int {}_{{\mu }_{{\text{O}}_2}^{\text{High}}}^{{\mu }_{{\text{O}}_{\text{2}}}^{\text{Low}}}{t}_{\text{ion}}d{\mu }_{{\text{O}}_2}$$(1)

Fig. 1.

Mechanism of EMF method. (Online version in color.)

Where ${\mu }_{{\text{O}}_2}$ is the oxygen chemical potential for each electrode, t_ion is the ionic transport number of the solid electrolyte, which is supposed to be unity in general. n is the number of electrons transferred in a half electrochemical reaction (mol) and F is the Faraday constant (F = 96500 C·mol⁻¹).

The more general form of Eq. (1) is   

$$E=\frac{RT}{4F}\text{ln}\frac{P_{{\text{O}}_2}^{\text{High}}}{P_{{\text{O}}_2}^{\text{Low}}}$$(2)

Where R is the gas constant (R = 8.314 J·mol⁻¹·K⁻¹) and T is the experimental temperature (K). General, the high potential electrode is reference gas and the low potential electrode is molten metal.

Hence, the oxygen partial pressure of molten metals ($P_{{\text{O}}_2}^{\text{M}}$) can be calculated by   

$$P_{{\text{O}}_2}^{\text{M}}=P_{{\text{O}}_2}^{\text{Ref}}\cdot \text{exp}\left(-\frac{4FE}{RT}\right)$$(3)

By using the equilibrium of dissolved oxygen into molten metals,   

$$\frac{1}{2}{\text{O}}_2(\text{g})=\underset{\_}{\text{O}}(X_{\text{O}}\mathrm{\hspace{0.17em} }\text{in}\mathrm{\hspace{0.17em} }\text{molten}\mathrm{\hspace{0.17em} }\text{metal)}$$(4)

the standard Gibbs energy for the dissolution of oxygen can be expressed by the equation below.   

$$\mathrm{\Delta }{G}^o=-RT\text{ln}\frac{f_{\text{O}}\cdot X_{\text{O}\mathrm{\hspace{0.17em} }\text{in}\mathrm{\hspace{0.17em} }\text{molten}\mathrm{\hspace{0.17em} }\text{metal}}}{{\left(P_{{\text{O}}_2}^{\text{M}}\right)}^{\raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$2$}\right.}}$$(5)

Where f_O is the activity coefficient of dissolved oxygen relative to 1 mass% Henrian standard, X_{O in molten metal} is the oxygen concentration in molten metal. Since the oxygen concentration in molten metal is extremely low, the activity coefficient of dissolved oxygen is assumed to be unity.

2.2. Innovation of Oxygen Sensor

A schematic diagram of the experiment setup showed in Fig. 2(a). The cathode for the oxygen sensor was a platinum electrode, formed by a platinum wire (99.98% Pt, φ0.30 mm × 700 mm) and a platinum mesh (99.98% Pt, 10 mm × 10 mm). As for the electrolyte part, two kinds of zirconia-based solid electrolytes were applied in the present research (produced by Nikkato Corporation), calcia-stabilized zirconia (CSZ, CaO–ZrO₂, φ15 mm × φ11 mm × 600 mm) for relatively lower temperature and magnesia-stabilized zirconia (MSZ, MgO–ZrO₂, φ15 mm × φ11 mm × 600 mm) for relatively higher temperature. The cathode was lightly attached to the bottom of the stabilized zirconia tube by platinum paste and sintered in an electric furnace at 1273 K to fix its connection. The covered area of Pt mesh was extremely small so that the temperature difference at the interface between the cathode and electrolyte could be negligible. Air was selected as a reference gas because of its convenience and low cost. One innovation of the present oxygen sensor was its anode part, which was made by specific metal for each experiment, respectively. Due to the temperature gradients in the furnace (Fig. 2(b)), only the front part of the metal rod, at which experimental temperature was over than melting temperature, would become liquid. The rest solid metal would connect with a pure metal wire by a groove on the top of the rod, regarding as a lead part. Furthermore, one spring part was made on the pure metal wire in case of the volume reduction of phase transformation. In the present work, we mainly focused on the front part of the liquid phase, where the temperature indicated a stable value with negligible minor fluctuations. Therefore, the temperature of the present oxygen concentration cell was regarded to be constant. Compared with other literature^3,4,5,6,7,8) listed in Table 1, it was noteworthy that an oxygen sensor with the liquid-solid electrode technique could specifically avoid possible contamination from wire metal element to target metals and improves the accuracy of the measurement.

