Volta Potential Evolution of Intermetallics in Aluminum Alloy Microstructure Under Thin Aqueous Adlayers: A combined DFT and Experimental Study

Abstract

In this work, first-principle density functional theory (DFT) was used to calculate the work function and Volta potential differences between aluminum alloy matrix and two intermetallic phases (Mg₂Si and Al₂Cu) with varying surface terminations as a function of adhering monolayers (ML) of water. The calculated data were compared with experimental local Volta potential data obtained by the scanning Kelvin probe force microscopy (SKPFM) on a commercial aluminum alloy AA6063-T5 in atmospheric environments with varying relative humidity (RH). The calculations suggest that the surface termination has a major effect on the magnitude and polarity of the Volta potential of both intermetallic phases (IMP’s). The Volta potential difference between the IMP’s and the aluminum matrix decreases when the surface is gradually covered by water molecules, and may further change as a function of adhering ML’s of water. This can lead to nobility inversions of the IMP’s relative to the aluminum matrix. The measured Volta potential difference between both IMP’s and their neighboring matrix is dependent on RH. Natural oxidation in ambient indoor air for 2 months led to a nobility inversion of the IMP’s with respect to the aluminum matrix, with the intermetallics showing anodic nature already in dry condition. The anodic nature of Al₂Cu remained with the introduction of RH, whereas Mg₂Si became cathodic at high RH, presumably due to de-alloying of Mg and oxide dissolution. The DFT calculations predicted an anodic character of both IMP’s in reference to the oxidized aluminum matrix, being in good agreement with the SKPFM data. The DFT and SKPFM data were discussed in light of understanding localized corrosion of aluminum alloys under conditions akin to atmospheric exposure.

1 Introduction

The interaction of water with any solid surface under ambient pressure conditions is a topic of profound scientific and technical importance. The literature includes experimental and theoretical studies in diverse research areas such as heterogeneous catalysis (Refs. [1, 2] are examples of recent publications), meteorology [3, 4], geology [5, 6], physical chemistry [7, 8], solar energy conversion [9,10,11], pharmacy [12], atmospheric chemistry [13], electrochemistry [14, 15], tribology [16, 17], and corrosion science [18, 19]. Advancements in experimental and computational techniques have resulted in considerable progress in our fundamental view of water–solid interactions, and the research field has been summarized in a few most comprehensive reviews [20,21,22].

In atmospheric corrosion, which is the most common and perhaps most complex form of corrosion, the presence of an ultra-thin ubiquitous aqueous adlayer is of fundamental importance [23, 24]. Atmospheric corrosion involves three phases (atmosphere, liquid, and solid) and two interfaces (atmosphere|liquid, liquid|solid), whereby the aqueous film (often referred to as aqueous adlayer) interacts with the atmosphere and the solid (metal) through a variety of simultaneously acting physicochemical processes [23]. Detailed experimental analysis has shown that most solid surfaces are moist when in contact with ambient air [23]. The amount of adhering water (coverage) depends on different surface-related properties, such as porosity, microstructure, surface hydroxylation, degree of oxidation, and the presence of hygroscopic species [23]. On a clean metal surface, the amount of water present at room temperature increases from a few equivalent ML’s of water at 40% RH to approximately ten ML’s at 90% RH [25].

Through instrumental developments and better access of techniques, such as the quartz crystal microbalance (QCM) [26, 27], surface vibrational spectroscopy [28,29,30], and local probing techniques [31, 32], our molecular understanding of the role of the aqueous adlayer in atmospheric corrosion has greatly improved. A technique of special interest in this context is the SKPFM which can provide information about the Volta potential of a locally-probed metallic area in nanometer resolution while the surface is typically in contact with the ambient air [33]. The Volta potential can provide meaningful information about the relative electrochemical nobility of microstructural phases during initial aqueous corrosion of metals or alloys [34, 35], yet without a clear understanding of the theory behind.

An important step was recently taken when first principle calculations were performed on IMP’s in an aluminum alloy, whereby the computational results of Volta potential differences between the IMP’s and the surrounding aluminum matrix showed good agreement with experimental data obtained by SKPFM [36]. In the current study these efforts have advanced one step further by comparing theoretically-deduced Volta potential differences calculated by DFT with local Volta potential differences measured on a commercial aluminum alloy (AA6063-T5) by SKPFM. DFT has been used to obtain Volta potential differences between two IMP’s (Mg₂Si and Al₂Cu with different surface atomic terminals) and the aluminum matrix in the presence of up to two ML’s of water. SKPFM was performed in the presence of well-controlled humidified air on a commercial aluminum alloy (AA6063-T5) to measure the difference in Volta potentials between the same IMP’s and the surrounding aluminum matrix at different RH. Finally, the computational and experimental data have been compared and discussed in light of understanding localized corrosion of complex aluminum alloys under conditions akin to atmospheric exposure.

