Corrosion Resistance of Plasma-Anodized AZ91 Mg Alloy in the Electrolyte with/without Potassium Fluoride

Materials Transactions, Vol. 5, No. 3 (29) pp. 671 to 678 #29 The Japan Institute of Metals EXPRESS REGULAR ARTICLE Corrosion Resistance of Plasma-Anodized AZ91 Mg Alloy in the Electrolyte with/without Potassium Fluoride Duck Y. Hwang*, Yong M. Kim and Dong H. Shin Department of Metallurgy and Materials Science, Hanyang University, Ansan 426-791, Korea Plasma Electrolyte Oxidation (PEO) behavior of was investigated in the electrolytes with/without potassium fluoride. Growth rate of coating thickness in the electrolyte containing potassium fluoride () was much higher than that in the electrolyte without potassium fluoride (). The oxide layer formed on in electrolyte with potassium fluoride and sodium silicate consisted of MgO, MgF 2 and Mg 2 SiO 4. Corrosion current density of oxide layer coated from the electrolyte with potassium fluoride was much lower than that of oxide layer coated from the electrolyte without potassium fluoride. From the result of EIS analysis, it was known that inner barrier layer in the oxide layer coated from the electrolyte with potassium fluoride had a good influence of the corrosion resistance of Mg alloy. The corrosion resistance curves of were similar to the thickness curves, indicating that the thickness of the oxide layer played an important role in corrosion resistance of. The oxide layer in the containing potassium fluoride was found to be a compact barrier-type passive film in presence of fluoride ions. The existence of the dense MgO and MgF 2 in the barrier layer had a favorable effect on the corrosion resistance of the formed from by PEO process. [doi:1.232/matertrans.mer28345] (Received September 24, 28; Accepted December 1, 28; Published January 28, 29) Keywords: magnesium alloy, plasma electrolytic oxidation, corrosion, oxide layer 1. Introduction Magnesium alloys have a superior strength-to-weight ratio, high dimensional stability, lower density and good electromagnetic shielding than several other alloys but relatively poor corrosion resistance, especially in acidic environments and the saltwater conditions, because Mg is electrochemically the most active metal. 1 4) Therefore, it is desirable to alter the surface properties of Mg and its alloys in order to improve its corrosion resistance. Various surface treatments such as electrochemical plating, conversion coating and anodizing have been used to increase the corrosion resistance of magnesium alloys. 1,5,6) However, conventional surface treatment method has the disadvantages of low corrosion resistance, complicated process of manufacture, low productivity and waste problems. Therefore it is worth developing new process of surface treatment to solve the problems. Plasma electrolytic oxidation (PEO) is one of the electrochemical surface treatment methods, which form the oxide layer on magnesium alloys in plasma state generated by applying extremely high voltage in a suitable electrolyte. 7) The structures of oxide layer on the Mg and Mg alloys fabricated by PEO process depend on various processing conditions, including chemical composition and concentration of electrolyte, electric parameters, alloy composition of substrate, pretreatment and post treatment. Especially, the chemical composition of the electrolyte exerts a considerable influence on the property and formation of effective oxide layer for Mg alloy. Multi-component electrolytes such as phosphate, silicate, and aluminate based on potassium hydroxide are used in PEO process in order to improve the corrosion resistance of magnesium alloys. In general, microstructures of oxide layer in PEO process consist of an outer porous layer and an inner barrier layer. 8,9) It was reported that the composition and *Corresponding author, E-mail: mir4181@hanyang.ac.kr quality of the barrier layer had a considerable influence on the corrosion resistance of coated Mg alloy. 1) Therefore, it is very important to control composition and quality of barrier layer during the PEO process. Therefore, it is imperative to select proper electrolyte compositions to increase the corrosion resistance of magnesium alloys. 3,11) Some researchers reported that the presence of F ion played an important role in the formation of oxide layer. 9,12) However, there is a lack of systematic study on the effect of F ion on corrosion resistance in Mg alloy coated by PEO process. In this study, the influence of potassium fluoride in the electrolyte on the structure of oxide layer of AZ91 Mg alloy was reported and the corrosion resistance was also evaluated by electrochemical analysis. 