Fig. 2.

Experiment setup (a) Schematic diagram of the experimental apparatus; (b) Temperature gradient in furnace with a set temperature of 1873 K.

Table 1. Various metallic lead wires in previous literature.

	Year	Types of metal	Types of lead wire
Alcock and Belford3,4)	1964	Pb	Ir
	1965	Sn
Iwase et al.5)	1979	Ni	LaCrO3
Rickert and Wagner6)	1966	Cu	Pt
Fitterer7)	1966	Steel	Pt
Kozuka et al.8)	1968	Cu	Mo

In the present work, the metal rod, connecting with a lead wire, was put carefully into the zirconia-based tube at first. Subsequently, the oxygen sensor was filled with argon gas (99.998%) and sealed by a heat-resistant adhesive agent of silicon resin and ceramic cement. The oxygen sensor was then heated up to the experimental temperature and held for several hours. Values of electromotive force in the open circuit were recorded by a digital monitor. Table 2 indicated the experimental condition in detail. Seebeck coefficients for platinum, copper, and nickel at different temperatures were reported by several researchers.^18,19,20) In the present lab-scale experiment, the electromotive force in the open circuit caused by the Seebeck effect was small enough to be negligible.

Table 2. Experimental conditions (oxygen sensor).

Molten metals	Reference gas	Solid electrolyte	Temperature	Holding time	Lead wire
Cu	air	CSZ	1461 K	3600 s	Cu wire
Ni		MSZ	1836 K	10800 s	Ni wire

2.3. Development of Oxygen Pump

Present work also developed a new oxygen pump for electrochemical deoxidation based on the new solid-liquid electrode technique. As shown in Figs. 3(a) and 3(b), an external constant current was added in the circuit. Electrons firstly flew along with the blue arrows, starting from cathode to the molten metal/solid electrolyte interface. Oxygen dissolution closed to the interface would absorb those electrons and form oxygen ion by Eq. (6).   

$$\underset{\_}{\text{O}}\text{(in}\mathrm{\hspace{0.17em} }\text{molten}\mathrm{\hspace{0.17em} }\text{metals)}+2e^{-}={\text{O}}^{2-}$$(6)

Fig. 3.

(a) Schematic diagram of the oxygen pump; (b) Mechanism of electrical deoxidation.

In accordance with the red arrows, oxygen ions would diffuse through the zirconia stabilized solid electrolyte, and reform the oxygen molecule by electron reaction Eq. (7).   

$${\text{O}}^{2-}=\frac{1}{2}{\text{O}}_2(\text{in}\mathrm{\hspace{0.17em} }\text{air})+2e^{-}$$(7)

In the present research, experiment setup was heated to a specific temperature and held for 3500 s to reach a steady state, followed by electrifying a 40-mA constant current for 3600 s to 7200 s to ensure the electromotive force tend to steady. Changes of electrical potential between two electrodes were recorded by the digital monitor. Details of the molten metals and electrifying time are given in Table 3.

Table 3. Experimental conditions (oxygen pump).

Molten metals	Constant current	Solid electrolyte	Temperature	Electrifying time
Cu	40 mA	CSZ	1463 K	7200 s
Cu		MSZ	1413 K	3600 s

3. Results

3.1. Determination of Free Energy for Oxygen Dissolution into Molten Metals by Using Oxygen Sensor

Electromotive force values of the present oxygen sensor were recorded under specific experiment conditions, reaching steady states after 1 hour and 3 hours with 0.6580 V and 0.4920 V for molten copper and molten nickel, respectively. Figure 4 showed the form of each metal rod before and after the experiment.

