Corrosion inhibition of iron surfaces with phosphatidic acid

1,2-dimyristoyl-sn-glycero-3-phosphate (DMPA) was purchased from Avanti Polar Lipids (Alabaster, AL, USA). A stock solution of the DMPA was prepared by dissolving 10 μmol (6.15 mg) of the DMPA in 750 μl chloroform and 250 μl methanol. Then 0.5 μmol lipid was measured out into a glass test tube, dried under a gentle stream of nitrogen gas and placed in a desiccator overnight. Then it was dissolved in 1ml PBS and gently sonicated to form a suspension of spherical bilayer vesicles for surface modification in the QCM.

Mass loss experiments were conducted using a Q-Sense E1 instrument (Q-Sense AB, Gothenburg, Sweden). Iron coated sensor chip QSX 319 with 100 nm Fe layer (nominal), packed in inert gas, was purchased from the same company. Substrate is an AT cut quartz crystal with a fundamental resonance frequency of 5 MHz. The frequency shift and energy dissipation were recorded at the 3^rd, 5^th, 7^th, 9^th, 11^th and 13^th eigenmodes; reported sensograms are for the 5th eigenmode. Corrosion as mass loss (dissolution of iron) was measured with nanogram accuracy using acetic acid (HAc) as the corrosive agent in aqueous solution. 0.1 M and 0.01 M HAc solutions were used at 25 °C in situ in the QCM chamber. Reference sensograms were recorded in deionised water (18.2 MΩcm) for 24 hrs.

Lipid deposition was performed in the QCM chamber in situ by introducing 0.5 μmol of DMPA suspended in 1 ml PBS. PBS was used to improve the stability of the liposomes used for deposition (27, 28, 29). First the chip was cleaned with deionized water and dried with nitrogen gas before assembling into the QCM chamber. For a typical experiment, water was flushed in the chamber for at least 5 min. A water baseline was recorded before introducing the PBS. After a PBS baseline DMPA liposome suspension was introduced and the frequency and the dissipation change was recorded in real time. Finally, a PBS wash was used to achieve the final baseline that was used to determine the amount of lipid attached to the surface. Experiments were repeated at least 3 times (i.e. n=3).

Adsorbed/removed mass and the oxidation rate were calculated using the Sauerbrey equation (30)

where ᐃm is the adsorbed mass in ng/cm², C is a constant (17.7 ng cm^-2 Hz^-1 at f = 5 MHz, the eigenfrequency of the unloaded crystal) and Δf is the change in frequency at n overtone of the crystal oscillation.

The chemical composition of the oxidized iron surface was characterized by X-Ray Photoelectron Spectroscopy (XPS) using a Kartos Axis Nova spectrometer (Kartos Analytical, Manchester, United Kingdom). A monochromatic Al K_α source (10 mA, 15 kV) was used at 7 x 10^-9 Torr pressure. The data were analysed using CasaXPS (version 2.3.15) and Origin Pro 2015 software.

The analysis of the iron 2p_3/2 peak is particularly challenging due to the series of satellite peaks present in each oxidation state, necessitating the use of synthetic model functions formed from a series of peaks (31, 32, 33). Given that on an oxidized iron surface at least three oxidation states are likely present with potential shifts due to changing chemical environment, quantitative analysis is not possible; however a qualitative assessment of the iron species can be attempted (34, 35) For this we followed the data published by Biesinger et a³³ that includes peak parameters for FeOOH (Table 1) to fit the experimental data using the

CasaXPS software. In this approach the peak positions, FWHM and relative intensities of the peak series for each species are fixed; only the intensities of the entire series are altered to fit the experimental data. As the table shows, there is a distinct shift between the Fe₂O₃ and FeOOH peak series that allows for qualitative assessment of the presence of the three iron species.

Layer thicknesses were measured using a Picometer Ellipsometer (Beaglehole Instruments, New Zealand). All measurements were conducted on dried samples. Variable angle scan method was used at λ = 633.0 nm wavelength for measuring the thickness of iron oxide. There are no dielectic parameters for DMPA in the database, therefore a material file was created based on Cauchy model (i.e. dielectric material). The Cauchy parameters were calculated from spectroscopic measurements on lipids at a wavelength range of 400-800 nm at 67 degrees incidence angle. The thicknesses of the DMPA layers were measured based on angle scans at 550 nm wavelength. Fitting for layer thickness was conducted in Thin Film Companion (TFC) software (Semiconsoft).

To establish a reference, corrosion of iron in acetic acid was monitored using QCM. Experiments were performed with 0.1M and 0.01M acetic acid at 25 °C, and the rates of iron dissolution were established. Fig. 1. shows and example of the experimental results, the effect of 10 mM acetic acid on the iron surface at 25 °C. After an initial period (~8 min) linear increase of the frequency from zero to 2.3 kHz is observed, indicating substantial mass loss, when the acid was flushed with deionized water to stop further dissolution. The initial resistance of the layer to dissolution is consistent with the presence of a thin passivating oxide layer that was breached at about 8 min. This pattern was observed in all corrosion experiments, with some variation of the time needed for activation. The rate of corrosion is a function of several factors, including temperature, acid concentration, as well as surface roughness of the iron coating and the thickness and composition of the surface oxide layer; accordingly the measured rate was varied. At 25°C for 0.01 M HAc solution it was typically ~ 6 Hz/min, or ~0.1 mg/min, and for 0.1 M HAc concentration the rate increased to ~400 Hz/min or ~7.1 mg/min. This is a suitable reference for the qualitative study of corrosion inhibition.