2. Experimental Procedures Commercial AZ91 ingot was used in this study. Prior to the experiments, plates with 3 5 2 mm 3 were mechanically polished to 1 grit emery paper finish, rinsed with de-ionized water, ultrasonically cleaned in ethanol, and finally dried in warm air. PEO process was conducted with 2-kW equipment which had a glass-vessel container with a sample holder as the electrolyte cell, a stainless steel used as the cathode, stirring and cooling system. Two kinds of electrolyte were used in this study. Table 1 lists the chemical compositions and constituents of electrolyte. Temperature of electrolyte was maintained at 2 to 3 C during PEO process. Applied current density is controlled at 1 A/dm 2. Table 1 Electrolyte compositions for PEO process in. Electrolyte KOH (M/L) KF (M/L) Na 2 SiO 3 (M/L).8.4.8.8.4

672 D. Y. Hwang, Y. M. Kim and D. H. Shin Surface morphology and cross-sectional images of the oxide layer were observed using a scanning electron microscope (HITACHI, S-48) with energy-dispersive spectroscopy (EDS). Phase analysis of the oxide layer was analyzed using X-ray diffraction with Cu-K radiation and excitation source at a grazing angle 2. Surface roughness of the oxide layer was determined using a laser scanning microscope. Thickness of oxide layer was measured from the crosssection morphologies. Cross-sectional TEM (TECHNAI G2 instrument) was used to characterize the oxide layer (3 nm thick) formed on. Samples for TEM observation were prepared using focused ion beam (FIB) milling. After the PEO process, the corrosion resistance of the specimen was subjected to a salt spray test for 72 ks. In compliance with ASTM standard B117, the chamber temperature was held at 35 C, and a salt solution of 5 mass% NaCl and ph ¼ 7: was used. Electrochemical impedance spectroscopy (EIS) and potentiodynamic polarization were utilized to evaluate the corrosion resistance of the coated and carried out in a 3.5 mass% NaCl solution using a Reference 6 potentiostat (Gamry Instruments, Warminster, PA, USA). After stabilization of the electrochemical testing system, the following parameters were used: the scanning rate of polarization was 1 mv/s, the EIS signal amplitude was 1 mv, and the frequency range was between.1 and 1 6 Hz. The electrochemical measurement was a conventional three-electrode cell with the sample as the working electrode, a carbon plate as the counter electrode, and a saturated calomel electrode (SCE) as the reference electrode. 3. Results and Discussions 3.1 Characteristics of oxide layer on Figure 1 shows voltage-time behaviors of in two different electrolytes. PEO process in s was typically divided into three stages regardless of composition and concentration of electrolyte. In the first stage of PEO process, the voltage increased linearly until it reached a breakdown voltage as depicted in Fig. 1. In first region, no apparent sparks were found on the metal surface, a thin transparent passive film was formed and oxygen resulting from oxidation of water and hydroxyl anions was evenly absorbed on the anode surface. In the second stage, a large number of small size sparks was observed as a white light scanning over the metal surface rapidly and randomly but distributed evenly over the whole metal surface. This point is breakdown voltage. The cell voltage increased with a rate slower than that in the initial linear stage after the breakdown voltage. In the third stage, steady sparking was established on the anode surface and the cell voltage reached a relatively stable value. Gradually, the spark size was increased, and density decreased. As shown in Fig. 1, it was observed that cell voltage of containing potassium fluoride was rapidly increased, compared to increasing rate of cell voltage of after occurring breakdown voltage. Final voltage of coated for 6 s was higher than that of. The properties of PEO process such as breakdown voltage, critical voltage, and final voltage strongly depended on composition and concentration of electrolyte. It was reported that the addition of compounds containing fluorine ion in the electrolyte helped to increase electrical conductivity of electrolyte. 