Fig. 4.

(a) & (b) Cu rod before and after EMF experiment; (c) & (d) Ni rod before and after EMF experiment.

By using inert gas fusion extraction infrared absorption analysis (LECO ONH836), we detected the oxygen concentrations in the front parts of the molten metals as well, showing 13.27 ppm for copper experiment and 1772 ppm for nickel experiment. In the present work, an equilibrium value was defined as the oxygen content for one liquid phase in the oxygen sensor front part, where the temperature was regarded to be constant. According to Eq. (3), the oxygen partial pressure showed 1.74 × 10⁻¹⁰ atm in molten copper and 4.84 × 10⁻⁷ atm in molten nickel. The Gibbs free energies for the oxygen dissolution into molten metals were calculated by using Eqs. (2), (3), (4), (5). Table 4 showed the comparison with other literature.^{1,21,22,23,24,25,26,27)}

Table 4. Results of EMF method and comparison with literature values.

	Year	Method	$\mathrm{\Delta }{G}^{°}$/kJ∙mol−1
Cu experiment
Present study			−58.14
Fischer and Ackermann1)	1966	EMF method (Pt electrode)	−59.63
Sano and Sakao21)	1955	Gas equilibrium method	−59.39
Belton and Tankins22)	1965	Gas equilibrium method	−53.89
Gerlach, et al.23)	1968	Gas equilibrium method	−57.94
Oberg, et al.24)	1973	EMF method (Ni–Cr electrode)	−57.96
Ni experiment
Present study			−85.37
Chiang and Chang25)	1976	Gas equilibrium method	−87.13
Kemori, et al.26)	1981	EMF method (LaCrO3 electrode)	−87.04
Jacob27)	1986	EMF method (Mo/MoO2 electrode)	−85.98

The results indicated a good agreement between the present work and previous research, suggesting that (i) this new liquid-solid electrode technique was extremely suitable for oxygen sensor experiment; (ii) A slight but unneglected reduction of $\mathrm{\Delta }{G}^{°}$ was caused by Seebeck effect especially at high experimental temperature; (iii) MSZ solid electrolyte indicated good feasibility at high-temperature electromotive force method research.

3.2. Results of Electrochemical Deoxidation

3.2.1. Electrochemical Deoxidation in CSZ Solid Electrolyte

Figure 5 below showed electromotive force changes of the deoxidation experiment by using CSZ solid electrolyte. At first, the electromotive force increased rapidly after adding a constant external current, starting from a steady value of 0.486 V. The value kept rising for around 7000 s and reached a steady platform again finally. After removing the external current, the line dropped immediately but back to a stable position soon, with a value of 0.692 V ultimately.

Fig. 5.

Electrical deoxidation in CSZ.

The equilibrium reaction of oxygen in molten copper was given below.¹⁾   

$$\frac{1}{2}{\text{O}}_2(\text{g})=\underset{\_}{\text{O}}\text{(in}\mathrm{\hspace{0.17em} }\text{molten}\mathrm{\hspace{0.17em} }\text{copper)}$$(8)

  

$$\mathrm{\Delta }{G}^{°}=-73\mathrm{\hspace{0.17em} }220+9.29\mathrm{\hspace{0.17em} }T\mathrm{\hspace{0.17em} }\text{J}/\text{mol}$$(9)

Substituting Eqs. (9) to (5), the oxygen concentration values in molten copper were calculated to be 276 ppm at first and 11 ppm at last. On the other hand, LECO ONH836 analyzer measured the mean oxygen concentration in the molten copper with the value of 11.68 ppm, demonstrating a considerable good agreement with the calculated final value.

The present work calculated the practical deoxidation quantity by assuming the front part of the metal rod, which worked as an electrode, as a half-sphere. In this case, based on the density of copper (8.96 × 10³ kg/m³) and the radius of the electrode (5 mm), the practical oxygen loss resulted in 6.22 × 10⁻⁴ g.

