Improved Responsivity and Sensitivity of Hydrogen Mapping Technique in Pure Iron Using WO₃ Thin Film by Control of Pd Intermediate Layer

Abstract

The objective in this study is to improve the responsivity and the sensitivity of the hydrogen mapping technique using the WO₃ thin film by the optimization of the Pd intermediate layer. Especially, the effect of the thickness of Pd on the responsivity and the sensitivity of the hydrogen detection during hydrogen charging was investigated. Pd and WO₃ thin films were coated on the detection side of the pure iron sheet by a magnetron sputtering system. No color change was observed on the hydrogen detection side of the specimens with the Pd intermediate layer more than 25 nm in thickness during the hydrogen detection test for 7.2 ks. The responsivity for the hydrogen mapping technique using WO₃ was essentially improved by decreasing the Pd thickness. In terms of the onset time of the average color change, the earliest response was obtained when the Pd thickness was 4 nm because of the uneven distribution of the color change in the case of the Pd thickness of 2.5 nm. The sensitivity for the hydrogen mapping technique using WO₃ was improved by decreasing the Pd thickness. Taking into account the responsivity, the sensitivity, and the spatial resolution comprehensively, the best thickness of the Pd intermediate layer seems to be 4 nm in this study.

1. Introduction

High strength steels are widely used in many fields, hydrogen embrittlement (HE) is, however, a serious problem for high strength steels. It is well-known that high strength steel bolts suffer from delayed fracture in atmospheric corrosion environments,^1,2,3) and the susceptibility of HE increases with the strength of steels.⁴⁾ The strategy to prevent high strength steel bolts from delayed fracture by the material design has been studied,⁵⁾ and a method to reduce hydrogen uptake into steels under atmospheric corrosion is required to ensure the safe and widespread use of high strength steels. Because hydrogen uptake is deeply sensitive to surface heterogeneity, a real-time mapping technique for obtaining the distribution of the hydrogen absorbed at the steel surface is essential for the suppression of hydrogen uptake.

The combined effect of the low solubility limit of hydrogen into steels and the tendency for diffusible hydrogen to escape from steels at room temperature makes it difficult to detect hydrogen in steels.⁶⁾ This explains the distinct lack of methods available for use in the detection of hydrogen. The hydrogen microprint technique,^7,8) silver decoration,^9,10) secondary ion mass spectrometry,^11,12) and the three dimensional atom probe¹³⁾ are capable of detecting local hydrogen with a high spatial resolution. However, these methods could not measure temporal changes in the distribution of hydrogen uptake. Scanning electrochemical microscopy¹⁴⁾ and scanning laser-enhanced electrochemical microscopy¹⁵⁾ have been reported as in situ hydrogen visualizing methods, but few reports have appeared to date. A surface potential mapping using a scanning Kelvin probe (SKP) and Kelvin probe force microscopy (KPFM) are being established as an in situ hydrogen mapping technique.^{16,17,18,19,20,21,22,23,24)} While the SKP and KPFM are powerful tools for the real-time mapping of hydrogen uptake experimentally, more convenient and simpler methods are required for the application in atmospheric environments.²⁵⁾

We reported an approach to detecting the distribution of hydrogen uptake into pure iron using a WO₃ thin film.²⁶⁾ WO₃ has the electrochromic property, and its optical property changed by the formation of H_xWO₃.^27,28) In our approach, Pd was electrochemically-coated, and the WO₃ film was subsequently formed on the hydrogen detection side of the pure iron sheet. After hydrogen was absorbed into the iron sheet by cathodic polarization on the hydrogen entry side, the color on the hydrogen detection side corresponding to the hydrogen charging area started to change. Because this real-time method to detect the distribution of hydrogen uptake which utilizes the phase transition from WO₃ to H_xWO₃ is simpler than the other hydrogen detection techniques reported to date, it is expected to be applied for various purposes. However, the improvement of the responsivity, the sensitivity, and the spatial resolution are required to realize practical applications. The spatial resolution strongly depends on the thickness of iron sheets. The responsivity and the sensitivity, on the other hand, are possible to be improved by the optimization of Pd and WO₃ films.

