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

Subscribe >
Skip to content

Developments in fusion welding of stainless steels (January 1994)

P Woollin

(Originally published in Welding & Metal Fabrication, 1994, Vol. 62, No. 1, January, pp 18, 20, 22, 24 & 26
Publishers - DMG Business Media Ltd - )


Steels which contain a minimum of around 12% chromium are usually referred to as 'stainless steels' as a result of their resistance to the formation of visible corrosion products when left exposed to the atmosphere. The reason for their corrosion resistance is the formation of a tenacious surface film which protects the underlying steel. A wide range of stainless steels has been developed during the last 80 years to meet the demands of service in highly corrosive media, and these alloys also offer attractive mechanical properties at temperatures from absolute zero to above 800°C. Consequently, the term 'stainless steel' now applies to several different classes of material with entirely different metallurgical characteristics and chemical compositions, which are used in diverse applications, from decorative trim on consumer goods to conditions requiring excellent corrosion resistance or mechanical properties. These steels have been developed through the use of alloying additions and processing routes designed to impart specific chemical, physical and mechanical properties, through microstructural control. Stainless steels are now available with martensitic, austenitic, ferritic and duplex austenitic/ferritic microstructures. Each of these families has characteristic properties, designed to meet specific needs.

Welding Stainless Steels

When joining any material to be used under demanding conditions, it is important to ensure that the joint has adequate properties to perform successfully in service. This is particularly true for the more highly developed stainless steels whose entire material design has been formulated to give optimum properties. When a fusion welding process is employed, the original microstructure of the fused zone is destroyed and a new structure develops, in a similar fashion to that of a casting. The resulting room temperature structure depends upon the phases developed during solidification and the extent of subsequent solid state transformations.

In the heat affected zone (HAZ) adjacent to the fusion zone, solid state transformations may occur, the extent of these being controlled by the alloy content and the thermal cycle. A range of thermal cycles are experienced in the HAZ, depending on distance from the fusion boundary, and, in multipass welds, some material is subjected to several cycles with varying peak temperature.

In autogenous welds, ie ones with no filler addition, or ones employing matching consumables the composition of the weld metal is approximately fixed apart from some minor loss (eg Ti) or gain of elements (eg N2 ). However, when non-matching filler is added, the resulting weld metal composition depends upon the filler type and the level of dilution, which is controlled by the welding process and the variables employed.

Another important factor in the response of a material to welding is the residual stress, developed as a result of contraction during cooling, which contributes to the formation of several defect types. The magnitude of these stresses is controlled by the thermal expansion coefficient of the material and the joint configuration, although the stress distribution may be modified by control of the welding process.

The aim of this article is to present the general current situation regarding fusion welding of stainless steels. They are considered as discrete families, in terms of the major microstructural constituents of the parent material. However, this does not imply that all steels within a group show identical welding characteristics.

Martensitic Stainless Steels

Steels which contain an appropriate balance of austenite stabilising (eg Ni, C) and ferrite stabilising elements (eg Cr) have transformation characteristics similar to those of carbon steels, ie ferrite is stable at low temperatures, austenite is stable at high temperatures and martensite formation is possible. As a result of the alloy content of these steels, martensite forms under typical cooling conditions and ferrite only forms during slow cooling. These steels combine good corrosion resistance with excellent strength.

The martensitic stainless steels can be divided into two general types, which have different welding characteristics. The first group (eg 410, 420 and 440 types), have relatively high carbon levels which impart high hardness and low toughness to the HAZs and autogenous/matching weld deposits. The toughness of the weldment is usually improved by use of postweld heat treatment (PWHT) after allowing the weld to cool sufficiently for martensite to form. However, these steels are prone to cracking under the action of welding residual stress and the internal stress associated with transformation to martensite. Cracking is especially triggered by the presence of hydrogen in the martensite, and it is essential that hydrogen levels are minimised, if cracking is to be avoided reliably. Weld area cleanliness is essential and the use of a process with low hydrogen potential, eg the tungsten inert gas (TIG) process, is recommended where practical. When using the manual metal arc (MMA) process, the electrodes should be baked immediately prior to use, to drive off hydrogenous material. Significant reductions of hydrogen content often require baking at 350-450°C [1] ). If hydrogen is present in the weldment, there are ways in which its effects may be minimised. Pre-heating in the range 200-300°C is normally employed to reduce the cooling rate and allow time for hydrogen to diffuse out of the weldment. Holding the weldment at 200-300°C for some time after welding has a similar beneficial effect, although it is essential that the weld area cools below the martensite transformation temperature before hydrogen release can occur. When joint strength is not required to match that of the parent material, austenitic consumables may be used as these have high hydrogen solubility and tend to reduce the hydrogen content of the HAZ.

