Microstructure and corrosion behaviour of laser surface melting treated WE43 magnesium alloy

Abstract

An attempt has been made to improve the corrosion behavior of WE43 magnesium alloy by laser surface melting (LSM) using a 10 kW continuous-wave CO₂ laser. The microstructure evolution of WE43 alloy after LSM treatment was analyzed by using scanning electron microscopy, energy-dispersive spectroscopy and metallographic microscope. The corrosion resistance of specimens was assessed by electrochemical and immersion tests. Results showed that the LSM treated WE43 alloy presented a uniform microstructure with refined grains, enriched alloying elements and redistributed intermetallic compounds. The LSM treatment effectively improved corrosion resistance of WE43 alloy, which was mainly associated with enrichment of alloying elements in α-Mg matrix and uniform distributions of the refined Mg₁₄Nd₂Y phase.

---

1. Introduction

Recently, magnesium (Mg) alloys display great potential for applications in automotive, aerospace and electronic industries owe to their low density and adequate specific strength.^1–4 However, Mg alloys, generally, reveal a poor corrosion resistance, which is mainly associated with their high chemical activity and the lack of a protective passive oxide film.^5,6 This disadvantage has been a major obstacle restricting their further application in aforementioned industries.

Corrosion of Mg alloys is essentially a surface degradation process, and it can be reduced by proper modification of the surface microstructure and/or composition.^7,8 Various surface treatments including conversion coating, electrochemical deposition, plasma electrolytic oxidation, ion implantation, gas phase deposition and laser surface treatment, are applied to improve corrosion resistance of Mg alloys.^8–12 Of these methods, laser surface melting (LSM) is one of the most effective techniques to modify surface properties, because it can homogenize and refine the microstructure, and dissolve the second phases.^13,14 A number of attempts have been made to investigate the effects of LSM treatment on the corrosion behaviour of Mg alloys.^15–24 It has been proved that the LSM treatment has positive influences on the corrosion resistance of AZ- and AM-type Mg alloys, such as AZ31, AZ61, AZ91 and AM60.^15–21 The enhanced corrosion resistance was mainly attributed to the Al enrichment, pronounced grain refinement and uniform distributions of Mg₁₇Al₁₂ phase. For the Mg alloys free of Al element, relatively rare researches^7,22–24 are carried out to explore the corrosion behaviour by LSM treatments. Majumdar⁷ and Khalfaoui²² studied the effect of LSM treatment on the corrosion resistance of WEZ and ZE41 Mg alloys, respectively. Results revealed that the LSM treatment improved their corrosion resistance, resulting from the refinement of grains and redistribution of the intermetallic particles. Guo et al.²³ observed similar improvement in case of WE43 alloy due to the dissolution of Mg₁₂Nd into α-Mg matrix. Unfortunately, they did not analyse corrosion mechanism in depth. However, Banerjee et al.²⁴ indicated that the corrosion resistance of ZE41 Mg alloy was not significantly improved by LSM treatment, although microstructural refinement was achieved. This is due to absence of beneficial alloying elements such as Al.

WE43 Mg alloy can offer the required creep resistance for long-term usage at elevated temperature (up to 250 °C), which exhibits important applications in the aerospace industries.²⁵ However, so far, little efforts have been made to examine the microstructure and corrosion behaviour of the LSM treated WE43 Mg alloy in detail. In this work, LSM treatment has been conducted on WE43 Mg alloy containing Y and Nd as the major alloying elements. Microstructural changes induced by LSM treatment are analysed and its effect on corrosion behaviour is also investigated.

2. Experimental

2.1 Laser surface melting process

WE43 Mg alloy, with the chemical composition shown in Table 1, was used as the substrate in this investigation. The LSM process was performed using a 10 kW transverse-flow continuous-wave CO₂ laser. Prior to LSM treatment, the specimens (50 mm × 50 mm × 7 mm) were ground with 1500 grit SiC paper, then sand blasted, washed with alcohol and dried in warm air. Argon served as a shielding gas to protect the specimens during laser process. The diameter of laser beam spot was 3 mm and the overlapped track was 50%. The laser power and scan speed were set as 3000 W and 1000 mm min⁻¹, respectively. After that, the specimens were cut into rectangular plates with dimensions of 30 mm × 20 mm × 7 mm, ground with 1500 grit SiC paper and polished with Al₂O₃ paste to get a smooth surface.