The Pd intermediate layer, especially, seems to be a key factor in improving the responsivity and the sensitivity, because no hydrogen uptake could be detected by the WO₃ film without the Pd layer for 43.2 ks.²⁶⁾ It is well-known that the hydrogen solubility of Pd is high,²⁹⁾ and Pd is one of hydrogen spillover catalysts for hydrogen storage.³⁰⁾ For these reasons, the Pd intermediate layer is thought to assist hydrogen permeation in the pure iron/Pd and Pd/WO₃ interfaces, and is essential for detecting hydrogen uptake by the WO₃ film. The objective in this study is to improve the responsivity and the sensitivity of the hydrogen mapping technique using the WO₃ thin film by the optimization of the Pd intermediate layer. The effect of the thickness of Pd on the responsivity and the sensitivity of the hydrogen detection was investigated.

2. Experimental

2.1. Specimen Preparation and Electrolyte

A commercial pure iron sheet 1 mm in thickness was used in this study. Table 1 shows the chemical composition of the pure iron. The pure iron sheet was cut into 25 mm × 25 mm coupons. All the pure iron sheets were heat-treated at 1073 K for 3.6 ks and were then furnace-cooled. The grain size of the pure iron was 10−100 μm. A small number of round inclusions (less than 5 μm in diameter) existed in the pure iron. The typical inclusion was Fe oxides containing no sulfur. Before the deposition of Pd and WO₃, the front and back sides of the sheet were mechanically ground with SiC papers through a 1500 grit, mirror-polished with a 1 μm diamond paste, and ultrasonically cleaned with ethanol.

Table 1. Chemical composition of the specimen (mass%).

	C	Si	Mn	P	S	N	O
Pure iron	0.002	<0.01	<0.01	<0.002	<0.001	0.012	0.022

A Pd thin film was coated on the detection side of the pure iron sheet by a radio frequency (RF) magnetron sputtering system. After the pure iron sheet was loaded into the sputtering chamber, the chamber of the sputtering system was evacuated to a pressure lower than 5 × 10⁻⁵ Pa, and the Ar gas was introduced into the chamber. The flow rates of the Ar gas was 5.1 × 10⁻⁷ m³ s⁻¹. The target was a Pd disc 50.8 mm in diameter, its purity was 99.9 mass%. The distance between the sheet and the target was 100 mm. The Pd sputtering was carried out at a pressure of 5.0 Pa. The RF power was 50 W. The Pd thickness was controlled by the sputtering time. After the end of the Pd sputtering process, the sputtering chamber was re-evacuated to a pressure lower than 2 × 10⁻⁴ Pa, and the subsequent deposition of WO₃ was conducted by reactive RF sputtering in a mixed Ar/O₂ discharge.³¹⁾ The flow rates of the Ar and O₂ gases were 1.7 × 10⁻⁷ and 8.3 × 10⁻⁸ m³ s⁻¹, respectively. The target was a W disc 50.8 mm in diameter, its purity was 99.95 mass%. The distance between the substrate and target was 70 mm. The WO₃ sputtering was conducted at a pressure of 4.5 Pa, and the RF power was 50 W. The sputtering time was 10.8 ks to reach 80 nm in thickness. The pure iron substrate was kept at 298 K, and rotated at 5 rpm (rotations per minute) during all the sputtering process.

Hydrogen charging (see next section) was performed in a naturally aerated 0.1 M H₂SO₄ solution at 298 K. The solutions were prepared using deionized water and analytical grade sulfuric acid.

2.2. Hydrogen Detection Test

Measurements for detecting hydrogen uptake into the specimen by the WO₃ film were taken. Figure 1 shows the schematic of the experimental setup for the observation of hydrogen mapping with the WO₃ film. The hydrogen entry side of the specimen was exposed to an acrylic electrochemical cell for hydrogen charging. The cell was sealed with an O-ring. The working electrode (W. E.) was the mirror-polished pure iron. The counter electrode (C. E.) was Pt, and the reference electrode (R. E.) was an Ag/AgCl (3.33 M KCl) electrode. In this paper, all the potential values refer to the standard hydrogen electrode (SHE). Potentiostatic polarization in 0.1 M H₂SO₄ was performed at −0.5 V for hydrogen charging. The surface of the specimen on the hydrogen detection side during hydrogen charging was monitored by an inverted optical microscope. The microscopic images (1600 × 1200 pixels, 24-bit color image) were captured at an interval of 60 s.

