1 Introduction

Effective handling of radioactive nuclear wastes is a major concern. Dry storage systems have been widely adopted for long-term storage of used nuclear fuels that consist of several stainless steel (SS) canisters [1,2,3]. Typically, 304L or 316-type austenitic stainless steels plates are fusion-welded together to fabricate the storage canisters. Even though these canisters provide a safer way of storing the used nuclear fuels, an underlying problem of chloride-induced stress corrosion cracking (SCC) of stainless steel canisters remains to be solved. Salts carried by the dust in the atmosphere may settle on the canisters and subsequently deliquesce as heat generated by radioactive decay declines over time. The deliquescence of salt deposit could induce various forms of corrosion attack, such as pitting and CISCC [4, 5]. Pitting corrosion occurs in austenitic SS readily in the presence of chloride solution [6]. The chloride-induced pits that are formed on austenitic SS can act as stress concentrators and further facilitate SCC [4]. Pitting in austenitic stainless steels and their weldments is influenced by various metallurgical factors. The microstructural features such as grain boundaries [7], grain size [8], presence of second-phase particles and inclusions like carbides or manganese sulfide (MnS) [9] can act as pit nucleation sites in austenitic stainless steel. Further, external factors such as the composition of chloride-containing electrolyte [10] and surface finish [11] influence the pitting tendencies in stainless steels. Chloride-induced pitting and stress corrosion cracking are the potential failure mechanisms of stainless steel canisters. Repair welding of SCC cracks is one of the strategies to expand the life span of the stainless steel canisters [12]. The problems with the fusion welding (such as chromium depletion and grain growth at the heat-affected zone (HAZ), segregation at the interdendrites) have been addressed by implementing a low heat input laser welding [13].

A solid-state friction stir welding (FSW) process for crack repair has been reported and showed promising results on the corrosion resistance of the crack-healed region [14]. However, another study showed that friction stir crack-repaired 304L specimens developed stress corrosion cracks on the weld nugget and along the advancing side of the weld when exposed to boiling magnesium chloride (155 °C) for 3 h [15]. On the other hand, friction stir-welded 316L SS showed a more noble pitting potential for the stirred zone than that of the base metal [16]. The increased corrosion resistance of the friction-stirred stainless steel weld nugget or stirred zone (SZ) was attributed to several factors such as: (a) finer grains developed due to severe plastic deformation that promote short-circuit diffusion of passive film-forming elements [17], (b) low-angle grain boundaries (2°–15°) [18,19,20], and (c) decreased donor density of passive films formed on the stirred zone [21]. Several investigators reported localized corrosion of friction stir-welded austenitic stainless steels (not necessarily 304 or 316 types), and the corrosion was attributed to features such as: (i) formation of delta ferrite in the SZ of nickel-free austenitic stainless steel and subsequent enrichment of ferrite-stabilizing elements in the delta ferrite that lead to localized depletion of Cr at the austenite/ferrite interface [22], (ii) tool wear debris that caused galvanic effect [23], (iii) formation of topologically close-packed (TCP) phases such as σ and χ [24], and (iv) high reactivity of the stirred zone or thermomechanically affected zone (TMAZ) due to a high defect density that rendered the passive film to be less protective [25].

The review of recent literature on the corrosion behavior of the friction stir-welded stainless steel materials showed both positive and negative attributes. This work focuses on the friction stir repair of an existing crack and the influence of repair welding on the corrosion behavior. To this effect, an artificial crack was introduced by the electric discharge machining (EDM), a narrow slit at the center of 304L SS coupons. The simulated EDM crack on the 304L SS coupon was repaired or healed by FSW by maintaining two different isothermal tool temperatures such as 725 and 825 °C. The electrochemical corrosion behavior of the friction stir repair-welded 304L specimens was investigated using cyclic and potentiostatic polarization and electrochemical impedance spectroscopic techniques in 3.5% NaCl solution (pH unadjusted) and compared with that of unaffected base metal specimens.

2 Experimental

2.1 Material and repair welding

Simulated cracks were introduced at the centre of the 304L coupons (330 mm (L) × 149 mm (W) × 12.7 mm (T)) by creating slots (5 mm deep and 0.33 mm wide) using EDM along the length of the plates, as illustrated in Fig. 1. Single-pass FSW was performed on plates to repair the simulated cracks at the Pacific Northwest National Laboratory (PNNL) using a gantry-type TTI FSW machine. A polycrystalline cubic boron nitride (PCBN) tool (MegaStir™) with a shoulder diameter of 36.8 mm and pin length of 5.7 mm was used for the FSW. The tool tilt angle was − 0.5°. The FSW was carried out at isothermal conditions by keeping the tool temperature constant at either 725 or 825 °C. The welding speed was 25.4 mm/s in both the temperatures. The welds were carried out in power-controlled mode. The tool speed changes continuously through the length of the weld to maintain a steady power and thus a steady temperature. A customized algorithm that was developed at PNNL was used for both FSW runs. The specialty of this algorithm is that it can maintain a desired weld temperature during the course of FSW tool travel. Fixed weld temperatures were achieved through varying the angular velocity of FSW tool using a feedback mechanism, at a user-specified weld power. PNNL has worked extensively in developing and fine-tuning the control parameters (proportional–integral–derivative PID constants) to ensure a steady weld temperature. (Maximum diversion was noted to be < 1 °C from the specified weld temperature.) Table 1 summarizes the welding parameters employed in this study. The welding parameters (such as rotational speed, vertical forging load, and spindle torque) were selected by the machine algorithm based on the specified welding power, isothermal condition, and linear welding speed. The welding speed was kept at 25.4 mm/min to maintain the lateral load on the PCBN tool in the acceptable range. Higher lateral load can cause premature tool failure.

