Corrosion resistance of layer-by-layer assembled polyvinylpyrrolidone/polyacrylic acid and amorphous silica films on AZ31 magnesium alloys

Abstract

A layer-by-layer (LbL)-assembled composite coating containing SiO₂ and a biocompatible polyvinylpyrrolidone (PVP) and polyacrylic acid (PAA) multi-layer, designated as SiO₂/(PVP/PAA)₅, was prepared on AZ31 Mg alloy via dip-coating. The surface morphology, microstructure and chemical composition of the coating were investigated using FE-SEM, FT-IR, XRD and XPS. The physical properties of the coating were characterized by scratch testing. The results demonstrated that the coating was amorphous and remarkably soft. PVP and PAA promoted the formation of Ca–P precipitates, and the amorphous silica film further enhanced corrosion resistance. The SiO₂/(PVP/PAA)₅ coating may be a promising surface modification for degradable Mg cardiovascular stents.

---

1 Introduction

Mg and its alloys hold promise as biomedical materials used in biodegradable implants as a result of their favourable biodegradability and elastic modulus close to human bone.^1–3 However, the extensive potential applications of Mg-based alloys is still limited by high degradation rates and consequent loss of mechanical integrity at pH levels between 7.2 and 7.4, and in the chloride-rich environments of physiological systems; biocompatibility is also a problem.⁴

Surface modification is the most common approach because of its simplicity, low cost and high performance. Surface modifications, including micro-arc oxidation (MAO) or plasma electrolytic oxidation (PEO),^5,6 hydrotalcite coating^7–9 and chemical conversion coatings,^10–12 tend to reduce degradation rates, improve alloy biocompatibility and enhance osseointegration between the coated implant and the surrounding tissue. However, several limitations have hindered the application of these technologies. For example, porous MAO coatings enhance the corrosion resistance of Mg–Li–Ca–Y alloy only in the initial stage but promote corrosion over longer times.⁶ The ceramic coating is too hard to meet the high ductility requirements for degradable Mg alloys.

In particular, layer-by-layer (LbL) self-assembly has attracted considerable attention because of its effective, economical and environmentally friendly characteristics, especially for biological applications.^13,14 Films of (heparin/chitosan)₁₀–(polyvinylpyrrolidone/poly(acrylic acid))₁₀ ((HEP/CHI)₁₀–(PVP/PAA)₁₀) fabricated on silicon wafers by LbL technology exhibit favourable antibacterial properties.¹⁵ In this sense, the fabrication of polyelectrolyte multi-layer coatings via LbL technology on Mg alloys may be a promising approach. But most of the polyelectrolyte multi-layer coatings are not efficient to improve the corrosion resistance of the Mg alloys. For examples, Kumta et al. show that the ceramic conversion coating and LbL polymeric composite coatings lead to a reduction in immediate and long-term corrosion of the substrate.^16,17

Biodegradable polymers such as polyvinylpyrrolidone (PVP) and polyacrylic acid (PAA) can be directly used as polyelectrolytes for chemical deposition.¹⁸ In particular, water-soluble and amphiphilic PVP has been used in a wide variety of applications such as medicine, pharmacology and cosmetics because of its good biocompatibility.^19,20 In addition, PAA is a weak polyacid that features a carboxylic acid functional group on each repeating unit. The carboxyl groups of the grafted PAA chains are capable of interacting with a variety of chemical functionalities (e.g., polyelectrolytes or biomolecules), exhibiting a large surface area. Thus, PAA has great potential application in cross-linked coatings.^21,22 These combination may be enhanced the corrosion resistance of the Mg alloys efficiently.

To date, amorphous coatings on Mg alloys have attracted scientific interest because of their outstanding mechanical behaviour and good corrosion resistance.²³ Amorphous coatings can be prepared by means of chemical or electroless deposition as well as sputtering. Ni–P amorphous coatings, plated on AZ31 Mg alloy via the electroless plating technique, offer higher corrosion resistance than the substrate alloy.²⁴ Additionally, Al–Ni–Mm–Fe amorphous and nanocrystalline composite coatings, prepared on AZ91 Mg alloy by high velocity arc spraying, exhibit a five-fold hardness increase compared to the substrate and good abrasion resistance.²⁵ Cu-Based amorphous composite coatings, fabricated on AZ91D Mg alloy by laser cladding, exhibit excellent wear resistance and a lower corrosion current density than the Mg alloy substrates.²⁶

Thus, amorphous coatings are potentially useful for as biomedical materials. Unfortunately, the above-mentioned coatings are not suitable for biomedical applications because the Al, Ni and Cu contents are potentially harmful to human tissues. Additionally, the corrosion resistance of the above amorphous coatings on Mg alloys is far insufficient for practical applications. Furthermore, the surfaces of these kinds of metallic amorphous coatings are too hard and inappropriate for stent coatings, which require high elasticity. Therefore, it is necessary to develop a soft amorphous coating on stents with good elasticity, corrosion resistance and biocompatibility.

Interestingly, functional silanes have been used to modify stainless steels and Ti alloys to provide surface functionality for post-chemical anchoring of biomolecules (e.g., collagen, heparin and fibronectin) to improve the corrosion resistance and biocompatibility of metallic implants. So far, a considerable number of research has been focused on silane-modified coatings on Mg alloys.^27,28 The hydrophobic outer layer of a silane coating enriched with Si–O–Si bonds lowers the water penetration rate and consequently enhances the corrosion performance.²⁹

The purpose of the study aims to improve the corrosion resistance and durability of coatings on Mg-based cardiovascular stents by developing a novel composite coating with polyelectrolyte multi-layers and soft silica film.

