Impedance of an Aluminum Electrode with a Nanoporous Oxide

Abstract

The methods of electrochemical impedance spectroscopy and electron microscopy with X-ray electron probe analysis are used to study two types of oxide films on aluminum: a smooth compact one and a porous one (nanooxide) formed by two-stage oxidation. An equivalent circuit is chosen that agrees well with the obtained impedance spectra and allows estimating the thickness and conductivity of barrier oxide layers and also the area occupied by nanopores for various film types.

INTRODUCTION

The characteristic features of aluminum include its relatively high corrosion stability related to formation of a compact passivating film of aluminum oxide on the surface that prevents the metal from directly interacting with the medium. The protective oxide film on the aluminum surface in the passive state is 5 to 100 nm thick [1]. To obtain oxide film with a regular cellular structure, the process of anodic aluminum oxidation (anodizing) is carried out. Depending on the process conditions, thick-layer and thin-layer anodizing are distinguished. Thick-layer anodizing [1–3] with differing pore filling is widely used in engineering and, depending on the conditions of synthesis, oxide can possess different properties (enhanced corrosion stability, wear resistance, hardness, etc.). In the case of thin-layer anodizing [4–8], nanoporous oxide with a regular cellular structure and pore diameter of 20–100 nm is formed on the aluminum surface and can serve as a support for deposition into nanopores of components imparting unique protective [9] properties to the material. Besides, transition metals are deposited into oxide nanopores to provide catalytic properties [10, 11]. Morphological features and properties of the obtained composites are largely determined by the properties of the cellular regular support structure [12]. In this connection, much attention is paid in the studies to investigation of the structure and properties of the nanoporous support itself. The electrochemical impedance technique is one efficient research method [13–15].

EXPERIMENTAL

Oxide films were formed on 150-μm-thick aluminum foil with the purity of 99.99 wt %. The foil was preannealed at 450°C for 1 h to remove internal stresses. Nanoporous oxide films were obtained using the method of two-stage anodizing [4] in the galvanostatic mode at the current density of 10 mA/cm² in 10% H₂SO₄ solution. Before the first stage of anodizing, samples were subjected to chemical etching in the solution of HF + HNO₃ + HCl + H₂O at the ratio of 1 : 10 : 20 : 69 for 10 min at 20°C. The galvanostatic treatment time was 2 h. Before the second stage, the samples were treated for 10 min in a solution containing 20 g/L of CrO₃, 35 mL/L of H₃PO₄ (85%) at 60–90°C. The galvanostatic treatment time was 1 h. Such treatment led to the formation of a film of nanoporous oxide with a regular cellular structure [5].

Studies of impedance on samples with compact smooth nanoporous oxide formed when an aluminum sample was placed into the solution was carried out using a complex including an IPC-ProMF potentiostat and an FRA-2 impedance meter in the frequency range of 0.05 to 3000 Hz. To eliminate the effect of the electrolyte on the processes occurring on the sample and changing the properties of oxide films (appearance of faradaic impedance as a result of charge transfer owing to electrochemical oxidation of aluminum on acidic and alkaline media, interaction of ions of electrolyte with the oxide film in the medium of chloride ions), impedance was measured in a sufficiently “soft” 0.5 N Na₂SO₄ solution. The studies were performed under a steady-state potential and under anodic polarization in the range of aluminum passivity with the ac amplitude of 20 mV. Measurements were performed according to the three-electrode scheme. Samples of oxidized foil were clamped in a frame providing a working surface in the form of a circle with the area of 1 cm² and reliable contact between the nonoxidized metal surface and the working electrode on the reverse side. The auxiliary electrode was platinum gauze with the area of ~10 cm². The potentials were measured versus a saturated silver–silver chloride electrode connected via a salt bridge, which eliminated the possibility of chlorides penetrating the working space. Equivalent circuits were analyzed and their elements were calculated using the DCS (Dummy Cell Solver) software developed by one of us, V.E. Kasatkin (Frumkin Institute of Physical Chemistry and Electrochemistry, Russian Academy of Sciences). This program allowed calculating the theoretical frequency dependence of impedance of the chosen circuit and, by comparing it with the experimentally obtained dependence, optimizing the values of the circuit elements by minimizing the sum of relative errors F between the experimental Ze and simulated Zm system impedance values for each of frequencies f_i in the studied range:

$$F = \sum\limits_{i = {{f}_{{\min }}}}^{{{f}_{{\max }}}} {(Z{{e}_{i}}} - Z{{m}_{i}}){\text{/}}Z{{e}_{i}},$$

((1))

where Ze_i is the experimental value of the system impedance, Zm_i is the model value of the system impedance, and f_min and f_max are, respectively, the minimum and maximum frequencies in the studied region.

