Welding Metallurgy

Welding Metallurgy

Joining of two metal elements by means of welding consists in the formation of a weld between these two elements. The weld is composed of molten, mixed and, eventually, solidified metal of the edges of welded elements as well as of filler metal (consumable). In principle, melting of metal during welding as well as its chemical and physical reacting with enclosing shield and slag followed by solidification and crystallization reflects melting of steel in a metallurgical furnace and its casting. However, being to some extent similar, welding also differs considerably differs from the afore-mentioned processes by being very fast and needing very limited volume of ingredients.

Welding is a complex process which involves gas-metal and slag-metal reactions, solidification, metallurgical reactions in the solid state, annealing and recovery, grain growth, precipitation, and phase transformation. These metallurgical phenomena control strength and ductility of the weld.

The space in which metallurgical processes take place comprises (i) fusing charge in the form of wire and coating or, alternatively, flux or core of powder-core electrode, (ii) liquid metal of the partially molten base, (iii) liquid and crystallizing part of weld, and (iv) the area between the electrode and the welded element filled with arc plasma, gases, and metal vapours. The distance between the end of fusing electrode and welded element amounts to several millimetres (mm).

The butt surface of the wire is an area of several reactions proceeding between melting metal and fusing coating. In the arc space, it is possible to observe drops of liquid metal moving from the end of fusing electrode towards the welded element. On the surface of the welded element, the aforesaid metal drops react with slag and gases filling the arc space. At the same time, alloying components of the coating enter the weld pool. Very high temperature and considerable difference in concentration of individual components as well as a significant area of the contact trigger very intense and dynamic reactions (far from equilibrium).

Welding metallurgy describes a microcosm (a situation regarded as condensing in miniature the characteristics of something much larger) of metallurgical processes occurring in and around a weld which influence the micro-structure, properties, and weldability of the material. Because of the rapid heating and cooling rates associated with majority of the welding processes, metallurgical reactions frequently occur under transient, non-equilibrium conditions.

The cooling rate and chemical composition affect the micro-structure of the welded joint. The mechanical properties of a welded joint depend on the micro-structure produced by welding. The micro-structure can vary from region to region in a weld. Micro-structure has a profound effect on the weld.

The metallurgical processes which affect the micro-structure are (i) melting and solidification, (ii) nucleation and growth, (iii) phase transformations, (iv) segregation and diffusion, (v) precipitation, (vi) recrystallization and grain growth, (vii) liquation mechanisms, (viii) embrittlement, and (ix) thermal expansion, contraction, and residual stress. The weld region consists of two zones namely (i) fusion zone and (ii) heat affected zone (HAZ). Fusion zone is further divided into two namely (i) composite zone, and (ii) unmixed zone. The HAZ is further divided into partially-melted zone (PMZ) and true HAZ. Fig 1 shows welding zones and evolution of properties of welded metal.

Fig 1 Welding zones and evolution of properties of welded metal

PMZ consists of (i) region separating the fusion zone from the true HAZ, (ii) transition from 100 % liquid at the fusion boundary to 100 % solid in the HAZ, (iii) localized melting normally observed at grain boundaries, and (iv) constitutional liquation of certain particles.

Grain boundary liquation in the PMZ results into segregation of solute / impurities to grain boundaries which depresses the local melting point. Also, the temperature gradient has a strong effect on the extent of melting.

The true HAZ zone is adjacent to the PMZ. In this zone, all the metallurgical reactions occur in the solid state. The zone is strongly dependent on weld thermal cycle and heat flow conditions. Heat input and heat flow have their effects. HAZ width is dictated by weld thermal conditions. The temperature gradient of the HAZ depends on heat input, and heat flow.

Metallurgical reactions in the true HAZ are solid-state metallurgical reactions. Solid state metallurgical reactions consist of recrystallization, grain growth, allotropic / phase transformations, dissolution / over-aging of precipitates, formation of precipitates, formation of residual stresses, and degradation of weldment properties which is frequently associated with the HAZ.

HAZ transformations in steels are functions of composition and cooling rate. Regions which form austenite during heating transform during cooling as ferrite, pearlite, bainite, martensite, or combinations of phases as per the CCT (continuous cooling transformation) diagrams. The nature of transformations is normally diffusion-controlled. In case of diffusion less, or shear-type transformation, it is either martensitic or massive.

Solid-state joining processes have no fusion zone, but can have a HAZ. These joining processes are friction welding, flash butt welding, diffusion welding, and explosion welding. In friction welding and flash butt welding, the base metal is heated until it is easily deformable, two ends of the joint are forged together, and hot base metal is extruded from the joint to form a flash.

Recrystallization has effect on the strength and ductility. Recrystallization promotes loss in strength and increase in ductility, while the grain growth promotes some additional softening.

Fusion zone is the region of the weld which is completely melted and resolidified. In this zone, the micro-structure is dependent on the composition and the solidification conditions and the local variations in the composition. It is distinct from the other regions of the weld. It can have three zones namely (i) composite zone, (ii) transition zone, and (iii) unmixed zone.

Unmixed zone consists of (i) narrow region adjacent to the fusion boundary, (ii) completely melted and resolidified base metal, and (iii) no mixing with the bulk fusion zone (composite region). Factors which influence the formation of unmixed zone are (i) base metal / filler metal composition, (ii) physical properties consisting of melting point, fluid viscosity, and miscibility, (iii) welding process, and (iv) process conditions consisting of heat input, and fluid flow. Unmixed zone is most prevalent in arc welding processes such as gas tungsten arc welding (GTAW), and gas metal arc welding (GMAW) etc. It is not observed in electron beam welding (EBW) and laser beam welding (LBW).

The micro-structure and the properties are considerably affected by the dilution of the fusion zone with the filler material. The quantity of the melted base metal is mixed with the filler material. It is expressed as percent base metal dilution of the filler metal. The dilution is 100 % is an autogenous weld, and 10 % to 40 % in arc welds. The dilution is controlled by the design of the joint, the process, and the parameters.

The weld pool shape depends on material properties, process parameters, and heat flow conditions. Material properties are melting point, thermal conductivity, surface tension, and Marangoni effect which is the mass transfer along an interface between two phases because of a gradient of the surface tension. Process parameters are heat-input, and travel speed. Heat flow conditions are full penetration in 2-dimension, and partial penetration in the 3-dimension.

Surface tension influences fluid flow. Surface tension of liquid is a function of composition and temperature and has Marangoni effect. It has influence of gradient on weld pool fluid flow. Negative gradient promotes outward flow and shallow penetration while the positive gradient promotes inward (downward) flow and good penetration. Sulphur and oxygen have strong influence on surface tension. Fig 2 shows surface tension as a function of temperature

Fig 2 Surface tension and temperature

Nucleation of solid during solidification can be homogeneous which is dependent on the critical radius size, and liquid undercooling, or can be heterogeneous which is dependent on the nucleation from existing substrate or particle and for which little or no undercooling is needed. Types of heterogeneous nucleation are dendrite fragmentation, grain detachment, nucleant particle formation, surface nucleation, and epitaxial nucleation. Epitaxial nucleation at the fusion boundary has (i) nucleation is from an existing solid substrate, (ii) crystallographic orientation of base metal ‘seed crystal’ is maintained, and (iii) growth is parallel to cube edge in cubic materials also called ‘easy growth’ directions. Fig 3 shows nucleation mechanisms.

