Effect of Formic Acid on Corrosion Behavior of STBA24 Low-Alloyed Steel and Its Weldment in Simulated Boiler Water Containing Chloride Ions

Abstract

The effect of formic acid (HCOOH) in simulated boiler water containing chloride ions (Cl⁻) on the corrosion behavior of STBA24 low-alloyed steel and its weldment was investigated using electrochemical corrosion tests, immersion corrosion tests and surface analyses. The addition of 100 ppm HCOOH into water containing 100 ppm Cl⁻ resulted in thicker films with poor corrosion resistance that were formed easily on both the base metal and the weldment of STBA24, even though pit initiation was inhibited within a short time. The results suggest that the presence of 100 ppm formic acid in the simulated boiler water containing 100 ppm chloride ions promotes corrosion of the materials tested.

Fig. 7 Time variations of the OCPs of the specimens.

1. Introduction

Typically corrosive species such as chloride ions (Cl⁻) are invariably mixed into the boiler water in thermal power plants, which promotes corrosion of the boiler component materials in the case of inadequate boiler water treatment or accidental seawater leakage. If boiler water is adulterated with a certain amount of Cl⁻ ions, then the pH of the boiler water decreases and the risk of corrosion increases.¹⁾ It has also been reported that the Cl⁻ in the boiler water, even under invariably high pH conditions, lowers the corrosion potentials of boiler tube steels and exhibits a corrosion promoting effect.²⁾

Many carbon steels and low-alloyed steels are used as boiler component materials in thermal power plants. To suppress the corrosion of these materials, pH maintenance of the boiler water as alkaline is the most effective approach because it stabilizes passivation films (mainly composed of magnetite; Fe₃O₄) on the surfaces of the steels in a deaerated boiler water and reduces the corrosion rate.³⁾ Therefore, the boiler water is typically adjusted to pH 9.0 or more using ammonia (NH₃). However, ammonia has high relative volatility (RV) and is easily transformed to the vapor phase, so that the pH of the boiler water decreases during plant operation and corrosion damage such as flow-accelerated corrosion (FAC) of the boiler tube materials frequently occurs. Therefore, to suppress corrosion damage such as FAC of boiler tube materials, there is a tendency to add organic amine additives such as ethanolamine (ETA; OHC₂H₄NH₂) into the boiler feed water for pH adjustment.⁴⁾ However, at high temperatures the organic amines decompose to produce organic acids such as formic acid (HCOOH) and acetic acid (CH₃COOH); therefore, the corrosion of boiler equipment materials due to these organic acids has also been of concern.^5,6)

It has thus become very important to protect the equipment materials in power plants from corrosion due to organic acids as well as typical corrosive species such as Cl⁻ in boiler water. In addition, various welded components are used for boiler equipment; therefore, there is also concern regarding corrosion caused by unevenness of weldment (WM) microstructures due to welding processes. However, the influence of these impurities in boiler water, particularly the combined effects of organic acids and Cl⁻ on the corrosion behavior of boiler equipment materials including WMs, has not been sufficiently clarified. In this work, the corrosion behavior of STBA24 low-alloyed steel and associated WMs for boiler tube materials in simulated boiler water with added Cl⁻ and HCOOH was investigated using electrochemical corrosion tests, immersion corrosion tests, and surface analyses.

2. Experimental

2.1 Materials and methods

The materials used in this work were STBA24 low-alloyed steel and its WM, which are used for boiler tubes in thermal power plants. Table 1 lists the chemical compositions of the base metal (BM) and the deposited metal (DM). Auto-TIG (tungsten inert gas) welding was applied for the WM, similar to actual boiler tube steels. The temperature between passes was 150°C or less, and post-weld heat treatment (PWHT) was not performed. Figure 1 shows optical micrographs of the different microstructures of the BM, DM, and the heat-affected zone (HAZ) of the WM.

Table 1 Chemical composition of STBA24 and its WM.

Fig. 1

Optical micrographs of STBA24 WM. (a) BM; (b) HAZ; (c) DM.

Specimens with a size of 35 × 15 × 2 mm³ were cut from the BM and WM specimens of unused boiler tubes. The widths of the DM and HAZ in the cut-out WM specimens were ca. 6 and 3 mm, respectively. Prior to the tests, the specimens were abraded using 150–800 grit emery papers and then degreased with acetone. Figure 2 shows the geometry of the specimens. In addition, specimens for polarization tests (Fig. 2(b)) were isolation-coated with liquid room-temperature-vulcanizing (RTV) rubber (Shin-Etsu Chemical, 1 Component RTV) on their surfaces, except for a reaction area of 10 × 20 mm².