7) Verdier 13) reported that the cell voltage of PEO process was important factors in the process parameter, especially influencing growth rate of oxide layer. Therefore, it was considered that potassium fluoride in the electrolyte played a considerable influence on growth and properties of the oxide layer during PEO process. The changes of coating thickness as a function of coating time in electrolytes with and without potassium fluoride is shown in Fig. 2. The growth rate of exhibited much faster than that of. When the coating time was 6 s, thickness of oxide layer coated in the and B was measured to be 12:69, 18.75 mm, respectively. Growth rate of oxide layer coated in the containing potassium fluoride was about 1.9 mm/min. Guo 11) reported that the cell voltage and time corresponding to the appearance of sparks on the anode surface was defined as breakdown voltage and ignition time strongly depended on the concentrations and constituents of the electrolyte and a strong influence on the growth rate of coating. Therefore, type of electrolyte was one of the important factors influencing growth rate of the oxide layer on. Voltage (V) 5 4 3 2 1 Fig. 1 Breakdown voltage Stage Stage Stage Critical voltage Current density : 1 A/dm 2 1 2 3 4 5 6 7 Time, t/sec Voltage-time curves of PEO process in. Coating thickness, t / µm 25 2 15 1 5 BathB 1 2 3 4 5 6 7 Coating time, t/sec Fig. 2 Change of coating thickness with coating time of PEO process in.

Corrosion Resistance of Plasma-Anodized AZ91 Mg Alloy in the Electrolyte with/without Potassium Fluoride 673 Surface roughness, R a / µ m 1.6 1.4 1.2 1..8.6.4.2 current density : 1 A/dm 2. 1 2 3 4 5 6 7 Coating time, t / sec Fig. 3 Variation of surface roughness with coating time of PEO process in. The changes of the surface roughness (R a ) as a function of coating time in two different electrolytes were shown in Fig. 3. There was no change of surface roughness during the initial stage of PEO process. Surface roughness increased with increasing coating time. When the coating time was 6 s, surface roughness of exhibited much rougher than that of. It was considered that surface morphology of exhibited more coarse pores due to the faster growth rate of oxide layer in, abovementioned in Fig. 2. It was believed that the electrolyte containing potassium fluoride had an influence of growth rate of oxide layer on. Figure 4 shows surface morphology of oxide layers in coated in two different electrolytes. In case of coated for 12 s, the traces of scratches which were formed during mechanical polishing before PEO process were observed and fewer pores were observed. Especially, wide gaps were observed at the interface between -Mg and phases and inside of phases (Fig. 4(a)). Such a big gaps might be generated by the difference of oxide formation during the PEO process for and phases. 14,15) Distribution of irregular and coarse pores of oxide layer coated from the for 6 s could be related to non-uniform distribution of and phase (Fig. 4(b)). Oxide layer coated from the for 12 s showed the typical surface morphology for PEO process (Fig. 4(c)). For the oxide layer coated from the for 6 s, the oxide layer showed surface morphology of dense crater-like microstructures with some round-shape shrinkage pores observed in the crater centers. Oxide layer in exhibited irregular tiny pores with less than 12 mm in diameter and sometimes coarse pores are locally existed (Fig. 4(b)). showed bimodal distribution of pores due to coexistence of both relatively coarse pores with about 3 mm and fine pores with less than 12 mm in diameter (Fig. 4(d)). The results of the EDS analysis for the oxide layer were shown in Fig. 5. The component of consisted of oxygen, silicon, magnesium, and aluminum. The two later elements came directly from the oxidation of substrate (Fig. 5(a)). As shown in Fig. 5(b), fluorine ion was detected in the in addition to having component of the. Figure 6 shows cross-sectional images of oxide layers in coated in two different electrolytes. In general, the oxide layer formed on the during the PEO process composed of two different layers; an outer porous layer and an inner barrier layer, which was very thin layer with the thickness of 13 nanometers. 8,16,17) The pores only existed in the outer porous layer and interconnected with each other, but did not crossover through the inner barrier layer to the substrate. A lot of pores in the porous layer were observed in both samples. Figure 7 shows the EDS analysis on the oxide layer/substrate interface in the coated for 1 min. There was a difference of the composition between the barrier layer and the outer layer, especially for the fluorine content (Fig. 7(d)). Fig. 4 Surface morphology of oxide layer of formed by PEO process at 1 A/dm 2 in electrolyte with/without potassium fluoride; (a)(b): samples coated from for 12 and 6 s, respectively, (c)(d): samples coated from for 12 and 6 s, respectively.

