3.2. Electrochemical Measurements
The relative OCP values determined for each of the cermet and single metallic phase compositions are presented in
Table 2. The OCP values for the metallic phase steels are less negative (i.e., more noble) than the corresponding cermets, which indicates better surface passivation and protection behaviour and, consequently, corrosion resistance [
38]. The OCP values for the cermets generally become more negative with increasing steel binder content, but on occasion show a slight deviation from this trend at the highest binder content. This is particularly apparent for TiC-316 L, where the mean OCP value for samples with 30 vol. % binder is less negative than for 10 vol. % binder samples. The more noble OCP values that were generally observed at the lowest steel binder content for each cermet can be attributed to the formation of a (partially) protective surface layer. This can be anticipated to be an oxide of titanium, and further evidence in support of this is presented in a subsequent section (
Post-Corrosion Chemical Analysis). It should be noted that TiC and related ceramics behave in a broadly similar manner to pure Ti, in that they form a TiO
2 protective oxide layer [
39,
40,
41].
Typical examples of the potentiodynamic polarisation plots obtained for each of the TiC–stainless steel cermets, as well as the baseline steels, are shown in
Figure 3.
Table 3 presents the accompanying quantitative data obtained after potentiodynamic polarisation experiments.
The results of the potentiodynamic polarisation have shown that the critical current density,
icrit, and “pseudo-passive” current density,
ipass, both increase with increasing steel binder content. This response relates to the passivation tendency of the cermets. The low critical anodic current density at the peak of each of the curves indicates that the specimens passivate quickly [
42], which indicates that the greater TiC fraction at the lowest binder contents is contributing to the corrosion resistance of the cermets. Similar results have also been reported by Sacks [
9], who noted an increase in
icrit with an increase in Co content during corrosion studies of WC-Co cermets. Sutthiruangwong and Mori also reported an increase in
icrit,
icorr, and
ipass with decreasing WC content [
14], which mirrors the observations presented in
Table 4.
From both
Figure 3 and
Table 4, it can be noted that the current responses in the passive region are relatively high to be viewed as a true passive material (i.e., of the order of 10
−2 A/cm
2) [
12]. Generally, for materials exhibiting “pseudo-passivity” at high electrochemical potentials, after reaching a critical current density, there will be a slight drop in current in the passivation region, as apparent in
Figure 3. The passive current density measured for each of the cermets is approximately four orders of magnitude higher than for a true passive material (i.e., 10 µA/cm
2). The probable reason for the high current observed in the pseudo-passive region is related to the formation of a weak, cracked, and/or porous oxide, thereby allowing the penetration of the electrolyte to the cermet surface [
11]. The presence of such an oxide scale, which still inhibits diffusion to a certain extent, leads to a limitation of the current density [
12].
To determine the corrosion current density,
icorr, and potential,
Ecorr, Tafel extrapolations were performed following the potentiodynamic polarisation tests; data obtained from these analyses is presented in
Table 4. It is apparent that both
icorr and
Ecorr for the cermets increase with increasing steel binder content. In comparison, the values of
icorr for each of the single phase metallic steels is lower than for the cermets, again indicating better corrosion resistance of the metallic phase by itself.
Based on the results of Tafel extrapolation, it was observed that the corrosion rate increases with binder content for all of the cermets (
Figure 4a). It is believed that this response arises from effectively increasing the TiC surface area, at lower binder volume fractions, with the TiC likely being semi-protected by an oxide surface layer [
41,
43]. These corrosion rate trends determined from the Tafel experiments are in general agreement with the observations of Sacks [
9], who studied the corrosion behaviour of WC-Co composites in tannic acid-based electrolytes, and reported an increase in the corrosion rate with increasing Co binder content. It was demonstrated that there was a preferential dissolution of the Co binder, while the WC grains retained their sharp facets, and are effectively not attacked during the corrosion tests [
9].
Using a broadly similar Tafel approach, Toma et al. determined aqueous corrosion rates for WC-Co and WC-CoCr, and demonstrated the former to be ~0.76 mm/year, which is lowered to ~0.32 mm/year through the addition of Cr [
44]. The best cermet coating systems for corrosion resistance are presently those based on Cr
3C
2, with a NiCr binder that has been demonstrated to exhibit aqueous corrosion rates as low as 0.008 mm/year [
45]. Some caution should be taken when evaluating these different materials, particularly in comparing bulk cermets and coatings, where coating porosity may play a role in degrading the corrosion properties. However, the present materials exhibit generally similar performance, in comparison to the Cr
3C
2-based cermets, when assessing electrochemically derived (i.e., Tafel) corrosion rates. When contemplating the electrochemical response of the present materials, it is important to consider which components might be playing an active role in corrosion, and which may be inert.
