Overview Welding

Welding: Solidification
and Microstructure
S.A. David, S.S. Babu, and J.M. Vitek

Editor’s Note: A hypertext-enhanced version of this article is cracking behavior of alloys. Sometimes, However, microstructure development
available on-line at www.tms.org/pubs/journals/JOM/0306/David-
0306.html it is convenient to consider the FZ as in the FZ is more complicated 15,16
a minicasting. Therefore, parameters because of physical processes that
Parameters that control the solidi- important in determining microstruc- occur due to the interaction of the heat
fication of castings also control the tures in casting, such as growth rate (R), source with the metal during welding,
solidification and microstructure of temperature gradient (G), undercooling including re-melting, heat and fluid
welds. However, various physical (∆T), and alloy composition determine flow, vaporization, dissolution of gasses,
processes that occur due to the interac- the development of microstructures in solidification, subsequent solid-state
tion of the heat source with the metal welds as well. Comprehensive reviews transformation, stresses, and distortion.
during welding add a new dimension of weld pool solidification based on These processes and their interactions
to the understanding of the weld pool profoundly affect weld pool solidifica-
solidification. Conventional theories tion and microstructure. In recent years,
of solidification over a broad range
During the past phenomenological modeling of welding
of conditions can be extended to 15 years, significant processes has provided unprecedented
understand weld pool solidification. In insight into understanding both the
certain cases, because of rapid cooling progress has been made welding process and the welded materi-
rate effects, it is not unusual to observe als. A variety of sophisticated models
nonequilibrium microstructures. Recent
in understanding the that employ analytical and numerical
developments in the application of solidification behavior approaches are capable of describing
computational thermodynamics and many physical processes that occur
kinetic models, studies on single- of the weld pool and during welding.15–25
crystal welds, and advanced in-situ During the past 15 years, significant
characterization techniques have led
the evolution of progress has been made in understand-
to a better understanding of weld microstructure in ing the solidification behavior of
solidification and microstructures. the weld pool and the evolution of
the [fusion zone]. . . . microstructure in the FZ. The applica-
INTRODUCTION
tion of computational thermodynamic
In welding, as the heat source interacts
This overview and kinetic tools has enhanced the
with the material, the severity of thermal [addresses] some of understanding of weld solidification
excursions experienced by the material behavior of complex multi-component
varies from region to region, resulting the current progress systems. Advanced in-situ characteriza-
in three distinct regions in the weldment tion techniques have enabled the
(Figure 1). These are the fusion zone
in understanding weld characterization of phase formation and
(FZ), also known as the weld metal, pool solidification. non-equilibrium effects during weld
the heat-affected zone (HAZ), and the pool solidification. The use of model
unaffected base metal (BM). The FZ alloy single crystals resulted in new
experiences melting and solidification, these parameters are available in the insight into the role of weld pool
and its microstructural characteristics literature.1,2 geometry and dendrite growth selection
are the focus of this article. Most knowledge of weld pool solidi- processes in the development of weld
The microstructure development in fication is derived from the extrapolation microstructure. This overview will
the FZ depends on the solidification of the knowledge of freezing of castings, address some of the current progress in
behavior of the weld pool. The principles ingots, and single crystals at lower understanding weld pool solidification.
of solidification control the size and thermal gradients and slower growth
WELD POOL SHAPE
shape of the grains, segregation, and the rates.1–6 In addition, rapid solidification
distribution of inclusions and porosity. theories have been extended to welds An important aspect of weld solidi-
Solidification is also critical to the hot- solidified at very high cooling rates.7–14 fication is the dynamics of weld pool

