Effect of environmental variables and main alloying elements on the repassivation potential of Ni–Cr–Mo–(W) alloys 59 and 686

3.1 Polarization curves

Polarization curves for alloys N06059 and N06686 recorded in NaCl and CaCl₂ solutions at different temperatures are shown in Figure 2. These tests were carried out to determine the anodic behavior of the alloys, avoiding crevice corrosion phenomena, although, in some instances, it could not be avoided (Rodríguez et al. 2010). The curves showed an extensive passive zone with low passive currents approximately independent of temperature. The potential at which the current increased after the passive domain (breakdown potential) depended on the temperature and the solution. The breakdown potential was associated with the onset of transpassive film breakdown in most cases, and with crevice corrosion initiation in other cases. For both alloys, a pseudo-secondary passive region (current density plateau before the transpassive domain) was observed between ∼0.4 V_SCE and 0.6 V_SCE. This was attributed to surface enrichment with Ni(OH)₂ in the passive film (Mishra and Shoesmith 2013). The most striking difference among polarization curves in the tested solutions was the transpassive potential, where a wider passive zone was observed in CaCl₂ solutions. With increased potential, a crevice corrosion breakdown and further repassivation were observed on the alloy N06059 at 60 °C. Crevice corrosion without further repassivation was observed for alloy N06059 at higher temperatures than 60 °C. Crevice corrosion was also found on the alloy surface under the top PTFE compression gasket. No crevice corrosion was observed on alloy N06686 under any tested condition. The absence of localized corrosion on alloy N06686 is consistent with its higher PRE value (Table 1). Although the Ca²⁺ solution used here corresponds to the highest aggressive ion concentration, a particular inhibitory effect of Ca²⁺ concerning Na⁺ is already known (Jiang et al. 2006). This inhibitory effect becomes less evident with increasing temperature, to the point that Carranza and Rebak (2009) did not observe any effect on E_R,CREV at 90 °C. The polarization curve of alloy N06686 in CaCl₂ 5 M at 117 °C deviates from the behavior observed at lower temperatures and shows a similar shape to those in NaCl solutions (Figure 2b). Deviation behavior is supposed that corresponds to a suppression of the Ca²⁺ inhibition by the high temperature on alloy N06686, and crevice corrosion on alloy N06059 may have overshadowed the apparent suppression of the Ca²⁺ inhibition.

Figure 2:

Polarization curves in deaerated 0.1–1 M NaCl and 5 M CaCl₂ at 30, 60, 90 and 117 °C for alloys: (a) N06059 and (b) N06686. Potential scan rate: 0.6 V/h.

3.2 Polarization curves in PD-GS-PD tests

Figure 3 shows PD-GS-PD tests performed on alloy N06686 in chloride solutions at different temperatures. The outcome of the PD-GS-PD tests depended on the alloy, the chloride concentration and the temperature. For both alloys and solution/temperature conditions, the measured PD-GS-PD curves exhibited a shape similar to one of the five reference curves shown in Figure 3 (Types 1, 2, 3, 4 and 5). The experimental conditions and the electrochemical behavior exhibited by each alloy are summarized in Table 2.

Figure 3:

Examples of behaviors exhibited during PD-GS-PD tests on chloride solutions: Types 1–5. The results correspond to alloy N06686, and the multiple curves are replicate tests to illustrate reproducibility.

At temperatures higher than 80 °C, both alloys showed a Type 1 behavior (Figure 3a), which is the most common outcome of the PD-GS-PD method. In those conditions, the crevice corrosion initiation potential (E_CREV) is indicated by a sudden increase of the current density. The potential decreases during the galvanostatic step, with a final repassivation and a positive hysteresis loop during the potentiodynamic polarization in the cathodic direction. Post-test LOM/SEM examination revealed crevice corrosion damage occurred in all the specimens. For specimens tested in CaCl₂ solutions, the appearance of the alloys was clean and without corrosion products, Figures 4a and 5a for alloys N06686 and N06059 at 90 °C. The localized attack on the specimen tested in CaCl₂ 5 M spread out from the covered area by the crevice former teeth. The localized attack was crystalline where grains and stacking faults were discernible in SEM images (Rebak 2005). For specimens tested in NaCl 0.1 and 1 M solutions, the attack was discontinued and full of corrosion products on the alloy surface and below the crevice former teeth, as shown in Figure 5b, c.