Fig. 1
figure 1

Image of the 304L stainless steel coupon with a simulated crack introduced by electric discharge machining

Full size image
Table 1 Friction stir welding parameters for isothermal tool temperature
Full size table

2.2 Microstructural characterization

Small specimens were cut from the FSW coupons and metallographically polished down to 0.05 µm alumina slurry using a vibratory polisher for 10 h (Pace Technologies GIGA-900). The mirror-finished samples were electrochemically etched using a 10 wt% oxalic acid aqueous solution. The etching was performed at 10 V for 20 s. Electron backscattered diffraction (EBSD) was performed in a JEOL JSL-6610LV SEM equipped with Hikari camera and TSL OIM v6 data collection software with 20 V accelerating voltage, 70° tilt, and 0.1 μm step size. This was to determine the average grain size in the stir zone and to map grain boundaries as well as misorientation. Grain size measurement was done using the mean linear intercept (MLI) method for all the samples.

2.3 Corrosion studies

The FSW 304L SS samples were cut down to a size of 1 cm2 and metallographically polished down to 1 µm finish. Corrosion characterizations were performed using a three-electrode setup and a potentiostat. A spiral Pt-wire of area 2 cm2 and an Ag/AgCl in saturated KCl electrode (− 46 mV Vs. SCE) were used as counter and reference electrodes. A 3.5% NaCl (0.6 M) solution is used as an electrolyte for the tests. No aeration or deaeration of electrolyte was carried out. In the beginning, the samples were dipped in the electrolyte in an open-circuit condition until steady-state corrosion potential (OCP) was achieved. After reaching the OCP, linear polarization tests were carried out by scanning the potential in the range of \(\pm\) 25 mV with respect to the OCP and at a 0.5 mV/s scan rate. Electrochemical impedance spectroscopy (EIS) tests were carried out on the samples at the OCP by varying the frequency from 10 kHz to 0.01 Hz. After EIS, cyclic polarization (CP) tests were performed in order to understand the pitting behavior. In the cyclic polarization, a forward potential scan was applied starting − 200 mV below the OCP at a scan rate 0.167 mV/s. The forward potential scan was reversed with a same scan rate when the apex current reached a current density of ~ 5 mA/cm2. After polarization studies, the samples were then thoroughly cleaned by ultrasonication in deionized water. Potentiostatic tests were performed on the freshly polished specimens at a potential of 0 V Ag/AgCl. During the potentiostatic conditioning, a current decay behavior was observed initially when a stable passive film was stable. An increase in the current with time signified passivity breakdown. The potentiostatic tests were continued until a current density of 200 μA/cm2 was reached. EIS was carried out at the beginning of the potentiostatic conditioning at 0 VAg/AgCl (referred to as pre-potentiostatic EIS), when a current density of > 20 μA/cm2 was reached (this condition is referred to as pit initiation), and at the end of potentiostatic test when the current density exceeded 200 μA/cm2 (referred to as pit growth). Mott–Schottky analysis was also carried out immediately after EIS. The Mott–Schottky analysis consisted of scanning the potential at 50 mV steps at every 2 s from 0 VAg/AgCl in the cathodic direction to the initial OCPS value and measuring the imaginary impedance at 1000 Hz. Samples were observed under optical microscope before and after the potentiostatic runs.

3 Results

3.1 Microstructures

The microstructure of the base metal (BM) is shown in Fig. 2. The optical microstructure in Fig. 2a shows equiaxed grains. The average grain size determined by the mean linear intercept method was 47 ± 16 µm as summarized in Table 1. Numerous annealing twins inside the grains are evident from the inverse pole figure map shown in Fig. 2b. A grain boundary mapping of the BM showed the presence of about 50% special boundary mostly in the form of Σ3 annealing twins. The fraction of low-angle grain boundaries was very small. The fraction of random grain boundaries could be considered as 47.4%.

Fig. 2
figure 2

Microstructures of 304L SS specimens: a optical microstructure of base metal (BM), b inverse pole figure ( IPF) map of the as-received 304L SS BM from electron backscattered diffraction (EBSD) analysis; c friction stir welded at 725 °C and d friction stir welded at 825 °C

Full size image

Figure 2c and d shows the shape of the stirred zones (SZ) welded in two different isothermal conditions such as 725 and 825 °C, respectively. The presence of spiral defects is evident in both the microstructures which is commonly known as ‘lazy-S.’ This lazy-S feature extends throughout the stir zone spanning the crown of the nugget and passing the root of the weld. This feature appears as an interface between the advancing and retreating sides that intermixed during the stirring operation [26, 27]. The lazy-S appeared because of breaking of the oxide layers which was initially formed on the artificial crack surfaces during the EDM process. A quick preliminary analysis using X-ray energy dispersion spectroscopy (EDS) of the lazy-S showed the presence of Cr-rich carbide and oxide particles. (Results are not shown here.) The number of curved interfaces in the bottom branch was more in the 725 °C sample as compared to the 825 °C one as shown in Fig. 2c and d. These lazy-S interface arcs were observed to be merging with the onion rings predominantly in the 725 °C sample. The presence of flow lines known as ‘onion rings’ could be observed at higher magnifications. The onion rings were predominantly found in the lower part of the SZ where material flow was influenced by the rotation of the pin. The ring spacing decreased as one moves away from the center toward the TMAZ interface. Formation of onion rings was explained by many researchers [28,29,30]. Disappearance of onion rings with increased tool rotation speed has also been reported in some previous studies [30]. The absence of onion rings in the top half of the SZ in both the samples can be attributed to intense material mixing driven by tool shoulder rotation [29].

