Study of fusion boundary microstructure and local mismatch of SA508/alloy 52 dissimilar metal weld with buttering

A SA508/Alloy 52 dissimilar metal weld (DMW) mock-up with double-sided Alloy 52 butterings, which is fully representative of Ringhals pressurizer surge nozzle DMW repair solution, was studied. The microstructure, crystal structure, elemental diffusion, carbide formation and macro-, micro-and nano-hardness of the SA508/ nickel-base Alloy 52 buttering fusion boundary (FB) were investigated. Three types of FBs were analyzed, i


Introduction
Dissimilar metal welds (DMWs) are commonly used for joining austenitic and ferritic components in the reactor coolant pressure boundary, typically using a nickel-based filler metal [1].However, due to the difference in crystal structure between body-centered cubic (BCC) bainitic reactor pressure vessel steel and face-centered cubic (FCC) austenitic weld metal, a crystallographic mismatch occurs at the fusion boundary (FB) and creates a significant chemical composition gradient, particularly in carbon and chromium, at the dissimilar metal interface.As a result, a complex microstructure and a physical and mechanical properties mismatch (e.g.corrosion and strength) form at the interface [2][3][4],.Furthermore, the dissimilarity of metal at the FB can cause residual stresses due to the thermal mismatch during weld solidification and subsequent cooling [5].The complex DMW interface contains various microstructural regions including heat-affected zone (HAZ), carbon-depleted zone (CDZ), carbon build-up area at FB, partially-mixed zone (PMZ) and carbide precipitation zone [6,7],.
DMWs can pose potential risks to the structural integrity of the nuclear power systems, structures and components [8][9][10],.However, information on the microstructural and fracture mechanical changes that occur at the FB following buttering, welding, and post-weld heat treatment (PWHT) is limited and highly dependent on individual cases [11].In particular, knowledge about the local strength mismatch at the interface between low alloy steel (LAS) and Ni-based alloy weld metal during PWHT and long-term aging is lacking [7,12],.Accurately characterizing the local mechanical behaviors of the different materials constituting the DMW is challenging [13,14],.As such, the influence of FB microstructure on DMW performance has become an increasingly important topic in the nuclear power community [14][15][16],.
Despite numerous studies conducted, several critical questions remain unanswered.The two-sided Alloy 52 buttering welding technique, followed by PWHT, exhibits different microstructures at the weld FBs compared to other welding techniques, such as narrow-gap (NG) weld without buttering [17][18][19],.Limited data are available regarding the behavior of DMWs manufactured using high-Cr Ni-based Alloy 52 weld metal [20], which replaces the earlier used Alloy 182/82 weld metals based on wrought Alloy 600 composition [15] Although Alloy 52 has higher resistance to environmentally-assisted cracking (EAC) than Alloy 182 [21], the electrochemical potential gap and the resultant galvanic corrosion susceptibility of Alloy 52/LAS is higher than that of Alloy 182/LAS [22].It is critical to understand the effect of welding techniques and materials on the detailed microstructures at the FBs and local mechanical properties for assessing and enhancing nuclear component integrity and developing repair welding solutions to ensure safe long-term operation of nuclear power plants (NPPs) [23,24],.Furthermore, the segregation of impurity elements during heat treatment and aging, such as phosphorus segregation in the HAZ [18,25], and carbide diffusion to the type-II boundaries, coupled with residual stresses and strains at the interface, are known to increase susceptibility to cracking and stress corrosion cracking (SCC) [26,27],.The roles played by FB structures and types in promoting or reducing weld-related cracking are yet to be fully understood [28].
The welding techniques and parameters utilized in DMW studies in most research programs differ from those employed in actual NPPs, which can lead to significant variations in microstructure and mechanical properties [29,30],.Thus, a comprehensive investigation into the microstructure of the FB and the local mismatch of NPP-relevant DMW mock-ups is necessary.In this study, a Ringhals SA508/Alloy 52 DMW mock-up consisting of Alloy 52 buttering on both sides of an Alloy 52 V-groove weld was studied, which fully represents a repair technology solution of actual NPP components.The two-sided Alloy 52 buttering welding technique applied was first of a kind in Europe.The FB of SA508/Alloy 52 buttering is the most critical component for the DMW's integrity, and therefore, it was the focus of this work.The microstructure, crystal structure, carbide evolution, elemental diffusion and macro-, micro-and nano-hardness of the bainitic SA508/austenitic Ni-base Alloy 52 buttering FB were investigated.The findings are crucial for comprehending the fracture mechanical behavior and cracking susceptibility of DMWs.