Figure 4:

SEM and macroscopic images of alloy N06686 specimens after testing in 5 M CaCl₂ at 90 °C (a and b are Type 1 and 2, respectively) and at 80 °C (c is Type 3).

Figure 5:

SEM and macroscopic images of alloy N06059 specimens after testing in 5 M CaCl₂ (a), 1 M NaCl (b) and 0.1 M NaCl (c) at 90 °C. All of them with Type 1 behavior.

Type 2 curves (Figure 3a) were observed in tests performed in CaCl₂ 5 M solution at temperatures from 60 to 90 °C for the alloy N06059 and at 90 °C for alloy N06686. The PD-GS-PD tests with Type 2 behavior showed similar E_CREV to Type 1 (Figure 3a), but a sudden increase in the recorded potential during the second step (galvanostatic step). The potential reached high values up to the transpassivity range and remained there throughout the entire stage. Mishra and Shoesmith (2014) observed similar behavior on the galvanostatic step of the PD-GS-PD test on the 686 alloy on NaCl 1 M at T < 80 °C. The potentiodynamic polarization in the cathodic direction (third stage of the method) showed a decreased current density (negative hysteresis loop) up to a reactivation peak and subsequent repassivation. The reactivation peak exceeded the current density value of the previous galvanostatic stage, up to 200 μA/cm² at +50 m V_SCE. Repassivation potentials were similar to those obtained on Type 1 (Figure 3a). LOM/SEM examination revealed a more profound and rough attack than Type 1 (Figure 4b). The unbright surface under the crevice formers showed an aggressive localized corrosion as the elevated charge circulated. Rebak (2005) characterized that surface with a spotty dull grey appearance, with isolated crystallographic etch pits. Rebak argued that this type of corrosion occurs when potentials above transpassivity are applied to the specimen in non-aggressive solutions, this is low chloride concentration or low temperatures. In contrast, this type of attack was observed with the highest chloride concentration in this work.

Type 3 curves (Figure 3b) were also observed in tests performed in CaCl₂ solution at 60 °C for the alloy N06059 and at temperatures of 80 and 90 °C for alloy N06686. The PD-GS-PD tests that showed Type 3 behavior did not show an E_CREV as Type 1 and 2. Instead, a wide passive range is observed until the current density threshold is obtained at transpassive potentials. The recorded potential during the second step (galvanostatic) did not show changes over time. The potentiodynamic polarization in the cathodic direction (third stage of the method) showed a similar behavior than in Type 2, following the current density path from the first stage of the technique until the reactivation peak. The reactivation peak did not exceed the galvanostatic current density value of the previous step. LOM/SEM examination revealed a rough attack similar to that of Type 2 (Figure 4c), even with a low-peak current density, and it is supposed that the charge circulated by the current density peak contributes to the observed crevice corrosion.

Type 4 curves (Figure 3c) were observed in tests performed in NaCl solution at temperatures from 60 to 80 °C for alloy N06059 and alloy N06686 at temperatures from 70 to 80 °C in NaCl 0.1 M and 60–80 °C in NaCl 1 M. The recorded potential during the galvanostatic step (second step) did not show changes over time, the same as Type 3 behavior. The potentiodynamic polarization in the cathodic direction (third stage of the method) showed higher current densities than the first stage, obtaining a positive hysteresis loop. No reactivation peaks were observed on NaCl solutions. LOM/SEM examination revealed a similar attack to Figure 5b, c (not shown).

Type 5 curves (Figure 3d) indicated the absence of localized corrosion. Type 5 curves were observed at 50 °C for alloy N06059 and at 60 °C in NaCl 0.1 M, 50 °C in NaCl 1 M and 70 °C in CaCl₂ 5 M for alloy N06686. LOM/SEM examination revealed no crevice corrosion. The recorded potential did not show any change over time during the galvanostatic step. No significant hysteresis between the forward (step 1) and reverse (step 3) curves were observed. This was indicated by the current density retracing the forward scan until finally achieving values lesser than those recorded during the forward scan.

