Welding New Stainless Steels for the Oil and Gas Industry

Presented at Corrosion NACE Expo 2000, 55th Annual Conference and Exposition, Orlando, USA
26-31 March 2000

Abstract

Stainless steels play an essential role in all sectors of the oil and gas industry, both upstream and downstream. In the past, this has predominantly entailed established grades, but the last 10 to 20 years have seen major changes in the design and use of stainless steels for oil and gas service. A range of low carbon 13%Cr martensitic, highly alloyed austenitic and two phase ferritic-austenitic steels has been developed offering improved properties and cost effectiveness for a wide range of duties, and the materials are being employed in increasing tonnages. However, the practical application of such steels is critically dependent on the use of welding for fabrication, since a welding operation can have a significant influence on the corrosion performance of the completed assembly.

The paper considers the welding behaviour of the new steel types relative to older grades. Potential problem areas are identified, with the aim of indicating the controls necessary so that full advantage can be taken of the attractive base metal properties.

Introduction

The diverse conditions experienced in oil and gas plant require close attention to material selection in order that the most cost effective and reliable materials can be employed as appropriate. This has led to a high level of awareness of the properties offered by stainless steels, and in consequence these materials play a major role in both upstream and downstream sectors of the oil and gas industry. In large part, they are employed in plant and associated equipment where the corrosion resistance of plain carbon or low alloy steels is inadequate, although the austenitic grades in particular find applications where their excellent elevated temperature or cryogenic mechanical properties are of advantage.

Much oil and gas technology is mature practice and thus use of stainless steels mainly entails well standardised grades, with austenitic and martensitic alloys being dominant and, historically, with less utilisation of ferritic alloys. However, the last 20 years or so have seen development of new alloys which differ appreciably from traditional materials in composition and microstructure, and these steels are being increasingly specified for oil and gas service. The present paper considers the welding characteristics of stainless steels of significance to the oil and gas industry, with an emphasis on the newer materials.

Martensitic Stainless Steels

Established Grades

Welding procedures for hardenable 13%Cr steels are designed primarily to avoid hydrogen fabrication cracking and to obtain adequate weld area toughness ^[1] . This normally involves both preheat prior to welding and postweld heat treatment (PWHT). Given appropriate control of temperature at all stages, it is a fairly straightforward operation to obtain sound joints in plain 13%Cr alloys giving the desired mechanical properties. Particularly stringent attention is necessary with the higher carbon grades such as AISI 420, but welding can nevertheless be carried out reliably to meet appropriate specification requirements.

For sour service, NACE MR0175 specifies maximum hardness levels for base metals, and requires that welded joints should meet similar hardness limits. There has been very little attention to determining the SSC behaviour of weldments in 13%Cr steels, and some reduction in weld metal behaviour might be expected, recognising the inevitability of alloy element segregation during solidification and of a high inclusion population. Nonetheless, compliance with base metal hardness limits has normally yielded satisfactory service of welded joints. For martensitic stainless steels, a postweld tempering operation is necessary to obtain the requisite hardness levels. This poses no particular problems with plain 13%Cr alloys, but some difficulty is commonly experienced by fabricators with low carbon grades such as F6NM in achieving the prescribed hardness maximum ^[2] .

These alloys contain nickel to ensure hardenability which leads to a significant reduction in the Ac ₁ temperature. Postweld heat treatment is therefore necessarily carried out at fairly low temperature, typically 580-620°C. While a three stage heat treatment is stipulated for the base metals in MR0175, this is seldom applied to welded joints, although a two stage procedure is common to try to achieve the requisite degree of softening. The MR0175 standard utilises the Rockwell system to describe weld area hardness, but this is not appropriate for welded joints, because the width of the transformed heat affected zone (HAZ) may be similar or smaller in size than the Rockwell C indent, so that material adjacent to the HAZ is also sampled during the test. Weld area hardness should, therefore, be assessed using a system with a smaller indent such as the Vickers or Rockwell N methods. However, it must be appreciated that the hardness conversion tables for ferritic steels in, for example, ASTM E140 are not appropriate for low carbon 13%Cr steels. Work at TWI supported the Hayes-Patrick ^[3] relationship for these materials ( Fig.1), which indicates that the NACE limit of 23HRC is equivalent to 275HV and not to 255HV as indicated by ASTM E140. Recognition of this conversion will facilitate achievement of welded joints in low carbon 13%Cr steels meeting the intent of MR0175, without the necessity for prolonged heat treatment at low temperature, and utilisation of very low carbon contents, both of which can lead to tensile properties being below specification minima after the heat treatment is completed.