2 DFT Calculations

The work function of bare aluminum (Al), one stable phase of Al oxide, α-Al₂O₃, and two IMP’s, namely Mg₂Si and Al₂Cu, as a function of adhering ML of H₂O were calculated. Then, from these data their corresponding Volta potentials, the Volta potential differences between the IMP’s and the Al matrix, and the Volta potential changes upon surface coverage of adhering H₂O were calculated. This allowed direct comparison of DFT calculations with Volta potentials measured by SKPFM.

Figure 1 displays the atomic structures of the calculated slabs for Al(111), α-Al₂O₃(0001), Mg₂Si(111) with Mg- and Si-terminals, Mg₂Si(110) with MgSi-terminal, and Al₂Cu(110) with Al2- (two outermost layers of Al) and Cu-terminals, which are constructed in a 2 × 2 surface area showing several atomic layers. In this way, H₂O molecules could be introduced onto the surface and the interactions between H₂O molecules and the substrate as well as among the H₂O molecules were taken into account in all calculations. The calculation slabs were chosen thick enough to approximate their bulk properties. Each surface of the slabs was separated by a large enough vacuum gap, keeping the interaction of the surfaces with other systems at zero. Then, to study the effect of humidity, the work functions of the systems with and without water adsorption were calculated with several possible configurations of H₂O molecules at water coverages (θ) at 0.25, 0.5, 1, and 2 ML of H₂O. One ML of H₂O refers to the ratio of the amount of water molecules per surface atom [37].

Fig. 1

Atomic structures (slab) of Al(111), α-Al₂O₃(0001), Mg₂Si and Al₂Cu with different surface terminals used in DFT calculations. Color of the atoms: Al in purple, O in red, Mg in green, Si in yellow, and Cu in orange

The DFT calculations were performed using the Dmol3 code based on the GGA-PW91 exchange–correlation energy functional [38]. DFT semi-core pseudopotentials (DSPP) were employed as the core electron treatment with double numeric basis sets and polarization functions (DNP). Brillouin-zone integrations were performed using a 6 × 6 × 1 k-point mesh size for all surfaces, except for the Al matrix for which a 3 × 3 × 1 k-point mesh size was used. A Meshfessel–Paxton smearing of 0.005 Hatree (Ha, 1 Ha = 27.2114 eV) was used, and dipole corrections were introduced to all surfaces. All surface configurations including those with adhering water molecules were optimized until the energy, residual force, and atomic displacement became less than 10⁻⁵ Ha, 0.002 Ha/Å (Ångstrom), and 0.005 Å, respectively.

Figure 2 shows an optimized atomic configuration of 1 ML of H₂O adhering on the surface of Mg₂Si(111) with Mg-terminal as an example. There are four surface Mg atoms at the terminal with four adhering H₂O molecules corresponding to 1 ML of H₂O. The most stable configuration of all structures, having the lowest total energy, was chosen to obtain the work function of the water-covered surfaces.

Fig. 2

Optimized atomic configuration of 1 ML of H₂O adhering on the surface of Mg₂Si(111)-Mg terminal: a side view, b top view. O atoms are shown in red, H in white, Mg in green, and Si in yellow colors

The work function was calculated according to following equation [39]:

$${\phi _M}={E_{vacuum}} - {E_{Fermi}}$$

(1)

where ${\phi _M}$ is the work function of the metal, ${E_{vacuum}}$ is the vacuum energy, and ${E_{Fermi}}$ is the Fermi level. The work functions of all bare surfaces have been calculated and reported in previous publication [36], which are in reasonable agreement with other calculations [40, 41]. The reader is referred to these works for detailed explanation of the DFT procedure.

The slab thickness for Mg₂Si(111)-Mg and Mg₂Si(110)-MgSi was kept to a minimum to limit the required calculating capacity for the water-covered systems. The work function was used to calculate the Volta potential according to following equation:

$${\psi _M}=\frac{{{\phi _M}}}{e}$$

(2)

where ${\psi _M}$ is the Volta potential of the metal and $e$ the unit charge of an electron. The understanding of the Volta potential is in accordance to the concept described by Moore and Hummel which, in essence, goes back to the definitions of Erich Lange [42].

For comparison with experimental measurement by SKPFM, the Volta potential difference between the IMP and the atomic force microscopy (AFM) probe $({\psi _{IMP}} - {\psi _{Probe}})$ or the difference between the matrix and the probe $({\psi _{Matrix}} - {\psi _{Probe}})$ was derived from calculated work function data according to Eqs. (3) and (4). More detailed description about the procedure for calculating Volta potentials of the metal with varying crystallographic orientations of different IMP’s can be read elsewhere [36].

$$\Delta \psi _{{Probe}}^{{IMP}}={\psi _{IMP}} - {\psi _{Probe}}=\frac{{{\phi _{IMP}} - {\phi _{Probe}}}}{e}$$

(3)

$$\Delta \psi _{{Probe}}^{{Matrix}}={\psi _{Matrix}} - {\psi _{Probe}}=\frac{{{\phi _{Matrix}} - {\phi _{Probe}}}}{e}$$

(4)

