Characterization of anodized and sealed aluminium by EIS

Characterization of anodized and sealed aluminium by EIS

Corrosion Science 45 (2003) 611–624 www.elsevier.com/locate/corsci Characterization of anodized and sealed aluminium by EIS J.J. Suay *, E. Gimenez,...

350KB Sizes 11 Downloads 96 Views

Corrosion Science 45 (2003) 611–624 www.elsevier.com/locate/corsci

Characterization of anodized and sealed aluminium by EIS J.J. Suay *, E. Gimenez, T. Rodrıguez, K. Habbib, J.J. Saura Ciencia de los Materiales e Ingenieria Metalurgica, Departamento de Tecnologıa, Campus de Riu Sec, Universidad Jaume 1, 12071 Castell on, Spain Received 12 September 2001; accepted 18 June 2002

Abstract Anodized and sealed aluminium samples exposed to different atmospheres for up to three years have been studied by means of electrochemical impedance spectroscopy (EIS). EIS was used to obtain detailed information concerning the electrochemical properties of the porous and barrier layer of anodized aluminium. An equivalent circuit that reproduces the a.c. impedance results of porous aluminium oxide films is proposed. The results reveal that the EIS technique is a good tool for obtaining detailed information on the influence of autosealing and the ageing process on anodized aluminium. This research shows that porous layer sealing quality increases over months and years (especially in the first 24 months) as the ageing proceeds when exposed to the natural atmospheric conditions. The analysis was completed with the aid of the SEM technique. Ó 2002 Elsevier Science Ltd. All rights reserved.

1. Introduction The anticorrosive properties of aluminium are strongly increased by the anodizing process, which produces a closed-packed porous array of columnar hexagonal cells normal to the substrate surface and separate from it by a barrier layer. The anodizing process is often done in sulphuric acid solutions with external electric power [1]. The porous structure of the anodic layers provide the aluminium with absorbent characteristics which increase its applications (for example anodized aluminium can be coloured by different techniques in order to use it in ornamental and household

*

Corresponding author. Tel.: +34-964-728146; fax: +34-964-728106. E-mail address: [email protected] (J.J. Suay).

0010-938X/03/$ - see front matter Ó 2002 Elsevier Science Ltd. All rights reserved. PII: S 0 0 1 0 - 9 3 8 X ( 0 2 ) 0 0 1 3 7 - 3

612

J.J. Suay et al. / Corrosion Science 45 (2003) 611–624

applications). Nevertheless, in those applications where the anodized aluminium is atmospherically exposed, sealing of the anodic layers is required because porosity decreases the corrosion resistance. The sealing process achieves an important reduction in the porosity of the anodic layer. Atmospheric exposure of anodized aluminium produces significant changes in the anodic layer (due to the autosealing process) increasing the material corrosion resistance [2–4]. The corrosion resistance depends on layer thickness and especially on the sealing quality, so good quality control on both parameters has maximum significance. There are many different experimental evaluation techniques which offer interesting information, but all of them have their limitations [4–6]. The application of electrochemical impedance spectroscopy (EIS) to the oxide aluminium layer characterization has many advantages. The results obtained up until now show that EIS is a powerful tool able to provide tiny pieces of information on the anodic aluminium layerÕs electrochemical properties, and how atmospheric exposure can modify the anodized samplesÕ properties [6–12]. The experimental impedance results obtained will be modelled using equivalent electrical circuits, and the study of the corresponding electrical parameters will be used to elucidate anodic and sealed layer electrochemical behaviour. This research will try to contribute to clarify the autosealing process in different atmospheric conditions by means of EIS. 1.1. Equivalent electrical circuits Since the pioneering work of Hoar and Wood [13] involving an impedance bridge, various equivalent circuits (ECs) have been proposed to model the response of porous aluminium oxide films under different conditions. Hoar and Wood suggested ECs to model unsealed and partially sealed anodized films [8]. Lorenz and other authors proposed the ‘‘passive pit model’’ (where it is assumed that pits have penetrated only the outer porous layer without damaging the barrier layer) and the active pit model (where it is assumed that pits have penetrated both the porous and barrier layer) [11,12,14–17]. As pointed out by Mansfeld and Kendig [11], the passive pit model can also be used to interpret the sealing mechanism.