Extra Low Carbon Alloys

Considerable development has taken place to produce 13%Cr steels potentially suitable for linepipe and similar applications in the as-welded condition. ^[4,5] The materials show the common feature that carbon is reduced to very low levels, typically around 0.01%. To date, materials marketed may be divided compositionally into two categories. First, relatively simple alloys have been produced with low total alloy content to reduce material cost. Second, materials are marketed with increased nickel and molybdenum contents to obtain improved mechanical properties and corrosion resistance ( Table 1).

Table 1. Typical compositions of low carbon martensitic stainless steels: major elements

Element, wt%
C	Cr	Ni	Mo
0.01	11	1.5	-
0.04	13	4	1
0.01	12	6	2.5

When fusion welded, low carbon plain 13%Cr steels will transform to ferrite at temperatures approaching the solidus in the high temperature HAZ, and the resultant coarse microstructure will be retained on cooling to normal ambient temperature, leading to reduction in toughness and intercrystalline corrosion resistance ^[1] . The tendency to develop a fully or partially ferritic HAZ depends primarily on the material composition, with welding conditions and attendant cooling rates having much less effect over the range of practical application. Historically, the propensity for a steel to form ferrite has been described by means of the Kaltenhauser ferrite factor ^[6] , but it is unlikely that this is particularly applicable for the wide range of chromium, nickel and molybdenum contents now offered in commercial alloys. Work ^[7] has enabled a tentative HAZ structure prediction diagram to be produced using compositional factors after Irvine et al ^[8] , as shown in Fig.2. Caution is necessary in use of this diagram since the available database is fairly limited, but it serves to illustrate the wide range of HAZ microstructures that can be obtained for variations in composition around a simple 13%Cr base.

At present, the environmental limits of application of the new 13%Cr steels have not been fully explored. The cheaper, more ferritic alloys have found limited use for mild media, and have been considered for sour service in view of their low HAZ hardenability, but attention has tended to concentrate on the more hardenable variants. Studies have shown that the mechanical properties of welded joints in these steels in the as-welded condition are very attractive ^[4,5,9] . However, the inevitable HAZ hardening means that a PWHT cycle is required to achieve tempering and meet NACE MR0175 hardness requirements. As with F6NM, it can be difficult to achieve the necessary softening, given the conflicting requirements between increasing the temperature to obtain rapid tempering yet avoiding reaustenisation. Some success has been demonstrated using short term thermal cycles at around 650°C, as might be applied for laybarge applications, but time and peak temperature must be carefully controlled, as illustrated by Fig.3. Further, the high alloy variants display appreciable solid solution hardening and resistance to softening on short term PWHT ^[10] , so that compliance with a 23HRC maximum is extremely problematic ( Table 2). At the same time, the steels may have adequate SSC resistance for satisfactory service to be obtained at above this hardness level; appropriate testing is necessary to clarify the situation, and this will entail careful consideration of the test environment to be employed ^[11] .