Following the outlined Tafel extrapolation approach provides an apparently clear dependency of corrosion rate as a function of binder content. However, Hochstrasser-Kurz et al. noted that the metallic component in their system, namely WC-Co, was selectively removed, and therefore the effective corroding area should be based on the metallic component only [
22]. Taking the same approach in the current work,
Figure 4b presents revised corrosion rate data, where only the metallic component is participating; here, it is assumed that the area fraction of the metallic component is equivalent to the volume fraction. In this instance, the corrosion rates are based on revised values for
icorr, due to the reduced cross-sectional area undergoing corrosion. These modified area values for
icorr, termed
icorrMA, were presented previously in
Table 4. This “normalisation” approach largely eliminates any obvious dependency of corrosion rate upon the steel binder content, although trends between steel binder compositions are still apparent.
When examining the corrosion mechanism for the present cermets, it can be initially postulated that the electrochemical response is related (at least in part) to a potential galvanic effect between the TiC and steel binder. Given a nominally constant grain size (
Figure 2a), as the binder content is reduced, the carbide–carbide contacts increase (
Figure 2b). In terms of the microstructure, there is a transition from a nominally continuous TiC framework, with isolated islands of steel, to a continuous steel structure, with isolated TiC grains (
Figure 1). This clearly results in a complex relationship between the steel binder fraction and the carbide/metal interfacial area. Looking specifically at the system components, the corrosion potential of a single-phase TiC ceramic, prepared from the same TiC powder as the present work (with some W contamination), in an equivalent 3.5 wt. % NaCl electrolyte was determined to be −0.176 V (vs. SCE) [
46]. Interestingly, that work showed the effects of powder composition, as a high-purity TiC sample exhibited a corrosion potential of +0.005 V (vs. SCE). It can consequently be anticipated that the difference in corrosion potentials between the TiC ceramic phase and the steel binder, in a conducting electrolyte, may potentially lead to galvanic activity at the interfaces between these two dissimilar materials, especially for the 410 L grade.
It can be seen from
Table 5 that, of the baseline metals, 316 L stainless steel has the best corrosion resistance, but when incorporated with TiC into a cermet structure, the TiC-304 L cermets have better corrosion resistance than their TiC-316 L and TiC-410 L counterparts (
Figure 4). This is likely related to a reduced difference in the electrochemical potentials and galvanic activity between the TiC and the steel binder. Generally, when two dissimilar materials are incorporated in a conducting electrolyte, and one is more noble (cathodic) while the other is more active (anodic), the difference in potential between the cathode and anode site will indicate the expected degree of galvanic corrosion. Consequently, the smaller the difference in potential, the lower the expected extent of galvanic corrosion [
40]. However, changes in electrolyte composition and temperature could also alter the potential positioning in the galvanic series [
40]. As noted in the previous paragraph, the corrosion potential of the present TiC (as a single-phase material) in a similar electrolyte to the current study was determined to be −0.176 V [
46], while those of 304 L, 316 L, and 410 L steels (shown in
Table 4) are approximately −0.204, −0.221, and −0.400 V (versus SCE), respectively. Comparing the difference in the OCP between TiC (cathodic) and the metallic phase (anodic), based on this principle, TiC and 304 L offer the least likely candidates for the formation of a galvanic couple due to exhibiting the lowest potential difference between the two phases. This is further investigated by the performance of the cyclic polarisation analysis. The results of cyclic analysis, such as characteristics of the hysteresis loops and the pitting (
Epit) and repassivation/protection potentials (
Eprot), was used to confirm the susceptibility of each composition toward localised corrosion, similar to those initiated in ceramic/metallic interface within the composition.
The cyclic polarisation responses of the cermets and the baseline steels are presented in
Figure 5. The pitting and repassivation (or protection) potentials (
Epit and
Eprot, respectively) are used to study the susceptibility of the materials to localised corrosion. The relative degree of “self-healing” ability of the surface film is then given by [
47,
48]:
The accompanying electrochemical results of the cyclic polarisation experiments are shown in
Table 5. Where the value of
Eprot is lower than
Ecorr, no protection will be observed, and this is noted as such in the table. It is typical that an increase in
Epit improves the resistance of the materials to pitting corrosion, while a decrease of ∆
E indicates a greater ability towards the self-healing of the passively formed surface film [
47]. By increasing the steel binder content, the pitting resistance of the cermets is improved (i.e.,
Epit increases with binder content).
For the TiC cermets prepared with 304 L and 316 L, positive values of ∆E in addition to anti-clockwise (−) and relatively small hysteresis loops, were observed for the cermets containing only 10 vol. % of the metallic binder phase. This behaviour is likely related to the protection rendered by the formation of a surface oxide at the comparatively higher TiC content (effectively 90% of the surface area). Further increases in amounts of 304 L or 316 L binder resulted in the formation of large, clockwise (+) hysteresis loops, which can be interpreted as an increased susceptibility of these compositions towards localised corrosion.
On the other hand, cyclic polarisation tests on the cermet compositions prepared with 410 L steel showed a very different behaviour. The cermets with a 410-L steel binder consistently exhibited anti-clockwise (−) hysteresis loops, with the smallest loop corresponding to the composition with 30 vol. % of the metallic binder phase. It was also observed that the calculated ∆E increased with an increasing binder content.