2. Experimental details 2.1. Materials Aluminium alloy EN AW (UNE 573-3) [18] samples commercialized as ALPUR 50-H24 were used. The alloyÕs composition is given in Table 1. The samples were anodized in an industrial process and conditions, in a sulphuric acid solution (18% i.w.), at 20 °C and 1.5 A/dm2 . Anodized aluminium samples were sealed in boiling water for 45 min. Samples of 50  100 mm2 size were exposed to different atmospheric conditions for up to three years.

J.J. Suay et al. / Corrosion Science 45 (2003) 611–624

613

Table 1 Aluminium alloy (EN AW 1050 a) composition (%) Si 0.25

Fe 0.40

Cu 0.05

Mn 0.05

Mg 0.05

Cr

Zn 0.07

Ti 0.05

Other 0.03

Al 99.50

Table 2 Environmental data for the five natural exposure sites Site

HTa (h/year)

SO2 (mg/(m2 day))

Cl (mg/(m2 day))

Castell on Grao de Castell on Villarreal Onda Morella

4449 6263 4427 4371 3792

4.386 4.021 2.987 3.117 3.060

4.036 18.739 6.075 4.331 4.436

a

HT holds for humectation time: cumulative periods having T > 0 °C and RH P 80%.

2.2. The atmospheric exposure test Five atmospheric exposure stations were set-up in Valencia Community (Spain): in Castell on, El Grao de Castell on, Villarreal, Onda and Morella, where their atmospheric conditions can be classified as: urban, marine, semi-industrial (both the third and fourth) and rural respectively, and all of them belonging to ISO C2 class [19]. In order to correlate materials corrosion and atmosphere corrosivity, steel samples were also exposed. Environmental characteristics of the five corrosion sites are given in Table 2. SO2 , chloride and humectation time (HT) values, are average values corresponding to the first year of atmospheric exposure. SO2 determination was carried out by the lead dioxide method, while chlorides by the wet candle method [20]. The ISO standards [19] establish an atmospheric classification in four categories depending on sulphur substances quantity (P0 to P3 ) represented by sulphur dioxide, and salinity (S0 to S3 ) represented by chlorides ions. All the atmospheric conditions studied belong to the lowest category of SO2 content (P0 ). On the other hand, the atmosphere of the Grao of Castell on had the highest chlorides level (18 mg/ (m2 day)), nevertheless all stations belong to the same category S1 . The marine atmosphere had the highest HT value, which is representative of high corrosion rates. 2.3. The test method The experimental impedance results were obtained on anodized sealed aluminium immersed in a K2 SO4 (3.5% i.w.) solution, without stirring and at 25  2 °C. 2.4. The test equipment The electrochemical impedance tests were carried out in an electrochemical cell where a substrate surface area of 14.4 cm2 was exposed to the solution. This cell was

614

J.J. Suay et al. / Corrosion Science 45 (2003) 611–624

the classical glass tube fixed on the metal surface. This cell holds the electrolyte, the reference electrode and the flexible graphite counter electrode. The impedance data was obtained at the open-circuit potential using a potentiostat and a frequency response analyzer. The impedance measurements were conducted over a frequency range of 100 kHz down to 104 Hz (five frequencies per decade) using a 15 mV amplitude of sinusoidal voltage in a Faraday cage to minimize external interferences. 3. Results Fig. 1 show SEM micrographs of the aluminium anodic layer surface corresponding to the intermediate layer which have been formed during sealing process.

Fig. 1. (a) The SEM micrograph of the reference aluminium layer surface sample not exposed to natural atmospheric conditions. (b) The SEM micrograph of aluminium anodic layer surface natural exposed for 26 months to urban atmospheric conditions (Castell on site). (c) The SEM micrograph of aluminium anodic layer surface natural exposed for 36 months to urban atmospheric conditions (Castell on site).

