3.1. Characterisation
For both materials, dendritic α-Mg grains were revealed (
Figure 1) with the second phase distributed along the grain boundaries in the form of a discontinuous network and decorated the interdendritic regions (bright contrast regions).
The EDS analysis of Areas 1 and 4 correspond to the matrix of Mg0.6Ca and Mg0.6Ca2Ag respectively (
Table 2). Points 2 and 3 (
Figure 1b) were located in the second phase of the Mg0.6Ca and revealed mainly a similar amount of Mg and Ca elements. In the case of the Mg0.6Ca2Ag alloy, two different EDS analyses (
Figure 1d, Points 5 and 6) showed the formation of particles with different contents of Ag.
TEM analyses of Mg0.6Ca alloy displayed the second phase (
Figure 2a) as an eutectic aggregate with lamellar morphology, probably formed by α-Mg/Mg
2Ca phases in accordance with the EDS analysis (
Table 3), the Mg–Ca phase diagram and other works on binary Mg–Ca alloys [
25,
32,
33,
34]. In addition, polygonal shape particles (
Figure 2b) containing impurities were found (Point 2,
Table 3).
In the case of the ternary alloy (Mg0.6Ca2Ag) (
Figure 3), TEM studies were not conclusive as the zones of the ion milled lamella where particles were located were too thick for generating an informative electron diffraction pattern.
In fact, the available data of TEM studies on Mg alloys containing silver are very scarce and all of them have been conducted on binary Mg–Ag alloys reporting the formation of different intermetallic compounds such as MgAg, MgAg
3 or MgAg
4 [
35,
36]. The present work revealed the formation of particles with different shapes (
Figure 3) and compositions (
Table 4), some of them were Ag-enriched (
Table 3, Point 2).
The local Volta potential difference (VPD) between constituents on a submicron scale was obtained using SKPFM. Surface potential maps and potential profile along with the SEM images of the studied area are displayed in
Figure 4 for the Mg0.6Ca and Mg0.6Ca2Ag alloys, respectively.
Figure 4a,d show the selected region of the Mg0.6Ca alloy where surface potential maps and profiles were acquired. Point 1 corresponds to the Mg
2Ca phase that shows a negative potential with respect to the α-Mg matrix (VPD of ~−50 mV). It is important to note that the available data regarding the electrochemical behaviour of Mg
2Ca is quite limited and controversial results are reported about the anodic [
37,
38] as well as cathodic behaviour [
19] of this phase, which have been mainly attributed to the differences in the composition of the alloy (e.g., presence of impurities) and matrix segregation. Points 2 and 3 (
Figure 4d) correspond to an impurity (Point 2) and to the Mg
2Ca phase (Point 3,
Table 4). As it was observed before, the latter shows an anodic behaviour (VPD ~−15 mV) compared to the matrix, whereas the presence of impurities reveals a slight cathodic behaviour with respect to the α-Mg matrix (VPD ~+20 mV) due to the presence of more noble elements such as Si, Fe or Al.
For the Mg0.6Ca2Ag alloy, the Volta potential profiles conducted in the inclusion (
Figure 5) reveal a different electrochemical response depending on the elemental composition analyzed by EDS (
Table 5). The area enriched in Ag (Ca/Ag ratio of 0.89) (
Figure 5b) revealed a cathodic behaviour (VPD ~+20 mV) in comparison with the α-Mg matrix, whereas the Ca/Ag ratio of 2.39 leads to an anodic performance with potential differences around VPD ~−27 mV. Ben et al. studied the electrochemical behaviour of magnesium–silver ternary alloys (MgZnAg) and reported a cathodic behaviour of MgAg secondary phase [
39], which appears to be similar to the behaviour of the areas enriched with Ag in Mg0.6Ca2Ag alloy of the present work.