Some PD-GS-PD tests performed in alloys N06059 and N06686 reached transpassive potentials during stage 1 and remained at those high potentials during stage 2. The transpassive zone of potentials is characterized by a loss of stability of the passive layer, where the electrochemical conversion of the Cr (III) oxide into soluble Cr (VI) species occurs (Hayes et al. 2006; Mishra et al. 2016). Henderson et al. (2019) observed that Ni (II) is the principal cation released during transpassive dissolution on Ni–Cr–Mo alloys, while some Cr but mainly Mo are retained, enriching the surface in these elements. The same authors observed that during repassivation, the Mo that enriched the metallic surface is subsequently released during repassivation. This release of Mo surface could be related to the sudden increase in current density, and the peak observed at [Cl⁻] = 10 M (Figure 3a, b) by a hypothetical depletion of this alloying element on the metallic surface after it was released during repassivation.

Figure 6 shows E_R,CREV from the PD-GS-PD method on alloys N06059 and N06686, with the temperature for NaCl 0.1 M, NaCl 1 M and CaCl₂ 5 M. The symbols are the average values, and the error bars represent the standard deviation. The CCT inferred for alloy N06059 from the present results was between 50 and 60 °C. For alloy N06686, crevice corrosion was observed at T ≥ 60, 70, and 80 °C for [Cl⁻] = 1 M, [Cl⁻] = 0.1 M and [Cl⁻] = 10 M solutions, respectively. From the present results, CCT for alloy N06686 was estimated between 50 and 60 °C, as for alloy N06059. A considerable lower CCT was obtained with the PD-GS-PD method compared to those obtained by Agarwal et al. (2000), which following ASTM G48 (10% FeCl₃ solutions), report CPT > 85 °C for both alloy N06059 and alloy N06686.

Figure 6:

E _R,CREV of alloys N06059 and N06686 as a function of temperature. Average E_R,CREV and standard deviation values are represented along with the fit of Equation (2).

Equation (2) states a relationship between E_R,CREV chloride concentration and temperature. This empirical equation was used at first by Dunn et al. (2006) for alloy C-22 (UNS N06022). In previous works, Equation (2) shows a good fitting for E_R,CREV of Ni–Cr–Mo–W alloys with temperature and chloride solutions (Hornus et al. 2014, 2015; Maristany et al. 2015). It was fitted to a range of the collected data for alloys N06059 and N06686. A, B, C and D are constants parameters that depend on each alloy. Figure 6 also shows the fits of Equation (2) with dash-line for the two alloys. The comparison of the E_R,CREV values of the tested alloys as a function of temperature and chloride concentration indicates that their corrosion resistance increased according to their corresponding PRE, being N06059 < N06686. When the temperature and chloride concentration increased E_R,CREV showed a lesser dependence on the environmental variables.

The parameters obtained by least-square fits of Equation (2) are shown in Table 3 for alloy N06059 and Table 4 for alloy N06686. The correlation coefficients (R²) obtained were 0.77 and 0.90 for alloys N06059 and N06686, respectively, which indicates that Equation (2) successfully represented E_R,CREV as a function of temperature and chloride concentration for the tested alloys. Although the correlation coefficient on alloy N06059 was lower than that of alloy N06686, the standard deviations observed in the fitted parameters (A, B, C, D) were at least one order of magnitude lower than the mean values, in all cases. Regardless of that a similar normal distribution of E_R,CREV is expected around the average values given by the fits, the high error bars on E_R,CREV values for alloy N06059 in NaCl 1 M at 70 °C and NaCl 0.1 M at 60 °C (Figure 6) could contribute to its low correlation coefficients.

3.3 Alloying effects on E_R,CREV

Dunn’s empirical equation for E_R,CREV has shown an excellent correlation for the repassivation potential with environmental variables such as temperature and chloride solution concentration. Their parameter A indicates the variation of E_R,CREV with log[Cl⁻] independently of the temperature, the parameter C indicates the variation of E_R,CREV with T independently of the log[Cl⁻], B shows the temperature-dependent variation of E_R,CREV with log[Cl⁻] and D an independent parameter. Nevertheless, all parameters depend on the alloy composition.