Materials
The DMW investigated is a mock-up provided by Ringhals (representative of Ringhals 3 pressurizer surge nozzle DMW repair solution), consisting of nozzle material SA508M Grade 2 Class 1, Ni-based Alloy 52 buttering weld, Alloy 52 V-groove weld, Alloy 52 buttering weld, joined to the austenitic stainless steel 316LN safe-end.The inner surface of the nozzle LAS is cladded with stainless steel 308 L. A diagram of the DMW structure is presented in Fig. 1.The LAS side with buttering weld was subjected to PWHT at 615 • C for 1 hour and 17 min with heating and cooling rates of 55 • C/h from 300 • C. The V-groove weld and stainless steel side buttering were performed in a work-shop afterwards.The buttering welds were made in a transverse orientation, while the Vgroove weld in a longitudinal orientation.The welding parameters are listed in Table 1.The chemical compositions of each material in the DMW were confirmed with glow-discharge optical emission spectroscopy (GD-OES) and are presented in Table 2.

Microstructural characterization
The metallographic microstructure was evaluated using light optical microscopy (LOM) after etching the specimen.The LAS was etched with Nital 3% and the Alloy 52 with aqua regia (HNO 3 + HCl + glycerol).The specimens for electron microscopy, nanoindentation and X-ray diffraction (XRD) measurement were mechanically polished with diamond suspension to 0.25 µm, followed by a final surface finish prepared with Buehler VibroMet vibratory polisher machine using MasterMet 2 noncrystallizing amorphous 0.02 µm colloidal silica suspension for 45 min.
Scanning electron microscope (SEM), electron backscatter diffraction (EBSD) and energy dispersive X-ray spectroscopy (EDS) were applied for microstructural characterization of the FBs.A Zeiss Crossbeam 540 equipped with a solid-state four-quadrant backscatter detector (BSD) and an EDAX Hikari Plus EBSD detector was used.The nearsurface FB microstructure was investigated using SEM secondary electrons (SE) and backscattered electrons (BSE) imaging techniques.BSE imaging was conducted at an acceleration voltage of 15 keV with a working distance (WD) of 6-7 mm and a probe current 1.5-3.0nA.EBSD mapping was conducted at an acceleration voltage of 15 keV and a WD of 13-15 mm with 70 • tilting and a probe current of 1.5 nA.The EBSD phase maps, inverse polar figures (IPF), Kernel average misorientation (KAM) maps by TSL OIM Analysis 8 software.To reduce the interaction volume between incident electrons and matrix, the EDS line scan with an accelerating voltage of 5 keV and a probe current of 1.5 nA was applied to obtain local chemical differences across the FB from LAS to the weld with a measurement step size of 200 nm.
Transmission electron microscope (TEM) was used to characterize the carbides.The specimen was cut by electrical discharge machining (EDM) to a thickness of ~0.5 mm and mechanically thinned to ~100 µm.The final thinning was carried out by twin jet electropolishing using a Struers TenuPol-5 electropolisher with an environmentally friendly salt-based electrolyte.The electrolyte consists of 1 M concentration of NaCl in ethylene glycol and 200 ml ethanol.Scanning TEM (STEM) highangle annular dark-field (HAADF) and bright-field (BF) images and chemical information of carbides were acquired on FEI Talos F200X STEM equipped with Super-X EDS system operating at 200 kV.

Macro-, micro-and nano-hardness measurements
A representative DMW cross-sectional specimen was prepared with EDM and polished to 0.25 µm finish for the macro-and micro-hardness measurements.After the hardness measurement, the specimen was etched to reveal the microstructure and indent locations in the DMW (Fig. 2).The Vickers macro-and micro-hardness measurements with loads of 10 kg (HV10), 1 kg (HV1) and 0.3 kg (HV0.3)across the FB were performed, using Struers DuraScan-80 hardness measurement device.The indentation spacing of HV10 was 0.75 mm in the HAZ and region close to the FB, and 2 mm in other regions.The indentation spacing of HV1 and HV0.3 was maintained at 0.5 mm.As shown in Fig. 2, four lines of HV10 were measured and indicated as L1, L2, L3, and L4.L1 was closest to the outer surface, L2 was closest to the inner surface with the cladding, L3 was in the middle of the wall thickness, and L4 below the cladding HAZ characterizing the near inner surface hardness across the SA508/Alloy 52 FB.HV1 and HV0.3 were measured across the FB in the middle of the wall thickness, adjacent to the HV10 L3 measurement line.
Nanoindentation maps were measured with Anton Paar UNHT, using a Berkovich diamond tip indenter.The detailed parameters for nanoindentation are given in Table 3.The nanoindentation mappings encompass a distance of ~120 µm in the SA508 side (over the CDZ) and ~50 µm in the Alloy 52 buttering side, as measured from the FB.The   N. Hytönen et al. step size in the vicinity of the FB is 2.5 µm, while it is 5 µm in the remaining mapping area.The line spacing is consistently 5 µm.