Fig. 2

Geometry of the specimens. (a) For immersion corrosion test and OCP measurement, (b) For polarization test.

2.2 Test water

The simulated all-volatile treatment (AVT) boiler water (pH9.5, dissolved oxygen (DO) < 7 ppb) was prepared as the base water according to JIS B8223.⁷⁾ First, 1.2 L pure ion-exchanged water was continuously bubbled with nitrogen (N₂) gas for 30 min or longer at room temperature. After the DO became ca. 20 ppb, 10 ppb hydrazine (N₂H₄) as well as Cl⁻ ions and HCOOH were added to the following two types of test water. Next, NH₃ solution was added to the water to adjust the pH to 9.5, and then the test water temperature was raised to 90°C. Note that the removal of DO was performed by the addition of 10 ppb N₂H₄ and also by the continuously bubbling with N₂ gas prior to the test and during the test at a flow rate of 100 mL/min.

Two types of test water (WQ1 and WQ2) were used in this work. Table 2 shows the quality of the test waters. WQ1 was prepared by the addition of 100 ppm Cl⁻ into the base water, while WQ2 was prepared by the addition of both 100 ppm Cl⁻ and 100 ppm HCOOH. Cl⁻ ions were added as powdered sodium chloride (NaCl). The temperature of the test waters was 90°C.

Table 2 Quality of the test waters.

2.3 Electrochemical corrosion test and surface analysis

Potentiodynamic polarization tests (cathodic polarization and anodic polarization) and open circuit potential (OCP) measurements were conducted using a potentiostat (Hokuto-Denko, HZ-3000) with three electrodes, the specimen as the working electrode, a platinum counter electrode, and a saturated KCl–Ag/AgCl reference electrode. Figure 3 shows a schematic diagram of the electrochemical corrosion test apparatus.

Fig. 3

Schematic diagram of the electrochemical corrosion test apparatus.

Cathodic polarization tests were performed on the BM specimens in the two different test waters. Polarization was started 30 s after the potential of the specimen in the test water became stable for 100 s. The specimens were polarized to −1.5 V (vs. Ag/AgCl) with a sweep rate of −20 mV/min. Anodic polarization tests were conducted on both the BM and the WM specimens. The specimens were firstly cathodic-treated for 10 min at −0.7 V (vs. Ag/AgCl) in the test water and then polarized with a sweep rate of 20 mV/min from the corrosion potentials to 1.5 V (vs. Ag/AgCl). After anodic polarization, the reaction surfaces were observed with a digital camera.

The OCP of the BM and WM specimens in the two test waters were continuously measured for 100 h. After OCP measurements, the specimen surfaces were observed with a digital camera and using scanning electron microscopy (SEM; Hitachi, S-4100). The films formed on the specimen surfaces were analyzed by electron probe microanalysis (EPMA; Shimadzu, EMPA-1610).

2.4 Immersion corrosion test

Immersion corrosion tests were performed on BM specimens for 100 h (4.17 d) in the WQ1 and WQ2 test waters to investigate the effect of formic acid on the corrosion rate in the boiler water containing Cl⁻ ions. After immersion tests, descaling was performed to remove the film (scale) formed on the specimens in the test water. Descaling was performed by immersing the test specimen in aqueous hydrochloric acid solution (35–37% HCl:pure water = 1:1) containing 0.05% thiourea for 15 s, and then washed with acetone.⁸⁾ The corrosion rates were calculated using eq. (1).   

\begin{equation} R = (W_{1} - W_{2})/St, \end{equation} (1)

where R (mdd; milligrams per square decimeter per day) is the corrosion rate, W₁ (mg) is the specimen mass before testing, W₂ (mg) is the specimen mass after descaling, S (dm²) is the entire surface area of the specimen, and t (d) is the immersion time.

3. Results and Discussion

3.1 Potentiodynamic polarization behavior

Figure 4 shows cathodic polarization curves of the BM specimens in the WQ1 and WQ2 test waters. At the beginning of polarization the specimen immersed in WQ2 showed a higher potential than that immersed in WQ1. The specimens in the two test waters showed almost the same current densities during cathodic polarization until ca. −0.7 V (vs. Ag/AgCl), after which the cathodic current densities of the specimens increased and the specimen in WQ2 exhibited a slightly higher cathodic current density than that in WQ1.

