Subscribe to our newsletter to receive the latest news and events from TWI:

Subscribe >
Skip to content

Fusion Zone Microstructure Associated With Embrittlement Of Subsea Dissimilar Joints

   
M. F. Dodge, H. B. Dong

Department of Engineering, University of Leicester, Leicester, UK

M. F. Gittos, T. Mobberley
TWI Ltd. Great Abington, Cambridge, UK

Presented at Proceedings of the 33rd International Conference on Ocean, Offshore and Arctic Engineering OMAE2014 June 8-13, 2014, San Francisco, CA, USA

Abstract

Within subsea oil and gas systems, nickel alloy filler metals are commonly used in the joining of ferritic steels with different mechanical properties. An example of this is the joining of low alloy steel (LAS) forged manifold hubs (e.g 8630 or F22) to low hardenability pipelines steels, such as X65). The joint is part of a two stage welding process that simplifies offshore installation. Initially, a weld deposit is made on the forging using a multi-pass ‘buttering’ technique, providing intermediate layers of a suitable material, such as Alloy 625. A postweld heat treatment (PWHT) is applied to the buttered forging, onshore, to temper the hard heat affected zone (HAZ). After machining a bevel into the buttering layer, a closure weld, applied offshore, is employed to join the pipeline to the forging. As the buttering layer and linepipe are not critically hardened by the closure weld, no PWHT is required.

To prevent corrosion, subsea systems of this kind, are subjected to cathodic protection, via aluminium based anodes. Whilst successful in protecting the ferritic parts within the manifold structure, a number of high profile failures has been attributed to the evolution and ingress of hydrogen, and its diffusion to, the fusion zone of the dissimilar weld.

To investigate the susceptibility of fusion zone microstructures to hydrogen embrittlement, three dissimilar weld samples were fabricated: 8630-Alloy 625, F22-Alloy 625 and F22-309LSi. The latter is not a combination normally used to complete joins of this type. Each specimen was the subject of thermodynamic and kinetic modelling studies using Thermo-CalcTM and DictraTM, with microscopic examination by scanning electron microscopy (SEM) combined with energy-dispersive X-ray (EDX) and electron backscattered diffraction (EBSD). Nano-scale features were also investigated by transmission electron microscopy (TEM). By comparing the fusion zones of the three different joints in both as-welded and heat-treated conditions, the metallurgical aspects relating to hydrogen embrittlement are revealed. The results indicate that precipitates, which form in a zone of carbon diffusion during PWHT, are particularly harmful in the presence of hydrogen. The formation of this hydrogen susceptible region is partly due to the initial solidification structure morphology, the formation of which is also discussed.

1. Introduction

1.1 Background

In subsea systems, dissimilar welds are used to facilitate the joining of low alloy steel hub forgings to micro-alloyed steel components. To simplify offshore installation, the weld is part of a multi-stage process. Initially, nickel alloy filler metals are deposited on LAS forgings, using a gas tungsten arc buttering procedure. PWHT, in the range of 650-675°C for 5-10 hours, is applied to temper the HAZ in the LAS. The buttered layers are then machined, creating a beveled face for offshore closure welding, using the same nickel alloy filler metal. The buttering layer ensures that the HAZ of the closure weld is far enough away from the forging to ensure that the joint with the lower hardenability linepipe requires no further heat-treatment.

The manifold structure is subjected to cathodic protection (CP) using aluminium based anodes. Whilst CP is effective in protecting the ferritic parts from corrosion and the majority of dissimilar hub joints have been in-service without complication, a small number of joints has failed at the dissimilar interface between the ferritic forging and austenitic weld material. CP has been associated with the evolution of hydrogen, resulting in hydrogen-assisted cracking of sensitive interface microstructures between the heat-treated buttering and LAS forging [1].

1.2 Fusion Zone Microstructure:

The major microstructural features observed in both F22 and 8630-Alloy 625 joints can be summarised, based on observations in a number of other studies [2–6]:

  • A coarse-grained HAZ.
  • A ‘feathery’ band of martensite, at the fusion boundary, between the two metals.
  • An apparently featureless, planar solidified region on the nickel-rich side of the dissimilar interface.
  • Cellular microstructures associated with instability, followed by breakdown of the planar solid/liquid (S/L) interface.
  • Columnar dendrites strongly oriented with the welding direction, together with Nb and Mo interdendritic precipitates.
  • Discontinuous partially mixed zones (PMZs) such as ‘swirls’ rich in parent metal within the edge of the fusion zone.
  • Penetrations of weld metal into the HAZ, along prior austenite grain boundaries (GBs) that appear as ‘fingers’ of fused metal.

These microstructural features are illustrated schematically in Fig.1.