Fig 3 Nucleation mechanisms

Multiple solidification modes (morphologies) are possible. The multiple modes are planar, cellular, cellular dendritic, columnar dendritic, and equiaxed dendritic. The solidification modes are controlled by temperature gradient in the liquid, solidification growth rate, and the composition.

Fig 4 Effect of parameters on solidification modes

Travel speed has considerable effect on the weld pool shape. Low travel speeds result into elliptical pool shape, curved columnar grains, gradual change in the growth rate, and composition. High travel speeds result into tear-drop pool shape, distinct centre-line, constant composition along majority of the solid-liquid interface.

In case of weld metal epitaxial nucleation, nucleation is from and existing solid substrate at the fusion boundary, crystallographic orientation of HAZ grain is maintained, and there are ‘easy growth’ directions parallel to cube edge in cubic materials.

After the nucleation, there is competitive growth. There is random orientation of base metal grains in polycrystalline materials. Growth is most favourable when easy growth direction is parallel to heat flow direction. Grains ‘compete’ depending on the orientation and also intersection of grains forms solidification grain boundaries.

Fusion zone boundaries (Fig 5) are differentiated by the composition and the structure. There are three cases. The first is the solidification sub-grain boundaries which are because of composition, and low angle mis-orientation. Boundaries between cells and dendrites (solidification sub-grains) composition is dictated by the solute re-distribution.  There is low mis-orientation between adjacent sub-grains i.e., low angle boundary.

Fig 5 Fusion grain boundaries

The second is solidification grain boundaries which are because of composition, and high or low angle mis-orientation. Boundary between packets of sub-grains results from competitive growth. The composition is dictated by solute redistribution. There is large mis-orientation across boundary at the end of solidification, i.e., high angle boundary. It is the most likely site for solidification cracking.

The third are migrated grain boundaries which are because of local variation in composition, and high angle mis-orientation. These are crystallographic component of the solidification grain boundaries. The grain boundaries migrate away from the solidification grain boundaries in the solid-state following solidification or during reheating. There is large mis-orientation across boundary i.e., high angle boundary. The composition varies locally. There is possible boundary ‘sweeping’ and segregation. There are liquation and ductility dip cracking.

The main processes which are taking place during welding are (i) oxidation and reduction, (ii) desulphurization and dephosphorization of liquid metal, (iii) supplying of alloying components to weld dissolution, and (iv) liberation of gases in liquid metal, crystallization, release of non-metallic inclusions, and formation of cracks.

Welding is connected with oxidation and reduction reactions resulting in impoverishment or enrichment of liquid metal in elements revealing variable mutual affinity with oxygen, depending on momentary conditions.

Heat flow during welding can strongly affect phase transformations during welding and hence the resultant micro-structure and properties of the weld. It is also responsible for weld residual stresses and distortion. The ability of the heat source to melt the base metal (as well as the filler metal) is of practical interest to the welder.

The melting efficiency cannot be increased indefinitely by increasing the welding speed without increasing the power input. For doing so, the power input is to be increased along with the welding speed. In the presence of a surface-active agent such as sulphur in steel, the weld pool can become much deeper even though the welding parameters and physical properties remain unchanged.

As both the heat input and welding speed increase, the weld pool becomes more elongated, shifting from elliptical to tear-drop shaped. The effect of the welding parameters on the pool shape is very significant. The low thermal conductivity in case of certain steel grades makes it more difficult for the weld pool to dissipate heat and solidify.

The cooling rate of the liquid pool is reduced considerably by pre-heating. Pre-heating is a normal practice in welding high-strength steels, since it reduces the risk of HAZ cracking. In multiple-pass welding, the inter-pass temperature is equivalent to the pre-heat temperature in single-pass welding.

Basic chemical reactions during fusion welding include gas-metal reactions and slag–metal reactions. These chemical reactions on the weld metal affect composition and mechanical properties. Nitrogen (N2), oxygen (O2), and hydrogen (H2) gases can dissolve in the weld metal during welding. These elements normally come from air, the consumables such as the shielding gas and flux, or the work-piece such as the moist or dirt on its surface. N2, O2, and H2 can considerably affect the soundness of the resultant weld.

The gas-metal reactions refer to chemical reactions which take place at the interface between the gas phase and the liquid metal. They include the dissolution of N2, O2, and H2 in liquid metal and the evolution of carbon monoxide (CO). In arc welding, a portion of the N2 molecules can dissociate (or even ionize) under the high temperature of the arc plasma. The atomic ‘N’ so produced can dissolve in the molten metal. As in the case of N2, a portion of the O2 and H2 molecules can dissociate (or even ionize) under the high temperature of the arc plasma. The atomic ‘O’ and ‘H’ so produced can dissolve in the molten metal.

The dissolution of mono-atomic, rather than di-atomic, N2 and H2 dominates in molten iron. The species concentration in the weld metal can be considerably higher than those calculated from dissolution of di-atomic molecules. Dissociation of such molecules to neutral atoms and ions in the arc leads to improved dissolution in the molten metal.

The majority of H2 absorption appears to take place around the outer edge of the weld pool, and monatomic H2 absorption dominates the contribution to the H2 content. This contradicts predictions based on Sievert’s law that the maximum absorption occurs near the centre of the pool surface where the temperature is highest. However, the dissolution process alone does not determine the H2 content in the resultant weld metal. Rejection of the dissolved H2 atoms by the solidification front and diffusion of the H2 atoms from the weld pool are also be considered. It is interesting to note that a considerable quantity of H2 absorbed by the liquid metal during welding leaves the weld metal immediately after the extinction of the arc.

For metals which neither dissolve nor react with N2, such as copper (Cu) and nickel (Ni), N2 can be used as the shielding gas during welding. On the other hand, for metals which either dissolve N2 or form nitrides (or both), such as iron (Fe), titanium (Ti), manganese (Mn), and chromium (Cr), the protection of the weld metal from N2 is to be considered.

The presence of N2 in the welding zone is normally a result of improper protection against air. However, N2 is sometimes added purposely to the inert shielding gas. Increasing the weld metal N2 content can decrease the ferrite content and increase the risk of solidification cracking.

The presence of N2 in the weld metal can considerably affect its mechanical properties. Iron nitride (Fe4N) has a needle like structure in a ferrite matrix. The sharp ends of such a brittle nitride act as ideal sites for crack initiation. The ductility and the impact toughness of the weld metal decrease with increasing N2 in the weld metal. N2 can decrease the ductility of Ti welds.

In the self-shielded arc welding process, strong nitride formers, such as Ti, aluminum (Al), silicon (Si), and zirconium (Zr), are frequently added to the filler wire. The nitrides formed enter the slag and N2 in the weld metal is hence reduced. However, the N2 contents of self-shielded arc welds can still be rather high, and other arc welding processes such as GTAW, GMAW, or SAW (submerged arc welding) are to be used if weld N2 contamination is to be minimized.

