1 Introduction

Duplex ferritic–austenitic stainless steels (DSS) are known for their two-phase microstructure that consist of nearly the same fraction of ferrite and austenite at room temperature [1, 2]. DSS are based on the Fe-Cr-Ni-N alloy system and are extensively used in a variety of fields where good mechanical properties combined with outstanding corrosion resistance are required. DSS can be considered an adequate substitute for austenitic stainless steels in many applications where mitigation of chloride-induced stress corrosion cracking (SCC) and pitting corrosion is paramount. The resistance against the latter serves as a basis for the major classification of the DSS. The quantitative value known as their pitting resistance equivalent (PRE) is calculated from the chromium, molybdenum, and nitrogen content as follows [3]:

$$PRE=wt.\%\;Cr+3.3\;\left(wt.\%\;Mo\right)+16\;\left(wt.\%\;N\right)$$
(1)

Molybdenum-free lean DSS, which typically contain 21–23 % chromium, 1–4 % nickel, 2–6 % manganese, and 0.15–0.3 % nitrogen, provide a cost-effective alternative to 304 and 316 austenitic stainless steels [4]. The historical background of these alloys is linked to the large fluctuation of the nickel price, which represents a significant proportion of the total material cost. To make the steel prices less sensitive to cost fluctuations, steel manufacturers have substituted nickel with manganese and nitrogen [5]. Common lean DSS alloys have a PRE between 22 and 30 [6]. The standard DSS, which is the most widely used in industry, contain approximately 21–25 wt.% chromium, 2–3 wt.% molybdenum, 4.5–6 wt.% nickel, ca. 2 wt.% manganese, and 0.15–0.25 % nitrogen. Their PRE lies between 30 and 39. Alloys with a PRE greater than 40 are generally classified as super DSS. They have been developed to meet stringent requirements against pitting and stress corrosion in chloride-containing environments [3]. These alloys typically have a chromium content of 25–28 wt. %, nickel content of 6–7 wt. %, molybdenum content of 3–4 wt. %, a manganese content of 1–1.5 wt. %, and nitrogen up to 0.4 wt.%. Various super DSS alloys have been subject of several studies [7,8,9,10,11], which primarily focused on the microstructural changes and the resulting altered mechanical properties experienced after welding. Although some of these changes occur in each type of DSS alloys, some are more typical for super DSS.

Generally, the chemical composition of the DSS and the annealing during the steel production are adjusted to achieve a balanced microstructure typically containing a ferrite fraction of 45–55 %. However, fusion welding results in a reformed microstructure due to metallurgical changes and the given cooling conditions, which fundamentally differ from those in a steelmaking process. The major challenge lies in the fully ferritic solidification of the DSS [1]. The austenite formation starts in the solid-state typically below 1200 °C, and its extent is determined by two factors: the cooling time between 1200 and 800 °C and the chemical composition of the underlying alloy.

To gain a comprehensive understanding of the microstructural changes, the fusion zone (FZ) and the heat-affected zone (HAZ) should be discussed separately. Unlike the HAZ, the FZ undergoes significant changes in its chemical composition. In electron beam welding (EBW), which is a widely used process for thick sections [12], a significant fraction of the nitrogen escapes the weld pool due to high-vacuum conditions which technically reduces the partial pressure of the nitrogen to zero. Nitrogen is a strong austenite stabilizer; its loss greatly contributes to a mostly ferritic microstructure in the weld metal (WM) in all welding processes even when the cooling time is comparably high [13,14,15]. Nitrogen loss also occurs in laser beam welding (LBW) if the shielding gas does not contain nitrogen. However, unlike in EBW, it can be compensated by adding nitrogen into the shielding gas, which was the subject of several studies [16, 17]. This is also the case when arc welding [18,19,20,21,22]. The beneficial effects of prolonged cooling can only be exploited if no additional nitrogen loss occurs. Otherwise, the negative effect of the larger nitrogen loss often overweighs the beneficial influence of a longer cooling time [23]. Since the super DSS contain a large amount of nitrogen to compensate the influence of the higher chromium and molybdenum contents in terms of the austenite formation, they are even more sensitive to this phenomenon.