Figure 1 – Summary of common features within the dissimilar LAS-Alloy 625 joints. Black regions indicate martensite-rich areas.
Figure 1 – Summary of common features within the dissimilar LAS-Alloy 625 joints. Black regions indicate martensite-rich areas.

1.3 Factors Affecting the Mechanical Performance

In the as-welded condition, the weld zone is a region of high residual stress and hardness [7]. PWHT is applied, principally, to reduce the HAZ hardness but some relief of residual stresses also occurs. However, during PWHT, carbon migrates from the ferritic steel into the austenitic weld metal [8]. Consequent carbon supersaturation [9]has been associated with martensite formation and precipitation of additional phases [5].

The morphology of the interface, is central to the susceptibility of the fusion zone, to hydrogen-assisted cracking. Martensite formation, which is commonly found at the interface between dissimilar welded joints, is particularly sensitive to hydrogen embrittlement [3]. Larger regions of martensite might also form in localised regions. For example, ‘swirls’ of parent metal may be swept into the weld metal where they solidify as ‘islands’ or ‘peninsulas’ [10]. Improvements in the environmental cracking performance of dissimilar joints, have been observed with PWHT [11]. However, excessive PWHT has also been shown to precipitate carbide phases, in the zone of planar solidification at the austenitic side of the dissimilar interface [12]. Carbide precipitation in the weld metal might be expected to affect its mechanical performance.

At the fusion boundary of the ferritic side, chromium/molybdenum-rich M23C6 precipitates have been observed within the lath-morphology, martensitic region [6]. In the Alloy 625 side, after prolonged PWHT, a band of chromium-rich M7C3 carbide forms [3], [4], [6]. The latter carbide has been associated with the embrittlement of these joints and is likely to affect performance in the ‘over-aged’ condition. Individual M7C3 carbides may act as hydrogen traps [8], providing a low energy fracture path that links together failure along brittle discontinuous PMZ [3].
Inhibiting the formation of the austenitic planar solidification zone is likely to prevent the formation of M7C3 carbide, however, it is unclear how to achieve this. Lower arc energies have been shown to reduce the amount of planar solidified regions but this is accompanied by an increase in the amount of martensitic swirls [13]. Another possible solution includes the use of different forging chemistries, the suitability of which will be explored in this paper.

2. Experimental Procedure

2.1 Materials and specimen preparation

Three dissimilar weld specimens, were provided in the as-welded condition. The composition of each is given in Table 1. Gas metal arc welding (GMAW) was used for all three joint types and the welding parameters are given in Table 2. The F22 and 8630 joints were selected, based on their current usage as hub forgings. The F22-309LSi joint is not a proposed solution to the embrittlement problem, but is included, for comparison of metallurgical aspects, that relate to failure.

Table 1 – Specified compositions (wt%) of the components of the three dissimilar joints.

 

Parent metals

Weld metals

 

8630
(AISI 8630)

F22
(ASTMA-182)

309
(ER309LSi)

Alloy 625
(ERNiCrMo-3)

C

0.28-0.33

0.08-0.15

max 0.03

0.01-0.1

Mn

0.70-0.90

0.03-0.60

1-2.5

0.125-0.5

Ni

0.40-0.70

0.5 max

12-14

58.0

Cr

0.40-0.60

2.0-2.50

23-25

20.0-23.0

Mo

0.15-0.25

0.90-1.10

max 0.75

8.0-10.0

Si

0.15-0.30

0.15-0.50

0.65-1.0

0.125-0.5

Nb

-

-

-

3.15-4.15

Fe

Bal.

Bal.

Bal.

5.0 max


Table 2 – welding procedure details for the dissimilar joints

Combination,
PM-weld

Pass
No

Method

Wire diameter,
mm

Current,
A

Voltage,
V

Polarity

Wire
Feed
Speed,
m/min

Welding speed,

m/min

Heat Input,
kJ/mm

8630—Alloy 625

1
2-n

GMAW
GMAW

1
1

150±15
170±15

22±3
24±3

DC + pulsed
DC + pulsed

3.6
4.2

0.32±0.05
0.35±0.05

0.495
0.56

F22-Alloy 625

1
2-n

GMAW
GMAW

1
1

150±15
170±15

22±3
24±3

DC + pulsed
DC + pulsed

3.6
4.2

0.32±0.05
0.35±0.05

0.495
0.56

F22-309

1
2-n

GMAW
GMAW

1.2
1.2

205
315

25.3±3
26.4±3

DC + pulsed
DC + pulsed

6.5
10

0.32±0.05
0.40±0.05

0.71
0.99


Sections from each joint were furnace heat-treated for 10 hours, followed by furnace cooling. Reference samples were instrumented with K-type thermocouples, in order to record sample temperature during PWHT. The F22 material was heat-treated at 650°C, whereas the 8630 joint was heat-treated at 675°C. Plots of the heat-treatments are shown in Fig.2.