674 D. Y. Hwang, Y. M. Kim and D. H. Shin Intensity Intensity Mg (a) 15 O 1 Si 5 Al Zn 1 2 3 4 5 6 25 Mg (b) 2 O 15 1 Si 5 F Al Zn K 1 2 3 4 5 6 Fig. 5 Energy-dispersive spectra of coated by PEO process; (a), (b). The EDS line scanning showed a fluoride enriched zone at the oxide layer/substrate interface. It was considered that fluorine ion in the electrolyte played an important role in the formation of barrier layer during the initial stage of PEO process. These observations were in agreement with the results of Cai and Liang. 9,12) It was also reported that the barrier layer had an influence on the corrosion resistance of magnesium alloys although the layer existed very thin. 18) In order to evaluate microstructure in the oxide layer in detail, the oxide layer was analyzed using cross-sectional TEM. Figure 8 shows cross-sectional TEM images of the sample in and B coated for 3 s. According to TEM observation in (Fig. 8(a), (b)), oxide layer of the mainly consisted of amorphous structure. A lot of pores in the oxide layer were observed on the whole. Sometimes, nanocrystalline structure was locally observed in the oxide layer (Fig. 8(a)). The SAD pattern in the oxide layer in was dim, indicating amorphous structure. The amorphous structure of outer layer might be caused by rapid solidification of oxide layer near surface. The plasma, which was generated as an arc-shape on the surface, could melt down the oxide layer causing the craters in the surface. If arcs fade out by depletion of ions in plasma, melted oxide would be solidified rapidly by heat loss through convection Fig. 6 Cross-section of oxide layer of in electrolyte with and without potassium fluoride coated for 6 s; (a) sample coated from, (b) sample coated from. Fig. 7 Cross-section SEM micrograph (a) and EDS spectra of Mg (b), O (c) and F (d) of in coated for 6 s.

Corrosion Resistance of Plasma-Anodized AZ91 Mg Alloy in the Electrolyte with/without Potassium Fluoride 675 Fig. 8 TEM micrograph of oxide layer of formed by PEO process at 1 A/dm 2 in electrolyte with/without potassium fluoride; (a), (b): samples coated from for 3 s, respectively, (c), (d): samples coated from for 3 s, respectively. and conduction to electrolyte. In this moment, oxide close to surface would be solidified much faster than inner layer. Therefore, outer layer could be formed as amorphous phase. Also, gases, existed inside of melted oxide by turbulence, could not escape, but they were trapped inside of oxide layer causing the big voids. It was observed that a lot of pores were existed in the inner layer on the substrate (Fig. 8(b)). Therefore, it was difficult to explain the difference between outer porous and inner barrier layer because of inner layer with the non-dense structure. In contrast, in case of, oxide layer consisted of porous and barrier layer (Fig. 8(c)). There was a clear line between porous and barrier layer. The barrier layer was very thin and dense layer without pores. The volume fraction of nanocrystalline structure in was much larger than that of (Fig. 8(c)). It was known from Fig. 8(d) that the substrate and inner layer was compatible each other and no pores was also observed. It was observed that the inner barrier layer contained higher amount of the fluorine ions, compared to the outer porous layer above-mentioned in Fig. 7. XRD patterns of the oxide layer of the coated from for 6 s are shown in Fig. 9. Oxide layer coated from containing fluorine ion consisted of MgO, Mg 2 SiO 4 and MgF 2. Liang 12) showed that electrolyte containing fluorine ion in the potassium hydroxide-silicate solution played an important role during the initial film formation. Therefore, it was indicated that fluorine ion in the electrolyte participates in the reaction and were incorporated into the oxide layer. It was considered that MgF 2 might be formed in the oxide layer due to the high chemical reactivity with magnesium substrate during the PEO process. Especially, it was known that volume fraction of the MgF 2 in the inner barrier layer existed more higher than that of the porous layer from the EDS line scanning. Intensity, I / a.u. 35 3 25 2 15 1 5 Mg MgO Mg 2 SOi 4 MgF 2 2 3 4 5 6 7 8 Scattering angle, 2Θ / degree Fig. 9 XRD pattern of in containing potassium fluoride coated for 6 s. 3.2 Corrosion resistance behavior of oxide layer on Corrosion resistance of the oxide layer was evaluated by electrochemical potentiodynamic polarization in 3.5 mass% NaCl solution. Figure 1 shows the potentiodynamic polarization curve of the samples in the electrolyte with and without potassium fluoride. It was well known that corrosion potential and current density of coated samples were often used to characterize corrosion protective property of the oxide layer. 