Devendranath Ramkumar et al. [9] reported the microstructure and mechanical properties of electron beam welded 2507 having a thickness of 6 mm. The microstructure of the weld metal had a ferrite number of 55–59 FN. The weld metal consequently showed a higher tensile strength but a lower impact toughness, compared to the base material. Saravanan et al. [10] investigated Nd:YAG laser welded super duplex stainless steel (2507) 2-mm-thick sheets and found that increasing heat input led to a decrease in austenite content. Applying 0.12 kJ/mm heat input, a mean austenite content of 38 % was achieved, but it dropped to 20 % when using 0.33 kJ/mm. This study supports the view that increasing heat input decreases the austenite fraction through promoting the nitrogen loss. Further challenges arise if the plate thickness increases. With larger plate thicknesses, critical heat input should be exceeded to prevent a significant amount of austenite dissolution in the HAZ [11]. This is often performed by reducing the weld speed and hence keeping the weld pool in molten phase longer, which in turn increases the nitrogen loss, leading to a higher ferrite fraction in the FZ. The biggest challenge in EB is that nitrogen loss cannot be avoided, necessitating new ways to promote austenite formation during cooling. Previous studies [24,25,26,27] showed that the application of nickel wire can be taken into consideration as an effective way to improve the phase balance in the weld metal in standard DSS, albeit it did not have any influence on the HAZ. The underlying study extends this approach to super duplex alloys which are richer in chromium, nickel, and nitrogen, examining the microstructural evolution and determining the resulting mechanical properties.

2 Materials and methods

2.1 Materials

1.4410 (UNS S32750) super DSS 20-mm-thick plates with a size of 300 mm × 150 mm were utilized as the base material for the experiments. To achieve the desire austenite formation within the weld metal, a high nickel content filler wire (S Ni 6059 or NiCr23Mo16 according to EN ISO 18274 [28]) with a diameter of 1 mm was used. High nickel content wire promotes the desired austenite formation. The nominal chemical compositions were determined on machined (base metal) and a re-melted (filler wire) specimen by means of optical emission spectroscopy (OES). Furthermore, the nitrogen content of the base metal war was measured by applying carrier gas hot extraction (CGHE). The results are shown in Table 1.

Table 1 Nominal composition of the base material and the filler wire
Full size table

2.2 Electron beam welding

The welding of the super DSS plates was carried out by utilizing a high-vacuum EBW machine with an integrated wire feeder unit and a maximal output power of 15 kW. The used welding machine and the welding layout are depicted in Fig. 1.

Fig. 1
figure 1

The electron beam welding machine (a) and the clamping setup of the specimens (b)

Full size image

The accelerating voltage (U) of 120 kV was kept constant during the entire experimental procedure. Analogous to that, the working distance (D) was also not changed during the studies, in order not to modify the aspect ratio of the electron optic, so that the resulting intensity distribution of the spot remained unaltered. Prior to the final welding, the adjustment of the welding current (Ib) and the travel speed (vw) took place, they amounted to 100 mA and 0.33 m/min in the final setup, respectively. The goal was there to achieve a stable penetration depth which on average is 2–3 mm larger than the thickness of the base material. The line energy (Q) — to be calculated by means of the U, Ib, and vw, — amounted to 2.18 kJ/mm. The resulting weld seam cross-section provided a basis for the calculation (by means of initial process adjustment steps) of the wire feed speed (vf). Previous experiments conducted on 2205 [26] showed that an average CrEq/NiEq ratio of 2.37 leads to a microstructure that consists of nearly the same amount of austenite and ferrite, on a macroscopic level at similar cooling conditions. The calculation is performed based on the chemical compositions, the weld cross-section, the wire diameter, and the relation between the welding speed and wire feed speed (as presented in [25]) neglecting any inhomogeneity of the dilution. It must to be noted that this balanced average results from a top region, which contains more than 50 % austenite, and a bottom region, which has less than 50 % austenite, due to a slight incomplete dilution. It is assumed that the same applies for the alloy 2507, even when considering a larger width depth ratio of the weld seam. The wire feed speed is set based upon this assumption, so that the average CrEq/NiEq ratio amounts to ca. 2.37, allowing a sufficient austenite formation in the root, even if the mixing between the filler wire and the base metal remains slightly incomplete. In terms of the wire feed, the contact tube and consequently the wire had a trailing angle of 30°.