The second class of martensitic stainless steels (eg 13%Cr, 4%Ni steel) have low carbon contents which limit hardness and reduce the risk of hydrogen cracking. These steels can often be welded successfully with little or no pre-heat and have excellent strength and toughness after tempering, in some cases as a result of deliberate alloy additions to promote precipitation hardening (eg 17-4PH). However, the tempering characteristics of these materials are not so straightforward as those of the higher carbon martensitics. For example, 13%Cr, 4%Ni martensitic steel is difficult to soften effectively through PWHT as re-transformation to austenite occurs at relatively low temperatures. Therefore, tempering must be performed for long periods at a temperature at which the kinetics of transformation are slow [2] , while two or more PWHT cycles may be needed to achieve maximum softening. With 13%Cr, 4%Ni and similar steels, the martensite start and finish temperatures may be low, to the extent that significant amounts of austenite may be retained at room temperature. Thus, if multistage PWHT is intended, the weldment should be cooled to below normal ambient, and possibly even sub-zero, to obtain maximum transformation to martensite prior to the final tempering treatment.

Austenitic Stainless Steels

The austenitic stainless steels were developed by the addition, to the Fe-Cr system, of alloying elements such as Ni and Mn, which expand the austenite phase field to room temperature. Numerous austenitic stainless grades are available, with compositions in the range 16-26%Cr, 8-30% Ni and various other alloying additions. The most common grades are the 300 series alloys, eg types 304 and 316. Austenite has a number of attractive features, including good ductility down to cryogenic temperatures and retention of useful strength to temperatures above 800°C.

The austenitic matrix shows relatively low alloy element diffusion rates and a high thermal expansion coefficient. Therefore, when compared with ferritic stainless steels, HAZ grain growth and solid state transformation tend to occur slowly, but welding shrinkage strains can be very high.

These austenitic steels may solidify directly to austenite and remain austenitic to room temperature, transform to austenite at high temperature after solidifying as ferrite, or show mixed solidification, depending upon composition. Those steels which show solid state transformation may retain some ferrite to room temperature when welded. The fraction of ferrite remaining in austenitic weld metals has been presented graphically as a function of composition on the Schaeffler diagram [3] and modifications to this which take into account a wider range of alloying elements [4-6] . Predictions can be made for a particular composition by categorising the alloying elements as 'ferrite stabilising' (eg Cr, Mo, Si and Nb) or 'austenite stabilising' (eg Ni, Mn, C and N) and calculating chromium and nickel 'equivalent' values which are then plotted on the diagram. Figure 1 shows the WRC diagram which was developed to cover modern stainless grades, including the duplex stainless steels [5,6] .


Fig.1. The WRC-1992 constitution diagram for stainless steel weldments [6]

These constitution diagrams were developed from MMA weld metals and are also appropriate to the other common arc welding processes. However, the microstructure developed may be sensitive to cooling rate, which controls the extent of solid state transformation. Consequently, laser and electron beam welds do not necessarily conform to the predictions of these diagrams [7,8] . Steels which normally solidify as austenite are unaffected by cooling rate but those which solidify as ferrite may show either higher or lower than predicted residual ferrite levels, depending on the composition and cooling rate.

