Influence of surface treatment on metal dusting resistance of welds

Welds of the commercial Ni‐based alloys 602 CA, 699 XA, 601 and 690 were exposed under harsh metal dusting conditions. The pit formation was compared for as‐welded, brushed, ground, glass bead blasted, steel shot peened and dry cathodic pickled surfaces. Surface treatments were applied by industrial partners. When (stainless) steels were involved in the surface treatment, Fe contaminations acted as catalyst sites for pit initiation. A fast metal dusting attack was also observed on dry cathodic pickled samples. A beneficial effect was attributed to deformation of the sub‐surface zone, as it provides diffusion paths for oxide formers (by grinding and glass bead blasting). In the same region, formation of α‐Cr (BCC) precipitates was observed. The best performance was found for samples where the weld bead was fully flattened by grinding to P40 grit. In addition, it was demonstrated that on‐site slurry aluminization further enhances the resistance against metal dusting attack.

○ brushed alloy 602 CA with Nicrofer S 6025 filler (FM 602 CA). [11] A strong attack along the heat-affected zone(HAZ), as reported for ○ alloy 304H with 308H filler, [12] ○ brushed, ground, sandblasted, and pickled alloy 600H with UTP068HH filler (FM 82), [11] ○ brushed alloy 601H with Nicrofer S 6025 filler (FM 602CA). [11] A stronger attack on the base plate than the weld, as reported for ○ alloy 800 with a "~alloy 600" filler (FM 82). [1] Similar attack on the weld and base plate, as reported for ○ welded alloy HR-160 with HR-160 filler. [9]nce, it is valuable to know for which alloys a stronger attack in the welded region is expected and how to prevent this local failure.For plant maintenance and to build cost-efficient new plants, lifetime optimization is the easiest to implement with as few changes in plant production and maintenance process as possible.One feasible option is to adjust the surface treatment of the welds.
Welds are typically brushed to remove the oxide scale after the welding process.Other methods include grinding, polishing, shot peening and/or etching.Gheno et al. [13] showed that the deformation induced by surface treatments was beneficial for the outward diffusion of Cr in a Ni-30Cr alloy at temperatures up to 600°C.In this temperature range, Cr diffusion along grain boundaries and dislocation pipes is faster than bulk diffusion.Hence, a stronger deformed zone provides more Cr for oxide scale formation.][16] However, residual impurities from the surface treatment, such as Fe from steel shot peening, may accelerate pit initiation. [17]n existing plants, or when other restrictions narrow down the choice of material, a coating on locations prone to metal dusting failure can increase the lifetime of the component.Aluminum diffusion coatings have been shown to be applicable on-site using an environmentally friendly slurry. [18,19]The slurry is applied onto the shot peened weld, and the required subsequent heat treatment can be performed along with a post-weld heat treatment. [20]Aluminide diffusion coatings have been successfully applied to protect the base material from metal dusting attack. [14,21]he alloys 602 CA and 699 XA were developed for applications in metal dusting atmospheres, where both alloys form protecting alumina and chromia scales. [10,22,23]Therefore, the aim of this study was to evaluate the best surface treatment for welds of commercial Ni-based alloys, with a focus on the mentioned alloys 602 CA and 699 XA.

| Alloys
The metal dusting resistance of commercial Ni-based alloys and their similar weld joints was studied.The effect of surface treatments was studied on VDM® Alloy 602 CA and VDM® Alloy 699 XA, and the influence of additional aluminization was also compared for VDM® Alloy 601 and VDM® Alloy 690.Two types of gas tungsten arc welds (GTAW) were studied: welds joining two plates of the same alloy and weld beads on a single plate without a filler.
VDM® Alloys 602 CA, 699 XA, 601 and 690 as well as VDM® FM 602 CA, 699 XA, and 52i were melted in an electric furnace.VDM® Alloy 699 XA and VDM® FM 699 XA were additionally electro-slag remelted (ESR).The compositions are given in Table 1 for the plates and Table 2 for the filler material.
Plates of 5 and 16 mm respectively were produced by hot rolling (partly with additional cold rolling), solution annealing and subsequent descaling by mill grinding.Welding wire was produced by hot rolling, descaling, cold drawing and annealing under reducing atmosphere.Joints were welded with the recommended filler materials: VDM FM 602 CA for VDM® Alloy 602 CA and 601, VDM FM 699 XA for VDM® Alloy 699 XA and VDM FM 52i for VDM® Alloy 690.