According to Faraday’s law, the mass of oxygen altered at the metal electrode was mathematically calculated by Eq. (10), showing that the removed oxygen contents could be 2.24 × 10⁻² g during the present experiment.   

$$\mathrm{\Delta }{m}_{\text{O}}=\frac{16It}{2F}$$(10)

Where I is the constant current (A), t is the electrifying time (s).

By combining both practical and calculated values, it suggested that the deoxidation efficiency was 2.78% for the electrochemical deoxidation in CSZ. The deoxidation efficiency was affected by (i) thermal diffusion of oxygen from the non-electrode segment to the electrode segment in liquid phase; (ii) limitation of the present solid electrolyte.

3.2.2. Electrochemical Deoxidation in MSZ Solid Electrolyte

Similarly, Fig. 6 displayed the result of the deoxidation experiment with MSZ solid electrolyte. The initial electromotive force value was 0.397 V. After electrifying for 3600 s, the electromotive force value significantly increased to 0.509 V. As a result, the initial oxygen concentration was calculated to be 1079 ppm while the final was 171 ppm. The mean oxygen concentration in molten copper was measured to be 175.95 ppm by using LECO ONH836 analyzer. Likewise, the practical deoxidation content of the electrode was 2.13 × 10⁻³ g while the calculated value based on Faraday’s law of electrolysis was 1.19 × 10⁻² g. Hence, the deoxidation efficiency was 17.9% for the experiment in MSZ. This result showed high feasibility for applying MSZ solid electrolyte in electrochemical deoxidation process.

Fig. 6.

Electrical deoxidation in MSZ.

4. Discussion

4.1. Limitation of Solid Electrolyte under the Practical Condition

Solid electrolyte plays a vital role in electromotive force experiment. As is mentioned before, the transport number of both CSZ and MSZ in the present work is assumed to be unity. However, under a practical condition, several effects being on the transport number of a solid electrolyte, for example, temperature and partial pressure, are usually hard to be handled and controlled because of the order of magnitude. Thus, in general, the transport number is more changeable and sensitive than that under the laboratory condition. In 1971, Patterson²⁸⁾ has reported a range of electrolytic domains for some known solid electrolytes, in which the transport numbers are always unity (Fig. 7). For instance, at 1473 K, the transport number of CSZ can be regarded to be unity only if the oxygen partial pressure in molten metal locates at the scope around 10⁻¹² atm to 10² atm. In addition, the electrolytic domains map also shows that for most part of solid electrolyte materials, the higher temperature they have, the narrower scope of the partial pressure can be calculated with a constant transport number. In electrochemical deoxidation research, the calculation shown above will be exceedingly complicated if the transport number is not unity, i.e., when the oxygen partial pressure in molten metal is over or under the scope. Hence, the present work gives two ideas to improve this deficiency and to estimate the electrochemical deoxidation efficiency more accurately.

Fig. 7.

The electrolytic domains for ranges of solid electrolytes with respect to temperature (Reference gas: air).

Firstly, a reference gas with lower oxygen concentration is suggested to overcome the limitation of the present oxygen sensor. Generally, the ionic transport number t_ion of a zirconia-based solid electrolyte is expressed as   

$$t_{\text{ion}}=\frac{{\sigma }_{\text{ion}}}{{\sigma }_{\text{ion}}+{\sigma }_{\text{e}}+{\sigma }_{\text{h}}}$$(11)

Where σ_ion is the electrical conductivity of the oxygen ion (S), σ_e and σ_h are the electrical conductivities of the electrons and electron holes of the electrolyte (S).

The relationship between the electrical conductivity and oxygen partial pressure can be expressed by using the equilibrium constant of the electron reactions in two potential terminals.