The austenitic steels show no other major phase change during cooling to room temperature and, therefore, normally no pre-heat is required when welding them. However, some precipitation may occur during cooling, eg the formation of chromium carbides on HAZ grain boundaries, leading to depletion of Cr and reduced corrosion resistance adjacent to the boundaries. Material may become 'sensitised' to intergranular attack (also known as 'weld decay') if sufficient carbide is formed. The risk of weld decay has reduced significantly in recent years through the development of low carbon grades (eg 304L, 316L) which contain ≤0.03%C, and 'stabilised' grades (eg 321, 347) containing either Ti or Nb, to form carbides preferentially to Cr. In both types of steel, the amount of free carbon in solution is sufficiently low to ensure that Cr carbide formation is minimal and therefore that sensitisation is not usually of practical significance during welding. Carbon levels in the standard austenitic grades have also been reduced in recent years, usually to levels of ≤0.05%, reducing the tendency for sensitisation so that this is very much less of a practical problem than was the case in the past.

Another form of precipitation which may be encountered is the development of intermetallic phases, eg sigma. This is not normally a problem in the common austenitic stainless steels, as a consequence of the relatively low diffusion rates, but can occur in the higher alloy grades, such as the 6% Mo superaustenitic steels, leading to some loss of ductility and toughness and reduced corrosion resistance. Therefore, restriction of arc energy, to prevent excessively slow cooling and minimise intermetallic formation, is recommended for high alloy grades [9] . Intermetallic formation occurs much more rapidly in ferrite than austenite, consequently excessively slow cooling and PWHT in the sigma formation temperature range should be avoided as far as possible for austenitic welds containing some ferrite.

During weld metal solidification, some segregation inevitably occurs. In materials which solidify as austenite, Si and Nb and impurity elements (eg P and S) segregate into the interdendritic regions, which therefore solidify at lower temperatures than the bulk of the steel. This may lead to the formation of interdendritic solidification cracks as a result of contraction during cooling [10] . This form of cracking can be controlled by restricting travel speed, thereby altering the solidification structure and promoting a more favourable segregation pattern. A related form of cracking may occur in reheated weld deposits, in which intergranular tears or liquation cracks may form under the action of residual stresses.

The most common means of avoiding solidification and liquation cracking is to choose a filler which gives some residual ferrite in the weld metal. This has proved extremely effective, although the mechanism by which the effect is achieved is still the subject of debate. Solidification to ferrite, rather than austenite, is thought to be of importance, as is the ability of ferrite to retain the elements which contribute to the cracking process. Consequently, nominally austenitic consumables are normally designed to yield weld metal containing some ferrite, unless residual ferrite would impair the ability of the component to perform in service, eg as a result of its magnetic properties, poor low temperature toughness or susceptibility to embrittlement at high temperatures.

A further aspect of segregation is of importance, particularly in the high alloy austenitic grades, such as the superaustenitic steels. These steels solidify as austenite and contain large amounts of Cr, Mo and N to impart high resistance to pitting in chloride environments. However, extensive segregation occurs on solidification, with Cr and especially Mo being rejected into the interdendritic regions, whilst N is retained in the dendrite cores [11] . Little of the inhomogeneity is removed by solid state diffusion under typical welding conditions. This segregation reduces the corrosion resistance of the interdendritic regions; consequently, if these materials are welded autogenously or with matching consumables, the weld metal has lower corrosion resistance than the parent material. This effect depends upon the bulk composition of the steel and is of minor importance in the common 300 series grades which have low alloy contents. Figure 2 shows corrosion of dendrite cores in autogenously welded 6% Mo steel. The common solution to the problem of segregation in such material is to weld with an overalloyed nickel-based filler. However, from work at The Welding Institute, the use of low arc energies, say ≤0.5kJ/mm, and nitrogen containing shielding gas, can give autogenous welds with corrosion resistance similar to that of welds made with overalloyed consumables [12] . Furthermore, the use of laser surface treatment has been demonstrated as a means of reducing compositional variation and increasing the corrosion resistance of autogenous welds to levels comparable with parent material [13,14] .


