Intermetallic Phases and Steel Corrosion Resistance

+ TWI Ltd
Granta Park, Great Abington, Cambridge, CB1 6AL, UK

* Now at Capcis Ltd
Capcis House, 1 Echo Street, Manchester, M1 7DP, UK

 

Stainless Steel World Conference The Hague, Netherlands, 13-15 November 2001
Manuscript number : P0185

Abstract

Superduplex, superaustenitic and nickel alloys have exceptional corrosion resistance and have proven performance in many aggressive environments. However, these highly alloyed materials are susceptible to the formation of intermetallic phases when exposed to thermal cycles in the range 600-1000°C as may typically be experienced during welding.

The study examined four alloys: UNS S31254, S32760, S32750 and N08825, which were welded, using an overmatching Ni-Cr-Mo filler in the case of the superaustenitic steel, nickel overalloyed superduplex fillers for the superduplex stainless steels and AWS E/ERNiCrMo-3 type consumables for alloy N08825. Welds were produced with different levels of arc energy representing conditions: (i) well within typical industrial practice, (ii) at the top end of typical industrial practice and (iii) above typical practice. The volume fraction of intermetallic phase was determined for each material and welding condition. The corrosion performance of the welds was assessed using short term critical pitting temperature tests, and also using long term tests in simulated service environments: chlorinated seawater, CO₂/O₂ brine, CO₂/H₂S brine and a severe refinery environment.

The results showed that using good industrial practice, welds containing less than 1% intermetallic were produced in superduplex stainless steel and between 1.4 and 2.4% weld metal intermetallic was formed in the highly alloyed nickel base weld metals. In the case of superduplex consumables, this increased to between 1.7 and 2.9% when using conditions above typical industrial practice. The increase in arc energy was associated with a reduction in critical pitting temperature. However, in simulated service environments, the welds showed no significant corrosion. Recommendations are made with regard to the setting of fitness for service criteria.

1. Introduction

The excellent localised corrosion resistance of superduplex and superaustenitic stainless steels and Ni-Fe-Cr alloys stems from the high levels of alloying elements, in particular Cr, Mo and N. However, these highly alloyed materials are susceptible to the formation of intermetallic phases, which form at temperatures between around 600°C and 1000°C. Such temperatures are experienced during a weld thermal cycle, giving the possibility of intermetallics being formed in weldments, especially for high alloy grades, where precipitation can begin within a few tens or hundreds of seconds ^[1,2] .

Highly alloyed stainless steels and Ni-Fe-Cr alloys are susceptible to localised corrosion, e.g. pitting, in aggressive chloride containing media and intermetallic precipitation may reduce the corrosion resistance significantly. Intermetallic phases are rich in alloying elements, notably Cr and Mo, compared to the surrounding matrix and therefore, as the intermetallic phase is formed, Cr and Mo depleted zone can be created around the particles. The extent of the alloy depletion will depend upon the thermal cycles experienced and will determine the reduction in corrosion resistance ^[1,2] . The aims of the current work were to quantify the variation of intermetallic phase formation over a range of welding conditions for superduplex, nickel and superaustenitic alloys, and to determine their effects on corrosion resistance in realistic service environments.

2. Experimental procedure

2.1. Materials

Four corrosion resistant alloys were selected for testing, i.e. one superaustenitic grade: UNS S31254, two superduplex grades: S32760 and S32750, and one Ni-Fe-Cr alloy: N08825. These are referred to as materials A to D respectively. The materials were in plate form, with a nominal thickness of 10mm. Table 1a shows their chemical compositions.

2.2.Welding

The TIG process was utilised for the root and second passes, whilst fill and capping passes were deposited using the MMA process. No preheating was used and interpass temperatures were restricted to below 150°C in all cases. The shielding and backing gases were both argon. Typical weld metal compositions for the consumables used are listed in Table 1b . An overmatching Ni-Cr-Mo filler was used in the case of S31254, nickel overalloyed superduplex consumables were used for the two superduplex grades, and AWS E/ERNiCrMo-3 type consumables were used for alloy N08825.