Fig. 4

Cathodic polarization curves for the BM specimens.

For STBA24 in the free immersion state in WQ1 and WQ2 with pH 9.5, the anodic and cathodic reactions that occurred on the specimen surfaces are shown in eqs. (2) and (3), respectively. The metal ions (Mⁿ⁺) are mostly Fe²⁺; therefore, the anodic reaction can be simply written as eq. (4).   

\begin{equation} \text{M} \to \text{M$^{\text{n+}}$} + \text{ne$^{-}$}, \end{equation} (2)

  

\begin{equation} \text{1/2O$_{2}$} + \text{H$_{2}$O} + \text{2e$^{-}$} \to \text{2OH$^{-}$}, \end{equation} (3)

  

\begin{equation} \text{Fe} \to \text{Fe$^{2+}$} + \text{2e$^{-}$}. \end{equation} (4)

The specimen in WQ2 showed a higher potential just before cathodic polarization than that in WQ1. This behavior was the same as that shown at the beginning of corrosion potential measurements in the following 3.2. During cathodic polarization until ca. −0.7 V (vs. Ag/AgCl), the cathodic current densities of the specimens in the two test waters were almost the same. Tsushima et al.⁹⁾ reported that the reduction reaction of Fe₃O₄ film on low-alloyed steels occurred in the test waters deaerated with N₂ during cathodic polarization. However, according to the potential-pH diagram of Fe,¹⁰⁾ the Fe-oxides in the present test waters with pH 9.5 were in a relatively thermodynamically stable state in this potential range. As the reduction reactions on the specimens in the two test waters occurred at the same lower rate until ca. −0.7 V (vs. Ag/AgCl), it is considered that there was almost no influence of the 100 ppm HCOOH in WQ2 on the reduction reactions.

However, after cathodic polarization until ca. −0.7 V (vs. Ag/AgCl), the cathodic reactions on the steel surfaces became the hydrogen generation reaction shown in eq. (5).^9,11,12) Figure 4 shows that in this region, the cathodic reaction rate was slightly higher in WQ2 than in WQ1.   

\begin{equation} \text{2H$_{2}$O} + \text{2e$^{-}$} \to \text{2OH$^{-}$} + \text{H$_{2}$}. \end{equation} (5)

Figure 5 shows anodic polarization curves of the BM and WM specimens in the two test waters. The appearance of the reaction surfaces after anodic polarization is shown in Fig. 6. Pitting corrosion was evident on all of the BM and WM specimens, and more pitting occurred in WQ1. Furthermore, all the BM and WM specimens in the two test waters showed passivation behavior, as shown by the polarization curves in Fig. 5. The specimens in WQ2 exhibited higher pitting potentials than those in WQ1. It is assumed that the formic acid in WQ2 suppressed the initiation of pitting corrosion. The formic acid in WQ2 dissociated according to the reaction shown in eq. (6). Therefore, both formate ions (COOH⁻) and ammonium formate (NH₄COOH) (eqs. (7) and (8)) were present in the WQ2 test water. It is considered that the competitive adsorption between COOH⁻ and Cl⁻ inhibited the destruction of the passivation film by Cl⁻ and therefore suppressed the initiation of pitting corrosion. It is also presumed that the physisorption of NH₄COOH inhibited corrosion in the WQ2 test water.^13,14)   

\begin{equation} \text{HCOOH} \to \text{COOH$^{-}$} + \text{H$^{+}$}, \end{equation} (6)

  

\begin{equation} \text{NH$_{3}$} + \text{H$_{2}$O} \to \text{NH$_{4}{}^{+}$} + \text{OH$^{-}$}, \end{equation} (7)

  

\begin{equation} \text{NH$_{4}{}^{+}$} + \text{COOH$^{-}$} \to \text{NH$_{4}$COOH}. \end{equation} (8)

Fig. 5

Anodic polarization curves for the BM and WM specimens.

Fig. 6

Appearance of the reaction surfaces after anodic polarization. (a) BM in WQ1; (b) BM in WQ2; (c) WM in WQ1; (d) WM in WQ2.

Furthermore, compared to the BM specimens, the WM specimens (shown as solid marks in Fig. 5) in the two test waters showed lower current densities in the passivation regions. It is suggested that the anodic reaction rate on the surface of the WM specimen decreased as the passivation film quickly formed on the WM specimen in each test water. This behavior is presumed to be due to the difference in chemical composition between the BM and the DM in the WM.