To further confirm the observations from cyclic polarisation, and to substantiate the possibility of localised corrosion, potentiostatic polarisation tests were conducted at applied voltages slightly above
Epit (
Figure 5); the specific polarisation voltages used for each sample are provided in
Table 6. All of the present materials showed an
increase in current with time during the potentiostatic scans above
Epit, indicating the probability of localised corrosion. These observations confirm the results of the cyclic polarisation tests shown previously. It can also be seen from
Figure 6 that the cermets with 10 vol. % 304 L and 316 L steel showed an increase in current with time. This again infers the superior capacity of the lower binder content cermets towards the better oxidation resistance of the passive film, when compared with the other cermets studied. The large oscillations observed in the potentiostatic curves for the pure 316 L and 304 L stainless steels presents evidence of localised corrosion, and an aggressive attack of the passive film by Cl
− ions.
3.3. Post-Corrosion Sample Characterisation
Typical SEM images of the cermets after potentiodynamic polarisation testing are shown in
Figure 7; in these examples, the cermets have been subjected to an accelerated corrosion through a voltage sweep to +1.5 V (versus SCE). Preferential dissolution of the binder is apparent, while the TiC grains remain largely unaffected following the tests. Even when the steel binder is removed, the TiC particles support each other to a large extent and form a continuous, rigid skeleton. However, some surface TiC grain removal does occur; the presence of remnant surface grains can be easily observed in the form of a retained, uniform flattened face (i.e., previously polished) on each individual grain (e.g.,
Figure 7a).
It can be clearly seen that at the highest binder contents, essentially all of the interconnecting metal binder content is removed from the near surface region, leaving clean and smooth TiC grains (
Figure 7b). In this instance, the surface TiC grains are also removed, as TiC particles are clearly rounded at the new surface (i.e., the previously polished grains are removed).
The typical effects of cyclic polarisation on the cermets and the accompanying baseline steels are shown in
Figure 8. It is apparent that there is heavy pitting on the baseline steel surfaces (
Figure 8d), which is in agreement with the experimental results of the cyclic polarisation tests, as shown in
Figure 5. In that instance, the values of
Eprot for the steels (i.e., the protection potential) are more active than
Epit, indicating the tendency for localised corrosion. The presence of pits also relates to the oscillatory response observed for the steels during the cyclic and potentiostatic polarisation experiments (
Figure 5 and
Figure 6, respectively), which indicates evidence of Cl
− ion attack [
39]. However, for the TiC–stainless steel cermets, there is only a selective attack occurring locally at the interface between the TiC and the steel binder, resulting in the removal of the binder itself. In this situation, the TiC ceramic particles are not affected in any obvious way. The initial stages of this corrosive process are clearly shown in
Figure 8b, with nominally lenticular-shaped voids appearing at the interface between the TiC and the steel phase (in this case, 316 L). Here, the bulk of the steel is still retained, but the initial stages of material loss are clearly associated with the interfacial region, indicating the dominance of a galvanic mechanism (rather than general corrosion), as outlined earlier. Close inspection of the TiC grains in the image shows that they invariably have consistent surface curvature, inferring that material loss arises from the steel binder (as predicted by the galvanic corrosion argument), which is the anodic portion of the galvanic couple. It is interesting to note that there is, broadly speaking, an association of the attacked regions with areas where there are closely neighbouring TiC grains. This indicates that residual stress and/or localised dislocation density in the steel may also play a role in corrosion in the interfacial region.
The lack of substantial evidence for pitting on the cermets can be postulated to be due to the beneficial protective effect offered by the formation of a (partially protective) surface oxide which, based on prior studies [
41,
42], is likely to be a titanium oxide (i.e., TiO
2) for the current cermets. From the present work, it can be clearly seen that there is a selective interfacial attack on the cermets, resulting in steel binder loss, while the TiC grains are unaffected. As noted previously, it is expected that the anodic area is the metallic side of the interface between the steel binder and hard TiC phase (where metallic dissolution takes place), whereas the adjacent ceramic side of interface is the cathodic region (where oxygen reduction takes place), although in the case of both 304 L and 316 L, the corrosion potential differences with the present TiC are small. Within the interface zone (anodic region), metal dissolution occurs, and increases the migration of chloride ions into the interface. The formation of metal cations may be anticipated to lower the pH (acidifying the solution) near the interface, due to the hydrolysis of metal cations, leading to the more intensive corrosion in this region. This theory could also potentially explain the localised corrosion at the interface of the cermets, acting in conjunction with galvanic attack. However, one further factor needs to be considered, in that there is some evidence of localised compositional change in these cermets during fabrication, as noted in an earlier study [
49]. It was shown, specifically for the TiC-316 L system, that the binder composition is modified, as Mo is removed to a large extent, and is incorporated into the outer rim of a complex core–rim structure that is formed on the TiC grains. This outer rim contains Mo and also W, which was present as an impurity in the starting TiC powder.