Microstructural characteristics of different heat-affected zones in welded joints of UNS S32304 duplex stainless steel using the GMAW process: analysis of the pitting corrosion resistance

1 Introduction

It is widely recognized that good corrosion resistance and high mechanical strength are achieved in duplex stainless steels (DSSs) when the fraction between the δ/γ phases is approximately one (Ghosh and Mondal 2008; Hertzman et al. 1997; Liou et al. 2002; Verma and Taiwade 2017; Wang et al. 2006). However, the thermal history, for example, in welding processes, causes a significant change in the balance between both phases. This imbalance of the phases may severely impair the properties, especially local pitting corrosion resistance, at the heat-affected zone (HAZ) and fusion zone of a welded joint (Jiang et al. 2013; Ogawa and Koseki 1989; Pettersson et al. 2019). A determined amount of reformed austenite is necessary to avoid the formation of detrimental chromium nitrides to retain a high resistance to corrosion and favorable properties in a welded structure, the weld metal and the HAZ (Karlsson et al. 1995). The microstructure and the properties of the weld metal are generally controlled by adjusting the filler material composition and heat input. The microstructure in the HAZ is determined by the welding thermal cycle and therefore is very sensitive to variations in the welding parameters (Hemmer and Grong 1999). During the DSSs welding procedure, the HT (high temperature)-HAZ usually solidifies fully as ferrite, and austenite precipitates from ferrite during cooling (Chehuan et al. 2014).

High cooling rates, associated with lower heat inputs in the welding procedure, will result in an excessive amount of ferrite in the heat-affected zone (HAZ) and consequently, the chromium nitride precipitation (Brytan et al. 2016; Karlsson 2012; Ramirez et al. 2004). The main consequence of this nitride precipitation is a notable degradation of ductility, toughness, and corrosion resistance (Lippold and Kotecki 2005).

On the other hand, slow cooling rates associated with higher heat inputs in the welding procedure, favor the recovery of the phase ratio (more reformed austenite) and ferrite grain growth (Chen and Yang 2002). Also, low cooling rates can favor the precipitation of intermetallic phases (σ, χ), although according to the lean DSS welding standards, these detrimental phases rarely occur (Chen and Yang 2002).

The pitting corrosion occurs in small localized areas (pits) on the metal surface. Once pitting corrosion initiates, it grows, leading to structural damage. Hou et al. (2022) studied the initiation mechanisms of pitting corrosion in the heat-affected zone (HAZ) of a commercial UNS S 31803 duplex stainless steel containing 0.004 % Ti. They found that most of the pitting corrosion occurred in the HT (high-temperature)-HAZ. In this study both Cr₂N and CrN were present in the ferrite phase, along with a Cr-depleted zone in the ferrite matrix adjacent to these Cr nitrides. However, the occurrence of pitting corrosion caused by the precipitation of CrN and Cr₂N could not be confirmed in this study. Sathirachinda et al. (2010) employed a scanning Kelvin probe force microscope (SKPFM) to measure the Volta potential differences between the Cr₂N precipitates and matrix in a super-duplex stainless steel (UNS S32750). The purpose of this investigation was to assess the effect of Cr₂N on the steelʼs resistance to pitting corrosion (Sathirachinda et al. 2010; Sathirachinda et al. 2011). They found that the presence of fine Cr₂N precipitates with a size smaller than 100 nm did not significantly affect the pitting corrosion of the steel when it was water-quenched. However, they also noted that larger nitrides could have a detrimental effect on the corrosion resistance due to a Cr-depleted zone that forms around them. Bettini et al. (2013) reported a similar result for UNS S32205 duplex stainless steel. Numerous previous studies have suggested that the Cr-depleted zone around the Cr₂N is a contributing factor to the initiation of pitting corrosion. (Kaneko et al. 2011; Nakamichi et al. 2008; Pervushin and Suito 2001; Yang et al. 2012; Zhang et al. 2016). Ito et al. (2022) using EC-AFM showed that the initiation site of the metastable pit is the chromium-depleted area around the chromium nitride (Cr₂N) at the HAZ of a UNS S32101 lean duplex stainless steel. This study demonstrated that chromium-depleted area exists in the vicinity of the precipitated chromium-nitrides (Cr₂N). Besides, this depletion was particularly remarkable when the chromium nitrides (Cr₂N) precipitate at the α/γ grain boundaries.