14 JOM • June 2003

development and its steady-state geom- the edge to the centerline of the weld. the stability of the solid-liquid interface
etry. Weld pool shape is important in Most of these microstructural features is mostly determined by the thermal and
the development of grain structure and can be interpreted by considering constitutional conditions (constitutional
dendrite growth selection process.6, 26-29 classical theories of nucleation and supercooling) that exist in the immediate
Thermal conditions in and near the growth. vicinity of the interface.35,36 Depending
weld pool and the nature of the fluid In welds, weld pool solidification on these conditions, the interface
flow have been found to influence the often occurs without a nucleation barrier. growth may occur by planar, cellular,
size and shape of the weld pool.16–18,24,25 Therefore, no significant undercooling or dendritic growth. Dendritic growth
Significant advances have been made of the liquid is required for nucleation of the solid, with its multiple branches,
in recent years to understand, in greater of the solid. Solidification occurs is shown in Figure 3. Another example
detail, the dynamics of the heat and fluid spontaneously by epitaxial growth on of changes in solidification morphology
flow in the weld and the subsequent the partially melted grains. This is the directly related to welding conditions is
development of the pool shape. To a case during autogenous welding. In shown in Figure 4. This figure shows a
large extent, convective flow in the certain welds, where filler metals are spot weld on a nickel-based superalloy
weld pool determines weld penetration. used, inoculants and other grain-refining in which the morphology changes from
For arc-welding processes, convection techniques are used in much the same cellular to dendritic as the growth
in the weld pool is mainly controlled way as they are in casting practices. velocity increases toward the center
by buoyancy, electromagnetic forces, In addition, dynamic methods for of the spot weld after the spot weld
and surface-tension forces. In actuality, promoting nucleation such as weld- arc is extinguished. The micrograph
depending on the interplay between pool stirring and arc oscillation have also shows the elimination of a poorly
various driving forces, the convective been used to refine the weld metal aligned dendrite, which is discussed in
flow could be simple or more complex solidification structure.2 Although the greater detail later. The criterion for
with a number of convective cells mechanisms of nucleation in weld metal constitutional supercooling for plane
operating within the weld pool, as are reasonably well understood, not front instability can be mathematically
shown in Figure 2. much attention is given to modeling this stated as:
Recent theoretical developments phenomenon. Often, weld solidification
G L ∆To
include the formulation of a free-surface models assume epitaxial growth and >
computational model to investigate for most of the cases the assumption R DL
coupled conduction and convection seems to be appropriate. However, to plane front will be stable (1)
heat-transfer models to predict not only describe the effects of inoculants, arc
weld pool geometry but also thermal oscillations, and weld pool stirring, heat
G L ∆To
profiles to estimate thermal gradients and mass transfer models18,24,25 have <
and cooling rates critical to determining to be coupled with either probabilistic R DL
solidification structure.25 In addition models such as cellular automata31–33 planar instability will occur (2)
to computational models, neural net or deterministic models using the
models have been applied to predict fundamental equations of nucleation as where GL is the temperature gradient
weld pool geometry.30 These models, described elsewhere.34 in the liquid, R is the solidification
which are empirical in nature, are During growth of the solid in the front growth rate, ∆To is the equilibrium
useful when applied to complex welding weld pool, the shape of the solid-liquid solidification temperature range (at
processes such as hybrid laser-arc interface controls the development of composition Co), and DL is the solute
welding.30 microstructural features. The nature and diffusion coefficient in liquid.
The temperature gradient and growth
MICROSTRUCTURE
rate are important in the combined
Unlike in casting, during welding, forms GR (cooling rate) and G/R since
where the molten pool is moved through they influence the scale of the solidifica-
the material, the growth rate and tion substructure and solidification
temperature gradient vary considerably morphology, respectively. Although the
across the weld pool. Geometrical method of using GR and G/R relations
analyses have been developed that relate to understand the solidification modes
welding speed to the actual growth rates is simple and elegant, modeling of
of the solid at various locations in the solidification morphology in a typical
weld pool.1,2,27 weld must consider other factors such
Along the fusion line the growth rate as fluid flow and the effect of base plate
is low while the temperature gradient Figure 1. A schematic diagram showing texture. Recent work on the in-situ
is steepest. As the weld centerline is the interaction between the heat source observation of weld pool solidification
and the base metal. Three distinct regions
approached, the growth rate increases in the weldment are the fusion zone,
using a transparent analog-metal system
while the temperature gradient de- the heat-affected zone, and the base has produced a greater understanding
creases. Consequently, the microstruc- metal. of the evolution of growth morphology
ture that develops varies noticeably from in welds.37

2003 June • JOM 15

 Solute distribution during weld ogy used in jet and land-based turbine
pool solidification is an important engines.44–46 Unlike in Fe-15Ni-15Cr
phenomenon resulting in segregation 10.5 single-crystal welds where the single
that can significantly affect weldability, 10.0 crystallinity of the weld was maintained,
microstructure, and properties. Studies 9.5 nickel-based superalloys are extremely
extending different solidification models prone to stray grain formation (as shown

Y-Vertical Axis (mm)