19) In general, it was reported that the oxide layer with a high corrosion potential and low corrosion current density exhibited a good corrosion resistance. 2) In case of the samples coated for 12 s, the corrosion current density of containing fluorine ion was lower than that of without fluorine ion. The tendency of the corrosion current density for the samples coated for 6 s was similar to the same as for 12 s. However, their corrosion potential remained almost at a constant. As the coating time increased,

676 D. Y. Hwang, Y. M. Kim and D. H. Shin Potential, E/ V vs SCE -1.2 coating time : 12 s (a) current density : 1 A/dm 2 uncoated -1.4-1.6-1.8-2. uncoated Z", mω m 2 2.x1 4 current density : 1 A/dm 2 coating time : 6 s 1.5x1 4 Fit 1.x1 4 5.x1 3. 1x1 4 2x1 4 3x1 4 4x1 4 5x1 4 Potential, E / V vs SCE -1. -1.2-1.4-1.6-1.8-2. -2.2 1-9 1-7 1-5 1-3 1-1 coating time : 6 s current density : 1 A/dm 2 uncoated Current density, I / ma/dm 2 uncoated 1E-11 1E-9 1E-7 1E-5 1E-3.1 Current density, I / ma/dm 2 Fig. 1 Potentiodynamic polarization curves of the coatings formed in two different electrolytes; (a) coated for 12 s, (b) AZ91 Mg alloy coated for 6 s. corrosion current density of in both samples decreased and polarization resistance increased. Such improvement of corrosion resistance with time was mainly related with increasing of thickness of coated oxide layers. Corrosion current density of an oxide layer coated from Bath B for 6 s was slower than that of oxide layer coated from for 6 s because the thickness of the oxide layer coated from was only 18:75 mm, which was more than thicker than that of the oxide layer coated from for 6 s which was 12:69 mm as shown in Fig. 2. This result clearly showed that the thickness of the oxide layer played an important role in corrosion resistance of the oxide layer in. Compared to the uncoated AZ91 Mg alloy, the corrosion current density in the with potassium fluoride coated for 6 s decreased by approximately five times and polarization resistance also increased by approximately three times. These data clearly showed that the corrosion resistance of resulted in the better corrosion protective property of in solution containing chloride ion. In order to get more information about the corrosion phenomenon and mechanism, AC impedance measurement was carried out for oxide layer on coated from electrolyte with and without potassium fluoride. The (b) Z', mω m 2 Fig. 11 Nyquist plot of oxide layer coated from the electrolyte with and without potassium fluoride for 6 s. Fig. 12 Equivalent circuit used for impedance data fitting of oxide layer. nyquist plot for the oxide layer coated for 6 s is shown in Fig. 11. Similar with the potentiodynamic polarization experiment, corrosion resistance of oxide layer coated from the electrolyte with potassium fluoride was superior to that of oxide layer coated from the electrolyte without potassium fluoride. The electrode processes were in series or parallel with each other due to the complicated structure of oxide layer on the Mg alloys coated by PEO process. So it was essential to develop the appropriate models for the impedance which could then be used to fit the experimental data and extract the parameters which characterized the corrosion process. The simplified equivalent circuit on the oxide layer coated from electrolyte with and without potassium fluoride is proposed as shown in Fig. 12. In the equivalent circuit, the electrolyte resistance (Rs) was in series with the unit of the oxide layer system. R p was outer porous layer resistance paralleled with constant phase element (CPE p ). The properties of the inner barrier layer were described by the resistance R b in paralleled with CPE b. In order to include a surface inhomogeneity factor and a possible diffusional factor, a more general constant phase element (CPE) was used instead of a rigid capacitive element. Based on equivalent circuit model in Fig. 12, the nyquist plot is best fitted and the fitting result is shown in Fig. 11 as solid lines passing through the testing results. The corresponding values of the equivalent elements are listed in Table 2. It was reported that low frequency range of impedance diagram characterized the inner layer properties and the high frequency range reflected outer layer. 21,22) The oxide layers obtained electrolyte with and without potassium fluoride showed the different EIS behavior. The fitting results really showed that the resistance of inner barrier layer against relatively thin layer was higher than the