XRD measurements
The different phases present in the FB region, i.e., BCC SA508, FCC Alloy 52 and BCC tempered martensite, were studied with XRD.A set of high-resolution reciprocal space maps (RSM) was collected from LAS to Ni-based alloy across the FB with Rigaku SmartLab diffractometer.The parallel beam size of about 200-250 µm was used with a 0.2 mm collimator on the incident beam and a two-dimensional (2D) detector without optics on the diffracted beam path.
In addition, wide-angle X-ray scattering (WAXS) was performed with a Xenoxs Xéuss 3.0 system.Selected TEM foils were also studied with WAXS in transmission mode with a Mo tube of 300 µm beam size.The 2D WAXS patterns were integrated and converted into 1D profiles, which were further refined using the MAUD program.

DMW macro-and microstructure
The macrostructure of the entire DMW is shown in Fig. 2 displays the macrostructure of the entire, where the LAS, cladding, buttering and Vgroove welds are indicated.Hardness measurements reveal variations in microstructural features across the FB from weld to HAZ, the LAS matrix and stainless steel cladding.Fig. 3(a) presents macro-hardness measurements (HV10) using four lines, namely L1-L4, as indicated in Fig. 2 (a).The L2 line, which is the closest to the inner surface, exhibits the hardness profile across the FB of stainless steel cladding/Alloy 52, showing only a slight increase in hardness adjacent to the FB.The macro-hardness across the FB of SA508/Alloy 52 buttering is measured by L1, L3 and L4.The hardness peak with a value of ~280 HV10 is found in the HAZ adjacent to the FB on the SA508 side.Beyond ~2.5 mm from the FB in the SA508 side, the macro-hardness is relatively uniform at an average of ~210 HV10.On the weld side, the macro-hardness values decrease until 1-2 mm from the FB.The macro-hardness in the buttering weld is comparable to that of the SA508 side but with a higher scatter due to the microstructure of the buttering weld beads.The hardness curves in the weld side between lines L1 and L2-L4 differ significantly, with the hardness curve near the outer surface showing an almost opposite trend of macro-hardness compared to the hardness in the midthickness or near the inner surface.The deviation in the hardness curves in the weld indicates the presence of internal strains in the V-groove root (L2) and buttering root (L1).
Fig. 3(b) displays the micro-hardness profiles of HV1 and HV0.3.The micro-hardness indents are located in the middle of the wall thickness across the FB of SA508/Alloy 52 buttering, as indicated in Fig. 2(b).The profiles exhibit micro-hardness peaks (320 HV1/HV0.3) in the HAZ approximately 1 mm from the FB.At a greater distance from the FB, the micro-hardness remains relatively constant (~220 HV1/HV0.3) for both SA508 and Alloy 52.The micro-hardness measurements suggest that the SA508 HAZ extends approximately 2 mm in width.
The microstructure of the FB of SA508/Alloy 52 buttering and its surrounding regions on both sides was investigated with the etched metallographic specimen.As shown in Fig. 4(a), the HAZ appears as a dark region adjacent to the FB, while ghost lines are visible as dark lines distributed throughout the bainitic SA508 material.The macro-and micro-hardness measurements indicate that the degree of macrosegregation is mild, as the ghost lines do not cause any significant macroscale deviation in the hardness values.Furthermore, the HAZ width is estimated to be approximately 2.3 mm, which is consistent with the hardness measurement results.Another etching solution was used to study the weld microstructure, specifically the weld beads in the buttering orientation, as shown in Fig. 4(b).The austenitic Ni-grains exhibit dendritic substructure and grow transversely across the weld beads.These grains are decorated with darker vertical micro-segregation areas at the intersection of the weld beads.

FB types
The microstructure of the SA508/Alloy 52 buttering FB was investigated using LOM (with SA508 side being etched) and SEM-BSE.Three different types of FB microstructures were identified, i.e., i) narrow FB, ii) tempered martensitic transition region and iii) wide PMZ, as shown in Fig. 5.The SEM-BSE analysis covered approximately 18 mm of FB where the majority of the FB was found to be narrow without any distinctive microstructure (Fig. 5(a-b)), which was defined as type-A and occupied ~80-85% of the FB.The type-A boundary area was narrow (<1 µm wide), where the microstructure changes from BCC to FCC without any  visible distinctive transition zone.The second most common boundary type observed was the feathery-like tempered martensitic transition microstructure, defined as type-B, which had a width of up to ~8 µm (Fig. 5(c, d)) and occupied up to ~15% of the FB.In optical imaging, the type-B FB appeared as a dark region with fine lath-like acicular microstructure and the weld side interface was not very distinct due to the lath-like microstructure at the FB.The third FB microstructure type was the wide PMZ (Fig. 5(e, f)), which had a width of up to ~25 µm and was the least common, found only at a few locations.It was defined as type-C, occupying only ~1-2% of the FB.The type-C FB region appeared light in color under LOM and there was a distinct and well-defined boundary for the transition zone, which was different from the type-B FB.As shown in Fig. 5, the three types of FBs appeared to associate with different levels of grain coarsening and widths of CDZ in the LAS HAZ, indicating a difference in the local heat flow and elemental diffusion between the FB types.The CDZ and grain size evolution adjacent to the FB were illustrated in Fig. 6.The carbon depletion appeared as empty ferrite grains in the coarse-grained region with a relatively small grain size of ~4 µm.The CDZ width is approximately 30 µm at the location of a type-A FB.Beyond the CDZ, the grain size decreased towards ~1 µm size and the carbide content increased.The present study focuses on investigating the specific features of type-B and type-C FBs, which are believed to cause local changes in properties and play a crucial role in maintaining the integrity of DMWs.In order to explore the microstructures of these FBs, various analytical techniques such as SEM-BSE imaging, EBSD, EDS, nanoindentation and XRD analysis were employed.Nanoindentation mappings were performed to reveal the influence of FB types B and C on the nano-hardness mismatch across the local DMW interfaces.The nano-hardness mapping and statistical data points with the average value are shown in Fig. 7.The mappings in Fig. 7(a,c) visually illustrate the variations in the properties across the two different FBs.The nanoindentation data revealed a conspicuous peak in the nano-hardness at the type-B FB area.The adjacent LAS side to the FB, however, did not show any significant softening due to carbon depletion.In the type-C FB, only a mild nano-hardness peak was observed for the PMZ FB, which appeared higher due to the contrast to the lower hardness level of the LAS side region adjacent to the FB.This ~30 µm region with low nano-hardness in the LAS is identified as CDZ.The overall hardness level for the type-C FB appeared lower with less scatter in Fig. 7(d) compared to the type-B FB  in Fig. 7(b).Contradictorily, a low bound nano-hardness region of ~10 µm was observed on the Alloy 52 buttering side in the type-B FB in Fig. 7 (b), which was not observed in the type-C FB in Fig. 7(d).
SEM-BSE, EBSD and EDS were applied to analyze the microstructures of the nano-indented regions of types B and C FB.The BSE images in Figs.8(a-c) and 9(a-c) reveal the microstructures of the type-B and the type-C boundaries, respectively.The corresponding EBSD maps, including the IPF map and phase map with high-angle grain boundaries (15 • -62.7 • ) and KAM analysis, are shown in Figs.8(d-f) and 9(d-f).The type-B boundary, with a width of 8 µm, exhibited a lath-like acicular grain microstructure.The laths, with a length of 2-8 µm and a diameter of 200-600 nm, grew towards the weld buttering in the heat flow direction, presumably starting from the favorably oriented fine grains at the bainitic HAZ.The interface between type-B FB and the Alloy 52 buttering weld metal was characterized by a wavy, feathery-like boundary.The type-B FB was dominated by the BCC phase in the phase map and the feathery-like structure was associated with tempered martensitic microstructure.The type-C PMZ FB in Fig. 9 exhibited a straight FB to the Alloy 52 buttering weld metal but with a distinct region of mixed phases of BCC and FCC in the transition region, which was approximately 25 µm wide.
A semi-quantitative analysis of the chemical composition using EDS line scan was conducted across the type-B and the type-C FBs.The location of the line scans was identified in the Figs.8(a The tempered martensite FB and the surrounding area as depicted in Fig. 8 were studied using high-resolution XRD with a fine beam size of 200-250 µm, aiming to verify the existence of tempered martensitic phase and determine its crystal structure, i.e., whether it is BCC or bodycentered tetragonal (BCT).By measuring from the fully BCC area (LAS) to the fully FCC area (weld metal), the FB was located by reaching 50% BCC of the strongest peak intensity.The first RSM at the FB in Fig. 11(a) revealed two prominent BCC rings (110 and 200) and one strong FCC 111 peak along with one minor FCC 111 peak.These peaks corresponded to the microstructure observed in Fig. 8(d) with randomly orientated small bainitic LAS grains and two large FCC grains.Two additional maps shown in Fig. 11(b-c) were collected on the FCC side to maximize the diffraction from the tempered martensite.These maps were taken from two adjacent locations along the FB to minimize the diffraction from the LAS matrix, as only a very weak BCC 110 ring was visible.The ring patterns in all three maps matched the cubic structure and no splitting was observed in the 110 and 200 rings, indicating the absence of a tetragonal crystal structure.Notably, the BCC 110 plane spacing continuously increased with distance from the FB, indicating that the BCC phase near the FB was under residual compressive stress, which could be attributed to the martensite transformation.Fig. 12(b) displays the integrated peaks of FCC 111 and BCC 110 from Fig. 11(a).The half widths at full maximum of these two peaks were compared and it was found that the BCC 110 peak was significantly broader than the FCC 111 peak.This observation suggests that BCC 110 belongs to the tempered martensite phase with a high density of lattice defects.As can be seen in Fig. 12(a), the grain size of the tempered martensite phase is ~1 µm, which should not cause any detectable broadening.The Kikuchi pattern from one large tempered martensite grain from the type-B FB is depicted in Fig. 12(c) whereas a perfect Kikuchi pattern from the LAS matrix far from the FB is presented in Fig. 12(d) for comparison purposes.The Kikuchi pattern for the tempered martensitic grain appears blurred and diffused compared to the well-defined sharp lines of the Kikuchi pattern of the LAS matrix, indicating the presence of internal distortion and lattice defects in the tempered martensitic grain.