Fig. 1.

Schematic of the experimental setup for observation of hydrogen mapping with a WO₃ film. (Online version in color.)

In the center of the specimen surface, two penetration holes were formed as a marker to match the position on the hydrogen entry side with that on the hydrogen detection side. The size of the hydrogen charging areas was approximately 2 mm × 1 mm. With the exception of the hydrogen charging area, the specimen surface on the hydrogen entry side containing the penetration holes were covered with an epoxy resin.

2.3. Specimen Observation and Analysis

The Fe/Pd/WO₃ interfaces of the specimen were observed by a transmission electron microscope (TEM) equipped with an energy-dispersive X-ray spectroscopy (EDS) system at an accelerating voltage of 200 kV. The sample for the TEM observation was prepared by focused ion beam (FIB) cross-sectioning. The samples were obtained by milling the surface on the hydrogen detection side with a gallium ion beam, and were lifted out using a manipulator. Prior to FIB sampling, carbon was coated on the hydrogen detection side of the specimen.

The specimen surface before and after the hydrogen detection test was taken by a digital camera. The thickness of the WO₃ film and the Pd intermediate layer was determined by the TEM images.

3. Results and Discussion

3.1. Hydrogen Detection for the Specimen with Pd Thickness of 4 nm

We examined the hydrogen detection test for the specimens with various Pd thicknesses (2.5, 4, 7.5, 25, 60, and 100 nm), and observed no color change on the hydrogen detection side of the specimens with the Pd intermediate layer 25, 60, and 100 nm in thickness at 7.2 ks after the start. In addition, we conducted the hydrogen detection test using the specimen without the Pd intermediate layer, and confirmed that no color change occurred on the hydrogen detection side at least for 36 ks. Results of the hydrogen detection test for the specimens with the Pd layer 2.5, 4, and 7.5 nm in thickness were analyzed in detail.

Figure 2 shows the surface appearance of the specimen with the Pd intermediate layer 4 nm in thickness before and after hydrogen charging for 6 ks taken by a digital camera. The images on the hydrogen entry side (Figs. 2(a) and 2(b)) was horizontally-flipped to correspond with those on the hydrogen detection side (Figs. 2(c) and 2(d)). All the images in Fig. 2 are enlarged views at the center of the specimen, and the two penetration holes were observed. Using these holes, the positions were aligned between the hydrogen entry side and detection side. The bright part in Fig. 2(a) was the hydrogen charging area before the detection test. The surrounding area of the bright part was covered with an epoxy resin. The inset in Fig. 2(b) shows a further enlarged view around the hydrogen charging area. After hydrogen charging for 6 ks (Fig. 2(b)), the yellow region was observed in the hydrogen charging area. This indicates the formation of corrosion products during hydrogen charging. In the hydrogen detection side before hydrogen charging (Fig. 2(c)), only the penetration holes were observed. The color of the surface was light blue. After hydrogen charging for 6 ks (Fig. 2(d)), a dark blue and circular shape was clearly observed near the penetration holes. Figure 2(d) exhibited the color change from light blue to dark blue occurred only around the part corresponding to the hydrogen charging area. Our previous research showed that the color change is caused by the formation of H_xWO₃,²⁶⁾ and this result indicates the local hydrogen uptake into the specimen could be visualized by the phase transition of WO₃.

Fig. 2.

Surface appearance of the specimen with the Pd intermediate layer 4 nm in thickness on (a, b) the hydrogen entry side and (c, d) the hydrogen detection side. The images were taken (a, c) before and (b, d) after hydrogen charging for 6 ks. The images in the hydrogen entry side (a, b) was horizontally-flipped. The inset in Fig. 2(b) is an enlarged view around the hydrogen charging area. (Online version in color.)