The surface morphology was controlled using the electronic microscopy technique. A raster electron microscopic complex (SEM) with a Quanta 650 FEG field cathode (FEI, Netherlands) was used in the secondary electron detection mode. The device was equipped with an addon for X-ray spectral microanalysis of the elemental composition of the analyzed surface.

EXPERIMENTAL RESULTS AND DISCUSSION

SEM Microscopy

Figure 1 shows electron microscopy images of the aluminum surface with (a) smooth oxide and (b) nanoporous oxide. One can see that, as opposed to the relatively smooth surface for case (a), nanoporous oxide forms a periodical honeycomb pattern with an elementary cell represented by hexagons framing a round pore (Fig. 1b). A closely packed structure of surface filling by hexagonal cells is observed (the pores occupy approximately 80% of the whole surface). The honeycomb surface structure of nanoporous oxide is characterized by the following parameters: the average pore diameter is 25 nm, the step (honeycomb structure period) on the surface is 34 nm, and the average diameter of the circumscribed hexagon circle is 30 nm.

Fig. 1.

Electron microscopy images of oxidized aluminum surface with (a) smooth oxide, (b) nanoporous oxide, (c) after electron beam etching, and (d) after electron beam etching.

The structure of the oxide honeycomb cells can be made more pronounced by etching by a high-intensity electron beam owing to electron-induced crystallization of nanoporous oxide elements. For this, the chosen anodized surface region was exposed to a high–intensity electron beam of the electron microscope (Fig. 1, the rectangular beam in the center). Figure 1c shows a fragment of an image of the electron-beam-treated surface. A 4- to 6-nm-thick transition region is found at the pore–oxide interface. It probably corresponds to the bilayer oxide structure at the pore walls. Obviously, the observed outer pore wall layer may be associated with intercalation of sulfate ions into the material of pore wall oxide.

As follows from analysis of X-ray spectral data (Fig. 2), the O, Al, and S elements are present on the surface. The atomic ratio between oxygen and aluminum is almost 1.7, which approaches the theoretical value (1.5 for Al₂O₃). The presence of S can be due to intercalation of electrolyte traces into the anodic film, its pores, and pore walls. It is known that pore walls formed in sulfuric acid are contaminated by acidic ${\text{SO}}_{4}^{{ - 2}}$ groups.

Fig. 2.

Results of X-ray spectral analysis of (b) the sample surface with (a) nanoporous oxide.

Figure 1d shows a 3D image of several neighboring honeycomb cells. One can see that electron-beam crystallization indeed leads to enhancement of the surface topography: the hexagonal network of intercell boundaries indeed becomes prominent between the neighboring honeycomb cells on the surface, where the oxide thickness is maximal. One can suggest that the selective “etching” of the remaining nanoporous oxide surface may be related to predominant crystallization and the resulting compaction of pore walls, where the oxide contains the maximum amount of defects owing to sulfate ions embedded in the course of anodizing.

Electrochemical Impedance Spectroscopy

Figure 3 shows phase angle-frequency and amplitude-frequency (Bode diagrams) characteristics of aluminum samples with smooth oxide in the studied solution at two values of potentials (corrosion potential appearing when the sample is placed into the solution and E = +200 mV). In both cases, the values of logarithm of the absolute impedance value decrease linearly at an increase in the frequency. Deviation of the linear dependence is observed in the low-frequency range in tests with polarization. At intermediate and high frequency values, the plots of sample dependences in the absence and in the presence of polarization practically coincide. Phase-frequency characteristics of samples in the intermediate and high frequency ranges also coincide. Phase-angle-shift independence of the frequency is observed at intermediate frequency values. The phase shift in this frequency range is close to 90°. The high-frequency cut in the phase-frequency characteristic can be related to manifestation of the inductive circuit component similarly to capacitors. In the low–frequency range, the decrease in the phase angle can be explained by the fact that the capacitive impedance component is inversely proportional to frequency and increases so much that current transport through the oxide in parallel to the capacitance (capacitance bypass by a parasite parallel resistance) starts to play a significant role. All the above data allow considering the system of aluminum with smooth oxide as a nonideal capacitor with leakage currents through oxide and presenting it as a circuit of a capacitor and a resistance connected in parallel. The capacitor is in this case formed by charges on the metal–oxide and oxide–solution interfaces and the dielectric spacer is the oxide film on the metal surface. In the case of anodic electrode polarization (Fig. 3, curve 2), the cutoff in the phase-frequency characteristic at a decrease in frequency occurs earlier (at higher frequency values). In the absence of any noticeable charge transfer on a passive electrode due to hindered electrochemical processes, practically the whole voltage drop (difference between the open circuit potential and polarization potential) occurs in the oxide film. This can create electric field in the oxide film accelerating migration phenomena, which can in fact cause a decrease in its resistance¹. As a result, the bypass effect of resistance in the system with polarization occurs at higher frequency values as compared to the open circuit.