Table 2. Hardness reduction of welded low carbon martensitic stainless steels after short term PWHT at 650°C/5 min ^[10]

Steel Type	HAZ area	Reduction of maximum hardness, HV5	Maximum hardness after PWHT, HV5
12Cr/6Ni/2Mo	Root*	0	317
	Cap*	42	303
11Cr/1.6Ni	Root**	86	227
	Cap**	76	237
*GTA  **SMA

Notwithstanding the excellent mechanical properties that have been achieved, it must be appreciated that many envisaged application areas for low carbon 13%Cr steels will entail placing in service an as-welded martensitic HAZ structure. This has not been common practice for martensitic stainless steels in the past, PWHT being normal to obtain weld area tempering and softening. There is the possibility that HAZs may be sensitised to intercrystalline attack on welding by precipitation of chromium-rich carbides on the prior austenite grain boundaries, with no healing by chromium diffusion during subsequent PWHT. Intergranular cracking of as-welded joints has been observed ^[10] in laboratory tests under sweet conditions, at least with lower alloy variants ( Fig.4). This potential problem is in fact recognised in industries other than oil and gas, from which experience suggests that practical failures arise only under specific environmental conditions, probably corresponding to a redox potential close to the matrix active/passive transition so that only sensitised grain boundaries are corroded. Such sensitisation may further require an 'additive' effect of multiple weld thermal cycles, since failures have been manifest principally at multipass welds or at weld junctions where the material undergoes local repeated heating into the sensitising temperature range. It is not clear how far sensitisation will constitute a practical hazard with as-welded low carbon martensitic 13%Cr steels in the oil and gas sector, and further work in this area is necessary.

Study is required also on the susceptibility of the transformed HAZ structure at a weld to hydrogen-induced environmental cracking. To date, the assumption has been made that compliance with NACE hardness limits is sufficient to ensure satisfactory service, but this must be confirmed in different media, with attention also to the possible adverse effect of hydrogen pick-up on external surfaces of subsea pipes from cathodic polarisation.

For the most part, the new low carbon 13%Cr steels have been welded with non-matching, ferritic-austenitic stainless steel consumables ^[4,5,9] , primarily to obtain good weld metal toughness although recent developments hold out the promise that nominally matching composition fillers will be available on a commercial basis ^[12] . The use of ferritic-austenitic consumables in principle has the further advantage of reducing the risk of weld area hydrogen cracking, since the relatively low diffusion rate of hydrogen in the ferrite-austenite matrix means that the risk of hydrogen diffusion and cracking in the heat affected zone is greatly reduced. At the same time, ferritic-austenitic consumables, both coated electrodes and solid wire, may offer a fairly high hydrogen potential over the molten pool, and HAZ cracking has been experienced in reeling applications where an appreciable strain is applied to the completed joint ^[13] . Evidently, notwithstanding the very low carbon contents and concomitant reduced hardening at welds, control of hydrogen cracking must be allowed for in design of welding procedures.

Austenitic Stainless Steels

Alloy Types

In general, austenitic stainless steels are among the easiest of steels to fusion weld, and they have been employed in welded form for a wide range of oil and gas applications with entirely satisfactory service being achieved. While specific problems continue to arise, such as avoidance of polythionic acid stress corrosion cracking ^[14] , the common austenitic stainless steels represent a thoroughly mature class of materials, and the guiding principles to be followed during welding are well defined.

There has, however, been a major development in austenitic stainless steels over the last 20 years. The particular benefits of increased molybdenum and nitrogen in enhancing resistance to chlorides have been recognised, and advances in steel production technology have been such that a wide range of highly alloyed austenitic stainless steels is now marketed. The first generation of such materials contained about 6% molybdenum and 0.2% nitrogen, but variants are now available with up to 7.5% molybdenum and 0.5% nitrogen ( Table 3).

Table 3. Typical compositions of high alloy austenitic stainless steels: major elements

UNS No.	Element, wt%
	Cr	Ni	Mo	N	Other
R20033	32	32	1.5	0.45	0.8Cu
S31254	20	19	6.2	0.2	0.8Cu
S34565	24	17	4.5	0.5	5.5Mn
N08367	21	24	6.5	0.22	-
S32654	24.5	22	7.5	0.5	3Mn

The 'superaustenitic' stainless steels have been used in the oil and gas context primarily for seawater systems, in which they offer excellent resistance to attack over a very wide range of flow rate. With chlorination to avoid fouling, the 6% molybdenum grades have been limited to about 30-40°C, but higher service temperatures are permitted with more highly alloyed materials.