Figure 2 – Thermal history of reference samples, during PWHT of: a) 8630-625, b) F22-625 and c) F22-309LSi dissimilar joints.
Figure 2 – Thermal history of reference samples, during PWHT of: a) 8630-625, b) F22-625 and c) F22-309LSi dissimilar joints.

2.2 Microstructural Analysis

After electro-discharge machining, all metallographic specimens were manually ground using 80 to 1000 SiC papers. Cloths were used for polishing, together with 3, 1 and 0.25μm diamond suspensions.  Final finishing included polishing using colloidal silica (OP-U).

The fusion zone microstructure, of all joints, was examined using an FEI Sirion SEM. Both backscatter (BS) and secondary electron (SE) modes were used. As hydrogen-assisted failure, within planar solidified zones, is commonly associated with the Steel-Alloy 625 joints, these joint types were also subjected to EBSD analysis, using a HKL Nordlys EBSD detector. Quantitative EDX analysis was performed, using a Princeton Gamma Tech detector. All SEM studies were performed using a 20kV accelerating voltage. All EDX linescans had 0.5µm point resolution.

Site specific TEM samples, were created using an FEI Quanta 3D dual beam system. The in-situ lift out procedure, described in [14], was adapted, to extract TEM thin-foils from polished specimens. A retractable, solid-state, backscatter detector facilitated the location of the dissimilar interfaces. A gaseous injection system was used to deposit protective Pt caps, above the thin-foil cross-sections. An Omniprobe micromanipulator, with a W needle, was used to attach the TEM foils, to Cu lift-out grids.

2.3 Thermodynamic modelling

Weld metal dilution was calculated, from the average nominal compositions of each joint, in 10% increments. For dilutions between 0 to 10% and 90 to 100%, calculations were made in 2% increments. Thermodynamic simulations were then conducted over the dilution range using the Thermo-Calc simulation software. The software enables the phase transformation temperatures, Scheil solidification range and phase equilibrium to be calculated.

The fusion zone microstructure, in a dissimilar 8630-Alloy 625 joint, has previously been evaluated by Alexandrov et al using similar methods [15], albeit with different thermodynamic databases. In this paper, the updated TCS Steels/Fe-Alloys Database Version 6.2 (TCFE6), was used in simulations above 50wt% iron. Simulations with 50wt% nickel or higher were conducted, using the Thermotech Ni-DATA database, version 7 (NIDAT7). Lowering the principal database component to near 50wt% carries increasing inaccuracy, due to the databases being evaluated, with substantially lower alloying content. Lower accuracy calculations are therefore indicated.

The Scheil-Gulliver solidification module, within the software package, was used to calculate the solidification mode. Solidification range was determined within the Scheil-Gulliver module, assuming equilibrium conditions. The Scheil-Gulliver module assumes the S/L interface is in equilibrium, and that no diffusion occurs within the solid (except for interstitial carbon diffusion). It also assumes complete mixing within the liquid phase. In order to calculate phase equilibria for the F22 and 8630 parent metalsthe gas, liquid, BCC, FCC, M23­C6, MC3, M6C, M5C2 and M3C­2 phases were added. The temperature was lowered from 2000 to 400°C and calculations made at 1°C intervals, until the solid fraction reached 0.99.

2.4 Diffusion Modelling

Phase and elemental distributions, resulting from diffusion, were calculated using Dictra. The TCFE6 database was used to assess phase equilibria, whilst the mobility database v.2 (MOB2) was used to assess the thermo-kinetic data. Simulations were carried out at an assumed PWHT temperature of 665°C for up to 10 hours. The observed FCC, BCC, M23C6 and M7­C3 phases were added to the simulation of the F22 and 8630-Alloy 625 joints. A Hashin-Shtrikman (H-S) homogenization function was used, in which the highest local volume fraction, becomes the alpha phase. Geometrically, the H-S function treats phases as concentric shells upon a matrix of a second phase material [16].

3. Results

3.1 Microstructural Analysis

The microstructures observed, in the three dissimilar joints, were consistent with the cross-section, illustrated schematically in Fig.1. The joints, had varying amounts of discontinuous PMZ, due to welding procedure and compositional differences. Therefore, to aid comparison, only continuous PMZ were selected for analysis. SEM BS electron images, together with EDX linescans, across the three fusion boundaries, are presented in Fig.3. Values corresponding to dilution of iron in the weld metal, are indicated on the micrographs. The arrows on the diagrams show the linescan traces, in which the compositions in the corresponding graphs, were measured.

