Flow modulated ionic migration during porous oxide growth on aluminium

Flow modulated ionic migration during porous oxide growth on aluminium

Electrochimica Acta 55 (2010) 7044–7049 Contents lists available at ScienceDirect Electrochimica Acta journal homepage: www.elsevier.com/locate/elec...

411KB Sizes 0 Downloads 40 Views

Electrochimica Acta 55 (2010) 7044–7049

Contents lists available at ScienceDirect

Electrochimica Acta journal homepage: www.elsevier.com/locate/electacta

Flow modulated ionic migration during porous oxide growth on aluminium M. Curioni ∗ , E.V. Koroleva, P. Skeldon, G.E. Thompson Corrosion and Protection Centre, School of Materials, The University of Manchester, Manchester M13 9PL, UK

a r t i c l e

i n f o

Article history: Received 21 April 2010 Received in revised form 28 June 2010 Accepted 29 June 2010 Available online 6 July 2010 Keywords: Electrochemical impedance spectroscopy Anodic oxides Ionic migration Anodizing Flow

a b s t r a c t The ability to tailor the morphology of porous anodic alumina (PAA) drives applications in the important architectural, aerospace, packaging, lithography and nanotechnology sectors with the mechanisms regulating oxide growth being pursued vigorously. Here, we have probed the barrier layer of PAA during film growth, using high voltage, in situ, electrochemical impedance spectroscopy. Our new findings account for aluminium oxidation at the metal/oxide interface, ionic migration and interaction of charge carriers through the layer, and oxide viscous flow under the field. Separation of these processes by appropriate modelling discloses that ionic migration, being driven by the electrical field only, is independent of the electrolyte. The incorporated electrolyte-derived species modify the mechanical properties of the oxide, by regulating the oxide displacement from the barrier layer. The approach developed provides the vital foundations for an integrated theory of porous film growth on a range of metals in various film-forming electrolytes. © 2010 Elsevier Ltd. All rights reserved.

1. Introduction Porous anodic alumina (PAA) films are generated by anodic polarization of aluminium in aqueous electrolytes and their properties can be tailored by selecting the anodizing conditions to obtain ordered templates [1–5], corrosion [6–9] or wear resistant layers and coloured surfaces [10–13]. During porous film growth, new material is generated within the barrier layer by migration of charged species (aluminium and oxygen ions) under the electric field. Previous work on barrier-type oxides [14,15] has revealed that (i) growth proceeds at an efficiency close to 100%, 60% and 40% of the oxide is generated close to the metal/oxide interface and to the oxide/electrolyte interfaces respectively [14], (ii) the anodic oxide is amorphous and migration proceeds by a co-operative transport mechanism [14] and (iii) ionic transport is the rate-determining step [15]. It has also been suggested that the oxide may exhibit semiconductive properties due to the nature and distribution of charge carriers during growth [15,16]. Electrochemical impedance spectroscopy has been applied rarely to the study of growing barrier-type films because the oxide thickness increases with the charge passed and, therefore, the use of a frequency sweep is inappropriate for such non-stationary system. However, De Wit et al. acquired a series of single frequency impedance measurements during linear voltage sweeps, obtaining potential dependent spectra [17]. The resultant Nyquist plots

∗ Corresponding author. Tel.: +44 0 161 306 5971; fax: +44 0 161 306 4865. E-mail address: [email protected] (M. Curioni). 0013-4686/$ – see front matter © 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.electacta.2010.06.088

revealed a capacitive arch and an inductive loop, which were associated respectively with the dielectric properties of the oxide and the occurrence of relaxation processes during growth. In electrochemical or semiconductive systems that display similarities with a growing anodic oxide (i.e. supra-linear current–voltage dependence [18,19], a reaction involving intermediates [20,21], DC biasing [18,19] or p–n junctions [22,23], inductive contributions are often revealed and interpreted as relaxation phenomena [17], change in concentration of adsorbed intermediates [20,21], or negative capacitance resulting from charge distribution [23]. Information regarding barrier-type films can be transferred to porous-type films, with the difference that porous film growth proceeds at an efficiency of about 60% due to ejection of outwardly mobile aluminium ions into the electrolyte. Further, pore formation accompanies ionic conduction and, after an initial transient, a barrier layer with a steady thickness is generated, enabling the use of a frequency sweep to study the growth. Concerning pore formation, much research on oxides generated with specific combinations of anodizing parameters has been undertaken, but an integrated theory of porous anodic film growth is currently awaited. O’Sullivan and Wood [24] pioneered the field by using transmission electron microscopy of fracture sections to reveal the oxide morphology. They concluded that the morphology results from a dynamic equilibrium between generation of oxide and thermally enhanced, field-assisted dissolution at the pore base. However, recent evidence, based on the ratio of the thickness of the oxide generated to the aluminium consumed and observation of the displacement of ion tracers within the barrier layer [25–29], has suggested that, for sulphuric, oxalic and phosphoric acids, the contribution to film growth from displacement of newly formed