Focus was given to the change in Volta potential differences between the IMP’s and the matrix and, therefore, the effect of the probe was omitted by subtracting Eq. (4) from Eq. (3). Then, the Volta potential difference between an IMP and the Al matrix at varying magnitude of water coverage $\theta$ was calculated according to following equation:

$$\Delta \psi _{{Matrix}}^{{IMP}}(\theta )=({\phi _{IMP}}(\theta ) - {\phi _{matrix}}(\theta ))~/e$$

(5)

where ${\phi _{IMP}}(\theta )$ and ${\phi _{matrix}}(\theta )$ are the work function of the IMP and the Al matrix at a certain magnitude of water coverage $\theta$, respectively. Selected representative terminals were included into the calculation to highlight the importance of surface terminals of the IMP. Only Al(111) orientation was chosen to represent the Al matrix to simplify the calculation. Moreover, for comparison with aged Al alloy samples (natural oxidation), α-Al₂O₃(0001) was also chosen to represent an oxide-covered surface of the alloy matrix.

3 Experimental

The effect of RH on Volta potential changes of the Al-matrix as well as the IMP’s was studied. Discussion is restricted to the IMP’s (Mg₂Si and Al₂Cu) as well as their Volta potential differences with respect to the alloy matrix in order to compare with DFT data. The influence of water adsorption on the AFM probe was disregarded. The focus was on the Volta potential difference between the IMP and the matrix and, therefore, the role of the probe was omitted.

3.1 Material Used

The extruded, commercial aluminum alloy AA6063-T5 was used in this work. The chemical composition (wt%) was 0.6% Mg, 0.45% Si, 0.5% Fe, 0.2% Zn, 0.15% Ti, 0.1% Cu, 0.1% Mn, < 1.5% others, and Al (bal.). Coupon specimens in dimensions of 10 mm × 10 mm × 2.5 mm were cut from the extruded bar and ground and polished to ¼ µm using diamond suspension, followed by an end-polishing using a suspension containing active oxide silica particles (OP-S) from Struers (Denmark) for 30 min (pH 9). The latter removes gently surface strains originating from the grinding and polishing treatment from the surface and yields smooth surfaces. The sample was stored in ambient air for 2 months for natural oxide growth (aging) to achieve steady-state conditions. Two indents were placed on the surface as fiducial markers using a scriber to easily locate the region of interest for correlative analyses.

3.2 Microstructure Characterization

The microstructure of the sample was characterized to identify IMP’s and other microstructural details for correlative Volta potential measurements. For this, the JEOL JSM-7001F scanning electron microscope (SEM) equipped with an Xmax energy-dispersive X-ray (EDX) silicon drift detector (SDD) with 80 mm² window size from Oxford Instruments was used. EDX was used for chemical analysis to determine phase identification. Information obtained from EDX measurements were interpreted in conjunction with existing X-ray diffraction (XRD) data and geometrical information about the IMP’s obtained from SEM analysis and literature reports. SEM imaging was performed in backscattered electron (BSE) and secondary electron (SE) modes. An accelerating voltage of 10–15 kV with high probe current was used for imaging the microstructure. High-resolution EDX measurements were acquired using AZtec V3.3 software from Oxford Instruments using an accelerating voltage of 5 kV with high probe current and high detection resolution (process time 5, dead time 50–60%).

3.3 Volta Potential Microscopy

The SKPFM technique was used to determine local Volta potentials of the microstructure. Volta potential measurements were carried out prior to SEM analyses in order to avoid carbon burnt onto the surface. The Dimension Icon atomic force microscope (AFM) from Bruker with OSCM-Pt R3 platinum-coated (20 nm) n-doped silicon probes from Olympus was used in amplitude modulation (AM-KPFM) mode (two-pass technique). The scan rate was between 0.35 and 0.5 Hz. The scan size varied between 60 µm × 60 µm to 10 µm × 10 µm, depending on the region of interest to be mapped. Each map contained 128 × 128 or 256 × 256 pixels, corresponding to a mapping time between 5 and 17 min with spatial resolutions between 50 and 200 nm. A lift height of 50 nm was used. A nullifying potential (drive routing) of 6000 mV was applied to the sample. No data flattening, smoothing, or inversion was done. An inversion was not needed because the nullifying bias was applied to the sample which is the same if the bias would have been applied to the tip followed by an inversion. The polarity and magnitude of all measured Volta potentials conform to the classical electrochemical concept in which high values are shown as electrochemically noble (cathodic) and lower ones as electrochemically active (anodic). The potential values in the maps are in reference to Pt.

Data interpretation was done according to the following equation in accordance to the description provided by Moore and Hummel [42]:

$${\psi _S}={\psi _{Pt}}+{V_{CPD}}$$

(6)

where ${\psi _S}$ is the local Volta potential of the sample, ${\psi _{Pt}}$ the Volta potential of the AFM probe (here Pt), and ${V_{CPD}}$ the contact potential difference i.e. the measured local Volta potential difference between the sample and the probe.