Welding current and travel speed were varied to produce welds with three different levels of arc energy, i.e. low (L): from 0.9 to 1.6kJ/mm, medium (M): between 1.1 and 2.0kJ/mm and high (H): between 1.9 and 3.2kJ/mm, representing conditions (i) well within typical industrial practice, (ii) at the top end of typical industrial practice and (iii) above typical industrial practice.

The welds were coded with a 'W', followed by 'A, B, C or D' to represent the base steel and then 'L, M or H'. Where more than one weld of each type was made, these were distinguished by a number, e.g. 1 or 2.

2.3. Characterisation

Two different point counting methodologies were adopted for determining the localised peak and average volume fraction of intermetallic phases in each weld: (i) To determine 'peak' localised precipitate levels: 16 measurement fields were counted on a prepared metallographic specimen, in a square 4x4 field array, in areas identified by prior microscopic examination as containing the highest levels, at a magnification of either X1250 or X1400. (ii) To determine average precipitate levels: 16 measurement fields were counted, using a 100 point (10x10) grid, at random locations within the root and second weld runs. The magnification used was either X1250 or X1400. The phase balance was also determined in each of the following areas of the superduplex materials: parent steel, weld metal cap, weld metal root, cap HAZ and root HAZ.

2.4. Short term ferric chloride corrosion tests

Critical pitting temperature (CPT) ferric chloride tests were conducted generally following ASTM G48 Method C ^[3] . The weld surfaces were left as-welded except for degreasing, and the specimens sides and ends were prepared to 120 and 1200 grit finish respectively. Exposure commenced at 27.5°C. The temperature increment was 2.5°C. This procedure was repeated until either metal loss was detected by visual inspection or weight loss exceeding 20mg was recorded. The first temperature at which pitting occurred was taken as the CPT.

2.5. Long term corrosion tests in simulated service environments

2.5.1. Test samples

The specimens were ~25 by 50mm by full thickness, so that the exposed surfaces included both weld cap and root, and cut edges. The weld surfaces were degreased, whilst the specimen edges and ends were prepared to 120 and 1200 grit finish respectively. Specimen mass was recorded before and after test. Following completion of the corrosion test, samples were examined visually and if signs of corrosion were observed, examination in a scanning electron microscope (SEM) was undertaken.

2.5.2. Chlorinated seawater test

Duplicate weld samples of the superaustenitic and both superduplex grades were immersed in natural seawater held at a constant temperature of 40°C. The test vessel was open to the atmosphere, allowing diffusion of oxygen into the test solution. A diluted solution of sodium hypochlorite (approximately 1%) was continuously added to the test solution, to achieve a free chlorine level between 0.5 and 1.0ppm throughout the test. Electrochemical potential was recorded during the exposure period.

After 36 and 62 days exposure, the specimens were removed and inspected visually under a binocular microscope. Mass loss measurements were taken. As no evidence of corrosion was found, the test temperature was increased from 40°C to 53°C on re-immersion after the second inspection until the end of the test on day 91. Following removal from the test solution, the specimens were cleaned in water, inspected under a binocular light and re-weighed.

2.5.3. CO₂/O₂ Brine test

The CO₂/O₂-containing brine was designed to simulate produced water systems where temperatures can be high (80-100°C) and small amounts of O₂ can be present. Specimens of superaustenitic steel (UNS S31254), one of the superduplex grades (UNS S32750) and the Ni-Fe-Cr alloy N08825 were suspended in deaerated 3%NaCl solution at 80°C, a gas mixture of CO₂ and 0.4%O₂ was introduced and bubbled through the solution throughout the 90 day test period. The total test pressure was 2.5 barg, which gave a partial pressure of CO₂ of 2 bar. Following completion of the test, the specimens were cleaned in water, inspected visually and re-weighed. The mass loss for each specimen was also determined after a second cleaning operation, using 40% hydrochloric acid to remove tarnishing ^[4] .