M. Curioni et al. / Electrochimica Acta 55 (2010) 7044–7049

7045

oxide material from the barrier layer towards the cell walls dominates dissolution-related effects. Such displacement is possible due to plasticization of the oxide material under the electric field and the action of growth and electrostriction stresses within the barrier layer. 2. Experimental Anodizing was undertaken on superpure (99.99 wt.%) aluminium specimens in stirred 0.3 and 0.6 M oxalic acid, 0.4 and 0.8 M sulphuric acid at 20 and 30 ◦ C in a three-electrode cell, with a saturated calomel electrode (SCE) as reference. Prior to anodizing, the specimens were degreased in acetone and in ethanol, rinsed in water and dried with cold air. For anodizing with simultaneous acquisition of the electrochemical impedance spectra, a Solartron Modulab potentiostat was employed. The acquisition of the impedance spectra at each potential was initiated after a steady anodizing current was achieved. In order to promote the rapid establishment of a relatively steady condition for the impedance measurement, a voltage sweep (1 V/min) was employed to achieve the potential of interest. Subsequently, a preliminary impedance spectrum, requiring about 4 min, was acquired at this potential (over a reduced frequency range), followed by a 2 min of potentiostatic anodizing. After the short potentiostatic anodizing, the final impedance spectrum was acquired. The stationarity of the system was confirmed by verifying that the current–time curves acquired during the final impedance measurement did not display significant DC drift and by verifying that the spectra acquired during the preliminary impedance measurement overlapped with the spectra acquired during the final impedance measurement over the frequency range available. The amplitude of the sinusoidal signal superimposed on the DC anodizing potential (in the range of 4–20 V) was 50 mV, with the frequency varied between 100 kHz and 0.001 Hz. The acquisition time for an impedance spectrum was approximately 3 h. Fitting of the impedance data was performed by using the Modulab software. 3. Results and discussion 3.1. Semiconductive nature of the growing oxide In order to fully interpret the data obtained by electrochemical impedance spectroscopy, the intrinsic behavior of the metal–oxide–solution system during oxide growth is considered initially. Fig. 1 displays a schematic view of the metal–oxide–electrolyte system during oxide growth. Oxidation of aluminium to aluminium ions is thermodynamically favoured; therefore, an accumulation of aluminium ions is expected within the oxide close to the metal interface, with a corresponding accumulation of electrons expected within the metal close to the oxide interface. When DC current is passed, oxide growth proceeds with additional aluminium atoms oxidized to aluminium ions, which are continuously injected into the oxide at the metal/oxide interface. Simultaneously, oxygen anions, generated from water adsorbed at the oxide/electrolyte interface, are injected into the oxide at the oxide/solution interface. When a given number of aluminium atoms (i.e. 10 in Fig. 1), are oxidized, a corresponding number of aluminium ions are generated at the metal/oxide interface. Of the 10 ions injected into the oxide at the aluminium/oxide interface, only 6 combine with a corresponding number of oxygen ions within the oxide (9), to form 3 units of Al2 O3 . 4 aluminium ions injected at the metal/oxide interface migrate further under the electric field with eventual ejection into the electrolyte. From Fig. 1, an excess of Al3+ positive charge carriers (similar to p-type semiconductor), resides close to the metal and an excess of O2− negative charge

Fig. 1. The metal–oxide system. Schematic view of the metal–oxide–electrolyte system during oxide growth (without accounting for the externally applied potential) with qualitative plots of ion concentration, resulting space charge, electric field and potential distribution.

