Formation and Growth of Iron Nuclei on Wüstite Surface at the Initial Stage of Reduction

Abstract

Formation and growth of iron nuclei on the (001) surface of Fe_1–xO by hydrogen ion bombardment were observed using scanning tunneling microscope (STM) and low energy electron diffraction (LEED) in order to understand the change in the nano-level structure of the Fe_1–xO surface at the initial stage of reduction. It has been found from the STM images that prior to the formation of iron nuclei, hydrogen ion bombardment increases the number of large dents called “depression” on the terrace of the Fe_1–xO surface, which may imply that O atoms are deficient on the surface. Simultaneously, the number of steps on the Fe_1–xO surface increases probably owing to the reconstraction of the surface structure. According to the LEED pattern, the (2×2) long range periodic structures have appeared which may also demonstrate that the O atom-deficient surface is produced by hydrogen ion bombardment. As the reduction proceeds, the iron nuclei with the shape of a slightly rounded truncated four-sided pyramid are preferentially formed on the surface area where the steps are densified. The side length and the height of the pyramids are 4–8 nm and 1.2–1.8 nm, respectively. Subsequently, the iron nuclei are also formed on the terraces.

1. Introduction

Iron ore is reduced to metallic iron by a consecutive reaction through Fe₃O₄ and wüstite. Among reactions from iron oxide to metallic iron, the reaction from wüstite to Fe has been intensively elucidated by many researchers because the reaction controls the overall reduction rate at alternative coke/ore layers in the blast furnace shaft around 1300 K. Inami and Suzuki^1,2) have investigated the reduction rate of dense wüstite plate to iron with CO–CO₂ and/or CO–CO₂–Ar gas mixtures and have found that the reduction rate is slow before the formation of metallic iron, while the rate becomes markedly faster after the formation of metallic iron. Their results imply that the formation of iron nuclei on the wüstite surface controls the reduction reaction rate from wüstite to Fe. In order to know the mechanism of the formation of iron nuclei on wüstite surface, it may be important to observe the change in the wüstite surface at the initial stage of reduction reaction at the nanometer level. Ishikawa et al.³⁾ have carried out the in-situ observation of the reduction process of wüstite to iron using hydrogen ion implantation in transmission electron microscope, and have found that particles having the size of several to dozens of nanometers precipitate on the wüstite, which are probably iron. They have also reported that the smaller precipitates have the crystallographic orientation related to that of the wüstite matrix, on the other hand, the larger precipitates do not have such a distinctive orientation. Their experimental method is an innovative approach. However, they have not investigated either the structure change in the wüstite surface during the reduction process or the preferential site for iron nucleation on the wüstite surface.

Wüstite (Fe_1–xO) is a nonstoichiometric compound which is stable above 833 K.⁴⁾ The measure of the deviation from stoichiometry, x, widely ranges from 0.05 to 0.15 depending on the temperature and the oxygen partial pressure. The crystal structure of the bulk form of Fe_1–xO is the rock-salt type structure. Fe vacancies and Fe³⁺ ions at interstitial sites make defect clusters.⁵⁾ According to the previous results obtained by X-ray diffraction, neutron diffraction, transmission electron microscope (TEM) and computer simulation, defect clusters are periodically arranged every 2.5–2.7a₀ (a₀ is the lattice parameter of the rock-salt type FeO).^{6,7,8,9,10,11,12)} However, there are only limited reports on the surface structures of bulk Fe_1–xO. The surface crystal structures of Fe_1–xO were only studied in terms of a thin film on a substrate using scanning tunneling microscope (STM), low energy electron diffraction (LEED) and/or Auger electron spectroscopy (AES).^{13,14,15,16,17,18,19,20)} The effect of the nonstoichiometry, i.e., Fe vacancies on the surface, was not referred in the previous reports despite the fact that the stoichiometric FeO (x = 0) cannot exist in the equilibrium state.