J.J. Suay et al. / Corrosion Science 45 (2003) 611–624

615

This layer does not allow us to see any more of the anodic layer structure (like the hexagonal cells). Fig. 1a corresponds to a sample which has not been exposed (reference sample). As it can be seen the surface structure is heterogenous and it is full of ‘‘cavities’’. Fig. 1b and c correspond to samples exposed in the Castell on station (urban atmosphere) for 26 and 36 months respectively. When the sample has been naturally exposed to Castell on atmosphere for 26 months, it can be seen that the ‘‘cavities’’ showed in the reference sample surface started to vanish due to the fact that they were being filled by corrosion products. The rest of the anodic surface starts to be covered by oxide particles too. Finally, when the samples had been naturally exposed to three years to atmospheric conditions, the cavities disappear (because oxide particles accumulation) and the surface is quite homogeneous and totally covered by an oxide particle layer.

Fig. 2. (a) The Nyquist plot of sealed anodized aluminium reference sample (j) and naturally exposed for on site atmosphere. (b) Bode plot periods of 9 ( ), 15 (M), 26 (O) and 35 (}) months to the Castell (impedance module) of sealed anodized aluminium reference sample (j) and naturally exposed for periods of 9 ( ), 15 (M), 26 (O) and 35 (}) months to the Castell on site atmosphere. (c) Bode plot (phase angle) of sealed anodized aluminium reference sample (j) and naturally exposed for periods of 9 ( ), 15 (M), 26 on site atmosphere. (O) and 35 (}) months to the Castell







616

J.J. Suay et al. / Corrosion Science 45 (2003) 611–624

SEM micrographs were also carried out on the aluminium samples cross-sections. It was verified that the alumina layer was continuous with almost constant thickness (27–30 lm) along the surface. The cavities observed seemed to be caused by irregularities in the aluminium substrate and by the presence of different alloy particles that produced an irregular growth of this layer. Only a few points of localised corrosion have been found in the anodized aluminium samples (and only in those samples exposed to marine atmospheric conditions for more than 15 months), so we consider that the influence of the localised corrosion can almost be ignored. Fig. 2a–c show Nyquist and Bode plots (impedance and phase angle) for the reference samples as well as the exposed ones after 9, 15, 26 and 35 months in the Castell on site. Atmospheric exposure significantly affects the medium and high frequency domaines. The Nyquist plots show that for anodized and sealed aluminium there is a high frequency arch and capacitive behaviour at lower frequencies. The high frequency arch increases as ageing time increases. The Bode plots show that there is a displacement of the minimum angle to lower frequencies as atmospheric exposure progresses. It can be seen that horizontal part of the impedance modulus in the Bode plots increases up twice its level (depending on the atmospheric station) for exposure time, and especially in the first 9 months. This process can be due to the anodic layer autosealing phenomena because of the atmospheric exposure.

4. Discussion The experimental electrochemical results or our reference samples are very similar to those measured by other research groups working on similar anodic and sealed aluminium samples [7]. These groups [7] have characterized by means of TEM these types of samples. The micrographs showed a semi-porous closed-packed array of columnar hexagonal cells consisting of a central pore surrounded by a thick oxide wall normal to the substrate surface. When sealed anodized samples are exposed to atmospheric conditions the anodized layer is able to absorb atmospheric humidity (the porous layers have absorbant properties even if they are sealed), so pores can be filled by water if they are not perfectly sealed. Also atmospheric exposure causes autosealing and the ageing process which affects significantly the alumina porous layer. As Gonzalez et al. proposed [4], the mouths of the pores in the alumina film complete their closing outside the sealing bath and during the ageing process (where the samples are exposed to natural atmospheric conditions). Possible explanations of the sealing and aging process could be: (1) The bottom of the pores (in the anodic layer) is filled during immersion in the sealing solution but the pore mouth is only partially blocked, so pores have the capacity to continue absorbing water during ageing in humid environments. There is a quite low impedance spectra.