Platinum is typically one of the most noble metals and is nobler than aluminum. However, oxide on the surface of aluminum and other metals can enhance the work function and consequently increase its Volta potential to nobler values [43]. If the sample is the routing drive (as in this paper), then ${\psi _S}$ is nobler than ${\psi _{Pt}}$ for positive ${V_{CPD}}$. Conversely, if ${V_{CPD}}$ is negative, then ${\psi _S}$ is less noble than ${\psi _{Pt}}$ (not applicable to the data reported in this paper, but an important complimentary information to understand the polarity convention of ${V_{CPD}}$). In this paper, areas on the sample with lower measured ${V_{CPD}}$ (Volta potential difference) indicate electrochemically-active (anodic) character, whereas higher ${V_{CPD}}$ imply higher electrochemical nobility (cathodic) with respect to each other. In this context, ${V_{CPD}}$ and the Volta potential difference between the local microstructural site and the probe are quantitatively the same and will be denoted as “Volta potential vs. Pt” in further communication. It should be noted that all measured potentials with their interpretations have been carefully checked by measuring against a known reference. The PFKPFM-SMPL reference sample from Bruker, which has discrete regions of Au and Al coating, was used for this purpose.

The Volta potential indicates the relative tendency of metals to corrosion and can help to assess the severity of localized corrosion in alloys with complex microstructures [42, 44,45,46]. In a galvanic couple, the phase with higher electrochemical nobility has typically higher Volta potentials and usually acts as the cathode, whereas that with lower Volta potentials acts as the anode [47, 48]. For a detailed interpretation of the polarity the reader is referred to previous works [47, 49, 52].

All SKPFM measurements were performed in controlled atmospheric environment, with the RH ranging between 25 and 95%. The sample was mounted in a humidity cell from Bruker and the humidity was provided by a Cellkraft P-2 humidifier using a flow rate of 200 nccm. The temperature and humidity were continuously recorded by a Sensirion SHTC1/SHTW2 USB sensor connected to a laptop. The cell temperature was kept constant at 23 °C. Volta potential measurements over a region of 60 µm × 60 µm containing an Al₂Cu IMP were performed at 25, 51, 81, 84, and 88% RH with ± 3% uncertainty. At least 30 min were given to equilibrate, and then the Volta potential differences were measured repeatedly to achieve reproducible and reliable data. Then, the relative humidity was increased to the next increment when no further changes were observed. The humidity was set back to ambient humidity (27% RH) to monitor remaining Volta potential changes.

Next, further Volta potential measurements over a region of 8 µm × 8 µm containing an Mg₂Si IMP was carried out on the same sample at 27, 60, 90, and 95% RH with ± 3% uncertainty. The humidity was also set back to ambient air at the end of the measurement to monitor remaining Volta potential changes. Potential (vs. Pt) and corresponding topography (height) maps were produced using Nanoscope V1.6 from Bruker. The data was not flattened. Potential line profiles were sectioned from selected sites and plotted in potential vs. distance charts, from which the Volta potential difference between the IMP and the alloy matrix was obtained. Interpretation of SKPFM data was done in accordance to previous works [49,50,51,52].

4 Results

4.1 DFT Calculations of the Work Function and Volta Potential Differences

Table 1 summarizes calculated work functions of all bare surfaces without and with H₂O adlayers (from 0 to 2 ML). For Al(111), a coverage of 0.25 ML of H₂O, which corresponds to a single water molecule present on the surface of the calculation slab, leads to an abrupt work function decrease from 4.16 to 3.05 eV. Further increase of the water coverage to 1 ML increases the work function to 4.16 eV. The work function changes only slightly when the water coverage is further increased to 2 ML of H₂O. α-Al₂O₃(0001) has a high work function (6.72 eV) when no water adsorbs, and its work function changes slightly (~ 0.7 eV at 0.25 ML) with the increase of water coverage. The work function of Al₂Cu(110) shows a small difference between Cu and Al₂ terminals, and a similar trend of obvious drop followed by a rise for both Cu and Al2 terminals with increasing H₂O coverage, i.e. decreasing first but then increasing again, having the lowest level at 0.5 ML H₂O for the Al₂ terminal and 1 ML H₂O for the Cu terminal, respectively.

Table 1 Calculated work functions (${\phi _M}$) of Al, α-Al₂O₃(0001), Mg₂Si, and Al₂Cu surfaces with different surface terminal as a function of adhering H₂O molecules and magnitude of water coverage (${\theta _{{{\text{H}}_2}{\text{O}}}}$)

In contrast, there is a significant influence of the surface terminal on the work function of Mg₂Si. Without H₂O on the surface, the work function of Mg₂Si-Si and Mg₂Si-Mg is 5.4 and 3.3 eV, respectively, clearly showing a large difference between both terminals. The work function of both terminals decreases with the increase of H₂O coverage. Mg₂Si with MgSi-terminal and Al(111) have similar work functions when no water is present, but show large differences when the surface is covered by water. The work function of Mg₂Si-MgSi decreases dramatically with 0.25 ML of H₂O adsorption, and shows alternating behavior with further rise of H₂O coverage. All surface terminals of Mg₂Si have similar work functions when covered with 2 ML of H₂O (converging trend).