2 Experimental

2.1 Materials and chemicals

The substrate used was an as-extruded Mg alloy AZ31 with nominal chemical composition (wt%): Al 2.5–3.0, Zn 0.7–1.3, Mn > 0.20 and the balance Mg. PVP ((C₆H₉NO)_n, MW = 40 000), PAA ((C₃H₄O₂)_n, MW = 800–1000), polyethylenimine or polyaziridine (PEI, (C₂H₅N)₁₄, MW = 600), and FAS (fluoroalkyl-silane, C₁₃H₁₃SiO₃F₁₇, MW = 568.3, 97.0%) were purchased from Qingdao Jingke Chemical Reagent Co, Ltd., China. The structures of PEI, PAA and PVP together with FAS are given in Fig. 1.

Fig. 1 Schematic representation of the preparation of the SiO₂/(PVP/PAA)₅ film.

2.2 Alkaline treatment

The substrates were cut into squares with dimensions of 20 mm × 20 mm × 3 mm, ground with SiC sandpaper from 400 to 2500 grit, washed with a solution of de-ionized (DI) water and alcohol and dried with warm air. The polished substrates were soaked in a 5 M NaOH solution at 60 °C for 2 h (ref. 16 and 17) and then cleaned thoroughly with DI water and dried at 80 °C for 1 h.

2.3 Coating preparation

The preparation process is schematically illustrated in Fig. 1. The LbL coatings were prepared via a dip-coating method; for each coating the substrates were dipped into the polymeric solution, incubated for 5 min and dried in air. Layers were generated in the following sequence: “ABCBCBCBCBC”. Solutions A and C contained the cationic polymers PEI (pH 10.9) at a concentration of 10 mg mL⁻¹ in DI water and PVP (pH 3.5) at a concentration of 5 mg mL⁻¹ in DI water. Solution B consisted of the anionic polymer PAA (pH 7.0) at a concentration of 10 mg mL⁻¹ in DI water. Five deposition cycles were performed to obtain (PVP/PAA)₅-modified substrates. The (PVP/PAA)₅ samples were then dipped in FAS solution (FAS : methanol : DI water = 1 : 10 : 90) for 1 h. The FAS-modified (PVP/PAA)₅ samples were heat treated at 120 °C for 24 h, yielding the SiO₂/(PVP/PAA)₅ samples.

2.4 Surface analysis

The surface morphology of the coatings was investigated using field emission scanning electron microscopy (FE-SEM, Nova Nano SEM 450, Netherlands). The chemical bonding of the coatings was confirmed by means of Fourier transform infrared spectroscopy (FT-IR, Nicolet 380, Thermo electron, US) and X-ray photoelectron spectroscopy (XPS, ESCALAB250, Thermo electron, US). The crystal structures of the samples were examined by X-ray diffraction (XRD, Rigaku D/MAX2500PC, Japan).

2.5 Corrosion testing

Potentiodynamic polarisation and electrochemical impedance spectroscopy (EIS) tests were performed on an electrochemical potentiostat (PAR Model 2273, Princeton Applied Research, US) in Hank's balanced salt solution (HBSS: 8.0 g L⁻¹ NaCl, 0.4 g L⁻¹ KCl, 0.14 g L⁻¹ CaCl₂, 0.1 g L⁻¹ MgCl₂·6H₂O, 0.35 g L⁻¹ NaHCO₃, 1.0 g L⁻¹ C₆H₆O₆ (glucose), 0.06 g L⁻¹ MgSO₄·7H₂O, 0.06 g L⁻¹ KH₂PO₄, 0.06 g L⁻¹ Na₂HPO₄·12H₂O). All electrochemical tests were conducted in a three-electrode configuration consisting of the sample as the working electrode (1 cm²), a platinum plate as the counter electrode and a saturated calomel electrode (SCE) as the reference electrode. Polarization curves were recorded at a sweep rate of 1 mV s⁻¹. The electrochemical parameters (open circuit potential (E_corr), corrosion current density (I_corr) and Tafel slopes) were fitted using the PowerSuite software. I_corr (mA cm⁻²) can be converted into the corrosion rate, P_i (mm per year) through:³⁰

P_i = 22.85I_corr (1)

Hydrogen evolution testing was performed according to the literature.³¹ The hydrogen evolution rate (HER), v_H (mL cm⁻² h⁻¹), is expressed as

v_H = V/(st) (2)

where V is the volume of hydrogen release, s and t are the area of the sample and immersion time, respectively. The HER, v_H (mL cm⁻² per day) can be converted into the corrosion rate, P_H (mm per year)³⁰ via:

P_H = 2.279v_H (3)

or

P_H = 54.696v_H (4)

when v_H is expressed in units of mL cm⁻² h⁻¹.

EIS measurements were acquired from 10⁵ Hz to 10⁻² Hz using a 5 mV amplitude perturbation. Hydrogen evolution was measured by placing the substrates in HBSS at 37 °C for 10 days with full surface exposure. Tests were conducted in triplicate for each condition.

2.6 Scratch testing

The adhesion strength and elastic modulus of the coatings were characterized using an MML Nanotest system with a Rockwell diamond probe with tip diameter of 25 μm. The nanoscratch test was performed at a scan velocity of 5 μm s⁻¹ by linearly increasing the load to 3 N, with an increased velocity of 10 mN s⁻¹ applied after 50 μm displacement, until the total scratch length reached 300 μm. An image of the scratch was recorded by a 3D optical profiler (Zeta-20).

3 Results

3.1 SEM morphologies

The SEM image in Fig. 2a indicates that the naturally air-formed oxide film on the as-prepared AZ31 substrate has a fish-scale-like nanostructure. In contrast, the (PVP/PAA)₅ film exhibits compact and homogeneous hexagonal nanoplates (Fig. 2b) with sizes of 30–60 nm and pore defects.⁹ The SiO₂/(PVP/PAA)₅ film displays a continuous, smooth and defect-free morphology (Fig. 2c) as a result of the amorphous SiO₂ formation, which was further confirmed by the following investigations.

Fig. 2 SEM images of the (a) AZ31 substrate, (b) (PVP/PAA)₅ and (c) SiO₂/(PVP/PAA)₅ coatings.