carriers (similar to n-type semiconductor), is present close to the film/electrolyte interface. Between the p-type oxide and the n-type oxide, charge recombination takes place, generating Al2 O3 units. The process of charge recombination requires that the positive charge carried by the aluminium ions equals the negative charge carried by the oxygen ions and, therefore, the behavior of this region is similar to an intrinsic (i-type) semiconductor [15,16]. Consequently, the system can be described as a metal-p–i–n junction, with additional qualitative consideration of the spatial distribution of charge, field and potential, assisting selection of the appropriate equivalent circuit (Fig. 2) for fitting of impedance data. From Fig. 1, it is evident that, as a result of the charge distribution, the potential is a non-monotonic function of position, increasing at the metal/film interface and decreasing across the p–i–n oxide. Through aluminium accumulation at the metal/oxide interface, an interfacial capacitance, CMO , similar to the double layer capacitance in aqueous electrolytes, and an associated parallel resistance RMO , similar to a charge transfer resistance for aqueous electrolytes, are expected. During oxide growth, the current passes from the

7046

Fig. 2. Equivalent circuit. Equivalent metal/oxide/electrolyte system.

M. Curioni et al. / Electrochimica Acta 55 (2010) 7044–7049

circuit

representative

of

the

metal towards the oxide i.e., when crossing the metal/oxide interface, from a region of more negative potential towards a region of more positive potential. Therefore, since resistance and capacitance are defined as positive for current passing from a region of more positive potential towards a region of more negative potential, negative values of CMO and RMO are expected thereby determining the inductive behaviour in the Nyquist impedance plots. Similar to RMO , accounting for the potential drop (VMO ) due to the metal/p-oxide interface, RJ relates to the potential drop (VJ ) due to the p–i–n oxide, resulting from the distribution of charge carriers within the oxide. The resistor RO accounts for the dissipative nature of ionic transport across the i-type oxide, and a Warburg element considers diffusion of the charged species within the oxide. The contribution of the Warburg element to the overall resistive behavior (see Additional Table 1) is always below 5% and, for simplicity, it is not discussed further. The capacitance of the barrier layer oxide, proportional to its thickness, is represented by the capacitor CBL . Finally, CTH accounts for an apparent capacitance related to the dynamic reorganization of the porous oxide under the oscillating potential associated with impedance probing. Specifically, for a dynamic condition at sufficiently low frequencies, i.e. below 0.1 Hz, the variation in potential is followed by a variation in the barrier layer thickness due to the proportionality between applied potential and barrier layer thickness [24,30]. Over the frequency range where cyclic thickening (or thinning) of the barrier layer is evident, a contribution to the current that is proportional to the time derivative of the potential appears [30], which is revealed as an apparent large capacitance. 3.2. Effect of experimental parameters on impedance spectra 4 experimental parameters have been examined, namely applied DC potential and electrolyte composition, concentration and temperature. Generally, the appearance of the complex impedance plots was similar for all the experimental conditions (Fig. 3). Thus, spectra with 3 time constants, comprising a high frequency capacitive arch, a medium frequency inductive loop and a low frequency capacitive arch, where always recorded. The low frequency capacitive arch was one order of magnitude larger than the high frequency capacitive arch, and the diameter of the inductive loop was approximately one-half of the high frequency capacitive arch. An increase in DC current, achieved either by increasing potential, temperature or electrolyte concentration, consistently reduced the diameter of all the 3 arches. Unreported experiments performed on variously pre-anodized specimens indicated that, once a steady condition is achieved, the impedance measurement senses only the barrier layer properties, irrespective of the previous anodizing history. Examples of experimental and calculated impedance spectra for various conditions are provided in Fig. 4. 3.3. Modelling and interpretation The barrier layer thickness calculated from CBL (relative dielectric constant εr = 9, Fig. 5), increased linearly with applied potential, independent of the experimental conditions, with a proportional-

Fig. 3. Impedance data. Nyquist plots obtained at 4, 8, 12 and 16 V DC in 0.4 M sulphuric acid at 20 ◦ C. In the inset the high frequency region of the spectra is enlarged.