9.0
to describe solute distribution during 8.5 in Figure 7). This phenomenon can be
weld solidification are summarized 8.0 attributed to constitutional supercool-
elsewhere.2 In describing the solute 7.5 ing46,47 or dendrite fragmentation48 ahead
distribution under dendritic growth 7.0 of the dendritic front that may nucleate
conditions, consideration should be 6.5 new grains. Recent studies suggest that
given to redistribution at the dendrite 0.1
0.1M/6
M/6 the constitutional supercooling may be
6.0
tip and in the interdendritic regions. 0.0 1.0 2.0 3.0 4.0 5.0 the controlling mechanism for stray
In welds, since the microstructures are X-Horizontal Axis (mm) grain formation.44,47
much finer in scale than in castings, the Figure 2. The calculated fluid-flow pattern
in a stainless-steel stationary arc weld pool NONEQUILIBRIUM
contribution to the total tip undercooling 25 s after the initiation of the arc. SOLIDIFICATION
due to the curvature effect is significant.2
The effect of increased undercooling Because of the rapid cooling rates
at the dendrite tip would be to solidify promote a columnar grain structure. encountered during welding, especially
at a composition closer to the overall Figure 5 shows clearly the grain growth during high-power-density processes
composition and thus reduce the extent selection process in an iridium alloy such as electron and laser-beam weld-
of microsegregation. Dendrite tip weld. Under certain conditions it is ing, it is not uncommon to observe
undercoolings in welds have been also possible to change the epitaxial nonequilibrium solidification effects.
estimated by measuring dendrite core columnar growth to equiaxed growth Most nonequilibrium features in welding
compositions for Al-Cu and Fe-Nb by inoculation or changing welding can be associated with two phenomena
systems after welding.38 For solute conditions.28,42,43 that take place as the solidification
distribution in the interdendritic regions growth velocities increase. First, the
SOLIDIFICATION OF
it may be sufficient to extend the partitioning of solute between solid and
SINGLE-CRYSTAL WELDS
solidification models for microsegrega- liquid, described by the partitioning
tion in castings to welds. This can be Studies on Fe-15Ni-15Cr single- coefficient k (= solid composition/liquid
achieved by the Schiel equation39 or crystal welds carried out during the last composition, both at the solid/liquid
modified Schiel equation that consid- ten years have advanced significantly the interface), is affected by growth rate such
ers the diffusion in the solid during fundamental understanding of weld that, as the growth velocity increases, k
welding.38,40 pool solidification.27–29 These studies deviates from the equilibrium value and
As mentioned earlier, since solidifica- have identified the effect of crystal- approaches a value of 1. Second, high
tion of the weld metal proceeds spontane- lography on the development of FZ growth velocities can lead to a change
ously by epitaxial growth of the partially microstructure. A geometrical model in the solidification mode and result in
melted grains in the base metal, the FZ has been developed that provides a nonequilibrium phase formation. It is
grain structure is mainly determined by three-dimensional relationship between noteworthy that these phenomena are
the base metal grain structure and the travel speed, solidification velocity, and closely interrelated.
welding conditions.2 Crystallographic dendrite growth velocity that predicts As discussed earlier, the solidification
effects will influence grain growth stable dendrite growth directions as a morphology also changes with growth
by favoring growth along particular function of weld pool shape and weld velocity and is influenced by the extent
crystallographic directions, namely the orientation. The regions of differently of solute partitioning and the phase that
easy growth directions.35,36,41 For cubic oriented dendrites develop because forms. In this section, nonequilibrium
metals, these easy directions are <100>. growth occurs along the preferred solute partitioning will be addressed, but
Which of these <100> directions will be <100> growth directions, and the choice even equilibrium solute partitioning can
selected, a fundamental question that is of which growth direction will prevail lead to nonequilibrium phase formation
important when welding single crystals, among the six possible variants is based because of residual microsegregation;
will be addressed later. Conditions for on the relation between weld pool shape this can be evaluated by the Scheil
growth are optimum when one of the and dendrite orientation. The model’s equation and its variants.
easy growth directions coincides with capability to predict microstructural Theories have been developed to
the heat-flow direction. Thus, among features in an Fe-15Ni-15Cr single- relate the degree of partitioning to the
the randomly oriented grains in a crystal electron beam weld made along growth rate.14 For high growth rates
polycrystalline specimen, those grains [100] on (001) plane is shown in that may be prevalent during welding,
that have one of their <100> crystal- Figure 6. reduced solute partitioning resulting
lographic axes closely aligned with Recently, these basic concepts have from a change in k can lead to a variety
heat-flow direction will be favored. been extended to commercial nickel- of effects including morphological
Without additional nucleation, this will based superalloy single-crystal technol- changes to plane front solidification,