2.5.4. CO₂/H₂S brine test

The H₂S-containing environment was chosen to match the NACE MR0175-99 recommended service limits for UNS S32760 ^[5] . The test solution comprised deaerated 12%NaCl, with an addition of 0.34 g/l of sodium bicarbonate to give a pH value of 4, at test temperature and pressure. Four point bend test specimens were taken from the two superduplex grades. The specimens had the root intact and were strained to give 0.2% plastic strain at the weld toe. After heating to test temperature, i.e. 100°C, the CO₂/H₂S/N₂ test gas mix was introduced to the vessel, providing CO₂ and H₂S partial pressures of 2 and 0.2 bar, respectively and the test lasted 90 days. At the end of the test period, the specimens were inspected as described in section 2.5.2 above.

2.5.5. Severe refinery test

The test involved preparation of NH₄Cl paste (160ml of distilled water and 330g of NH₄Cl), which was placed in a glass container inside an autoclave. Approximately one litre of distilled water was poured into the autoclave. After sealing, the system was purged with N₂ gas to remove oxygen prior to testing. Two sets of specimens were employed: one set was placed in the paste, the other was suspended in the vapour. All four materials were included in the test. The autoclave was heated to a temperature of 148°C and was pressurised to 18 barg with N₂ gas. The test duration was one month. The water in the bottom of the vessel provided approximately 20% humidity in the vapour phase, assuming a water vapour partial pressure of 4.5 bar. Following unloading of the autoclave, the specimens were washed in water, visually inspected for evidence of corrosion under a binocular microscope and re-weighed.

3. Results

3.1. Metallography

3.1.1. Material A: superaustentic steel - UNS S31254

Each of the weld metals of material A contained widespread interdentritic intermetallic phases, the volume fraction of which generally increased with increasing arc energy, Tables 2 and 3 . The peak volume fraction increased from 1.8% to 2.4%. Most dense precipitation was in re-heated areas. The weld cap and bands in the lower parts of re-heated passes showed less precipitation. The average volume fraction of intermetallics was lower, but showed a similar trend to the peak volume fraction results increasing from low arc energy (1.4%) to high arc energy (1.8%). Figure 1 shows a typical microstructure of the weld root in the high arc energy weld. The intermetallic particles were irregular in shape and typically 1µm to 4µm in all dimensions. No HAZ precipitation could be quantified, although, there was evidence of preferential etching of the HAZ grain boundaries and at higher arc energies some 'spottiness' and thickening of the boundaries when etched.

 

3.1.2. Material B: superduplex steel - UNS S32760

All welds in material B showed a typical ferrite-austenite structure, with islands of austenite in a ferrite matrix. Some small patches of secondary austenite were present, typically lying between weld passes, and most numerous for the deliberately 'abusive' weld procedures, i.e. the medium and high arc energy. The ferrite volume fractions for each of the welds are shown in Table 4 . The reported values were within the expected range ^[6] .

Welds of medium and high arc energy both contained intermetallic phases in a region encompassing the root and 2nd weld passes. In these samples, the peak volume fractions were 0.8% and 1.7% respectively in the middle of the areas of precipitation ( Table 2 ). Random point counting gave average volume fractions in the root and second pass of 0.6 % and 0.5% respectively ( Table 3 ). The low arc energy weld contained only a few particles in the weld root, possibly intermetallics. Figure 2 shows the root bead microstructure in sample WBH1. The intermetallic particles were irregular in size and morphology, Fig.2b, and tended to be elongated on the austenite-ferrite interfaces. Typical examples were 1-2µm wide and up to around 10µm in length.

Fig.2. Microstructure of the root of WBH1 UNS S32760 welds

The HAZ contained precipitates, with morphology suggesting that some were nitrides and some intermetallics. All of the different arc energy samples showed similar high temperature HAZ features with groups of etch pits within ferritic grains and decoration of ferrite sub-grain boundaries. These are consistent with chromium nitrides.

3.1.3. Material C: UNS S32750

Each weld contained a small fraction of intermetallic phases, predominantly in the second weld run and some small regions of secondary austenite towards the weld cap. The high temperature HAZ showed limited fine intragranular nitrides, but no intermetallic phases were found. The ferrite volume fraction in each weld is shown in Table 4 and was within the expected range ^[6] .

The peak level of intermetallic phases was highest in. the medium arc energy weld. This was determined to be 2.6% c.f. 0.8% in the low arc energy sample and 1.3% in high arc energy sample ( Table 2 ). However, random point counting revealed lower average volume fractions, which increased with increasing weld arc energy, Table 3 . As for S32760, typical precipitates were 1-2µm wide and up to around 10µm in length.