During the EBW, the multi-beam technique was applied to provide a controlled spatial decoupling between the melting the wire and the main full penetration welding. The same configuration with minor differences was already used in a previous work [26]. As opposed to the LBW, the physical splitting of the beam is not possible. The welding in different positions in the same time is instead realized by a high-frequency beam deflection, so that a sufficient process stability is given in every position. In the first zone, the beam is defocused and melts the wire. The beam deflection was configured such that it creates two distinct zones without a real physical separation of the weld pool. The beam spends 10 % of the total time in zone 1 and 90 % in zone 2 while continuously jumping back and forth with a frequency of (FSwitch) 10 kHz. The energy ratio (P) is given from the dwell times in the respective zone since the beam current remains unaltered. The second zone creates a wide weld seam, enabling also a sufficient mixing between the base metal and the filler wire, due to the large oscillation width and frequency. Here, an oscillation pattern having a shape of an “infinity” is applied (Fig. 2), which has been developed and investigated thoroughly in previous experiments [26, 27]. The pattern possesses a length of 5 mm (SWx) and a width of 4.0 mm (SWy) and oscillates with a frequency (F) of 500 Hz. Its central point was retracted by 2.5 mm (DCx) in the x-direction related to the neutral beam axis having the coordinate of (0, 0). One additional feature was that the beam has a different focal position (f) depending on the respective zone. This was achieved by a so-called dynamic lens, which allows the modification of the focal position in just a small fraction of a millisecond. In zone 1, a stronger defocusing of +60 mm (focal point above the surface) is applied. This gives a better process stability — through enlarging the beam diameter on the surface — in cases where the wire is slightly bent, and therefore, its tip lightly deviates from the (0, 0) point. In zone 2, a defocusing of +22.5 mm was used which further widened the weld pool and reduced the spatter formation compared to a focused beam. An overview of the multi-beam parameters in detail is shown in Table 2.

Fig. 2
figure 2

Schematic illustration of the beam oscillation parameters

Full size image
Table 2 The applied parameters of the multi-beam technique in the electron beam welding
Full size table

2.3 Metallographic and metallurgical investigations

The study primarily focused on the microstructure formation in the FZ and the HAZ. After the welding, the plates were cut into pieces using a band saw. The specimens dedicated for the metallographic investigations were firstly ground and polished then etched with Beraha-II reagent (70 ml distilled water + 30 ml HCl + 0.5 g K2S2O5), coloring the different phases. Altogether, two metallographic specimens were extracted from the plate: one from the beginning of the weld seam (pos. A) and a second one from the end (pos. B) in order to gain information about the process stability (Fig. 3).