Figure 1

DMPA was deposited on the iron coated sensor chip surface using the liposome deposition method. Experiments were conducted at 25 ºC. A typical sensogram is shown in Fig. 2. First deionised water was introduced into the chamber until the sensor signals stabilized. Next PBS was introduced for 10 min

Figure 2

to obtain a baseline. This was followed by the DMPA liposome suspension that lead to a sudden decrease of frequency, indicating mass uptake, as the DMPA liposomes attached to the surface. PBS rinse was applied after the DMPA deposition to remove excess liposomes, followed by osmotic stress treatment with pure water to burst open any remaining liposomes. PBS was reintroduced to measure the final mass of the deposit layer. The frequency change for surface bound lipid was ~ -20 Hz in PBS; this value is consistent with a single bilayer surface coating of a tightly bound layer (25, 26). 0.01 M acetic acid was introduced at 1.7hrs; the slight change in frequency reflects the different viscosity of the acetic acid solution (i.e. absence of hydrated ions of the NaCl). It is clear that no mass loss was recorded over one hour, unlike in the benchmarking experiments where acetic acid started dissolving the iron chip within minutes. Hence the DMPA bilayer coverage on the surface inhibited corrosion.

In some of the experiments a slow baseline drift was observed

upon addition of the acid solution, which slowed down over time, asymptotically approaching a stable value (Fig. 3). That and the magnitude of the change (~4 Hz) rules out corrosion as the cause; it is more likely that the top leaflet of the lipid bilayer is removed, parallel with further compaction of the bottom (surface bound) leaflet. If this is the case, the exposed surface has to be hydrophobic, opposed to the bilayer with the exposed phosphate groups that is hydrophilic. Hydrophobicity of the coating offers further advantage in restricting water access and reducing viscous drag in applications.

Figure 3

To confirm that the resulting surface is hydrophobic as expected if a single leaflet membrane is present, contact angle measurements were performed before and after the coating. A water drop was stabilized on the chip surface at room temperature and the contact angles were measured as in Figure 4. Before the coating, the water drop preferentially wetted the

Figure 4

iron oxide and the measured contact angle was ~51º (Fig. 4 A). After coating and exposure to acetic acid, the contact angle was about 107º as the water did not spread on the surface (Fig. 4 B).

In the hydrophobic case, the surface energy can be calculated. The work of adhesion is related to the contact angle by the Young-Dupré equation (1) (36, 37).

where γ_l is the surface tension of water, θ is the contact angle. The methyl and methylene moieties of the lipid phase are apolar with low polarizability, therefore even at the water interface the contribution of polar interactions to the total work of adhesion can be neglected. In the absence of polar interactions the work of adhesion at an interface can be also approximated from the surface free energies of the materials that form the interface:

where γdsandγd1${{\gamma }^{\text{d}}}_{\text{s}}\,\text{and}\,{{\mathbf{\gamma }}^{\text{d}}}_{1}$are thedispersion force contributions to the surface free energy (38). For water γd1${{\mathbf{\gamma }}^{\text{d}}}_{\text{1}}$is ~20mJm^-2 whereas for apolar materials, such as hydrocarbons, γds${{\mathbf{\gamma }}^{\text{d}}}_{\text{s}}$makes up 100% of the surface free energy. Hence, for the water-lipid interface this approximation provides a reliable estimate. By substituting equation (2) in the Young-Dupré eq (1) it is possible to calculate the surface tension of the solid-vapor (~solid-vacuum) interface, that is, the surface energy of the solid:

Hence the surface free energy of the lipid monolayer-coated surface is ~32,4 mJm^-2 (θ = 107°) which is comparable to the surface energies of bulk hydrocarbons (e.g. for octadecane γ = 28 mJm^-2). The same approximation cannot be applied to the hydrophilic case. The low surface free energy confirms the removal of the top leaflet of the lipid bilayer, leaving the alkyl chains exposed.

As an additional confirmation of the thickness of the DMPA coating, ellispometry was used. First the thickness of the iron oxide layer was determined after oxidation with acetic acid. The measurements were done on the same QCM chips used for the corrosion experiments. Varied angle method was used to measure the thickness at 633 nm wavelength. The thickness of the iron oxide was measured to be 5.5 nm and this value was then used to parametrize the layer stack for the measurements on DMPA.

The thicknesses of the lipid layers were measured before and after exposure to acetic acid. The measured ellipsometric thickness of the lipid deposit before acid treatment was 4.2 nm, which is consistent with the thickness of a DMPA bilayer. The thickness after acid treatment was reduced to 2.0 nm which is close to the theoretical and practical value of the self-assembled monolayer of C14-PA (39). This result is further proof that upon exposure to acetic acid a monolayer of DMPA remained on the iron oxide surface.

The formation of phosphate-based self-assembled monolayers on transition metal oxide surfaces is based on a condensation reaction between the phosphate moiety and hydroxyl groups on the surface (32, 42). It can be assumed that a hydrated metal oxide surface contains hydroxyl moieties; to gain more certainty the surface of an iron coated sensor was characterized by XPS. A representative spectrum of the iron 2p_3/2 peak is shown in Fig. 5 (black line).

Figure 5

Synthetic model functions have been constructed from the characteristic peak series of FeO, Fe₂O₃ and FeOOH in CasaXPS that also employs a range of built-in factors to perform peak

fitting optimized for XPS data analysis. As stated above, in the fit only the intensities of the peak series (i.e. 4 parameters) were fitted: within each series the peak positions, FWHM and relative peak intensities were maintained constant. None of the composite functions could fit the peak accurately when combined only with the Fe⁰ peak. Therefore the analysis of the spectra confirms qualitatively the presence of iron hydroxide besides the Fe²⁺ and Fe³⁺oxide species.