The color change over time on the detection side was monitored by the inverted optical microscope. Figure 3 shows the optical microscope images of the specimen with the Pd intermediate layer 4 nm in thickness on the hydrogen detection side during hydrogen charging. The dashed-line in the image before the test corresponds to the hydrogen charging area on the hydrogen entry side. At the beginning of hydrogen charging, the surface on the hydrogen detection side was light blue, and no visible color change was observed for 1.8 ks. After 2.4 ks, the color around the center of the region corresponds to the hydrogen charging area slightly changed to dark blue, and a circular shape gradually emerged as time went by. At 3.6 ks, the dark blue and circular shape was clearly recognized. In the colored region, the degree of color change appeared a little uneven. After 4.2 ks, the circular shape became larger with time. At 5.4 ks, the color change expanded outside the hydrogen charging area, which is because the hydrogen atoms absorbed from the hydrogen charging area diffuse through the specimen radially.

Fig. 3.

Optical microscope images of the hydrogen detection side of the specimen with the Pd intermediate layer 4 nm in thickness during hydrogen charging for 6 ks. The dashed-line in the image before the hydrogen detection test corresponds to the electrode area for hydrogen charging. (Online version in color.)

In order to quantitatively evaluate the color change on the hydrogen detection side during hydrogen charging, the RGB values were extracted. Figure 4 shows the time variation of the RGB values of the specimen with the Pd intermediate layer 4 nm in thickness on the hydrogen detection side during hydrogen charging. To obtain the average degree of the color change, the RGB values were extracted from the area corresponding to the hydrogen charging part on the hydrogen entry side indicated in the dashed-area of Fig. 3. All the color values clearly decreased after ca. 2 ks, with the most distinct change occurring in the R value. Even though the color change between 0 to 1.8 ks could not be visibly observed in Fig. 3, a slight decrease in the R value could be detected at 1.8 ks. This suggests that by referring to the R value, the responsivity of the hydrogen detection can be evaluated with good sensitivity.

Fig. 4.

Time variation of RGB values of the hydrogen detection side of the specimen with the Pd intermediate layer 4 nm in thickness during hydrogen charging. The average RGB values in the area corresponding to the hydrogen charging part on the hydrogen entry side (the dashed-area in Fig. 3) were calculated. (Online version in color.)

3.2. Hydrogen Detection for the Specimens with Pd Thickness of 2.5 and 7.5 nm

To investigate the effect of the Pd thickness on the hydrogen detection properties, the color change of the specimens with the Pd layer 2.5 and 7.5 nm in thickness was analyzed. Figure 5 shows that the optical microscope images of the specimen with the Pd intermediate layer 7.5 nm in thickness on the hydrogen detection side during hydrogen charging for 7.2 ks. No visible color change was observed for 2.4 ks. At 3 ks, the color change from light blue to dark blue was slightly observed as indicated by the black arrows. After 3.6 ks, the color change gradually spread around, and gradually emerged as time went by. At the end of the hydrogen detection test (7.2 ks), the colored area was not the circular shape, and the degree of color change appeared rather uneven, as distinct from the specimen with the Pd thickness of 4 nm shown in Fig. 3. In addition, the dark blue at 7.2 ks seemed to be lighter than that at 5.4 ks in Fig. 3. Figure 6 shows that the optical microscope images of the specimen with the Pd intermediate layer 2.5 nm in thickness on the hydrogen detection side during hydrogen charging for 7.2 ks. The color change from light blue to dark blue was initially observed at 1.8 ks as indicated by the black arrows. After 2.4 ks, many dark blue-colored spots were emerged, and gradually spread around. As shown in the image at 3 ks, the color change with an extremely uneven distribution was observed. At the end of the hydrogen detection test (7.2 ks), the color change expanded outside the hydrogen charging area, and was not the circular shape. The dark blue at 7.2 ks appeared deeper than that at 5.4 ks in Fig. 3. These color changes in Figs. 3, 5, and 6 were organized in terms of the average R value.

Fig. 5.

Optical microscope images of the hydrogen detection side of the specimen with the Pd intermediate layer 7.5 nm in thickness during hydrogen charging for 7.2 ks. The dashed-line in the image before the hydrogen detection test corresponds to the electrode area for hydrogen charging. (Online version in color.)

Fig. 6.