Corrosion Resistance of Plasma-Anodized AZ91 Mg Alloy in the Electrolyte with/without Potassium Fluoride 677 Table 2 Equivalent circuit data of oxide layer coated from the electrolyte with and without potassium fluoride for 6 s. R s (mm 2 ) CPE p (mfm 2 ) CPE1-P R p (mm 2 ) R b (mm 2 ) CPE b (mfm 2 ) CPE2-P 13.66 8.49E-5.63 8.28E3 4.93E4 5.74E-4.86 18.4 1.46E-4.35 4.67E3 8.8E5 2.73E-6.92 corresponding value of the outer porous layer. When chloride ions penetrated through the outer porous layer reached the inner barrier layer, penetration process of chloride ion into barrier layer was a slow diffusion reaction for its much compact structure layer. This showed that inner barrier layer in the oxide layer coated from the electrolyte with potassium fluoride had a good influence of the corrosion resistance of Mg alloy. This layer mainly composed of MgO and MgF 2. It was considered that fluorine ion played an important role on the property of barrier layer and corrosion resistance of. Although growth rate of oxide layer in the was higher than that of due to the presence of potassium fluoride ion in the electrolyte, most of the layer growth was focus on the growth of porous layer as shown in TEM results. Therefore, it was considered that corrosion resistance of the oxide layer on was dominantly improved due to the structure and composition of barrier layer rather than thickness of oxide layer. The results of salt spray test for 72 ks in this study are shown in Fig. 13. The result of salt spray test was in agreement with the corrosion resistance of potentiodynamic polarization tests. Severe filiform corrosion was observed in surface of in coated for 12 s. In case of the sample in coated for 1 min, pitting corrosion was locally showed in the surface of the sample. It was revealed that this kind of structure of oxide layer cannot provide perfect protection for their substrate. Z. Shi 23) suggested that corrosion mechanism of coated by PEO process could be attributed to the microgalvanic effect between the and the phases in the Mg Al alloys in 5 mass% NaCl solution. Above-mentioned in Fig. 4, therefore, it was considered that corrosion of the sample in was occurred at two-phase boundaries due to the microgalvanic corrosion. In contrast, it was observed that the surface of the in was entirely not corroded regardless of coating time. Only pitting corrosion was locally observed in coated for 12 s. The corrosion resistance curves of are similar to the thickness curves, indicating that the thickness of the oxide layer played an important role in corrosion resistance of the oxide layer in. It was believed that the corrosion resistance of depends on the existence of fluoride compound in the oxide layer. Compact barrier-type passive film in containing potassium fluoride was formed due to the presence of fluorine ions. 4. Conclusions Fig. 13 Appearance of coated after salt spray test (72 ks). (1) PEO coating behavior of AZ91Mg alloy in the different electrolytes with and without potassium fluoride was investigated. Growth rate of coating in exhibited much higher than that of. The oxide layer formed on in electrolyte with potassium fluoride and sodium silicate consisted of MgO, MgF 2 and Mg 2 SiO 4. (2) The corrosion resistance of the sample coated from electrolyte with potassium fluoride was superior to that of the sample without potassium fluoride. As a result of salt spray test, it was observed that the surface of the in was entirely not corroded regardless of coating time. The corrosion resistance curves of were similar to the thickness curves, indicating that the thickness of the oxide layer played an important role in corrosion resistance of the oxide layer in. (3) The corrosion resistance of fabricated by PEO process depended on the existence of the fluoride compound in the oxide layer. The oxide layer formed in in electrolyte containing potassium fluoride was found to be a compact barrier-type passive film in presence of fluoride ions. The existence of the dense MgO and MgF 2 in the barrier layer might have a favorable effect on the corrosion resistance of the coated from by PEO process. Acknowledgements This research was supported by a grant from the Center for Advanced Materials Processing (CAMP) of the 21st Century Frontier R&D Program funded by the Ministry of Knowledge Economy (MKE), Republic of Korea.