Fig. 3
figure 3

Schematic layout of the specimen extraction

Full size image

The metallographic investigation of the joints covered the examination of the weld geometry and the detailed investigation of the FZ and the HAZ. Altogether, five images were captured from the top, middle, and bottom region of the FZ, respectively, to gain an adequate amount of information. An incident light microscope was used for this step. The calculation of the ferrite content was performed by a custom-made script as presented in [26]. This runs an automated histogram-based image segmentation and a subsequent analysis [30], which allows the precise quantification of each phase as they are colored differently. The secondary focus lied on the distribution of the nickel in the weld metal. Therefore, an EDS mapping was performed on the specimen which was extracted from the pos. A. To conduct this measurement, a scanning electron microscope (SEM) with an integrated EDS detector was used. Altogether, 120 high-resolution image sequences were combined into a mosaic image, creating a panoramic view rich in detail which allowed a qualitative analysis of the weld dilution. Three line scans were furthermore performed across the weld metal to obtain semi-quantitative information about the nickel content and its inhomogeneity lateral to the weld penetration. Moreover, six standard notched specimens (Charpy-V) were extracted from the center of the plate. The first 3 specimens were extracted from the bottom region of the weld metal and the remainder from the top region, respecting the welding direction. The Charpy-V impact tests were carried out at −46 °C (requirement of the NORSOK M-601 [31], NORSOK M-630 [32], and EN ISO 17781 [33]) conforming to the EN ISO 148-1 [34] and evaluated according to the EN ISO 17781 [33]. Besides the impact value, the lateral expansion (increase in thickness of the specimens around the zone of fracture due to plastic deformation) was also measured by using a screw gauge. Additionally, a sample for carrier gas hot extraction (CGHE) was taken from the weld metal.

3 Results and discussion

3.1 Microstructure

Figures 4 and 5 depict the weld geometry and the microstructure in pos. A and pos. B, respectively. In both cases, parallel fusion lines can be seen, resulting from the “infinity”-shaped beam oscillation as observed in previous studies [26, 27]. The process led to a stable penetration depth. One striking characteristic is the cold lapping at the cap, which occurs in both cases and requires either a subsequent re-melting or mechanical removal. When analyzing the weld seams, it becomes apparent that the upper region of the fusion line is characterized by a considerable proportion of unmixed filler wire, which appears white after color etching. It can be assumed that it is a general issue when using the described welding technique, and its degree varies along the weld direction due to small process instabilities induced by the relatively high welding speed and weld depth. Similar observations apply to the HAZ. In pos. A, it is only visible in the upper and middle region. In pos. B, the weld seam has in contrast a more pronounced HAZ in the bottom and the middle region, although no visible HAZ occurs adjacent to the partially unmixed zone in the top region. In terms of the weld metal, the microstructure exhibits a balanced austenite-to-ferrite ratio with a slight dominance of the austenitic phase. Higher ferrite fractions amounting to 50–60 % can be found solely in the bottom region. All regions are characterized by the presence of Widmanstätten austenite and allotriomorphic austenite, while intragranular austenite was only observed in the bottom region.

Fig. 4
figure 4

Weld geometry and the microstructure of the EB welded SDX 2507 (pos. A) after color etching, highlighting the HAZ and the FZ

Full size image
Fig. 5
figure 5

Weld geometry and the microstructure of the EB welded SDX 2507 (pos. B) after color etching, highlighting the HAZ and the FZ

Full size image

Figure 6 presents a summary from the acquired ferrite fraction values in the different regions of the weld metal in pos. A and pos. B. The region of ferrite fractions accepted by the NORSOK M-630 [32] is marked green. All ferrite fraction values in the root and hence their average fulfill the requirements. In contrast, some ferrite values, especially in pos. B, drop below 35 % in the middle region. By analyzing the values in the upper region, it has to be noted that all values in pos. B lie between 38 and 52 % and thus satisfy the requirements. In contrast, pos. A is characterized by low ferrite fraction values which partly drop below 30 %. By taking a holistic view of the obtained values, the austenite fraction in average lies above that of the ferrite. It can be assumed that a small reduction (5–10 %) of the wire feed speed could lead to a microstructure where the ferrite fraction remains between 35 and 65 % in all regions.