3.1.4. Material D: Ni-Fe-Cr alloy UNS N08825

All of the weld metals contained second phase interdendritic particles, which was fairly uniform throughout the weld metal. The volume fractions of the intermetallic phase particles are displayed in Tables 2 and 3 , and increase slightly with increasing arc energy.

Examination in the SEM confirmed the presence of second phase particles of irregular morphology and with dimensions up to 5µm. The HAZ precipitation could not be quantified in terms of size or volume fraction, but the grain boundaries in this zone showed preferential etching when compared to parent material, becoming increasingly pronounced at higher arc energy.

3.2. Ferric chloride testing

The results are displayed in Fig.3. The superaustenitic steel samples gave the highest CPT values, up to 55°C, whilst pitting was first observed on the superduplex alloys at temperatures between 35 and 40°C. Generally, as the weld arc energy increased, the CPT was reduced, although there was little difference between the medium and high arc energy samples. The parent material critical pitting temperatures calculated from published data ^[3] were found to be approximately 60°C. In all cases, the measured weld values were below the parent material value, as would be expected.

 

The location of attack was found to vary with the UNS S31254 specimens mainly suffering corrosion on the specimen sides at the fusion line. For the two superduplex grades, attack was again on the specimen sides, generally in the root or second/third pass weld metal region.

3.3. Long term corrosion tests in simulated service environments

3.3.1. Chlorinated seawater test

The weight change results, displayed in Fig.4, were small, which indicates that no significant corrosion occurred during the test period. In addition, there was no significant difference between the low and high arc energy weld results. The mass loss increased with time and the maximum recorded mass loss for the 91 day exposure was 5.7mg. This represents a very low uniform corrosion rate of <0.001mm/yr. The visual examination results support the weight change data, as the inspection after 36 days exposure only revealed some weldment discolouration and light etching, which revealed the weld metal.

The corrosion potentials that were recorded during the test period are plotted in Fig.5a, b and c. The initial potential readings were approximately -200 to -300mV SCE, but during the first few days of the test, the corrosion potentials of the specimens increased to values above 0 mV. During the three month test, potentials of +500 to +600mV were recorded on all the alloys. However, the potentials were not steady and significant fluctuations were observed. Typical behaviour is shown in the plot of time versus corrosion potential, Fig.5d, which was recorded after 19 days exposure.

Fig.5. Corrosion potentials during the chlorinated seawater tests:

 

3.3.2. CO₂/O₂ brine test

After exposure, all of the test specimens appeared tarnished brown, with some black discoloration around the weld cap. There was no visual evidence of pitting corrosion or of any crevice corrosion where the specimens had been supported, but some corrosion at weld spatter on the cap had occurred.

The mass loss results are given in Table 5 and are generally low, indicating no corrosion attack with the exception of WCL1 and WCH2-F2. The results after light brushing showed a small weight gain, probably reflecting the build-up of the Fe/Cr oxide on the weldment surface. Further oxidation of the weld oxides during the corrosion test could lead to an increase in weight and a brown discoloration. The mass losses recorded after cleaning in HCl suggested that these oxides were removed. The losses recorded for specimens WCL1 and WCH2-F2 were not accompanied by any visual evidence of corrosion and are considered unlikely to be the result of corrosion attack, but more likely to be caused by the removal of weld oxides and weld spatter. A similar magnitude mass loss was observed in a ferric chloride test, during initial exposure.

3.3.3. CO₂/H₂S brine test

Mass loss data are given in Table 6 . The visual examination performed after the test did not reveal any evidence of corrosion attack and all the specimens were similar in appearance. All the specimens suffered some mass loss during the test period (Table 6). The maximum value was 34mg on specimen WBH1-F4, but as for the CO₂/O₂ brine test, as there was no visual evidence of corrosion, the mass loss is believed to arise from loss of weld oxide. Although the differences were small, high arc energy welds gave higher weight loss than the low arc energy welds. The S32760 steel welds showed greater weight loss than the S32750 steel.