| Surface treatments and slurry coating
Surface treatments were performed by industrial partners.Details of the surface treatments are given in Table 3.Generally, the surface treatments were followed by wire-cutting to the final dimensions, however slurry aluminization was applied after wire-cutting.
Coupon samples were machined to 20 × 10 mm.Joint welds were made from 5 mm thick plates, while the weld bead samples were wire-cut to 4-5 mm sample thickness by removing part of the back side.All wire-cut surfaces were ground to a P500 grit (699 XA bead 2) or P1200 grit (other samples) finish.In the following, all alloy names are named only with their numbers/letters short form.
Depending on the type of treatment, the surface roughness changes.The roughness parameters Ra and Rz of 699 XA samples were measured with a 2Dprofilometer (Mahr MahrSurf GD 26) and can be compared in Table 4.The initial surface roughness of the plates after mill grinding is shown for comparison.

| Metal dusting tests
Quasi-isothermal exposures were conducted in flowing gas using a horizontal tube furnace for elevated pressures

DECHEMA research institute
T A B L E 1 Alloy compositions in wt.%, most elements were measured using X-ray fluorescence spectroscopy, C, S, and N were measured using combustion analysis and B using atomic emission spectroscopy.as described in Madloch et al. [25] The pressure vessels consisted of horizontal tubes of Centralloy ET 45 micro or Haynes HR-235.Samples were placed in separate alumina crucibles within the tubes.The system was purged with Ar (≤2 ppm O 2 ) overnight before heating to 620°C.When the temperature was reached, the aggressive gas was released into the furnace and the pressure was increased to the test level.The conditions for each exposure experiment are given in Table 5, the main difference between the two is the gas composition.The 699 XA bead 2 samples were exposed in gas 2, while all other samples were exposed in gas 1.

VDM® alloy
Exposures were ended by releasing the pressure and cooling the furnace and then flushing with Ar (≤2 ppm O 2 ).Both gases are highly aggressive, as has been shown in the referenced previous studies.Samples were cleaned in an ultrasonic bath with water and ethanol after each exposure to remove the accumulated coke.The total exposure times were between 480 and 985 h.

| Analysis
Cross-sections were prepared from selected initial, aluminized and exposed samples using conventional metallographic methods, including Ni-plating, polishing down to 1 µm and etching with V2A etchant at 50°C or Marble's reagent.Elemental distribution maps of the cross-sections were obtained on an electron probe microanalysis (EPMA) Jeol JXA-8100 instrument with a spatial resolution of about 0.5 µm.Electron backscatter diffraction (EBSD, EDAX Velocity) was performed in a scanning electron microscope (SEM; Hitachi SU5000).The latter was also used to generate surface images with secondary electrons and backscattered electron and elemental distribution maps using energydispersive X-ray spectroscopy (EDS, EDAX Octane Elite).Elements were identified using the following bands: Ni Kα1, Cr Kα1, Fe Kα1, and Si Kα1.

| Weld microstructures
The initial microstructures of the welds including the weld bead, HAZ and base metal are given in Figure 1.
The microstructures seen for the 602 CA and 699 XA weld beads are representative of all of the studied welds.Within the welds, classic dendritic solidification and segregation are evident.The HAZ showed either a similar grain size to the base metal (for 602 CA), or grain coarsening (for 699 XA).Primary carbides of 602 CA and segregation bands are visible in the base metal.

| Alloy comparison
The effect of microstructure and composition on pit growth can be better evaluated using laboratory surface finishes, where the effect of impurities is minimalized when compared with industrial treatments.To achieve this, the attack on the side surfaces was compared for each alloy weld joint.Figure 2 shows an overview of these side surfaces (ground to P1200 grit in the lab) of the four tested joints after 985 h exposure in gas 1.
A few pits formed on 602 CA on the base metal and within the HAZ; however, it cannot be definitively stated if the absence of pits on the weld is related to the weld microstructure or if pits would eventually develop given a longer exposure time.No pits were apparent on the 699 XA side surfaces.On 601, strong pitting was observed on the base metal, while only a few pits formed on the weld and HAZ.The higher resistance of the weld can be attributed to the slightly higher alloyed filler metal with 25 instead of 23 wt.%Cr and 2 instead of 1.4 wt.% Al.Conversely for alloy 690, a few pits formed in the weld and HAZ, while the base alloy was not attacked.This T A B L E 4 Roughness of 699 XA samples after different surface treatments, mill ground and steel shot peened sample from bead 2, rest from bead 1.