Carbide analysis
A qualitative investigation of carbides was conducted using SEM-BSD imaging at various distances for the type-B and type-C FBs in the LAS HAZ and in the Alloy 52 weld.The analysis close to the type-B FB in Fig. 13(a) shows tempered martensitic microstructure with a comparatively high concentration of intragranular carbides.At a distance of ~35 µm from the FB on the LAS side (Fig. 13(b)), the carbide concentration decreases significantly in the CDZ.The concentration of carbides increases in the HAZ and beyond the CDZ, where the carbides are both intra-and intergranular.As seen in Fig. 13(c), on the weld side, with a distance of ~40 µm from the FB, only a few carbides are present in the grain boundary.However, with a distance of 100 µm from the FB (Fig. 13(d)), a relatively large cluster of intergranular carbides is observed at the junction of a weld grain boundary.
In Fig. 13(e), the density and size (~50 nm) of carbides observed in a type-C PMZ FB are comparable to those of the tempered martensitic FB.The width of CDZ depicted in Fig. 13(f) for the type-C FB is greater than that of type-B FB.In Fig. 13(g), the CDZ transitions to the HAZ at a distance of ~48 µm from the FB for a type-C FB.The distribution of carbides in the type-C FB is more clustered with regions of depletion, while in the CDZ transition of a type-B FB shown in Fig. 13(b), the carbide distribution is more uniform, but at a lower density.On the weld side, intergranular carbides consisting of ~300 nm long lamellas are revealed at a distance of 13 µm from the PMZ FB, as shown in Fig. 13(h).
In a type-I boundary resulting from the epitaxial growth of grains in the FB region [20] adjacent to the FB, the carbide density is higher near the FB compared to a typical weld grain boundary.STEM images and elemental maps presented in Fig. 14 revealed the presence of two types of carbides within the LAS HAZ.The predominant carbides are Fe-and Mn-rich with a small amount of Cr, having round morphology with a size ranging from a few nm to over 100 nm.These    carbides are distributed both along the grain boundaries and within the grains.Additionally, a second category of carbides, enriched in Mo, has been observed, which are primarily of small size ranging from a few nm to a few tens of nm.Analogous to the (Fe, Mn, Cr) carbides, these Morich carbides can be both inter-and intragranular.The concentration of Cr within the (Fe, Mn, Cr) carbides is considerably lower than that of Mn content.In contrast, the content of Mn and Cr in the Mo-rich carbide is nearly equivalent.The normalized at% profiles of C, Cr, Mn and Mo are illustrated in Fig. 14, allowing for a comparison of the relative concentrations in the two carbide types.
The Rietveld refinement result of the WAXS patterns observed in the HAZ is presented in Fig. 15.The patterns can be indexed as BCC phase (the major reflections), M 3 C carbides and Mo 2 C carbides.In the investigated sample, the M 3 C carbides (i.e.(Fe, Mn, Cr) carbides) and the Fig. 13.Carbide evolution across the (a-d) type-B and (e-h) type-C FBs.Analysis from locations in the LAS HAZ with distances of (a) 6 µm (close to FB) and (b) 36 µm (in CDZ transition zone) from the type-B FB, and (e) at the PMZ FB, (f) at 8 µm (in CDZ), and (g) at 48 µm (in HAZ) from the type-C FB.Analysis from locations in the Alloy 52 weld with distances of (c) 40 µm and (d) 100 µm from the type-B FB and (h) at 13 µm from the type-C FB.For all analysis from locations in the LAS HAZ, FB is at the right side of the images whereas from locations in the Alloy 52 weld, FB at the left side of the images.