J.J. Suay et al. / Corrosion Science 45 (2003) 611–624

617

(2) The pores are virtually filled up at the start of the ageing process at moderate or high RH. Aluminium and others hydroxides precipitate at pore mouths and all the quality thresholds are reached. The corrosion resistance increases. There is a remarkable increase in the impedance value. Whereas at low RH water desorption can take place. (3) Inside pore changes continue to take place throughout the ageing process. The ageing mechanism involves degradation, gelling, agglomeration and a precipitation process. We have tried to study the sealing quality and the anodic layer changes, for the autosealing and the ageing process produced by atmospheric exposure, by means of electrochemical impedance spectra. An EC has been proposed in order to fit experimental data and to be able to analyze the alumina layerÕs physicochemical changes.

5. Results from models The anodic process will produce a compact packed hexagonal cell, located in a perpendicular plane to the substrate, and each one of the cells with a central pore of 100–200 nm of diameter, with a surrounding wall of 100–200 nm in thickness and everything separate from the substrate by the barrier layer. The hexagonal cells and barrier layer are made from anhydride alumina with different sulphate ions Al2 (SO4 )3 from the anodizing bath. On the other hand, the porous layer is full of hydrate alumina (mono- or trihydrate or a mixture) plus absorbed water [4–8]. Finally there is one last layer (the intermediate layer), which is formed by diffusion from the exterior, especially in the sealing process, and slowly in atmospheric exposure [1]. It is thought that significant changes in the porous layer are to be expected because of the atmospheric exposure (water absorption and the transformation of anhydride alumina to hydrate alumina inside the cells) [7], and in the intermediate layer [21]. Big changes in the porous cell walls and barrier layer are not expected. Figs. 3 and 4 show an equivalent proposed circuit to model the EIS experimental results of the anodic aluminium layer. The parameter Rsol represents electrolyte used in the test resistance. R1 tries to simulate the presence of electrolyte in the pores and faults of the intermediate layer. One of the parallel branches which are present in the circuit is formed by the resistance Rpw and the associate capacitance Cpw and represent the walls of the hexagonal cells with uniform or nearly uniform dielectric properties. Taking into account that Rpw is usually so high that it prevents the passage of current across it we will use EC of Fig. 3 for the modelisation with only a capacitance Cpw in this branch. The other branch will be formed by the resistance R1 of the electrolyte in the pores and defects of the intermediate layer, in series with the pores as a whole (the pore and the barrier layer lying at their bases). The electrochemical properties of the pore filling are represented by Rp and Cp , while the properties of the barrier layer are represented by Rb and Cb .

618

J.J. Suay et al. / Corrosion Science 45 (2003) 611–624

Fig. 3. The proposed EC to model the impedance spectra of the sealed anodized aluminium layer.

Fig. 4. The physical significance of the passive elements forming the proposed EC.

Fig. 5a and b show the fitting of the experimental results for an anodized and sealed aluminium aged in Castell on site for different time with the proposed EC. As it can be seen in the Nyquist and Bode plots (Fig. 5a and b respectively), the model results are significantly similar to the experimental data.

J.J. Suay et al. / Corrosion Science 45 (2003) 611–624

619

Fig. 5. (a) The Nyquist plot showing experimental and model data of samples exposed to Grao de Castell on atmosphere for periods of 9 (experimental ( ) and modelized ( )), 26 (experimental (N) and modelized (M)) and 36 months (experimental (j) and modelized ()). (b) The Bode plot showing experimental and model data of samples exposed to Grao de Castell on atmosphere for 36 months (experimental data full symbol and modelized data open symbol).





Resistance R1 (Fig. 6) is directly related to those intermediate layer changes as result to atmospheric exposure. Parameter R1 vary markedly with the extent of sealing and the growing of the intermediate layer thickness (both process are closely related and depend on ageing time). Parameter R1 doubles in size for all types of atmospheres when the samples have been exposed up to 39 months. R1 increases markedly during the first 20 months, reaching values around 5 MX and being almost constant after this exposure time. This change can be due to a better sealing of pores because hydroxides precipitation increases with exposure time and/or to a

620

J.J. Suay et al. / Corrosion Science 45 (2003) 611–624

Fig. 6. The changes in the R1 parameter with exposure time and atmospheric site (Castell on (), Grao ( ), Villarreal (M), Onda (O) and Morella (})).