Weld Metal Behaviour

When high alloy stainless steels are welded, the main corrosion concern is the fact that segregation of alloying elements inevitably takes place during solidification of the weld metal ^[15] . This is especially the case for molybdenum, and the initially solidified dendrite cores can be significantly depleted in this element relative to the inter-dendritic regions which are last to solidify. Chromium will segregate in a similar sense to molybdenum, although to a lesser degree, and the result is that the corrosion resistance of the solidified weld metal will be determined by the dendrite centres which are depleted in elements primarily responsible for corrosion resistance. In consequence, the corrosion resistance of nominally matching weld metal is inferior to that of the surrounding base steel ( Fig.5) ^[15,16] . The effect is most marked in high molybdenum grades, and, with low molybdenum alloys such as R20033, matching composition weld metal can give corrosion properties more akin to those of the parent material ^[17] .

Corrosion resistance of high alloy deposits can be further reduced by loss of nitrogen from the solidifying weld pool. The effect is most pronounced with inert gas shielded welding, such as gas tungsten arc (GTA) root runs in pipe for example, and is very much less with the use of coated electrodes. A counter-measure is the addition of low nitrogen contents to the shielding gas for GTA welding, and, with 6% molybdenum alloys, the resultant nitrogen pick-up by the molten pool can be such that the joint corrosion resistance approaches parent material performance ^[18] . At the same time, complete retention of weld metal nitrogen becomes more difficult as the absolute nitrogen content of the material increases. With GTA welding, a practical limit exists on the nitrogen content of the argon shielding gas to avoid erosion of the tungsten electrode and arc instability. This limit is generally considered to be below 5% nitrogen content of the gas shroud, although somewhat higher levels may be permissible with plasma arc welding. Studies by Woollin have shown that, even with 10% nitrogen in the gas shield, it may be difficult to achieve in weld metal the 0.5% nitrogen content associated with some commercial parent steels ^[17] ( Fig.6).

To avoid preferential weld metal corrosion with high alloy austenitic stainless steels, it has become normal practice to employ an overalloyed consumable ^[19] : segregation still occurs, but the minimum alloy content at any point in the deposit remains at least equivalent to the base steel. For this purpose, nickel-base consumables such as AWS ENiCrMo-3 or -4 are employed, largely because of their ready commercial availability, with the choice of filler depending on the specific grade of base steel being welded. Using this approach, weldments having a remarkably high pitting resistance can be achieved, with, for example, critical pitting temperature in an ASTM G48 type test being around 100°C for very highly alloyed materials ( Fig.5). It is, however, essential that sufficient filler be added to avoid excessive dilution and maintain a high weld metal alloy content. Hence, open gap joint preparations are strongly preferred, although this complicates production of root runs. Using overmatching fillers, the service behaviour of welds in superaustenitic steels for oil and gas purposes has been excellent. Pitting in seawater systems has been experienced in weld runs of excessive dilution, but only at very high temperatures and chlorination levels ^[20] .

Fusion Boundary and HAZ

Welding the superaustenitic steels, especially with nickel-alloy fillers that may have a melting point somewhat below that of the base metal, can lead to a small region of base metal at the fusion boundary that is melted but not incorporated into the bulk deposit. This unmixed zone (UMZ) is in effect an autogenous weld with inevitable segregation of molybdenum and, if present at the weld surface, will reduce corrosion resistance ( Fig.7). Debate exists regarding the conditions causing such an unmixed zone ^[21,22] , and its practical significance, but available evidence indicates that it is most marked with the GTA process, and less severe with flux processes. Nitrogen loss from the UMZ is probably negligible, so that, even given the fact of segregation, the UMZ does not have the same detrimental effect as autogenous welding. Nevertheless, its possible formation must be recognised, and, even with overmatching fillers, some reduction in weldment performance compared to base metal must be expected. The practical significance of any such loss must be investigated by undertaking appropriate corrosion tests on welded joints made reproducing the procedure to be used in practice.