In the low potential terminal, the electrical conductivity of the electrons corresponds to the oxygen partial pressure,   

$${\sigma }_{\text{e}}\propto {\left(P_{{\text{O}}_2}^{\text{Low}}\right)}^{\raisebox{1ex}{$-1$}\!\left/ \!\raisebox{-1ex}{$4$}\right.}$$(12)

In addition, in the high potential terminal, σ_h is expressed as below, predominating with the oxygen partial pressure $P_{{\text{O}}_2}^{\text{High}}$.   

$${\sigma }_{\text{h}}\propto {\left(P_{{\text{O}}_2}^{\text{High}}\right)}^{\raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$4$}\right.}$$(13)

Therefore, reduction of the oxygen partial pressure in the high potential terminal, erasing the difference between the two electrodes, can make the ionic transference number of the zirconia-based solid electrolyte be assumed to be unity or closer again.

Secondly, it is also achievable to replace the present solid electrolyte to whose electrolytic domains are suitable for tested metal. For example, Yttria-doped thoria (YDT, Y₂O₃–ThO₂) shows a more fit scope for measuring lower oxygen partial pressure in molten metals in the electrolytic domains map. Subbarao et al.²⁹⁾ showed that the ionic transference number was about 1 almost from 800°C to 1400°C. Rao and Tare³⁰⁾ also measured the safe lowest limits of oxygen pressure for YDT (15 mol% YO_1.5) are 10⁻³² atm at 927°C and 10⁻²⁴ atm at 1175°C, indicating a good validity under low oxygen pressure. Yttria-stabilized zirconia (YSZ, Y₂O₃–ZrO₂) also showed a higher ionic conductivity than CSZ at the same temperature,^31,32,33) suggesting that it is easier to transport the oxygen ions from the molten metal terminal to the reference gas terminal in the deoxidation experiment.

However, how to decrease the cost of electrolyte development is the primary problem that practical application is facing. MSZ, as one attempt in the present research, shows a good agreement in the electrochemical deoxidation experiment, but its stability and the feasibility for molten metal with lower oxygen partial pressure are still needed to be verified by further research.

4.2. Application for Practical Manufacturing

Although the technology of oxygen sensor is quite mature in various engineering fields, the idea of oxygen pump is still worthwhile developing to improve the quality of products for practical manufacturing. Figure 8 shows one possible example of oxygen pump application in the steelmaking process. In the continuous casting process, inserting the oxygen pump technology in the ladle can remove oxygen content in molten steel, resulting in a further and ultimate deoxidation.

Fig. 8.

Idea of oxygen pump application in the steelmaking process.

5. Conclusion

In this study, we design a new solid-liquid coexistent oxygen sensor and apply it on the EMF method with molten copper and molten nickel to verify its feasibility. The results show that:

(1) The newly developed oxygen sensor is feasible for EMF experiment because of a good agreement between the present result and previous literature. It is also noteworthy that the $\mathrm{\Delta }{G}^{°}$ for the oxygen dissolution into molten metals can be influenced by Seebeck effect at high temperature, causing a slight but unneglected deviation.

(2) Based on the newly developed apparatus, we further propose a new electrochemical deoxidation approach. The remarkable results show that both CSZ and MSZ are feasible for the deoxidation process.

(3) We also discuss the limitation of the present work in a practical situation. Although we get a good agreement presently, we are not able to assert that CSZ or MSZ is the most suitable material in our deoxidation project due to the lack of result. Two possible improvements are suggested in this study as follows.

(i) To change a reference gas with lower oxygen concentration;

(ii) To explore new solid electrolytes which are suitable for the metal with ultimate low oxygen concentration.

(4) We propose a possible application of this study in the steelmaking process at last. It shows a high potential that we can achieve a further and ultimate deoxidation by this cheaper and cleaner approach.

Acknowledgement

The authors are grateful to the Elements Strategy Initiative Center for Magnetic Materials for the financial support to this research in the Elements Strategy Project, lauched by Ministry of Education, Culture, Sports, Science and Technology (MEXT).