3.2 Corrosion potential and the film formed

Figure 7 shows the time variations of OCP for the BM and WM specimens in the two test waters. The BM specimen in WQ1 showed a lower OCP in the first few hours after the start of test, but this became stable and higher than for the WM specimen until 100 h of measurement. It is considered that the passivation film formed on the surface of the BM specimen has better corrosion resistance for a longer time than that of the WM specimen in WQ1.

Fig. 7

Time variations of the OCPs of the specimens.

Both the BM and WM specimens in WQ2 exhibited higher OCPs after the start of the corrosion potential tests. It is considered that the competitive adsorption between COOH⁻ and Cl⁻ suppressed the attack on the passivation film from Cl⁻. The physisorption of NH₄COOH was also considered to inhibit corrosion in the WQ2 test water.¹³⁾ Note that the missing region of the OCP curve for BM in WQ2 around 55 h was due to a temporary error in the recorder. However, similar to the results of the anodic polarization tests, the corrosion inhibiting effect of the formic acid additive in WQ2 was evident within a short time for STBA24 and its WM. The BM and WM specimens in WQ2 showed sharp potential drops after 57 h and 50 h of measurement, respectively, as shown in Fig. 7. It is considered that the sharp potential drops correspond to the progress of corrosion due to destruction of the passivation film and also pitting corrosion. The time to the sharp potential drop for the BM specimen was slightly longer than that for the WM specimen, which suggested that the stability of the film formed on the WM specimen was somewhat lower. Similar sharp potential drop behavior of boiler equipment materials in simulated boiler lay-up waters has been reported by Hirano et al.²⁾

Figure 8 shows the appearances of the specimens after OCP measurement for 100 h. The BM specimens exhibited pitting corrosion in both test waters, and a thicker film (scale) was formed on the specimen tested in WQ2. For the WM specimens in both test waters, no clear pits were observed on the DM zones, whereas a large pit was observed at the boundaries of the HAZ and BM on the specimen tested in WQ2. It is presumed that the growth of the large pit is one of the main reasons for the sharp potential drop after 50 h of OCP measurement. The susceptibility to pitting corrosion at the boundary of HAZ and BM is considered to be due to the unevenness in the microstructure and composition at these boundaries.^15–17) Furthermore, the presence of coarse and fine grains in the microstructure of the HAZ increased the susceptibility to pitting corrosion, while the lower susceptibility of the DM to pitting corrosion was due to the slightly higher chromium content in the region.

Fig. 8

Appearance of the specimens after the 100 h OCP measurements. (a) BM in WQ1; (b) BM in WQ2; (c) WM in WQ1; (d) WM in WQ2.

Figure 9 presents SEM images of the BM specimen surfaces after 100 h of OCP measurements in both test waters. Thick and cracked film (scale) was clearly observed on the surface of the specimen tested in WQ2. Figure 10 shows the surfaces of each zone (BM, HAZ, and DM) for the WM specimens tested in the two test waters. In each test water the BM zone of WM specimen exhibited the same surface pattern with the BM specimen shown in Fig. 9. No obvious differences were observed among the BM, HAZ, and DM zones in each test water. Similar to the BM specimen tested in WQ2 (Fig. 9(b)), thick and cracked films were also observed on the WM specimen tested in WQ2.

Fig. 9

SEM images of the BM specimen surfaces after OCP measurements for 100 h. (a) In WQ1; (b) In WQ2.

Fig. 10

SEM images of the WM specimen surfaces after OCP measurements for 100 h. (a) BM in WQ1; (b) HAZ in WQ1; (c) DM in WQ1; (d) BM in WQ2; (e) HAZ in WQ2; (f) DM in WQ2.

The films formed on the surfaces of the specimens after OCP measurement for 100 h were analyzed using EPMA. Figure 11 shows the EPMA results of the WM specimens tested in both test waters. Similar to the results of SEM observation, no obvious differences were confirmed among the BM, HAZ, and DM zones for either test water. The EPMA results suggest that the films formed on the BM, HAZ, and DM zones of the WM specimen in WQ1 are composed mainly of Fe-oxides such as Fe(OH)₂ and Fe₃O₄, which explains the relatively stable potential behavior of both the BM and WM specimens during the 100 h OCP measurements. In contrast, a large amount of oxygen (O) atoms was detected on the BM, HAZ, and DM zones of the WM specimen tested in WQ2, which corresponds to the formation of the thick film (scale) observed. Compared to the components of the films formed in WQ1, the films formed in WQ2 contained a large amount of carbon (C) and chromium (Cr) atoms.