Regions of partial mixing, were observed as transitional changes from the parent to weld metal. PMZ width, is determined by factors including heat-input and filler metal feed rate. Therefore, the two Alloy 625 joints, having been fabricated under the same welding conditions, can be compared. In the F22-joint, the PMZ width was approximately twice the width of that seen in the 8630 joint, although the PMZ width is likely to vary across thickness of the weld, due to localised differences in weld pool mixing.

Figure 3 - Low magnification SEM BS electron images across the fusion boundaries of the joints in the as-welded condition. Accompanying EDX linescans (yellow arrow) are shown. a) F22-Alloy 625, b) 8630-Alloy 625 and c) F22-309LSi dissimilar joints.
Figure 3 - Low magnification SEM BS electron images across the fusion boundaries of the joints in the as-welded condition. Accompanying EDX linescans (yellow arrow) are shown. a) F22-Alloy 625, b) 8630-Alloy 625 and c) F22-309LSi dissimilar joints.

Higher magnification linescans, from the same continuous PMZ regions as in Fig.3, are presented in Fig.4. The most significant composition changes occur within the planar zone of both of the Alloy 625 butterings. Of the two joints, the steepest composition gradient can be seen to occur across the 8630-Alloy 625 interface and was associated with a reduced planar solidified zone width. In the 8630 and F22-Alloy 625 interfaces, the planar zone width was approximately 8μm and 32μm, respectively.

Figure 4 - SEM BS electron images of the dissimilar joint fusion boundaries (as-welded condition), overlaid with EDX chemical distribution linescans, for the three dissimilar welds: a) F22-Alloy 625, b) 8630-Alloy 625 and c) F22-309LSi
Figure 4 - SEM BS electron images of the dissimilar joint fusion boundaries (as-welded condition), overlaid with EDX chemical distribution linescans, for the three dissimilar welds: a) F22-Alloy 625, b) 8630-Alloy 625 and c) F22-309LSi

The nickel content of all three joints decreases from the weld metal PMZ towards the parent metals. During cooling, the reduced nickel content results in a weaker driving force for stabilisation of austenite. Together with the high carbon content of the forging, this creates a scenario whereby martensite is likely to form, even under slow cooling conditions.

The Schaeffler diagram, Fig.5, can be used to predict microstructure formation, in the transition region between the two dissimilar metals, under cooling conditions associated with arc welding. The Ni and Cr weight percent (wt.%) equivalents, for the three dissimilar welds, as derived using equations 1 and 2, are superimposed on the diagram.

Creq = %Mo+1.5(%Si)+0.5(%Nb)    eq.1

Nieq = %Ni+30(%C)+0.5(%Mn)       eq.2

Figure 5 – Schaeffler diagram (adapted from [17]) with superimposed lines for the composition range of the three dissimilar joints.
Figure 5 – Schaeffler diagram (adapted from [17]) with superimposed lines for the composition range of the three dissimilar joints.

The PMZ consists of all compositions connecting each pair of points together. In all three pairings, the composition traverses through a martensitic region. A complete assessment of the chromium and nickel equivalencies over the whole dilution range is given in the appendices, together with predicted phase percentages. Due to the higher nickel concentration of the joints prepared with Alloy 625, a steeper composition gradient is found within the PMZ, compared to the 309LSi combination. Consequently, there is a reduced compositional band, over which, martensite is stable, hence resulting in the formation of a thinner band of martensite.

In the as-welded condition, martensite was present in its virgin form, whereas after PWHT, the structure presented a tempered morphology. Although the width of the martensitic band varied along the length of the interface, due to weld pool mixing, in general the F22-Alloy 625 system had a broader martensitic band than that of the 8630 joint. Differences in martensite width between the F22 and 8630-Alloy 625 joints are attributed to the higher nickel equivalency in the 8630 joint, as well as higher hardenability. The F22-309LSi dissimilar weld exhibited the widest martensitic region. In the Alloy 625 joints, interface martensite typically measured less than 1μm wide, whereas in the F22-309LSi combination the same structure measured approximately 4μm wide, Fig 4c.

TEM samples were created across the fusion boundary in continuous PMZ regions, Fig.6. The most significant difference between the two LAS-Alloy 625 joints was found in the planar solidified region. In the F22 forging, no new precipitates were found, however in the 8630 system, predominantly chromium based carbides were present after 10 hours of PWHT, Fig 6b. The carbides were identified as M7C3 by EDX and are well documented as forming in heat-treated LAS-Alloy 625 dissimilar joints. The precipitation occurred approximately 200nm from the apparent BCC/FCC boundary, though the width of the carbide band was not ascertained, as the TEM thin-foil did not encompass the whole width of the planar solidified region. The presence of carbides in this sample is an important observation, as previous studies have found that M7C3 carbides promote the embrittling effects of hydrogen [11] [2].