Cross-sectional images of the SiO₂/(PVP/PAA)₅ film are shown in ESI Fig. 1.† The two images display the presence of a dense and uniform SiO₂ coating on the substrate. The coating, with a thickness of approximately 3.19 ± 0.41 μm, exhibits an unambiguous Si signal (Fig. 3b). Therefore, the SiO₂ coating totally covered the surface during the assembly process.

Fig. 3 (a and b) XPS survey and C_1s peak of the (PVP/PAA)₅ film, (c and d) XPS survey and Si_2p peak of the SiO₂/(PVP/PAA)₅ film, and (e and f) FT-IR spectra and XRD patterns of the (PVP/PAA)₅ and SiO₂/(PVP/PAA)₅ films.

3.2 Surface analysis

An XPS survey (Fig. 3a) of the (PVP/PAA)₅ film indicates the presence of O, Mg, C and N. The signal at 400.0 eV is assigned to the nitrogen atoms of the PVP pyrrolidone ring. For the C_1s (Fig. 3b), the peak at 288.8 eV is ascribed to the COOH group, implying the presence of PAA molecules.³² And the Mg is detected from the (PVP/PAA)₅ film by XPS due to the thickness of the multifilm is really thin.

In comparison with Fig. 3a, the Mg peaks of the XPS survey of the FAS-treated samples (Fig. 3c) disappeared, indicating that the thickness of the amorphous SiO₂ film is much greater than the detection depth of XPS. The Si_2p peak at 103.7 eV (Fig. 3d) is attributable to Si in SiO₂.^33,34 For the SiO₂/(PVP/PAA)₅ film, the absence of fluoride compounds indicates the formation of an amorphous SiO₂ film as a result of the interaction between the FAS and the hydroxyl groups derived from the (PVP/PAA)₅, the explicit reasons could be explored in the further study.¹⁷ Note that, the little fluorocarbon compound is beneficial for oxygen carriers and drug delivery.³⁵ Also, some researches aimed on the modification of fluoride to improves the corrosion resistance and biocompatibility of the Mg alloys.^36,37 The XPS results were further confirmed by means of FT-IR.

The FT-IR spectrum (Fig. 3e) of the (PVP/PAA)₅ film exhibits an absorption peak at 1718 cm⁻¹, corresponding to the carboxyl group of PAA;³⁸ the absorption peak at 1450 cm⁻¹ is attributed to the ring vibration of pyridine in PVP.¹⁵ These peaks confirm the formation of the (PVP/PAA)₅ film. For the SiO₂/(PVP/PAA)₅ film, the broad absorption between 1000 cm⁻¹ and 1200 cm⁻¹ is attributed to an asymmetric Si–O–Si stretching band and the major peak at 1080 cm⁻¹ is assigned to the silica network.³⁴ Thus, we infer that an amorphous SiO₂ film formed during the FAS self-assembly process.

The XRD patterns of the (PVP/PAA)₅ and SiO₂/(PVP/PAA)₅ films are given in Fig. 3f. Compared with the (PVP/PAA)₅ film. A broadened characteristic diffraction peak at approximately 22° was observed, suggesting the existence of an amorphous SiO₂ structure in the film.^34,39 This, taken with the above results, confirmed the formation of a amorphous SiO₂-bearing film, which may have crucially impacted the improvement of corrosion resistance of the Mg alloy.

3.3 Electrochemical results

The polarisation curves in Fig. 4 indicated that the Mg substrate was susceptible to corrosion in HBSS. The cathodic Tafel slope (β_c) was obtained from the cathodic polarization curve at 50 mV vs. SCE around the E_corr. The breakdown potential E_b in the anodic branch revealed the breakdown of the oxide film on the alloy. The I_corr (9.22 × 10⁻⁷ A cm⁻²; corrosion rate = 0.021 mm per year) of the (PVP/PAA)₅ film was reduced by an order of magnitude from that of the AZ31 substrate (1.19 × 10⁻⁵ A cm⁻²; corrosion rate = 0.272 mm per year). The I_corr of the SiO₂/(PVP/PAA)₅ film was further decreased as low as 1.26 × 10⁻⁸ A cm⁻²; corrosion rate = 2.88 × 10⁻⁴ mm per year (Table 1). Simultaneously, the SiO₂/(PVP/PAA)₅ film shifted E_corr towards the passive direction.

Fig. 4 Polarisation curves of the (a) AZ31 substrate, (b) (PVP/PAA)₅ and (c) SiO₂/(PVP/PAA)₅ films in HBSS.

Table 1 Electrochemical parameters of the polarisation curves

Samples	Ecorr (mV vs. SCE)	Icorr (A cm^−2)	βc (mV per decade)	βa (mV per decade)	chi	Eb (mV vs. SCE)
Substrate	−1487	1.19 × 10^−5	227.19	659.21	0.21	−1270
(PVP/PAA)5	−1597	9.22 × 10^−7	296.14	278.88	0.64	−1450
SiO2/(PVP/PAA)5	−1379	1.26 × 10^−8	472.77	199.77	3.67	−1302

---

It is notable that the I_corr ratio of bare AZ31 : SiO₂/(PVP/PAA)₅-coated Mg is 944, indicating the good protection of the SiO₂/(PVP/PAA)₅ film. Similar results were determined in our previous studies. The I_corr ratio of the AZ31 : LDH-coated alloy is 466.⁴⁰ The I_corr ratio of the AZ31 substrate : silane-modified hydroxide zinc carbonate film on AZ31 alloy is approximately 1000 (2.64 × 10⁻⁵ A cm⁻²/2.61 × 10⁻⁸ A cm⁻²).²⁹

The good corrosion-resistant performance of the SiO₂/(PVP/PAA)₅ film can be further confirmed by EIS. The EIS spectra of the samples as a function of immersion time in HBSS at 37 °C are shown in Fig. 5. For the (PVP/PAA)₅ film, two semicircles merged into one large semicircle, corresponding to an increase in impedance and indicating an improvement in corrosion resistance. Additionally, the low-frequency impedance modulus |Z| is one of the parameters used to evaluate the corrosion resistance of different samples in Bode plots (Fig. 5c). A larger |Z| indicates better corrosion protection performance.^41,42 The |Z| increased for the coated AZ31 samples, indicating the beneficial effect of the LbL-assembled films on the corrosion resistance of the Mg alloy. The FAS-modified sample had the highest |Z| value, demonstrating that the SiO₂/(PVP/PAA)₅ film greatly enhanced the corrosion resistance of the alloy. The |Z| in Bode plots confirms the observed Nyquist plot results. The results are also in pronounced agreement with the I_corr of the polarisation curves (Fig. 4) and are further supported by the hydrogen evolution studies (Section 3.4).