The calculated Volta potential difference between the IMP (Mg₂Si and Al₂Cu with different surface terminals) and Al(111) and α-Al₂O₃(0001) is shown in Fig. 3. Apparently, Mg₂Si and Al₂Cu can be either anodic or cathodic relative to Al(111) (Fig. 3a), depending on the surface terminal and the level of H₂O coverage. The potential difference between the IMP’s for all terminals and Al(111) decreases gradually with the increase of water coverage from 0.25 to 1 ML. Al₂Cu(110) with both terminals and Mg₂Si(111) with Si-terminals are cathodic with respect to Al(111) at 0.25 ML. With the increase in water coverage, their nobilities invert from cathodic to anodic with respect to Al(111). Al₂Cu(110)-Cu becomes slightly cathodic again upon coverage of 2 ML. Seemingly, the Volta potential difference becomes negative for both phases with all terminals, indicating anodic character of the IMP’s relative to Al(111) when the surface is fully covered by water (≥ 1 ML). The Volta potential difference between the IMP’s and α-Al₂O₃(0001) are negative, implying that the IMP’s have anodic character in reference to the α-Al₂O₃(0001). Their anodic nobility remain for all magnitudes of water coverage (Fig. 3b).

Fig. 3

Calculated Volta potential difference of IMP’s (Mg₂Si and Al₂Cu with different surface terminals) relative to Al(111) (a) and α-Al₂O₃(0001) (b) as a function of H₂O coverage. Negative values imply anodic character (red background) and positive values imply cathodic character (blue background) of the IMP’s relative to the matrix

In a nutshell, the DFT calculations suggest that the Volta potential difference between the IMP’s (Mg₂Si and Al₂Cu) and Al(111) decreases when the surface of the IMP is covered by a ML of adhering H₂O. The IMP’s are anodic with respect to the Al-matrix at water coverages of 1–2 ML of H₂O, whereas they always show anodic character relative to α-Al₂O₃(0001) at all water coverage calculated here. A striking result is that Mg₂Si(111)-Si changes from a strong cathodic to a strong anodic character when the magnitude of H₂O coverage increases from 0.25 to 2 ML. Aged aluminum (oxide) has apparently large Volta potential differences as contrasted to bare aluminum (freshly polished), having also large impact on the relative nobilities among microstructural constituents (IMP’s) which may differ from each other.

4.2 Microstructure

The microstructure of the alloy contains Al_4.01Mn(Fe,Cr,Ti)Si_0.74, Al_1.69Mg₄Zn_2.31, FeAl₃, Al₂Cu, ZnAl₂, and Mg₂Si IMPs, all IMP’s embedded in the extruded matrix, as confirmed by XRD and high-resolution SEM analyses. Fe- and Cr-rich Al_4.01MnSi_0.74 constitutes the majority of all IMP’s and they are block-shaped with sizes of 1–10 µm. Al_1.69Mg₄Zn_2.31 and FeAl₃ are also block-shaped and have similar dimensions to Al_4.01MnSi_0.74. Al₂Cu IMP’s are spherical with diameters of 50–400 nm. Mg₂Si also appears spherical with sizes typically ≤ 500 nm. The fraction of Mg₂Si is low, having a fraction of ≈ 0.1 vol%. In the SEM images, all precipitates show brighter contrast than the matrix except for Mg₂Si, which exhibits a darker contrast due to a lower mean atomic number than the matrix. All IMP’s except for Al₂Cu contain Si, Fe, Mn, and Cr, (Fig. 4).

Fig. 4

Numerous inter- and transgranular IMP’s embedded in the microstructure shown in the SEM image (left) with corresponding EDX maps of an aluminum-manganese silicide and a copper aluminide (top) and of a magnesium silicide (bottom)

4.3 Volta Potential Measurements

The local Volta potentials (vs. Pt) measured over a region containing Al₂Cu in dry air (27% RH) are shown in Fig. 5a–d. The Al₂Cu has a spherical shape with ca. 250 nm in diameter and is seen to protrude the surface by $~ \approx 25\,{\text{nm}}$, indicating higher hardness than the matrix (Fig. 5a). In general, the exposure conditions exhibit a strong influence of the measured potentials. The measured Volta potentials over the Al₂Cu show lower values than that over the matrix in ambient air (27% RH), indicating anodic character of the Al₂Cu in reference to the matrix. The increase of RH decreases the Volta potential difference between the matrix and the Al₂Cu. The matrix seems to be more influenced than the Al₂Cu in humid environment. Figure 6 summarizes the change of the Volta potential difference measured between the Al₂Cu and the neighboring aluminum matrix as a function of RH. Apparently, the anodic character of Al₂Cu relative to the neighboring aluminum matrix remains in humid air, but decreases with the increase of RH.