16 JOM • June 2003

 be found even under less extreme
solidification conditions. 50 Current
research focuses on the quantitative
prediction of these transitions from
equilibrium to nonequilibrium con-
ditions by numerical modeling of
weld solidification in multicomponent
alloys.
MODELING WELD
SOLIDIFICATION
In addition to heat and fluid-flow
models used for welding, additional
modeling techniques are now avail-
able that can help describe the phase
evolution during weld solidification.
Foremost among these are computa-
tional thermodynamic models for
multicomponent systems that can predict
the primary solidification phases, the
solidification phases that may form as
a result of solute partitioning during
Figure 3. A scanning-electron micrograph showing the development
of dendrites in a nickel-based superalloy single-crystal weld. solidification, and the stability of these
phases as the weldments are cooled to
room temperature. For example, one
changes in the solidification phase, austenite solidification. such program, ThermoCalc,51 has been
and less segregation in the weld Another example of nonequilibrium used to calculate a phase diagram for
microstructure. An example is shown solidification in a low-alloy steel is a hypothetical Fe-20Cr-8Ni-xN (wt.%)
in Figure 8a, where an autogenous presented in the section on in-situ alloy as a function of temperature and
laser weld was made on a 312 stainless- observations. It is also noteworthy that chromium content for two different
steel weld overlay pad. The laser-weld the laser-welded microstructure does nitrogen concentrations, x = 0.01% and
microstructure is fully ferritic, which not show any dendritic structure; x = 0.1% (Figure 9a and b, respectively).
reflects the fact that minimal partition- this is another example of the solidifica- The plots show that at 20% chromium,
ing during solidification prevented tion morphology changing to planar for both 0.01% nitrogen and 0.1%
secondary austenite formation, as found solidification at high growth rates. nitrogen, the primary solidification will
in the weld overlay. In this case, the Extremely high growth rates are not occur by δ-ferrite. However, the phase
rapid cooling conditions during laser necessary to produce nonequilibrium stability following solidification is quite
welding also prevented solid-state solidification. A series of experiments different. In the case of the low-nitrogen
transformation of the solidified ferrite in which welds were made across stainless steel, at 800°C, a mixture of
to austenite. dissimilar stainless steels showed ferrite and austenite is expected while a
Numerous examples of nonequilib- that nonequilibrium solidification can fully austenitic structure is predicted for
rium solidification in austenitic stainless
steels have been documented over the
years.8–11,49 An example is shown in
Figure 8b. In this case, the micrograph
is of an autogenous laser weld on a
308 stainless-steel weld overlay. The
base material (weld overlay), shown
on the left, shows the typical weld
Figure 4. An optical
microstructure in this material consist- micrograph shows
ing of austenite and residual ferrite. the change in den-
This is produced by primary ferrite drite morphology from
cellular to dendritic
solidification followed by secondary as the growth velocity
austenite solidification and ferrite increases toward the
transformation to austenite during center of spot weld
(from bottom to top)
solid-state cooling. The laser-weld after the spot weld arc
microstructure is completely different. is extinguished.
It is a fully austenitic microstructure
20 µm
produced by nonequilibrium primary