Oxygen in the weld metal can come from the air, the use of excess O2 in oxy-fuel welding, and the use of O2-containing or CO2-containing shielding gases. It can also come from the decomposition of oxides, especially SiO2 (silica), MnO (manganese oxide), and FeO (ferrous oxide) in the flux and from the slag-metal reactions in the weld pool.

In GMAW of steels the addition of O2 or carbon dioxide (CO2) to argon (Ar), e.g., Ar-2 % O2, helps stabilize the arc, reduce spatter, and prevent the filler metal from drawing away from (or not flowing out to) the fusion line. CO2 is widely used as a shielding gas in FCAW (flux cored arc welding), the advantages being low cost, high welding speed, and good weld penetration.  CO2 can decompose under the high temperature of the welding as given by the equations (i) CO2 (g) = CO (g) + 1/2 O2(g), and (ii) CO = C (s) + 1/2 O2 (g).

Oxygen can oxidize the carbon (C) and other alloying elements in the liquid metal, modifying their prevailing role, depressing hardenability, and producing inclusions. The oxidation of C is represented by the equation C + O = CO (g).

The oxidation of other alloying elements in slag-metal reactions, forms oxides which either go into the slag or remain in the liquid metal and become inclusion particles in the resultant weld metal. The gas composition in oxy-acetylene welding of mild steel has effect on the weld metal composition and properties. When too much O2 is used, the weld metal has a high O2 level but low C level. On the other hand, when too much acetylene (C2H2) is used, the weld metal has a low O2 level but high C level (the flame becomes carburizing). In either case, the weld mechanical properties are poor. When the O2-C2H2 ratio is close to 1, both the impact toughness and strength (proportional to hardness) are reasonably good.

If oxidation results in excessive inclusion formation in the weld metal or considerable loss of alloying elements to the slag, the mechanical properties of the weld metal can deteriorate. The strength, toughness, and ductility of mild steel welds can all decrease with increasing O2 contamination. In some cases, however, fine inclusion particles can act as nucleation sites for acicular ferrite to form and improve weld-metal toughness.

The presence of H2 during the welding of high-strength steels can cause H2 cracking. H2 in the welding zone can come from several different sources such as the combustion products in oxy-fuel welding, decomposition products of cellulose-type electrode coverings in shielded metal arc welding (SMAW), moisture or grease on the surface of the work-piece or electrode, and moisture in the flux, electrode coverings, or shielding gas.

In SMAW, high-cellulose electrodes contain much cellulose, (C6H10O5)x, in the electrode covering. The covering decomposes upon heating during welding and produces a gaseous shield rich in H2, for example, 41 % H2, 40 % CO, 16 % H2O, and 3 % CO2 in the case of E6010 electrodes (an all position, cellulosic electrode which has a quick-starting, steady, and deep penetrating arc). On the other hand, low-H2 electrodes contain much CaCO3 (calcium carbonate) in the electrode covering. The covering decomposes during welding and produces a gaseous shield low in H2, for example, 77 % CO, 19 % CO2, 2 % H2, and 2 % H2O in the case of E6015 electrodes. As such, to reduce weld metal H2, low-H2 electrodes are required to be used.

The weld H2 content can be reduced in several ways. First is the avoiding of H2-containing shielding gases, including the use of hydro-carbon fuel gases, cellulose-type electrode coverings, and H2-containing inert gases. Second is the drying of the electrode covering and flux to remove moisture and clean the filler wire and work-piece to remove grease. Third is the adjusting of the composition of the consumables if feasible. Increase of the CaF2 (calcium fluoride) content in the electrode covering or the flux has been reported to reduce the weld H2 content. This reduction in H2 has been ascribed to the reaction between H2 and CaF2. Fourth is the using of the post-weld heating to help H2 diffuse out of the weld.

The thermo-chemical slag-metal reactions refer to thermo-chemical reactions which take place at the interface between the molten slag and the liquid metal. Examples of such reactions are decomposition of metal oxides in the flux, oxidation of alloying elements in the liquid metal by the O2 dissolved in the liquid metal, and desulphurization of the weld metal.

In the high-temperature environment near the welding plasma, all oxides are susceptible to decomposition and produce O2. It has been found that the stability of metal oxides during welding decreases in the order of (i) CaO (calcium oxide), (ii) K2O (potassium oxide), (iii) Na2O (sodium oxide) and TiO2 (titanium oxide), (iv) Al2O3 (alumina), (v) MgO (magnesia), and (vi) SiO2 and MnO. FeO is not included but can also be expected to be rather unstable.

In fluxes of low FeO content (less than 10 % FeO), SiO2 and MnO are the primary sources of O2 contamination and the stability of metal oxides in welding is not directly related to their thermo-dynamic stability. CaF2 reduces the oxidizing potential of welding fluxes because of the dilution of the reactive oxides by CaF2 rather than to reactivity of the CaF2 itself and considerable losses of Mn can occur by evaporation from the weld pool because of the high vapour pressure of Mn.

The FeO additions, at the expense of MnO, increase the extent of O2 transfer to the weld metal. This is since, FeO is less stable than MnO and hence decomposes and produces O2 in the arc more easily than MnO. The CaO additions at the expense of MnO decrease the extent of O2 transfer to the weld metal since CaO is more stable than MnO. The CaF2 additions at the expense of MnO also decrease the extent of O2 transfer to the weld metal but more significantly.

Hence, the flux composition can affect the weld metal composition and hence mechanical properties rather considerably. The loss of alloying elements can be made up by the addition of ferro-alloy powder (e.g., Fe-50 % Si, and Fe-80 % Mn) to SAW fluxes or SMAW electrode coverings. In doing so, the alloying element recovery, i.e., the percentage of the element transferred across the arc and into the weld metal, is to be considered. The recovery varies considerably from element to element. In SMAW, for example, it can be around 100 % for Ni and Cr, 75 % for Mn, 70 % for Nb, 45 % for Si, and 5 % for Ti.

Excessive weld metal O2 and hence oxide inclusions can deteriorate weld metal mechanical properties.

The driving forces for fluid flow in the weld pool include the buoyancy force, the Lorentz force, the shear stress induced by the surface tension gradient at the weld pool surface, and the shear stress acting on the pool surface by the arc plasma. The arc pressure is another force acting on the pool surface, but its effect on fluid flow is small, especially below 200 A, which is normally the case for GTAW.

Because of the intense heating of pool surface, evaporation from the weld pool can be considerable with some alloying elements. Mn has a much higher vapour pressure than Fe, which explains Mn losses from laser welds of stainless steels. Evaporation can also occur as metal droplets transfer from the filler wire to the weld pool through the arc, considering the very high temperature of the arc.

The use of fluxes in GTAW has been found to dramatically increase weld penetration in steels and stainless steels. The flux normally consists of oxides and halides, and it is mixed with acetone or the like to form a paste and painted as a thin coating over the area to be welded.

Residual stresses are stresses which exist in a body if all external loads were removed. They are sometimes called internal stresses. Residual stresses which exist in a body which has previously been subjected to non-uniform temperature changes, such as those during welding, are frequently called thermal stresses. The development of residual stresses can be explained by considering heating and cooling under constraint.

The expansion and contraction of the weld metal and the adjacent base metal are restrained by the areas farther away from the weld metal. Hence, after cooling to the room temperature, residual tensile stresses exist in the weld metal and the adjacent base metal, while residual compressive stresses exist in the areas farther away from the weld metal.