In µm
Mill might be due to a similar (inverse) reason, the filler metal of 690 is slightly less enriched in oxide formers than the base metal (27 instead of 29 wt.%Cr).

| Alloy 602 CA
An overview of 602 CA weld bead samples with different surface treatments before and after exposure for a total time of 830 h in gas 1 is given in Figure 3.By grinding with P40 grit, the bead was flattened, while its shape remained after brushing, glass bead blasting and dry cathodic pickling.Pit formation was observed on the weld bead and HAZ of the brushed and dry cathodic pickled weld starting in the first 100 h (images not shown).Smaller spots of carbon accumulation were visible on the glass bead blasted sample after 830 h.The weld bead ground with P40 grit was not attacked, while the side surfaces ground in the lab to P1200 grit all show pits (Figure 2).
In Figure 4, etched cross-sections of respective samples in the initial, surface-treated state and after 350 h exposure show the microstructure of the weld bead near the surface.
Grinding and glass bead blasting resulted in deformation zones about 20 and 100 µm deep, respectively.After exposure, typical metal dusting pit growth was observed on the brushed and dry cathodic pickled sample.A carburized zone accompanied the pit inward growth and coke formation.In contrast, carbon-rich "islands" accumulated at the surface of the glass bead blasted sample.EDS of the surface of these areas revealed typical carbon filament formation with Ni-rich particles (not shown).Both the ground 40 and glass bead blasted samples formed precipitates in the region of the deformation zone (as indicated in Figure 3).EPMA identified these to be Cr-rich (shown in detail below, see Figure 10).

| Alloy 699 XA
Figure 5 shows 699 XA weld bead samples before and after exposure for up to 820 h in gas 1.To compare the attack of the two gas compositions, steel shot peened samples of the same plate were exposed in both gases.In gas 1, the sample (without a weld bead) was exposed for 680 h.Similar to 602 CA, the bead was flattened during grinding, but not from other surface treatments.Pit formation was observed on the brushed and dry cathodic pickled samples after the first exposure of only 100 h (images not shown).No attack or carbon deposition was visible on the ground 40 and glass bead blasted samples after 820 h.The attack on the steel shot peened plate started within the first 240 h, and was clearly visible after 680 h.
Cross-sections were prepared from the samples exposed for 360 h and are compared with the initial states in Figure 6.
Similar to 602 CA, the deformed zones of the ground 40 sample (around 25 µm) and glass bead blasted sample (around 70 µm) showed a high amount of precipitates after exposure.EPMA showed these to be Cr-rich (not shown).Typical metal dusting pit formation was observed on the brushed and dry cathodic pickled samples.Note that in the etched cross-section of the initial dry cathodic pickled sample, the dendritic segregation upon cooling is more emphasized.The impact of dendritic segregation on C diffusion is most likely negligible in this case.
In a second test series, 699 XA bead 2 samples were exposed to gas 2, which has a different gas composition and slightly elevated pressure when compared to gas 1 (see Table 5).An overview of the samples before and after exposure is given in Figure 7.
The dark surface tarnish of the as-welded sample before exposure is a confirmation of the oxide formation in this region.After exposure, that sample showed strong pit formation along the HAZ and a few pits on the weld bead and base metal.The brushed sample had fewer, but still some pits formed along the HAZ and on the base metal.The ground 60 sample showed very few pits.Unlike the ground 40 sample of 699 XA bead 1, the weld bead was not fully flattened here.Hence, less material was removed by the surface treatment.The strongest metal dusting attack occurred on the steel shot peened sample, with a vast amount of pits on the weld bead and even more on the base metal.
Cross-sections of the exposed 699 XA bead 2 samples are shown in Figure 8.
The steel shot peened sample showed precipitate formation in the deformation zone below the surface.The brushed and ground 60 sample showed a similar, but thinner, zone only a few µm below the surface.EDS elemental maps of 699 XA after different surface treatments, but before any exposures, are shown in Figure 9.
Depending on the material used for the surface treatment, different elements remained on the cleaned surfaces.On the as-welded sample, oxides were detected locally.No residuals were found on the ground 40 samples.The Al-and O-rich trace on the brushed sample could be a remaining oxide layer from the welding process.From glass bead blasting, some beads became attached to the surface, and the surface remained partially covered with a Si-, Na-, Mg-and O-rich layer from the glass beads.In the dry cathodic pickling, W-, Al-, Ce-and O-rich particles adhered to the material.Fe-impurities on the brushed and steel shot peened sample can be attributed to the steels or stainless steels respectively used for the surface treatments.In a previous study, [17] Fe enrichment was also observed on steel shot peened 699 XA in comparison with mill ground 699 XA.Relatively lower amounts of Fe on the glass bead blasted and dry cathodic pickled samples can also be impurities from the process.Fe catalyzes carbon accumulation and absorption at the surface and thus accelerates pit formation.
Not shown here is the distribution of Cu, which was detected locally on the brushed, ground 40 and glass bead blasted sample.It could originate from the alloy itself (which is in this case improbable, because of a Cu content of <0.1%), or be deposited by cross-contamination in the industrial equipment.Alloying with Cu inhibits carbon deposition. [27]hus Cu impurities are not problematic for the metal dusting resistance, but due to their local appearance they are not expected to have a positive impact.