Obtaining an equation for E_R,CREV that involves the environmental parameters and the main alloying elements would help to quantify the individual effects of main alloying elements on E_R,CREV. Linear behaviors of different corrosion parameters such as Critical Pitting Temperature (CPT) and Critical Crevice Temperature (CCT) with the PRE value were observed by various researchers on Ni alloys and stainless steels (Sarmiento Klapper et al. 2017; Sedriks 1996). Sosa Haudet et al. (2012) obtained E_R,CREV by the PD-GS-PD method for several Ni–Cr–Mo–W alloys in chloride solutions. They observed a linear behavior of the E_R,CREV with the PRE values on NaCl 1 M and CaCl₂ 5 M at 60 °C. Similar linear shapes of E_R,CREV with PRE were reported by Zadorozne et al. (2012) on Ni–Cr–Mo–W alloys on NaCl solutions. The linear relationships observed by Sosa Haudet et al. and others could be extended from a phenomenological point of view to different temperatures and chloride concentrations as follows:

where ER,CREVA1,B1,C1,D1 and ER,CREVA2,B2,C2,D2 represent Equation (2) and PRE corresponds to Equation (1). The extended or explicit form of Equation (3) for fitting is shown as follows:

Figure 7 shows the fitted Equation (4) for alloys N06059 and N06686 (this work) and alloys N06625, N06022, N07022, and N10362 (Hornus et al. 2014) as a function of temperature and chloride solution concentration. Equation (4) shows a complex interdependence of environmental variables ([Cl⁻] and T) and the content of alloying elements (PRE) to determine E_R,CREV. The parameters obtained by least-square fits of Equation (4) are shown in Table 5. A similar fit shape was observed with Equation (2) for alloys N06059 and N06686 and the other Ni–Cr–Mo–W alloys. The correlation coefficient (R²) obtained with Equation (4) was 0.83 for the six alloys, which is better than the one obtained for alloy N06059 (R² = 0.77) with Equation (2) but lower than that observed for alloys N06625 (R² = 0.90), N06022 (R² = 0.92), N07022 (R² = 0.92), N10362 (R² = 0.98) (Hornus et al. 2014) and N06686 (R² = 0.90). Figure 7 shows a high dispersion of the repassivation potential for all alloys with a chloride concentration less than or equal to 1 M. The statistical dispersion of the repassivation potentials observed in 10 M chloride solutions is considerably lower than for the other chloride concentrations. Miller and Lillard (2019) studied the nucleation and growth stages of crevice corrosion by potentiostatic methods measuring 50–1000 µA in alloy 625 in 1 M NaCl, and they observed that the completion of the growth stage exceeded 48 h. According to them, the repassivation during the growth stage is the reason for the high dispersion in repassivation potentials. The galvanostatic step on the PD-GS-PD method lasts only 2 h, and a rapid decrease in potential has been observed for high chloride concentrations, getting slower for the more dilute solutions, as in Figure 8. In general, for [Cl⁻] ≤ 1 M, the potential recorded during the galvanostatic stage does not finally stabilize, contributing to the high dispersion of E_R,CREV. Figure 7 shows the fitted Equation (4) successfully represented E_R,CREV as a temperature and chloride concentration function for the tested alloys depending on its PRE value.

Figure 7:

Individual E_R,CREV for alloys N06059 and N06686 (this work) and for alloys N06625, N06022, N07022, and N10362 (Hornus et al. 2014) alloys as a function of temperature. Dash lines correspond to the fit of Equation (4).

Figure 8:

Galvanostatic step of the PD-GS-PD method for alloys N06059 and N06686 at 90 °C in chloride solutions with different concentrations.

The terms of Equation (4) were regrouped to compare its fitted parameters with those of Equation (2) (individual fitted for each alloy) according to:

where A′ = A1 + A2 * PRE, B′ = B1 + B2 * PRE, C′ = C1 + C2 * PRE and D′ = D1 + D2 * PRE. Figure 9 shows the fitted parameters of Equations (2) and (5) as a function of PRE. It can be seen that the parameters A (affected by chloride concentration) and D (independent of chloride concentration and temperature) showed a specular behavior with a different sign. The same is observed with parameters B and C. Figure 9 shows that the parameters of the fitted Equation (2) are very similar to those of Equation (5), except for those corresponding to the N06059, N06686 and N10362 alloys. The main difference between the three alloys that did not follow Equation (5) parameters is their composition’s null or low Co content. Co alloy is added to Ni alloys to increase the resistance to carburization and sulfidation as Co increases the solubility of C in Ni-based alloys (Davis 2000). The unmatched parameters for alloys N06059, N06686 and N10362 (indicated in Figure 9 with dot-lines) could suppose a Co effect that is not contemplated in the PRE equation.