Figure 3a and b shows inverse pole figure (IPF) maps of the SZ of FSWed samples at 725 and 825 °C, respectively. In both the FSW conditions, the SZ showed significant refinement in grain size as compared to BM. The grain boundary character distribution was different in the SZ compared to that of BM as summarized in Table 2. The fraction of low-angle grain boundary (LAGB) increased by an order of magnitude after severe plastic deformation during the FSW. This is attributed to the stirring process at 725 °C which has left some unrecrystallized part in the form of substructure. A decrease in the fraction of Σ3 and Σ9 fraction has also been noticed in the SZ as compared to that of BM. The severe plastic deformation could cause disappearance of the coincident lattice points in the boundary plane. Furthermore, migration of high-angle boundary (HAGB) during recrystallization could have resulted in a sharp drop in the number of annealing twins in the SZ. A marginal decrease in Σ3 fraction has also been noted on the SZ-advancing side which is mostly because of higher deformation associated with that side as compared to the retreating side [31]. The 825 °C SZ showed higher mean grain size than the 725 °C SZ which could be associated with the grain growth that was more pronounced at 825 °C than at 725 °C. Grain size distribution analysis of 825 °C SZ revealed that the area fraction of submicron size grains was less than 5% throughout the SZ. The amount of LAGB in 825 °C SZ was less than that obtained in the 725 °C SZ because of higher tool rotation [32]. At a lower rpm condition, due to frictional heat, the dislocations could easily group to form low-angle boundaries (referred to as polygonization) in larger numbers than that occurred in a higher rpm condition because of the low energy condition. The low-angle boundaries are sub-grain structures and consist of linear dislocation arrays. A low-angle boundary moves as a unit under shear stress [33].

Fig. 3
figure 3

Microstructures (inverse pole figure (IPF) map from EBSD) of 304L SS friction-stirred top surface: a friction stir welded at 725 °C (mean grain size 2.4 ± 0.9 μm) and b friction stir welded at 825 °C (mean grain size 4.2 ± 1.6 μm)

Full size image
Table 2 Summary of grain structure analyses of base metal and friction-stirred zone (SZ) of 304L SS from the EBSD data. Low-angle (2°–15°) grain boundaries are denoted as LAGB. The balance is % of random boundaries
Full size table

3.2 Corrosion under open-circuit condition

The plots of open-circuit potential as a function of time are shown in Fig. 4a and c. The steady-state open-circuit potentials of the specimens in 3.5% NaCl solution are summarized in Table 3. Open-circuit potential (OCP) or the corrosion potential is a mixed potential at which the rate of anodic reactions equals the rate of cathodic reaction. Since there was no deaeration of the electrolyte, the predominant cathodic reaction could be reduction of dissolved oxygen in the solution, and the anodic reaction was oxidation of the stainless steel. Formation of a passive film on the specimen will increase the polarizability and increase the anodic Tafel slope which, in turn, will shift the OCP in the more positive direction if the cathodic reaction rates do not change. Therefore, the values of OCP will be an indication of the reactivity or passivity of the specimens in a given environment. The more positive the OCP, the more passive will be a material. The OCP shifted in the negative direction with the increase in time and reached a plateau value for all the specimens. The OCP of SZ-825 °C stirred sample is the most negative followed by SZ-725 °C. Therefore, it can be stated that the SZ-825 sample is the most corrosion prone in open-circuit condition, whereas the base metal is the least corrosion prone in the 3.5% NaCl solution. Also, the OCP of SZ-825 sample showed significant fluctuation before reaching a quasi-steady-state conditions as shown in Fig. 4c. The OCP or corrosion of the 304L SS base metal was found to be around − 380 mV. This value is consistent with previously reported OCP value in the same aqueous sodium chloride medium. A similar value of − 372 mV for the OCP was reported by Heakal et al. [34]. The linear polarization resistance is another indicator to describe the corrosion resistance of a metal. The higher the polarization resistance, Rp, the higher the resistance toward corrosion. The Rp values for the different specimens are listed in Table 3, which follows the same trend observed for the OCP values. The 304L SS base metal showed the highest resistance toward the polarization or better corrosion resistant than that of SZ-725 and SZ-825 specimens. Figure 4c shows the Tafel plots of the 304L base metal and FSW samples in 3.5% NaCl solution. The Tafel slopes of cathodic (βc) and anodic (βa) branches were evaluated by considering the linear portion of the log i vs. potential plots. The corrosion current density icorr is related to cathodic and anodic Tafel’s slope (βc, βa) and Rp (ohm. cm2) values and can be expressed as [35]:

$$i_{{{\text{corr}}}} \, = \,\left( {\beta_{c} *\beta_{a} } \right) \, / \, \left( {2.303 \, \left( {\beta_{c} \, + \,\beta_{a} } \right) \, R_{p} } \right).$$
(1)

The values of βc, βa, Rp, and icorr are summarized in Table 3. The FSW specimens showed steeper anodic and cathodic Tafel slopes than the base metal indicating an increased polarizability after plastic deformation, possibly due to the presence of more active sites for film formation on the specimens. The SZ-825 specimen showed the highest corrosion rate among the specimens investigated because the icorr value was influenced by the lower Rp and steeper Tafel slopes.