ity constant close to 1.05 nm/V in agreement with cited values [24]. For the specimens anodized in 0.3 M oxalic acid 30 ◦ C at 16 and 20 V, slightly lower values of barrier layer thickness were revealed, suggesting an increase of electric field during growth under high voltage and high temperature conditions. The potential drops due to the metal–oxide interface and the p–i–n junction within the oxide, calculated by multiplying the DC current by the values of RM and RJ obtained from the fitting of the impedance data, respectively decreased and increased linearly with the applied potential, but irrespective of other experimental parameters (Fig. 6). The value of the resistance, RO , representative of the process of ionic migration, was higher for oxides formed in oxalic acid than for sulphuric acid under similar anodizing conditions and decreased with increasing applied potential, temperature, or electrolyte concentration (Fig. 7). RO was generally one order of magnitude higher than RM and RJ (see Additional Table 1). The oxide resistivity, i.e. the ratio between RO and the barrier layer thickness, only displayed a dependence on the DC current, decreasing with increase of current with a log–log relationship (Fig. 8). 3.4. Flow modulated ionic migration During porous oxide growth, the migration of charged species across the barrier layer region of the porous oxide is responsible for the generation of new oxide material. However, in parallel with the ionic migration process, the viscous displacement of the newly formed oxide from the barrier layer towards the cell walls plays a key role. The overlap of these 2 processes determines the geometry and the growth rate of the porous oxide films, and the resulting electrical properties that are measured during growth. In order to understand further the growth process, the intrinsic electrical properties of the oxide, regulating only ionic migration, must be separated from the extrinsic electrical properties that arise from the dynamic equilibrium between ionic migration and viscous displacement. If this is achieved, it is then possible to explain the influence of electrolyte nature, determining incorporation of acid anion species within the oxide film, on ionic migration and viscous displacement. Our work unveils that the intrinsic electrical properties, regulating ionic migration, are completely independent of the electrolyte nature within the experimental window inspected. Specifically, the potential drops across the metal/oxide interface and the p–i–n oxide are affected only by the value of the applied potential, indicat-

M. Curioni et al. / Electrochimica Acta 55 (2010) 7044–7049

7047

Fig. 4. Fitting of impedance data. Experimental and calculated impedance spectra for anodizing in (a) 0.6 M oxalic acid at 20 ◦ C, and 20 V, (b) 0.4 M sulphuric acid at 30 ◦ C and 4 V, (c) 0.3 M oxalic acid at 30 ◦ C and 8 V, (d) 0.8 M sulphuric acid at 20 ◦ C and 8 V.

ing that ion incorporation does not modify the electrical properties and the distribution of charge. Further, by disclosing the relation between RO and the DC current, and revealing that RO is one order of magnitude greater than RJ and RMO , it is confirmed, from a purely electrical viewpoint, that the rate-determining process during oxide growth is ionic migration across the barrier layer oxide. However, the variation of RO with the applied potential, showing a dependence on all the experimental parameters, suggests that RO contains contributions from the overlapping processes of ionic migration and mechanical displacement. Therefore, RO cannot be considered strictly an intrinsic electrical property of the oxide, but is the result of the combined electrical and the mechanical

Fig. 5. Barrier layer thickness. Variation of the barrier layer thickness with applied potential for all the selected conditions.

behaviour. However, by normalizing RO to the barrier layer thickness, the oxide resistivity can be calculated. As shown, the value of oxide resistivity only relates to the DC current and, consequently, it can be considered an intrinsic property that is unaffected by the electrolyte nature or flow regime. The log–log dependence of the oxide resistivity on the DC current is identical to the dependence observed in p–i–n diodes [31]. In p–i–n diodes, generally used as current-controlled resistors, a layer of intrinsically semiconductive silicon is placed between the p-type and n-type layers. Due to the increased resistivity of the i-layer with respect to the p and n layers, the potential drop is largely localized across the i-layer, and charge

Fig. 6. Potential drops within the oxide. Potential drops due to the metal/oxide junction (RM × IDC ) and to the p–i–n junction (RJ × IDC ) as a function of the applied potential.