Optical microscope images of the hydrogen detection side of the specimen with the Pd intermediate layer 2.5 nm in thickness during hydrogen charging for 7.2 ks. The dashed-line in the image before the hydrogen detection test corresponds to the electrode area for hydrogen charging. (Online version in color.)

Figure 7 shows the time variation of the average R value and its standard deviation on the hydrogen detection side during hydrogen charging. The results of the specimens with the Pd thickness of 2.5, 4, 7.5, and 25 nm were plotted in Fig. 7. The average R value was extracted from the area corresponding to the hydrogen charging part on the hydrogen entry side, as indicated in the dashed-areas of Figs. 3, 5, and 6. In this study, the degree of dispersion for the color change was evaluated by the standard deviation of the R value. In Fig. 7(a), in the case of the Pd thickness of 2.5 and 4 nm, both R values started to decrease at 1.8 ks. In comparison with 4 nm, the average R value change for the Pd thickness of 2.5 nm appeared to decrease slowly. This is thought to be because of the color change with an extremely uneven distribution as shown in Fig. 6. As the thickness of the Pd intermediate layer increased from 4 nm, the onset time of the color change became large, and the specimens with Pd thickness of 25 nm showed no color change for 7.2 ks. From Fig. 7(b), the decrease of the R values in Fig. 7(a) was followed by the increase of its standard deviations for the specimens. In the case of Pd thickness of 4 nm, the standard deviation was the smallest, and the time to return to the original values was the shortest of the three. In terms of the dispersion for the color change, the best thickness of the Pd intermediate layer was 4 nm.

Fig. 7.

Effect of Pd thickness on the time variation of (a) R value and (b) its standard deviation on the hydrogen detection side during hydrogen charging. The average R value was extracted from the area corresponding to the hydrogen charging part on the hydrogen entry side (the dashed-areas in Figs. 3, 5, and 6). (Online version in color.)

3.3. Effect of Pd Thickness on Responsivity and Sensitivity

From the results so far, the effect of the Pd thickness on the responsivity and the sensitivity of the hydrogen detection was analyzed. The responsivity was evaluated by the onset time of the color change. In this study, the time when the decrease of the R value in Fig. 7 exceeded 20 was defined as the onset time of the average color change. In addition, the onset time of the local color changes at Point A in Fig. 3, Point B in Fig. 5, and Point C in Fig. 6 was also obtained. Point A, B, and C are the initial spots for the visible color change observed in Figs. 3, 5, and 6, respectively. And, color difference was used for the evaluation of the sensitivity. In this study, the color difference was calculated with the average RGB values obtained from the images at the start and end of hydrogen charging in Figs. 3, 5, and 6 as follows:   

$$\text{Color difference}=\sqrt{{(R_e-R_0)}^2+{(G_e-G_0)}^2+{(B_e-B_0)}^2}$$

where R₀, G₀, and B₀ were the average RGB values extracted from the images before the test in Figs. 3, 5, and 6. The R_e, G_e, and B_e were also the average RGB values extracted from the images at the end of hydrogen charging. The average RGB values were obtained from the area corresponding to the hydrogen charging part on the hydrogen entry side (the dashed-areas in Figs. 3, 5, and 6).

Figure 8 shows the onset time of the color change and the color difference after hydrogen charging on the hydrogen detection side as a function of the thickness of the Pd intermediate layers. When the Pd thickness was 4 nm, the onset time of the average color change was the smallest. On the other hand, the onset time of the color change at Point A, B, and C increased as increasing the Pd thickness, that is, the responsivity is improved by decreasing the Pd thickness. The difference between the average and the local color change for the specimen with the Pd thickness of 2.5 nm is thought to be caused by the uneven distribution of the color change observed in Fig. 6. The color difference increased with decreasing the Pd thickness. The sensitivity is also improved by decreasing the Pd thickness. However, in the case of the Pd thickness of 2.5 nm, the degree of the dispersion for the color change was too large, and a detrimental effect of the spatial resolution is estimated. Taking into account the responsivity, the sensitivity, and the spatial resolution comprehensively, the best thickness of the Pd intermediate layer seems to be 4 nm in this study.

Fig. 8.