Because of solidification shrinkage and thermal contraction of the weld metal during welding, the work-piece has a tendency to distort. There are several types of weld distortions. The welded work-piece can shrink in the transverse direction. It can also shrink in the longitudinal direction along the weld. Upward angular distortion normally occurs when the weld is made from the top of the work-piece alone. The weld tends to be wider at the top than at the bottom, causing more solidification shrinkage and thermal contraction at the top of the weld than at the bottom. Hence, the resultant angular distortion is upward.

Several techniques can be used for reducing weld distortion. Reducing the volume of the weld metal can reduce the quantity of angular distortion and lateral shrinkage. The use of EBW or LBW can minimize angular distortion. Balancing welding by using a double-V joint in preference to a single-V joint can help reduce the angular distortion. Placing welds about the neutral axis also helps reduce distortion. There are three other techniques for reducing weld distortion. These are (i) pre-setting, (ii) elastic pre-springing, and (iii) pre-heating, thermal management, and post-weld heating. Pre-setting is achieved by estimating the quantity of distortion likely to occur during welding and then assembling the job with members preset to compensate for the distortion. Elastic pre-springing can reduce angular changes after the removal of the restraint. Pre-heating, thermal management during welding, and post-weld heating can also reduce angular distortion.

Failure can occur in welds under repeated loading. This type of failure, called fatigue, has three phases namely crack initiation, crack propagation, and fracture. A simple fatigue stress cycling can result in the formation of intrusions and extrusions at the surface of a material along the slip planes. A discontinuity-point in the material (e.g., inclusions, porosity) can serve as the source for a slip to initiate. A series of intrusions and extrusions at the free surfaces because of the alternating placement of metal along slip planes eventually become severe enough and initial cracks form along slip planes. The direction of crack propagation is along the slip plane at the beginning and then becomes macroscopically normal to the maximum tensile stress. There are several factors which affect the fatigue behaviour, such as material properties, joint configuration, stress ratio, welding procedure, post-weld treatment, loading condition, residual stresses, and weld reinforcement geometry.

In developing any fatigue behaviour criteria for welding, the severity of joint geometry is probably the most critical factor. The more severe is the geometry, the lower is the fatigue strength. The severity level of the longitudinal butt weld is lowest since both the weld and the base metal carry the load. The severity level of the cross shaped, on the other hand, is highest since the welds alone carry the load and the parts are joined perpendicular to each other.

It is well known that stress raisers tend to reduce fatigue life, namely, the so called, notch effect. Stress raisers can be mechanical, such as toes with a high reinforcement, lack of penetration, and deep undercuts. They can also be metallurgical, such as micro-fissures (micro-cracks), porosity, inclusions, and brittle and sharp intermetallic compounds.

A corrosive environment (such as salt water) can frequently reduce fatigue life. This is called corrosion fatigue. It has been reported that the damage can be almost always higher than the sum of the damage by corrosion and fatigue acting separately.

Some basic solidification concepts include solute redistribution, solidification modes, constitutional super-cooling, micro-segregation and banding, the dendrite-arm or cell spacing, and the solidification path. When a liquid of uniform composition solidifies, the resultant solid is seldom uniform in composition. The solute atoms in the liquid are redistributed during solidification. The redistribution of the solute depends on both thermodynamics, i.e., the phase diagram, and kinetics, i.e., diffusion, undercooling, fluid flow, and so on.

The development of the grain structure in the fusion zone depends on the welding parameters. There are several mechanisms and techniques for grain refining. The grain structure of the fusion zone can considerably affect its susceptibility to solidification cracking during welding and its mechanical properties after welding.

In fusion welding, the existing base-metal grains at the fusion line act as the substrate for nucleation. Since the liquid metal of the weld pool is in intimate contact with these substrate grains and wets them completely, crystals nucleate from the liquid metal upon the substrate grains without difficulties. When welding without a filler metal (autogenous welding), nucleation occurs by arranging atoms from the liquid metal upon the substrate grains without altering their existing crystallographic orientations.

Epitaxial growth during fusion welding can also occur when the work-piece is a material of more than one phase. In epitaxial growth, all dendrites grow from each grain point in one direction, and this direction varies from one grain to another.

When welding with a filler metal (or joining two different materials), the weld metal composition is different from the base metal composition, and the weld metal crystal structure can differ from the base metal crystal structure. When this occurs, epitaxial growth is no longer possible and new grains are to nucleate at the fusion boundary.

When the base metal and the weld metal show two different crystal structures at the solidification temperature, nucleation of solid weld metal occurs on heterogeneous sites on the partially melted base metal at the fusion boundary. The fusion boundary shows random mis-orientations between base metal grains and weld metal grains as a result of heterogeneous nucleation at the pool boundary. The weld metal grains follow or do not follow special orientation relationships with the base metal grains they are in contact with, namely, orient themselves so that certain atomic planes are parallel to specific planes and directions in the base-metal grains.

The grain structure near the fusion line of a weld is dominated either by epitaxial growth when the base metal and the weld metal have the same crystal structure or by nucleation of new grains when they have different crystal structures. Away from the fusion line, however, the grain structure is dominated by a different mechanism known as competitive growth.

During weld metal solidification grains tend to grow in the direction perpendicular to pool boundary since this is the direction of the maximum temperature gradient and hence maximum heat extraction. However, columnar dendrites or cells within each grain tend to grow in the easy-growth direction. Hence during solidification, grains with their easy-growth direction essentially perpendicular to the pool boundary grow more easily and crowd out those less favourably oriented grains. This mechanism of competitive growth dominates the grain structure of the bulk weld metal.

The weld pool becomes tear-drop shaped at high welding speeds and elliptical at low welding speeds. Since the trailing pool boundary of a tear-drop shaped weld pool is essentially straight, the columnar grains are also essentially straight in order to grow perpendicular to the pool boundary. On the other hand, since the trailing boundary of an elliptical weld pool is curved, the columnar grains are also curved in order to grow perpendicular to the pool boundary.

Axial grains can also exist in the fusion zone. Axial grains can initiate from the fusion boundary at the starting point of the weld and continue along the length of the weld, blocking the columnar grains growing inward from the fusion lines. Like other columnar grains, these axial grains also tend to grow perpendicular to the weld pool boundary. With a tear-drop shaped pool, only a short section of the trailing pool boundary can be perpendicular to the axial direction, and the region of axial grains is, hence, rather narrow. With an elliptical weld pool, however, a considerably longer section of the trailing pool boundary can be perpendicular to the axial direction, and the region of axial grains can hence be considerably wider. Axial grains have been reported in Al alloys, austenitic stainless steels, and iridium (Ir) alloys.

The mechanisms of nucleation of grains in the weld metal help understand the micro-structure of the material around the weld pool. There are three possible mechanisms for new grains to nucleate during welding. These mechanisms are (i) dendrite fragmentation, (ii) grain detachment, and (iii) heterogeneous nucleation.

Weld pool convection can in principle cause fragmentation of dendrite tips in the mushy zone. These dendrite fragments are carried into the bulk weld pool and act as nuclei for new grains to form if they survive the weld pool temperature. It is interesting to note that this mechanism has been referred frequently as the grain refining mechanism for weld metals though without proof.