Figure 6 – TEM images of the three dissimilar welds after PWHT, showing a) 8630-Alloy 625 in the as-welded condition, b) 8630-Alloy 625 after PWHT. c) F22-Alloy 625 in the as-welded condition, d) F22-Alloy 625 after PWHT and e) F22-309LSi after PHWT.
Figure 6 – TEM images of the three dissimilar welds after PWHT, showing a) 8630-Alloy 625 in the as-welded condition, b) 8630-Alloy 625 after PWHT. c) F22-Alloy 625 in the as-welded condition, d) F22-Alloy 625 after PWHT and e) F22-309LSi after PHWT. Broken lines delineate the boundary between martensite and weld metal.

EBSD analysis of the interfacial regions, was used to determine phase constitution and grain texture. For clarity, the phase maps are overlaid with associated electron band-contrast images in Fig.7. The polished sections were analysed using a 0.5μm sampling interval. Inverse pole figure plots are given relative to the observed surface plane. In each plot, the BCC phase is shown in blue, whilst FCC is shown in red.

In both of the Alloy 625 dissimilar joints the iron-rich parent metal was determined to be BCC. In continuous PMZ regions, Fig.7 a and b, the BCC/FCC boundary appears to be nominally flat, however the same is not true within discontinuous PMZ regions, Fig.7c and d. The interface presented in Fig. 7c, shows a discontinuous PMZ, in which the  BCC/FCC boundary is irregular. These penetrations are likely to have occurred during welding, as opposed to during PWHT. The penetrations are shown to have an FCC structure before PWHT, but are likely to remain so after PWHT, due to the selection of PWHT temperature.

The parent metal and molten metal, within the GB penetration, are likely to lie within the steep composition gradient. Both near-interface sides of the fusion zone, are predicted to form austenite, initially, with overlapping austenite temperature ranges, Table 3 (appendices) and Fig.8. In accordance with Alexandrov et al [15], the solidification sequence and temperature ranges of both sides may promote epitaxial solidification of the weld metal from the neighbouring steel HAZ grains.

Figure 7 - Phase constitution (overlaid with band contrast image) and EBSD maps: a and b) F22-Alloy 625 joint, continuous PMZ in the as-welded condition, c) F22-Alloy 625 joint, discontinuous PMZ in the as-welded condition, d) 8630-Alloy 625 joint af
Figure 7 - Phase constitution (overlaid with band contrast image) and EBSD maps: a and b) F22-Alloy 625 joint, continuous PMZ in the as-welded condition, c) F22-Alloy 625 joint, discontinuous PMZ in the as-welded condition, d) 8630-Alloy 625 joint after 10 hours of PWHT showing a swirl adjacent to a continuous PMZ.

In all of the EBSD maps presented, a planar solidified zone is seen, regardless of PMZ type. Grains within the planar zone had a high aspect ratio; the long sides were aligned parallel to the dissimilar interface. High angular misorientation, is observed between planar zones and neighbouring dendritic regions. Elongated grains, perpendicular to the fusion boundary, indicated grain growth towards the centre of the weld pool, during solidification.

3.2 Thermodynamic modelling

Figure 8 - Thermo-Calc prediction of phase equilibrium in a) the F22 and b) the 8630 parent metals.
Figure 8 - Thermo-Calc prediction of phase equilibrium in a) the F22 and b) the 8630 parent metals.

Predicted phase equilibrium calculations for the 8630 parent metal, Fig.8, shows that above the martensite transformation temperature, complete dissolution of the M23C6 and M7C3 carbide phases is possible. Chemistry differences, between the 8630 and lower carbon content F22 forging, result in dissimilarities in phase content; cementite is not stable in the F22 forging, whereas M6­C carbide is not stable in the 8630 forging. At a typical PWHT temperature of 650°C, complete dissolution of M23C6 is not predicted in the F22 forging.

Predicted solidification ranges, for the three dissimilar joints, are given in Fig.9. Overlaid on the graph, is the percentage of δ-ferrite at solidification, over the dilution range. For the 8630 joint, the undiluted parent metal initially solidifies as 36% δ-ferrite, reducing to 14% at 98% dilution of the filler wire. In the F22-Alloy 625 joint, δ-ferrite formation is extended further into the dilution range. In its undiluted state, a phase percentage of 84% δ-ferrite was calculated. 27% δ-ferrite was calculated at 92% dilution. In the F22-309LSi joint, δ-ferrite extends throughout the whole dilution range, though is reduced by mixing with the 309 filler metal.

