Free radical-induced grafting from plasma polymers for the synthesis of thin barrier coatings

Abstract

Plasma polymer films (PPF) are attracting a great deal of attention for application in various fields due to several remarkable properties, such as good adhesion to different substrates, improved mechanical/chemical stability and a high surface reactivity. This reactivity, associated with the presence of free radicals and originating from the PPF growth mechanism based on many fragmentation and recombination reactions, is often, however, a potential source of trouble. Oxidation of the PPF promptly begins in aerobic conditions via reactions of surface free radicals with oxygen molecules and causes a deterioration of its intrinsic properties in the surface region leaving a nonspecifically functionalized surface in the long-term. Recently a novel approach to functionalize plasma polymer films through the grafting reaction initiated from free radicals trapped on the PPF surface was developed. The present work investigates the potential to employ such an approach in a corrosion protection context. Characterization methods, including Electrochemical Impedance Spectroscopy (EIS) tests, demonstrate that the controlled consumption of surface free radicals via polymer grafting, instead of oxidation, has a beneficial effect on the corrosion protection behavior of the PPF layer deposited on clad 2024 aluminum alloy.

---

Introduction

Recent advancements in science and technology have placed higher demands on engineering materials in terms of quality, quantity and, particularly, properties.¹ When a piece under service is expected to have certain bulk properties while exhibiting, at the same time, different surface properties, surface technology offers encouraging solutions.^2–4 In almost every aspect of life, on macro, micro and nano-scales, thin films synthesized by various deposition methods, e.g. spin coating, solvent casting and electro-deposition techniques as well as plasma-based technologies, help to provide a material with a particular property at the surface while maintaining its valuable intrinsic bulk characteristics.^5–8 When focusing specifically on plasma-based thin films, various applications, ranging from hard coating on cutting tools to optical coatings on glass and dielectric layers for microelectronic devices, are already commonly encountered and many new potentially exploitable routes can be envisioned.^9–13 Among such coatings, organic-like films, commonly referred to as plasma polymer films (PPF) and grown by plasma-enhanced chemical vapor deposition (PECVD) technique, or speaking more pointedly by its particular type – plasma polymerization, are attracting growing interest in the fields of biomedicine and microelectronics.^9,14,15 Due to the specifics of the synthesis process, based on the principles of competitive ablation and polymerization (CAP) explained in details elsewhere,^16,17 PPF exhibit an intriguing set of unique properties. They can be deposited with a thickness ranging from a few nanometers to microns and are characterized by a strong adhesion to a large variety of substrates, including metals, oxides and other organic materials.^18,19 The microstructure of the PPF, exhibiting a high degree of cross-linking, is very different from conventional polymers which are composed of the repetition of original monomer molecules along macromolecular chains. This high level of cross-linking is the reason for their enhanced stability as well as good chemical and mechanical resistances.^3,4,18 Specific properties, related mostly to a desirable composition or a controlled cross-linking degree, can be tuned in a controlled manner by varying the precursor type and/or deposition parameters.^16,18,20 Among many remarkable features of the PPF there is one that needs to be handled very carefully – plasma polymers exhibit a highly reactive surface due to the presence of free radicals generated during the film synthesis.^21,22 These free radicals are responsible for an uncontrollable surface oxidation referred to as ‘plasma polymer ageing’ and taking place inevitably in aerobic conditions.^23,24 This process, progressing with time, leaves the surface randomly functionalized and poorly suited for any subsequent controllable chemical modification. However, surface free radicals, when treated delicately, can be beneficially exploited for grafting of a specific chemical group^25,26 allowing, thus, for a uniform surface alteration to be attained. More importantly, it has been reported that the free radical-induced polymerization of a monomer put in direct contact with a fresh reactive PPF surface can result in a controllable modification of PPF surface properties without affecting its bulk.^27–30

Due to a recently demonstrated interest for plasma polymers in a corrosion protection domain^31,32 as well as the necessity to find eco-friendly substitutes for toxic chromate-based conversion coatings,^33,34 plus considering the good adhesion of PPF to metals and their reinforced (by cross-linking) microstructure, the aim of the present work is to investigate protective properties of a plasma polymer with a controllably modified surface. In the frame of the study a protective bilayer coating on clad 2024 aluminum alloy was developed and characterized in terms of barrier corrosion protection potential. Hexamethyldisiloxane (HDMSO) was chosen as a precursor for the PPF synthesis and a careful control of deposition parameters allowed for a hybrid organic–inorganic plasma polymer rich in free radicals to be produced. A necessary prerequisite for a successful surface functionalization via free radicals, i.e. a sufficient surface free radical density of the HMDSO-based PPF, was tuned and addressed quantitatively by the means of NO chemical derivatization with subsequent X-ray Photoelectron Spectroscopy (XPS) analysis. A reasonable compromise between barrier corrosion protection behavior and surface free radical density allowed a promising candidate for the subsequent grafting procedure with ethyl hexyl acrylate (EHA) to be selected. The characterization of each layer by XPS and successive individual contributions to the barrier protection enhancement of the entire bilayer coating, measured by Electrochemical Impedance Spectroscopy (EIS), are reported. An explanation of enhanced protective properties originating from the addition of a thin grafted layer on the PPF is proposed on the base of Quartz Crystal Microbalance (QCM) measurements.