To evaluate the susceptibility of the different phases or regions to pitting corrosive attack in duplex stainless steels is common to use the pitting resistance equivalent number (PREN). PREN is a theoretical way of comparing the pitting corrosion resistance of the different phases or regions, based only on their chemical compositions. (Atamert and King 1991; Garfias-Mesias et al. 1996; Tan et al. 2009; Yang et al. 2013; Zhang et al. 2012). As a result, the Cr-depleted areas surrounding nitrides exhibit lower PREN values, leading to greater susceptibility to pitting corrosion compared to the Cr-rich regions within the ferritic matrix. Additionally, modifications in the volumetric proportions of the constituent phases can affect the concentration of their stabilizing elements, consequently influencing the PREN value of each phase. The pitting corrosion resistance in DSSs has been extensively studied using different techniques, such as gravimetric tests (ASTMG48), potentiodynamic polarization techniques (ASTMG5), critical pitting temperature (CPT) measurements (ASTM G150), transmission electron microscopy, atomic force microscopy (AFM), Auger electron spectroscopy (AES), among others (Cervo et al. 2010; Massoud et al. 2013; Maurice and Marcus 2012, 2018; Maurice et al. 1996, 1998; Strehblow et al. 2011). Most of these studies have been carried out on pure metals, simple alloys and annealed DSSs. However, an in-depth understanding of the development of pitting corrosion in alloys with multiple alloying elements still needs to be included. It is of particular interest the regions where different thermal cycles have operated.

The knowing on pitting corrosion in the regions of the heat-affected zone (HAZ) in duplex stainless steel (DSS) is limited, and previous efforts aimed at addressing this issue have focused on the formation of a simple HAZ for a given heat input. Several studies have examined the formation of a simple heat-affected zone (HAZ) for a given heat input through thermal cycle simulations using thermo-mechanical simulators or numerical methods. However, in actual gas metal arc welding (GMAW) processes that employ a high argon mixture, the weld bead takes on a unique shape. A distinctive inflection occurs in the fusion line, resulting in the formation of distinct regions within the HAZ. Previous studies have not considered the formation of these HAZ regions in UNS S32304 lean duplex stainless steels in relation to the shape of the weld bead. This study aims to examine how the microstructural characteristics of three distinct regions in the HAZ affect the pitting corrosion resistance of the lean duplex stainless steel welded joints, after undergoing the GMAW process.

2 Materials and methods

Single pass welds were made on 300 × 150 × 10 mm lean DSS plates by the GMAW process. A Sandvik solid wire 22.8.3.L with 1.2 mm diameter (equivalent to AWS ER2209) was used as filler metal. Table 1 shows the chemical composition of the UNS S32304 lean DSS and the filler metal.

The GMAW process was performed under reverse-polarity direct current in the constant current mode. An 98 % Ar + 2 % CO₂ shielding gas was used in the welding procedure. The welded joints were made with two heat input (Q) levels (1.5 and 2.5 kJ/mm, respectively). The details of welding parameters are given in Table 2.

Cross-sectional specimens of the welded joints were sanded with abrasive papers of successive grades up to 1200 grit and then polished with 1 µm diamond paste. The samples were then chemically etched by immersion for 30 s in a modified Beraha reagent (5 mL·HCl, 35 mL·H₂O, 0.18 g·K₂S₂O₅). The microstructures of the specimens were examined by optical and scanning electron microscopy (OM and SEM). The average fractions of austenite and ferrite at the three HAZ regions were determined by digital image processing (ImageJ software) in approximately 20 optical micrographs.

Thin-foil samples for TEM were prepared by cutting a 200 mm-thick slice from the two regions of the HAZ for each heat input. From these samples, 3 mm diameter disks were punched. The disks were electropolishing in a Tenupol-5 apparatus, using a solution of 10 % perchloric acid and 90 % ethylic alcohol at 0 °C and 20 V.

Similar simulated thermal cycles were performed using a Gleeble 3800 thermo-mechanical simulator to model the regions of the real welded HAZ. The applied thermal cycles are schematically shown in Figure 1. A peak temperature of 1350 °C was reached at a rate of 350 °C/s. A holding time of 2 s at this temperature followed by five cooling rates up to 250 °C to simulate the HAZ with varying heat inputs.