Fig. 5

Volta potential measurements of Al₂Cu and Mg₂Si in the area shown in Fig. 4(left) as a function of RH in reference to their neighboring aluminum matrix: a Topography map, b corresponding Volta Potential differences map obtained in ambient air (27% RH) of the same region shown in a, c at 90% RH, and d at 95% RH of a region containing Al₂Cu (measurement area was 60 µm × 60 µm); e BSE-SEM micrograph showing a Mg₂Si intermetallic, b corresponding topography map of the same region shown in a, corresponding Volta potential differences maps obtained at f 25% RH, g 84% RH, h 88% RH (measurement area was 7 µm × 7 µm). All potentials are in reference to Pt

Fig. 6

Measured Volta potential difference (in average) between IMP’s (Mg₂Si and Al₂Cu) and their neighboring Al-matrix as a function of RH. Negative values imply anodic character (red background) and positive values imply cathodic character (blue background) of the IMP relative to their neighboring Al-matrix

Next, the Volta potential change of an Mg₂Si was analyzed (Fig. 5e–h). In ambient air, the Mg₂Si exhibits lower Volta potentials than the matrix, indicating an anodic character. The average Volta potential difference between the Mg₂Si and the matrix is 20 mV. Furthermore, an increase of RH decreases the electrochemical nobility of both phases, with the matrix apparently more than the silicide. The nobility of the silicide is seen to flip over from anodic to cathodic at 81% RH. Figure 6 summarizes the mean Volta potential difference of Mg₂Si and the neighboring matrix. The silicide is seen to remain cathodic with respect to the neighboring matrix at elevated RH.

5 Discussion

The DFT calculations provide theoretical explanations of the influence of adhering ML of H₂O on the work function and Volta potential of aluminum as well as of Mg₂Si and Al₂Cu. The calculations suggest that water adlayers largely influence the nobilities of all microstructure constituents. Water adsorbed on the surface reduces the Volta potential differences between the IMP’s and Al, which can lead to nobility inversions of the IMP’s in reference to Al(111) when the surface is fully covered with water.

The orientation of the molecule dipole is a key factor that influences the work function of metals when exposed to molecular adsorption [53]. The calculations have shown that the work function of each surface decreases when water adsorbs on the surface (Table 1), which indicates that the net orientation of water molecules is pointing upwards [54, 55]. The optimized configuration of 1 ML of water on Al(111) has two H₂O molecules laying upright and two laying downwards. The work function of all atomic terminals of Mg₂Si decreases with increasing water coverage, converging to values between 2.6 and 2.7 eV (Table 1).

The configuration of adsorbed H₂O on Mg₂Si(111)-Si surface suggests that the initial adsorption of H₂O leads to a rearrangement of the surface atoms (Fig. 7). Interaction of the first adhering H₂O molecules with the metal atoms results in a displacement of bulk Mg atoms towards the outermost surface, whereas Si atoms are shifted slightly towards the bulk interior. Consequently, this leads to a compositional change (relaxation) of the outermost surface of the Mg₂Si(111)-Si terminal, showing a mixed structure of Mg and Si atoms. This may be the reason for the same work functions (2.6 and 2.7 eV) at a water coverage of 2 ML.

Fig. 7

Side view of an optimized atomic structure of Mg₂Si(111)-Si terminal (a), and rearranged surface atoms with one adsorbed H₂O molecule (b). Si and Mg atoms are shown in yellow and green colors, respectively. The arrows indicate that the Mg atom is pulled towards the surface whereas the Si atom is repulsed towards the bulk interior

Bare Mg₂Si(111)-Si, Al₂Cu(110)-Al2, and Al₂Cu(110)-Cu were initially cathodic with respect to Al(111), but they became anodic upon H₂O adsorption at 1 ML. In contrast, Mg₂Si(111)-Mg and Mg₂Si(110)-MgSi remained anodic relative to Al(111) with increasing H₂O coverage. The nobility inversion from cathodic to anodic relative to Al(111) for Mg₂Si(111)-Si, Al₂Cu(110)-Al₂, and Al₂Cu(110)-Cu upon water adsorption is most-likely related to the rearrangement of surface atoms (relaxation), causing a compositional variation of the outermost surface due to the interaction of water molecules with the metal atoms (Figs. 7, 8). This explanation may also be applied to all other phases irrespective of their atomic terminal.

Fig. 8

Top view of optimized atomic structures of bare Al₂Cu(110)-Al2 terminal (a) and Al₂Cu(110)-Cu terminal (c), and corresponding surface configurations with 1 ML of H₂O on Al₂Cu(110)-Al2 (b) and Al₂Cu(110)-Cu (d). Al atoms are in purple, Cu in orange, H in white, and O in red colors. The d-spacings (Å) are highlighted in brackets and show inter-atomic distances along the dashed lines

The optimized structures of Al₂Cu(110)-Al₂ and Al₂Cu(110)-Cu with 1 ML of H₂O (Fig. 8) suggest that adhering water molecules may have caused some relaxation of the outermost surface atoms. The d-spacing of the surface atoms showed a contraction of 0.1 and 0.05 Å for Al₂Cu(110)-Al2 and Al₂Cu(110)-Cu terminals, respectively, when water was adsorbed. Thus, adsorbed water on the surface can alter the structure of the outermost surface of the metals investigated, and may lead to nobility changes. This provides an alternative explanation for the different polarities for various IMP’s reported in aluminum and magnesium alloys [56] in addition to de-alloying and oxide formation/dissolution, often proposed in the literature.