Experimental part

Materials

The substrate used in the current study is an AA2024 aluminum alloy cladded with a layer of AA1050 (0.7 wt% Si + Fe, 0.1 wt% Cu, 0.05 wt% Mn, 0.1 wt% Zn, 0.03 wt% Ti). The total sheet thickness is 1 mm with a clad layer thickness of 40 μm polished to a final roughness of 30–40 nm. The choice of the material, Al clad 2024, for this study was driven by its wide application in aerospace due to the high mechanical properties of AA2024, arising from Cu alloying, and good intrinsic corrosion protection properties of AA1050 brought by the protective native alumina layer present at the surface.³⁵ Therefore, given that the electrochemical behavior of the bare substrate is rather stable (no corrosion activity), the deposition of coatings on this substrate will allow only the barrier properties of synthesized layers to be evaluated.

2-Ethylhexyl acrylate (EHA, VWR, 99%) was passed through a basic alumina column to remove the stabilizing agent prior to use. Chloroform (Alfa Aesar, 99+%), isopropanol (MERCK, 99.8%), hexamethyldisiloxane (HMDSO, VWR, 99.7%), nitric oxide (Air Liquide, 99.9%), azobisisobutyronitrile (AIBN, Acros, 98%) and tetrahydrofuran (THF, VWR, for analysis) were used as received.

HMDSO-based PPF synthesis

HMDSO-based PPF were deposited on 4 × 4 cm² Al clad 2024 substrates (acetone/methanol pre-cleaned) using a balanced radio-frequency (RF) magnetron sputtering (MS) of a Cerium target (99.9% purity, 76 mm in diameter and 3 mm in thickness) as plasma source. The sputtering process was performed in HMDSO–Ar gas mixture atmospheres. The choice of the target material, Cerium, is motivated by the reported beneficial effects of Ce compounds incorporation on the corrosion protection performance of barrier coatings.^36,37 However, the current work does not pursue the goal of exposing or studying potential Ce-related self-healing effects given the type of substrate selected (non-reactive) and the barrier nature of deposited films. The experimental setup including the deposition chamber as well as other vacuum chambers is presented in Fig. 1.

Fig. 1 Schematic drawing of the experimental setup comprising the introduction chamber, pre-treatment chamber, PVD chamber and derivatization/grafting chamber.

The magnetron was fed by an RF power source (ATX-600 from Advanced Energy) equipped with a corresponding matching network. The target-to-substrate distance was fixed at 50 mm. A base pressure in the order of 10⁻⁵ Pa was achieved by a combination of scroll and turbomolecular pumps. The Ar and HMDSO gas flows were independently controlled by separate mass flow controllers while the working pressure of 1.33 Pa was kept constant during deposition with the help of a throttle valve. Prior to the PPF synthesis Al substrates were pre-treated for 10 min in an Ar/O₂ discharge (25 sccm each) at the pressure of 6.67 Pa with the help of an induction coil-generated RF power of 100 W in a separate vacuum chamber (Fig. 1). The transfer to the PPF deposition chamber was subsequently carried out under vacuum (<10⁻⁵ Pa) during less than 60 s. For all the samples a thickness of ∼100 nm was measured by a mechanical profilometer Dektak 150 from Veeco.

Derivatization tests

After the PPF synthesis samples were quickly transferred (<3 min) in vacuum (<10⁻⁵ Pa) into a separate chamber (Fig. 1) with a base pressure below 10⁻³ Pa for derivatization tests. Chemical labeling of the surface free radicals was performed with nitric oxide (NO), which was introduced into the chamber until a pressure of 10⁴ Pa (Pirani gauge controlled) was reached by blocking out of a turbo pump, according to the procedure established and reported previously.^29,30

Grafting experiments

Grafting experiments were performed in the same vacuum chamber as derivatization tests (Fig. 1). EHA solution was passed through a basic alumina column to remove the stabilizing agent before being introduced into the chamber through the valve located above the substrate-holder after degasing for 5 min with Ar in order to remove residual oxygen. Pressure was increased with Ar to slightly below atmospheric pressure prior to the introduction of the monomer in order to avoid violent spraying of EHA upon contact with the base vacuum of the chamber and to ensure a complete coverage of the PPF sample by the monomer solution during 1 h. The chamber was preheated to 50 °C by an external heater to guarantee an efficient polymerization reaction. After polymerization samples were washed with chloroform for 10 min so that any physically adsorbed layers could be removed before subsequent analyses.

X-ray Photoelectron Spectroscopy (XPS)

The chemical composition of PPF was evaluated by XPS using a Versaprobe PHI 5000 hemispherical analyzer from Physical Electronics with a base pressure below 10⁻⁷ Pa. The X-ray photoelectron spectra were collected at the take-off angle of 45° with respect to the electron energy analyzer, operating in constant analyzer energy (CAE) mode (23.50 eV). The spectra were recorded with monochromatic Al Kα radiation (15 kV, 25 W) with a highly focused beam size of 100 μm. The energy resolution was 0.7 eV and the binding energy scale of the spectra was calibrated with respect to the aliphatic component of the C 1s peak at 285 eV. Surface charging was compensated by a built-in electron gun and an argon ion neutralizer. In order to address the actual bulk chemical composition of the films, that might be disguised by the inevitable surface contamination during sample transportation, a 2 minute pre-treatment with an Ar ion beam (2 keV) was carried out in the same vacuum chamber prior to the analysis.