Figure 9 - Thermo-Calc prediction of the solidification temperatures for the three dissimilar weldments, using the Scheil-Gulliver module (together with δ-ferrite solidification %): a) 8630-Alloy 625, b) F22-Alloy 625 and c) F22-309LSi dissimilar int
Figure 9 - Thermo-Calc prediction of the solidification temperatures for the three dissimilar weldments, using the Scheil-Gulliver module (together with δ-ferrite solidification %): a) 8630-Alloy 625, b) F22-Alloy 625 and c) F22-309LSi dissimilar interfaces.

For all three dissimilar joints, a decrease in the dilution of the filler metal by the LAS parent metal, results in a similar decrease in both liquidus and solidus temperatures. Due to the relative compositional similarity between the materials, the F22-309LSi joint has a lower liquidus-solidus temperature range, throughout the dilution range. In general, however, the solidification range increases with decreasing dilution of the filler metal.

EDX measured compositional gradients indicate, that in the planar solidified region of the F22-Alloy 625 joint, the dilution range is approximately 100 to 68%. In the 8630-Alloy 625 joint, planar solidified region dilution is between 100 and 46% dilution. Comparing the solidification range with these measured values, a sharp solidification temperature gradient would be expected within the planar zone after welding. The results suggest that, in the F22-Alloy 625 joint, from the parent metal at the fusion boundary, a 32µm thick planar band is predicted to exist, which has a liquidus temperature gradient of approximately 150°C. In the 8630 joint, the planar band is predicted to be narrower, at approximately 10µm, and has a liquidus temperature gradient of approximately 150°C.

Figure 10 - Thermo-Calc prediction of carbon activity for: a) the parent metals and b) the weld metals. Typical PWHT temperature ranges are indicated.
Figure 10 - Thermo-Calc prediction of carbon activity for: a) the parent metals and b) the weld metals. Typical PWHT temperature ranges are indicated.

The temperature dependency of carbon thermodynamic activity, for the parent and weld metals, is presented in Fig.10. High values of carbon activity are present in both the parent metals, when compared to the Alloy 625 and 309LSi filler metals. Carbon is likely to diffuse in the direction of its thermodynamic activity gradient. High carbon activity in the parent metals, combined with high concentration gradients, will encourage diffusion of carbon from the forging, towards the weld metal. The higher the heterogeneity between the diffusion couple, the higher the likelihood of carbon redistribution during PWHT. The activity of carbon in F22 is seen to increase with temperature from 400 to 780°C. Conversely, over the same temperature range, the carbon activity decreases considerably, in the 8630 material. Despite the contrasting carbon activity trends, at typical PWHT temperatures, the carbon activity in the 8630 material is approximately 12 times higher than in the F22.

At typical PWHT temperatures, carbon activity in the weld metals remains comparatively low. This is attributed to the diffusivity in austenite being considerably lower than in ferrite and the carbon solubility being higher. The reduced nickel content in the 309LSi, results in a higher carbon activity at typical PWHT temperatures, compared to the Alloy 625.

3.2 Diffusion Modelling

Quantitative measurements of carbon concentration profiles are difficult due to the low atomic number of carbon, however, they were modeled in Dictra, as shown in Fig.11, albeit as an abrupt transition between the dissimilar metals. In practice, chemical gradients between the two dissimilar metals are present, in which a transitional increase in chromium is observed. Chromium is known to lower the chemical potential of carbon [18], [19]. Thus, a discontinuity in chemical potential of carbon would be expected across the interface. In the models, diffusion during PWHT resulted in 1.5 and 2.25wt% carbon in the weld metals of the F22 and 8630-Alloy 625 joints, respectively. The driving force for diffusion stems from a higher diffusivity of carbon in ferrite than in austenite.

Diffusion of carbon into the weld metal also resulted in the formation of a carbon-depleted region in the ferritic side, which results in a reduction in HAZ hardness.

Carbon diffusion into the austenitic side of the Alloy 625 joints resulted in the solubility limit being reached, causing the precipitation of new phases. M23C6 carbide is predicted to form, in the austenitic side of both dissimilar joint types. The formation of this carbide, has been observed in LAS-Alloy 625 joints [2], but not in the austenitic side. MC3 carbide is predicted to form, within the weld metal, of the 8630-Alloy 625 joint. This is an interesting observation, as the simple diffusion couple does not account for the morphology and chemical gradients that are present in real dissimilar welds. However, the prediction of M7C3 formation, after 10 hours PWHT, in the 8630 weld, but not in the F22-Alloy 625 joint, is consistent with TEM observations.