3.3.4. Severe refinery water test

The mass loss results, which are displayed in Table 7 , are very low. Visual examination confirmed that minimal corrosion had occurred.

4. Discussion

4.1. Precipitation characteristics

Over the arc energy range examined, the volume fraction of intermetallic phase in nickel based weld metals increased by only up to 50%, whilst superduplex weld metals showed an increase from less than 0.5% to around 2.6% maximum. It should be noted that the highest arc energies employed were well beyond the manufacturers' recommended levels and equally beyond the maximum levels that would typically be expected for industrial welding. Perhaps of greater significance is the fact that individual particle sizes varied little between the different arc energy levels. For superduplex weld metal, particles tended to grow along austenite/ferrite boundaries and become more numerous but the width of particles varied little with arc energy. Nickel based weld metals generally showed increased numbers of particles rather than larger particles at higher arc energy. Single phase weld deposits transform to intermetallic phase in alloy enriched interdendritic areas and dendrite size was relatively constant for the welds examined, typically requiring order of magnitude changes in cooling rate for significant changes to occur. In superduplex deposits, intermetallic tends to form in thin areas of ferrite between two islands of austenite. As austenite essentially does not transform to intermetallic phase, the particle size is limited to some extent by the spacing of austenite units.

4.2. Corrosion resistance

4.2.1. Ferric chloride tests

A general reduction in the CPT values was observed for the three alloys A, B and C as the arc energy increased. However, the reduction was not large; significant drops in CPT compared to parent plate material behaviour, up to 40°C, have been reported ^[7] for volume fractions of 1 to 4% intermetallic phase produced in superduplex steel by isothermal heat treatment (Fig.6). However, the scatter in this published plot of drop in CPT versus volume fraction of intermetallic is considerable and this is consistent with the size of individual precipitates being more influential than the volume fraction ^[1,2] . The temperature at which the precipitate is formed will also have an effect. For pitting resistance to be reduced, the depleted layer must be of sufficient size to contain a stable pit nucleus and the width of the depleted layer and the extent of alloy depletion are a function of diffusion rates. Hence, depleted layer width has been suggested to be the factor controlling corrosion resistance, rather than volume fraction. For example, the size of a stable pit nucleus might be 0.1µm ^[8] . It has been shown that precipitates produced by isothermal ageing, which are larger than those formed during welding, are more detrimental to CPT, for a given volume fraction ^[1] .

 

The present data is included with the published data in Fig.6 and it falls within the scatter band. The data has been plotted twice; firstly by using published parent material CPT values to calculate the reduction in CPT (the parent material value was ~60°C for all these alloys S31254, S32760 and S32750) and the 'peak' intermetallic volume fraction results, and secondly, by taking the difference between the low and high arc energy weld ferric chloride results and using 'average' intermetallic volume fraction data. It must be recognised that when comparing the weld and parent material CPT values, the significant reduction is not entirely the result of intermetallic formation. The two methods of depicting the data in Fig.6 represent the two extremes that might be considered.

Whilst intermetallic precipitation was more pronounced in the nickel weld deposits, they were sufficiently overalloyed that substantial loss of corrosion resistance would be required to bring the weld metal down to the parent steel level. Hence, in the superaustenitic and Ni-Fe-Cr alloy weldments, the corrosion resistance was much more likely to be controlled by the HAZ or fusion-line, where precipitation or formation of an unmixed zone (UMZ), with associated Mo segregation, may occur. The formation of a UMZ, which consists of melted parent material which has not mixed with the consumable, and its potentially limiting effect on corrosion resistance for welds in UNS S31254 has been well documented ^[9] . In the superaustenitic alloy the attack was observed on the specimen side at the fusion-line and is believed to be associated with the presence of a UMZ which has previously been found to be the area of lowest pitting resistance for welds in this material ^[9,10] . Furthermore, the reduction in corrosion resistance that is associated with the presence of welding oxide and the chromium depleted layer on the root and cap surface appears to be less significant than the effect of the UMZ in this instance.