Electrochemical impedance spectroscopy (EIS)

EIS measurements were performed on uncoated and coated Al substrates. A conventional three-electrode cell was used. The sample under investigation constituted the working electrode. The counter electrode was a platinum plate and all potentials were measured with respect to an Ag/AgCl reference electrode (+197 mV/SHE). In order to minimize external interference on the system, the cell was placed in a Faraday cage. The sample immersion area into 0.1 M NaCl electrolyte solution was ∼0.79 cm². Impedance measurements were performed over the frequency range from 100 kHz to 10 mHz with an amplitude signal voltage of 5 mV. The impedance spectra were obtained using a Parstat 2273 device controlled by the Powersuite software.

Quartz crystal microbalance

EG&G Quartz Crystal Analyzer QCA917 was used as an oscillating circuit and analyzing device in order to measure the frequency variation of the quartz crystal coated with thin films upon exposure to a saline medium (0.1 M). The piezoelectric quartz crystal selected for the QCM analysis was a square-shaped crystal model QA-A9M-AU (Ametek, USA). The resonance frequency of the quartz crystal was 9 MHz in AT-cut (polished surface) with gold-plated metal electrodes on both sides and an electrode area of 0.2 cm². The oscillator frequency measurement was performed by a frequency counter with an accuracy of ±1 Hz. The frequency variation can be correlated to mass variation of the electrode and for the piezoelectric quartz in use, the Sauerbrey equation³⁸ was defined (1):where Δm is the adsorbed mass in g; ΔF is the frequency shift in Hz; F₀ is the fundamental frequency of the quartz: 9 MHz; A is the total sensitive surface of the electrodes: 0.2 cm². The constant 0.44 has units of g MHz cm⁻². The theoretical frequency decrease caused by the adsorption of 1 ng substance is about 1 Hz.

During the PPF deposition and immersion into EHA solution, electrical contacts were protected in order to avoid their degradation and corrosion. Only one side of the quartz microbalance sensor was coated with the material under study, namely the PPF with grafted chains of poly(EHA). PPF was deposited on a gold QCM sensor and immersed into EHA solution for 1 h at 50 °C. A neat PPF sample without grafted chains was also investigated for comparative reasons. After washing, the immersed PPF was dried using nitrogen and introduced into the QCM device. Both PPF samples, with and without grafted chains, were then submerged into a saline solution for 3.5 h at ca. 25 °C and subjected to the same set of experiments. Upon their withdrawal, films were dried and analyzed again in order to detect any mass uptake or loss induced by immersion in the saline solution.

Results and discussion

HMDSO-based PPF

The first layer of the multilayer coating, namely the PPF with a thickness of 100 nm, was deposited on Al clad 2024 substrates in a vacuum process by a hybrid plasma-based technique combining Plasma Enhanced Chemical Vapor Deposition (PECVD) and Magnetron Sputtering (MS) as reported for other organic-like materials.³⁹ An organometallic precursor, hexamethyldisiloxane (HMDSO), providing simultaneously a carbon reservoir, as the source of radicals for the polymerization reaction, and atomic species responsible for the formation of metal-oxide barrier layer, was selected for the PPF synthesis. In addition, this silicon-based precursor rich in carbon can be an ideal choice to illustrate the universality of the grafting method. By proper tuning of the deposition parameters, the amount of C-centered radicals, capable of initiating the free radical polymerization of 2-ethylhexyl acrylate (EHA) monomer, might be controlled while retaining, at the same time, the good barrier properties reported for silica-based PPF.^40,41 Other interesting features of the PPF, such as good adhesion to metallic substrates and highly cross-linked structure, could be also beneficial in the surface protection context. The amounts of carbon, silicon and oxygen in the HMDSO-based PPF are highly dependent on the deposition parameters, particularly on power and pressure. In the current work powers ranging between 200 and 400 W and a pressure of 1.33 Pa were studied in order to demonstrate the feasibility of this approach. Therefore, the variation of chemical composition in the mixed Si–C PPF would allow one to evaluate its impact on the barrier properties in pursuit of an optimum, particularly taking into account the reported probable self-healing mechanisms associated with incorporation of Ce salt compounds.^36,37