Like other high alloy stainless steels, the superaustenitic grades are potentially sensitive to intermetallic formation at around 600-1000°C, and intermetallic precipitation has been reported in the HAZs of welded joints ^[23] . In principle, this can reduce corrosion resistance by an alloy element depletion mechanism analogous to the problem of 'weld decay' from carbide formation in common 300 series materials. However, intermetallic phases have not so far been manifest as a significant service problem, probably because of two reasons. First, precipitation rate is reduced by the high nitrogen contents in the steels ^[24] . Second, it is normal good practice to weld superaustenitic materials using low arc energy and interpass temperature ^[19] , so that the time spent on cooling in the critical temperature range is not unduly protracted. Precipitation rate may be increased in high molybdenum alloys, and steel manufacturer's recommendations on maximum arc energy must be followed, as illustrated by the intergranular HAZ attack in a high heat input weld shown in Fig.8.

Ferritic austenitic stainless steels

Alloy Types

The impact of ferritic-austenitic stainless steels on the oil and gas industry has been considerable. First developed for resistance to chloride stress corrosion cracking, their high strength and resistance to CO ₂ attack led to widespread application for linepipe to avoid the uncertainties associated with corrosion inhibition. This was followed by extensive use in process plant, especially offshore to reduce topside weight. The materials first employed contained about 22%Cr as exemplified by UNS S31803, but development to produce more highly alloyed 'superduplex' grades has resulted in remarkable corrosion resistance ^[25] and consequent application for seawater and other services ( Table 4).

Table 4. Typical compositions of ferritic-austenitic stainless steels: major elements

UNS No.	Element, wt%
	Cr	Ni	Mo	N	Other
S31500	18.5	5	2.5	0.1	-
S32304	23	4	-	0.1	-
S31803	22	5.5	2.8	0.13	-
S32205	22	5.5	3	0.17	-
S32550	25	6	3	0.18	1.8Cu
S32750	25	7	4	0.26	-
S32760	25	7	3.5	0.23	0.8W, 0.8Cu
S39274	25	7	3.2	0.30	2W, 0.5Cu

Transformation Behaviour

In contrast to the case with martensitic and austenitic steels, the microstructure developed at a fusion weld in duplex grades can differ very significantly from that in the base material ^[26] . The primary solidification phase is ferrite, and, given the fairly rapid cooling associated with a weld thermal cycle, diffusional transformation to austenite can be suppressed, leading to a predominantly ferritic structure in the fusion zone on cooling to room temperature. The HAZ is heated to temperatures close to the solidus, inducing transformation from the original two phase microstructure to ferrite, and again this is retained on cooling ( Fig.9).

The properties of ferritic-austenitic steels are highly dependent on the phase balance. Pitting resistance is normally considered to be optimum when the ferrite and austenite are present in roughly equal proportions ( Fig.10) ^[27] ; high weld area ferrite levels have induced pitting in offshore service, and can significantly reduce resistance to sulphide stress cracking ( Fig.11) ^[28] , and possibly also chloride induced stress corrosion ^[29] . High ferrite levels have also caused failure in high pressure hydrogen applications ^[30] . In the weld metal, suppression of a ferritic structure is readily achieved by increasing the nickel content relative to base steel, and this practice is the norm for commercial welding consumables. Considering the HAZ, it has in the past been necessary to specify a minimum heat input to ensure that cooling is sufficiently slow for adequate austenite reformation to take place ^[26] . However, development has led to production materials with increased nitrogen content relative to earlier alloys (circa 0.15-0.16% as compared to 0.12-0.13%) and this has greatly improved the tolerance of the materials to welding with low heat input, although the very rapid cooling associated with power beam welding (laser, electron beam) must still be regarded with caution. In consequence of these changes, satisfactory ferrite-austenite balance can be consistently achieved for a range of welding procedures.