The authors’ group has analyzed the surface crystal structures of bulk Fe_1–xO using STM and LEED in order to study the influence of Fe vacancy defect clusters on surface structures. Watanabe et al. have observed a mesh-like long period arrayed structure as well as some depressions in the mesh structure on a surface of polycrystalline Fe_1–xO.²¹⁾ Masaki et al. have also found from the observation of a polycrystalline Fe_1–xO surface that zigzag step edges run along the <110> directions and high-index planes are composed of (001) facets.²²⁾ Masaki et al. have also been succeeded in the observation of a single crystalline Fe_1–xO surface and have considered that periodically arrayed dents forming the mesh-like structure correspond to Fe vacancy pairs.²³⁾

On the other hand, the change in the nanostructure of the Fe_1–xO surface at the initial stage of reduction reaction has not been observed yet. The aim of this study is to observe the formation and growth of iron nuclei on a single crystalline Fe_1–xO surface at the initial stage of reduction reaction at the nanometer level using STM and LEED. Hydrogen gas was used as a reducing gas. It is known that the reaction between gas and solid hardly takes place in an ultrahigh vacuum. Therefore, the reaction was generated by irradiating ionized hydrogen gas onto the Fe_1–xO surface.

2. Experimental

2.1. Sample Preparation

Powder mixtures of Fe (99.9% purity) and Fe₂O₃ (99.99% purity) corresponding to a Fe_0.91O stoichiometry of the protoxide were pressed and formed into a rod (12 mm in diameter and 50 mm in length). It was sintered in vacuo at 1373 K for 24 h and then rapidly cooled at room temperature. A single crystal of Fe_1–xO was grown from the sintered rod in an optical floating zone furnace (Canon Machinery Inc., Desktop Single Crystal Growth) in a flow of pure argon (99.99% purity) deoxidized by zirconium sponge heated at 1073 K. The (001) orientation of the single crystal was analyzed using the X-ray pole figure (Rigaku Co. RINT-TTR-3C/PC & MPA-2000) and cut into pieces of 1 mm in thickness. The Fe_1–xO(001) samples on an alumina plate were placed in an electrical resistance furnace and equilibrated at 1373 K for 24 h in a flow of a CO–CO₂ mixture (P_CO/P_CO2 = 1.96, which is equivalent to the oxygen partial pressure of 1.01×10^–8 Pa). The Fe_1–xO(001) samples were then quenched to room temperature.

The annealed samples were subjected to the powder XRD technique (Rigaku Co. RINT-TTR-3C/PC) to confirm the phase purity of Fe_1–xO and to determine the lattice parameter. The lattice parameter of Fe_1–xO was calculated to be 4.295 Ǻ by means of the Nelson-Riley function using diffraction peaks of (111), (200), (220), (311), (222), (400), (331) and (420). The chemical composition of the sample was determined to be Fe_0.92O from the relation between the lattice parameter and the chemical composition of Fe_1–xO.^24,25,26,27)

The Fe_0.92O(001) samples were formed into a reed shape (3 mm × 13 mm) and polished using several diamond pastes with different grain sizes and the final polishing was performed using a diamond paste with 0.5 μm grains and then rinsed in acetone using an ultrasonic bath for 30 min.

2.2. Surface Observation of a Fe_0.92O Sample before Hydrogen Ion Bombardment

The surface of a Fe_0.92O sample was analyzed in an UHV chamber equipped with STM (Unisoku Co. USM-1100SX3) and LEED/AES optics (OCI Inc. BDL800IR). Figure 1 shows a photograph (a) and a schematic illustration (top view) (b) of UHV system equipped with STM and LEED/AES optics. The system is set on an anti-vibration table, consisting of three chambers; the lord-lock chamber, the preparation chamber and the observation chamber. The attainable degrees of vacuum for the lord-lock chamber, the preparation chamber and the observation chamber are 1.3×10^–5, 2.6×10^–8 and 1.3×10^–8 Pa, respectively. STM is installed in the observation chamber, and LEED as well as sputter ion gun are incorporated in the preparation chamber. A sample is introduced into the lord-lock chamber. Then, it is transferred into the preparation chamber and annealed in vacuo to evaporate the adsorbed matter from the sample surface. The STM observation is carried out in the observation chamber.

Fig. 1.

(a) Photograph and (b) schematic illustration (top view) of UHV system equipped with STM and LEED/AES optics in the present study.

In this study, after a Fe_0.92O sample was mounted on a STM sample holder, a STM observation was carried out so as to confirm that abradant diamond particles did not remain on the sample surface. Subsequently, the sample was heated up to 1273–1323 K and annealed at the temperature for 3 min in vacuo by passing a direct current through the sample to evaporate the adsorbed matter from the sample surface. The temperature of the sample was measured by a two-color pyrometer through the quartz window of the chamber. After the sample was rapidly cooled down to room temperature, STM, LEED and AES analyses were performed to observe the nanostructure of sample surface, to elucidate the periodicity of surface structure and to confirm the surface cleanness, respectively.