Firstly, Al clad 2024 substrates were pre-treated by Ar–O₂ plasma in a pre-treatment chamber (Fig. 1) in order to remove the surface carbon contamination and to improve thus the adhesion with the film to be deposited.⁴² Then they were transferred under vacuum to the chamber where the PPF synthesis was performed. The effect of the deposition parameters, such as input power and HMDSO flow rate, on the atomic concentration of Ce and the surface radical density were studied. For quantification their values were deduced from the respective Ce 3d and surface N 1s (after NO chemical labeling) XPS spectra. Fig. 2 shows their evolution as a function of sputtering power for PPF depositions with both 5 and 8 sccm HMDSO flow rates. In the studied power range the chemical composition of the HMDSO-based PPF (C and O) as determined after 2 minutes of surface erosion with Ar gun samples, remained relatively stable, being ∼60 and 15 at.%, respectively. At the same time, Si and Ce amounts undergo interdependent changes – the increase of one at the expense of the other. The increase of Ce content accompanied by the decrease of Si is associated with a higher energy input into the discharge and its effect on the target poisoning. With the flow rates used (5 and 8 sccm of HMDSO) the surface of the Ce target is covered by a layer of Ce oxide/carbide which causes a decrease in the sputter yield as compared to the metallic target. Therefore, a stronger Ar ion bombardment is necessary in order to effectively knock out Ce atoms from the target in poisoned state. While at low power (200 W) the fragmentation of the precursor is providing the main source of building blocks for the PPF and relatively few Ar ions are created, at higher powers sputtering of the poisoned target by more abundant Ar ions becomes increasingly significant leading to the substitution of a part of the Si atoms by Ce in the growing film. This is clearly observed for the depositions with 8 sccm HMDSO flow rate when at 200 W a greater part of input energy is spent on the precursor fragmentation while already at 300 W it is sufficient to generate enough Ar ions for an effective sputtering of the poisoned target. This combination of PECVD mechanisms with those of conventional magnetron sputtering allowed one to vary the amount of Ce from 0 to 14 at.%. A considerable reduction of Ce content in the PPF deposited with 8 sccm HMDSO flow rate is attributed to the fact that with the increased HMDSO fraction in the gas mixture the poisoning of the target occurs faster and more Ar ions are required to start the efficient Ce target bombardment. According to recent works,^36,37 low amounts of Ce compounds with an oxidation state of III or IV and which are free to migrate inside the films where they were incorporated, might be associated with an enhanced corrosion protection potential. Therefore the PPF samples obtained at 200 and 300 W and exhibiting ∼0, 3 and 8 at.% of Ce were selected for the evaluation of corrosion behavior (which are henceforth addressed as the samples A, B and C, respectively).

Fig. 2 Ce content in the PPF (squares) and surface free radical density (circles) as the function of deposition power for the HMDSO flow rate of 5 sccm (full symbols) and 8 sccm (hollow symbols).

A brief look on the structure of the PPF can be beneficial for a better understanding of the electrochemical data presented shortly hereafter. XPS spectra of the main constituent elements, shown in Fig. 3, are representative of the film bulk since they were acquired after an Ar gun pre-treatment aimed at removing surface carbon contamination.

Fig. 3 XPS Si 2p, O 1s and Ce 3d spectra of the PPF samples A, B and C.

The Si 2p peak (per se an unresolved doublet) of the sample A (0 : 25% of Ce : Si) can be split into two major components: the lower binding energy contribution is close to the value of a pure Si–C bond while the higher binding energy contribution is associated with oxygen-containing Si bonds.⁴³ With the decrease in Si content (samples B and C, 3 : 22 and 8 : 17, respectively) the carbide component becomes more dominant, as Ce, which has a much higher affinity to oxygen,^44,45 outruns Si in bond formation with available oxygen. Ce 3d spectra for the same samples B and C exhibit two clearly distinguishable pairs of spin–orbit doublets, characteristic of Ce³⁺,⁴⁴ confirming that Ce is present in its oxidized state. As for the O 1s peak, in the absence of Ce (sample A) it is composed of a single component that can be collectively attributed to O bonds in a silane/Si–C matrix. Given the energy resolution of the XPS machine of ∼0.7 eV and the close proximity of O–C and O–Si binding energy values, this assumption seems reasonable in the current context. For the samples B and C, with the increase of Ce content, the second component at lower binding energy starts to grow and compete. It is attributed to the energetically more preferable bonds that O starts to form with Ce rather than with Si. XPS spectra in Fig. 3 serve the goal of demonstrating various interdependencies of chemical bonds (Si bound to O and C, O bound to Si and Ce) and presumably indicate that HMDSO-based PPF have a cross-linked C–O–Si/Ce structure where each element forms numerous bonds with other constituents. XRD data (not shown here) revealed no crystalline phases detected for all three samples suggesting that the PPF samples are, furthermore, amorphous. This finding is important for interpreting electrochemical data because in the absence of grain boundaries between crystalline grains no short circuit diffusion paths are provided for the penetration of corrosive species on their way to the substrate and the barrier properties are expected to be superior.⁴⁶

Good barrier properties as well as active corrosion protection, the latter of which could be possibly provided by the incorporation of Ce, are not the only criteria for optimization of this layer. The other important prerequisite is the surface free radical density, which has to be high enough to guarantee sufficient grafting of the subsequent organic layer and consequently the full covalent coverage of the surface by the polymerized acrylic molecules. The evaluation of the density of free radicals on the surface of the PPF was performed with the help of NO chemical derivatization in combination with XPS analysis according to the procedure established earlier^29,30,47 and the results are presented in Fig. 2 along the right axis. The values of radical density have been determined from the atomic concentration of nitrogen, derived from the XPS N 1s peak, using the formula (2):

where at.% N is the concentration of nitrogen in at.%, N_A the Avogadro constant in mol⁻¹, M the molar mass of nitrogen in g mol⁻¹, ρ the density of the PPF in g cm⁻³ and l the XPS analysis depth in cm. The calculation was carried out assuming a plasma polymer density of 0.9 g cm⁻³.²²