Figure 6 shows the plan view and cross-sectional coating morphologies of Mg0.6Ca/PEO and Mg0.6Ca2Ag/PEO coatings. In both alloys the coating surface presents a typical crater-like porous morphology associated with the sites of discharge channels, gas evolution and rapid solidification phenomena [
24,
40]. In both Mg0.6Ca/PEO and Mg0.6Ca2Ag/PEO coatings the surface Ca/P ratio is relatively high (1.42 and 1.53, respectively) compared with the inner regions of the coating (
Table 6), although lower than that of biological hydroxyapatite (1.67) [
41], and both Ca and P contents increase towards the coating/electrolyte interface. The surface enrichment of biomaterials (e.g., of Ti alloys) in Ca and P is well known to improve the initial cell response [
42].
The cross-sectional images (
Figure 6c,d) reveal relatively uniform coatings with thicknesses in the range of 32–35 μm. It can be observed that both coatings are constituted by three layers. A thin barrier layer (less than 1 µm) adjacent to the substrate is mainly composed of Mg, O and F and Ag (the latter only in the case of MgCaAg,
Table 6). An intermediate region with small pores constitutes ~40% of the coating thickness. An outer, more compact, region contains a few but relatively large pores.
It is worth mentioning that EDS analysis for the MgCaAg alloy did not detect the presence of Ag in the coating surface, whereas the Ag content in the inner regions did not exceed ~0.2 at.%, suggesting that a rather limiting if any antibacterial effect can be expected at the initial stage of the implantation.
The XRD analyses of uncoated materials revealed peaks corresponding to α-Mg for both alloys (
Figure 7). In the case of uncoated Mg0.6Ca, no intermetallic/secondary phases were detected probably due to their negligible amount. Although, in our previous study of a similar alloy (cast Mg0.8Ca) a formation of Mg
2Ca phase was confirmed [
25]. Most of the diffraction peaks of Mg0.6Ca2Ag were the same as those for Mg0.6Ca except that there were small peaks in a low 2θ angles region corresponding to binary and ternary intermetallic phases as Mg
2Ca, MgAg
4, Mg
54Ag
17 [
25,
36,
43] and Ca
2MgAg
3 (
Figure 7, inset). A further systematic TEM and electron diffraction study would be necessary in order to confirm the presence of these phases.
On the other hand, PEO coatings revealed peaks corresponding to the substrate, and the formation of crystalline phases such as MgO, MgF
2, CaF
2 and Ca
5(PO
4)3F was detected. The MgO is formed due to electrolytic oxidation of the substrate and the other phases are formed due to plasma-chemical reactions between the ions of the electrolyte and the substrate inside the microdischarge channels [
44].
3.2. Hydrogen Evolution Measurement
Figure 8a,b show the hydrogen evolution volume and hydrogen evolution rate for Mg0.6Ca and Mg0.6Ca/PEO after 60 days of immersion in SBF at 37 °C.
Figure 9 shows the volume of evolved hydrogen for Mg0.6Ca2Ag and Mg0.6Ca2Ag/PEO after 4 days of immersion. As expected, the uncoated material initially exhibited a high amount of hydrogen with an evolution rate of 3.86 mL/cm
2 week after 1 week of immersion, with progressive decrease to 1.02 and 1.11 mL/cm
2 week after 6 and 8 weeks of immersion, respectively. The decrease of hydrogen evolution rate is due to the generation of a corrosion products layer, which acts as a partially protective coating.
Mg0.6Ca/PEO showed a considerably reduced hydrogen evolution rate during the first 4 weeks (~1.11 mL/cm
2 week); however, after that time both non-coated and PEO-coated materials reached a similar degradation rate. Further increase of the hydrogen evolution rate was evident after 6 weeks (1.75 mL/cm
2 week) due to the loss of protective properties of the inner PEO layer and increased electrochemical activities in the substrate/coating interface [
45,
46,
47]. The latter degradation rate corresponds to 1.9 mg/(cm
2 week) of mass loss or 11 μm/week of thickness loss.
Both Mg0.6Ca2Ag and Mg0.6Ca2Ag/PEO exhibited an extremely high degradation rate (
Figure 9), corresponding to ~60 or ~20 mL/cm
2, respectively, and the experiments were stopped after 4 days of immersion.