| Cr-rich precipitates
Grinding and shot peening deform the material, and a high dislocation density is visible in the zone below the surfaces in the initial states (see Figure 4 for 602 CA and Figure 6 for 699 XA).Precipitate formation was observed and EPMA revealed these to be Cr-rich, as exemplified for 602 CA in Figure 10 and for 699 XA in Figure 11.
The as-welded and the brushed samples showed the lowest surface deformation.After exposure, no changes in the sub-surface zone were visible when compared to the bulk microstructure in the EPMA elemental maps.The deformed zone from glass bead blasting was deeper than from grinding, and more precipitates form in the former sample.Some precipitates showed an increased carbon content, however, direct phase identification is not possible as quantitative carbon detectability with EPMA is limited.Based on calculations with the software JMatPro (MatPlus GmbH), the alloy 699 XA bead 2 composition would form phase fractions of 16.2% γ', 7.2% BCC (α-Cr) and 0.44% M 23 C 6 in thermodynamic equilibrium (excluding additional oxygen or carbon from the atmosphere).Cr is mostly present in the BCC phase (α-Cr), which supposedly is what most of the precipitates are.The identification of the precipitates as BCC phase was confirmed by EBSD measurements, see Figure 12.
In addition to the α-Cr precipitates, the EBSD maps show fine grains below the surface.The high dislocation density resulted in a recrystallized microstructure after exposure.The remaining dislocations can affect the image quality of the measurement (darker areas represent a lower image quality) and high residual stress is apparent by the color gradients in the grains at the bottom.

| On-site aluminization
Aluminide diffusion coatings were applied to all four alloys on top of the weld joints.The aluminized samples were exposed to gas 1, together with brushed samples of the same weld joints (Figure 13).
Metal dusting pits formed on the brushed 602 CA, 601 and 690 weld samples.A strong attack of the weld joint occurred for 602 CA, while for 601 the pitting was most significant at the HAZ.For 690, all regions of the sample were attacked, while the bottom of the weld showed a stronger attack than the top.Small black spots in the grinding grooves of the 699 XA base plate indicated the start of carbon deposition.No macroscopic pits formed on the brushed 699 XA sample during the 985 h of exposure.On the aluminized samples, single pits formed on the main surfaces of the 601 and 602 CA weld samples, mostly along the edges of the samples.In general, the aluminized samples show superior metal dusting resistance over the brushed samples.
The top row in Figure 14 shows cross-sections of the initial coatings after heat treatment.The bright outer region is an Al-rich (Ni, Cr, Fe)Al 3 with darker (Ni, Cr, Fe) 2 Al 3and (Ni, Cr, Fe) 2 Al 5 -precipitates above a continuous (Ni, Cr, Fe)Al-layer.The phases were determined using EPMA measurements.
The bottom row in Figure 14 shows the aluminum diffusion coatings after exposure.For all alloys, the (Ni, Cr, Fe)Al layer grew during exposure.Generally, the study confirmed that some welds are preferentially attacked and thus need special attention to prevent failure by metal dusting.Therefore, as exemplified by the strong pit formation along the HAZ of the aswelded alloy 699 XA bead in Figure 7, welds should not be exposed without a surface treatment, such as brushing, grinding, glass bead blasting, etc., as the oxides remaining from the welding process are non-protective.
Exposures were performed in two different gas compositions (see Table 5) at 620°C and elevated pressure.Samples of 602 CA, 699 XA, 601 and 690 were exposed in gas 1.In a previous study using the same condition, [26] samples ground to P1200 grit showed pit formation starting in the first 240 h (602 CA), 240 h (601), and 480 h (690).699 XA, also ground to P1200 grit, was not attacked in this gas even after 1507 h.This confirms the better performance of 699 XA in comparison with the attacked samples of 602 CA, 601, and 690, all ground to P1200 grit in Figure 2. Alloy 699 XA was additionally exposed in gas 2, in which samples of 699 XA ground to P500 grit showed formation of a few pits in the first 480 h, [8,17] implying a more aggressive attack in gas 2. In a previous study, [17] it was also shown that the pit initiation may have significantly changed with surface treatment.Study of 699 XA welds again confirms these findings (see Figure 7).As the surface treatment changes the catalytic sites and oxide scale formation, it is important to understand the mechanisms of each treatment to choose the best possible method.In the following, these different aspects are discussed.