Table 1 EDS elemental analysis of as-received, LSM treated alloy and appointed phases in Fig. 1

Specimens	Content of elements (wt%)
	Mg	Y	Nd	Gd	Zr
WE43	91.37	4.84	2.82	0.42	0.55
α-Mg matrix	95.56	3.44	0.86	0.33	0.10
1	77.21	5.38	15.89	1.22	0.31
2	25.85	67.90	2.64	1.27	2.34
3	38.02	0.00	0.39	0.05	61.55
α-Mg matrix (LSM)	90.98	5.07	2.77	0.52	0.62

---

2.2 Microstructure characterization

The as-received and LSM treated WE43 alloy were etched using a solution of 10 ml acetic acid, 4.2 g picric acid, 10 ml H₂O and 70 ml ethanol. The metallurgical microstructures were observed by Zeiss Axio Imager A2m microscope. The microstructural analysis and corresponding element distribution were conducted using a scanning electron microscope (SEM, JSM-5600LV, JEOL) equipped with energy-dispersive spectrometer (EDS, KEVER).

2.3 Electrochemical and immersion tests

Electrochemical impedance spectroscopy (EIS) and potentiodynamic polarization tests were conducted to evaluate the corrosion resistance of the as-received and LSM treated WE43 alloy. Electrochemical measurements were performed using an Autolab PGSTAT302N electrochemical workstation at room temperature. A typical three-electrode cell system, with a platinum plate as the auxiliary electrode, an Ag/AgCl (saturated with KCl) electrode as the reference electrode and the specimen as the working electrode was used. The specimens with an area of 0.5 cm² were exposed to the 0.62 M NaCl solution (3.5 wt%). After 0.5 h of immersion, EIS tests were operated at open circuit potential over the frequency from 100 000 to 0.01 Hz with the signal amplitude of 10 mV. Potentiodynamic polarization tests were performed at a scanning rate of 1 mV s⁻¹ after immersion for 1 h. For each specimen, electrochemical tests were performed in duplicate to ascertain the reproducibility of results.

Prior to immersion test, the specimens were sealed by polyurethane with the working surface exposed. For immersion test, the specimens were suspended vertically in NaCl solutions with chloride ion concentrations of 0.01 M, 0.1 M and 0.62 M, respectively. The evolved hydrogen was collected by the combination of the funnel and the burette above the specimen. The burette was initially filled with corrosion solution, which was gradually displaced by the evolved hydrogen during immersion test. The experiment device used to collect hydrogen was similar to that described in the literature.^26,27 After certain immersion periods, the surface appearances of samples were taken by a digital camera. After that, the specimens were rinsed with distilled water to remove salt deposits, and then dried in warm air. A scanning electron microscope was used to observe the surface and cross-sectional corrosion morphologies of the specimens. To calculate the weight loss, the corrosion products were removed with a chromic acid solution (200 g L⁻¹ CrO₃ + 10 g L⁻¹ AgNO₃),²⁶ rinsed with ethanol, dried in air, and finally weighed.

3. Results

3.1 Microstructure

Fig. 1 presents optical micrograph of the as-received WE43 alloy after etching. The WE43 alloy mainly consisted of α-Mg grains with size in the range of 20–55 μm and longitudinal precipitate along the grain boundaries. In addition, small amounts of cuboidal and spherical precipitates distributed randomly in α-Mg matrix. The longitudinal, cuboidal and spherical precipitates were labeled as 1, 2 and 3, respectively, and their chemical compositions were determined by SEM-EDS analysis in Table 1. The longitudinal precipitate with an approximate composition of 77.21 Mg-5.38 Y-15.89 Nd (wt%) could be identified as Mg₁₄Nd₂Y phase^28,29 and the over-high Mg content in the composition of Mg₁₄Nd₂Y phase was influenced by α-Mg matrix. The cuboidal precipitate consisting of 67.90 wt% Y corresponded to Y-rich phase.²⁸ The spherical precipitate was the Zr-rich phase (Zr content is as higher as 61.55 wt%). The element distribution in WE43 alloy was analysed using X-ray mapping as shown in Fig. 2. It can be seen that the alloying elements (Y, Nd, Gd and Zr) mainly concentrated on intermetallic compounds.