Fusion boundary microstructure and mismatch
A DMW with double-sided buttering and a V-groove weld between the bainitic LAS and austenitic stainless steel safe-end was investigated.The focus of this study was on the FB of the SA508 LAS/Alloy 52 buttering.The FBs studied were classified into three types based on their microstructural features and the width of the transition zone.The metallurgical and mechanical mismatch of the different FB types were analyzed.During the multi-pass welding, a low local heat input was found to improve the toughness properties by reducing the cooling rate and minimizing the HAZ grain coarsening [31].Moreover, the subsequent beads may have a slight variation in chemical composition (e.g.C or Cr contents) compared to the first beads at the interface due to the mixing of the weld pool with the matrix by forming swirls and dilution zones, which causes some variations in the FB microstructure.
The three types of FB microstructures were found to associate with varying FB widths, chemical dilution, width of the CDZ and carbide density in the LAS HAZ, indicating variations in local heat flow and cooling profiles.Type-B FB constituted approximately 15% of the analyzed FB whereas only 1-2% of the FB was categorized as type-C FB.The type-B FB is likely a result of re-melting of metals during multi-pass welding, followed by rapid cooling that led to the formation of martensite microstructure, which was tempered by subsequent weld beads.
The low nano-hardness of the Alloy 52 buttering (~10 µm) can be  attributed to a relatively high local heat input, resulting in carbon diffusion across the transition zone.Nanoindentation measurements revealed that type-B FB with tempered martensite phase showed a peak in nano-hardness.In contrast, no significant peak in nano-hardness was observed in the type-C PMZ.The nano-hardness peak associated with the hardened microstructure, i.e., tempered martensite, results in a stronger strength mismatch between the weld and LAS than in the type-C FB.The PMZ is distinct in the indentation map presented in Fig. 7(c).However, the absence of an evident nano-hardness peak in Fig. 7(d) and the nearly constant chemical composition of alloying elements at the PMZ relates to a stronger mixing of materials and a different solidification structure with lower cooling rate compared to the type-B FB.
The microstructure observed in the Type-C PMZ is similar to that often seen in swirls that form during welding when the turbulent weld pool enters the base material and partially melted metal solidifies.However, the welding orientation and technique effectively inhibit swirl formation in the investigated buttered weld.
A high local heat input, e.g. between two weld beads, results in an increase in Gibbs free energy and a stronger elemental diffusion, leading to dilution between the buttering weld and the LAS [15].These findings are consistent with the results in this work.Furthermore, a lower bound nano-hardness of ~30 µm in the LAS CDZ was observed in type-C FB, but not in type-B FB, revealing the effect of annealing and stress relief during the buttering process on the LAS side.Higher internal stresses are left in the LAS side in the tempered martensitic FB and the division between the tempered martensite and PMZ is supported by the EBSD analysis and the EDS line scan.The mixing of BCC and FCC phases and chemical compositions in the PMZ is relatively more significant compared to the type-B FB.Similar elemental dilution curves have been reported by Hou et al. [5] and Wang et al. [32].The relatively high C content in the LAS side lowers the starting temperature of martensite transformation.The local heat input and rapid cooling rate, along with carbide precipitations at the FB, create a favorable system for a diffusion-less process of martensite formation [6].Carbon migration from the LAS to the weld side and dissolution of primary carbides enhances the carbon equivalent near the interface at high austenitizing temperatures, enhancing hardenability and martensite formation [16].Chen et al. reported that a gradual change of the alloy elements' contents across the FB may result in the formation of martensite at the FB, whereas a sharp change will not lead to martensite formation [33].The orientation relationship between the weld metal and the formed martensite is described by a Kurdjumov-Sachs (K-S) relationship [33].The formation of martensite is mainly controlled by the diffusion of the alloy elements and residual stress resulting from welding [2].The tetragonality of martensite is lost in this study due to tempering and the tempered martensite FB showed a BCC structure in the XRD analysis.If the carbon content in the FB is below 0.18 wt%, tempered martensite is reported to have a cubic crystal structure, which has been confirmed experimentally by Liu et al. [34], and is in agreement with the current observation.Furthermore, all bands in type-B FB Kikuchi pattern were diffused compared to the sharp Kikuchi bands in the LAS metal matrix, indicating lower band contrast or coincidence index values due to distortion at the type-B FB.This finding correlates with high residual strains and high density of dislocations in the tempered martensite phase.