| Effect of weld composition and microstructure
A stronger attack of welds can be caused by differences in composition, microstructure, and residual stress, but also by a surface treatment.Pits do not grow at the same region (weld/HAZ/base metal) for the side surfaces (Figure 2, ground to P1200 grit) compared with the respective main surfaces (Figure 13, brushed) of the joint samples.This implies that the impact of the surface preparation may outweigh the compositional and grain/ microstructural influences.Thus, the pit initiation time is highly influenced by the surface treatment.Without a surface treatment, the microstructure had a large impact, where the larger grain size of the HAZ is likely the most influential, limiting Cr grain boundary outward diffusion to form a protective scale. [10]Nevertheless, the alloy composition determines the best metal dusting resistance which is achievable with a surface treatment.

| Effect of Fe impurities
Steel shot peening increases the Fe content of the surface zone, as shown in. [17]The element maps of unexposed surfaces in Figure 9 confirmed these results.Fe-rich particles were present on the brushed (with a stainless steel wire brush) and steel shot surfaces, and in lower amounts also on the glass bead blasted and dry cathodic pickled surfaces.][30] Hence, pit initiation is promoted by a higher Fe content and fast and stable oxide formation is necessary to counteract this effect.Contact between steels as well as stainless steels and Nibased parts exposed to metal dusting atmospheres should thus be minimized.This can be achieved by choosing appropriate handling and surface preparation methods and minimizing the amount of Fe-impurities.

| Deformation depth and α-Cr precipitation
The grinding and shot peening surface treatments induced dislocations in zones 20-100 µm below the surface, depending on the method and alloy (see Figure 4 for 602 CA and Figure 6 for 699 XA).The deformed zone partially recrystallizes during exposure at 620°C, as can be seen in the EBSD map for the steel shot peened 699 XA bead in Figure 12.The higher grain boundary density and remaining dislocations enhanced Cr and Al outward diffusion and therefore the formation of a protective scale which enhances the metal dusting resistance. [10,13,15,17]This is especially remarkable as the typical metal dusting attack results in formation of coke islands with Ni-rich particles distributed in filamentous carbon.However, with deformation the carbon accumulation remained at the surface and did not grow into the material and form a pit in the studied time.Instead, the highly deformed zone may promote the formation of an oxide scale below the accumulated carbon to continue to protect the alloy (see Figure 10).
The high density of dislocations and grain boundaries have no effect on C inward diffusion, because the large interstitial spacing in the FCC lattice allows rapid bulk diffusion of C in this temperature region. [23]As a further note, there was even an observable difference in the metal dusting resistance for different grit finishes, as seen when comparing alloy 602 CA in Figures 1 and 2-the P1200 grit side surface was attacked, but the P40 grit material showed no pit formation even after longer exposure.
The high dislocation density and residual stress can also be a factor to promote α-Cr precipitate formation.Other studies on high Cr Ni-based alloys also showed α-Cr formation after annealing at 950°C for 100 h [31] or at 540°C for 5000 h, [2] demonstrating that this is a slow, diffusioncontrolled process which a high dislocation density could promote.A study on the correlation between deformation, subsequent α-Cr formation and metal dusting pit formation showed that the impact of the near-surface deformation improving oxide scale formation outweighed any potential negative effect from Cr depletion of the matrix by α-Cr formation for high Cr-containing alloys. [17]4 | Effect of surface structure Surface finishes are typically easily characterized by their surface roughness.However, there is no correlation between the overall roughness (Table 4) and pit formation (Figure 5) of alloy 699 XA.For example, the ground and dry cathodic pickled surface have similar Ra and Rz, but a strong attack is seen on the latter, while none occurred on the former.The microscopic view of the surfaces in Figure 15 helps to explain the difference.
Within the individual grinding grooves, the surfaces of the ground sample are relatively flat, while the dry cathodic pickled sample has almost a porous appearance.The larger surface area provides more reaction sites with the gas while requiring a higher availability of Cr and Al to form a F I G U R E 15 Secondary electron images of ground 40 and dry cathodic pickled surface of 699 XA bead 1 (not exposed) at 1.5 and 5 k magnification.
protective oxide.In addition, dry cathodic pickling does not induce any deformation below the surface, hence no additional Cr and Al diffusion paths are formed and thus giving a much lower metal dusting resistance.The method of dry cathodic pickling is at an early stage of development.Therefore, it should be noted that the results on metal dusting behavior could be potentially different with further development of the process.