The Pt-Ir tips were used for the STM observation. All STM data were obtained in the constant current mode. The positions of the tips were calibrated by comparison with the size and the step height of the Si(111)-(7×7) structure.²⁸⁾ The bias voltage applied to the sample was set to be 0.15–2.30 V, where Fe sublattice can be observed as bright dots.

2.3. Reduction of Fe_0.92O Samples by Hydrogen Ion Bombardment

Prior to the reduction experiment, Fe_0.92O samples were annealed in the same manner as aforementioned in the section 2.2 in order to evaporate the adsorbed matter from the sample surface. Subsequently, while the samples were heated at 1073 K, hydrogen gas was introduced to the chamber causing the total pressure to be 1.3×10^–6 Pa. Hydrogen gas was ionized by the sputter ion gun, and ionized hydrogen gas was irradiated onto the sample surface. Each time the total irradiation duration reached 60, 120, 300, 600, 1200, 1800, 3600 and 7200 s, the hydrogen gas flow was stopped, and STM, LEED and AES analyses were carried out after making sure that the degree of vacuum returned to 1.3×10^–8 Pa or less. As for the sample after 7200 s of hydrogen ion bombardment, the sample surface was lustered, indicating that metallic iron was produced on the surface, which was also confirmed by the electron probe microanalyzer (EPMA). The surface was also too rough to be imaged by STM, therefore, LEED analysis was only conducted. In fact, the reduction progress does not only depend on the hydrogen ion irradiation duration, but also varies depending on the positions of an identical sample. This is because a sample has a temperature distribution during reduction experiment, which is inevitable owing to the apparatus problem.

3. Results and Discussion

3.1. Surface Observation of a Fe_0.92O Sample before Hydrogen Ion Bombardment

Figure 2 shows LEED patterns from the Fe_0.92O sample after annealing at 1273 K for 3 min in UHV. The FeO(001)-p(1×1) diffraction spots and the <110> direction streaks can be observed, which correspond to the rocksalt-type FeO structure and step edges running along the <110> directions, respectively.

Fig. 2.

LEED patterns from the Fe_0.92O (001) surface after annealing at 1273 K for 3 min in UHV with the change in the incident electron energies of (a) 91.3 eV, (b) 83.3 eV, (c) 72.4 eV.

Figure 3 shows STM images of the Fe_0.92O surface after annealing at 1273 K for 3 min in UHV. It can be seen that the surfaces consist of many steps having the zigzag shaped edges along <110> direction. It has been found from the line profile that the step height and the distance between two adjacent steps vary in the ranges between 0.3 to 1.0 nm and between 5 to 30 nm, respectively. There are relatively flat places between adjacent steps, which are termed “terrace” here. On the terraces, the relatively periodically arrayed dents (small dark dots), the intervals of which are 0.9–1.2 nm, can be observed (see Fig. 3(c)). Addition to these, scarce larger dark parts are also observed, which correspond to “depressions” reported by Watanabe et al..²¹⁾ Figure 4 shows a magnified STM image of Fe_0.92O surface, more clearly showing the mesh-like structure.^21,22,23) The dents have corn shape, which correspond to Fe vacancy pairs.²³⁾

Fig. 3.

(a), (c) STM images of the Fe_0.92O (001) surface after annealing at 1273 K for 3 min in UHV taken at sample bias 0.35 V, tunneling current 0.45 nA. (b) Line profile between A and B in (a).

Fig. 4.

17.85 nm × 17.85 nm STM image of the Fe_0.92O (001) surface at center of the sample annealed at 1273 K for 3 min.

3.2. Structure Change in Fe_1–xO Sample Surface Caused by Hydrogen Ion Bombardment before the Formation of Iron Nuclei

As described below, the inspection of STM images (Figs. 5, 6, 7) demonstrate that the structure change in Fe_1–xO sample surface during the reduction by hydrogen ion bombardment goes through the following two stages: (i) O atoms are removed from the sites bonded with fewer Fe atoms, that is, O atoms at step edges and/or O atoms adjacent to Fe vacancy pairs. (ii) Fe²⁺ ions on the surface where O atoms are deficient diffuse inside as well as to surrounding area because Fe atom-exposed surfaces are electrically unstable.

Fig. 5.

STM images of the Fe_1–xO (001) surface after 60 s of hydrogen ion bombardment. Sample biases and tunneling currents are (a) 0.35 V, 0.30 nA, (b) 0.30 V, 0.30 nA and line profile between A and B in (b).

Fig. 6.