Weld pool convection can also cause partially melted grains to detach themselves from the solid–liquid mixture surrounding the weld pool. Like dendrite fragments, these partially melted grains, if they survive in the weld pool, can act as nuclei for the formation of new grains in the weld metal.

Foreign particles present in the weld pool upon which atoms in the liquid metal can be arranged in a crystalline form can act as heterogeneous nuclei. Nucleation of solid weld metal on heterogeneous sites has been observed on the partially melted base metal at the fusion boundary when the weld metal and the base metal differ in crystal structure.

The weld pool surface can be undercooled thermally to induce surface nucleation by exposure to a stream of cooling gas or by instantaneous reduction or removal of the heat input. When this occurs, solid nuclei can form at the weld pool surface. These solid nuclei then grow into new grains as they shower down from the weld pool surface because of their higher density than the surrounding liquid metal.

The formation of equiaxed grains is improved by higher heat inputs and welding speeds. Equiaxed grains can form a band along the centre-line of the weld and block off columnar grains as the heat input and welding speed are increased. As the heat input and the welding speed are increased, the temperature gradient (G) at the end of the weld pool is reduced. Also, as the welding speed is increased, the solidification rate (R) of the weld metal is also increased. The ratio G/R is to be decreased and the constitutional super-cooling in front of the advancing solid-liquid interface is, hence, to be increased.

The transition to an equiaxed grain structure is because of the existence of a sufficiently long constitutionally under-cooled zone in the weld pool. However, the transition is not because of the constitutional super-cooling alone. In fact, it is observed that equiaxed grains in the fusion zone form by heterogeneous nucleation is aided by constitutional super-cooling.

The weld metal grain structure can affect its mechanical properties considerably. The weld metal ductility drops greatly when the columnar grains are pointed to the weld centre-line, i.e., when the grains become nearly normal to the tensile axis. Also, the weld metal tensile strength increases as the quantity of equiaxed grains increases.

The formation of fine equiaxed grains in the fusion zone has two main advantages. First, fine grains help in reducing the susceptibility of the weld metal to solidification cracking during welding. Second, fine grains can improve the mechanical properties of the weld, such as the ductility and fracture toughness in the case of steels and stainless steels. Hence, much effort has been made to try to grain refine the weld fusion zone. This includes the application of grain refining techniques which have been originally developed for casting.

The different techniques which have been used to control the weld metal grain structure are (i) inoculation which involves the addition of nucleating agents or inoculants to the liquid metal to be solidified, (ii) different dynamic grain refining techniques, such as liquid pool stirring, mould oscillation, and ultrasonic vibration of the liquid metal, and include similar techniques like weld pool stirring, arc oscillation, and arc pulsation, which have been applied to fusion welding, (iii) stimulated surface nucleation in which a stream of cool Ar gas is directed on the free surface of molten metal to cause thermal under-cooling and induce surface nucleation, (iv) manipulation of the orientation of columnar grains in the welds by low-frequency arc oscillation, and (v) welding under the high gravity produced by a centrifuge welding system and hence eliminating the narrow band of non-dendritic equiaxed grains along the fusion boundary.

As constitutional super-cooling increases, the solidification mode changes from planar to cellular and from cellular to columnar dendritic, and equiaxed dendritic as the degree of constitutional super-cooling at the pool boundary increases. Heterogeneous nucleation aided by constitutional super-cooling promotes the formation of equiaxed grains in the weld metal.

While the solidification mode can vary from one weld to another, it can also vary within a single weld from the fusion line to the centre-line. This is explained by growth rate (R) and the temperature gradient (G). The ratio G/R decreases from the fusion line toward the centre-line. This suggests that the solidification mode can change from planar to cellular, columnar dendritic, and equiaxed dendritic across the fusion zone.

The spacing between dendrite arms or cells, just as the solidification mode, can also vary across the fusion zone. The cooling rate (G x R) is higher at the weld centre-line and lower at the fusion line. This suggests that the dendrite arm spacing decreases from the fusion line to the centre-line since the dendrite arm spacing decreases with increasing cooling rate.

The variation in the dendrite arm spacing across the fusion zone can be further explained with the help of thermal cycles. The cooling time through the solidification temperature range is shorter at the weld centre-line and longer at the fusion line. As such, the cooling rate through the solidification temperature range increases and the dendrite arm spacing decreases from the fusion line to the centre-line. The solidification micro-structure gets finer from the fusion line to the centre-line.

The heat input and the welding speed can affect the solidification mode of the weld metal considerably. The solidification mode changes from planar to cellular and dendritic as the ratio G/R decreases. The higher is the heat input under the same welding speed, the lower is the temperature gradient and hence the lower is the ratio G/R. Hence, at higher heat inputs, G/R is lower and dendritic solidification prevails, while at lower heat inputs, G/R is higher and cellular solidification prevails.

The heat input and the welding speed can also affect the spacing between dendrite arms and cells. The dendrite arm spacing or cell spacing decreases with increasing cooling rate. As compared to arc welding, the cooling rate in EBW or LBW is higher and the weld metal micro-structure is finer. Under the same heat input (Q), the cooling rate increases with increasing welding speed (V). Hence, at higher welding speeds, the cooling rate is higher and the cells are finer, while at lower welding speeds, the cooling rate is lower and the cells are coarser.

The cooling rate increases with decreasing heat input-welding speed ratio Q/V. This ratio also represents the quantity of heat per unit length of the weld. Hence, the dendrite arm spacing or cell spacing can be expected to increase with increasing Q/V or quantity of heat per unit length of the weld.

The finer is the dendrite arm spacing, the higher is the ductility and yield strength of the weld metal and the more effective is the post-weld heat treatment, because of the finer distribution of inter-dendritic eutectics.

The temperature gradient (G) ahead of the solid-liquid interface can also be increased because of the smaller distance between the heat source and the pool boundary. Hence, the product G x R or the cooling rate is increased considerably by the action of arc oscillation. This explains why the micro-structure is finer in the oscillated arc weld. The micro-structure in oscillated arc welds is much more uniform than that in welds without oscillation. In oscillated arc welds, the weld centre-line is no longer a location where the cooling rate (or G x R) is clearly at its maximum.

Post-solidification phase transformations, when they occur, can change the solidification micro-structure and properties of the weld metal. It is, hence, essential that post-solidification phase transformations be understood in order to understand the weld metal micro-structure and properties. There are two major types of post-solidification phase transformations in the weld metal. The first involves the ferrite-to-austenite transformation in welds of austenitic stainless steels, and the second involves the austenite-to-ferrite transformation in welds of low-carbon, and low-alloy steels.

The dendrites or cells in the weld metal are not always visible. First, considerable solute partitioning does not occur during solidification if the partition ratio is too close to 1. The micro-segregation, especially solute segregation to the inter-dendritic or inter-cellular regions, in the resultant weld metal can be too little to bring out the dendritic or cellular structure in the grain interior even though the grain structure itself can still be very clear. Second, if solid-state diffusion occurs rapidly, micro-segregation either is small or is homogenized quickly, and the dendrites or cells in the resultant weld metal can be unclear. Third, post-solidification phase transformations, if they occur, can produce new micro-structures in the grain interior and / or along grain boundaries and the sub-grain structure in the resultant weld metal can be over-shadowed.