Figure 1:

Welding simulated thermal cycles corresponding to five heat inputs (Q).

Potentiodynamic anodic polarization tests were performed on simulated HAZ specimens with two heat inputs: 1 and 3 kJ/mm. Simulated HAZ specimens was used because it was not possible to obtain actual specimens of the correct dimensions from all three regions of the HAZ in the welded joints for the potentiodynamic polarization test.

The potentiodynamic anodic polarization measurements were carried out in a glass three-electrode cell containing 3.5 wt.% NaCl solution at 25 and 60 °C. To simulate the operational conditions of pipelines that use this duplex stainless steel (DSS), the test was conducted at a temperature of 60 ± 1 °C. A platinum wire and a saturated calomel electrode (SCE) were used as the counter and reference electrodes, respectively. The specimens of the simulated HAZs and base metal acting as working electrode were embedded in epoxy resin with an exposed working area of 0.25 cm². Prior to each electrochemical measurement was conducted, the working electrode was sanded with abrasive papers of successive grades up to 1200 grit and then polished with 1 µm diamond paste. Then, the polished working electrode was degreased with ethanol, rinsed with distilled water and dried. Also, before each corrosion test, the open circuit potential (OCP) was recorded for 30 min to stabilize the corrosion potential. The potentiodynamic polarization measurements were carried out at a scan rate of 1 mV/s from the OCP to 1000 mV SCE above the OCP. To ensure the reproducibility of the results, experiments were repeated at least three times under the same experimental condition.

The pitting corrosion of the welded specimens was studied using ASTM G48 gravimetric test (ASTM Standard G48-03 2003). After welding, six previously polished coupons of 20 × 15 mm (for each heat input) were initially polished and immersed in a ferric chloride solution (100 g·FeCl₃·6H₂O in 900 mL·H₂O) at 23 °C for 72 h. After this immersion, specimens were rinsed with water, dipped into acetone in an ultrasonic cleaner, and air-dried. Subsequently, the coupons were examined to see the locations of the pits. To show the place where the pits originate, the microstructure of the widest HAZ region corresponding to welded joints with a heat input of 1.5 kJ/mm was revealed. The procedure consisted of immersing the previously polished specimens in the ferric chloride solution for 24 h. After this immersion, the specimens were carefully polished with 1 µm diamond paste, immersed in acetone in an ultrasonic cleaner and lightly etched with Beraha reagent to reveal the microstructure of the widest HAZ region.

Vickers microhardness measurements were performed randomly on the different morphological forms of austenite and ferrite grains in the selected HAZ regions of the welded specimens with 1.5 and 2.5 kJ/mm as heat inputs. The specimen surfaces were carefully polished before indenting. A load of 0.010 kg (HV0.010) was applied for 10 s. The load of 0.010 kg was used to avoid the grain boundary effects on the hardness values due to the small size of the austenite morphologies. Between 7 and 10 impressions were used for each microhardness determination.

3 Results

Figure 2 shows the macrographs of the single pass welds cross-section with heat inputs of 1.5 and 2.5 kJ/mm, respectively. A single-pass weld was performed to avoid the formation of secondary austenite and unwanted intermetallic phases that can occur in multi-pass welds. The inflection of the fusion line can be seen in the macrographs originating the three HAZ regions. These regions have been indicated in Figure 2 as A, B, and C depending on their location and thickness in the HAZ.

Figure 2:

GMAW single pass welds of the UNS S32304 DSS with (a) 1.5 kJ/mm and (b) 2.5 kJ/mm as heat inputs. The three HAZ regions (A, B and C) are schematically represented in the welded joint with heat input of 2.5 kJ/mm.

Figure 3 shows the original and digitally processed optical light images of the three HAZ regions of the welded joints with both heat inputs. It is observed in the micrographs that the selected HAZ regions have different ferrite grain sizes and thicknesses since the three regions are subject to distinct cooling rates. In all the HAZ regions of the welded joints with both heat inputs, the three austenite morphologies were observed, the grain boundary austenite (GBA), Widmanstätten austenite (WA), and intragranular austenite (IGA). No partially dissolved austenite was observed in these high-temperature HAZ regions, indicating that the austenite was formed from a fully ferritized matrix.