Fig. 4
figure 4

Open-circuit potential as a function of time of stainless steel specimens in 3.5% NaCl solution at room temperature: a 304L base metal, b friction-stirred zone at isothermal tool temperatures at 725 °C and c 825 °C samples. d The polarization plots for evaluating Tafel slopes

Full size image
Table 3 The comparison of OCP and linear polarization resistance values for the base metal, SZ at 825° C, and SZ at 725° C friction stir crack-repaired samples
Full size table

Figure 5a shows the EIS data as Nyquist plots and Bode plots which are shown in Fig. 5b. The (depressed) semicircles observed in the Nyquist plots reflect the electrochemical interfacial behavior influenced by the material properties including microstructural inhomogeneity, presence of inclusions, etc. [36], which is fundamentally related to the corrosion behavior of the material. The EIS data were fitted with several electrical equivalent circuits (EECs) using the commercially available software program called Echem Analyst available through Gamry Instruments, Inc., USA, and a suitable model was chosen based on the goodness of fitting value. The EIS data were fitted with an EEC having two time constants. Even though the Nyquist plots appear to have only one time constant, the Bode plots exhibited two time constants as seen from the phase angle plots, specifically that of the base metal. After examining several types of EECs with one and two time constants, the EEC illustrated in Fig. 5c was considered to represent the EIS data better. The corresponding values of the equivalent electrical circuit (EEC) elements are listed in Table 4. The fitted values of the charge transfer resistance, R1, obtained from Table 4, are in a good agreement with the Rp value listed in Table 3. In the equivalent circuit, R0 represents the electrolyte resistance, R1 represents the charge transfer resistance or polarization resistance, and Q1 represents interfacial leaky capacitance whose impedance (Z) is given by the relation: Z = 1/[(jω)aQ1] where j = √− 1, ω = frequency, and the exponent ‘a’ represents the surface heterogeneity or continuously distributed time constants for charge transfer when a < 1. The second ‘RC’-type loop consisting of R2 and Q2 components represents the defect structure of the passive film. The Q2 could be related to the lumped capacitance developed due to high grain boundary area of the thin passive film. An epitaxial-type growth of passive film is assumed here, and therefore the grain size of the passive film is correlated with the grain size of the substrate. It is noted that the Q2 values of FSW specimens were three orders of magnitude higher than that of base metal specimen. The EIS behavior of the base metal did not include the circuit elements R3 (a resistor) and W1 (Warburg element) representing diffusion-controlled impedance behavior. These elements were part of the EIS behavior of FSW specimens which could be associated with the high fraction of low-angle grain boundaries in the substrate. The alignment of dislocations that formed low-angle grain boundaries in the substrate could be a source of misorientation dislocations at the metal/film interface [37]. Burgers vector of these dislocations will be normal to the interface plane (that is the reason it is called misorientation) and create monoatomic ledges at the interface. Anion vacancies could be created when climb occurs in these dislocations. Therefore, the climb of misorientation dislocations results in localized growth of oxide [37]. This entire process including the dislocation climb is assumed to be diffusion-controlled and represented by the Warburg element. The oxide growth process occurring by lateral climb of misorientation climb is considered not to induce stress in the metal substrate [37]. The EIS spectra followed a similar trend that is observed in the linear polarization studies. These findings clearly give a better idea that the base metal is more resistant to corrosion compared to the crack-repaired samples by friction welding.

Fig. 5
figure 5

Results of electrochemical corrosion tests of 304L base metal, and friction-stirred zone at isothermal tool temperatures at 825 °C and 725 °C samples (referred to as 850 °C and 725 °C, respectively) in 3.5% NaCl solution at room temperature. Electrochemical impedance spectra a Nyquist plots at OCP, b Bode plots at OCP, and c electrical equivalent circuit

Full size image
Table 4 The values of electrical equivalent circuit (EEC) elements fitted with the EIS data of the specimens after 1-h immersion in 3.5% NaCl solution at OCP
figure a
Full size table

3.3 Cyclic polarization

Figure 6 shows the cyclic polarization plots for the three different specimens. The cyclic polarization plot gives us an idea about the pitting behavior of the material. It is well established that 304L-type austenitic steels are prone to the pitting corrosion in the presence of sea water (3.5% NaCl) [2, 35, 38]. The cyclic polarization test carries the information about the pitting protection potential (Epp), the breakdown or transpassive potential (Eb), and passive current density (ipass). During the forward potential scan in the anodic direction, the current density does not increase significantly or remains almost a constant in the case of the SZ-825 sample, until a threshold potential is reached, as shown in Fig. 6. A protective surface film forms during polarization at this potential window. Thickening of passive film is anticipated as the potential increases which could be associated with accumulation of point defects and incorporation of chloride ions in the passive layer. After reaching a threshold potential, the current density increased sharply indicating a shoulder or knee in the polarization plot. This shoulder potential is denoted as passivity breakdown potential (Eb) beyond which the passive film is unstable and is not protective. Pits are considered to initiate and grow when polarization continued above this potential. After reaching a predetermined current density (5 mA/cm2 in this study), the direction of the polarization is reversed. If the pits were quickly passivated, the reverse log I-V plot would trace the forward log I-V plot without showing any hysteresis. On the other hand, if active dissolution was occurring within the pits, then the current density during the reverse scan would be higher than that of forward scan for a given potential. This would show a hysteresis behavior as shown in Fig. 6. After reaching a critical potential during the reverse scan, the pits repassivate completely and exhibit a lower current density than that of the forward scan. At this potential, the reverse log I–V plot crosses over the forward plot. The crossover potential is denoted as pitting protection potential (EPP). It is noted that the EPP shows negative potentials and is closer to the corrosion potential. The EPP value will change as a function of apex current density (at which the polarization direction is reversed), scan rate, and pH of the solution. Increase in apex current density, and decrease in pH of the solution will result in more negative EPP values. More positive EPP values could be observed for shallow pits since these can be repassivated faster than pits with high aspect ratio. The EPP is the potential below which no new pits can initiate. At potentials more positive than the Eb or breakdown potential, the pits can form and grow exponentially causing a severe degradation of the material [39]. Passive current density (ipass) provides the information about the passivity of the specimens; the lower the current density, the better is the passivity and higher the corrosion resistance. The exchange current densities of oxygen reduction reaction and iron dissolution were calculated from the cathodic and polarization plots, respectively. The i0,ORR was estimated by extrapolating the linear portion of the cathodic polarization to the redox potential of the oxygen reduction reaction (E0 ORR = 0.7055 VAg/AgCl). Here, oxygen reduction was the only reaction considered to occur as the cathodic reaction on the specimen surface. The anodic reaction is the dissolution of the stainless steel. However, for the exchange current density estimation, it is assumed that iron is the predominantly dissolving element, and therefore the linear portion of the anodic plot was extrapolated to the redox potential of Fe → Fe2+  + 2e−; E0 = − 0. 64 VAg/AgCl. The exchange current densities give an idea about the general corrosion behaviors of the materials. Higher exchange current density indicates higher activity for corrosion.