Figure 10 shows the macro degradation photos of the specimens. Both coated and non-coated Mg0.6Ca alloy (
Figure 10a,b) presented a uniform corrosion with similar loss of material and dimensions. Mg0.6Ca2Ag also showed a generalized, but heavily heterogeneous corrosion morphology for the non-coated specimens (
Figure 10c) and localized corrosion for the PEO-coated ones; both were found completely disintegrated after about 10 days of immersion.
Following the immersion, a heavy generalized corrosion was observed in all cases (
Figure 11) and PEO coatings were evidently detached, and, in the case of the Mg0.6Ca2Ag/PEO system, this had already occurred by the 4th day of immersion. A complete loss of adhesion is not a typical behaviour for PEO-coated Mg alloy, as was previously demonstrated by the authors in [
25], and a 40–50 μm-thick coating can be expected to remain mostly adhered even after 8 weeks of immersion with a thick corrosion product layer developing underneath it.
Some researchers of Mg–Ca systems have reported that for Ca content below 1.25% the corrosion process is driven from a general mechanism to localized corrosion due to severe electrochemical activity [
18]. Other studies suggest that when the amount of Ag in MgAg alloys is increased [
16] (1.15 mm/year for Mg2Ag to 1.43 mm/year for Mg6Ag) the Ag-containing second phase does not play an important role in microgalvanic activities, as the responsibility for these phenomena fall to the impurities [
48,
49]. However, in our case the corrosion rate was 7.78 mm/year for Mg0.6Ca2Ag and 3.64 mm/year for Mg0.6Ca2Ag/PEO. The elevated degradation rate was observed in the present study for Mg0.6Ca in comparison with other Mg–Ca systems [
25] and especially for Mg0.6Ca2Ag.
It is evident that, in both alloys, intermetallic particles disclosed some regions with active behaviour that depended on the Ca/Ag ratio in the region (
Figure 4 and
Figure 5). Such regions present galvanic micro-couples with a minimum anodic surface compared to a large cathodic surface of α-Mg causing dissolution of the intermetallic particles. It can be clearly appreciated from
Figure 1a,c that in the Mg0.6Ca2Ag alloy the grain boundary network is much more decorated with intermetallic particles than in Mg0.6Ca. Consequently, dissolution of the grain boundaries can lead to an easy fall-out and loss of the cathodic α-Mg grains or clusters of grains (
Figure 12). This anodic behavior of the particles may explain the extremely fast degradation rate of the Mg0.6Ca2Ag alloy.
3.3. Fluoride Release
Figure 13 shows progressive F
− ion release from Mg0.6Ca/PEO during 24 h of immersion in 0.9% NaCl. After 24 h of immersion, 33.757 µg/cm
2 of F
− was released, which corresponds to 1.35 ppm or 71 μM of F
− for the volume of the immersion solution. According to the solubility products,
Ksp, of fluoride containing crystalline phases in the coating, MgF
2, CaF
2, and Ca
5(PO
4)
3F (5.16 × 10
−11, 3.58 × 10
−11, and 8.6 × 10
−61, respectively [
50]), the dissolution of these compounds will produce up to a maximum of 0.88 mM of free F
−. Hence, it would be reasonable to expect a further increase of F
− with time until a solubility limit is reached. In our previous work, it was demonstrated that this limit can be reached in 12 weeks [
33]. For comparison, according to The World Health Organization (WHO), fluoride content in drinking water should not exceed 1.22 mg/L (~1 ppm), the optimal daily fluoride consumption for an adult is between 1.4–3.4 mg·day
−1, whereas the average fluorine concentration in the human blood plasma is 19 ppb [
51]. Therefore, the fluoride liberation from Mg0.6Ca/PEO in a 24-h period appears to be safely within these guidelines without a danger of intoxication. Further studies are needed in order to evaluate the potential antibacterial effect of the fluoride in the coating.