Higher energy causes phenomena such as the precursor fragmentation in plasma, the interaction of the growing PPF front with plasma particles and the vacuum ultraviolet (VUV) radiation all to become stronger. Fig. 2 shows that for both series of depositions with different HMDSO flow rates (5 and 8 sccm) the free radical density increases with power remaining superior for the lower precursor flow rate. For the same deposition power, a lower flow rate leads to a longer residence time of precursor molecules in plasma allowing kinetically for a more extensive fragmentation to take place. At the same time, a higher energy input (for the same flow rate) causes phenomena, such as the precursor fragmentation in plasma, the interaction of the growing PPF front with plasma particles and the VUV radiation, all to intensify. This, in turn, leads to increased bond scissions in precursor molecules producing higher amounts of free radicals. It is worthy to note that the radical generation does not reach a threshold in the studied power range indicating that for these depositions surface free radicals do not undergo recombination reactions resulting in the leveling of their density as observed in our previous studies with other precursors.^29,30,47 The fact that the same powers used in the cases of e.g. isopropanol²⁹ and HMDSO do not give rise to similar radical densities may be attributed not only to different chemical structures of these precursors but also to different energy inputs per molecule. The difference of Yasuda's parameters,¹⁷ Y_HMDSO = 0.37Y_isopropanol, indicates that it is necessary to provide considerably more energy to HMDSO discharge in order to have a comparable energy input per molecule and hence, to fragment it heavily. While the detailed investigation of the HMDSO discharge is not in the scope of the current study, the examined power range seems reasonable enough in order to select a few candidates, taking into account that the further power increase would, most probably, lead to a gain not only in the free radical density but also in the Ce content, which is undesirable. The samples A, B and C chosen on the sole consideration of Ce content exhibit radical densities of ∼2.4, 4.1 and 6.1 × 10¹⁴ spin per cm², respectively. It should be mentioned that the grafting of the subsequent poly(EHA) organic layer is expected for all the samples because they all demonstrate free radical density values higher than the maximum value (2.3 × 10¹⁴ spin per cm²) obtained for the isopropanol-based PPF – a value that resulted in sufficient grafting.³⁰

Electrochemical impedance spectroscopy is used to evaluate the barrier properties and the electrochemical behavior of the HMDSO-based PPF films deposited on Al substrates. The evolution of impedance modulus and phase diagrams as a function of immersion time in 0.1 M NaCl solution for the HMDSO-based PPF samples A, B and C (containing, respectively, 0, 3 and 8 at.% Ce) on Al substrates as well as for the uncoated Al substrate are presented in Fig. 4.

Fig. 4 Bode diagrams in modulus (black) and phase (red) of an uncoated Al substrate, HMDSO-based PPF samples A, B and C deposited on Al substrate after (■) 1 hour, (□) 6 hours, (•) 7 days, and (✗) 32 days of immersion in 0.1 M NaCl solution.

Electrochemical behavior of the uncoated Al substrate remains rather stable during the whole experiment duration due to the presence of a protective alumina layer on the surface of the metallic substrate, low amount of intermetallics and low surface roughness.³⁵ The low frequency modulus corresponding to the total corrosion resistance of the system is in the order of 10⁵ to 10⁶ Ω cm⁻². As can be seen on impedance diagrams, independently of the Ce content, the presence of PPF improves the corrosion protection of the substrate. This enhancement of corrosion protection is characterized by an increase of the low frequency modulus to values around 10⁶ to 10⁷ Ω cm². However, when considering chemical composition, there are notable differences. The Ce-free sample A demonstrates the highest low frequency modulus around 10⁷ Ω cm², one order of magnitude higher in comparison to the uncoated Al substrate. This efficient protective behavior can be associated with the dense amorphous cross-linked C–O–Si matrix providing a supplementary barrier against the penetration of corrosive species. For samples B and C, the presence of Ce in PPF affects their barrier properties causing the low frequency modulus to be centered slightly above 10⁶ Ω cm², a bit higher than the one observed for the uncoated Al substrate but inferior to the one of the Ce-free sample A. Suggestedly, a considerably higher atomic radius of Ce atoms (185 pm) in comparison to C (70 pm) and Si (110 pm) induces localized defects in the amorphous network facilitating, therefore, the diffusion of corrosive species. For longer immersion times no evident self-healing phenomena are revealed for Ce-containing films as might be expected from Ce incorporation. This is probably due to the fact that Ce atoms are covalently bound in the PPF network (as indicated by XPS analyses) and are not capable of forming ions which can migrate and precipitate on the corrosion site. Moreover, since the substrate used in the current study is quite stable in NaCl solution, the possible self-healing mechanism of Ce ions cannot be clearly observed or put in evidence and only the barrier properties are evaluated.

Table 1 summarizes deposition conditions and obtained characterization data for samples A, B and C.

Table 1 Deposition parameters and measurement results for samples A, B and C

	Power (W), HMDSO flow rate (sccm)	Ce/Si (at.%)	Spin concentration (spin per cm^2)	Low frequency modulus after 7 days of immersion into 0.1 M NaCl solution (Ω cm^2)
Sample A	200, 8	0/25	2.4 × 10^14	1.0 × 10^7
Sample B	200, 5	3/22	4.1 × 10^14	2.0 ×10^6
Sample C	300, 5	8/17	6.1 × 10^14	2.0 × 10^6

---

In light of presented results, the HMDSO-based PPF sample A (0 at.% of Ce) was selected for the further study. Its capacity to initiate polymerization reaction of the EHA monomer via surface free radicals (2.4 × 10¹⁴ spin per cm²) is addressed in the following section.