intermediate layer thickness increment (which can be observed in Fig. 1a–c). As exposure time increases precipitate particles start covering the alumina layer surface making cavities disappear and finally forming after years of exposure a homogeneous anodic layer surface. Intermediate layer changes (assessment by R1 evolution) go slower in rural atmospheric conditions (Morella site), as a probable consequence of a more inactive atmosphere (in this site lower HT and contaminant values were recorded than at other sites). The capacity Cpw can be calculated from the expression Cpw ¼ ee0 ðS=dÞ, being S ¼ 1 cm2 , e ¼ 10 (alumina permittivity) and e0 ¼ 8:854  104 F/cm (vacuum dielectric constant) and d, the layer thickness. Cpw values were found to be from 2  109 to 6  1010 F (Fig. 7) which permit us to calculate d values until 14.7 lm. Experimental d measurements by micrography on different samples (exposed and reference samples) gave us values from 27 to 30 lm which are more than double than those found using the models. Nevertheless, the contamination of anhydrous alumina by sulphate ions can increase the anodic layer permittivity, which makes the Cpw calculated value quite plausible. The values of Cb go from 1 to 10 lF (Fig. 8) with exposition time. If a permittivity of 20 is also assumed for the barrier layer, Cb values would lead to a barrier layer thickness of 17.7 nm maximum. This is consistent with the result expected from an anodizing voltage of 15 V (the films thus formed are widely accepted to be 1.0–1.4 nm V1 thick) [3]. As can be seen 1000 times thicker porous layer give, approximately, 1000 times smaller Cpw values. The parameter Cp must be the capacitance of the pore filling (Fig. 9), in parallel to the hexagonal cell walls. As hydration and the ageing process develops (i.e. as the exposure time increases), the anhydrous alumina reacts with absorbed water giving hydrated alumina. As there is a free water content reduction inside pores because hydrated alumina formation, Cp decreases (the hydrated alumina permittivity is

J.J. Suay et al. / Corrosion Science 45 (2003) 611–624

621

Fig. 7. The changes in the Cpw parameter with exposure time and atmospheric site (Castell on (), Grao ( ), Villarreal (M), Onda (O) and Morella (})).



Fig. 8. The changes in the Cb parameter with exposure time and atmospheric site (Castell on (), Grao ( ), Villarreal (M), Onda (O) and Morella (})).



lower than alumina þ free water permittivity). This effect is more significant in those samples exposed to atmospheres with lower HT (Villarreal, Onda and Morella sites). Nevertheless Cp value remains almost constant after three years of exposition in Grao de Castell on site. It can be due to the HT high values found in this site which is very close to the sea (the HT high value can promote a major quantity level of free water inside the pores). Because the time constant results in characteristics frequencies in the nHz range which are quite far from the lowest frequency used in the tests (1 mHz), estimated (extrapolated) Rb values will be highly uncertain. However Rb is higher to the GX cm2 range (Fig. 10). If this order of magnitude and a barrier layer thickness of

622

J.J. Suay et al. / Corrosion Science 45 (2003) 611–624

Fig. 9. The changes in the Cp parameter with exposure time and atmospheric site (Castell on (), Grao ( ), Villarreal (M), Onda (O) and Morella (})).



Fig. 10. The changes in the Rb parameter with exposure time and atmospheric site (Castell on (), Grao ( ), Villarreal (M), Onda (O) and Morella (})).



20 nm are accepted, then the barrier layer resistivity will be qb ¼ 1  109 X cm2 =20  107 cm ¼ 5  1014 X cm, which is consistent with the data for alumina in the Goodfellow catalogue [8]. Porous layer resistance (Rp ) value (Fig. 11) is observed to be six orders of magnitude lower than the associated barrier layer resistance (Rb ). Therefore, the pore filling is much better conductor than the cell walls and barrier layer. As it can be seen in Fig. 11 Rp value increase with exposure time until 1 order of magnitude for all atmospheric sites after three years of exposure (this result is quite similar to that

J.J. Suay et al. / Corrosion Science 45 (2003) 611–624

623

Fig. 11. The changes in the Rp parameter with exposure time and atmospheric site (Castell on (), Grao ( ), Villarreal (M), Onda (O) and Morella (})).



found by Gonz alez et al. in [4]). However this increase is significantly localised in the 15 first months. In Fig. 11 can be seen too that Rp increases goes slower in samples exposed to Morella site (rural station) than in the rest. A possible explanation is that alumina conversion to hydrated alumina is retarded because of the lower HT in the local environment.