7048

M. Curioni et al. / Electrochimica Acta 55 (2010) 7044–7049

Fig. 7. Oxide resistance. Resistance to ionic migration, RO , as a function of the applied potential.

transport takes place under the resulting electric field. In the case of growing aluminium oxide, the thickness of the i-type region is variable, since the barrier layer oxide thickness is proportional to the applied potential. Depending on the experimental conditions and, therefore, on the combined ionic transport-mechanical flow regime, the growth of the oxide occurs at a fixed value of current for a given DC potential. From the log–log dependence of the resistivity on the current, the system is stable from the electrical viewpoint; the barrier layer thickness increases rapidly to a maximum value, Thmax , corresponding to a negligible growth rate in the absence of mechanical flow. When Thmax is attained, the resistivity and, therefore, the resistance are very high and ionic migration is negligible. Simultaneously, the porous film morphology, if any, is determined, since the pore diameter and curvature directly correlate with the barrier layer thickness. If for some reason the barrier layer becomes locally thinner, the electric field increases and, consequently, the current also increases. This produces a substantial drop in oxide resistivity, which promotes a further passage of current. Clearly, the passage

of current generates new oxide, which contributes to an increase of barrier layer thickness. Therefore, since the intrinsic electrical properties of the oxides are not affected by the electrolyte composition from the electrical viewpoint, it can be concluded that the barrier layer oxide can rapidly approach Thmax , determining the porous oxide morphology, irrespectively of the electrolyte. When electrolyte ions are incorporated into the oxide, viscous displacement of the oxide from the barrier layer region become possible [25–28]. As shown in our work, this process does not affect the intrinsic electrical properties of the oxide and, for a given potential, an oxide with a barrier layer thickness equal to Thmax , is always obtained. As a direct consequence, the compressive stresses acting on the barrier layer are independent of the electrolyte, since they are related only to the applied potential and to the barrier layer geometry. Under this condition, the oxide viscosity becomes the rate-determining property and the mechanical displacement of oxide material from the barrier layer towards the cell walls can be considered to be an attempt to reduce the barrier layer thickness. However, as detailed previously, the electrical system can promptly respond by lowering the resistivity and increasing the growth rate, to achieve Thmax . The sulphate-containing oxides have reduced viscosity compared with the oxalate-containing oxides and flow faster under the electrostriction force, as evident from the reduced values of RO . Overall, the electrical properties of the oxide are independent of the electrolyte and determine the barrier layer thickness and the resulting pore geometry, while the oxide mechanical properties modulate the ionic migration across the barrier layer, determining the anodizing current and, ultimately, the growth rate. 4. Conclusions The mechanism of porous oxide growth has been investigated in situ by high voltage electrochemical impedance spectroscopy to disclose the relation between experimental parameters, electrical properties of the growing oxide and flow regime. Manipulation of the electrochemical impedance data enables separation of the intrinsic electrical properties of the growing oxide, due only to the oxide nature and independent of electrolyte composition, from the extrinsic properties, resulting from the overlapping of the processes of ionic migration and mechanical displacement. It is disclosed that the anodic oxide can be represented as a system described by a metal-p–i–n junction, with characteristic intrinsic potential drops, revealed readily by the impedance measurement. From the electrical viewpoint, it is revealed that the process of ionic transport across the oxide is the rate-determining step during the growth. Further, the oxide intrinsic electrical properties are independent of the electrolyte, providing the reason for the observed the potential dependent porous film morphology. Importantly, the electrolyte nature modifies the mechanical properties of the oxide by regulating the oxide displacement from the barrier layer towards the cell walls. The rate of displacement modulates ionic transport by determining the anodizing current and the film growth rate. Acknowledgement The authors acknowledge the financial support provided by the Engineering and Physical Sciences Research Council (U.K.) Portfolio Award, Light Alloys Towards Environmentally Sustainable Transportation. Appendix A. Supplementary data

Fig. 8. Oxide resistivity. Oxide resistivity, o = RO /barrier layer thickness, as a function of the DC current.

Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.electacta.2010.06.088.

