Materials Science & Metallurgy Master of Philosophy, Materials Modelling,

Course MP10, Process Modelling, H. K. D. H. Bhadeshia

Lecture 5: Fusion Welding

Introduction

The purpose of the next three lectures is to illustrate how one aspect of a process
of great importance in industry, might be modelled. The first lecture introduces the
process in the context of steels, the second lecture introduces steel microstructure and
the third a model for the fusion zone of the weld microstructure.
Fusion welding is of greatest importance in the fabrication of engineering struc-
tures. There are many ways in which the process is carried out, but all of them involve
the deposition of a small amount of molten steel within a gap between the components
to be joined. When the steel solidifies, it welds the components together. Fig. 1 il-
lustrates the process. The heat source can be varied, but common sources include the
electric arc, resistance to electrical current, lasers, electron beams and friction. The
source of metal can be a manually fed wire, wire fed automatically from a spool, metal-
lic powder or metallic foil. In autogenous welds there is no external source of metal.
The process achieves very high temperatures so the region subjected to heat must
be protected from the environment. Typical methods of protection include shielding
with inert gas, shielding with flux powder which melts to form a protective slag, and
welding in a vacuum.

Fig. 1: Schematic illustration of the welding process.

1
 Fig. 2: A multipass arc weld illustrating the heterogeneous mi-
crostructure. The plate is about 20 mm thick. Photograph courtesy
of L.–E. Svensson.

The metallurgy of the welded joint can be categorised into two major regions, the
fusion zone and the heat–affected zone (Fig. 2).
The fusion zone represents both the deposited metal and the parts of the steel
component melted during the process, and is a solidification microstructure. The heat–
affected zone, on the other hand, represents those regions in the close proximity of the
weld, where the heat input during welding changes the microstructure without melting
the steel. Virtually every aspect of phase transformations in steels is relevant to the
subject of welding. There is an opportunity for a whole series of transformations to
occur successively as the weld cools from the liquid state. Because of time limitations,
we shall focus only on the as–deposited microstructure of the fusion zone.

Iron

The two main crystal structures of iron are illustrated in Fig. 3. Austenite with the
face–centred cubic structure is stable at temperatures in the range 911.5– 1394.0 ◦ C,
whereas ferrite (body–centred cubic structure) is stable at temperatures less than
911.5 ◦ C or greater than 1394 ◦ C to the melting temperature. Thus, ferrite is the first
solid–phase to form on cooling liquid iron; it then undergoes solid–state transformation
into austenite, which on further cooling reverts to ferrite. This sequence is essentially
correct for the vast majority of steels, which are alloys of iron.

2
 1
z = 2- z=0,1

cubic-close body-centered
packed cubic

Fig. 3: Projections of the crystal structures of austenite and ferrite,

along their respective z –axes, which are normal to the plane of the
diagram. The dark atoms have z –axis co–ordinates 0 and 1, whereas
the lighter atoms are at 12 z .

Weld Solidification

We have seen that iron is ferritic at temperatures just below the melting point.
As it cools, the ferrite then transforms to austenite, only to revert back to ferrite
on continued cooling. Most steels contain modest concentrations of alloying elements,
and hence show similar crystal structure changes as pure iron. Weld deposits therefore
begin solidification with the epitaxial growth of columnar ferrite from the hot grain–
structure of the parent plate at the fusion surface (Fig. 4). The grains are anisotropic
because they grow along the direction of heat flow. Those grains with their < 100 >
directions parallel to the heat flow direction grow fastest and hence stifle the growth
of unsuitably oriented grains. The width of the columnar grains therefore increases
with distance away from the fusion boundary.
As already pointed out, the ferrite undergoes a solid–state transformation into
austenite as the temperature decreases. The austenite nucleates at the ferrite grain
boundaries and develops into a columnar austenite grain structure which strongly
resembles that of the original ferrite grains (Fig. 4).
The shape of the columnar grains is like that of space–filling hexagonal prisms.
The grains are typically about 100 µ m wide and about 5000 µ m in length. This
is quite unlike an equiaxed grain structure, and because of the fewer grain junctions
involved, allows the hardenability of a weld to be larger than that of a wrought alloy.