Figure 3:

Original and digitally processed optical light micrographs of the three HAZ regions in the welded joints with both heat inputs. The GBA, IGA and WA are highlighted by yellow, red and blue colors, respectively.

The austenite fractions were quantified based on their morphologies in the digitally processed optical images at each region of the HAZ. Figure 4 shows the austenite fractions considering their morphological forms, in each HAZ region of the welded joints with heat inputs of 1.5 and 2.5 kJ/mm respectively. The average ferrite grain sizes and HAZ thicknesses in each region are shown in Table 3.

Figure 4:

Fractions of the austenite morphologies at the HAZ regions of the welded joints performed with the GMAW process with 1.5 kJ/mm and 2.5 kJ/mm as heat inputs.

Figure 4 shows that the amount of austenite in the HAZ regions of the single-pass welded joints using the above parameters in the GMAW process strongly depends on the cooling rate. The GBA determines the total austenite amount in each HAZ region. The widest HAZ region (B) has the largest ferritic grain size and, therefore, the smallest GBA fraction. Conversely, this HAZ region has higher IGA and WA fractions.

Figure 5 shows the microhardness values of each austenite morphological form and ferrite in the HAZ regions of the welded joints. In this Figure is appreciated that the GBA hardness in region C is lower than that of region B. The hardness values of the IGA and WA depend on their volume fractions in each HAZ region for both heat inputs. The GBA hardness values at the three HAZ regions differ more with increasing the heat input. In addition, the hardness of the morphological forms of the austenite in all HAZ regions are higher than the hardness of the ferritic matrix for both heat inputs. A slight increase in the hardness of the ferrite corresponding to the C region with the lowest heat input can be perceived.

Figure 5:

Microhardness values of each austenite morphological form and ferrite in the HAZ regions of the welded joints.

To obtain the cooling times and cooling rates between 1200 and 800 °C in the HAZ regions of the welded joint (narrow regions), equivalent simulated HAZs of the same DSS were performed with a Gleeble simulator, where the thermal cycles are accurately determined. It is known that the ferrite grain growth is controlled by the dissolution of austenite during heating (Atamert and King 1992). As the spacing between the austenite grains is small in the base metal of the studied lean DSS (as-received condition), the ferrite grain growth is insignificant during heating. Also, the high heating rates to reach the peak temperature (∼1350 °C determined by the ThermoCalc software for this alloy) and the short holding time at this temperature do not allow an appreciable growth of the ferritic grain. Therefore, the differences in the ferritic grain sizes at the HAZs of the real and simulated welds are determined by the cooling rate from the peak temperature up to approximately 1200 °C. The subsequent growth of the ferritic grain is hindered at 1200 °C by the GBA formation. Hence, it is justified that similar grain sizes between a HAZ region in the welded joint and a simulated HAZ (for a given heat input) have similar cooling times regardless of how the heat was dissipated. Knowing the correlation between the ferrite grain sizes and cooling times (Δt_12/8) at simulated HAZs, the cooling rates and cooling times in each HAZ region of the real welded joint can be determined. Figure 6 shows the average ferrite grain sizes, heat inputs, and cooling times between 1200 and 800 °C corresponding to the simulated HAZs of the UNS S32304 lean DSS.

Figure 6:

Average ferrite grain sizes, heat inputs and cooling times between 1200 and 800 °C at the simulated HAZs of the UNS S32304 lean DSS.

An excellent linear fit resulted from the correlation between Δt_12/8 and the average ferrite grain size <d_δ> corresponding to the simulated HAZs of the UNS S32304 lean DSS:

As described above, the Δt_12/8 values of the three HAZ regions in the welded joints were calculated using equation (1), where the average ferrite grain diameters <d_δ> of each HAZ region were determined experimentally. Table 4 shows the Δt_12/8, Δt_8/5 (cooling time between 800 and 500 °C), and their equivalent heat inputs of the different HAZ regions for both heat inputs.

The specimens of the welded joints were immersed in a ferric chloride solution for 72 h. After immersion, the welded joints showed pitting corrosion in certain regions of HAZ. The macrographs (Figure 7) show that pitting corrosion is located in specific HAZ regions and correlates with each heat inputʼs cooling rate.