Several factors have effect on the development of micro-structure of the weld metal. These are (i) the weld metal composition, (ii) the cooling time from 800 deg C to 500 deg C, the weld metal O2 content, and the austenite grain size.

Chemical inhomogeneities in the weld metal which include solute segregation, banding, inclusions, and gas porosity can occur in the weld metal. Solute segregation can be either micro-segregation or macro-segregation. Micro-segregation refers to composition variations across structures of microscopic sizes, e.g., dendrite arms or cells. Macro-segregation, on the other hand, refers to variations in the local average composition (composition averaged over several dendrites) across structures of macroscopic sizes, e.g., the weld.

In case of micro-segregation, alloying elements with an equilibrium segregation coefficient of less than 1 tend to segregate toward the boundary between cells or dendrite arms, and those with higher than 1 tend to segregate toward the core of cells or dendrite arms. Micro-segregation can have a considerable effect on the solidification cracking susceptibility of the weld metal.

Micro-segregation can be reduced considerably by solid-state diffusion during and after solidification. Hence, micro-segregation measured after welding does not represent the true micro-segregation during welding, which is more relevant to solidification cracking.

In addition to solid-state diffusion, micro-segregation can also be affected by the extent of dendrite tip undercooling. The difference between the equilibrium liquidus temperature and the dendrite tip temperature is the total under-cooling, which can be divided into four parts, namely (i) concentration-induced under-cooling, (ii) curvature-induced under-cooling, thermal under-cooling, and (iv) kinetic under-cooling.

The solute rejected by the dendrite tip into the liquid can pile up and cause under-cooling at the dendrite tip, similar to constitutional super-cooling at a planar growth front. The equilibrium liquidus temperature in a phase diagram is for a flat solid-liquid interface, and it is suppressed if the interface has a radius of curvature like a dendrite tip. Thermal under-cooling is present where there is a significant nucleation barrier for the liquid to transform into solid. The kinetic under-cooling, which is normally negligible, is associated with the driving force for the liquid atoms to become attached to the solid. It has been observed that the higher is the velocity of the dendrite tip, the smaller is the radius of the dendrite tip, and the larger the under-cooling at the dendrite tip.

In addition to micro-segregation across dendrites, micro-segregation can also exist in the weld metal as a result of banding during weld pool solidification. Banding in welds can cause perturbations in the solidification structure as well as the solute concentration.

Banding in the weld metal can occur because of a number of reasons. Fluctuations in the welding speed during manual welding or arc pulsing during pulsed arc welding can cause banding. However, even under steady-state welding conditions, banding can still occur, as seen by the surface rippling of the weld. Beside fluctuations in the welding speed and the power input, the mechanisms which also have been proposed are (i) solidification halts because of the rapid evolution of latent heat caused by high solidification rates during welding, (ii) oscillations of weld pool metal because of uncontrollable variations in arc stability and the downward stream of the shielding gas, and (iii) fluctuations in weld pool turbulence because of electro-magnetic effects.

Inclusions and gas porosity tend to deteriorate the mechanical properties of the weld metal. Gas-metal and slag-metal reactions can produce gas porosity and inclusions in the weld metal and affect the weld metal properties. Inclusions can also result from incomplete slag removal during multiple-pass welding.

Dissimilar metal welding is frequently encountered in welding, where a filler metal different in composition from the base metal is used or where two base metals different in composition are welded together. In dissimilar metal welding, the region near the fusion boundary frequently differs considerably from the bulk weld metal in composition and sometimes even micro-structure and properties. The region is called the unmixed zone, filler-metal-depleted area, partially mixed zone, intermediate mixed zone, and hard zone. It has been observed in different welds, including stainless steels, alloy steels, Al alloys, and super-alloys.

Inhomogeneities in the region along the fusion boundary have been reported to cause problems, including H2 cracking, corrosion, and stress corrosion cracking. Martensite frequently exists in the region in C  steels or alloy steels welded with austenitic stainless-steel fillers. This is since the weld metal composition here can be within the martensite region of the constitutional diagrams.

Weld pool convection can normally mix the weld pool well to minimize macro-segregation across the resultant weld metal. However, in single-pass dissimilar-metal welding, macro-segregation can still occur if weld pool mixing is incomplete. In multiple-pass dissimilar-metal welding, macro-segregation can still occur even if weld pool mixing is complete in each pass. Macro-segregation can occur in a dissimilar weld between two different base metals because of insufficient mixing in the weld pool.

Solidification cracking, which is observed frequently in castings and ingots, can also occur in fusion welding. Such cracking is intergranular, i.e., along the grain boundaries of the weld metal. It occurs during the terminal stage of solidification, when the tensile stresses developed across the adjacent grains exceed the strength of the almost completely solidified weld metal. The solidifying weld metal tends to contract because of both solidification shrinkage and thermal contraction. The surrounding base metal also tends to contract, but not as much, since it is neither melted nor heated as much on the average. Hence, the contraction of the solidifying metal can be hindered by the base metal, especially if the work-piece is constrained and cannot contract freely. Hence, tensile stresses develop in the solidifying weld metal. The severity of such tensile stresses increases with both the degree of constraint and the thickness of the work-piece.

The different theories of solidification cracking are effectively identical and embody the concept of the formation of a coherent inter-locking solid network which is separated by essentially continuous thin liquid films and hence ruptured by the tensile stresses. The fracture surface frequently reveals the dendritic morphology of the solidifying weld metal. If a sufficient quantity of liquid metal is present near the cracks, it can ‘back-fill’ and ‘heal’ the incipient cracks.

Metallurgical factors which have been known to affect the solidification cracking susceptibility of weld metals include (i) the solidification temperature range, (ii) the quantity and distribution of liquid at the terminal stage of solidification, (iii) the primary solidification phase, (iv) the surface tension of the grain boundary liquid, and (v) the grain structure. All these factors are directly or indirectly affected by the weld metal composition. The first two factors are affected by micro-segregation during solidification. Micro-segregation in turn can be affected by the cooling rate during solidification. In fact, in austenitic stainless steels, the primary solidification phase can also be affected by the cooling rate.

Generally speaking, the wider the solidification (freezing) temperature range, the larger the ‘solid + liquid’ region in the weld metal or the mushy zone and hence the larger the area which is weak and susceptible to solidification cracking. The solidification temperature range of an alloy increases as a result of either the presence of undesirable impurities such as sulphur and phosphorus in steels and nickel-base alloys or intentionally added alloying elements.

It is interesting to note that, because of the steep angle of support between columnar grains growing from opposite sides of the weld pool, welds made with a tear-drop shaped weld pool tend to be more susceptible to centre-line solidification cracking than welds made with an elliptical-shaped weld pool. A steep angle seems to favour the head-on impingement of columnar grains growing from opposite sides of the weld pool and the formation of the continuous liquid film of low-melting-point segregates at the weld centre-line. As a result, centre-line solidification cracking occurs under the influence of transverse contraction stresses.