2003 June • JOM 17

 the high-nitrogen alloy in equilibrium at
the same temperature. Such calculations
are simple and can be used to identify
the effect of alloy composition on the
phase stability during and after weld
solidification. Perhaps the greatest
benefit that results from these models is
that the calculations can be performed
easily for complex multicomponent
systems with ten or more constituents.
Kinetics models based on diffusion-
controlled growth can be integrated with
computational thermodynamics models
200 µm to provide valuable information on the
Figure 5. Epitaxial and columnar growth near the fusion time evolution of the microstructure.52
line in an iridium alloy electron-beam weld. The figure also For example, in the case of welding,
shows the grain-growth selection process of the grains
from the fusion line. calculations can be made to identify
the effect of cooling rate on the final
microstructure.
Such calculations were made for the
two Fe-20Cr-8Ni-xN alloys described
above. The calculations assumed a
half-dendrite arm spacing of 100 µm
and a cooling rate of 10 Ks–1. The model
considered a peritectic solidification
mode, with primary ferrite formation
and secondary austenite formation at
the ferrite/liquid interface. The results
of the calculations are shown in Figure
9c and 9d, where the phase fractions
are plotted versus time. In the case of
the high-nitrogen welds, the austenite
growth into ferrite phase was found
a 200 µm to increase rapidly after ~35 s. Thus,
Figure 6. (a) An Fe-15Cr-15Ni single- the diffusion-controlled growth models
crystal electron-beam weld made allow the calculation of the amount
along [100] direction on (001) plane,
and (b) the calculated dendritic of δ-ferrite that may be retained after
growth pattern for a similar weld solidification and the description of
orientation in (a). the weld microstructure evolution in
stainless steels to a certain extent.
These calculations can be repeated
for different weld cooling rates and
dendrite arm spacings to evaluate the
effect of welding process parameters
on the microstructure.
As noted in the previous section,
b
nonequilibrium solidification may take
place at higher cooling rates and
solidification growth rates. Recent
advances in interface-response function
models53 can be used to evaluate the
Figure 7. An optical micro- phase selection during solidification in
graph of overlapping laser multicomponent steels by coupling them
spot welds on PWA-1480
single-crystal nickel-based
with computational thermodynamic
superalloy showing the software. The interface-response func-
formation of stray grains at tion model evaluates the dendrite tip
the center of the weld.
radius, tip temperature, and partition
100 µm
coefficients as a function of interface
velocity for various competing phases

18 JOM • June 2003

and determines which solidification
phase is kinetically favored. The next
step in the modeling of weld solidifica-
tion is to couple computational thermo-
dynamic, diffusion-controlled growth
models, crystallographic geometry
models,27 and cellular automata54 models
to depict the fine details of microstructure
morphology as a function of composition
and welding process parameters.
IN-SITU OBSERVATIONS
Modeling activities must be accompa-
nied by careful experimental measure- a 40 µm
ments in order to validate the models.
Figure 8. Photo-
Traditionally, the evaluations of models micrographs of high-
have been made by post-weld character- speed laser welds
ization of solidification microstructures showing (a) fully
ferritic microstructure
using optical microscopy and analytical in type-312 stainless
electron microscopy. However, interpre- steel with negligible
tation of weld behavior by examination secondary austenite
formation and (b)
of welds at room temperature is often nonequilibrium aus-
incomplete and complicated by phase tenitic microstructure
transformations that take place upon in type-308 stainless
steel without any
cooling. There is a growing need to ferrite formation.
monitor solidification microstructure
in-situ during weld cooling. Many b 40 µm
techniques are currently available to
observe the weld solidification features
in-situ, including high-speed, high-
resolution photography on real materi-
als55 or on metal analog transparent
systems, 37 and time-resolved x-ray
diffraction (TRXRD) with synchrotron
radiation.56
Recent results from metal analog
transparent systems, combined with
detailed numerical heat transfer models
and solidification theories, led to the
identification and analysis of instabilities
at the liquid-solid interface while
welding at high speeds.37 Additional a b
work has focused on nonequilibrium
phase selection during weld solidifica-
tion in an Fe-C-Al-Mn steel by means of
in-situ observations using the TRXRD
technique. 57,58 In this research, the
equilibrium primary solidification phase
is δ-ferrite and this was confirmed by
TRXRD measurements on slowly cooled
spot welds. However, under rapid
cooling conditions, the TRXRD mea- c d
surements showed the formation of
primary austenite (Figure 10). Research Figure 9. Quasi-binary diagrams showing liquid, austenite, and δ-ferrite phase regions
in stainless steels has shown that it is in Fe-Cr-Ni alloy systems with (a) 0.01 wt.% nitrogen and (b) 0.1 wt.% nitrogen. The
calculated variation of phase fraction as a function of cooling time from 1,750 K using a
possible to form nonequilibrium primary diffusion-controlled growth model for Fe-Cr-Ni alloy systems with (c) 0.01 wt.% nitrogen
austenite under rapid solidification and (d) 0.1 wt.% nitrogen.
conditions but this is the first time such
a phenomenon was observed in a