STM image of a terrace of the Fe_1–xO (001) surface after 120 s of hydrogen ion bombardment. Sample bias and tunneling current are 0.40 V and 0.30 nA.

Fig. 7.

STM images of the steps on the Fe_1–xO (001) surface after hydrogen ion bombardment of (a) 120 s and (b) 300 s at 1073 K. Sample biases and tunneling currents are (a) 0.50 V, 0.25 nA and (b) 0.60 V, 0.20 nA.

Figure 5 shows STM images of the Fe_1–xO surface after 60 s of hydrogen ion bombardment. As shown in Fig. 5(a), in most parts, the images are similar to those of Fe_1–xO surface before hydrogen iron bombardment, consisting of many steps and terraces. However, in some parts, collapsed steps can be observed, as shown in Fig. 5(b). It has been found from the line profiles that a collapsed step is elevated above the level of surrounding by ca. 0.8 nm. This is considered to be due to an apparent phenomenon caused by an increase in tunnel current between a Pt-Ir tip and the sample where the density of Fe atoms becomes larger by removing O atoms.²⁹⁾ Namely, the collapsed step may correspond to the O atom deficient area produced by removing O atoms from step edges.

Figure 6 shows a STM image of a terrace of the Fe_1–xO surface after 120 s of hydrogen ion bombardment. It can be seen that the number of the dents (small dark dots) decreases, while the number of depressions (larger dark parts) increases. The depressions may be formed by removing O atoms sited in the vicinity of Fe vacancy pairs.

Figure 7 shows STM images of Fe_1–xO surfaces after 120 s (a) and 300 s (b) of hydrogen ion bombardment. It can be seen that as increasing the irradiation duration, the steps are densified and the step intervals become shorter. This may be due to the diffusion of Fe²⁺ ions from the O-atom deficient surface: The change in the step structure with time is schematically shown in Fig. 8. When the Fe_1–xO surface in Fig. 8(a) is reduced, O atoms at the step edges are removed preferentially because these O atoms are bonded with fewer Fe atoms. As a result, more Fe atoms become exposed on the Fe_1–xO surface than O atoms, foming Fe atom-exposed surface such as (111) surface (see Fig. 8(b)). This corresponds to the STM image of Fig. 5(b). Since such Fe atom-exposed surfaces are electrically unstable, Fe atom-exposed surfaces are easily reconstructed to the electroneutral surfaces, i.e., the (001) surface by diffusing Fe²⁺ ions inside as well as to surrounding area, and reversely diffusing Fe vacancies to the surface. The reconstruction of the surface may lead to the step densification, as shown in Fig. 8(c).

Fig. 8.

Schematic illustrations of the step structure change with time.

3.3. Formation and Growth of Iron Nuclei

It has been found from STM images that nanocrystals appear on the Fe_1–xO surface after the reduction sufficiently progresses by hydrogen ion bombardment.

Figure 9 shows a STM image of the Fe_1–xO surface after 60 s of hydrogen ion bombardment, which shows nanocrystals on the surface. The nanocrystals seem to have the shape of a slightly rounded truncated four-sided pyramid, the side length and the height of which are 4–8 nm and 1.2–1.8 nm, respectively. It is found that the sides of truncated pyramid shaped nanocrystals run along the <110> directions. In fact, Silly and Castell have investigated the structure and morphology of self-assembled iron nanocrystals supported on a SrTiO₃(001)-c(4×2) substrate using STM, and have reported that Fe nanocrystals have a truncated pyramid shape with a bcc structure.³⁰⁾ In the bcc structure, the lowest energy facets of Fe are the (001) and the (011) facets, which means that the equilibrium truncated pyramid shape has a (001) top facet and four (011) side facets. In fact, the commensurate interface crystallography between wüstite and α-Fe is reported to be (100)_wüstite||(100)_α-Fe, [110] _wüstite||[100]_α-Fe,³¹⁾ which is in good agreement with the facts that truncated pyramid shaped nanocrystals formed on Fe_1–xO (001) surfaces have a (001) top facet, and that the sides of pyramids run along the <110> directions. Therefore, it can be considered that the nanocrystals in Fig. 9 are iron.

Fig. 9.

(a) STM image of the Fe_1–xO (001) surface after 60 s of hydrogen ion bombardment. (b) Line profile between A and B in (a).