Fig. 6
figure 6

Ferrite fractions in the different regions of the weld metal

Full size image

In order to formulate more detailed statements regarding the HAZ, images were captured from its different zones using a magnification of 1000× and are shown in Fig. 7. Even in such a way, it cannot be completely avoided that the images contain small sections of the WM or the BM (especially in the bottom region); hence, no quantified analysis was performed. In pos. A, a balanced microstructure can be observed in the top (Fig. 7a) and bottom region (Fig. 7c); the HAZ in the middle region (Fig. 7b) is rather ferritic. When comparing the results to those obtained during previous studies on 2205 (depicted in Fig. 7g, h, and i) [35], it becomes visible that the HAZ, especially in the upper region, contains significantly more austenite in the case of 2507 despite the same welding parameters. Similar observations apply to pos. B with the addition that no real HAZ can be seen in the top region. Instead, a partially mixed zone appears in the proximity of the fusion line, where an incomplete dilution of the filler wire can be recognized.

Fig. 7
figure 7

Different regions of the HAZ captured with a magnification of 1000× (a pos. A/top region; b pos. A/middle region; c pos. A/bottom region; d pos. B/top region; e pos. B/middle region; f pos. B/bottom region; g reference welding on 2205/top region; h reference welding on 2205/middle region; i reference welding on 2205/bottom region)

Full size image

The investigation shows that the addition of nickel-based filler wire in electron beam welding of 2507 led to a balanced microstructure, attributed to the improved austenite formation. The microstructural development in the WM is in accordance with the phenomena experienced in prior studies of 2205 standard DSS [25,26,27]. Regarding the HAZ, however, a significant difference became apparent as compared to previously investigated weldments in 2205 [26]. In common standard DSS, the HAZ may become predominantly ferritic when power beam welding without post-weld heat treatment and may therefore be of concern [25,26,27, 36]. In contrast, the HAZ in this study was mostly narrow and generally possessed a balanced austenite-to-ferrite ratio. Of note is the fact that the microstructure of the super DSS is more resistant to microstructural changes when experiencing further thermal cycles with peaks below the δ solvus line. On one hand, the degree of ferritization in super DSS is suppressed due to the higher nitrogen content of the base metal by increasing the δ+γ → δ start temperature [11, 36, 37]. At a given thermal cycle, the time for a diffusion-driven phase transformation is thus much more limited, and only a tiny fraction of the BM — in the direct proximity to the fusion line — reaches this temperature at all. On the other hand, if noticeable austenite–ferrite transformation still occurs, its reformation upon cooling is strongly promoted through to the previously mentioned high nitrogen content, resulting in a balanced microstructure [36]. Based on the obtained results, further aspects open. To improve the weld appearance, re-melting of the cap with TIG or a simultaneous EBW can be taken into consideration. This step induces a change in the microstructure due to a second thermal cycle, which leads to a further loss of nitrogen. Therefore, it can be the subject of future research.

3.2 Chemical composition of the weld metal

A particular emphasis lied on the analysis of the mixing of the filler wire and the base metal. Especially the mixing in the vertical direction from the weld cap to the root and its homogeneity were central questions of the study, due to the large material thickness and the external wire feeding. The EDS mapping performed on the metallographic specimen extracted from pos. A reveals information regarding the filler metal mixing behavior. Fig. 8 presents the reconstructed image, where the nickel distribution is visualized by means of color contouring. Lime green color highlights the regions rich in nickel, while the areas with low nickel content appear black. Additionally, three line scans were performed using the raw data. Due to the wide oscillation pattern, as well as the high oscillation frequency, a nearly homogeneous distribution of the nickel was achieved in the weld metal. By analyzing the map from the cap to the root, altogether, four regions of interests (ROI) are described. The first feature which can be seen is the cold lapping on the cap side (ROI “A”), which is associated with a high amount of unmixed filler wire indicated by the bright lime color.