It should be borne in mind that DFT has limitations when modeling water-covered systems. The temperature in DFT calculation is 0 K and, therefore, all the considered structures are statically frozen. The considered metal and IMP’s are simple systems with well-defined structures. Moreover, up to 2 ML of water were considered in the calculations only due to limitations in the model and computational tools. In reality, commercial alloys are usually polycrystalline materials and, therefore, crystallographic orientations as well as surface defects can complicate the calculations. Furthermore, the amount of adsorbed water can be far more than 2 ML in ambient environment, in particular at high RH. Thick water adlayers may further alter the Volta potentials and, hence, affect the electrochemical nobilities of the aluminum alloy microstructure. The surface of the sample used for SKPFM experiments was much more complex than that considered for the DFT calculations. The alloy was always in contact with the ambient environment and, therefore, was continuously exposed to gas molecules in air, which certainly led to the formation of oxides/hydroxides on the surface and, perhaps, also to early stages of atmospheric corrosion. All these aspects could not be taken into consideration in the DFT calculations, and therefore, a direct one-by-one comparison of DFT with SKPFM data is not possible. However, both DFT and SKPFM showed good agreement in the change of Volta potential as a function of water-exposed environment.

Both the DFT calculations and SKPFM measurements indicate that there are large Volta potential differences between the IMP’s and the aluminum matrix. The magnitude of the Volta potential difference between dissimilar microstructural constituents is a quantitative measure for micro-galvanic corrosion [35, 36, 47, 49, 50, 52, 57,58,59,60,61,62]. Therefore, a driving force for localized attack can be expected which indeed is commonly observed in aluminum and magnesium alloys [34, 63,64,65,66].

The Volta potential measurements show that Al₂Cu in the aged sample is anodic relative to the alloy matrix and remains anodic during exposure to humidified air. Mg₂Si was also anodic with respect to the matrix but became cathodic when the humidity reached ≥ 81% RH. Its polarity remained cathodic when the RH was set back to ambient air, clearly indicating a modified surface caused by water adsorption or corrosion. Introduction of water, obviously, affects the electrochemical nobility of all phases. The DFT calculations suggest that different surface terminals of IMP’s may exhibit different polarity, and relaxation and/or reconstruction of water-covered surfaces could be further reasons for nobility inversions, which is predicted by the calculations and observed by the measurements.

Oxide formation, oxide dissolution, and de-alloying phenomena can also change the Volta potential of a phase. In dry atmosphere, the surface of metals is usually covered by a dense and protective (passive) metal-oxide layer which impedes corrosion reactions [67,68,69]. It has been reported that the surface of Mg₂Si can form a passive oxide layer [70,71,72], which grows with time and results in an enhanced electrochemical nobility in both ambient air and corrosive environments [71]. The investigated sample was exposed to laboratory air for 2 months, allowing the formation of oxide/hydroxide on the surface (aging). The Volta potential measured by SKPFM is a weighted average of the Volta potentials from all surface constituents, with a depth resolution of a few Ångstroms to a few nanometers [73]. Hence, the contribution to the Volta potential of the surface oxide/hydroxide and other adhering species is predominant, and could have altered the electrochemical nobility of both the IMP’s and the aluminum matrix.

Volta potential differences of ca. 300 mV between Mg₂Si and the aluminum matrix have been reported [74]. However, in this work, small differences of maximum 25 mV were measured between the Mg₂Si and the alloy matrix, and the Mg₂Si was anodic in reference to the matrix. The calculation show that the IMP’s are cathodic with respect to aluminum when no water is present. The experimental results show, however, anodic properties of both IMP’s with respect to the alloy matrix. The oxide/hydroxide formed on the surface during laboratory air exposure is most likely the reason for the apparent discrepancy. SKPFM was performed at 25–27% RH, and, therefore, one ML of water may have been further adhering on the surface. DFT calculations in dry conditions (no water) are, therefore, not directly comparable with measured data in lab ambient conditions.

The DFT calculations show that all IMP’s exhibit anodic character with respect to the oxide-covered matrix with and without adsorbed water (Fig. 3b), which is in good agreement with the measurement results. The α-Al₂O₃ may not be entirely comparable with the naturally-grown oxide on the commercial alloy but certainly reflects the nature of oxidized aluminum and the change of its Volta potential. Therefore, the calculations with α-Al₂O₃ show comparable Volta potential differences (in the qualitative sense) between the IMP’s and the aluminum matrix with SKPFM data measured on the commercial alloy. Hence, the nobility inversion seen in both calculation and measurements can be explained by the formation of a surface oxide, which led to an increase of the work function of aluminum. Likewise, the formation of oxide/hydroxide on IMP’s can also enhance their nobility, which could further explain the enhanced nobility of the Mg₂Si compared to that reported elsewhere [40, 74, 75].