Fig. 1 Optical micrograph of the as-received WE43 alloy.

Fig. 2 SEM micrograph and corresponding element distribution of the as-received WE43 alloy.

Fig. 3 exhibits the cross-section and surface micrographs of LSM treated WE43 alloy. A continuous and homogeneous modified layer formed on the top surface of WE43 alloy (Fig. 3a). The modified depth was approximately 1.15 mm. The surface microstructure revealed highly refined grain structure with uniform dispersion of fine Mg₁₄Nd₂Y phase (Fig. 3b). The average grain size decreased to 6.4 μm. Besides, relatively small amounts of Y-rich and Zr-rich phases still remained within the melted layer. Fig. 4 presents the element distribution of the LSM treated WE43 alloy. It was obvious that Mg and the alloying elements (Y, Nd, Gd and Zr) in the modified layer distributed more uniformly than that in the as-received one, which proved the redistribution of intermetallic compounds. The chemical composition of α-Mg matrix after LSM treatment is shown in Table 1. Compared to the as-received alloy, the content of all alloying elements in the α-Mg matrix was increased, especially the Y and Nd elements, manifesting the dissolution of the intermetallic compounds. In summary, LSM treatment resulted in refined grains, enriched alloying elements and redistributed intermetallic compounds in the modified layer of WE43 alloy.

Fig. 3 The cross-sectional (a) and surface (b) micrographs of LSM treated WE43 alloy.

Fig. 4 SEM micrograph and corresponding element distribution of LSM treated WE43 alloy.

3.2 Electrochemical results

The potentiodynamic polarization curves of the as-received and LSM treated WE43 alloy in 0.62 M NaCl solution are shown in Fig. 5. The corrosion potential (E_corr) and the corrosion current density (i_corr) derived from the polarization curves are listed in Table 2. The E_corr of the LSM treated WE43 alloy was approximately 141 mV more positive than that of the as-received one. The i_corr value of WE43 alloy was reduced from 2.93 × 10⁻⁵ to 8.51 × 10⁻⁶ after LSM treatment. These results indicated that the corrosion resistance of WE43 alloy was effectively improved by LSM treatment. Meanwhile, the LSM treated alloy revealed a wider passive region at anodic branch of the polarization curve as compared to the as-received one. The change in corrosion behaviour was ascribed to the enrichment of alloying elements, which provides passive characteristics to the melted surface.¹⁷

Fig. 5 Potentiodynamic polarization curves of the as-received and LSM treated WE43 alloy in 0.62 M NaCl solution.

Table 2 The results of potentiodynamic corrosion tests in 0.62 M NaCl solution

Specimen	Ecorr (mV vs. Ag/AgCl)	icorr (A cm^−2)
As-received	−1412	2.93 × 10^−5
LSM	−1271	8.51 × 10^−6

---

EIS test was performed to obtain further information on corrosion resistance of the as-received and LSM treated WE43 alloy. As shown in Fig. 6a and b, the EIS spectrums of both specimens are characterized by three loops, i.e. a capacitive loop in the high frequency, another capacitive loop in the medium frequency and an inductive loop in the low frequency. Based on the characteristics of EIS spectrum, an equivalent circuit is proposed in Fig. 6a. In the equivalent circuit,^30,31 R_s represents the solution resistance, R₁ is resistance of naturally formed passive film on the substrate exposed to the corrosive electrolyte in parallel with the film layer capacitance C₁, and R₂ represents charge transfer resistance of the faradaic process on the metal surface in parallel with double layer capacity C₂. To account for the inductive behaviour, an inductor L in series with a resistance R₃ was introduced in this model. The Nyquist and Bode plots are fitted with this equivalent circuit and the fit results are shown in Fig. 6a and b. The values of the fitted circuit elements are summarized in Table 3. It can be found that the LSM treated WE43 alloy exhibited much higher resistance values (R₁, R₂ and R₃) than the as-received one, implying an increase in corrosion resistance after the LSM treatment. These results were in good agreement with the investigations of polarization curves.

Fig. 6 Experimental and fitting results of (a) Nyquist and (b) Bode plots of the as-received and LSM treated WE43 alloy in 0.62 M NaCl solution. The illustration is equivalent circuit for fitting the EIS data.