Figure 11 - Calculated phase distribution and carbon weight percentage values for the LAS-Alloy 625 joints after 10 hours of PWHT at 665°C. a) F22-Alloy 625 b) 8630-Alloy 625 and c) 8630-Alloy 625 showing the formation of M7C3.
Figure 11 - Calculated phase distribution and carbon weight percentage values for the LAS-Alloy 625 joints after 10 hours of PWHT at 665°C. a) F22-Alloy 625 b) 8630-Alloy 625 and c) 8630-Alloy 625 showing the formation of M7C¬3.

4. Discussion

The formation of embrittled weld fusion boundary regions, is determined primarily by the welding process and parent/weld metal material combination used. In the F22-309LSi combination, the relative similarity between the parent and weld metals, reduces the chemical gradient with consequent broadening of zones exhibiting transitional microstructures. This is most evident, when comparing interface martensite; the highly dissimilar 8630-Alloy 625 combination, produced the narrowest martensite band and planar solidified region. The steep weld metal composition gradient, appeared linked to planar solidified zone width. The F22-Alloy 625 weld has a considerably broader planar zone and a less sharp compositional change, than the 8630-Alloy 625, despite the use of a similar welding procedure. The steepest composition gradients, might be expected within the planar zone, due to the fact that larger composition changes occur in planar solidified regions, compared to cellular or dendritic areas [20].

During PWHT, carbon saturation within the planar zone is accompanied by the formation of M7C3. The width of the planar zone, might therefore play an important role in the formation of embrittling carbides. However, the extent of carbon diffusion, beyond the planar region, has not yet been determined.

Diffusion modelling using Dictra, has successfully predicted M7C3 carbide formation within the first 15μm of the Alloy 625 weld metal, which has been confirmed via TEM observations The band of carbide is predicted to form in this region, despite implementation of a simple diffusion couple model, in which the planar solidified morphology, was not taken into account.

As well as redistributing residual stresses, PWHT also tempers virgin martensite and increases the ductility of the heat affected zone. Synchronously, peak hardness associated with interface martensite may be reduced, provided the PWHT temperature is kept below the Ac1 transformation temperature. Thus, it is proposed that during PWHT there are  a number of different mechanisms taking place: 1) martensite tempering 2) redistribution of residual stress 3) the formation of phases in the interfacial transition zone which might exacerbate hydrogen embrittlement, such as M7C3 and 4) some degree of de-carburisation of the forging HAZ. The latter point suggests that lengthy PWHTs, in which a significantly de-carburised zone forms within the HAZ, can cause loss of strength and toughness.

5. Conclusions

The following conclusions can be drawn from this work:

  • Along with welding procedure, the chemical composition of the forging and weld metal influence the formation of hydrogen-assisted cracking, susceptible microstructures. Within the planar zone of LAS-Alloy 625 welds, a steep composition gradient exists.
  • Martensite can be successfully tempered during industry standard PWHTs, however, diffusion of carbon might result in the formation of the detrimental M7C3 carbide phase, within the planar solidified zone. At typical PWHT temperatures, there is a reduced carbon activity in F22, compared that in 8630 forgings. The extent of carbon diffusion into the weld was also reduced.
  • Hydrogen-assisted cracking of the planar zone, might be mitigated by altering forging chemistry. Unlike the 8630-Alloy 625 combination, PWHT of the F22-Alloy 625 joint for 10 hours, did not result in the formation of M7C3. On this basis, the F22 forging represents a possible solution to the problem of hydrogen-assisted cracking within the planar solidified zone, of heat-treated LAS-Alloy 625 welds.

Acknowledgments

This work was jointly funded by TWI Ltd., Cambridge, and MintWeld, a European Commission Seventh Framework Programme (FP7) sponsored project (contract No: NMP3-SL-2009-229108).