Fig. 5 (a) Nyquist plots of the samples with and without an immersion of 240 h in HBSS, (b) the high magnitude of the square frame in (a), (c) Bode plots of the samples with and without an immersion of 240 h in HBSS, and equivalent circuits for EIS: (d) the AZ31 substrate and (PVP/PAA)₅ film without an immersion of 240 h (e) the AZ31 substrate and (PVP/PAA)₅ film with an immersion of 240 h (f) the SiO₂/(PVP/PAA)₅ films with and without an immersion of 240 h in HBSS.

These experimental data could be accurately fitted to the equivalent circuit R_s(Q₁(R_c1(CR_ct))) and R_s(Q₁(R_c1(Q₂(R_c2(CR_ct))))) (Fig. 5d and e). However, for the SiO₂/(PVP/PAA)₅ film, a semicircle capacitive loop in the first high frequency region corresponds to the layer of amorphous SiO₂; the large second loop corresponds to the LbL-assembled (PVP/PAA)₅ films, which may be sealed with the amorphous SiO₂. The inductive loop in the low frequency region implies the delamination of the amorphous SiO₂ film. The experimental data can be accurately fitted to the equivalent circuit R_s(Q₁(R_c1(Q₂(R_ct(C(R_LL)))))) (Fig. 5f).

In these models, R_s represents the solution resistance between the reference electrode and working electrode, R_ct is the charge transfer resistance relating to the electrochemical reaction, and R_c is the coating resistance (virtual pore resistance) paralleled with a constant phase element (CPE, Y_CPE(ω) = 1/Z_CPE = Q_a(jω)ⁿ).⁴³ Based on the proposed equivalent circuit models and the properties of the composite coating, EIS curves are best fitted, and the corresponding values of the equivalent circuit parameters are listed in Table 2. And according to MAO coating on AZ31 substrate, the large values of L are possible.⁴²

Table 2 Parameter values of the electrical equivalent circuits of the samples with and without an immersion of 240 h

Samples	Rs (Ω cm^2)	Q1 (Ω^−1 s^n cm^−2)	n1	Rc1 (Ω cm^2)	Q2 (Ω^−1 s^n cm^−2)	n2	Rct (Ω cm^2)	C (F cm^−2)	Rc2 (Ω cm^2)	RL (Ω cm^2)	L (H cm^2)
Substrate	71.72	8.71 × 10^−6	0.86	—	—	—	249.90	9.44 × 10^−7	1080	—	—
(PVP/PAA)5	82.75	3.40 × 10^−6	0.81	—	—	—	3727	1.17 × 10^−4	1840	—	—
SiO2/(PVP/PAA)5	99.98	1.16 × 10^−8	0.88	5319	3.37 × 10^−7	0.58	6.54 × 10^4	2.47 × 10^−9	—	9.89 × 10^5	4160
Substrate-240 h	54.29	7.42 × 10^−6	0.69	208.80	1.79 × 10^−5	0.70	1994	2.16 × 10^−4	1705	—	—
(PVP/PAA)5-240 h	63.44	6.26 × 10^−6	0.51	2946	3.93 × 10^−6	0.79	1.27 × 10^4	7.34 × 10^−7	4.52 × 10^4	—	—
SiO2/(PVP/PAA)5-240 h	83.85	3.74 × 10^−7	0.48	3.99 × 10^4	1.01 × 10^−8	0.88	1.67 × 10^4	8.93 × 10^−9	—	7.41 × 10^5	8384

---

Generally, a higher R_ct value implies a lower dissolution rate. The lower R_ct of the Mg substrate (approximately 0.25 kΩ) suggests a high corrosion rate. The increased value of R_ct (approximately 3.73 kΩ) for the LbL-assembled films emphasizes the beneficial effect of the (PVP/PAA)₅ film. Moreover, the R_ct value for the SiO₂/(PVP/PAA)₅ film markedly increased to approximately 65.44 kΩ, demonstrating that the SiO₂/(PVP/PAA)₅ coating greatly improved the corrosion resistance of the alloy.^44,45

The (PVP/PAA)₅ films formed a corrosion system that consisted of three phases and two interfaces (HBSS solution/corrosion product layer/LbL-assembled (PVP/PAA)₅ film) after 240 h immersion in HBSS. The R_c1 indicated the formation of a partially covered corrosion product layer on the sample surfaces. The R_c2 represents the resistance of the LbL-assembled film. Both R_c1 and R_c2 significantly increased with the immersion time because of the formation of a protective corrosion product layer. The R_ct for the (PVP/PAA)₅ films notably improved from 3.73 kΩ cm² to 12.70 kΩ cm², demonstrating a significant improvement in corrosion resistance. For the SiO₂/(PVP/PAA)₅ films, an increase in R_c1 reveals the hydrolysation of the amorphous SiO₂ coating and the formation of the corrosion product layer. The inductive loop, which is best fitted as an equivalent inductance L in the equivalent circuits, is ascribed to the continuous peeling-off of the amorphous SiO₂. The R_ct scenario is supported by a decrease from 65.44 kΩ cm² to 16.73 kΩ cm², demonstrating the good corrosion resistance of the composite coating during a longer immersion time. Interestingly, it can be observed that n₁ for (PVP/PAA)₅ and SiO₂/(PVP/PAA)₅ are close to 0.85, indicating a capacitive behaviour. With an immersion of 240 h, n₁ for (PVP/PAA)₅-240 h and SiO₂/(PVP/PAA)₅-240 h decreased close to 0.5, which may indicate a diffusion process. These results may be ascribed to the changes of micro-environment between the HBSS solution and the modificated surface during the corrosion process, which resulted in the peeling-off of the amorphous SiO₂ and the penetration of the caustic ion.