M. Curioni et al. / Electrochimica Acta 55 (2010) 7044–7049

References [1] S. Ono, M. Saito, H. Asoh, Electrochimica Acta 51 (2005) 827. [2] W. Lee, K. Schwirn, M. Steinhart, E. Pippel, R. Scholz, U. Gösele, Nature Nanotechnology 3 (2008) 234. [3] H. Masuda, F. Hasegwa, S. Ono, Journal of the Electrochemical Society 144 (1997) L127. [4] S. Ono, M. Saito, M. Ishiguro, H. Asoh, Journal of the Electrochemical Society 151 (2004) B473. [5] W. Lee, R. Ji, U. Gosele, K. Nielsch, Nature Materials 5 (2006) 741. [6] P.G. Sheasby, R. Pinner, The Surface Treatment and Finishing of Aluminium and its Alloys, ASM International, Materials Park, OH, 2001. [7] X. Zhou, G.E. Thompson, G. Potts, Transactions of the Institute of Metal Finishing (UK) 78 (2000) 210. [8] M. Curioni, P. Skeldon, G.E. Thompson, J. Ferguson, Proceedings of the Advanced Materials Research 38 (2008) 48. [9] M. Curioni, P. Skeldon, E. Koroleva, G.E. Thompson, J. Ferguson, Journal of the Electrochemical Society 156 (2009) C147. [10] H.W. Wang, P. Skeldon, G.E. Thompson, Tribology Transactions 42 (1999) 202. [11] M. Ward, D.R. Gabe, R.J. Latham, R.H. Dahm, Transactions of the Institute of Metal Finishing 81 (2003) 122. [12] M. Takaya, K. Hashimoto, Y. Toda, M. Maejima, Surface and Coatings Technology 169–170 (2003) 160. [13] M. Maejima, K. Saruwatari, M. Takaya, Surface and Coatings Technology 132 (2000) 105. [14] M.M. Lohrengel, Materials Science and Engineering R: Reports 11 (1993) 243. [15] J.W. Diggle, T.C. Downie, C.W. Goulding, Chemical Reviews 69 (1969) 365. [16] Y. Sasaki, Journal of Physics and Chemistry of Solids 13 (1960) 177.

7049

[17] H.J. De Wit, C. Wijenberg, C. Crevecoeur, Journal of the Electrochemical Society 126 (1979) 779. [18] J. Shulman, Y.Y. Xue, S. Tsui, F. Chen, C.W. Chu, Physical Review B—Condensed Matter and Materials Physics 80 (2009). [19] J. Shulman, S. Tsui, F. Chen, Y.Y. Xue, C.W. Chu, Applied Physics Letters 90 (2007). [20] I. Epelboin, M. Keddam, Journal of the Electrochemical Society 117 (1970) 1052. [21] S.L. Wu, M.E. Orazem, B. Tribollet, V. Vivier, Journal of the Electrochemical Society 156 (2009) C214. [22] I. Mora-Sero, J. Bisquert, F. Fabregat-Santiago, G. Garcia-Belmonte, G. Zoppi, K. Durose, Y. Proskuryakov, I. Oja, A. Belaidi, T. Dittrich, R. Tena-Zaera, A. Katty, C. Levy-Clement, V. Barrioz, S.J.C. Irvine, Nano Letters 6 (2006) 640. [23] J. Bisquert, G. Garcia-Belmonte, A. Pitarch, H.J. Bolink, Chemical Physics Letters 422 (2006) 184. [24] J.P. O’Sullivan, G.C. Wood, Proceedings of the Royal Society of London Series A, Mathematical and Physical Sciences 317 (1970) 511. [25] S.J. Garcia-Vergara, P. Skeldon, G.E. Thompson, H. Habazaki, Electrochimica Acta 52 (2006) 681. [26] P. Skeldon, G.E. Thompson, S.J. Garcia-Vergara, L. Iglesias-Rubianes, C.E. BlancoPinzon, Electrochemical and Solid-State Letters 9 (2006) B47. [27] S.J. Garcia-Vergara, P. Skeldon, G.E. Thompson, H. Habakaki, Applied Surface Science 254 (2007) 1534. [28] S.J. Garcia-Vergara, L. Iglesias-Rubianes, C.E. Blanco-Pinzon, P. Skeldon, G.E. Thompson, P. Campestrini, Proceedings of the Royal Society A: Mathematical Physical and Engineering Sciences 462 (2006) 2345. [29] J.E. Houser, K.R. Hebert, Nature Materials 8 (2009) 415. [30] M. Curioni, P. Skeldon, G.E. Thompson, Journal of the Electrochemical Society 156 (2009) C407. [31] BAP50-03, General purpose PIN diode data sheet, http://www.nxp.com/documents/data sheet/BAP50-03.pdf.