2003 June • JOM 19

 28. S.A. David et al., Metall. Trans. A, 21A, (1990), p.
1753.
29. J.M. Vitek et al., Int’l Trends in Welding Science and
Tech., ed. S.A. David and J.M. Vitek (Materials Park,
OH: ASM Int., 1993), p. 167.
30. J.M. Vitek et al., Sci Technol. Weld. Joining, 6 (5)
(2001), p. 305.
31. B. Radhakrishnan and T. Zacharia, Modeling and
Control of Joining Processes, ed. T. Zacharia (Miami,
FL: Am. Weld. Soc., 1994), p. 298.
32. Ch.-A. Gandin, M. Rappaz, and R. Tintillier, Metall.
Trans., 24A (1993), p. 467.
33. W.B. Dress, T. Zacharia, and B. Radhakrishnan,
Modeling and Control of Joining Processes, ed. T.
Zacharia (Miami, FL: Am. Weld. Soc., 1994), p. 321.
34. S.S. Babu et al., Mater. Sci. Technol., 11 (1995), p.
186.
35. M.C. Flemings, Solidification Processing (New York:
McGraw Hill, 1974).
36. W. Kurz and D.J. Fisher, Fundamentals of
Solidification (Aedermannsdorf, Switzerland: Trans-
Tech. Publications, 1986).
37. R. Trivedi et al., J. Appl. Phys., 93 April (2003) p.
4,885.
38. J.A. Brooks and M.I. Baskes, Advances in Welding
Science and Technology, ed. S.A. David (Materials
Park, OH: ASM Int., 1986), p. 93.
39. E. Scheil, Z. Metall, 34 (1942), p 70.
Figure 10. An image representation of time-resolved x-ray diffraction data that shows the 40. T. Matsumiya et al., Nippon Steel Technical Report
formation of primary austenite (fcc) from liquid during rapid cooling. 57 (1993), p. 50.
41. W.F. Savage, C.D. Lundin, and A. Aronson, Weld J,
44 (1965), p. 175.
42. T. Ganaha, B.P. Pearce, and H.W. Kerr, Metall.
Trans, 11A (1980), p. 1351.
low-alloy steel. In these steels, in-situ p. 185. 43. H.W. Kerr and J.C. Villefuerta, Metal. Sci. of Joining,
measurements are particularly valuable 8. J.M. Vitek, A. DasGupta, and S.A. David, Metall. ed. K.J. Cieslak et al. (Warrendale, PA: TMS, 1991), p.
Trans., 14A (1983), p. 1833. 11.
since behavior at elevated temperatures 9. S.A. David, J.M. Vitek, and T.L. Hebble, Weld J., 44. J.M. Vitek et al., to be published in proceedings of
is masked by subsequent solid-state 66 (1987), p. 289s. Thermec 2003, (Switzerland: Trans Tech Publishers)
transformation of ferrite to austenite and 10. S. Katayama and A. Matsunawa, Proc. of Int’l Madrid, Spain.
Congress on Application of Lasers and Electro Optics, 45. S.A. David et al., Sci. Technol. Weld. Joining, 2 (2)
austenite to martensite. Time-resolved vol. 4 (Boston, MA: Laser Institute of America, (1997), p. 79.
x-ray diffraction measurements have 1984), p. 60. 46. J.M. Vitek, S.A. David, and L.A. Boatner, Sci. Technol.
proven to be ideal for identifying 11. J.W. Elmer, S.M. Allen, and T.W. Eager, Metall. Weld. Joining, 2 (3) (1997), p. 109.
Trans. 20A (1989), p. 2117. 47. M. Gäumann, R. Trivedi, and W. Kurz; Mater. Sci,
competing phase-transformation mecha- 12. R. Trivedi and W. Kurz, Acta Metall. 34 (1986), Eng., A 226-228 (1997), p. 763.
nisms under nonequilibrium weld- p. 1663. 48. T.M. Pollock and W.H. Murphy, Metall. Mater. Trans.,
cooling conditions. This technique has 13. W.J. Boettinger and S.R. Coriell, Mater. Sci. Eng., 27A (1996), p. 1081.
65 (1984), p. 27. 49. J.M. Vitek and S.A. David, Laser Materials
been applied to other alloy systems and 14. M.J. Aziz, J. Appl. Phys., 53 (1982), p. 1158. Processing IV, ed. J. Mazumder, K. Mukherjee, and B.L.