Fig. 3.

Bode diagrams for aluminum samples with smooth oxide: (1) the dependence of phase angle θ on the frequency for aluminum samples at the corrosion potential, (2) dependence of phase angle θ on the frequency for aluminum samples at the polarization of (E = +200 mV), (3) dependence of absolute impedance |Z| on the frequency for aluminum samples at the corrosion potential, and (4) dependence of absolute impedance |Z| on the frequency for aluminum samples at polarization (E = +200 mV).

The Nyquist diagrams of the studied electrodes (Fig. 4) have the form of a weakly pronounced semicircle. If one considers only them, one can suggest different equivalent circuits, including a circuit with a Warburg element. However, taking the above discussion into account, calculations with respect to the Nyquist diagram should be carried out using an equivalent circuit including serial resistance R₀ and a circuit of constant phase element CPE and resistance R connected in parallel. In this circuit, R₀ corresponds to the resistance of the electrolyte and capacitor plates, resistance R corresponds to current leakage through oxide, and CPE parameters characterize the capacitive properties of the system and capacitor nonideality. The calculated values of the parameters are presented in Table 1.

Fig. 4.

Nyquist diagrams: (1) aluminum sample with smooth oxide at the corrosion potential, (2) aluminum sample with smooth oxide at the polarization of E = +200 mV, (3) aluminum sample with nanooxide at the corrosion potential, and (4) aluminum sample with nanooxide at the polarization of E = +200 mV. Dots represent experimental data; curves correspond to the data calculated on the basis of the equivalent circuit.

Table 1.   Tables of values of parameters calculated according to the equivalent circuit

As follows from the table, electrolyte resistance R₀ is small for samples with smooth oxide and is independent of the electrode polarization within the limits of experimental error. Exponent n remains practically unchanged. The equation for the CPE element of the impedance takes the form of ${{Z}_{{{\text{CPE}}}}} = {1 \mathord{\left/ {\vphantom {1 Y}} \right. \kern-0em} Y}{{(j2\pi f)}^{n}},$ where Y is the frequency-independent preexponential factor, j is an imaginary unit, f is the signal frequency, n is the exponent determining the character of the frequency dependence (–1 ≤ n ≤ 1). The value of n in the table is about 0.9, which is close to 1, but this does not allow considering the CPE element quite as a capacitor with lumped parameters. This is possibly related to a certain spatial diffusion of charge due to the presence of the diffusion region of the double electric region in the electrolyte. Amplitude component Y of the CPE element has formally the dimension of Ω C^–n, but in the case of n close to unity it can be interpreted in the capacitance C units (farads). Parameter Y drops linearly at an increase in polarization. In all probability, this is related to the linear growth of the oxide thickness. Resistance R decreases practically by half at the start of polarization and then remains in the range of 0.5–0.7 MΩ within the limits of experimental error. One can assume that a decrease in resistance related to appearance of electric field in the oxide is compensated in the course of the further polarization by the corresponding oxide thickening.

One can estimate the oxide layer thickness using the data presented in the table and the expression for capacitance C:

$$C = \frac{{{{\varepsilon }_{0}}\varepsilon S}}{d},$$

((2))

where ε₀ is the electric constant, ε is the relative dielectric penetrability of the medium, S is the sample surface area, and d is the layer thickness.

For the reference value of relative dielectric constant of Al₂O₃, which is about 10, and capacitance in the range of 3.3 to 2.8 μF at the sample area of 1 cm², one can calculate that the oxide thickness changes from 2.7 to 3.1 nm, as dependent on polarization. Judging by the calculated oxide thickness values, one can assume that it is the so-called “barrier layer.”