Table 3 EIS fitting results for specimens in 0.62 M NaCl solution

Specimen	C1 μF cm^−2	R1 Ω cm^2	C2 μF cm^−2	R2 Ω cm^2	L Ω s cm^2	R3 Ω cm^2
As-received	3	268	2.62	428	5102	220
LSM	3	1268	1337	433	52 411	2295

---

3.3 Immersion tests

Fig. 7a shows hydrogen evolution data of the as-received and LSM treated WE43 alloy immersed in 0.01, 0.1 and 0.62 M NaCl solutions. For all specimens, hydrogen evolution volume increased with immersion time elapsing, although the rate of hydrogen evolution was different. Corrosion rates of specimens (P_H, mm per year) were calculated according to the hydrogen evolution rates (V_H, ml cm⁻² per day)³² and presented in Fig. 7b. It can be found that the corrosion rates of WE43 alloy, as-received or LSM treated, increased with immersion time and chloride ion concentration. Moreover, the LSM treated specimen exhibited much lower corrosion rate than the received one in the same concentration of NaCl solution. These suggested that LSM treatment enhanced the corrosion resistance of WE43 alloy, regardless of the chloride ion concentration.

Fig. 7 Hydrogen evolution data (a) and corresponding corrosion rates (b) for the as-received and LSM treated WE43 alloys immersed in 0.62 M NaCl solution.

Table 4 lists the corrosion rates after immersion in 0.62 M NaCl solution for different time. The P_i,EIS was instantaneous corrosion rate evaluated from EIS data (Fig. 5) using the low frequency limit as a good measure of the polarisation resistance and using values of the Tafel slopes.³³ The P_i was instantaneous corrosion rate calculated from Tafel extrapolation using i_corr (Table 2).³² The P_W was average corrosion rate measured from weight loss.³² As shown in Table 4, the value of P_W was very close to that of P_H in the same immersion time. However, the values of P_i,EIS and P_i were much lower than that of P_H in the same immersion time. These indicated the electrochemical technique, either EIS data or Tafel extrapolation, could not provided good measurements of the corrosion rate of WE43 alloy.

Table 4 The corrosion rates of the as-received and LSM treated specimens after immersion in 0.62 M NaCl solution for certain time at room temperature

Specimens	0.5 h		1 h		24 h
	Pi,EIS (mm per year)	PH (mm per year)	Pi (mm per year)	PH (mm per year)	PW (mm per year)	PH (mm per year)
As-received	0.50	9.49	0.67	10.54	100.68	128.18
LSM	0.089	6.52	0.19	8.37	40.16	45.63

---

Fig. 8 gives the photographs of the as-received and LSM treated specimens after immersion in 0.01, 0.1 and 0.62 M NaCl solutions for certain time. For all specimens, the surface was characterized by local corrosion, although the extent of corrosion was different. The extent of corrosion damage of specimens, revealing a similar change trend as that of corrosion rates, increased with immersion time and chloride ion concentration. In addition, compared to the as-received one, the LSM treated specimens showed relatively light corrosion damage within the same immersion time and chloride ion concentration. It should be noted that the as-received specimens in 0.1 and 0.62 M NaCl solutions suffered from severe corrosion damage, a mass of corrosion products fell off from the alloy surface and thus the alloy surface became bumpy.

Fig. 8 Photographs of the as-received (a₁–c₁) and LSM treated (a₂–c₂) WE43 alloy after immersion in NaCl solutions of different chloride for different time.

To more clearly observe the corrosion features of the as-received WE43 alloy, the surface and cross-sectional micrographs of the specimens after immersion in 0.1 M NaCl solution for certain time are analysed. As shown in Fig. 9, the as-received alloy showed typical pitting corrosion. After 24 h immersion, the sample surface featured discrete corrosion damage (Fig. 9a). With the extending of immersion time, several adjacent corrosion pits merged together to form a large one. At the same time, the corrosion penetrated gradually into the interior of substrate. Up to 120 h, lots of corrosion products accumulated at the sample surface with the depth more than 470 μm (Fig. 9e). However, there were many cracks in the corrosion layer, indicating the limited corrosion protection to alloy. Interestingly, corrosion developed heterogeneously in the interior of the alloy. As shown in Fig. 9a₂–e₂, corrosion attack mainly focused on α-Mg matrix (grey region) while intermetallic grains (bright region) remained unchanged. Moreover, the Mg₁₄Nd₂Y phase identified by blue arrows could serve as a barrier to prevent the corrosion progress (Fig. 9a₂ and b₂).