Without the presence of stresses acting on adjacent grains during solidification, no cracking can occur. Such stresses can be because of the thermal contraction or solidification shrinkage or both. Austenitic stainless steels have relatively high thermal expansion coefficients (as compared with mild steels) and, hence, are frequently prone to solidification cracking.

The degree of restraint of the work-piece is another mechanical factor of solidification cracking. For the same joint design and material, the higher is the restraint of the work-piece, there is more likelihood for the solidification cracking to occur.

Weld metals of crack-susceptible compositions are to be avoided. In autogenous welding where no filler metal is used, and the weld metal composition is determined by the base-metal composition. For avoiding or reducing solidification cracking, base metals of susceptible compositions are to be avoided. When a base metal of a crack-susceptible composition is to be welded, then a filler metal of a proper composition can be selected to adjust the weld metal composition to a less susceptible level.

The weld metal Mn content can affect considerably the solidification cracking. It is frequently kept high enough to ensure the formation of MnS (manganese sulphide) rather than FeS (ferrous sulphide). This is since the high melting point and the globular morphology of MnS tend to render sulphur less detrimental. At relatively low C levels, the solidification cracking tendency can be reduced by increasing the Mn/S ratio. However, at higher C levels (i.e., C ranging from 0.2 % to 0.3 %) increasing the Mn-S ratio is no longer effective. In such cases lowering the weld metal C content, if permissible, is more effective. Fig 6 shows the effect of Mn-S ratio and C content on solidification cracking susceptibility of C steel weld metal.

Fig 6 Effect of Mn-S ratio and carbon content

One way of lowering the weld metal C content is to use low-C electrodes. In fact, in welding high-C steels one is frequently needed to make the first bead (i.e., the root bead) with a low-C electrode. This is since the first bead tends to have a higher dilution ratio and a higher C content than subsequent beads. A high C content is undesirable since it promotes not only the solidification cracking of the weld metal but also the formation of brittle martensite and, hence, the post solidification cracking of the weld metal. Hence, in welding steels of very high C contents (e.g., higher than 1 % C), extra steps are needed to be taken for avoiding introduction of excessive quantities of C from the base metal into the weld metal.

Welds with coarse columnar grains are frequently more susceptible to solidification cracking than those with fine equiaxed grains. Hence, it is desirable to grain refine the weld metal.

Magnetic arc oscillation has been reported to reduce solidification cracking. Transverse arc oscillation at low frequencies can produce alternating columnar grains. This type of grain structure can be effective in reducing solidification cracking. Columnar grains which reverse their orientation at regular intervals force the crack to change its direction periodically, hence making crack propagation difficult. A minimum crack susceptibility exists at a rather low frequency, where alternating grain orientation is most pronounced. This frequency can vary with the welding speed.

The use of high-intensity heat sources (electron beam or laser beam) considerably reduces the distortion of the work-piece and hence the thermally induced strains. Less restraint and proper pre-heating of the work-piece can also help reduce strains. Local heating decreases the quantity of plastic straining resulting from the welding operation and produces a less stressful situation behind the weld pool.

The weld bead shape can also affect solidification cracking. When a concave single-pass fillet weld cools and shrinks, the outer surface is stressed in tension. The outer surface can be considered as being pulled toward the toes and the root. However, by making the outer surface convex, pulling toward the root actually compresses the outer surface and offsets the tension caused by pulling toward the toes. Hence, the tensile stresses along the outer surface are reduced, and the tendency for solidification cracking to initiate from the outer surface is lowered. However, it is pointed out that excessive convexity can produce stress concentrations and induce fatigue cracking or H2 cracking at the toes. In multiple-pass welding, solidification cracking can also initiate from the weld surface if the weld passes are too wide and concave. The weld width-to-depth ratio can also affect solidification cracking. Deep narrow welds with a low width-to-depth ratio can be susceptible to weld centre-line cracking. This is because of the steep angle of the support between columnar grains growing from opposite sides of the weld pool. This type of cracking is frequently observed in deep and narrow welds produced by EBW and SAW.

The grain boundary liquid has a tendency to solidify essentially upward and toward the weld regardless of its location with respect to the weld. This directional solidification is caused by the high-temperature gradients toward the weld during welding. It has been normally accepted that the grain boundary liquid between two neighbouring grains solidifies from both grains to the middle between them. However, the micrographs show that it solidifies from one grain to the other i.e., in the direction upward and toward the weld. This, in fact, suggests grain boundary migration in the same direction. However, if the grains in the PMZ zone are very thin or very long, there is not much grain boundary area facing the weld. Hence, solidification of the grain boundary liquid is still directional but just upward.

As the grain boundary liquid solidifies, solute atoms are rejected by the solid into the liquid if the equilibrium partition coefficient is less than 1. The grain boundary liquid solidifies first as a solute-depleted, but finally as eutectic when the liquid composition reaches eutectic composition.

The alpha band along the eutectic grain boundary is planar, namely, without cells or dendrites. This suggests that the solidification mode of the grain boundary liquid is planar. Although planar solidification of the grain boundary liquid pre-dominates in the partially melted zone, cellular solidification can also occur. These cellular alpha bands share two common characteristics. First is that they are frequently located near the weld bottom. Second is that on an average, they appear considerably thicker than the planar alpha bands nearby. These characteristics can be because of the lower vertical temperature gradient (G) in the area or back-filling of liquid from the weld pool.

Since a thicker grain boundary liquid has to solidify faster, the vertically upward solidification rate (R) is higher. The lower G/R in the area suggests a higher chance for constitutional super-cooling and hence cellular instead of planar solidification. However, it is to be noted that planar solidification changes to cellular solidification gradually, and a thinner grain boundary liquid does not have enough room for the transition to take place. Hence, planar grain boundary solidification does not necessarily mean that G/R is high enough to avoid cellular solidification.

In case of a cast iron weld, the partially melted zone has austenite, ferrite, and graphite phases. This area tends to freeze as white iron because of the high cooling rates and becomes very hard.

The PMZ can suffer from liquation cracking, loss of ductility, and H2 cracking. Liquation cracking, which is cracking induced by grain boundary liquation in the PMZ during welding, is also called PMZ or hot cracking. The susceptibility of the PMZ to liquation cracking can be evaluated using several different methods, such as Varestraint testing, circular-patch testing and hot-ductility testing, etc.

Liquation cracking in the PMZ is intergranular. Liquation cracking can also occur along the fusion boundary. The presence of a liquid phase at the intergranular fracture surface can be either evident or unclear.

Since the PMZ is weakened by grain boundary liquation, it cracks when the solidifying weld metal contracts and pulls it. Majority of the Al alloys are susceptible to liquation cracking. This is because of their wide PMZ (because of wide freezing temperature range and high thermal conductivity), large solidification shrinkage (solid density significantly higher than liquid density), and large thermal contraction (large thermal expansion coefficient). The solidification shrinkage of Al is as high as 6.6 %, and the thermal expansion coefficient of Al is roughly twice that of iron base alloys.