Effect of microstructure and mismatch
A comprehensive understanding of metallurgical boundaries, phases and properties at FB regions can lead to a more reliable assessment of the structural integrity of DMW joints.However, existing codes do not mandate such knowledge [35].This chapter discusses the effect of FB microstructure and mismatch on the structural integrity of DMWs.In the present study, the macro-and micro-hardness peak were discovered in the LAS HAZ adjacent to the FB, while the nano-hardness peak was observed within the FB, specifically for the type-B FB.Nevertheless, in a study on SA508/Alloy 52 NG DMW mock-up without buttering, the nano-hardness peak was reported to be located in the weld metal side [1], at an interface lacking any distinct FB microstructure, i.e., similar to type-A.The nano-hardness for the NG-DMW aligns well with the position of carbon peak value [3,7],.In comparison to the current weld with buttering, the difference in nano-hardness mismatch may be attributed to differences in the welding technique and parameters, welding orientation and duration of PWHT, causing differences in microstructure and elemental diffusion at the FB.The welding parameters significantly impact the solidification, cooling rate, internal stresses and mixing of metals, which influence the formation of a solidification microstructure at the FB.The orientation of the buttering weld is perpendicular to the typical V-groove or NG weld [36].During the long PWHT of up to 24 h in the NG mock-up, more carbon in the LAS side diffused from the CDZ to the weld metal side [1].As the currently investigated buttered mock-up was subjected to relatively short PWHT, the carbon accumulation is mainly concentrated at the FBs.Adjacent to the hardness peak is a region of decreased hardness due to the carbon depletion, i.e., a soft zone.
The mismatch in local mechanical properties in the FBs play a critical role in determining the local fracture resistance and susceptibility to cracking of DMW [12].In previous studies [37,38], it has been observed that there are abrupt deviations in the path of cracks initiated in DMWs, which occur due to the mismatch in mechanical strength and local microstructures.Nevasmaa et al. [37] suggested that increased strength mismatch results in lowered fracture resistance in regions close to the interface between the LAS and the Alloy 52 weld of DMW.This reduction in fracture resistance was attributed to the increased metallurgical constraint and crack (growth) driving force, accentuated by the high local mismatch state and inherent local inhomogeneity of the near-interface regions [37].Under conditions of high local strength mismatch, the strength and toughness properties of neighboring microstructures become increasingly important and can dictate the overall fracture behavior of a DMW.
The present study investigated the fracture behavior of DMW using single edge-notched bend specimens (5 mm thick and 10 mm wide) in the temperature range of -130 • C to -180 • C. Preliminary study of fracture surfaces and cross-sections of brittle fracture toughness specimens was performed for analyzing the role of FB microstructure in fracture path deviation.A horizontal cross-section of a specimen with the notch nominally placed at 0.3 mm from the FB in the HAZ and a variety of FB microstructures is revealed in Fig. 16.A LOM image after etching (Fig. 16(a)) showed a white dilution zone in the middle along the FB that extended into a swirl.The horizontal cross-section was prepared near the brittle fracture crack initiation site, where the crack propagated mainly along the FB and in the HAZ.It was observed that the weld bead solidification and FB microstructures had an effect on the crack path.The fracture behavior was found to vary depending on the FB microstructure.When a swirl or a distinct type-B FB microstructure (Fig. 16(b,c)) was present, the fracture deviated into the HAZ.However, when a type-A FB microstructure (Fig. 16(d)) was present, the crack propagated very close to or along the FB.Further investigations are ongoing to investigate the effect of carbides on the fracture behavior (as brittle fracture primary initiator), how the local tempered martensite influences the interface toughness and whether the HAZ is weakened due to elemental diffusion.It was concluded that the varying FB microstructure made the brittle fracture path propagation less predictable.
Martensite at the FB has been reported to correlate with the formation of type-II boundaries, which are known to be prone to cracking or FB debonding [39].The allotropic transformation in the base metal has been reported to result in the evolution of type-II boundaries that are parallel to the FB [20].Wang et al. studied the microstructure and fracture behavior of a buttered and a V-groove DMW of SA508 and Alloy 52 [12,30],.They observed a larger area of lath martensite at the base material/weld interface due to a high heat input, along with significant type-I and type-II boundaries on the weld side adjacent to the interface [12,30],.Type-II boundaries were also observed in both buttering and NG welds, which was linked to the evolution of the weld pass structure [3,6],.Ming et al. also reported that a high heat input and long high-temperature duration welding process can promote the elemental diffusion, grain boundary migration, and the formation of type-I and type-II boundaries, resulting in a wider CDZ and decreased cracking resistance [17].However, in the present study, no type-II boundaries and negligible amount of type-I boundaries were observed in the buttering weld side, even along the tempered martensitic FBs.This absence of type-II boundaries reduces the risk of SCC or impurity segregation during aging, potentially enhancing the structural integrity of the LAS/buttering weld interface [21].
The susceptibility of weld metals to ductility dip cracking (DDC) is a significant concern in welding, as it can negatively impact the lifetime and reliability of key components [26].DDC occurs when intergranular solid-state cracking forms between migrated grain boundaries due to sufficient activation energy during multi-pass welding and reduced ductility in the temperature range of 0.5-0.8 of the melting point [40].However, no DDC was observed in the investigated DMW.This could be attributed to the use of double-sided buttering technique with specific parameters, which reduces the susceptibility to DDC.