Fig. 11

EPMA results for the WM specimens after OCP measurements for 100 h. (a) BM in WQ1; (b) HAZ in WQ1; (c) DM in WQ1; (d) BM in WQ2; (e) HAZ in WQ2; (f) DM in WQ2.

Cathodic polarization of the BM specimens revealed almost no influence of the 100 ppm HCOOH on the cathodic reactions that occurred on the steel surface in WQ2. Therefore, the corrosion potentials of the specimens corresponded mainly to the anodic reactions in the two test waters. The anodic reactions during the OCP measurements in WQ2 were suppressed by the structural adsorption of NH₄COOH and the quick formation of a thicker film that exhibited the same corrosion inhibiting action as that with anodic polarization; therefore, the specimens exhibited higher OCPs after the start of tests. However, it is presumed that the hydrolysis of HCOO⁻ in WQ2 (eq. (9)) progressed with relatively long time OCP measurements.¹⁸⁾   

\begin{equation} \text{8HCOO$^{-}$} + \text{2H$_{2}$O} \to \text{8CO$_{2}$} + \text{O$_{2}$} + \text{12H$^{+}$} + \text{20e$^{-}$}. \end{equation} (9)

The oxygen (O₂) and carbon dioxide (CO₂) generated from the eq. (9) promote the anodic reactions. It is therefore considered not only Fe-oxides and Cr-oxides,¹⁹⁾ but also iron carbonate (FeCO₃) and chromium carbonate (Cr₂(CO₃)₃) are easily formed during the OCP measurements in WQ2.¹⁸⁾ These corrosion products adhere to the specimen surfaces; therefore, the film formed became thicker. On the other hand, the quick formation of the thicker film indicates that a certain magnitude of corrosion progressed in WQ2. Furthermore, it is presumed that the corrosion protection properties decreased as the film (scale) became thicker, so that both the BM and WM specimens showed drops in OCP. In addition, it is presumed that the cracks formed were due to dehydration when the specimens were placed in air after the OCP measurements.

3.3 Corrosion rate in the test water

The effect of formic acid on the corrosion rate of STBA24 low-alloyed steel in WQ2 was investigated. Figure 12 shows the corrosion rate obtained by immersion tests of the BM in the test waters for 100 h (4.17 d). The corrosion rate in WQ2 (17.18 mdd) was higher than that in WQ1 (7.85 mdd). It is typically desirable to set the number of test specimens to three or more to measure the corrosion rate under the same test conditions to improve statistical significance;²⁰⁾ however, only one specimen was used in each test water to simply compare the tendency of the change in corrosion rate due to the presence or absence of formic acid. The authors have also conducted immersion corrosion tests on STBA12 low-alloyed steel (mass%; 0.19C–0.17Si–0.46Mn–0.007P–0.005S–0.48Mo–bal. Fe) for boiler tubes in the same test waters,²¹⁾ and confirmed that the average corrosion rate in WQ2 (118.8 mdd) was higher than that in WQ1 (54.7 mdd), which is consistent with the tendency of the STBA24 steel. It should also be noted that STBA12 low-alloyed steel contains no chromium; therefore, it showed a higher corrosion rate than STBA24 in each test water. In WQ2 with the hydrolysis of HCOO⁻ (eq. (9)), the film (scale) composed of Fe-oxides, Cr-oxides, FeCO₃, and Cr₂(CO₃)₃ was easily formed on the steel surface, so that corrosion proceeded readily. The above anodic polarization measurements showed that formic acid suppressed the initiation of pitting corrosion on the specimens in the simulated boiler water with added Cl⁻, and similar results have shown that a small amount of formic acid inhibited the pitting corrosion of the boiler tube steels.²²⁾ Conversely, the 100 h immersion corrosion tests confirmed that formic acid promoted corrosion. Therefore, it is considered that the presence of formic acid is a factor that promotes the corrosion of boiler equipment materials composed of low-alloyed or carbon steels over long-term service in the boiler water containing Cl⁻.

Fig. 12

Corrosion rate of BM specimen in both test waters.