Free radical polymerization/grafting onto HMDSO-based PPF

Grafting experiments with EHA were performed on the HMDSO-based PPF sample A, containing no Ce, exhibiting surface free radicals of 2.4 × 10¹⁴ spin per cm² and showing improved corrosion protection behavior. Subsequently to its synthesis the PPF sample was immediately transferred under vacuum (in order to avoid quenching of free radicals with oxygen) to the polymerization chamber (Fig. 1) where it was put in direct contact with the liquid EHA monomer for 60 minutes. This monomer was chosen among others due its high polymerization ability and its wide use in coating industry.⁴⁸ During the immersion time, as shown in the example of the isopropanol-based PPF,³⁰ surface free radicals are able to react with EHA molecules via covalent addition onto their carbon–carbon double bonds and initiating free radical-induced polymerization. Upon removal of the sample from the polymerization chamber it was thoroughly washed with CHCl₃, a good solvent of poly(EHA) capable of removing any physisorbed monomer or non-grafted polymer chains (resulting mainly from radical transfer side reactions) via selective solubilization, and leaving covalently anchored polyacrylate chains intact on the PPF surface.

C 1s XPS peaks of the as-deposited PPF and the PPF subjected to immersion into EHA solution are presented in Fig. 5.

Fig. 5 C 1s XPS spectra of the HMDSO-based PPF (dash line) and the PPF with grafted poly(EHA) chains (solid line).

The large peak centered at ∼285 eV represents various C bonds in the cross-linked plasma polymer matrix collectively and is observed in both cases. The main difference in the two spectra arises from the separate peak at ∼289 eV, absent for the initial HMDSO-based PPF and appearing after the grafting procedure. This peak is attributed to the carbonyl group COO, characteristic of poly(EHA). Its observation reveals the presence of covalently bound poly(EHA) chains on the PPF surface confirming that the surface free radicals, accessible to monomer molecules, successfully grafted and initiated EHA polymerization.

EIS data for the HMDSO-based PPF sample (A) with grafted poly(EHA) chains is shown in Fig. 6.

Fig. 6 Bode diagrams in modulus (black) and phase (red) of the HMDSO-based PPF (sample A) with the grafted poly(EHA) layer after (■) 1 hour, (□) 6 hours, (•) 3 days, (○) 21 days and (✗) 35 days of immersion in NaCl solution.

Remarkably enough, the addition of grafted poly(EHA) layer on top of the HMDSO-based PPF film induces further stabilization of anticorrosive properties of the system. An increase of the low frequency modulus to ∼10⁸ Ω cm² as compared with ∼10⁷ Ω cm² for the blank PPF (without grafted chains) as well as more stable modulus and phase angle values are observed over the frequency range 100 kHz to 10 mHz. This enhanced electrochemical behavior (particularly notable in the phase plot characterized by a more stable capacitive trend) observed over the entire frequency range can be associated with the protective effect of the grafted polyacrylate layer and the fact that the surface reactivity (free radicals) of the PPF was intentionally consumed for grafting of polymer chains preventing otherwise inevitable surface oxidation upon exposure to the air. Indeed, once the PPF was taken out of the chamber the surface free radicals reacted randomly with omnipresent oxygen following multiple reaction pathways and generating a chemically inhomogeneous surface.²³ At the same time, one hour of immersion into the EHA solution allows the radical reactivity to be intentionally used for the initiation of polymerization of EHA molecules in a directional surface modification process. It is worth pointing out that even after 35 days of immersion in NaCl solution the low frequency modulus of the PPF surface-covered with the covalently grafted polyacrylates remains fairly superior to the one recorded for the blank PPF. This durable stabilization of corrosion protection performance is particularly notable considering the very low thickness of the grafted polyacrylate layer (∼3.5 nm as evaluated in a previous study where a similar COO amount, ca. 6%, was obtained for the same XPS take-off angle of 45°).³⁰

In order to understand better the effect of the poly(EHA) layer on barrier properties and to validate a potential application of the covalently grafted polyacrylate layer on the PPF for reducing its excessive surface reactivity, we investigated the stability of the system (PPF + grafted layer) upon immersion into a saline solution (0.1 M), similar to the one used for EIS measurements, by QCM analysis. The HMDSO-based PPF (sample A) with and without covalently grafted poly(EHA) chains were analyzed in order to shed some light on the role of these grafted chains. Each sample underwent a cycle of 3.5 h in saline water; analyses in dry conditions before and after the whole experiment cycle were also performed. QCM, a useful tool to study mass changes of the system via the frequency variation response, was employed in order to investigate the difference in the film behavior upon immersion in water arising from the presence or absence of the grafted poly(EHA) chains. Fig. 7 shows the results of the QCM analysis performed for the HMDSO-based PPF with and without the grafted polyacrylate layer for 1000 s of immersion.

Fig. 7 Effect of the saline water on the behavior of the PPF (black) and the PPF with grafted poly(EHA) chains (red).