3.4 Hydrogen evolution

Interestingly, the HER (Fig. 6) as a function of immersion time exhibited two stages: an initial increase, corresponding to the dissolution of the substrate and the degradation of the coated samples, and a subsequent decrease caused by the formation of a corrosion products such as phosphate precipitates that partially protected the substrate as supported in the next section. The HER values of the materials can be ranked in increasing order as follows: SiO₂/(PVP/PAA)₅ < (PVP/PAA)₅ < Mg alloy.

Fig. 6 HER as a function of immersion time for the AZ31 substrate, (PVP/PAA)₅ and SiO₂/(PVP/PAA)₅ films in HBSS.

After an initial immersion of 11 h of the AZ31 substrate, the HER increased up to 0.01 mL cm⁻² h⁻¹ or 0.547 mm per year, greater than the corrosion rate of 0.272 mm per year derived from the polarisation curves (Fig. 4). The HER is 0.00315 mL cm⁻² h⁻¹ (0.172 mm per year) for the (PVP/PAA)₅ coating, and 0.00173 mL cm⁻² h⁻¹ (0.0945 mm per year) for the SiO₂/(PVP/PAA)₅ coating. These rates reveal a 5.8-fold increase in corrosion resistance of the SiO₂/(PVP/PAA)₅ coating compared with the substrate. Notably, at an immersion of 61 h, the HER of the substrate peaked (0.02672 mL cm⁻² h⁻¹ (1.46 mm per year)), and the value of the (PVP/PAA)₅ coating was 0.00902 mL cm⁻² h⁻¹ (0.49 mm per year) and of the SiO₂/(PVP/PAA)₅ coating was 0.00373 mL cm⁻² h⁻¹ (0.20 mm per year). The corrosion resistance gap between the SiO₂/(PVP/PAA)₅ coating and its substrate increased to 7.3-fold.

The HER of the substrate is reasonable compared with the results (approximately 0.3 mm per year) for pure Mg.³⁰ Basically, the HER, similar to the weight loss rate, actually designates the corrosion rate of Mg alloys during the immersion period. In fact, the I_corr of Mg alloys fitted by Tafel extrapolation from polarisation curves cannot reflect the true corrosion rate because of anodic hydrogen evolution and dissolution of Mg.⁴⁶ The phenomenon is related to the negative difference effect (NDE).^47,48 In addition, the formation of the corrosion product Mg(OH)₂ leads to a marked decrease in the HER of the substrate during the immersion period. The longer the substrate is immersed, the more clearly the HER decreases. As a result, the I_corr do not agree completely with the corrosion rates evaluated from hydrogen evolution.⁴⁹

3.5 Nanoscratch behaviour of the composite coatings

The adhesion strength of the coatings was investigated by nanoscratch tests, which can be used to measure critical load during adhesive failure of the coating/substrate system.⁵⁰ ESI Fig. 2† illustrates the relationships between the depth, load and sliding displacement, which consists of a ramped-load scratch topography (curve 1) with a progressively increasing load from zero to the maximum value shown by a proportional loading plot (curve 2). The critical point of coating failure is characterized by an abrupt change with a continual fluctuation in the displacement of the probe, and the load at this point is defined as the critical load for adhesive failure, which can be used to quantify the adhesive force between the coating and the substrate.⁵⁰ The critical load represents the comprehensive capability of the coating to withstand delamination depending on the characteristics of the coating, substrate and interface.

The 3D optical profilometry image of the nanoscratch on the SiO₂/(PVP/PAA)₅ coating is shown in ESI Fig. 2a.† The critical load of the SiO₂/(PVP/PAA)₅ coating is 199.92 ± 16 mN (ESI Fig. 2b†), representing the adhesion between the composite coating and the substrate. The adhesion strength of the coating is much less than 692.9 mN for the Al₂O₃ sol–gel coating on the AZ31 alloy and 1020 mN for the ceramic MAO coating on the AZ91 alloy because of the presence of a metallic bond between the MAO coating and its substrate.^50,51 The spinodal points in the scratch topography represent the bottom of the coating and indicate that the SiO₂/(PVP/PAA)₅ coating has a thickness of approximately 3.16 ± 0.8 μm, which is agreed with the value obtained by SEM (ESI Fig. 1a†). Nanohardness (H) and Young's modulus (E) values calculated as a function of indentation depth are presented in Table 3. The E (9.88 ± 0.05 GPa) of the coating is twice that of hard rubbers (5 GPa) and almost one half of that of Pb metal (16–18 GPa).⁵² These results imply that the coating is notably soft and compliant in comparison with the MAO coating on the AZ91 alloy with a hardness of 4.2 GPa and an elastic modulus of 71.1 GPa.⁵¹

Table 3 Nanohardness (H), Young's modulus (E), the H/E ratio, and the parameter H³/E² for the SiO₂/(PVP/PAA)₅ coating

Sample	H (GPa)	E (GPa)	H/E	H^3/E^2 (GPa)
SiO2/(PVP/PAA)5	0.33 ± 0.04	9.88 ± 0.05	3.34 × 10^−2	3.68 × 10^−4

---

4 Discussion

4.1 Corrosion resistance

The I_corr values were compared with those reported in the literature for similar films (ESI Fig. 3†).^16,25,27 It is apparent that the single LbL coating and the single silane coating improve the corrosion resistance to a limited degree. Although the effect of the Ni coating is satisfactory, it is not suitable for biomedical use. This result indicates that the corrosion resistance of the SiO₂/(PVP/PAA)₅ composite coating is better than that of other films such as biodegradable polymers and amorphous silica.