Fig. 6
figure 6

Results of cyclic polarization tests of 304L base metal, and friction-stirred zone at isothermal tool temperatures at 825 °C and 725 °C samples (referred to as 850 °C and 725 °C, respectively) in 3.5% NaCl solution at room temperature

Full size image

Table 5 summarizes the values of Epp, Eb, ipass, and i0 for three different samples derived from the cyclic polarization data. This comparison revealed that among the three specimens investigated, the base metal shows the lowest Epp and the SZ-725 °C sample the highest Epp. The Epp value of base metal (− 152 mV) was similar to the previously reported values [40]. The EPP values observed on all three specimens were about 200 mV more positive than that of their corrosion potentials in the 3.5% NaCl solution. This observation indicated that the presence of oxidizers in the 3.5% NaCl environment that increases the corrosion potential of the 304L would lead to pitting. FSW did not increase the pitting resistance. On the other hand, FSW decreased the breakdown potential (Eb) due to the presence of lazy-S, as shown in Table 5. Both the base metal and SZ-725 specimens showed the same order of passivation current density, while the SZ-825 specimen showed a marginally higher passivation current density than other two cases. The SZ-725 specimen showed an increase in the current density at potentials anodic to − 80 mVAg/AgCl (primary transpassive potential) that reached a peak value of 211 μA/cm2 at 30 mVAg/AgCl. The increase in current density could be due to localized attack on the lazy-S feature of the ZS-725 specimen. The SZ-825 specimen also consisted of a lazy-S feature with a higher surface area than that of the lazy-S of the SZ-725. Therefore, attack on the more localized lazy-S of the SZ-825 specimen which started at potentials more anodic to 40 mV continued monotonically, revealing a transpassive behavior. On the other hand, the SZ-725 specimen repassivated after showing a peak current because of a relatively smaller area and less localized nature of the lazy-S present in the SZ-725 specimen. However, the second passivation showed much higher current density (~ 110 μA/cm2). Because of the secondary passivation, the transpassive potential or passivity breakdown potential of the SZ-725 specimen could be considered as 350 mVAg/AgCl. However, the primary transpassive potential of the SZ-725 was around − 80 mVAg/AgCl as listed in Table 5. It should be noted that the theoretical oxygen evolution potential in the 3.5% NaCl (pH 5.5) is around 700 mVAg/AgCl. Therefore, the increase in the current density at high anodic potentials corresponds to the passivity breakdown and not because of the oxygen evolution reaction. The cyclic polarization results showed only marginal difference in the pitting protection potentials and the transpassive potentials of the specimens. The resistance to pitting corrosion of the specimens followed more or less a similar trend that was observed for general corrosion at the open-circuit condition. The results of polarization resistance and cyclic polarization did not give an unequivocal point of view on how the plastic deformation by FSW process affected the corrosion resistance of the material. On the other hand, the exchange current densities for oxygen reduction reaction (ORR) and iron oxidation exhibited some dependency of plastic deformation. For example, the i0,ORR and i0,Fe of SZ-825 C were two and three orders of magnitude larger than those of base metal, respectively. The ORR activity of SZ-725 was similar to that of base metal, while the i0,Fe was similar to that of SZ-825 C. Therefore, it can be concluded that heavy plastic deformation increased the activity for iron oxidation by three orders of magnitude when compared to that of annealed base metal.

Table 5 Summary of the cyclic polarization results of the base metal, SZ at 825 °C, and SZ at 725 °C
Full size table

In order to better understand the effect of microstructural changes occurred in the FSW process, the results of potentiostatic tests are presented in the next section.

3.4 Potentiostatic passivation

Potentiostatic tests are carried out at a constant potential of 0 V vs Ag/AgCl on each sample in order to understand the behavior of the passive film forming on the surface when immersed in the 3.5% NaCl solution. The 0 V vs Ag/AgCl was selected for the potentiostatic conditioning because this potential was more positive than the pitting protection potentials (Epp) of all the samples, and therefore pitting can be easily initiated. The time taken for pit initiation may be used for ranking the materials prepared under different conditions. In the potentiostatic tests, the current was recorded as a function of time. Formation of an impervious passive film resulted in a current decay behavior. The presence of any current spikes indicated events of localized rupture of the passive film. Instead of decaying, if the current started to rise, then pit initiation and growth were considered. Figure 7a–c shows the I-t plot recorded during the potentiostatic test. An initial decay of current was observed for the first few hundred seconds in all the specimens. The absence of abrupt current spikes during the initial decay period indicated that the passive film was stable. After the initial passivation, the current slowly started increasing. If the current density was higher than 20 µA/cm2, pit initiation was considered to have occurred. This value of current density was chosen to signify pit initiation because the cyclic polarization tests showed that the passivation current density was typically around 10 μA/cm2. When the current density reached about 80 μA/cm2, the test was interrupted and EIS and Mott–Schottky analyses were carried out to evaluate the electrochemical properties of the passive film after pit initiation. After these tests, the potentiostatic conditioning continued until reaching a current density of 180 µA/cm2 or more, which considered as pitting growth stage. The pit initiation (to reach 80 μA/cm2) and growth (to reach 180 μA/cm2) times were recorded for each sample and presented in Table 6. The base metal specimens showed longer pit initiation time than other two specimens. Pit growth occurred faster in the SZ-725 C specimen than in other two cases, whereas the SZ-825C showed the longest time required to reach a current density of 180 μA/cm2 at 0 V Ag/AgCl at room temperature.