Figure 7:

Pitting corrosion at the HAZ regions after 72 h of immersion in a ferric chloride solution. (a) Welded joint specimen with 1.5 kJ/mm as heat input, (b) welded joint specimen with 2.5 kJ/mm as heat input. A scheme of the HAZ regions was made on these macrographs.

The macrograph corresponding to heat input of 1.5 kJ/mm (Figure 7 [a]) shows that the pitting corrosion has occurred in the region B of the HAZ and, to a lesser extent, in the region A. No pitting corrosion was detected in the C HAZ region of the welded joint for this heat input. However, pitting corrosion is observed in all regions of the HAZ when the welded jointʼs heat input is 2.5 kJ/mm (Figure 7 [b]).

Figure 8 shows an optical image of the microstructure corresponding to the B HAZ region of the welded joint, with a heat input of 1.5 kJ/mm after being immersed in the ferric chloride solution for 24 h. It can be seen in this image, that the pits (showed by arrows) are mostly located in the ferrite phase and in the ferrite/IGA interfaces.

Figure 8:

Optical image of the microstructure of the B HAZ region corresponding to the welded joint with a heat input of 1.5 after the immersion in the ferric chloride solution for 24 h. The pits are shown by arrows in the micrograph.

The potentiodynamic anodic polarization measurements at 25 and 60 °C in a NaCl solution showed a slight decrease of both, E‾pit (average pitting potential) and ΔE (average pitting nucleation resistance) in the simulated HAZ with a heat input of 3 kJ/mm, with respect to the simulated HAZ with 1 kJ/mm as heat input. This slight decrease judging by the error bounds can be considered not very significant, Table 5 and Figure 9 (a) and (b). However, the reduction of E‾pit and ΔE is very significant in the simulated HAZs with respect to the base metal. Thereby, the base metal shows higher corrosion resistance that the simulated HAZs with both heat inputs. The pitting nucleation resistance is the difference between the pitting potential and the corrosion potential (ΔE = E‾pit−E‾corr_.) and represents the amplitude of the passive domain. Higher ΔE indicates higher resistance to localized corrosion. Furthermore, as the simulated HAZ specimens, with both heat inputs, showed relatively low fractions of reformed austenite (18 and 22 wt.% respectively), the Cr and Mo contents are more diluted in ferrite by an increase in the volume fraction of this phase with respect to the as-received steel.

Figure 9:

Potentiodynamic anodic polarization curves in a 3.5 wt.% NaCl solution at 25 °C (a) and 60 °C (b) at the simulated HAZs with 1 and 3 kJ/mm as heat input.

Thin films of the B and C HAZ regions corresponding to the welded joint with 1.5 kJ/mm heat input after the GMAW procedure were obtained and analyzed by transmission electron microscopy, TEM. The C HAZ region showed two chromium nitride morphologies. The plate-shaped chromium nitride morphology identified as cubic CrN and the rod-shaped morphology corresponding to the Cr₂N. The cubic chromium nitrides CrN with the plate morphology predominates in this C HAZ region. The B HAZ region only showed chromium nitrides Cr₂N with the rod morphology. These Cr₂N were concentrated in sub-boundaries of the bigger ferrite grains at this HAZ region of the welded joint. Figure 10 shows typical B and C HAZ regions where two precipitate morphologies were observed.

Figure 10:

Plate and rod chromium nitride morphologies in the B and C regions of the HAZ at the welded joint with 1.5 kJ/mm as heat input.

4 Discussion

The austenite nucleates at the δ grain boundaries during continuous cooling from the fully ferritic region. This first austenite, called GBA or allotriomorphic austenite, covers the entire δ–δ boundaries around 1200 °C (Atamert and King 1992; Liou et al. 2002; Wang et al. 2006). The GBA nucleation and growth at ferritic boundaries hinder ferrite grain growth and consume much of the surrounding nitrogen. At lower temperatures, the WA is formed in a wide range of temperatures, between 1000 and 800 °C (Ohmori et al. 1995). The WA formed at higher temperatures better redistributes the alloying elements and grows depending on the ferrite grain size. The WA is fragmented when it nucleates and grows at the lowest temperatures to better alloying elements partition (Ohmori et al. 1995; Wu et al. 2019). The WA fragments were considered as IGA in the present work. At greater undercooling, after the WA formation, the IGA nucleates and grows. These characteristics were observed in the micrographs of the HAZ regions (Figure 3), since in the ferrite grains where WA are in higher volume fraction, there is not a strong presence of IGA and vice versa. The IGA formation requires greater undercooling due to the low nitrogen content in the remaining ferrite and fewer favorable sites for its nucleation.