References

  1. J. D. Burk, C. L. Ribardo, and B. P. America, “OTC 20401 Thunder Horse – Materials, Welding and Corrosion Challenges and Solutions,” in Offshore Technology Conference, 2010.
  2. J. Fenske, “Microstructure and Hydrogen Induced Failure Mechanisms in Iron-Nickel Weldments,” 2010.
  3. J. Fenske, M. Hukle, B. Newbury, J. R. Gordon, R. Noecker, and I. Robertson, “Hydrogen Induced Mechanical Property Behaviour of Dissimilar Weld Metal Interfaces,” in International Conference on Offshore Mechanics and Arctic Engineering, 2011, pp. 1–8.
  4. J. Fenske, I. M. Robertson, R. Ayer, M. Hukle, D. Lillig, and B. Newbury, “Microstructure and Hydrogen-Induced Failure Mechanisms in Fe and Ni Alloy Weldments,” Metallurgical and Materials Transactions A, vol. 43, no. 9, pp. 3011–3022, Jun. 2012.
  5. M. F. Gittos, “Resistance of Dissimilar Joints Between Steel and Nickel Alloys to Hydrogen-Assisted Cracking,” in NACE, 2008, pp. NACE2008–08095.
  6. M. Tong, G. Duggan, J. Liu, Y. Xie, M. Dodge, L. Aucott, H. Dong, R. L. Davidchack, J. Dantzig, O. Barrera, A. C. F. Cocks, H. Kitaguchi, S. Lozano-Perez, C. Zhao, I. Richardson, A. Kidess, C. R. Kleijn, S. Wen, R. Barnett, and D. J. Browne, “Multiscale, Multiphysics Numerical Modeling of Fusion Welding with Experimental Characterization and Validation,” JOM, vol. 65, no. 1, pp. 99–106, Nov. 2012.
  7. G. F. Li, E. A. Charles, and J. Congleton, “Effect of post weld heat treatment on stress corrosion cracking of a low alloy steel to stainless steel transition weld,” vol. 43, pp. 1963–1983, 2001.
  8. V. Olden, “The effect of PWHT on the material properties and micro structure in inconel 625 and inconel 725 buttered joints,” International Conference on Offshore Mechanics and Arctic Engineering, vol. 3, pp. 109–115, OMAE2003–37196, 2003.
  9. P. Kvaale and G. Rorvik, “Experience with Qualification and use of Stainless Steels in Subsea Pipelines,” in International Conference on Offshore Mechanics and Arctic Engineering, 2004, pp. 945–952, OMAE2004–5136.
  10. Y. K. Yang and S. Kou, “Weld-Bottom Macrosegregation Caused by Dissimilar Filler Metals,” Welding Journal, vol. 86, pp. 379–387, 2007.
  11. M. F. Dodge, H. B. Dong, and M. Milititsky, “Environment – Induced Cracking In Weld Joints In Subsea Oil And Gas Systems - Part II,” in International Conference on Offshore Mechanics and Arctic Engineering, 2013.
  12. M. F. Dodge, H. B. Dong, M. Milititsky, R. P. Barnett, V. F. Marques, and M. F. Gittos, “Environment–Induced Cracking in Weld Joints in Subsea Oil and Gas Systems - Part I,” in International Conference on Offshore Mechanics and Arctic Engineering, 2012, p. 305.
  13. V. C. M. Beaugrand, L. S. Smith, and M. F. Gittos, “Hydrogen Embrittlement of 8630M/625 Subsea Dissimilar Joints: Factors That Influence the Performance,” in International Conference on Offshore Mechanics and Arctic Engineering, 2009, pp. 227–236.
  14. S. Lozano-Perez, “A guide on FIB preparation of samples containing stress corrosion crack tips for TEM and atom-probe analysis.,” Micron (Oxford, England : 1993), vol. 39, no. 3, pp. 320–8, Jan. 2008.
  15. B. T. Alexandrov, J. C. Lippold, J. W. Sowards, a. T. Hope, and D. R. Saltzmann, “Fusion boundary microstructure evolution associated with embrittlement of Ni–base alloy overlays applied to carbon steel,” Welding in the World, vol. 57, no. 1, pp. 39–53, Dec. 2012.
  16. H. Larsson and L. Höglund, “Multiphase diffusion simulations in 1D using the DICTRA homogenization model,” Calphad, vol. 33, no. 3, pp. 495–501, Sep. 2009.
  17. A. L. Schaeffler, “Constitution Diagram For Stainless Steel Weld Metal,” Metal Progress, vol. 56(11), pp. 680–680b, 1956.
  18. J. N. Dupont and R. E. Mizia, “Review of Dissimilar Metal Welding for the NGNP Helical-Coil Steam Generator,” Idaho Falls, Idaho, 2010.
  19. K. Laha, K. S. Chandravathi, K. B. S. Rao, and S. L. Mannan, “An Assessment of Creep Deformation and Fracture Behavior of 2 . 25Cr-1Mo Similar and Dissimilar Weld Joints,” vol. 32, no. January, pp. 115–124, 2001.
  20. L. P. Karjalainen, “Weld fusion boundary structures in aluminium and Al-Zn-Mg alloy.,” Zeitschrift fur Metallkunde, 1979. 

Appendices

Table 3 - Summary of dilution calculations and Thermo-Calc solidification analysis.  Percentage BCC and FCC were calculated under equilibrium conditions.  Liquidus and Solidus temperatures were calculated using the Scheil-Gulliver module (under equilibrium conditions).  Percentage δ-ferrite was calculated using the Scheil-Gulliver module under non-equilibrium conditions.

Table 3

For more information please email:


contactus@twi.co.uk