For the two superduplex grades, attack was also on the specimen sides, generally in the root weld metal region or around mid-thickness, equivalent to the second/third passes. These areas were associated with the highest concentration of intermetallic phases. The results indicate that the root surface, which is normally the surface that is exposed to the environment, has a better pitting resistance in a G48 test than the weld metal on the specimen sides. This probably reflects the fact that intermetallic phases do not tend to form on the root surface but concentrate further into the weld, although in the high arc energy S32760, weld intermetallic phase was observed adjacent to the weld root toes. However, attack was not observed at that location.

The ferric chloride test is a very severe accelerated corrosion test and although it is used to rank corrosion performance of alloys, the significance of any comparison with corrosion resistance in quite different service conditions, e.g. as employed in the long term corrosion tests, is open to question.

4.2.2. Effect of intermetallic phase on corrosion resistance in service environments

No severe corrosion was observed in the long-term tests conducted in simulated service environments. The test environments represented service-like conditions and were considered a reasonably severe test of the weldment performance.

The chlorinated seawater test commenced at 40°C, 10°C above the NORSOK standard limit ^[11] and because no corrosion was detected the temperature was increased to 53°C. It is recognised that the NORSOK limit is considered severe and that the test specimens may have been 'conditioned' to some degree by exposure to the lower test temperature, but overall it is considered that the test has demonstrated the corrosion resistance of the welded specimens to chlorinated seawater.

Previous tests in CO₂/O₂-containing brine have been reported by Rogne and Johnsen ^[12] . They considered that at 80°C and 200ppb O₂, localised corrosion on S31254 might initiate and a maximum service temperature of 60°C was recommended for welded S31254, if 200ppb O₂ is present. The test conditions used were therefore more severe than those recommended, yet no corrosion was observed.

The CO₂/H₂S brine and NH₄Cl tests were based on conditions from NACE ^[5] and industrial experience respectively and were considered fairly demanding. The bend samples subjected to the sour test did not show any evidence of corrosion or cracking and the unstrained samples showed no significant corrosion. Similar behaviour was observed for the NH₄Cl tests. Modest mass losses, up to 65mg, were observed on a number of specimens. However, there was no visual evidence of localised corrosion on any of these specimens and it is believed that the mass loss arose largely from removal of weld spatter either during the test or post-test cleaning.

In summary, both the low and high arc energy weld samples had sufficient corrosion resistance to withstand the chlorinated seawater, CO₂/O₂ brine, CO₂/H₂S and NH₄Cl test environments. Therefore, materials with intermetallic phase of similar type and at comparable volume fractions would be considered suitable for service in these environments. Corrosion attack of the high arc energy specimens might have been expected if the welding process had produced intermetallic phase that significantly reduced corrosion resistance. However, the short-term tests indicated that the effect of the intermetallic was fairly small.

The intermetallic phase volume fractions produced in this work were low when compared to levels generated by isothermal ageing, but the range of arc energy was broad and went well beyond current recommended welding practice.

4.3. Practical implications

For optimum superduplex weldment corrosion resistance, welding conditions should be restricted, e.g. to 0.5 to 1.5kJ/mm arc energy and <100°C interpass temperature, to minimise intermetallic phase formation. Superaustenitic alloy UNS S31254 and Ni-Fe-Cr alloy N08825, do not require such close control of welding conditions to minimise weld metal intermetallic formation, but lower arc energy gives improved corrosion resistance, perhaps by minimising HAZ precipitation and UMZ segregation.

Metallographic examination for intermetallic phase is typically a part of weld procedure qualification for superduplex stainless steels. When some particles are formed, point counting is often performed, and attempts made to set an allowable upper limit, typically around 1-2%. However, the size of the precipitates is perhaps of significance as larger particles are expected to produce greater depleted zones. From this work, welds with 2-3% intermetallic phase may be considered for corrosion resistant service. Demonstration of fitness for purpose should be on the basis of realistic simulated worst case service tests, rather than short-term quality control type tests.