3
 Fig. 4: Illustration of the epitaxial growth of columnar grains from
the fusion boundary. These grains are initially ferrite, but then trans-
form into columnar grains of austenite which resemble the original
microstructure.

Solidification does not occur under equilibrium conditions during welding.

Solidification–induced chemical segregation, and compositional variations due to un-
controlled changes in the welding conditions make the solidification microstructure
inhomogeneous. The amplitude of these variations becomes larger as the alloy concen-
tration increases.

Mineral fluxes or inert gas shrouds are employed in order to protect the hot metal
against environmental attack during welding. Such protection is not entirely effective,
with the result that the oxide content of welds tends to be much larger than that of
wrought steel (Table 1). The oxide particles are entrapped in the fusion zone during
solidification. As discussed later, these nonmetallic particles serve as heterogeneous
nucleation sites and hence are of considerable importance in the evolution of the mi-
crostructure. Table 1 reveals some other interesting differences between the plate and
weld compositions. The copper concentration of the weld is large because in this case,
the welding wire has a copper coating to enable better electrical contact. The silicon
concentration in the weld is larger than both the wire and the plate, because the ex-
cess silicon is acquired by decomposition of the protective flux. These observations
emphasise the complexity of the welding process, in which the chemical composition
of the final weld deposit depends on many variables, including the plate, wire and flux
compositions.

4
 C Mn Si Cu Al N O
Plate 0.21 1.0 0.2 0.05 0.04 0.01 0.004
Wire 0.14 1.5 0.2 0.31 0.01 0.01 0.001
Weld 0.16 1.1 0.3 0.16 0.01 0.01 0.053

Table 1 A comparison of the chemical composition (wt.%) of a sub-

merged arc weld with that of the plate being welded, and the wire
used as the consumable electrode. The welding conditions used were
34 V, 900 Amps (D.C. +ve), at a welding speed of 0.005 m s−1 , with
a calcium silicate flux.

Mechanism of Solid–State Transformation

Before explaining the actual series of solid–state transformations in steels, it is

essential to understand the atomic mechanisms involved. These two crystal structures
can transform into one another by two essential mechanisms, reconstructive and dis-
placive. In the former case, the bonds are all broken and the atoms rearranged into
the new pattern by the random motion of atoms in a way which minimises free energy.
In displacive transformations, the pattern in which the atoms are arranged is altered
by deformation, without diffusion; there is consequently a great deal of strain involved
in the transformation.
The atomic arrangement in a crystal can be altered either by breaking all the
bonds and rearranging the atoms into an alternative pattern (reconstructive trans-
formation), or by homogeneously deforming the original pattern into a new crystal
structure (displacive transformation), Fig. 5.
In the displacive mechanism the change in crystal structure also alters the macro-
scopic shape of the sample when the latter is not constrained. The shape deformation
during constrained transformation is accommodated by a combination of elastic and
plastic strains in the surrounding matrix. The product phase grows in the form of thin
plates to minimise the strains. The atoms are displaced into their new positions in a
coordinated motion. Displacive transformations can therefore occur at temperatures
where diffusion is inconceivable within the time scale of the experiment. Some solutes
may be forced into the product phase, a phenomenon known as solute trapping. Both

5
 Fig. 5: The main mechanisms of transformation. The parent crys-
tal contains two kinds of atoms. The figures on the right represent
partially transformed samples with the parent and product unit cells
outlined in bold. The transformations are unconstrained in this illus-
tration.

the trapping of atoms and the strains make displacive transformations less favourable
from a thermodynamic point of view.
It is the diffusion of atoms that leads to the new crystal structure during a re-
constructive transformation. The flow of matter is sufficient to avoid any shear com-
ponents of the shape deformation, leaving only the effects of volume change. This
diffusion is necessary even when transformation occurs in pure iron [1]. In alloys, the
diffusion process may also lead to the redistribution of solutes between the phases in
a manner consistent with a reduction in the overall free energy. Fig. 6 is a summary
of the main transformations that occur in steels.

6
Fig. 6: A summary of the essential characteristics of solid–state trans-
formations in steels.