4. Conclusions

1. (1)    From the results of potentiodynamic polarizations, it was confirmed that formic acid in simulated boiler water with 100 ppm Cl⁻ added can suppress the initiation of pitting corrosion within a shorter time on both the BM and WM of STBA24. Conversely, after 100 h of immersion corrosion tests, it was confirmed that formic acid promoted corrosion.
2. (2)    During the 100 h OCP measurements, both the BM and WM of STBA24 exhibited sharp potential drops after tens of hours in the WQ2 test water. It is presumed that as Fe-oxides, Cr-oxides, FeCO₃, and Cr₂(CO₃)₃ were easily formed on the materials in WQ2, the film formed (scale) became thicker and the corrosion protection properties decreased. Furthermore, the WM showed somewhat lower corrosion resistance during OCP measurements for 100 h in both of the test waters. This behavior is considered to be mainly due to unevenness in the microstructure and composition.
3. (3)    Based on these results and the corrosion rate measurement results, it is suggested that formic acid in chloride-containing boiler water is a factor that promotes the corrosion of STBA24 and its WM over long-term service.

Acknowledgment

The research was supported by a Kakenhi Grant-in-Aid for Scientific Research (C) (No. 20K04158) from the Japan Society for the Promotion of Science (JSPS).

REFERENCES

• 1)  Thermal and Nuclear Power Engineering Society: Hatsuden-Pranto no Fushoku to Sono Boshi (Corrosion of Power Plant and Its Prevention) (in Japanese), (Thermal and Nuclear Power Engineering Society, Tokyo, 1997) p. 49.
• 2)   M.  Hirano,  S.  Itaba,  T.  Minami,  L.-B.  Niu and  H.  Takaku: Zairyo-to-Kankyo 54 (2005) 55–60.
• 3)  H.H. Uhlig: Corrosion Handbook, (John Wiley and Sons, New York, 1961) p. 525.
• 4)   N.A.  Parker and  A.G.  Howell: PowerPlant Chem. 15 (2013) 212–215.
• 5)   C.  Lertsurasakda,  P.  Srisukvatananan,  L.  Liu,  D.  Lister and  J.  Mathews: PowerPlant Chem. 15 (2013) 181–189.
• 6)  D.H. Moed, A.R.D. Verliefde, S.G.J. Heijman and L.C. Rietveld: Proc. of ICPWS16, (London, 2013) PWS-034.
• 7)  Japanese Standards Association: Water conditioning for boiler feed water and boiler water; JIS B8223, (Japanese Standards Association Tokyo, 2015).
• 8)  Japanese Standards Association: Testing method for corrosivity of industrial water; JIS K0100, (Japanese Standards Association Tokyo, 1990).
• 9)   T.  Tsushima,  M.  Suzuki,  N.  Akao,  N.  Hara and  K.  Sugimoto: Zairyo-to-Kankyo 52 (2003) 92–101.
• 10)  M. Pourbaix: Atlas of Electrochemical Equilibrium in Aqueous Solutions, (Pergamon Press, New York, 1966).
• 11)   T.  Tsushima,  N.  Hara and  K.  Sugimoto: Zairyo-to-Kankyo 52 (2003) 545–553.
• 12)   T.  Shibata,  M.  Seo,  K.  Sugimoto,  T.  Tsuru,  S.  Fujimoto and  H.  Inoue: Zairyo-to-Kankyo 54 (2005) 2–8.
• 13)   S.N.  Raicheva,  B.V.  Aleksiev and  E.I.  Sokolova: Corros. Sci. 34 (1993) 343–350.
• 14)   S.A.  Ali and  M.T.  Saeed: Polymer 42 (2001) 2785–2794.
• 15)   K.  Nishimaki and  H.  Uchiyama: J. Surf. Finish. Soc. Jpn. 9 (1958) 132–137.
• 16)   H.  Miyuki: J. Jpn. Welding Soc. 77 (2008) 641–645.
• 17)   T.  Nakane,  L.-B.  Niu,  S.  Oishi and  H.  Takaku: Netsu Shori (J. of the Jpn. Soc. for Heat Treatment) 50 (2010) 614–619.
• 18)   G.E.  Badea,  D.  Ionita and  P.  Cret: Mater. Corros. 65 (2014) 1103–1110.
• 19)   L.B.  Niu,  T.  Nakane and  H.  Takaku: J. Japan Inst. Metals 76 (2012) 203–209.
• 20)  Japan Society of Corrosion Engineering: Corrosion Handbook (in Japanese), (Maruzen, Tokyo, 2000) p. 569.
• 21)   L.-B.  Niu and  Y.  Umano: unpublished.
• 22)   T.  Nakane,  L.-B.  Niu,  S.  Oishi and  H.  Takaku: J. Japan Inst. Metals 74 (2010) 565–571.