678 D. Y. Hwang, Y. M. Kim and D. H. Shin REFERENCES 1) J. E. Gray and B. Luan: J. Alloys Comp. 336 (22) 88 113. 2) L. Rudd, C. B. Breslin and F. Mansfeld: Corros. Sci. 42 (2) 275 288. 3) R. Ambat, N. N. Aung and W. Zhou: Corros. Sci. 42 (2) 1433 1455. 4) G. Song, A. Atrens, D. ST John, X. Wu and J. Nairn: Corros. Sci. 39 (1997) 1981 24. 5) H. Umehara, M. Takaya and Y. Kojima: Mater. Trans. 42 (21) 1691 1699. 6) A. Yfantis, I. Paloumpa, D. Schmeiber and D. Yfantis: Surf. Coat. Technol. 151 (22) 4 44. 7) A. L. Yerokhin, X. Nie, A. Leyland, A. Matthews and S. J. Dowey: Surf. Coat. Technol. 122 (1999) 73 93. 8) V. Birss, S. Xia, R. Yue and R. G. Rateick Jr.: J. Electrochem. Soc. 151 (24) B1 1. 9) Q. Cai, L. Wang, B. Wei and Q. Liu: Surf. Coat. Technol. 2 (26) 3727 3733. 1) S. J. Xia, R. Yue, R. G. Rateick Jr. and V. I. Birss: J. Electrochem. Soc. 151 (24) B179 B187. 11) H. Guo and M. Z. An: Appl. Surf. Sci. 246 (25) 229 238. 12) J. Liang, B. Guo, J. Tian, H. Liu, J. Zhou and T. Xu: Appl. Surf. Sci. 252 (25) 345 351. 13) S. Verdier, N. van der Laak, S. Delalande, J. Metson and F. Dalard: Appl. Surf. Sci. 235 (24) 513 524. 14) Z. Shi, G. Song and A. Atrens: Corros. Sci. 47 (25) 276 2777. 15) O. Khaselev and J. Yahalom: J. Electrochem. Soc. 145 (1998) 19 193. 16) Z. Shi, G. Song and A. Atrens: Corros. Sci. 47 (25) 276 2777. 17) O. Khaselev, D. Weiss and J. Yahalom: Corros. Sci. 43 (21) 1295 1397. 18) A. L. Yerokin, V. V. Lyubimov and R. V. Ashitkov: Ceram. Int. 24 (1998) 1 6. 19) E. Gulbrandsen, J. Tafto and A. Olsen: Corros. Sci. 34 (1993) 1423 144. 2) Y. Zhang, C. Yan, F. Wang, H. Hou and C. Cao: Surf. Coat. Techol. 161 (22) 36 43. 21) F. Mansfeld and M. W. Kendig: J. Electrochem. Soc. 135 (1988) 828 833. 22) S. C. Chung, J. R. Cheng and S. D. Chiou: Corros. Sci. 42 (2) 1249 1268. 23) Z. Shi, G. Song and A. Atrens: Corros. Sci. 48 (26) 3531 3546.