Onset time of the color change and color difference after hydrogen charging on the hydrogen detection side with the thickness of Pd intermediate layers. The average value of the onset time of the color change was obtained from Fig. 7. Point A, B, and C are shown in Figs. 3, 5, and 6, respectively. The color difference was calculated with the average RGB values obtained from the images at the start and end of hydrogen charging in Figs. 3, 5, and 6. (Online version in color.)

3.4. TEM Observation of Fe/Pd/WO₃ Interfaces

In order to discuss the effect of the Pd thickness on the hydrogen detection properties, TEM observations were made on the hydrogen detection side. Figure 9 shows the cross-sectional TEM images of Fe/Pd/WO₃ interfaces of the specimens with the Pd intermediate layer 4 nm and 2.5 nm in thickness. In Fig. 9(a), the difference of the contrast was clearly observed. The left area was pure iron, and the right one was WO₃. WO₃ appeared to have an extremely fine crystal structure. Two layers were observed between Fe and WO₃. From EDS analysis, the layer in the WO₃ side was Pd. The Pd layer 4 nm in thickness was smoothly and evenly formed. The layer in the Fe side was identified as an air-formed native Fe oxide by EDS analysis. The native oxide of Fe is possible to affect the detection properties, such as the responsivity, the sensitivity, and the spatial resolution. The influence of the native oxide of Fe on the hydrogen detection will be discussed in our next paper.

Fig. 9.

Cross-sectional TEM images of Fe/Pd/WO₃ interfaces of the specimens with the Pd intermediate layer (a) 4 nm and (b) 2.5 nm in thickness.

In Fig. 9(b), the Pd layer 2.5 nm in thickness was observed between the Fe oxide and WO₃. Small contrasts appeared to exist in the Pd layer, and the layer seemed to be consisted of nano-particles of Pd ca. 2 nm in diameter. In this case, regions where the WO₃ was directly formed on the pure iron surface, was thought to be produced. Because the existence of Pd is essential to the hydrogen detection using WO₃, the heterogeneity of the Fe/Pd/WO₃ interfaces seemed to affect the uneven distribution of the color change in Fig. 6. In the case of the Pd thickness of more than 4 nm, smooth and even Pd layer is expected to form. Because the hydrogen solubility of Pd is high²⁹⁾ and the diffusion coefficient of hydrogen in Pd is small,³²⁾ it is thought that the responsivity and the sensitivity of the hydrogen detection was enhanced as the Pd thickness decreased.

In this paper, we clarified the effect of the thickness of the Pd intermediate layer on the responsivity and the sensitivity of the hydrogen detection. Based on the results, the responsivity and the sensitivity for the hydrogen mapping technique using WO₃ was well-improved. However, the spatial resolution is still insufficient for practical applications such as hydrogen mapping for living steel structures and analysis of microstructural effect on hydrogen uptake. Further study is on-going to improve the spatial resolution for the hydrogen mapping technique using WO₃.

4. Conclusions

(1) No color change was observed on the hydrogen detection side of the specimens with the Pd intermediate layer more than 25 nm in thickness during the hydrogen detection test for 7.2 ks.

(2) The responsivity for the hydrogen mapping technique using WO₃ was improved by decreasing the Pd thickness. However, the onset time of the average color change was the smallest when the Pd thickness was 4 nm because of the uneven distribution of the color change in the case of the Pd thickness of 2.5 nm.

(3) The sensitivity for the hydrogen mapping technique using WO₃ was improved by decreasing the Pd thickness.

(4) Taking into account the responsivity, the sensitivity, and the spatial resolution comprehensively, the best thickness of the Pd intermediate layer seems to be 4 nm in this study.

(5) The heterogeneity of the Fe/Pd/WO₃ interfaces is thought to affect the uneven distribution of the color change.

Acknowledgments

This work was performed under the support of a Grant-in-Aid for Challenging Exploratory Research from Japan Society for the Promotion of Science (grant No. 26630357) and the ISIJ Research Promotion Grant from The Iron and Steel Institute of Japan.