Formation of intermetallic phases on welding poses a greater problem with ferritic-austenitic steels than with the superaustenitic alloys. Such transformation is most marked with the superduplex materials because of both the high level of alloying and the rapid substitutional element diffusion permitted by the ferrite phase. There is therefore no doubt that a maximum arc energy and interpass temperature must be specified when welding ferritic-austenitic steels to avoid unacceptably slow cooling and extended times in the intermetallic formation temperature range ^[26] . Intermetallic phases reduce toughness and corrosion resistance ( Fig.12) in a range of media, and may also enhance any tendency for SSC ( Fig.13). The risk of forming intermetallics during welding is now well recognised, and limits on arc energy and interpass temperature are given in typical manufacturers' literature. However, particularly with the superduplex grades, the material must be in the optimum treatment condition prior to welding. In production of forged components etc involving reheating, it is possible for intermetallic nuclei to exist in the base steel if quenching from solution treatment temperatures is not correctly carried out. The properties of the base steel may meet specifications such as ASTM A923, but imposition of a weld thermal cycle can lead to rapid growth of the intermetallic nuclei and considerable reduction in properties ^[31] . This can be manifest only during weld procedure qualification and, in such cases, it can be very difficult to limit total heat input to the weld and maintain a reasonable level of productivity.

Nitrogen Effects

As with the superaustenitic steels, nitrogen loss can occur from gas-shielded ferritic-austenitic weld pools ^[32] . However, even with the trend to increased base metal nitrogen contents, this can be negated by fairly low additions of nitrogen to have gas shroud, say ca 2% ( Fig.14). Alternatively, the 22%Cr duplex grades can be welded, at least for most runs, using superduplex consumables to achieve overalloying, and this measure is fairly commonly employed to achieve high deposit corrosion resistance.

Nitrogen in stainless steels is an extremely potent solution hardening element, particularly in cold worked material. During welding, the material in the weld area undergoes constrained expansion and contraction, leading to increased dislocation density, and the resultant hardening in nitrogen-containing steels can be sufficiently marked that it is difficult to meet hardness limits as specified in NACE MR0175. This has been recognised with superduplex materials for some time, especially in multipass welds made with low heat input and small weld beads to avoid intermetallic formation, but with resultant repeated thermal cycling ( Fig.15) ^[33] . The situation is complicated by the fact that there is no standardised conversion system between Vickers and Rockwell measurement for ferritic-austenitic steels and direct application of, for example, ASTM E140 has led to considerable problems in qualifying procedures without exceeding 275HV taken as equivalent to 28HRC. As for the 13Cr/4Ni martensitic alloys, use of a ferritic steel correlation is not correct. Fig.16 was developed ^[34] specifically for ferritic-austenitic steels, and this would indicate that, if 28HRC is to be maintained for superduplex grades, weld area hardness of up to about 310HV should be taken as acceptable. Following the increasing nitrogen contents in 22%Cr grades to facilitate achievement of a satisfactory phase balance, a similar problem of weld area hardening has emerged for these lower alloy materials. At the same time, it must be realised that hardness levels in MR0175 were derived originally from studies of deliberately cold worked materials in severe downhole applications. They should not necessarily be directly applied to welded joints, and certainly there seems to be no reported SSC failure of welded ferritic-austenitic steel associated with the hardening which inevitably takes place in the weld area.

Concluding Remarks

To some degree, it can be argued that the impetus for the new stainless steels has come mainly from the offshore sector to avoid CO ₂ attack and for seawater systems. However, the oil and gas industry is far from static, and the materials are finding increased application in all areas. Indeed, the extent to which new steels have been pursued to enhance cost effectiveness and reliability is remarkable, and this would not have been possible without clear recognition of the central role played by welding in fabrication. For the various grades of martensitic, superaustenitic, and ferritic-austenitic steels, the essential consequences of a welding operation are now fairly well understood, and it is clear that the different material characteristics must be recognised at the outset of fabrication, and appropriate procedural controls implemented at all stages. The superaustenitic and ferritic-austenitic steels have reached the more mature position in oil and gas service, and the weldment properties achievable have been reasonably well defined. At the same time, further study remains necessary to define the limits of use and enable welding to be carried out with maximum productivity, and this is especially the case for low carbon martensitic alloys for which experience is as yet more limited.