5. Conclusions

1. Generally the intermetallic volume fraction increased with increasing arc energy in superduplex, superaustenitic stainless steels and nickel alloys.
2. Welds in superduplex steel (UNS S32760 and S32750) had weld metal intermetallic volume fractions less than 1% when welded within typical good industrial practice limits.
3. Welds produced with high arc energy, up to 3.2kJ/mm, i.e. above typical industrial practice, had maximum weld metal intermetallic volume fractions, between 1.7% and 2.9%.
4. Interdendritic intermetallic precipitates formed in highly alloyed nickel base weld metals. When typical good industrial practice was followed, the volume fraction of intermetallic was between 1.4 and 2.4%.
5. Little variation of intermetallic particle size with arc energy was observed.
6. The corrosion resistance of weld specimens containing intermetallic precipitates produced by welding with arc energy well above typical industrial limits, was reduced by a modest margin in short-term pitting tests, but in simulated service environments, no difference in corrosion performance was recorded, as none of the samples showed significant corrosion.

6. Acknowledgements

This work was funded by Avesta Polarit, BP Exploration Operating Co Ltd, DERA, ESAB AB, Esso Engineering (Europe) Ltd, Fabrique de Fer de Charleroi SA and Shell UK Exploration and Production. The support of the sponsors is gratefully acknowledged. The authors also wish to thank all their colleagues at TWI who assisted with the practical work.

7. References

1. Francis R: 'Discussion on influence of σ phase on general and pitting corrosion resistance of SAF 2205 duplex stainless steel'. Brit Corr J 1992 27 (4) 319-320.
2. Gooch T G: 'Corrosion behaviour of welded stainless steel'. Weld J 1996 75 (5) 135-154.
3. ASTM G48-97 ^{ε 1} : 'Standard test methods for pitting and crevice corrosion resistance of stainless steels and related alloys by use of ferric chloride solutions'.
4. NACE Publication 3M182: 'Corrosion Testing of Chemical Cleaning Solvents'.
5. NACE MR0175-99: 'Standard material requirements for sulphide stress cracking resistant metallic materials for oilfield equipment'.
6. Gunn R N: 'Duplex stainless steels'. Abington Publishing, Cambridge, 1997, 118.
7. Gunn R N: 'Effect of thermal cycles on the properties of 25%Cr duplex stainless steel plates - preliminary studies'. TWI Members Report 505, 1995.
8. Mattin S P: 'Nucleation of corrosion pits in stainless steel'. PhD dissertation, Cambridge University, 1994.
9. Gooch T G and Elbro A C: 'Welding corrosion resistant high alloy austenitic stainless steel'. Proc Corrosion in Natural and Industrial Environments: Problems and Solutions, Grado, Italy, NACE InternationalItalia, Monza, 1995, 351.
10. Stenvall P, Liljas M and Wallen B: 'Performance of high molybdenum superaustenitic stainless steel welds in harsh environments', Proc Corrosion '96, Denver USA, NACE Houston, 1996, paper 419.
11. NORSOK Standard M-001 Rev. 2 November 1997: 'Materials Selection'.
12. Rogne T and Johnsen R: 'The effect of CO ₂ and O ₂ on the corrosion properties of UNS S31254 and UNS S31803 in brine solution'. Proc Corrosion '92, NACE International, Houston, paper 295.

 

Table 1a Chemical analyses of each of the parent materials.

Material	Element, % (m/m)											PRE N/PRE W
	C	Si	Mn	P	S	Cr	Cu	Mo	Ni	W	N
S31254 (A)	0.014	0.40	0.40	0.020	<0.001	20.1	0.67	6.23	18.0	0.05	0.210	44.0
S32760 (B)	0.03	0.38	0.64	0.027	0.003	25.4	0.62	3.53	7.1	0.64	0.203 +	41.4
S32750 (C)	0.02	0.21	0.78	0.025	0.002	24.8	0.29	3.84	6.8	0.06	0.264	41.7
N08825 (D) **	0.010	0.30	0.69	0.018	<0.001	22.5	1.85	3.31	bal	0.05	0.010	33.6

**Ti 0.62     ^**Nb 0.01     ^**Fe 30.5     ^**Al 0.09

 