Conclusions
This work investigates the microstructure, crystal structure, elemental diffusion, carbide formation and hardness mismatch of the interface of ferritic SA508/austenitic Ni-base Alloy 52 buttering in a double-sided buttered DMW mock-up.The mock-up, provided by Ringhals, is the first-of-a-kind in Europe and fully representative of an actual power plant component.
• Three types of FBs were analyzed: a type-A narrow FB (~80-85% of whole FB and with a width <1 µm), a type-B feathery-like tempered martensitic FB with fine lath-like microstructure (~15% of whole FB and with a width of up to 8 µm) and a type-C wide PMZ FB (~1-2% of whole FB and with a width of up to 25 µm).The width of CDZ and coarse-grained region adjacent to the FB depends on the FB type.The width of CDZ ranges from ~20 µm to ~50 µm.The widest CDZ was observed for type-C FB and the narrowest for type-B FB. • The macro-and micro-hardness peak was found in the LAS HAZ while the nano-hardness peak was observed within the FB, particularly for the type-B FB.The high cooling rate and carbide precipitations at the type-B FB contribute to the nano-hardness peak with the presence of tempered martensite at the transition zone, resulting in a significant strength mismatch between the weld and LAS.On the other hand, the relatively high local heat input induced diffusion at the type-C PMZ FB promotes the mixing of BCC and FCC phases and grain coarsening in LAS adjacent to the FB, resulting in low-level nano-hardness in Alloy 52 buttering and nearly constant chemical composition in the PMZ.• XRD analysis revealed the presence of BCC and FCC crystal structures at the LAS and weld metal interface of type-B FB, but no tetragonal martensite phase was observed.The relative volume fraction of martensite at the FB is low and likely lost its tetragonality due to tempering.• Carbide density and distribution vary with the distance from the FB in the LAS side, with the highest density at the FB and significantly reduced density in the CDZ adjacent to the FB.Beyond the CDZ and ~100 µm into the HAZ, the carbide density increases again to that of the regular distribution of carbides in a LAS bainitic microstructure.
Main carbides found at FBs are M 3 C ((Fe, Mn, Cr) carbides) and Mo 2 C carbides.• The FB microstructure appears to have an effect on brittle fracture crack path, as the crack deviates into the LAS HAZ with a swirl or type-B tempered martensitic FB, while with a type-A narrow FB, the crack propagates along or adjacent to the FB.

Fig. 1 .
Fig. 1.The illustrative figure showing (a) the location of the weld in the surge nozzle safe-end and (b) the DMW structure.All dimensions are in millimeters.

Fig. 2 .
Fig. 2. (a) Macrostructure of the etched DMW and HV10 macro-hardness measurement lines L1-L4.(b) The micro-hardness measurement lines HV1 and HV0.3 in the middle of the wall thickness of DMW (next to the HV10 L3 line).

Fig. 4 .
Fig. 4. Etched metallographic specimens showing (a) microstructure of the HAZ and ghost lines in SA508 and (b) weld beads of the Alloy 52 buttering weld.

Fig. 5 .
Fig. 5. Three types of FB microstructures (a-b) type-A straight FB, (c-d) type-B tempered martensite and (e-f) type-C PMZ.CDZ width appear more light color adjacent to the FB in (a, b, c) optical images and FB widths are marked in (b, d, f) SEM-BSE images.

Fig. 6 .
Fig. 6.(a) HAZ and (b) CDZ at a type-A FB.The grain size is ~4 µm adjacent to the FB with carbon depletion.From ~30 µm from the FB beyond the CDZ, the grain size decreases closer to ~1 µm.

Fig. 7 .
Fig. 7. Nanoindentation results for the (a-b) type-B tempered martensite and (c-d) type-C PMZ FBs.The type-B tempered martensite FB exhibits a nano-hardness peak in the FB.
) and 9(a) for the type-B and the type-C FBs, respectively.The EDS results displayed in Fig.10indicate the profiles of the major alloy elements of Fe, Ni, Cr and Mn.An increase in Fe content and a decrease in Ni and Cr contents were observed closer to the FB, leading to the formation of a dilution zone from the buttering weld metal side to the FB.The change in the chemical composition across the type-B tempered martensite FB exhibited an almost linear profile for Fe and Ni concentrations, which are the primary constituents of SA508 and Alloy 52, respectively.In contrast, a drastic change in the concentrations of alloying elements was observed from SA508 to PMZ and from PMZ to Alloy 52 with sharp changes in the profile across the type-C FB.The chemical composition of the alloying elements within the type-C PMZ was found to be nearly constant, with approximately 75 wt% of Fe, 15 wt% of Ni and 5 wt% of Cr.

Fig. 11 .
Fig. 11.Reciprocal space mapping on type-B FB at the tempered martensite area.(a) At the FB; (b) at the Alloy 52 side about 120 μm to the FB; (c) at the Alloy 52 side about 120 μm to the FB and 500 μm above the measured place in (b).

Fig. 12 .
Fig. 12.(a) SE image of type-B FB with tempered martensite.(b) Integrated XRD peaks from Fig. 10(a).(c) Kikuchi pattern from spot marked in (a).(d) Kikuchi pattern of the BCC LAS matrix far from the FB.

Mo 2 C
carbides (i.e.Mo-riched carbide) have a volume fraction of 1.83% and 0.22%, respectively, which are consistent with TEM results.

Fig. 14 .
Fig. 14.STEM images (HAADF and BF) and elemental maps of carbides in SA508 at ~83 µm distance from the FB.The peaks represent the elemental evolution along the arrow marked in C -Cr-Mn map.

Table 1
Welding parameters for the investigated DMW.

Table 2
Chemical composition of the LAS nozzle and Ni-based weld metals materials provided by Ringhals.The GD-OES analysis was applied to verify the chemical compositions in SA508 and Alloy 52 buttering weld.