Earlier, we have thoroughly analyzed the dependences of the barrier layer thickness for nanoporous and smooth aluminum oxides on the bath voltage in various electrolytes. It was found for sulfuric acid that the barrier layer thickness increases by 1.0 nm per 1 V of voltage. Under our conditions of formation of nanoporous oxides, the final bath voltage is about 15 V. Therefore, the overall barrier layer thickness is about 15 nm. The discrepancy is most probably due to the surface roughness. Comparison of the calculated and actual thickness can serve as a measure of the roughness factor that in our case yields the value of 5. Therefore, the electrochemical impedance technique can be used for calculation of the surface roughness factor.

Even in the absence of data on the plate charge and, accordingly, electric field intensity in oxide at the corrosion potentials, one can nevertheless estimate the gain in electric field intensity and charge under exposure to polarization. Despite the growing oxide thickness at an increase in polarization, the gain in field intensity and, accordingly, plate charge increases linearly at an increase in potential.

The character of frequency dependences of phase shift angle and absolute impedance on the electrode with aluminum nanooxide (Fig. 5) is similar to that on aluminum with smooth oxide. The plot of the dependence of absolute impedance on the logarithm of frequency is also linear and its slope corresponds to the slope of straight lines on nanooxide-free aluminum samples. However, the absolute impedance value is higher. The range in which the phase shift angle is independent of the frequency is closer to 90° on aluminum with nanooxide than on aluminum with smooth oxide. For the sake of illustration, Fig. 6 shows the values of angle and absolute impedance on polarization for the both electrodes. It should be noted that polarization does not affect these parameters, but the difference between electrodes with a surface nanostructure and without it are rather noticeable by the absolute value of parameters. In the case of the electrode with nanooxide (Fig. 5), the low-frequency cutoff of the phase-frequency characteristic occurs at lower frequencies. This is characteristic both for the open circuit and for a polarized electrode. Therefore, this system can also be described as a nonideal capacitor with the correction that nonideality of the capacitor is less pronounced.

Fig. 5.

Bode diagrams for aluminum samples with nanooxide: (1) the dependence of phase angle θ on the frequency for aluminum samples at the corrosion potential. (2) dependence of phase angle θ on the frequency for aluminum samples at the polarization of E = +200 mV, (3) dependence of absolute impedance |Z| on the frequency for aluminum samples at the corrosion potential, and (4) dependence of absolute impedance |Z| on the frequency for aluminum samples at polarization E = +200 mV.

Fig. 6.

Dependence of phase angle and absolute impedance value for aluminum samples on polarization at the frequency of 100 Hz: (1) the dependence of absolute impedance |Z| on polarization for aluminum samples with nanooxide, (2) dependence of absolute impedance |Z| on polarization for aluminum samples with smooth oxide, (3) dependence of phase angle θ on polarization for aluminum samples with nanooxide, and (4) dependence of phase angle θ on polarization for aluminum samples with smooth oxide.

The curves of samples with nanooxide on the surface in the Nyquist diagram (Fig. 4) also take the shape of an open semicircle. Using in calculations the same equivalent circuit as for aluminum with smooth oxide, one can calculate the parameters of circuit elements (Table 1). The values of resistance R and the amplitude component of the CPE element for samples with nanooxide decrease with polarization, same as in the case of smooth oxide. Here, resistance R is higher than in the first case, while capacitance is lower. On the assumption that the barrier layer located at the pore bottom has the same nature and thickness, as the smooth oxide barrier layer, the both effects can be explained by the size factor. The thick nanooxide region connected in parallel to the nanopore barrier layer in the equivalent circuit, owing to large resistance and small capacitance, produced no noticeable effect on the measured impedance. Therefore, only the pores themselves must be considered. The area occupied by pores corresponds only to a fraction of the overall surface. That is, while the whole surface takes part in capacitor formation in the case of smooth oxide, only a certain fraction of the surface participates in the case of nanoporous oxide. Hence, the overall resistance of samples with nanopores becomes higher and capacitance becomes lower in proportion to the surface fraction occupied by pores. At this assumption, one can calculate the nanooxide surface fraction occupied by pores. It is about 0.25 of the overall oxide surface area. This corresponds to the fraction occupied by pores in a regular hexagonal structure, in which the pore hexagon is surrounded by oxide hexagons [16].

CONCLUSIONS

Thus, studies in a sufficiently “soft” Na₂SO₄ solution favoring stable aluminum passivation show that the metal–oxide system can be interpreted using a simple equivalent circuit that allows reliably determining the system parameters. Impedance studies show that nanooxide behaves similarly to smooth oxide, but with a surface area smaller than the visible one. The data of impedance measurement allow estimating both the barrier layer thickness and the relative area occupied by nanopores.