4.2 Self-assembly mechanism of (PVP/PAA)₅ and SiO₂/(PVP/PAA)₅ coatings

The deposition process of (PVP/PAA)₅ and SiO₂/(PVP/PAA)₅ coatings on the Mg alloy substrate are schematically illustrated in ESI Fig. 4.†

(I) Mg²⁺ ions are readily released from the Mg substrate into the primary coating solution, simultaneously evolving H₂:

Mg → Mg²⁺ + 2e⁻ (5)

2H₂O + 2e⁻ → 2OH⁻ + H₂↑ (6)

Mg + 2H₂O → Mg(OH)₂ + H₂↑ (7)

(II) PEI used as an insulating polymer exhibits good adhesion to the substrate, assisting in molecular electrostatic self-assembly for multi-layer formation and providing a physical barrier for electrons transfer.⁴⁴ As a result, it is possible to obtain a robust PEI coating on the Mg substrate to increase positive charges.

(III) In the primary PAA solution (pH 7.0), the carboxyl groups are negatively charged, promoting the adherence of PAA and PEI via an electrostatic interaction. Additionally, PVP, as a water-soluble inertial polymer, is positively charged at a pH value of 3.5. Thus, the PVP molecules link with the PAA molecules through electrostatic interactions (ESI Fig. 4a†).

(IV) Silicon is partially positively charged when bonded to carbon as a result of its low electron negativity. These silicon atoms are thus easily attacked by OH⁻, leading to the formation of silanol groups, Si–O–Si networks and an amorphous SiO₂ structure through dehydration condensation (ESI Fig. 4b†).²⁷ The solution pH increases with decreasing H⁺ ion concentration.

≡Si–O–R + H₂O → ≡Si–OH + ROH (8)

≡Si–OH + 3/2H₂O → Si(OH)₄ (9)

As a polar molecule, Si(OH)₄ polarizes the water molecules such that a highly dispersed amorphous SiO₂ system forms.

Si(OH)₄ → mSiO₂·nH₂O (10)

4.3 Corrosion mechanism

The FT-IR spectra of the substrate, (PVP/PAA)₅, and SiO₂/(PVP/PAA)₅ samples immersed in HBSS for 40, 120 and 240 h are shown in Fig. 7. The peaks at 576 cm⁻¹ and 1058 cm⁻¹ are ascribed to the deformation vibration of PO₄³⁻.⁵³ Increasing intensity of the absorption peaks indicates that the (PVP/PAA)₅ coating promotes the formation of phosphates over time. Moreover, the peak at 1102 cm⁻¹ for the SiO₂/(PVP/PAA)₅ film designates the presence of amorphous SiO₂.³⁴ The decrease in intensity confirms the attack from HBSS.

Fig. 7 FT-IR spectra of the (a) AZ31 substrate, (b) (PVP/PAA)₅ and (c) SiO₂/(PVP/PAA)₅ films in HBSS after an 40, 120 and 240 h immersion.

Fig. 8 shows the XRD patterns of the substrate and the (PVP/PAA)₅ and SiO₂/(PVP/PAA)₅ coatings immersed in HBSS for 10 days. Phosphates such as Ca₁₀(PO₄)₆(OH)₂ (HA) and Ca₃(PO₄)₂ formed on the substrate and the (PVP/PAA)₅ coating. The gradual increase in the peak intensities of HA and Ca₃(PO₄)₂ indicates that (PVP/PAA)₅ promotes the formation of the Ca–P precipitates. Additionally, remaining amorphous SiO₂ on the SiO₂/(PVP/PAA)₅ coating implies that the amorphous SiO₂ film has a passive effect on the improvement in corrosion resistance.

Fig. 8 XRD patterns of the AZ31 substrate, (PVP/PAA)₅ and SiO₂/(PVP/PAA)₅ films in HBSS after 240 h.

Fig. 9 illustrates the SEM morphologies and corresponding EDS spectra of the substrate (a–f), the (PVP/PAA)₅ (g–l) and the SiO₂/(PVP/PAA)₅ (m–r) coatings after immersion in HBSS for 40, 120 and 240 h. For the substrate, phosphate corrosion products precipitate (Table 4) and a dry river-bed morphology formed on the surface (Fig. 9a, c and e). However, this attack was comparatively minor for the (PVP/PAA)₅ film (Fig. 9g, i and k). Interestingly, there was no visible attack on the surface of the SiO₂/(PVP/PAA)₅ film (Fig. 9m, o and q). The EDS spectra of the corrosion surfaces on all three materials indicate the presence of C, O, Zn, Mg, Al, Ca, P, S and Cl. In particular, the Ca and P contents increased with increasing immersion time, implying the existence of compounds such as HA and Ca₃(PO₄)₂, in agreement with the FT-IR and XRD results (Fig. 7 and 8). For the substrate and the (PVP/PAA)₅ film, the high Ca and P contents (Table 4) confirmed that PVP and PAA promoted the formation of the Ca–P precipitates. Nevertheless, Si was observed on the surface of the SiO₂/(PVP/PAA)₅ film (Fig. 9n, p and r), and low Ca and P contents indicated that the amorphous SiO₂ film might not be subject to attack from the solution.

Fig. 9 FE-SEM images and corresponding EDS spectra of the (a–f) AZ31 substrate, (g–l) (PVP/PAA)₅ and (m–r) SiO₂/(PVP/PAA)₅ film in HBSS after an immersion of 40, 120 and 240 h.