exciting new insight into issues relating 15. S.A. David and T. DebRoy, Science, 257 (1992), Mordike (Warrendale, PA: TMS 1994), p. 153.
to weld solidification issues is being p. 497. 50. H.K.D.H. Bhadeshia, S.A. David, and J.M. Vitek,
16. T. DebRoy and S.A. David, Reviews of Modern Mater. Sci. Technol. 7 (1991) p. 50.
achieved.59 Physics, 67 (1) (1995), p. 85. 51. B. Sundman, B. Jansson, and J.O. Andersson,
17. S.A. David and T. DebRoy, MRS Bulletin, XIX Calphad, 9 (1985) p. 1.
ACKNOWLEDGEMENTS (1) (1994), p. 29. 52. J. Agren, ISIJ International, 32 (1992), p. 291.
18. T. Zacharia et al., Metall. Trans., 20A (1989), 53. S. Fukumoto and W. Kurz, ISIJ International, 38
This research is sponsored by the p.957. (1998), p. 71.
Division of Materials Sciences and 19. K. Hong, D.C. Weckman, and A.B. Strong, Trends 54. U. Dilthey, V. Pavlik, and T. Reichel, Mathematical
Engineering, Office of Basic Energy in Welding Research, ed. H.B. Smartt, J.A. Johnson, Modeling of Weld Phenomena 3, ed. H. Cerjak (London:
and S.A. David (Materials Park, OH: ASM Int., 1996), Institute of Materials, 1997) p. 85.
Sciences, U.S. Department of Energy, p. 399. 55. A.C. Hall et al., Proceedings of the 11th International
under Contract DE-AC05-00OR22725 20. R.T.C. Choo and J. Szekely, Weld J., 73 (1994), Conference and Exhibition on Computer Technology in
with UT-Battelle, LLC. p. 255. Welding (Columbus, Ohio, 2001).
21. W. Pitscheneder et al., Weld J, 75 (3) (1996), 56. J.W. Elmer, J. Wong, and T. Ressler, Metall. Mater.
References p. 71s. Trans., 29A (1998) p. 276.
22. Y. Dong et al., Weld J., 76 (10) (1997), p. 442s. 57. S.S. Babu et al., J. Proc. Roy. Soc. A., 458 (2002),
1. W.F. Savage, Welding World, 18 (1980), p. 89. 23. J. Goldak et al., Mathematical Modeling of Weld p. 811.
2. S.A. David and J.M. Vitek, Inter. Mater. Review, Phenomena 3, ed. H. Cerjak (London: Institute of 58. S.S. Babu et al., Acta Materialia, 50 (2002), p. 4763.
34 (5) (1989), p. 213. Materials, 1997), p. 543. 59. J.W. Elmer, T. Palmer, and S.S. Babu, Adv. Mater.
3. G.J. Davies and J.G. Garland, Inter. Mater. Review, 24. T. DebRoy, Proceedings of the Julian Szekely Proc.,160 (2002) p. 23.
20 (1975), p. 83. Symposium on Materials Processing, ed. H.Y. Sohn,
S.A. David, S.S. Babu, and J.M. Vitek are with Oak
4. F. Matsuda, T. Hashimoto, and T. Senda, Trans. Natl. J.W. Evans, and D. Apelian (Warrendale, PA: TMS,
Ridge National Laboratory.
Res. Inst. Met (JPN), 11 (1) (1969), p. 83. 1997), p. 365.
5. K.E. Easterling, Introduction to Physical Metallurgy 25. T. DebRoy et al., Mathematical Modeling of Weld For more information, contact S.A. David, Oak Ridge
of Welding (London: Butterworths, 1983). Phenomena 6, ed. H. Cerjak (London: Institute of National Laboratory, Metals & Ceramics Division,
6. S. Kou, Welding Metallurgy, Second edition (New Materials, 2002), p. 21. Building 4508, MS 6095, Oak Ridge, Tennessee
York: John Wiley & Sons, Inc., 2002). 26. S.A. David and C.T. Liu, Weld J., 61 (1982), p. 157s. 37831-6095; (865) 574-4804; fax (865) 574-4928;
7. R. Mehrabian, Inter. Metall. Review, 27 (1982), 27. M. Rappaz et al., Metall. Trans. 20A (1989), p. 1125. e-mail [email protected].

20 JOM • June 2003