| Protection by aluminizing
Better performance of 602 CA, 601 and 690 welds with Al diffusion coatings when compared to brushed weld joints is attributable to two factors: the negative impact of Feimpurities from brushing and improved protection by the Al-reservoir for alumina formation.
According to a study of Speck et al., [32] the resistance of NiAl in metal dusting conditions is superior when compared to Ni 3 Al and Ni with 3.75 wt.% Al.Hence, in the development of aluminum diffusion coatings, longterm formation of NiAl is desired, while Ni 3 Al and the brittle Al-rich phases should be avoided.While Ni 3 Al is observed to form in aluminum diffusion coatings at 900°C within 1000 h, [33] the reduced diffusion rates at lower temperatures impede its formation.Agüero et al. [34] showed that minimal Ni 3 Al formed on aluminized alloy 800HT during service for ~50,000 h at ~600°C despite the high Al gradient.Hence, no negative effect of Ni 3 Al on the long-term performance of the aluminized coatings on the tested alloys is expected.
A critical factor for these aluminide diffusion coatings is crack formation, for example, along sample edges or induced by the brittle (Ni, Fe, Cr) 2 Al 5 phase.Optimization of the heat treatment for each alloy could reduce the initial amount of this phase.In the presented study, heat treatment was not varied for the alloys.Despite this, the results already show a higher resistance of the aluminized welds when compared to the brushed surfaces, and the formation of the desired (Ni, Cr, Fe)Al layer was achieved.Hence, on-site aluminization is a promising method to protect welds of alloys, for example, 601, in harsh metal dusting conditions.

| CONCLUSION
In summary, the following factors must be considered when choosing a surface treatment for Ni-based alloy welds for service in metal dusting environments: • Tarnish and non-protective oxides formed during welding should be removed, as they enhance pit formation.
• The final surface treatment step should include as little steel and other Fe-containing materials (i.e., stainless steel) if possible.Contamination from the surface treatment medium may catalyze carbon deposition.• Deformation of the subsurface zone is beneficial.
During service/exposure, a fine-grained microstructure is formed, which allows rapid Cr and Al outward diffusion.• Surface treatments resulting in a high surface area, such as a semi-porous layer, should be avoided.The best protection is achieved when the surface is covered with a protective oxide scale, formed from the bulk.
The surface treatments used within this study can be ranked for their metal dusting resistance as follows: Rough grit grinding > glass bead blasting > brushing > steel shot peening ≈ dry cathodic pickling Additional tests over even longer durations and of further surface treatments would be recommended to verify and refine the database.For an even further improved protection of the weld, on-site aluminization was demonstrated to provide an Al reservoir for the formation of an alumina layer and could be applied for the most aggressive service conditions.