Figure 10 shows STM images of the Fe_1–xO surface after 120 s (a) and 600 s (b) of hydrogen ion bombardment, which shows nanocrystals on the steps (a) and on the terraces (b). Firstly, iron nuclei are formed on the surface area where steps are densified, and array along the step directions. As the reduction reaction progresses, iron nuclei are formed on the terraces as well. The reason that iron nuclei are preferentially formed on the step-densified area is as follows: As illustrated in Fig. 8, the step densification is caused by the structural reconstruction of Fe atom-exposed surfaces such as the (111) surface by diffusing Fe²⁺ ions inside as well as to surrounding area, and reversely diffusing Fe vacancies to the surface. Nevertheless, the deviation from stoichiometric FeO, i.e., x in Fe_1–xO becomes smaller little by little around those areas and the Fe composition eventually exceeds the Fe/FeO equilibrium, yielding iron nuclei. As the reduction reaction progresses, the composition at the terrace also exceeds the Fe/FeO equilibrium, and iron nuclei are formed.

Fig. 10.

STM images of the Fe_1–xO (001) surface after hydrogen ion bombardment of (a) 120 s and (b) 600 s at 1073 K. Sample biases and tunneling currents are (a) 0.50 V, 0.25 nA and (b) 0.56 V, 0.30 nA.

3.4. (2×2) Long Range Periodic Structure Formed by Hydrogen Ion Bombardment

Figure 11 shows LEED patterns from the Fe_1–xO surfaces after 1200 s (a) and 3600 s (b) of hydrogen ion bombardment. It can be seen from Fig. 11(a) that (2×2) diffraction spots are superposed to the FeO(001)-p(1×1) diffraction spots due to the rocksalt-type FeO structure for a pattern from the Fe_1–xO surfaces after 1200 s of hydrogen ion bombardment. This means that a long range periodic structure appears addition to the rocksalt-type FeO structure after 1200 s of hydrogen ion bombardment. On the other hand, (2×2) diffraction spots disappear after 3600 s of hydrogen ion bombardment, as shown in Fig. 11(b). Inspection of Fig. 11(a) implies that the (2×2) long range periodic structure has the same crystallographic orientation as that of the FeO structure, and that the period length of (2×2) structure is twice as long as that of the FeO structure, i.e., twice as long as the distance between two adjacent Fe (or O) atoms, which corresponds to ca. 0.607 nm. Therefore, it can be considered that the (2×2) long range periodic structure is not due to iron nuclei formed on the Fe_1–xO surface as the lattice constant of α-Fe is reported to be 0.286 nm.³²⁾ The (2×2) long range periodic structure may be associated with the O deficient FeO surface; one possible structure producing (2×2) diffraction spots in Fig. 11(b) is shown in Fig. 12. The reason (2×2) diffraction spots disappear after 3600 s of hydrogen ion bombardment may be that iron nuclei covers the O atom-deficient FeO surface. However, further study is required to determine the exact structure of (2×2) diffraction spots in the future.

Fig. 11.

LEED patterns from the Fe_1–xO (001) surface after hydrogen ion bombardment of (a) 1200 s and (b) 3600 s at 1073 K.

Fig. 12.

Schematic image of one possible structure producing (2×2) diffraction spots.

4. Conclusions

It has been reported that the formation of iron nuclei on the wüstite surface controls the reduction reaction rate from Fe_1–xO to Fe. In order to know the mechanism of the formation of iron nuclei on the wüstite surface, it may be important to observe the change in the Fe_1–xO surface at the initial stage of reduction reaction at the nanometer level. In this study, formation and growth of iron nuclei on the (001) surface of Fe_1–xO by hydrogen ion bombardment were observed using scanning tunneling microscope (STM) and low energy electron diffraction (LEED). The followings are the results.

(1) It has been found from the STM images that prior to the formation of iron nuclei, hydrogen ion bombardment increases the number of large dents called “depression” on the terrace of the Fe_1–xO surface, which may imply that O atoms are deficient on the surface. Simultaneously, the number of steps on the Fe_1–xO surface increases probably owing to the reconstraction of the surface structure.

(2) STM images also show that as the reduction proceeds, the iron nuclei with the shape of a slightly rounded truncated four-sided pyramid are preferentially formed on the surface area where the steps are desified. The side length and the height of the pyramids are 4–8 nm and 1.2–1.8 nm, respectively. Subsequently, the iron nuclei are also formed on the terraces.

(3) According to the LEED pattern, the (2×2) long range periodic structures have appeared which may demonstrate that the O atom-deficient surface is produced by hydrogen ion bombardment. However, further study is required to determine the exact structure of (2×2) diffraction spots.

Acknowledgement

This project was financially supported by a research promotion grant of the Iron and Steel Institute of Japan (ISIJ).