The introduction of humid air led to water adsorption on the surface of the alloy, which resulted in reconstruction and relaxation of outermost surface atoms (Figs. 7, 8). Water adsorption may have caused further surface reactions, such as passivation or de-alloying, which is known to modify the surface nature of metals. Passivation can increase the nobility of a metal and protect the metal from dissolution. On the other hand, Mg₂Si in aluminum alloys has been shown to de-alloy in Mg in neutral solutions and to become enriched in Si, resulting in an increase of electrochemical nobility [70]. This might further explain the observed polarity inversion from anodic to cathodic. Mg is the most active element in alloy AA6063 and plays a vital role in the nobility of Mg₂Si. The thickness of water adlayer is more than 2 ML at high RH, and it is therefore likely that de-alloying of Mg in Mg₂Si occurred during the exposure to humid air, which can explain the observed nobility inversion from anodic to cathodic with respect to the neighboring matrix. DFT calculations did not consider de-alloying phenomena at elevated levels of water coverage. Post-exposure surface examination by both AFM and SEM showed that the sample surface changed its appearance, manifested by newly-formed oxidation products on large IMP’s in the vicinity of the Mg₂Si (not shown here). The Mg₂Si was not protruding the surface anymore and became slightly recessed indicating that localized corrosion (de-alloying) had indeed occurred during the exposure to humid air.

Most of the IMP’s, including Al₂Cu, in the microstructure usually show cathodic character with respect to the matrix when the material is freshly-polished, which is typically reported in the literature. Natural oxidation, however, as shown in this paper can lead to nobility inversions within the microstructure so that the IMP’s, including the aluminide, become anodic with respect to the aluminum matrix. Similar observations were reported by Leblanc et al. who monitored a polarity change of the measured Volta potential over an Al₂CuMg IMP when the surface was sputter-etched (oxides removed) [76]. Zhou et al. [63] reported de-alloying of Al from Al₂Cu with enrichment of Cu on the outermost surface, in-line with other works [77]. Frankel et al. showed that the oxide formed on IMP’s could alter their electrochemical nobility upon exposure to chloride-containing aqueous solutions, leading to polarity inversions of local Volta potentials [76]. In this study, de-alloying of Al in Al₂Cu may have been possible but it is not expected to occur to a large extent. The trend of the change of the measured Volta potential differences between the Al₂Cu and the matrix as a function of RH is in good agreement with the DFT calculations. DFT calculations showed that Al₂Cu can be either anodic or cathodic with respect to Al(111) depending on the surface terminal and the level of H₂O coverage. Al₂Cu(110)-Cu can be cathodic at water coverages of 2 ML, whereas Al₂Cu(110)-Al2 can be anodic with respect to Al(111), and therefore, it is possible that the Al₂Cu had an Al2 terminal.

In summary, DFT Volta potential calculations have provided valuable information about local electrochemical nobilities of a multi-phase aluminum alloy microstructure. SKPFM measurements yielded useful information for the assessment of the tendency for micro-galvanic corrosion induced by microstructure heterogeneities, as reported extensively in the literature. The nature of the IMP’s with varying surface terminal, humidity, and local surface reactions, such as de-alloying and oxide growth or dissolution, can obviously alter the nobility of a phase and even lead to nobility inversions within the microstructure. The Volta potential gives only an indication for the corrosion propensity of a certain phase in the microstructure. What is anodic in one electrolyte could be cathodic in another and, therefore, one should be careful in the interpretation of Volta potential data (there is further a time-dependent nobility change possible during exposure). As demonstrated in this study, the combination of DFT and SKPFM can provide complementary information that is important for a better understanding of localized corrosion of aluminum alloys and may be used to predict the tendency for localized corrosion, in particular in applications akin to atmospheric exposures.

6 Conclusions

First-principle DFT calculations of work functions and Volta potentials have been performed for Al(111) and Mg₂Si and Al₂Cu with different atomic surface terminals as a function of adsorbed water. The calculation results have been compared with local Volta potential data measured by SKPFM on a commercial Al alloy AA6063-T5. The following conclusions can be drawn:

• 1. DFT calculations are in good agreement with SKPFM data for comparable conditions, both showing similar trends in changes of Volta potential differences between microstructural constituents as a function of adhering water.

• 2. Surface terminals can largely affect the electrochemical nobility of Mg₂Si and Al₂Cu, and, depending on the terminal, these can be either anodic or cathodic with respect to the aluminum matrix.

• 3. According to the calculations, water adsorption causes a reduction of the Volta potential difference between microstructural constituents, which can further lead to an inversion of their relative electrochemical nobilities at higher magnitudes of water coverage (≥ 1 ML of H₂O).

• 4. SKPFM measurements show anodic character of Al₂Cu and Mg₂Si relative to the matrix in ambient air. Al₂Cu remains anodic at all humidities while Mg₂Si becomes cathodic upon exposure to ≥ 81% RH, with the latter due to de-alloying and oxide dissolution.

• 5. The surface oxide has a large influence on the Volta potential and can lead to nobility inversions within the microstructure.