Notably, at the beginning of the immersion (<200 s), the exposure of both samples to water causes a certain decrease of frequency meaning that the film adsorbs water progressively becoming more and more swollen.³⁸ After 200 s the as-deposited PPF continues to adsorb water as indicated by a considerable frequency decrease during the whole experiment. Once poly(EHA) chains, characterized by a relative hydrophobic behavior with a contact angle of ∼114°, are grafted on the PPF, the decrease of frequency is considerably less pronounced confirming that the consumption of the PPF free radicals by the grafting of an acrylic thin layer affects the PPF behavior in contact with saline water causing the PPF surface to exhibit a lower reactivity. It was already reported that even a grafting of a few nm of polymer chains on a PPF can influence the macroscopic behavior of a PPF.^5,7 It is of interest to report that on a longer immersion scale the frequency of PPF with grafted chains decreases with a trend similar to the as-deposited PPF (not shown here) probably due to the water absorption by the edges of crystal not covered by the polymer chains but also subjected to immersion.

It is worth mentioning that a conformational equilibrium is reached for both samples after an immersion time of 3 h in water once the frequency is stabilized. Upon drying, the frequency returns back globally to its initial value meaning that in both cases no considerable water uptake takes place confirming the good barrier properties observed by EIS measurements. Nevertheless, it should be noted that the PPF without grafted chains remains slightly more swollen than the PPF with grafted chains and that it is characterized by a 177 ng higher mass uptake.

Conclusions

In the current work a bilayer thin film coating, synthesized on an Al substrate by a combination of plasma-based and wet chemistry deposition routes and exhibiting enhanced barrier corrosion protection properties, is investigated. The plasma polymer film, synthesized from an HMDSO precursor, was selected as the first layer. EIS analyses reveal that the incorporation of Ce in the PPF, with the help of the balanced magnetron with a metallic target of cerium, altered the corrosion properties of this layer. The Ce-free PPF sample demonstrates a promising corrosion protection behavior (increase of the low frequency modulus by approximately one order of magnitude as compared to the Al substrate). Free radicals present on the surface of a plasma polymer (∼100 nm thick) were first addressed quantitatively by the means of chemical derivatization with NO labeling agent and then were successfully used for the subsequent free radical-induced polymerization of an acrylic monomer, the 2-ethylhexyl acrylate (EHA), that formed the second layer (few nm). Despite its markedly low thickness, the grafted layer further improved the barrier properties of the bilayer coating, as indicated by an increase of the low frequency modulus by one more order of magnitude as compared to the Al substrate covered by the PPF alone. Results of QCM analyses point out that the grafting of the thin poly(EHA) layer on the PPF provides a means of controlling its surface reactivity via directional consumption of surface free radicals. Reported scientific findings suggest that grafting of short polymer chains on the PPF surface provides extra barrier properties against corrosion via generation of a chemically homogenous, non-reactive surface in contrast to the nonspecifically functionalized, inhomogeneous surface of the oxidized PPF.

Acknowledgements

The authors are grateful to the ‘Région Wallonne’ and European Community (FEDER, FSE) in the frame of ‘Pôle d’Excellence Materia Nova’, portfolio ‘Revêtements Multifunctionnels’ and in the excellence program OPTI²MAT for financial support. CIRMAP thanks the ‘Belgian Federal Government Office Policy of Science (SSTC)’ for general support in the frame of the PAI-6/27. Authors also thank Dr Freddy Benard for his general help and we would like to acknowledge Dr Driss Lahem for his help for QCM experiments.