References

• 1)   T.  Inami and  K.  Suzuki: Tetsu-to-Hagané, 80 (1994), 699.
• 2)   T.  Inami and  K.  Suzuki: Tetsu-to-Hagané, 81 (1995), 1037.
• 3)   N.  Ishikawa,  M.  Song,  K.  Mitsuishi,  K.  Furuya,  Y.  Watanabe and  T.  Inami: ISIJ Int., 44 (2004), 2029.
• 4)   A.  Muan and  E. F.  Osborn: Phase Equilibria among Oxides in Steelmaking, Addison-Wesley, MA, (1965), 28.
• 5)   W. L.  Roth: Acta Cryst., 13 (1960), 146.
• 6)   P. J.  Manenc: J. Phys. Radium, 24 (1963), 447.
• 7)   F.  Koch and  J. B.  Cohen: Acta Cryst., B25 (1969), 275.
• 8)   B.  Andersson and  J. O.  Sletnes: Acta Cryst., A33 (1977), 268.
• 9)   C.  Lebreton and  L. W.  Hobbs: Radiat. Effects, 74 (1983), 227.
• 10)   T.  Ishiguro and  S.  Nagakura: Jpn. J. Appl. Phys., 24 (1985), L723.
• 11)   T. R.  Welberry and  A. G.  Christy: Phys. Chem. Miner., 24 (1997), 24.
• 12)   L.  Minervini and  R. W.  Grimes: J. Phys. Chem. Solids, 60 (1999), 235.
• 13)   A. J.  Pignocco and  G. E.  Pellissier: J. Electrochem. Soc., 112 (1965), 1188.
• 14)   C.  Leygraf and  S.  Ekelund: Surf. Sci., 40 (1973), 609.
• 15)   G. W.  Simmons and  D. J.  Dwyer: Surf. Sci., 48 (1975), 373.
• 16)   D.  Cappus,  M.  Habel,  E.  Neuhaus,  M.  Heber,  F.  Rohr and  H. J.  Freund: Surf. Sci., 337 (1995), 268.
• 17)   Y. J.  Kim,  C.  Westphal,  R. X.  Ynzunza,  Z.  Wang,  H. C.  Galloway,  M.  Salmeron,  M. A.  Van Hove and  C. S.  Fadley: Surf. Sci., 416 (1998), 68.
• 18)   S.  Shaikhutdinov,  M.  Ritter and  W.  Weiss: Phys. Rev., B62 (2000), 7535.
• 19)   G.  Ketteler and  W.  Ranke: J. Phys. Chem., B107 (2003), 4320.
• 20)   E. L.  Lopes,  G. J. P.  Abreu,  R.  Paniago,  E. A.  Soares,  V. E.  de Carvalho and  H. D.  Pflannes: Surf. Sci., 601 (2007), 1239.
• 21)   T.  Watanabe,  K.  Kitamura,  T.  Yamashita and  K.  Nagata: Steel Res. Int., 81 (2010), 814.
• 22)   Y.  Masaki,  T.  Watanabe,  T.  Yamashita,  M.  Hayashi and  K.  Nagata: ISIJ Int., 49 (2009), 74.
• 23)   Y.  Masaki,  T.  Fujii,  M.  Hayashi and  K.  Nagata: ISIJ Int., 51 (2011), 203.
• 24)   R. L.  Levin and  J. B.  Wagner, Jr.: Trans. Metal. Soc. AIME, 236 (1966), 516.
• 25)   E. R.  Jette and  F.  Foote: J. Chem. Phys., 1 (1933), 29.
• 26)   V.  Cirilli and  C.  Brisi: Ann. Chim. Appl., 41 (1951), 508.
• 27)   F. K.  Foster and  A. J. E.  Welch: Trans. Faraday Soc., 52 (1956), 1626.
• 28)   K.  Takayanagi,  Y.  Tanishiro,  M.  Takahashi and  S.  Takahashi: Surf. Sci., 164 (1985), 367.
• 29)   T.  Shimizu and  M.  Tsukada: Solid State Commun., 87 (1993), 193.
• 30)   F.  Silly and  M. R.  Castell: Appl. Phys. Lett., 87 (2005), 063106.
• 31)   S.  Taniguchi: Bull. Iron Steel Inst. Jpn., 13 (2008), 35.
• 32)   G. B.  Cho,  M.  Yamamoto and  Y.  Kameda: J. Appl. Phys., 41 (2002), 5713.