Table 1b Welding consumables

Parent material	Consumable type	Typical weld metal chemical composition Elements, wt%
		C	Si	Mn	Cr	Ni	Mo	Other
S31254	Ni-Cr-Mo wire	0.02	0.20	0.5	23.0	60.0	16.0	-
(A)	Ni-Cr-Mo electrode	0.02	0.3	0.7	25.0	Bal	15	-
S32760	Superduplex wire	0.015	0.4	0.7	25	9.3	3.7	W 0.6
(B)								Cu 0.7
								N 0.23
	Superduplex electrode	0.03	0.3	0.7	25	9.3	3.6	W 0.7
								Cu 0.7
								N 0.23
S32750	Superduplex TIG wire	0.02	0.3	0.4	25.0	9.5	4.0	N 0.25
(C)	Superduplex electrodes	0.03	0.5	1.0	25.0	9.5	3.6	N 0.22
	Ni-Cr-Mo electrodes	0.03	0.2	0.2	26.5	Bal	14
N08825	ER NiCrMo-3	<0.03	<0.02	<0.5	22	65	9	Nb 3.5
(D)	E NiCrMo-3	<0.03	0.4	0.4	21.0	64	9.5	Nb 3.3
								Fe 3.0

Table 2 Peak volume fraction of intermetallic phase (%) counted systematically in a 4x4 field array.

Parent material	Peak volume fraction (%)
	Low arc energy	Medium arc energy	High arc energy
A (UNS S31254)	1.8 (0.70)	2.1 (0.7)	2.4 (0.8)
B (UNS S32760)	0.2 (0.4)	0.8 (0.7)	1.7 (1.0)
C (UNS S32750)	0.8 (0.7)	2.6 (1.1)	1.3 (0.8)
D (UNS N08825)	2.3 (0.7)	2.8 (0.6)	2.7 (0.8)

Figures in parentheses are 95% confidence limits

Table 3 Volume fraction of intermetallic phase (%) counted randomly across 16 fields.

Parent material	Average volume fraction (%)
	Low arc energy	Medium arc energy	High arc energy
A (UNS S31254)	1.4 (0.8)	1.9 (0.9)	1.8 (0.8)
B (UNS S32760)	0.3 (0.2)	0.6 (0.5)	0.5 (0.4)
C (UNS S32750)	0.2 (0.2)	0.3 (0.2)	0.4 (0.5)
D (UNS N08825)	2.4 (0.6)	2.4 (0.5)	2.9 (0.8)

Figures in parentheses are 95% confidence limits

Table 4 Ferrite volume fraction (%) in each superduplex weld.

Parent steel	Weldment ID	Ferrite %
		Parent material	Cap	Root	Cap HAZ	Root HAZ
B (UNS S32760)	WBL1		49 (7)	46 (5)	62 (6)	55 (7)
	WBM1		51 (7)	40 (4)	60 (7)	60 (7)
	WBH1	53 (7)	48 (7)	42 (3)	67 (6)	58 (8)
C (UNS S32750)	WCL1		44 (5)	41 (3)	56 (5)	44 (4)
	WCM1	47 (7)	43 (5)	45 (4)	55 (5)	49 (6)
	WCH1		44 (4)	44 (5)	51 (5)	45 (5)

Figures in parentheses are 95% confidence limits

Table 5 Mass change results for CO₂/O₂ brine corrosion test

Material	Sample ID	Mass change after brushing lightly (mg)	Mass change after cleaning in HCl (mg)
A (UNS S31254)	WAL1-F4 WAM1-F2 WAH1-F4	+10 +10 +13	-9 -20 -6
C (UNS S32750)	WCL1 WCH2-F2	+0 +12	-44 -65
D (UNS N08825)	WDL1 WDH1	+12 +11	-7 -8

Table 6 Mass loss results for the CO ₂/H ₂S brine corrosion test

Material	Sample ID	Mass loss (mg)
B (UNS S32760)	WBL1-F4 WBM1-F2 WBH1-F4	-21 -13 -34
C (UNS S32750)	WCL1-F4 WCH1-F3 WCH2	-4 -8 -5

Table 7 Mass loss results for the severe refinery water corrosion test

Material	Sample ID	Mass loss (mg)
A (UNS S31254)	WAL-1P WAL-1V WAH-1P WAH-1V	-6 -5 -8 -3
C (UNS S32750)	WCL-1P WCL-1V WCH-1P WCH-1V	-9 -9 -8 -3
D (UNS N08825)	WDL-1P WDL-1V WDH-1P WDH-1V	-4 -3 -1 -2