It is known from the literature (Haghdadi et al. 2018; Morales et al. 2019) that intragranular austenite (IGA) formed at lower temperatures, with increasing cooling rate, develops interfaces with specific orientation relationships (OR) (K–S/N–W). Conversely, interfaces with random orientation relationships will predominate in IGA formed at higher temperatures. The character of the interfaces between the ferrite and different austenite morphologies constitutes a microstructural characteristic that influences the resistance to pitting corrosion at the HAZ of welded joints in DSS.

The behavior shown in Figure 3 is a direct consequence of the cooling rate influence on the ferrite grain size, the GBA, IGA, and WA fractions in each region of the HAZ. In this way, region B, which has the largest δ grain size and, therefore, less boundary area, has the lowest GBA fraction (Ferreira and Hertzman 1992). However, this region that experienced the slowest cooling rate has the highest IGA and WA fractions. Thus, as the GBA fraction predominates in the total amount of austenite, the C region of the HAZ has the highest volume fraction of the γ phase.

The hardness values of the three morphological forms of austenite, Figure 5, depend on the region in which they are found. The GBA hardness is lower in the C region and higher in the B region. This behavior is related with the austenite fractions in each region. In the C region, the GBA fraction is the highest due to the smaller ferrite grain size. It is given by the high cooling rate. Then, the GBA in this HAZ region must be less enriched in nitrogen.

On the other hand, as the volumetric fractions of IGA and WA are lower in the A and C HAZ regions, their hardness values are slightly higher. Higher ferrite grain sizes significantly reduce the fraction of GBA, which consumes the nitrogen from its surroundings. This GBA is then harder in the wider B HAZ region of the welded joint. During cooling, after the GBA formation, the chromium nitride precipitation, that also consumes nitrogen, occurs due to the low solubility of nitrogen in ferrite below 900 °C. Therefore, nucleation of the WA (at low temperatures) and IGA will consume the remainder of nitrogen in ferrite. At higher heat input (2.5 kJ/mm) or lower cooling rate, the IGA and WA volume fractions increase. Consequently, these austenite morphologies will be less enriched in nitrogen and have a lower hardness.

The diffusion distances of nitrogen atoms in the δ phase during continuous cooling, between 1200 and 800 °C, was calculated for analyzing the incidence of the chromium nitrides in the pits formation. These diffusion distances of nitrogen atoms in the δ phase were carried out using the equation (2) (Kobayashi et al. 2000; Liao 2001).

where tf is the cooling time, D(T) is the volume diffusion coefficient, which can be expressed by the following equation:

being D₀ = 1.13 × 10⁻⁶ m²/s and the activation energy E = 83(1−14.03/T) kJ/mol for nitrogen. The temperature T can be expressed as T = T₀−vt, where T₀ is the initial temperature (1200 °C) and v (°C/s) is the cooling rate.

Table 6 shows the diffusion distances of nitrogen atoms in ferrite between temperatures of 1200 and 800 °C during continuous cooling. These diffusion distances are compared to half the average diameters of the ferrite grains in each HAZ region.

It is interesting to note that in the C HAZ region for both heat inputs, the average distance that nitrogen atoms can travel in the time interval between 1200 and 800 °C is approximately half the average radius of the ferrite grain.

On the other hand, the average distances traveled by nitrogen atoms in this Δt_12/8 practically coincide with the average radii of the ferrite grain in the rest of the HAZ regions. Conversely, the C HAZ region was the least affected by pitting corrosion.