F
I G U R E 1 Light microscope images of etched cross-sections of 602 CA (left) and 699 XA (right) brushed bead welds.The microstructures of the welds and base metals at the surface are given in higher magnification below.[Color figure can be viewed at wileyonlinelibrary.com]F I G U R E 2 Macroscopic images of side surfaces (ground to P1200 grit) of weld joints after 985 h exposure in gas 1. [Color figure can be viewed at wileyonlinelibrary.com]

F
I G U R E 3 Macroscopic images of alloy 602 CA weld bead samples before and after 350 and 830 h exposure in gas 1.Note that the weld bead is central for all samples except the glass bead blasted (black horizontal arrow).[Color figure can be viewed at wileyonlinelibrary.com]F I G U R E 4 Light microscope images of etched cross-sections of 602 CA weld beads with different surface treatments before and after 350 h exposure in gas 1.All images show the weld bead.The cross-sections of exposed samples were prepared with a Ni-plating.[Color figure can be viewed at wileyonlinelibrary.com]

F I G U R E 5
Macroscopic images of 699 XA weld bead 1 samples before and after 360 and 820 h exposure in gas 1.The steel shot peened sample is a sample from 699 XA plate without a weld bead and was exposed only up to 680 h.[Color figure can be viewed at wileyonlinelibrary.com]F I G U R E 6 Light microscope images of etched cross-sections of 699 XA bead 1 samples before and after 360 h exposure in gas 1.All images show the weld bead.[Color figure can be viewed at wileyonlinelibrary.com]

F I G U R E 7
Macroscopic images of 699 XA bead 2 samples before and after exposure in gas 2 for 480 h.[Color figure can be viewed at wileyonlinelibrary.com]F I G U R E 8 Light microscope images of etched cross-sections of 699 XA bead 2 welds after 480 h exposure in gas 2. All samples have a Ni-plating.All images show the weld bead.[Color figure can be viewed at wileyonlinelibrary.com]F I G U R E 9 Backscattered electron images and energy-dispersive X-ray spectroscopy elemental maps of 699 XA bead surfaces (bead 1: as-welded, brushed, ground 40, glass bead blasted, dry cathodic pickled; bead 2: steel shot peened).[Color figure can be viewed at wileyonlinelibrary.com]F I G U R E 10 Backscattered electron images and electron probe microanalysis elemental map of brushed (top row), ground 40 (second row) and glass bead blasted (bottom row) 602 CA weld bead after 350 h exposure in gas 1.All images show the weld bead.[Color figure can be viewed at wileyonlinelibrary.com] Below the (Ni,Cr,Fe)Al layer, Cr was enriched up to 40-70 at.%.No formation of Ni 3 Al was observed.The coating on alloy 601 shows a few larger open pores, which could be due to remaining glass beads from the surface preparation before the slurry was applied.The aluminum diffusion coating still formed below these pores/glass beads, and thus no carbon deposition was observed.

F
I G U R E 11 Backscattered electron images and electron probe microanalysis elemental map of as-welded (first row) 699 XA bead 2 after 480 h exposure in gas 2 and of brushed (second row), ground 40 (third row) and glass bead blasted (bottom row) 699 XA weld bead 1 after 360 h exposure in gas 1.All images show the weld bead.[Color figure can be viewed at wileyonlinelibrary.com]F I G U R E 13 Light microscope images of brushed and aluminized weld joints before and after exposure in gas 1 for 985 h.The insert in the brushed 690 alloy after 985 h exposure shows the bottom side of the weld.[Color figure can be viewed at wileyonlinelibrary.com]F I G U R E 12 Electron backscatter diffraction maps of steel shot peened 699 XA bead 2 after 480 h exposure in gas 2. [Color figure can be viewed at wileyonlinelibrary.com]

F
I G U R E 14 Cross-sections of aluminum diffusion coatings on weld joints in the initial state after heat treatment (top row) and after exposure in gas 1 for 985 h.All images show the weld bead.Phases were determined using EPMA.Dashed arrows indicate the regions with mixed (Ni, Cr, Fe)Al 3 (bright), (Ni, Cr, Fe) 2 Al 3 and (Ni, Cr, Fe) 2 Al 5 ; the solid line arrows indicate the (Ni, Cr, Fe)Al layer.Note that for the exposed alloys, (Ni, Cr, Fe)Al 3 was measured only in the linescan of the coating on 602 CA.Horizontal cracks (602 CA and 690 after exposure) can form during hot embedding.[Color figure can be viewed at wileyonlinelibrary.com] Surface treatments.
T A B L E 3