Fig. 8
figure 8

EDS map in pos. A, showing the distribution of nickel in the cross-section of the weld seam (a) and line scans in the top (b), middle (c), and root region (d)

Full size image

On the fusion line, a brighter region (ROI “B”) indicates a significantly elevated nickel content compared to the weld metal, which results from the incomplete dilution between the filter wire and the base metal. Fig. 8b shows the nickel distribution along the line “b,” which crosses this ROI. When analyzing the nickel content on a macroscopic scale and in a semi-quantitative way, its content lies between 10 and 12 wt. % with a scatter comparable to the base metal. The aforementioned incompletely mixed zone on the fusion line manifests itself in a significant peak both in the individual readings and the Gaussian filtered data. Taking the middle of the weld seam (ROI “C”) under discussion, a color contrast can be observed between the areas close to the fusion line and that near the centerline of the weld seam. A line scan along (Fig. 8c) the track “c” reveals distribution of the nickel content across the weld seam. By analyzing the smoothed data, the slight decrease of the nickel content in the proximity of the centerline becomes clearly visible. Fig. 8d shows further quantitative information about the ROI “D,” where regions with higher and lower nickel contents alternate.

To sum up, the average nickel content in the weld metal was 10–11 wt. %, which is approximately 3–4 wt. % higher than that of the base material. This enriching has proven itself to be sufficient to promote the austenite formation, so that a balanced microstructure was formed. Due to this high nickel content, stress corrosion cracking in the weld metal may be of concern and can also be the subject of further experiments.

Additional to the nickel distribution, the nitrogen loss was also determined by means of carrier gas hot extraction (CGHE). After the welding, the nitrogen content of the re-melted sample dropped to 0.20 wt. %. With reference to the initial nitrogen content of the base metal (0.27 wt. %, determined by means of CGHE), this loss corresponds to ca. 30 %.

3.3 Mechanical properties

The additional examinations involved the investigation of the mechanical properties. Figure 9 presents the hardness distribution in pos. A (Fig. 9a) and pos. B (Fig. 9b). The hardness values of the weld metal are generally higher than that of the base metal. In the bottom region, a significant increase of the hardness values can be seen as a result of the higher ferrite content as well as the rapid cooling. This applies for all regions. In the middle and the upper region, the hardness is (with some exception) 10–15 HV higher than that observed in the base metal and lies below that of the bottom region. In a few cases, the hardness is above 290 HV, but even the highest reading (309 HV) lies below the limit (350 HV for an individual reading) defined by the API TR 938-C [38].

Fig. 9
figure 9

The hardness distribution in in pos. A (a) and in pos. B (b)

Full size image

Figure 10 reveals the acquired impact energy values, as well as the lateral expansions of the different samples and FZ regions. The rectangular bars represent the mean values, while the error bars show the minimum and the maximum values. As can be seen, the impact energy values are between 210 and 230 J, fulfilling the requirements formulated by various standards [31,32,33]. The top region exhibits a slightly higher toughness and broke consequently with a larger lateral expansion than that of the bottom region.

Fig. 10
figure 10

The obtained impact energy (a) and lateral expansion (b) values

Full size image

4 Conclusions

In this study, 2507 DSS plates were welded with EB applying a multi-beam technique and a nickel-based filler wire. The major findings of the study can be summarized as follows:

  • The multi-beam technique combined with an “infinity”-shaped oscillation achieved high-quality mixing between the filler wire and the base material. The EDS map revealed that the fusion zone has insignificant inhomogeneity regarding its nickel content.

  • The deliberated addition of nickel resulted in a balanced microstructure, even in the lower regions of the fusion zone. The ferrite fraction of the root was 50–60 %, where the microstructure in the middle and upper region of the FZ contained 30–50 % ferrite.

  • The HAZ was generally narrow and had a balanced microstructure due to the high nitrogen content of the base metal.

  • The hardness of the weld metal was 10–30 HV higher than that of the base material.

  • The impact values of the weld metal at −46 °C were 210–230 J and exceeded those of the base material (specimens taken transverse to the rolling direction) by 50–70 J.