It is well established in the literature that high cooling rates result in less reformed austenite and more precipitated chromium nitrides (Liao 2001; Liou et al. 2002; Yang et al. 2011). While low cooling rates corresponding to high heat inputs provide a more significant amount of reformed austenite, decreasing the chromium nitride precipitation. It is recognized that the chromium nitride precipitation in ferrite promotes pitting corrosion by chromium depletion in its surroundings, causing a weakening of the passive film (Matsunaga et al. 1998; Ogawa and Koseki 1989; Omura et al. 2000; Sridhar and Kolts 1987). This depletion leads to a local reduction of the PREN, and the precipitates may become preferential corrosion sites electrochemically (Matsunaga et al. 1998). According to this reasoning, the C HAZ region should be the lowest resistant to pitting corrosion because it has the highest cooling rate and, consequently, should increase nitride precipitation.

Conversely, the corrosive tests in the welds for both heat inputs (Figure 7) showed a particular situation. With the slowest cooling rate, the B HAZ region showed less reformed austenite than the C HAZ region, with the highest cooling rate. TEM analysis of these HAZ regions in the welded joint, with a heat input of 1.5 kJ/mm, showed precipitation of CrN nitrides predominantly in the C HAZ region. In contrast, in the B HAZ region, only Cr₂N precipitates were observed (Figure 10). Under non-equilibrium conditions at the highest cooling rates, both nitrides can precipitate simultaneously (Hertzman et al. 1986; Jargelius-Pettersson et al. 1995; Liao 2001; Yang et al. 2011). The nucleation of the metastable cubic chromium nitride (CrN) in ferrite is crystallographically favored having a greater driving force for its precipitation compared with the Cr₂N formation at temperatures below 1000 °C (Jargelius-Pettersson et al. 1995). Also, it was confirmed that the fraction CrN/(CrN + Cr₂N) increased in 2205 welds with an increasing cooling rate (Omura et al. 2000). These authors showed that more significant quantities of CrN were precipitated at faster cooling rates, and it was noted that CrN formed the greater part of the chromium nitrides when the cooling rate was faster than 100 °C/s. Analyzing the free energy of formation of the CrN (Jargelius-Pettersson et al. 1995; Omura et al. 2000), it was concluded that chromium depletion around such CrN nitride is small, and the effect on pitting corrosion is not significant. In Table 6, only the C HAZ region corresponding to the welded joint with 1.5 kJ/mm as heat input, where the cooling rate was 160 °C/s, did not show pitting corrosion. The other HAZ regions, with slower cooling rates, showed pitting corrosion at different intensities. The B HAZ region had the lower resistance to pitting corrosion. This HAZ region had the lowest amount of GBA and the highest amount of IGA. The WA fraction in this HAZ region was not significantly higher than in the other HAZ regions. Only a slight increase was observed due to the larger ferrite grain sizes. TEM analysis of the B HAZ region in this welded join with a heat input of 1.5 kJ/mm, showed intense Cr₂N precipitation at sub-boundaries in the bigger ferrite grains. This microstructural characteristic must be a factor that contributes to explaining the behavior of the pitting corrosion resistance at the HAZ regions of the welded joints using the GMAW process described above.

Another essential aspect to consider is the IGA fraction and its formation kinetics. The microstructural constituent interfacesʼ role in pitting corrosion resistance has been well documented (Li and Li 2004; Nazarov and Thierry 2007; Örnek et al. 2021). The IGA formed at higher cooling rates is characterized by having semi-coherent interfaces with the ferrite according to specific orientation relationships. However, the IGA formed at slower cooling rates, random orientation relationships predominate where the coherence between phases is lost. Incoherent interfaces are favorable sites for the initiation of pitting corrosion (Vignal et al. 2006; Yang et al. 2011). Figure 4 shows that the IGA fraction is higher in B HAZ region, where the cooling rate was lower. In addition, the sizes of the IGA particles in this region were the largest and presumably less enriched in nitrogen according to the microhardness values (Figure 5). In this sense, a greater amount of IGA particles with low nitrogen and the predominance of random interfaces in this B HAZ region constitute another microstructural aspect that favors the initiation of pitting corrosion.