References

• 1)   W. A.  Fischer and  W.  Ackermann: Arch. Eisenhüttenwes., 37 (1966), 43 (in German).
• 2)   K.  Kiukkola and  C.  Wagner: J. Electrochem. Soc., 104 (1957), 379.
• 3)   C. B.  Alcock and  T. N.  Belford: Trans. Faraday Soc., 60 (1964), 822.
• 4)   T. N.  Belford and  C. B.  Alcock: Trans. Faraday Soc., 61 (1965), 443.
• 5)   M.  Iwase,  S.  Miki and  T.  Mori: J. Chem. Thermodyn., 11 (1979), 307.
• 6)   H.  Rickert and  H.  Wagner: Electrochim. Acta, 11 (1966), 83.
• 7)   G. R.  Fitterer: JOM, 18 (1966), 961.
• 8)   Z.  Kozuka,  K.  Suzuki,  T.  Oishi and  J.  Moriyama: J. Jpn. Inst. Met., 32 (1968), 1132.
• 9)   K.  Kawakami: Sanyo Tech. Rep., 14 (2007), 22 (in Japanese).
• 10)   K. E.  Oberg,  L. M.  Friedman,  W. M.  Boorstein and  R. A.  Rapp: Metall. Trans., 4 (1973), 75.
• 11)   S.  Yuan,  U.  Pal and  K. C.  Chou: J. Electrochem. Soc., 141 (1994), 467.
• 12)   D.  Yuan and  F. A.  Kröger: J. Electrochem. Soc., 116 (1969), 594.
• 13)   H. W.  Koo,  H. J.  Park,  G. M.  Choi and  H. G.  Lee: ISIJ Int., 47 (2007), 689.
• 14)   W.  Kim,  D. J.  Min,  Y. S.  Lee and  J. H.  Park: ISIJ Int., 49 (2009), 1882.
• 15)   W.  Kim and  D. J.  Min: Steel Res. Int., 82 (2011), 195.
• 16)   G. H.  Zhang,  K. C.  Chou and  F. S.  Li: ISIJ Int., 54 (2014), 2767.
• 17)   Z.  Hasham,  U.  Pal and  K. C.  Chou: J. Electrochem. Soc., 142 (1995), 469.
• 18)   N.  Cusack and  P.  Kendall: Proc. Phys. Soc., 72 (1958), 898.
• 19)   L.  Abadlia,  F.  Gasser,  K.  Khalouk,  M.  Mayoufi and  J. G.  Gasser: Rev. Sci. Instrum., 85 (2014), 095121.
• 20)   R. A.  Howe and  J. E.  Enderby: J. Phys. F: Met. Phys., 3 (1973), L12.
• 21)   K.  Sano and  H.  Sakao: J. Jpn. Inst. Met., 19 (1955), 435 (in Japanese).
• 22)   G. R.  Belton and  E. S.  Tankins: Trans. Metall. Soc. AIME., 233 (1965), 1892.
• 23)   J.  Gerlach,  J.  Osterwald and  W. Z.  Stichel: Z. Metallkd., 59 (1968), 576.
• 24)   K. E.  Oberg,  L. M.  Friedman,  W. M.  Boorstein and  R. A.  Rapp: Metall. Trans., 4 (1973), 61.
• 25)   T.  Chiang and  Y. A.  Chang: Metall. Trans. B, 7 (1976), 453.
• 26)   N.  Kemori,  I.  Katayama and  Z.  Kozuka: J. Chem. Thermodyn., 13 (1981), 313.
• 27)   K. T.  Jacob: Metall. Trans. B, 17 (1986), 763.
• 28)   J. W.  Patterson: J. Electrochem. Soc., 118 (1971), 1033.
• 29)   E. C.  Subbarao,  P. H.  Sutter and  J.  Hrizo: J. Am. Ceram. Soc., 48 (1965), 443.
• 30)   A. V.  Ramana Rao and  V. B.  Tare: Scr. Metall., 5 (1971), 813.
• 31)   D. W.  Strickler and  W. G.  Carlson: J. Am. Ceram. Soc., 47 (1964), 122.
• 32)   R. W.  Vest and  N. M.  Tallan: J. Am. Ceram. Soc., 48 (1965), 472.
• 33)   R. W.  Vest and  N. M.  Tallan: J. Appl. Phys., 36 (1965), 543.