Fig.2. Corrosion of dendrite cores in a 6%Mo superaustenitic stainless steel autogenous weld deposit after exposure to ferric chloride solution

Ferritic Stainless Steels

By increasing the proportion of ferrite stabilising elements (eg Cr, Al, Ti) and limiting those which stabilise austenite (ie Ni, C), stainless steels have been developed which are essentially ferritic at all temperatures between the melting point and room temperature. The common ferritic grades either have Cr contents of ≥15% or have around 13%Cr with low carbon content and small additions of strong ferrite stabilising elements, such as Al or Ti. These steels can have a price advantage over austenitic materials, as they contain little or no nickel, and they are also immune to chloride stress corrosion cracking (SCC), which affects austenitic steels. Grades containing up to 30%Cr and 4%Mo have been developed for use in aggressive environments where the resistance to SCC may be exploited.

However, the ferrite structure has poor low temperature toughness and poor high temperature strength when compared to austenite. In fact, welds in the ferritic stainless steels show a ductile to brittle transition which may be around room temperature depending on the grain size and section thickness. They also have higher alloy element diffusion coefficients and lower thermal expansion coefficients than austenitic steels. Hence, grain growth and solid state transformations tend to occur relatively rapidly, homogenisation of segregation may occur during a typical weld thermal cycle and residual stresses are lower than in austenitic steels.

The low cleavage resistance is primarily a result of the coarse structure developed, which tends to increase the toughness transition temperature, although some martensite may form, leading to further loss of toughness. This usually restricts successful welding to section thicknesses of just a few millimetres. The toughness of ferritic stainless steel weldments cannot be recovered, as a consequence of the lack of a high temperature phase change, through which structural refinement could be achieved. Some small improvement in toughness may be brought about by tempering any martensite, at around 800-850°C, but this has a relatively minor effect. If an austenitic filler is used, brittleness may be restricted to the HAZs. Fully ferritic steels with improved toughness in the as-welded condition have been developed, through reduction of the interstitial impurity content to very low levels and through the addition of small amounts of nickel, although this may reduce the SCC resistance.

In some predominantly ferritic steels, a small amount of austenite forms at high temperatures and may transform to martensite on cooling. This phenomenon has been exploited in recent years to develop 12%Cr low carbon ferritic/martensitic steels potentially with better weldability than either ferritic or martensitic steels. This has been achieved through careful compositional design leading to the formation of a controlled amount of austenite (and hence martensite) in the HAZ, which can restrict grain growth ( Figure 3). The hardness and detrimental effect on toughness of the martensite is limited by the low carbon levels. However, achieving consistently good HAZ toughness has proved complex, with a number of steel companies adopting slightly different approaches to alloy design. Work at The Welding Institute concluded that, in general, the ferrite grain size controlled the HAZ toughness and that increasing the martensite level had a slightly detrimental effect when martensite was the minor phase [15]. More recently, it was found that alloys designed to promote a fully martensitic HAZ gave very good toughness, as illustrated in Fig.4 [16].


Fig.3. The HAZ of a 12%Cr low carbon ferritic/martensitic steel [16]


Fig.4. The effect of martensite content on HAZ Charpy impact resistance of 12%Cr low carbon ferritic/martensitic steel [16]

Precipitation processes, including sigma and carbide formation, generally occur more rapidly in ferritic steels than in austenitic steels. Carbide, nitride and carbonitride precipitation may also occur in ferritic steels at C and N levels which would have no effect on austenitic steel. Such precipitation causes localised reduction in corrosion resistance, eg sensitisation to intergranular attack, especially in the coarse grain HAZs. This effect is generally most significant in the higher chromium grades, which are more likely to be exposed to aggressive environments.

Duplex Ferritic/Austenitic Stainless Steels

Ferritic/austenitic stainless steels contain approximately equal proportions of the two phases, as a result of the balance of alloying elements. At present, the most common grades are based on the approximate compositions 22%Cr, 5%Ni, 2-3%Mo, 0.1-0.2%N and 25%Cr, 7%Ni, 3-4%Mo, 0.2-0.3%N. The higher alloy grades, often referred to as 'superduplex' steels, may also contain other alloying additions to enhance corrosion resistance. The most attractive features of the duplex grades are their excellent corrosion properties, including resistance to chloride SCC, high strength, and good toughness. This combination of properties is dependent upon the approximately 50:50 phase balance. Therefore, the welding metallurgy of these materials is dominated by the need to maintain an appropriate phase balance.