Fig. 7
figure 7

Results of potentiostatic tests at 0 VAg/AgCl in 3.5% NaCl solution at room temperature. a Base metal, b stirred zone (SZ) under 725 °C isothermal condition, and c SZ at 825 °C

Full size image
Table 6 The pit initiation time and pit growth time at 0 VAg/AgCl for the base metal, 825 °C and 725 °C friction stir crack-repaired samples
Full size table

EIS and Mott–Schottky analysis were performed in pre-potentiostatic (after stabilizing at the OCP and conditioning at 0 VAg/AgCl), pit initiation (after reaching 80 μA/cm2), and pit growth conditions in order to determine the defect structure of the passive film. The EIS spectra (Nyquist plots) of the specimens are shown in Fig. 8a–c. The corresponding Bode plots are shown in Fig. 9a–c. A general fitted electrical equivalent circuit is shown in Fig. 8d, but not all elements were present in each test condition. The impedance of the base metal specimen decreased at every step of evaluation as shown in Fig. 8a and Fig. 9a. The frequency at which the peak in the phase angle occurred shifted to higher values as the pit initiation and growth were observed as shown in Fig. 9a. The range of frequencies for capacitor-like behavior decreased with the pit initiation and pit growth. Table 7 summarizes the values of the EEC components. The EIS spectra of the base metal specimens could be fitted with a single EEC of a porous film type with two RQ loops. The Q1 value representing the interfacial capacitance decreased during pit initiation and growth as compared to that of the starting condition. On the other hand, Q2, which is associated with the defect structure, increased with the increase in time under potentiostatic conditioning. At very low frequencies, the effective resistance could be given as summation of all the resistor values. The effective resistance decreased with an increase in the exposure time which indicated degradation of the passive film or passivity breakdown. Figures 8b and 9b show EIS spectra of the SZ-725 °C specimens conditioned at 0 VAg/AgCl at three different time intervals. The impedance increased when the current density reached a value of 80 μA/cm2 in contrast to that observed with the BM specimens. The increase in the impedance indicated that the passive film of the SZ-725 °C specimen was not impaired in spite of the increase in the current density. The observed current density was attributed to the dissolution of the lazy-S region. Microstructure of the specimen observed after the test supported this claim which will be discussed in a later section. The EIS data were fitted with an EEC, and the values of the EEC elements are summarized in Table 8. The EEC of the potentiostatic condition is different from that of the OCP condition because it includes an inductor (L1) element in the RQ loop. The inductor represents the adsorption of chloride ions in the defect structures. Point defects such as anion vacancies (VO2+) could be preferential sites for chloride ion adsorption following the reaction [41]:

$$V_{{\text{O}}}^{2 + } \, + \,{\text{Cl}}^{ - } \, \to \,{\text{Cl}}_{{\text{O}}}^{ + }.$$
(2)
Fig. 8
figure 8

Results of Nyquist plots (potentiostatic EIS data) at 0 VAg/AgCl in 3.5% NaCl solution at room temperature in three different time intervals such as at t = 0 (pre-potentiostatic), after pit initiation, and pit growth: a base metal, b stirred zone (SZ) under 725 °C isothermal condition, c SZ at 825 °C, and d equivalent electric circuit (EEC) fitting the EIS data

Full size image
Fig. 9
figure 9

Results of Bode plots (potentiostatic EIS tests) at 0 VAg/AgCl in 3.5% NaCl solution at room temperature in three different time intervals such as at t = 0 (pre-potentiostatic), after pit initiation, and pit growth: a base metal, b stirred zone (SZ) under 725 °C isothermal condition, c SZ at 825 °C, and d equivalent electric circuit (EEC) fitting the EIS data

Full size image
Table 7 The values of electrical equivalent circuit (EEC) elements fitted with the EIS data of the base metal (BM) specimens after potentiostatic tests at 0 VAg/AgCl in 3.5% NaCl solution
Full size table
Table 8 The values of electrical equivalent circuit (EEC) elements fitted with the EIS data of the SZ-725 °C specimens after potentiostatic tests at 0 VAg/AgCl in 3.5% NaCl solution
Full size table

The presence of an inductor in the EIS of FSW specimens but not in the BM specimens indicated that the lazy-S defect played a crucial role in the adsorption phenomenon. The adsorption has a unit of ohm.s, and therefore, at low frequencies the impedance due to an inductor is very low. At low frequencies, the impedance is determined mainly by the resistor elements. The impedance values of the SZ-725 °C increased during ‘pit-initiation’ and decreased significantly during the pit growth condition.

Figures 8c and 9c show EIS spectra of the SZ-825 °C specimens under potentiostatic conditioning at 0 VAg/AgCl. The impedance behavior followed a similar trend of that observed with the SZ-725 °C specimens, where the impedance increased during the ‘pit-initiation’ period probably because the dissolution of the lazy-S zone did not result in passivity breakdown. When the current density was ~ 200 μA/cm2, the impedance decreased significantly. Table 9 lists the values of the components of EEC fitted with the EIS data. The impedance of the SZ-825 °C specimen was lower than that of SZ-725 °C under each test condition.

Table 9 The values of electrical equivalent circuit (EEC) elements fitted with the EIS data of the SZ-825 °C specimens after potentiostatic tests at 0 VAg/AgCl in 3.5% NaCl solution
Full size table

Figure 10a–c shows the Mott–Schottky (M-S) plots of the BM, SZ-725 C, and SZ-825 C specimens carried out at different time intervals during the potentiostatic conditioning at 0 VAg/AgCl. Passive films of the specimens showed an n-type semiconductivity because of the positive slopes observed in the M-S plots. The intercept of the linear region at 1/C2 = 0 could be considered as the flat band potential. It is noted that the flat band potential did not change with the increase in the exposure time in all the specimens. The flat band potential gives an idea about the Fermi energy level of the passive film, and the potential at which no band bending occurs and no space charge layer is present. The information on flat band potential, along with the band gap energy of the passive film, and charge carrier density, can be used to estimate the stability of the passive film under anodic or cathodic polarization conditions [42]. The energy level of the conduction band minimum (CBM) of the passive film can be calculated based on the flat band potential [43]. The energy level of the CBM determines the energetics of water reduction reaction which is one of the cathodic reactions of stainless steel corrosion. Negative shift of the flat band potential will enhance the water or hydrogen reduction reactions. Charge carrier density of the passive film is calculated from the slope of the linear portion (closer to the film formation potential) and listed in Table 10. The FSW specimens showed less negative flat band potential than that of base metal. The charge carrier density of the passive film formed on the BM specimen was less than that of FSW specimens under identical test conditions and was in the range of 3.7–6.1 × 1020 cm−3. The charge carrier density increased with an increase in the exposure time for all the specimens indicating aging effect of the passive film. The charge carrier density of the SZ-725 C increased from 5.6 × 1020 cm−3 to 6.0 × 1020 cm−3 after about 30 min of conditioning at 0 VAg/AgCl and to 7.2 × 1020 cm-3 after about 4200 s of total exposure time. The SZ-825 °C specimen also showed a marginally higher carrier density than that of the SZ-725 °C specimen in each test condition.

Fig. 10
figure 10

Mott–Schottky plots of specimens conditioned at 0 VAg/AgCl in 3.5% NaCl solution at room temperature. a Base metal, b stirred zone (SZ) under 725 °C isothermal condition, and c SZ at 825 °C

Full size image
Table 10 Charge carrier densities and flat band potentials of the passive film formed under potentiostatic conditioning at 0 VAg/AgCl at different time intervals
Full size table

Figure 11a–c shows the optical microstructures of the specimens after potentiostatic conditioning. The specimens were slightly etched to reveal the grain boundaries so that the pit initiation sites could be revealed. It is seen from Fig. 11a that the pits were initiated inside the grains in the case of the base metal. Figure 11b and c reveals preferential attack on the lazy-S zones of the SZ-725 °C and SZ-825 °C specimens, respectively. Figure 12a–d shows the SEM images of the microstructures of the specimens under similar test conditions as that of Fig. 11. Delta ferrite stringes are observed in the BM specimens as shown in Fig. 12b. It is noted that pits initiated inside the grains randomly as shown in Fig. 12a. Pit initiation was not observed either inside the delta ferrite stringes or at the interface between the ferrite/austenite. Therefore, compositional gradient of chromium or other ferrite-stabilizing elements did not deleteriously affect the pitting resistance of this material. Pit-like features could be observed on the lazy-S zone of the SZ-825 °C specimen as illustrated in Fig. 12c and d. Figure 13a and b shows the optical micrographs of the SZ-825 °C specimen before and after potentiostatic conditioning. The pit size and distribution could be seen at a current density of 200 μA/cm2 in the SZ-825 °C.

Fig. 11
figure 11

Optical micrographs of specimens after conditioned at 0 VAg/AgCl in 3.5% NaCl solution at room temperature. a Base metal, b stirred zone (SZ) under 725 °C isothermal condition, and c SZ at 825 ºC. The samples were slightly etched to reveal the grain boundaries

Full size image
Fig. 12
figure 12

SEM micrographs of specimens after conditioned at 0 VAg/AgCl in 3.5% NaCl solution at room temperature. a Base metal, b base metal showing no pitting at delta ferrite stringes or ferrite–austenite interfaces, c preferential attack on the lazy-S region of the SZ at 825 °C, and d higher magnification of the lazy-S region shown in figure (c). The samples were slightly etched to reveal the grain boundaries

Full size image
Fig. 13
figure 13

Optical micrographs of SZ-825 °C specimen before and after potentiostatic conditioning at 0 VAg/AgCl in 3.5% NaCl solution at room temperature. a Before testing, and b after pit growth

Full size image

4 Discussion

The experimental results discussed above indicate that overall corrosion resistance of the FSW specimens has significantly reduced compared to that of the base metal 304L SS. The polarization resistance (Rp) values obtained from both linear polarization and electrochemical impedance spectroscopy is a strong indicator of the base metal being more corrosion resistance than the other two samples. Grain size is one of the most important microstructural features that affects the corrosion properties in stainless steels. Grain refinement in 304L-type austenitic stainless steel was found to improve corrosion resistance in many studies [7, 44,45,46,47]. Sabooni et al. observed that the grain refinement did not affect the corrosion potential (Ecorr) of 304L SS in 3.5% NaCl solution, but the corrosion rate (icorr) and passive current density (ipass) were significantly reduced and Rp increased with the smaller grain size [44]. A widely posited theory behind the enhanced corrosion resistance with the grain refinement is that increased grain boundary area with the smaller grain size promotes a better diffusion passage for the chromium to the formation of chromium-rich passive film on the surface. Contrary to the findings of the improved corrosion resistance on grain refinement in 304L SS, we have observed that the smaller grain size samples, both SZ-725 °C and SZ-825 °C (1–4 µm), showed a lower corrosion resistance (Rp of 3–7 kohm) compared to the coarse-grained (~ 44 µm) base metal (Rp ~ 16 kohm). The factors such as grain boundary character [8, 48], the presence of delta ferrite in austenitic stainless steel [49, 50], and residual stress [51, 52] which result from different thermomechanical routes during processing stainless steel are widely considered to control the corrosion resistance. Singh et al. observed that below a critical grain size (~ 16 µm), the grain boundary structure did not play any significant role in improving the corrosion resistance [53]. Therefore, the lower polarization resistance (Rp) value in fine-grained SZ-725 °C and SZ-825 °C samples could be attributed to the attack on the lazy-S zone. A comparison between two different grain sizes, 15 µm and 38 µm, showed that the coarser grain size had a higher pitting resistance [7]. In the same study, it was observed that the corrosion potential was not affected by the grain size. Similar observation was made by Atapour et. al, who reported that the ultrafine-grained 304L SS (500 nm grain size) had much higher pitting resistance than that of the larger grain size (1–12 µm) ones, but the grain size reduction had a negligible effect on the corrosion potential [54]. A statistical method was employed by Aghug and the coworkers [55] in order to investigate the effect of grain size on the pitting resistance in 304L SS (5–28 µm) in 3.5% NaCl solution. They reported that the grain size had no effect on the pitting potential, but the grain refinement reduced the frequency of metastable pit formation. However, the smaller grain size increased the kinetics of metastable pit dissolution and the probability of the metastable pits to grow into a stable pit. Our study shows that the specimens with the smallest (SZ-725 °C) have the most noble pitting or breakdown potential (Eb), but the SZ-825 °C which has a relatively fine grain size (~ 4 µm) shows the most active pitting potential, whereas the coarse-grained base metal (~ 50 µm) has an intermediate Eb value. Therefore, general and pitting corrosion resistance of the austenitic stainless steels cannot be controlled by just controlling the grain size. The grain boundary structures such as low-angle grain boundaries and special grain boundaries (coincidence site lattices) are considered to improve the general and pitting corrosion resistance of the austenitic stainless steels. The presence of Ʃ3 grain boundaries in the microstructure of austenitic stainless steels was considered to enhance the pitting resistance [56]. These grain boundaries are known as special grain boundaries with the low interfacial energies and reported to inhibit the pitting corrosion in other alloys as well [57]. It was shown that the 304 SS with high fraction of Ʃ3 grain boundaries and lower fraction of random high-angle grain boundaries (RHAGB) exhibited a superior pitting resistance to the 304 samples having lower Ʃ3 grain boundary fraction [7]. The pits formed preferentially on the RHAGB. Our results show pit formation on the grain interior for the base metal. The special grain boundaries not only improve the pitting resistance but also the intergranular and general corrosion in the austenitic stainless steels [58, 59]. Our result shows that the Ʃ3 grain boundary fraction has significantly reduced to 1–1.5% in the SZ-725 °C and SZ-825 °C samples from the 48% in the 304L base metal. In line with the idea discussed before, it can be concluded that the lower Ʃ3 grain boundary fraction detrimentally affected the corrosion resistance properties of SZ-725 °C and SZ-825 °C samples. Along with the lower Ʃ3 grain boundary fraction, the presence of the lazy-S regions further lowered the corrosion resistance in those samples. On the other hand, the FSWed specimens had a high fraction of low-angle grain boundaries that counteracted the detrimental effect of the lazy-S zones. As discussed in the earlier section, the linear array of dislocations that constitute low-angle boundaries could act as misorientation dislocations at the metal/oxide interface and promote formation of oxide without stressing the substrate. Therefore, low-angle boundaries will be helpful for formation of stress-free passive films that will be more stable and enhance both pitting and stress corrosion cracking resistance.

5 Conclusions

Friction stir welding was carried out on 304L coupons with a simulated crack at two different tool temperatures such as 725 and 825 °C. Corrosion behavior of the friction stir repair-welded specimens of 304L SS was evaluated in the 3.5% NaCl solution and compared with the base metal by carrying out cyclic polarization, chronoamperometry, and electrochemical impedance spectroscopy. Based on the results, the following conclusions could be drawn:

The base metal of 304L SS had a mean grain size of 47 ± 16 μm. The grain size decreased to 2.4 ± 0.9 and 4.2 ± 1.4 μm upon friction stirring at 725 and 825 °C respectively. The base metal had a high fraction (~ 50%) of special grain boundaries such as Σ3 and Σ9. The friction stirring significantly reduced the special boundaries (< 10%), but increased fraction of low-angle grain boundaries from 25 to 37–43%. Furthermore, a preformed oxide layer that was present on the artificial crack surfaces was broken during the friction stirring and agglomerated as a spiral defect structure called ‘lazy-S’ due to its S shape.

These complex changes in the grain size and grain boundary character and the presence of the lazy-S defect structure after friction stirring negatively affected the pitting resistance of the specimens.

The FSW specimens showed lower polarization resistance and lower transpassive potential than those of the base metal specimens. However, the pitting protection potentials of the FSW specimens were higher than that of the base metal.

The pitting behavior of the FSW specimens was influenced more by the preferential attack on the lazy-S region than by passive film breakdown.

The flat band potentials of the passive film formed on the FSW specimens were more positive than that of base metal specimen which could lead to sluggish water or hydrogen reduction reaction as a cathodic reaction for the overall corrosion process. The charge carrier density of passive film formed on the FSW specimens was higher than that of the base metal.

The higher fraction of low-angle grain boundaries present in the FSW specimens could supply more number of misorientation dislocations at the metal/film interface which could form anion vacancies by a climb process leading to the formation of oxide at these locations without stressing the substrate. Therefore, low-angle boundaries will be helpful for formation of stress-free passive films that will be more stable and enhance both pitting and stress corrosion cracking resistance.