Figure 9:

Fitted parameters of Equations (2) and (5) as a function of PRE. Dotted lines show the PREs of alloys N06059, N06686 and N10362.

From Equation (4) and the fitted parameters (Table 5), we analyzed the effect of incremental values of Cr, Mo and W on E_R,CREV. The derivatives of E_R,CREV concerning Cr, Mo and W were calculated and evaluated for each alloy, chloride concentration and temperature. Figure 10 shows dE_R,CREV/d[%Cr], dE_R,CREV/d[%Mo] and dE_R,CREV/d[%W] in weight percent for 0.1–10 M chloride concentration as a function of temperature. The three derivate behaviors are the same with the Y-axis modification. The largest effect on repassivation potentials with increasing Cr, Mo or W alloys would be observed at low temperatures and chloride ion concentrations. In a particular case, there is a temperature for which the increase in repassivation potential would be independent of the chloride ion concentration (∼84.1 °C) and a chloride ion concentration for which the repassivation potential would be independent of temperature (∼3.36 M). At 85 °C (where slopes are independent of chloride concentration), the Cr–Mo–W contribution to the E_R,CREV on Ni-based alloys was about 5–6 mV/%Cr, 17–18 mV/%Mo and ∼9 mV/%W.

Figure 10:

dE _R,CREV/d[%Cr] (black), dE_R,CREV/d[%Mo] (red) and dE_R,CREV/d[%W] (green) in weight percent for 0.1–10 M chloride concentration as a function of temperature.

Sosa Haudet et al. (2012) obtained dE_R,CREV/dPRE from several Ni-based alloys in NaCl 1 M and CaCl₂ 5 M at 60 °C. As we could not represent directly dE_R,CREV/dPRE on Figure 10, we can obtained them as dE_R,CREV/d[alloy] = dE_R,CREV/dPRE × dPRE/d[alloy] with alloy as %Cr, %Mo or %W. The variation of E_R,CREV with the alloying elements from Sosa are in excellent agreement with this work and almost superimposed on our results (Figure 10, Sosa Haudet 2012). Sosa Haudet et al. (2015) also studied the effect of the alloying elements in E_R,CREV for NaCl 1 M at 60 °C with artificial neural networks. Their results are in Figure 10 (Sosa Haudet 2015) with an overestimation effect for W and sub-estimation for Cr and Mo concerning the present work. It is important to note that a large amount of data and training are necessary for artificial neuronal networks to obtain reliable values. Igual Muñoz et al. (2004) studied the E_R,CREV on the alloys 33 and 31 in LiBr solutions at a scan rate of 0.5 mV/s and different temperatures. The variation of E_R,CREV with Mo are close to those obtained in the present work, even increasing with the temperature (Figure 10, Muñoz). The Mo effect on E_R,CREV reported by Muñoz is higher than in the present work. The less aggressive bromide solutions plus the overlapped effect of increased Mo with a Cr falling could explain its behavior even when Mo is not as effective in inhibiting Br⁻ localized corrosion as it is for inhibiting Cl⁻ localized corrosion (Kappes 2019). Ha et al. (2017) observed a higher W effect on E_R,CREV than the actual work on duplex stainless steels at 90 °C with a scan rate of 2 mV/s.

Each derivate from Equation (4) gives us information about the behavior of E_R,CREV with the different variables and parameters. Classical mathematical analysis tells us that the gradient of a function allows us to observe the direction of maximum growth of the function (Marsden et al. 1993). Considering the triple Cr–Mo–W as if it were a vector entity, it is possible to obtain the vector components (in this case, the relation among them) that maximize E_R,CREV. The “alloy vector” of maximum growth of E_R,CREV are dependent on the environmental variables (chloride concentration and temperature) that become independent of them when normalized. The ratio of alloys that maximizes the value of E_R,CREV is 1:3.3:1.65 for Cr, Mo and W, respectively. It should be noted that the ratio of the alloy that maximizes E_R,CREV has the same factors as the PRE (Equation 1). The optimal alloying ratio would be independent of the initial composition of the alloy since it has constant values, i.e., an alloying increase of 1:3.3:1.65 in Cr, Mo and W on alloy N06022 would increase the E_R,CREV values in a maximum way, the same as it would on alloy N10362.