References

• 1)   H.  Berns and  W.  Theisen: Ferrous Materials, Springer-Verlag, Berlin, (2008), 178.
• 2)   E.  Akiyama: ISIJ Int., 52 (2012), 307.
• 3)   J.  Kinugasa,  F.  Yuse,  M.  Tsunezawa and  M.  Nakaya: ISIJ Int., 56 (2016), 459.
• 4)   T.  Kushida: ISIJ Int., 43 (2003), 470.
• 5)   T.  Omura: ISIJ Int., 52 (2012), 267.
• 6)   M.  Koyama,  M.  Rohwerder,  C. C.  Tasan,  A.  Bashir,  E.  Akiyama,  K.  Takai,  D.  Raabe and  K.  Tsuzaki: Mater. Sci. Technol., 33 (2017), 1481.
• 7)   J.  Ovejero-García: J. Mater. Sci., 20 (1985), 2623.
• 8)   K.  Ichitani,  M.  Kanno and  S.  Kuramoto: ISIJ Int., 43 (2003), 496.
• 9)   T.  Schober and  C.  Dieker: Metall. Trans. A, 14 (1983), 2440.
• 10)   J.  Yao and  J. R.  Cahoon: Metall. Trans. A, 21 (1990), 603.
• 11)   T.  Tsuru and  R. M.  Latanision: Scr. Metall., 16 (1982), 575.
• 12)   K.  Takai,  Y.  Chiba,  K.  Noguchi and  A.  Nozue: Metall. Mater. Trans. A, 33 (2002), 2659.
• 13)   J.  Takahashi,  K.  Kawakami,  Y.  Kobayashi and  T.  Tarui: Scr. Mater., 63 (2010), 261.
• 14)   R. F.  Schaller,  S.  Thomas,  N.  Birbilis and  J. R.  Scully: Electrochem. Commun., 51 (2015), 54.
• 15)   T.  Haruna,  S.  Ohtsuka and  T.  Shibata: ISIJ Int., 43 (2003), 482.
• 16)   S.  Evers,  C.  Senöz and  M.  Rohwerder: Sci. Technol. Adv. Mater., 14 (2013), 014201.
• 17)   S.  Evers and  M.  Rohwerder: Electrochem. Commun., 24 (2012), 85.
• 18)   S.  Evers,  C.  Senöz and  M.  Rohwerder: Electrochim. Acta, 110 (2013), 534.
• 19)   G.  Williams,  H. N.  McMurray and  R. C.  Newman: Electrochem. Commun., 27 (2013), 144.
• 20)   R. F.  Schaller and  J. R.  Scully: Electrochem. Commun., 40 (2014), 42.
• 21)   H.  Katayama: ISIJ Int., 56 (2016), 478.
• 22)   C.  Senös,  S.  Evers,  M.  Stratmann and  M.  Rohwerder: Electrochem. Commun., 13 (2011), 1542.
• 23)   M.  Koyama,  A.  Bashir,  M.  Rohwerder,  S. V.  Merzlikin,  E.  Akiyama,  K.  Tsuzaki and  D.  Raabe: J. Electrochem. Soc., 162 (2015), C638.
• 24)   R. F.  Schaller and  J. R.  Scully: Electrochem. Commun., 63 (2016), 5.
• 25)   S.  Ajito,  T.  Hojo,  M.  Koyama,  K.  Fujita and  E.  Akiyama: Int. J. Hydrog. Energy, 45 (2020), 25580.
• 26)   Y.  Sugawara,  Y.  Sakaizawa,  A.  Shibata,  I.  Muto and  N.  Hara: ISIJ Int., 58 (2018), 1860.
• 27)   S. K.  Deb: Philos. Mag., 27 (1973), 801.
• 28)   B. W.  Faughnan,  R. S.  Crandall and  P. M.  Heyman: RCA Rev., 36 (1975), 177.
• 29)   Y. C.  Huang,  K.  Fujita and  H.  Uchida: Bull. Jpn. Inst. Met., 18 (1979), 694 (in Japanese).
• 30)   S. K.  Konda and  A.  Chen: Mater. Today, 19 (2016), 100.
• 31)   H.  Kaneko,  K.  Miyake and  Y.  Teramoto: J. Appl. Phys., 53 (1982), 4416.
• 32)   H.  Iwaoka,  M.  Arita and  Z.  Horita: Acta Mater., 107 (2016), 168.