Fig. 9 Surface (a₁–e₁) and cross-section (a₂–e₂) corrosion morphologies of the as-received WE43 alloy after immersion in 0.1 M NaCl solution for different time.

Fig. 10 presents the surface and cross-sectional micrographs of the LSM treated WE43 alloy during the immersion tests. In contrast to the as-received alloy, some raised corrosion products existed on the alloy surface after immersion for 24 h (Fig. 10a₁). A larger corrosion pits could be observed until the 96 h of immersion. Meanwhile, the cross-sectional micrographs (Fig. 10a₂–e₂) indicated that the corrosion depth was much smaller than that of the as-received WE43 alloy in the same immersion time. The results of immersion tests further proved the positive effect of LSM treatment on the corrosion resistance of WE43 alloy.

Fig. 10 Surface (a₁–e₁) and cross-section (a₂–e₂) corrosion morphologies of the LSM treated WE43 alloy after immersion in 0.1 M NaCl solution for different time.

4. Discussion

Based on the discussion above, the different corrosion characteristics between the as-received and LSM treated WE43 alloy can be schematically illustrated in Fig. 11. As shown in Fig. 1, the as-received WE43 alloy mainly consisted of α-Mg and Mg₁₄Nd₂Y phases, as well as small amounts of Y-rich and Zr-rich phases. The Y-rich and Zr-rich phases exhibited positive volta potential differences with respect to the α-Mg matrix, of about +50 and +170 mV, respectively.²⁹ When the WE43 alloy was exposed to corrosion solution, these phases acted as effective cathodes to initiate micro-galvanic electrochemical corrosion of the adjacent α-Mg matrix.²⁹ In addition, the large size Mg₁₄Nd₂Y phase preferentially acted as a galvanic cathode to accelerate the dissolution of the α-Mg matrix, though there was only +15 mV volta potential difference between Mg₁₄Nd₂Y phase and α-Mg matrix. The corrosion was triggered by the dissolution of the α-Mg matrix adjacent to intermetallic compounds (Fig. 11a). As immersion time prolonged, corrosion aggravated and the area of localized damage broadened in vertical and horizontal directions. Because the Mg₁₄Nd₂Y phase could serve as a barrier to prevent the corrosion progress, the distribution of the corrosion products was not uniform in thickness. Meanwhile, adjacent corrosion pits would merge together. Finally, the surface of the as-received WE43 alloy was fully coved by corrosion products (Fig. 11b).

Fig. 11 Schematic diagrams for the corrosion process of the as-received (a, b) and LSM treated (c, d) WE43 alloy.

For the LSM treated WE43 alloy, some of intermetallic compounds (Mg₁₄Nd₂Y, Y-rich and Zr-rich phases) dissolved into α-Mg matrix, which significantly decreased the volume fraction of effective cathodes. On the other hand, the enrichment of alloying elements made corrosion potential of α-Mg matrix shift to the positive direction, lowering the corrosion susceptibility of α-Mg matrix. In addition, once the corrosion was initiated, the refined Mg₁₄Nd₂Y phases distributed evenly and mainly served as a barrier to impede corrosion process. Compared to the as-received WE43 alloy, the LSM treated specimen suffered slighter corrosion damage within the same immersion time (Fig. 11c and d).

5. Conclusions

(1) The WE43 alloy mainly consisted of α-Mg matrix and Mg₁₄Nd₂Y phase, as well as small amounts of Y-rich and Zr-rich phases.

(2) After LSM treatment, a modified layer with the depth of 1.15 mm formed on the top surface of WE43 alloy. In modified layer, grains were refined, alloying elements were enriched in α-Mg matrix and intermetallic compounds were dissolved and redistributed.

(3) The LSM treated WE43 alloy revealed better corrosion resistance than the as-received one, due to the combined effect of the lowering corrosion susceptibility of α-Mg matrix and the barrier effect of refined Mg₁₄Nd₂Y phase.

Acknowledgements

The financial support from National Natural Science Foundation of China (Grant No. 51241006) and the “Hundred Talents Program’’ of Chinese Academy of Sciences (J. Liang) was gratefully acknowledged.