Notes and references

1. L. Nicole, C. Laberty-Robert, L. Rozes and C. Sanchez, Nanoscale, 2014, 6, 6267–6292.
2. A. S. H. Makhlouf and I. Tiginyanu, Nanocoatings and ultra-thin films: technologies and applications, Woodhead Publishing, 2011.
3. D. L. Smith, Thin Film Deposition – Principles and Practice, McGraw-Hill, New York, 1995.
4. M. Ohring, Materials Science of Thin Films – Deposition and Practice, Academic Press, San Diego, CA, 2002.
5. D. M. Mattox, Handbook of Physical Vapor Deposition (PVD) Processing, Elsevier, 2007.
6. A. C. H. M. L. R. S. o. C. Jones, Chemical vapour deposition: precursors, processes and applications, RSC Pub., Cambridge, UK, 2009.
7. P. M. Martin, Handbook of Deposition Technologies for Films and Coatings, 2010.
8. S. Banerjee, T. K. Paira and T. K. Mandal, Polym. Chem., 2014, 5, 4153–4167.
9. S. C. Deshmukh and E. S. Aydil, J. Vac. Sci. Technol., B: Microelectron. Nanometer Struct.--Process., Meas., Phenom., 1996, 14, 738–743.
10. J. Müller, B. Rech, J. Springer and M. Vanecek, Sol. Energy, 2004, 77, 917–930.
11. Z. Bo, Y. Yang, J. Chen, K. Yu, J. Yan and K. Cen, Nanoscale, 2013, 5, 5180–5204.
12. P. Attri, B. Arora and E. H. Choi, RSC Adv., 2013, 3, 12540–12567.
13. L. Xu, Z. Fang, P. a. Song and M. Peng, Nanoscale, 2010, 2, 389–393.
14. S. Bhatt, J. Pulpytel, M. Mirshahi and F. Arefi-Khonsari, Polymer, 2013, 54, 4820–4829.
15. K. Vasilev, N. Poulter, P. Martinek and H. J. Griesser, ACS Appl. Mater. Interfaces, 2011, 3, 4831–4836.
16. J. Friedrich, Plasma Processes Polym., 2011, 8, 783–802.
17. H. K. Yasuda, Plasma Processes Polym., 2005, 2, 293–304.
18. H. Biederman and Y. Osada, Plasma Technology, Elsevier Science Publishers, Amsterdam, 1992.
19. M. E. Alf, A. Asatekin, M. C. Barr, S. H. Baxamusa, H. Chelawat, G. Ozaydin-lnce, C. D. Petruczok, R. Sreenivasan, W. E. Tenhaeff, N. J. Trujillo, S. Vaddiraju, J. Xu and K. K. Gleason, Adv. Mater., 2010, 22, 1993–2027.
20. A. Michelmore, D. A. Steele, D. E. Robinson, J. D. Whittle and R. D. Short, Soft Matter, 2013, 9, 6167–6175.
21. H. Biederman and Y. Osada, Plasma Polymerization Processes, Elsevier Science, 2001.
22. R. Wilken, A. Holländer and J. Behnisch, Macromolecules, 1998, 31, 7613–7617.
23. T. R. Gengenbach and H. J. Griesser, J. Polym. Sci., Part A: Polym. Chem., 1999, 37, 2191–2206.
24. J. D. Whittle, R. D. Short, C. W. I. Douglas and J. Davies, Chem. Mater., 2000, 12, 2664–2671.
25. M. M. M. Bilek, Appl. Surf. Sci., 2014, 310, 3–10.
26. A. Michelmore, D. A. Steele, J. D. Whittle, J. W. Bradley and R. D. Short, RSC Adv., 2013, 3, 13540–13557.
27. D. O. H. Teare, W. C. E. Schofield, R. P. Garrod and J. P. S. Badyal, Langmuir, 2005, 21, 10818–10824.
28. B. W. Muir and A. Tarasova, Plasma Processes Polym., 2012, 9, 884–889.
29. S. Ershov, F. Khelifa, P. Dubois and R. Snyders, ACS Appl. Mater. Interfaces, 2013, 5, 4216–4223.
30. F. Khelifa, S. Ershov, Y. Habibi, R. Snyders and P. Dubois, ACS Appl. Mater. Interfaces, 2013, 5, 11569–11577.
31. V. Barranco, J. Carpentier and G. Grundmeier, Electrochim. Acta, 2004, 49, 1999–2013.
32. G. Grundmeier, P. Thiemann, J. Carpentier and V. Barranco, Surf. Coat. Technol., 2003, 174–175, 996–1001.
33. S. M. Cohen, Corrosion, 1995, 51, 71–78.
34. M. W. Kendig and R. G. Buchheit, Corrosion, 2003, 59, 379–400.
35. D. A. Brevnov, G. V. Rama Rao, G. P. López and P. B. Atanassov, Electrochim. Acta, 2004, 49, 2487–2494.
36. K. A. Yasakau, M. L. Zheludkevich, S. V. Lamaka and M. G. S. Ferreira, J. Phys. Chem. B, 2006, 110, 5515–5528.
37. R. G. Buchheit, S. B. Mamidipally, P. Schmutz and H. Guan, Corrosion, 2002, 58, 3–14.
38. W. Steinmann, Z. Physik, 1961, 163, 92–107.
39. F. Henry, F. Renaux, S. Coppée, R. Lazzaroni, N. Vandencasteele, F. Reniers and R. Snyders, Surf. Sci., 2012, 606, 1825–1829.
40. E. Vassallo, A. Cremona, L. Laguardia and E. Mesto, Surf. Coat. Technol., 2006, 200, 3035–3040.
41. K. Teshima, H. Sugimura, Y. Inoue, O. Takai and A. Takano, Appl. Surf. Sci., 2005, 244, 619–622.
42. Y. Qingsong, in Silanes and Other Coupling Agents, CRC Press, 2009, vol. 5, pp. 151–180.
43. M. R. Alexander, R. D. Short, F. R. Jones, W. Michaeli and C. J. Blomfield, Appl. Surf. Sci., 1999, 137, 179–183.
44. E. J. Preisler, O. J. Marsh, R. A. Beach and T. C. McGill, J. Vac. Sci. Technol., B: Microelectron. Process. Phenom., 2001, 19, 1611–1618.
45. A. Pereira, M. Blouin, A. Pillonnet and D. Guay, Mater. Res. Express, 2014, 1, 015704.
46. Y. Gao, Z. J. Zheng, M. Zhu and C. P. Luo, J. Mater. Sci. Eng. A, 2004, 381, 98–103.
47. S. Ershov, F. Khelifa, V. Lemaur, J. Cornil, D. Cossement, Y. Habibi, P. Dubois and R. Snyders, ACS Appl. Mater. Interfaces, 2014, 6, 12395–12405.
48. F. Khelifa, M.-E. Druart, Y. Habibi, R. Rioboo, M. Olivier, J. De Coninck and P. Dubois, J. Mater. Chem. A, 2013, 1, 10334–10344.

---

Footnote

† These authors contributed equally.

---