Table 4 Contents of Ca and P elements after various immersion time in HBSS, wt%

Time (h)	Substrate		(PVP/PAA)5		SiO2/(PVP/PAA)5
	Ca	P	Ca	P	Ca	P
40	6.47	8.11	13.43	14.68	0.51	0.64
120	9.01	14.04	15.25	16.38	0.60	0.94
240	16.63	15.71	22.96	19.67	5.81	4.98

---

The corrosion mechanism of the SiO₂/(PVP/PAA)₅ coating is schematically illustrated in ESI Fig. 5.† In HBSS, the hydrogen phosphate (HPO₄²⁻) and dihydrogen phosphate (H₂PO₄⁻) ions were transformed into phosphate (PO₄³⁻) ions through the following ionic reactions:⁵⁴

H₂PO₄⁻ + OH⁻ → HPO₄²⁻ + H₂O (11)

HPO₄²⁻ + OH⁻ → PO₄³⁻ + H₂O (12)

Then, Ca²⁺ and PO₄³⁻ ions led to the deposition of Ca₃(PO₄)₂ precipitates by the following reactions:

3Ca²⁺ + 2PO₄³⁻ → Ca₃(PO₄)₂ (13)

The SiO₂/(PVP/PAA)₅ film had much greater corrosion resistance because the dense and amorphous SiO₂ coating could block migration access of H₂O and chlorides from the Mg substrate and inhibit the attack from the solution. Additionally, amorphous SiO₂ could be delaminated in HBSS (Fig. 5).

Additionally, the polyelectrolytes easily led to calcification in a solution of sodium ions.⁵⁵ The lactam group in PVP is strongly polar, causing the formation of the Ca²⁺–PVP coordination compound, which is easier to combine with PO₄³⁻ in HBSS, thus promoting the formation of crystalline Ca–P precipitates. Furthermore, the polymer chains have a stronger affinity with Ca–P products, resulting in the formation of a new corrosion-resistant phosphate film on the Mg substrate. The observed increases in the intensity of PO₄³⁻ in FT-IR (Fig. 7) and Ca–P precipitates in XRD (Fig. 8) on the substrate also illustrated the promotion of calcification for the (PVP/PAA)₅ film:

10Ca²⁺ + 6PO₄³⁻ + 2OH⁻ → Ca₁₀(PO₄)₆(OH)₂ (14)

Therefore, the SiO₂/(PVP/PAA)₅ film on the AZ31 alloy displays good corrosion resistance.

5 Conclusions

(1) A soft SiO₂/(PVP/PAA)₅ film has been successfully prepared by LbL assembly. The film has a critical load of approximately 199.92 ± 16 mN, a thickness of approximately 3.16 ± 0.80 μm, a hardness of 0.33 ± 0.04 GPa and a Young's elastic modulus of 9.88 ± 0.05 GPa.

(2) EDS, FT-IR, XPS and XRD results confirmed the formation of (PVP/PAA)₅ and amorphous SiO₂ films on the surface. The results of the electrochemical and hydrogen evolution tests demonstrate that the LbL assembled SiO₂/(PVP/PAA)₅ film delivers good corrosion protection as a result of the formation of an amorphous silica layer.

(3) PVP and PAA effectively promote the formation of Ca–P precipitants.

(4) The LbL assembled SiO₂/(PVP/PAA)₅ film may be a promising candidate for Mg-based cardiovascular stents.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (51571134), the Scientific Research Foundation of Shandong University of Science and Technology (SDUST) for Recruited Talents (2013RCJJ006) and the SDUST Research Fund (2014TDJH104). We thank Dr Zhenlin Wang at Chongqing University of Technology for his assistance with the scratch test.