Acknowledgements

The author thanks colleagues at TWI for assistance in producing this paper.

References

Copyright NACE, 2000

1	Folkhard H et al,	'Welding Metallurgy of Stainless Steels,' pub by Springer-Verlag, Wien, New York, 1988.
2	Gooch T G,	Weld J, 74, 7, (1995) p213.
3	Hayes C and Patrick D H,	Metallography, 6, 2, (1983) p229.
4	Okazawa T et al,	Proc 12th Int. Corr Congress, Houston, NACE International, 1993, p2464.
5	Enerhaug J, Eliasson S L and Kvaale P E,	CORROSION '97, New Orleans, NACE International, 1997, paper 60.
6	Kaltenhauser R H,	Met Eng Quarterly, 11, 2, (1971) p41.
7	Gooch T G, Woollin P and Haynes A G,	Proc 'Supermartensitic Stainless Steels '99', Brussels, BWI, 1999, p188.
8	Irvine K J, Crowe D J and Pickering F B,	JISI, 195, (1960), p386.
9	Duthie S A,	Offshore Technology Conference, Houston 1998, paper OTC 8716.
10	Woollin P, Noble D N and Lian B,	EPRG/PRCI 12th Biennial Joint Technical Meeting on Pipeline Research, Groningen, The Netherlands, May 1999.
11	Drugli J M et al,	vide ref.7, p315.
12	Karlsson L et al,	vide ref. 7, p172.
13	Rørvik G, Kvaale P E and Akselsen O M,	vide ref. 7, p196.
14	Nagashima E, Matsumoto K and Shibata K,	CORROSION '98, San Antonio, 1998, NACE International, Paper 592.
15	Suutala and Kurkela,	Proc 'Stainless Steels '84', Gothenburg, 1984, The Metals Society, London, p240.
16	Bonnefois B et al,	34 ieme Journee du Cercle d'Etudes des Métaux, 'Nitrogen as alloying or strengthening element', Ecole National Superieure des Mines de Saint Etienne, France, 1995.
17	Woollin P,	Proc 'Stainless Steel World '99', The Hague, KCI Publishing, 1999.
18	Ginn B J and Gooch T G,	vide ref.4, p2895.
19	Rabensteiner G,	Welding in the World, 27, 1/2, 1989, p2.
20	Gunn R N and Plummer R W,	Proc 'Eighth Middle East Corrosion Conference', Bahrain, 1998, Bahrain Society of Engineers/NACE International, p157.
21	Gooch T G and Elbro A C,	Proc. 'Corrosion in Natural and Industrial Environments: Problems and Solutions', Grado, NACE Italia, 1995, paper 13.
22	Stenvall P, Liljas M and Wallén B,	CORROSION '96, Denver, NACE International, 1997, paper 419.
23	Liljas M, Holmberg B and Ulander, A,	vide ref.12, p323.
24	Charles J et al,	Proc. 'Corrosion and Materials Offshore', Oslo, NITO, 1994.
25	Charles J,	Proc. 'Duplex '94', Glasgow, TWI Abington UK, 1994, Keynote Paper I.
26	Gooch T G,	Proc. 'Duplex Stainless Steels '91', Beaune, Les Éditions de Physique, Paris, 1991, p325.
27	Ogawa T and Kuseki T,	Welding Journal, 68, 5, (1989), p1815.
28	Kudo T, Tsuge H and Morishi T,	Corrosion, 45, 10, (1989), p831.
29	Gooch T G,	Welding in the World, 24, 7/8, (1986), p148.
30	Singh A K, Bereczky E L and Bagdasarian A J,	vide ref.22, paper 16.
31	Gunn R N,	Proc. 'Duplex stainless steels '97', The Hague, KCI Publishing, 1997, p335.
32	Gunn R N and Anderson P C J,	vide ref.25, paper 30.
33	Baxter C F G, Irwin J and Francis R,	Proc Third ISOPE Conference, Singapore, 1993, Vol.IV, p401.
34	Gunn R N,	Unpublished work at TWI, 1992.