6. Conclusions The EIS technique offers the possibility of studying carefully the natural sealing and ageing process in anodized aluminium. This technique allows you to monitor changes in the intermediate, barrier and porous layers during the ageing process. The changes in the passive elements over time due to different atmospheric exposure showed that the ageing process significantly affects the anodic layer in the first 24 months of exposure, while after three years of exposure, the parameter values are very similar in all atmospheric locations tested (urban, rural, marine and industrial sites). It can be observed in this study that the anodic layer quality improves with exposure time in all types of sites, so anodized and sealed aluminum anticorrosion performance is increased due to sealing and natural ageing process.

Acknowledgements The authors wish to thank X.R. N ovoa, M. Izquierdo, J. Ortega, J.L. Godes and E. Romero for their dedication and contribution to this investigation. This work was supported by Fundaci on Caixa––Castell o (P1A95––12 Project).

624

J.J. Suay et al. / Corrosion Science 45 (2003) 611–624

References [1] A.W. Brace, The Technology of Anodizing Aluminium, Tecnicopy Ltd., Stonenhouse, Gloucesterhire, Great Britian, 1979, pp. 1–19. [2] M.R. Kalantary, D.R. Gabe, D.H. Ross, Plat. Surf. Finish. 80 (1993) 52. [3] M.R. Kalantary, D.R. Gabe, D.H. Ross, J. Appl. Electrochem. 22 (1992) 268. [4] J.A. Gonz alez, S. Feliu, J.A. Bautista, E. Otero, J. Appl. Electrochem. 29 (1999) 845. [5] A. Bautista, V. L opez, E. Otero, R. Lizarbe, J.A. Gonzalez, Rev. Metalurg. 34 (1998) 57. [6] R. Lizarbe, V. L opez, E. Otero, J.A. Gonzalez, Rev. Metalurg. 26 (1990) 359. [7] J.A. Gonz alez, V. L opez, A. Bautista, E. Otero, J. Appl. Electrochem. 29 (1999) 229. [8] F. Mansfeld, G. Zhang, C. Chen, Plat. Surf. Finish. 84 (1997) 72. [9] R. Lizarbe, J.A. Gonzalez, E. Otero, V. L opez, Aluminium 69 (1993) 548. [10] F. Mansfeld, M.W. Kendig, Corrosion 41 (1985) 490. [11] F. Mansfeld, M.W. Kendig, J. Electrochem. Soc. 135 (1998) 828. [12] J. Hizing, K. Juettner, W.J. Lorenz, W. Paatsch, Corros. Sci. 24 (1984) 945. [13] T.P. Hoar, G.C. Wood, Electrochim. Acta 7 (1962) 333. [14] J. Hitzing, K. Juettner, W.J. Lorenz, W. Paatsch, J. Electrochem. Soc. 133 (1986) 887. [15] K. Juettner, W.J. Lorenz, W. Paatsch, Corros. Sci. 29 (1989) 279. [16] K. Juettner, Electrochim. Acta 35 (1990) 1501. [17] F. Mansfeld, Electrochim. Acta 35 (1990) 1533. [18] UNE-EN 573-3: Aluminio y aleaciones de aluminio. Composici on quımica y forma de productos de forja, Composici on quımica, Agosto 1994. [19] ISO 9225: Corrosion of metals and alloys, Corrosivity of atmospheres classification, first ed. International Standards Organization, 1992. [20] ISO 9225: Corrosion of metals and alloys, Measurement of pollution, first ed. International Standards Organization, 1992. [21] F. Mansfeld, Corrosion 44 (1988) 856.