Notes and references

1. Y. F. Zheng, X. N. Gu and F. Witte, Mater. Sci. Eng., R, 2014, 77, 1–34.
2. R. Zeng, W. Dietzel, F. Witte, N. Hort and C. Blawert, Adv. Eng. Mater., 2008, 10, B3–B14.
3. M. Moravej and D. Mantovani, Int. J. Mol. Sci., 2011, 12, 4250–4270.
4. A. Atrens, M. Liu and N. I. Z. Abidin, Mater. Sci. Eng., B, 2011, 176, 1609–1636.
5. R. C. Zeng, L. Y. Cui, K. Jiang, R. Liu, B. D. Zhao and Y. F. Zheng, ACS Appl. Mater. Interfaces, 2016, 8, 10014–10028.
6. R. C. Zeng, W. C. Qi, Y. W. Song, Q. K. He, H. Z. Cui and E. H. Han, Front. Mater. Sci., 2014, 8, 343–353.
7. J. Chen, Y. Song, D. Shan and E. H. Han, J. Mater. Sci. Technol., 2015, 31, 384–390.
8. J. K. Lin, J. Y. Uan, C. P. Wu and H. H. Huang, J. Mater. Chem., 2011, 21, 5011–5020.
9. R. C. Zeng, Z. G. Liu, F. Zhang, S. Q. Li, H. Z. Cui and E. H. Han, J. Mater. Chem. A, 2014, 2, 13049–13057.
10. R. C. Zeng, F. Zhang, Z. D. Lan, H. Z. Cui and E. H. Han, Corros. Sci., 2014, 88, 452–459.
11. K. Bai, Y. Zhang, Z. Fu, C. Zhang, X. Cui, E. Meng, S. Guan and J. Hu, Mater. Lett., 2012, 73, 59–61.
12. Q. Wang, L. Tan, W. Xu, B. Zhang and K. Yang, Mater. Sci. Eng., B, 2011, 176, 1718–1726.
13. G. Decher, J. D. Hong, J. Schmitt, G. Decher, J. D. Hong and J. Schmitt, Thin Solid Films, 1992, 210–211, 831–835.
14. G. Decher, Science, 1997, 277, 1232–1237.
15. B. L. Wang, K. F. Ren, H. Chang, J. L. Wang and J. Ji, ACS Appl. Mater. Interfaces, 2013, 5, 4136–4143.
16. O. Nicole, L. Boeun, E. Nathan, C. Benjamin, K. Sangeetha, R. Abhijit and P. N. Kumta, Acta Biomater., 2013, 9, 8704–8713.
17. S. Kunjukunju, A. Roy, M. Ramanathan, B. Lee, J. E. Candiello and P. N. Kumta, Acta Biomater., 2013, 9, 8690–8703.
18. H. Y. Zhang, Y. Fu, D. Wang, L. Y. Wang, Z. Q. Wang and X. Zhang, Langmuir, 2003, 19, 8497–8502.
19. H. A. Patel, S. Shah, D. O. Shah and P. A. Joshi, Appl. Clay Sci., 2011, 51, 126–130.
20. H. A. A. El-Rehim, E. A. Hegazy, F. H. Khalil and N. A. Hamed, Nucl. Instrum. Methods Phys. Res., Sect. B, 2007, 254, 105–112.
21. A. Solís-Gómez, M. G. Neira-Velázquez, J. Morales, M. A. Sánchez-Castillo and E. Pérez, Colloids Surf., A, 2014, 451, 96–105.
22. Z. Jiang, Z. J. Jiang, Y. Shi and Y. Meng, Appl. Surf. Sci., 2010, 256, 6473–6479.
23. M. Evan and X. Jian, Nat. Mater., 2009, 8, 855–857.
24. X. Shu, Y. Wang, C. Liu, A. Aljaafari and W. Gao, Surf. Coat. Technol., 2015, 261, 161–166.
25. Z. B. Zhang, X. B. Liang, Y. X. Chen and B. S. Xu, Phys. Procedia, 2013, 50, 156–162.
26. K. J. Huang, Y. Li, C. S. Wang, C. S. Xie and C. R. Zhou, Trans. Nonferrous Met. Soc. China, 2010, 20, 1351–1355.
27. X. Liu, Z. Yue, T. Romeo, J. Weber, T. Scheuermann, S. Moulton and G. Wallace, Acta Biomater., 2013, 9, 8671–8677.
28. L. Liu, P. Li, Y. Zou, K. Luo, F. Zhang, R. C. Zeng and S. Li, Surf. Coat. Technol., 2016, 291, 7–14.
29. R. C. Zeng, L. J. Liu, T. T. Pang, F. Zhang, W. W. Zhang, S. Q. Li, H. Z. Cui and E. H. Han, Acta Metall. Sin. (Engl. Lett.), 2015, 28, 373–380.
30. Z. Shi and A. Atrens, Corros. Sci., 2011, 53, 226–246.
31. R. C. Zeng, Y. Hu, S. K. Guan, H. Z. Cui and E. H. Han, Corros. Sci., 2014, 86, 171–182.
32. J. Garcia-Torres, D. Sylla, L. Molina, E. Crespo, J. Mota and L. Bautista, Appl. Surf. Sci., 2014, 305, 292–300.
33. P. Kościelniak, J. Mazur, J. Henek, M. Kwoka, Ł. Pawela and J. Szuber, Thin Solid Films, 2011, 520, 927–931.
34. S. Yu, A. Ito, R. Tu and T. Goto, Surf. Coat. Technol., 2011, 205, 2818–2822.
35. M. P. Krafft, Adv. Drug Delivery Rev., 2001, 47, 209–228.
36. C. M. Weber, R. Eifler, J. M. Seitz, H. J. Maier, J. Reifenrath, T. Lenarz and M. Durisin, Acta Biomater., 2015, 18, 249–261.
37. F. Witte, J. Fischer, J. Nellesen, C. Vogt, J. Vogt, T. Donath and F. Beckmann, Acta Biomater., 2010, 6, 1792–1799.
38. D. Yin, G. Jian, X. Yang, H. Jian, Z. Zhou and H. Yong, Adv. Powder Technol., 2014, 25, 244–249.
39. L. Xin, Q. Shi, J. Chen, W. Tang, N. Wang, Y. Liu and Y. Lin, Mater. Charact., 2012, 65, 55–61.
40. F. Zhang, Z. G. Liu, R. C. Zeng, S. Q. Li, H. Z. Cui, L. Song and E. H. Han, Surf. Coat. Technol., 2014, 258, 1152–1158.
41. M. I. Jamesh, G. Wu, Y. Zhao, D. R. Mckenzie, M. M. M. Bilek and P. K. Chu, Corros. Sci., 2015, 91, 160–184.
42. X. J. Cui, X. Z. Lin, C. H. Liu, R. S. Yang, X. W. Zheng and M. Gong, Corros. Sci., 2014, 90, 402–412.
43. P. Zoltowski, J. Electroanal. Chem., 1998, 443, 149–154.
44. A. Zomorodian, M. P. Garcia, T. M. E. Silva, J. C. S. Fernandes, M. H. Fernandes and M. F. Montemor, Acta Biomater., 2013, 9, 8660–8670.
45. T. S. Lim, H. S. Ryu and S. H. Hong, Corros. Sci., 2012, 62, 104–111.
46. D. Seifzadeh and A. R. Hollagh, J. Mater. Eng. Perform., 2014, 23, 4109–4121.
47. G. L. Song and A. Atrens, Adv. Eng. Mater., 1999, 1, 11–33.
48. R. C. Zeng, W. Ke, Y. B. Xu, E. H. Han and Z. Y. Zhu, Acta Metall. Sin. (Engl. Lett.), 2001, 37, 673–685.
49. Z. Shi, M. Liu and A. Atrens, Corros. Sci., 2010, 52, 579–588.
50. Z. L. Wang and R. C. Zeng, Trans. Nonferrous Met. Soc. China, 2010, 20, 665–669.
51. J. Chen, R. C. Zeng, W. J. Huang, Z. Q. Zheng, Z. L. Wang and J. Wang, Trans. Nonferrous Met. Soc. China, 2008, 18, 361–364.
52. X. C. G. X. Hu and Y. H. Rong, Fundamentals of Materials Science, Shanghai Jiaotong University Press, 2010, pp. 168–170.
53. R. C. Zeng, L. Wang, D. F. Zhang, H. Z. Cui and E. H. Han, Front. Mater. Sci., 2014, 8, 230–243.
54. X. Zhang, X. W. Li, J. G. Li and X. D. Sun, ACS Appl. Mater. Interfaces, 2014, 6, 513–525.
55. P. Ander and M. Kardan, Macromolecules, 1984, 17, 2431–2436.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c6ra08613f

---