The slight difference in the average values of the E_pit and ΔE showed in the potentiodynamic polarization curves (Table 5 and Figure 9) of the simulated HAZs with heat inputs of 1 and 3 kJ/mm is due to the cooling rates imposed by the programmed thermal cycle. As the simulated HAZ with a heat input of 1 kJ/mm has a higher cooling rate, the ferrite phase will be more saturated in nitrogen, and a higher fraction of chromium nitrides will precipitate. In Figure 6, the simulated HAZ with a heat input of 1 kJ/mm has a cooling time of 6 s between 1200 and 800 °C (given by the thermal cycle). The average cooling rate between these two temperatures is 66.7 °C/s, much lower than the 100 °C/s reported by Omura et al. (2000), where the precipitation of cubic CrN predominated above this cooling rate. On the other hand, the cooling rate between the same temperatures (1200 and 800 °C) in the simulated HAZ with a heat input of 3 kJ/mm is lower than the experimented in the simulated HAZ with a heat input of 1 kJ/mm. Therefore, in both simulated HAZs precipitated predominantly the Cr₂N. Zhang et al. (2015) have shown that the mechanisms that cause metastable pits are controlled by the chromium depleted zones around second phases. Figure 9 (a) displays metastable pits between 300 and 480 mVSCE.

The chromium nitrides nucleation is favored by increasing the cooling rate but its growth rate increases when the cooling rate decreases. This explains the slight difference between both potentiodynamic anodic polarization curves. The HAZ with a heat input of 1 kJ/mm will have a higher fraction of the Cr₂N small particles, while the HAZ with a heat input of 3 kJ/mm will have the Cr₂N more coalesced. The Cr-depleted zones around the Cr₂N will be greater in the simulated HAZ with a heat input of 3 kJ/mm, becoming more sensitive to pitting corrosion despite having a lower volumetric fraction of Cr₂N.

The significant difference between the values of E‾pit and ΔE corresponding to the base metal and simulated HAZs is a consequence of the structural changes experienced in the steel during the thermal cycle. These structural changes are basically caused by the precipitation of Cr₂N and the character of the interfaces of the new austenite morphologies in the simulated HAZs.

As the different HAZ regions in real welded joints have low austenite fractions (Figure 4), the Cr and Mo contents of ferrite in these HAZs regions are lower than in the parent steel, leading to a decrease in the PREN of the ferrite phase. So, the Cr-depleted ferrite zones around the nitrides have a much lower PREN than the ferrite matrix and therefore lower corrosion resistance. Figure 8 shows that the pits appear selectively inside the ferrite grains, where the precipitation of chromium nitrides occurs, and in some interfaces between the ferrite and IGA particles.

Based on the results obtained, the corrosion resistance of real heat affected zone (HAZ) regions is primarily determined by the size, structure and fraction of chromium nitrides and IGA particles. In Table 4, all HAZ regions in real welded joints with equivalent heat inputs equal to or greater than 1 kJ/mm exhibited pitting corrosion after being immersed for 72 h in a ferric chloride solution. This indicates that at higher heat inputs or lower cooling rates, there is a greater ferrite grain size and larger Cr₂N nitrides. The growth of IGA particles at lower cooling rates promotes random interfaces with the ferrite matrix. Therefore, the passive film in the HAZ regions, where the concentration of Cr₂N nitrides and the largest IGA particles is highest, is weakened by these microstructural characteristics.

5 Conclusions

After a meticulous analysis of the results obtained in the HAZ regions of welded joints with a single pass and two heat inputs, it was concluded:

1. Three HAZ regions appear in single-pass welded joints when the fusion line presents an inflection using the GMAW procedure. Each HAZ region in these welded joints showed different corrosion resistance after immersion in ferric chloride for 72 h.

2. The microstructural characteristics of the three HAZ regions determined their resistance to pitting corrosion. The HAZ region that was cooled at the highest rate (160 °C/s) had the highest austenite fraction, smallest ferritic grain size and highest CrN fraction. This HAZ region was the most resistant to pitting corrosion during the ferric chloride immersion tests.

3. The IGA fraction in the widest HAZ region was the highest, presumably with a predominance of incoherent boundaries of high energy. These high energy boundaries constitute preferential places for pits nucleation.

4. The widest HAZ region of the welded joint with both heat inputs has the lowest pitting corrosion resistance due to the Cr₂N nitrides concentration in the bigger ferrite grain sizes and predominance of incoherent boundaries of the IGA particles.

5. The highest cooling rate is linked with thinner HAZ region. The predominance of the CrN nitride precipitation in this HAZ region had a negligible effect on the pitting corrosion resistance.