References

1. B. L. Mordike and T. Ebert, Mater. Sci. Eng., A, 2001, A302, 37.
2. Y. Kojima, Mater. Trans., 2001, 42, 1154.
3. M. K. Kulekci, Int. J. Adv. Des. Manuf. Technol., 2008, 39, 851.
4. G. Song and A. Atrens, Adv. Eng. Mater., 1999, 1, 11.
5. A. Atrens, G. L. Song, F. Y. Cao, Z. M. Shi and P. K. Bowen, J. Magnesium Alloys, 2013, 1, 177.
6. G. L. Song and A. Atrens, Adv. Eng. Mater., 2003, 5, 837.
7. J. D. Majumdar, R. Galun, B. L. Mordike and I. Manna, Mater. Sci. Eng., A, 2003, 361, 119.
8. A. Atrens, G. Song, M. Liu, Z. Shi, F. Cao and M. Dargusch, Adv. Eng. Mater., 2015, 17, 400.
9. J. E. Gray and B. Luan, J. Alloys Compd., 2002, 336, 88.
10. Y. C. Su, Y. B. Lu, Y. C. Su, J. J. Hu, J. S. Lian and G. Y. Li, RSC Adv., 2015, 5, 56001.
11. B. Li and Y. Han, RSC Adv., 2015, 5, 46109.
12. H. Zhao, S. Cai, Z. T. Ding, M. Zhang, Y. Li and G. H. Xu, RSC Adv., 2015, 5, 24586.
13. A. Singh and S. P. Harimkar, JOM, 2012, 64, 716.
14. Y. C. Guan, W. Zhou, H. Y. Zheng and Z. L. Li, Appl. Phys. A, 2010, 101, 339.
15. C. C. Liu, J. Liang, J. S. Zhou, L. Q. Wang and Q. B. Li, Appl. Surf. Sci., 2015, 343, 133.
16. Z. Q. Cui, H. X. Shi, W. Wang and B. S. Xu, Trans. Nonferrous Met. Soc. China, 2015, 25, 1446.
17. C. Taltavull, B. Torres, A. J. Lopez, P. Rodrigo, E. Otero, A. Atrens and J. Rams, Mater. Des., 2014, 57, 40.
18. S. R. Paital, A. Bhattacharya, M. Moncayo, Y. H. Ho, K. Mahdak, S. Nag, R. Banerjee and N. B. Dahotre, Surf. Coat. Technol., 2012, 206, 2308.
19. A. E. Coy, F. Viejo, F. J. Garcia-Garcia, Z. Liu, P. Skeldon and G. E. Thompson, Corros. Sci., 2010, 52, 387.
20. R. K. Singh Raman, S. Murray and M. Brandt, Surf. Eng., 2007, 23, 107.
21. G. Abbas, Z. Liu and P. Skeldon, Appl. Surf. Sci., 2005, 247, 347.
22. W. Khalfaoui, E. Valerio, J. E. Masse and M. Autric, Optic Laser Eng., 2010, 48, 926.
23. L. F. Guo, T. M. Yue and H. C. Man, J. Mater. Sci., 2005, 40, 3531.
24. P. Banerjee, R. K. Singh Raman, Y. Durandet and G. McAdam, Corros. Sci., 2011, 53, 1505.
25. P. Mengucci, G. Barucca, G. Riontino, D. Lussana, M. Massazza, R. Ferragut and E. Hassan Aly, Mater. Sci. Eng., A, 2008, 479, 37.
26. Z. M. Shi and A. Atrens, Corros. Sci., 2011, 53, 226.
27. R. L. Xin, B. Li, L. Li and Q. Liu, Mater. Des., 2011, 32, 4548.
28. T. Rzychoń and A. Kiełbus, Journal of Achievements in Materials and Manufacturing, 2007, 21, 31.
29. A. E. Coy, F. Viejo, P. Skeldon and G. E. Thompson, Corros. Sci., 2010, 52, 3896.
30. G. Baril and N. Pébère, Corros. Sci., 2001, 43, 471.
31. M. Curioni, F. Scenini, T. Monetta and F. Bellucci, Electrochim. Acta, 2015, 166, 372.
32. Z. Shi, M. Liu and A. Atrens, Corros. Sci., 2010, 52, 579.
33. Z. Shi, F. Cao, G. Song and A. Atrens, Corros. Sci., 2014, 88, 434.

---