H2-induced cracking is inter-granular cracking in the PMZ and the adjacent region in the fusion zone where mixing between the filler and the weld metal is incomplete. The creation of liquated films on the grain boundaries in the PMZ provides preferential paths along which H2 from the weld metal can diffuse across the fusion boundary. This is because liquid iron can dissolve around three times to four times more nascent H2 than the solid, hence making the liquated grain boundaries serve as ‘pipelines’ along which H2 from the weld metal can readily diffuse across the fusion boundary. When these segregated films resolidify, they not only are left super-saturated with H2 but also show a higher hardenability because of the solute segregation. Hence, they serve as preferred nucleation sites for H2-induced cracking.

Remedies for the problems associated with the partially melted zone can be grouped into four categories namely filler metal, heat source, degree of restraint, and base metal. Liquation cracking can be reduced by selecting the proper filler metal. The size of the PMZ and hence the extent of PMZ liquation can be reduced by reducing the heat input. Liquation cracking and H2-induced cracking in the PMZ are both caused by the combination of a susceptible micro-structure and the presence of tensile stresses. The sensitivity of the PMZ to both types of cracking can be reduced by decreasing the degree of restraint and hence the level of tensile stresses. Liquation cracking can be reduced by selecting the proper base metal for welding if it is feasible. The base-metal composition, grain structure, and micro-segregation can affect the susceptibility of the PMZ to liquation cracking considerably.

When impurities such as sulphur and phosphorus are present, the freezing temperature range can be widened rather considerably. The widening of the freezing temperature range is because of the lowering of the incipient melting temperature, which is effectively the same as the liquation temperature in the sense of liquation cracking.

The course are the grains, the less ductile is the PMZ. Also, the course are the grains, the less is the grain boundary area and hence the more concentrated are the impurities or low-melting-point segregates at the grain boundary. Hence, a base metal with coarser grains is expected to be more susceptible to liquation cracking in the PMZ. Cracking is much more severe with coarse grains.

PMZ cracking is more severe in welds made transverse to the rolling direction than those made parallel to the rolling direction.

In the welding of as-cast materials, the PMZ is particularly susceptible to liquation cracking because of the presence of low-melting-point grain boundary segregates. Upon heating during welding, excessive grain boundary liquation occurs in the PMZ, making it highly susceptible to liquation cracking.

Carbon steels and alloy steels are more frequently welded than any other materials because of their widespread applications and good weldability. In general, C steels and alloy steels with higher strength levels are more difficult to weld because of the risk of H2 cracking.

Metals can be strengthened in several ways, including solution hardening, work hardening, precipitation hardening, and transformation hardening. The effectiveness of the last three methods can be reduced significantly by heating during welding in the area called the heat-affected zone (HAZ), where the peak temperatures are too low to cause melting but high enough to cause the micro-structure and properties of the materials to change considerably.

The effect of work hardening is completely gone in the fusion zone because of melting and is partially lost in the HAZ because of recrystallization and grain growth. These strength losses are to be taken into account in structural designs involving welding. Severe HAZ grain growth can result in coarse grains in the fusion zone because of epitaxial growth. Fracture toughness is normally poor with coarse grains in the HAZ zone and the fusion zone.

The loss of strength in the HAZ can be explained with the help of thermal cycles. The closer to the fusion boundary, the higher the peak temperature becomes and the longer the material stays above the effective recrystallization temperature. Under rapid heating during welding, the recrystallization temperature can increase since recrystallization needs diffusion and diffusion takes time. Since the strength of a work-hardened material decreases with increasing annealing temperature and time, the strength or hardness of the HAZ decreases as the fusion boundary is approached.

Grain growth in the HAZ can also be explained with the help of thermal cycles. The closer to the fusion boundary, the higher the peak temperature becomes and the longer the material stays at high temperatures. Since grain growth increases with increasing annealing temperature and time, the grain size in the HAZ increases as the fusion boundary is approached.

Welding parameters affect HAZ strength. Both the size of the HAZ and the retention time above the effective recrystallization temperature increase with increasing heat input per unit length of the weld, i.e., the ratio of heat input to welding speed. Hence, the loss of strength in the HAZ becomes more severe as the heat input per unit length of the weld is increase.

The HAZ in a C steel can be related to the Fe-C phase diagram, as shown in Fig 7a and Fig 7b, if the kinetic effect of rapid heating during welding on phase transformations is neglected. The HAZ can be considered to correspond to the area in the work-piece that is heated to between the lower critical temperature A1 (the eutectoid temperature) and the peritectic temperature. Similarly, the PMZ can be considered to correspond to the areas between the peritectic temperature and the liquidus temperature, and the fusion zone to the areas above the liquidus temperature. The Fe-C phase diagram and the CCT diagrams for heat treating C steels can be useful for welding as well, but some fundamental differences between welding and heat treating are to be recognized.

Fig 7 Welding of carbon steel

The thermal processes during the welding and heat treating of a C steel differ from each other considerably, as shown in Fig 7c and Fig 7d. First, in welding the peak temperature in the HAZ can approach 1,500 deg C. In heat treating, however, the maximum temperature is around 900 deg C, which is not much above the upper critical temperature A3 needed for austenite (gamma) to form. Second, the heating rate is high and the retention time above A3 is short during most welding processes (electroslag welding being a notable exception). In heat treating, on the other hand, the heating rate is much slower and the retention time above A3 is much longer. The A1 and A3 temperatures during heating are frequently referred to as the Ac1 and Ac3 temperatures, respectively.

For steels containing higher quantities of carbide-forming elements, such as Cr, Ti, vanadium (V), tungsten (W), and molybdenum (Mo), the effect of the heating rate becomes more pronounced. This is since the diffusion rate of such elements is orders of magnitude lower than that of C and also since they hinder the diffusion of C. As a result, phase transformations are delayed to a greater extent.

The combination of high heatment rates and short retention time above Ac3 in welding can result in the formation of inhomogeneous austenite during heating. This is since there is not enough time for C atoms in austenite to diffuse from the prior pearlite colonies of high C contents to prior ferrite colonies of low C contents. Upon rapid cooling, the former can transform into high-C martensite colonies while the latter into low-C ferrite colonies. Hence, the micro-hardness in the HAZ can scatter over a wide range in welds made with high heating rates.

As a result of high peak temperatures during welding, grain growth can take place near the fusion boundary. The slower is the heating rate, the longer is the retention time above Ac3 is and hence the more severe grain growth becomes. In the heat treatment, however, the maximum temperature used is only around 900 der C for avoiding the grain growth.

The CCT diagrams for welding can be obtained by using a weld thermal simulator and a high-speed dilatometer which detects the volume changes caused by phase transformations. However, since CCT diagrams for welding are frequently unavailable, those for heat treatment have been used. These two types of CCT diagrams can differ from each other because of kinetic reasons. For example, grain growth in welding can shift the CCT diagram to longer times favouring transformation to martensite. This is since grain growth reduces the grain boundary area available for ferrite and pearlite to nucleate during cooling. However, rapid heating in welding can shift the CCT diagram to shorter times, discouraging transformation to martensite. Carbide-forming elements such as Cr, Mo, Ti, V, and niobium (Nb), when they are dissolved in austenite, tend to increase the hardenability of the steel. Because of the sufficient time available in heat treatment, such carbides dissolve more completely and hence improves the hardenability of the steel. This is normally not possible in welding because of the high heating rate and the short high-temperature retention time encountered in the HAZ.

Leave a Comment