The duplex steels solidify to ferrite and undergo solid state transformation to austenite on cooling. The parent material phase balance is usually achieved by solution annealing in the two phase region followed by quenching to room temperature. However, when welding, the cooling rate is relatively rapid through the two phase region and solid state transformation to austenite is limited. Consequently, autogenous welds and ones made with matching consumables tend to be highly ferritic and require annealing to achieve a desirable duplex microstructure. More commonly, consumables are used which are overalloyed in Ni to promote austenite formation and give a suitable phase balance as-welded [17] , as shown in Fig.5. Nonetheless, increased ferrite levels in the HAZ must be expected. Therefore, very low arc energies and associated rapid cooling should be avoided, to allow austenite development in the HAZ. However, a range of precipitates, including intermetallics (eg sigma phase) and nitrides, may form under excessively slow cooling conditions, particularly in the higher alloy grades ( Fig.6). These reactions may lead to loss of corrosion resistance and toughness in both the weld metal and HAZ [18] . Consequently, high arc energies and interpass temperatures are also undesirable. As a result, duplex steels should be welded within an arc energy 'window', appropriate to the alloy composition concerned.


Fig.5. A 22%Cr duplex stainless steel weld metal deposited using consumables overalloyed in Ni


Fig.6. A schematic TTT curve for precipitation in duplex stainless steels, illustrating the temperature ranges over which the various phases may form and the effects of increasing alloy content [18]


For material in the range 10-15mm thick, an arc energy range of 0.5-2.0kJ/mm and maximum interpass temperature of 225°C have been recommended for 22%Cr steel, whilst for superduplex steels figures of 0.5-1.2kJ/mm and 150°C have been suggested [19,20].


The stainless steels are a diverse range of materials, all with good corrosion resistance, but which can be divided into a number of quite different metallurgical categories. The weldability of these materials is controlled by their microstructural features, not their 'stainless' nature. Therefore, it is essential that the characteristics of each type are appreciated. Sound welds may be produced in each type of stainless steel provided that appropriate procedures are employed, although the austenitics are generally the easiest to weld.


1 Castro R J and de Cadenet J J: 'Welding Metallurgy of Stainless and Heat Resisting Steels', Cambridge University Press, 1975.
2 Fischer R B and Larson J A: Trans American Foundrymens Soc, 90, 1982, p103.
3 Schaeffler, A L: Met Prog, November 1949, 56, p680.
4 DeLong W B: Met Prog, February 1960, 77, p98.
5 McCowan C N, Siewert T A and Olson D L: WRC Bulletin 342, April 1989.
6 Kotecki D J and Siewert T A: Welding Journal, May 1992, 71(5), p171-s.
7 David S A, Vitek J M and Hebble T L: Welding Research Supplement, October 1987, p289-s.
8 David S A and Vitek J M: International Materials Reviews, 34(5), 1989, p213.
9 Gooch T G: Proc 8th Annual North American Welding Research Conf, Columbus, USA, 1992, session 2, paper 1.
10 Gooch T G: Proc Conf Weldability of Materials, ASM, Detroit, USA, 1990, p31.
11 Marshall P I and Gooch T G: Corrosion, 49(6), 1993, p514.
12 Ginn B J, and Gooch T G: Proc 12th International Corrosion Congress, NACE, Houston, USA, 1993.
13 Nakao Y, Nishimoto K and Zhang W-P: Proc 5th International Symposium of the JWS, 1990, p935.
14 Nakao Y and Nishimoto K: IIW Doc IX-1666-92, 1992.
15 Gooch T G and Ginn B J: Welding Journal, 69(11), 1990, p431-s.
16 Ginn B J: TWI unpublished work, 1992.
17 Fager S A: Proc Conf Duplex Stainless Steels '91, Beaune, France, 1991, p403.
18 Charles J: Proc Conf Duplex Stainless Steels '91, Beaune, France, 1991, p3.
19 Gooch T G: Proc Conf Duplex Stainless Steels '91, Beaune, France, 1991, p325.
20 Gunn R N: Proc 3rd International Offshore and Polar Engineering Conference, Singapore, June 1993.


For more information please email: