See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/260038120

Fundamentals of Solidification

Chapter · February 2004

DOI: 10.31399/asm.hb.v09.a0003724

CITATIONS READS
219 25,337

2 authors:

Doru M. Stefanescu Roxana Ruxanda

The Ohio State University Copeland
425 PUBLICATIONS 7,532 CITATIONS 27 PUBLICATIONS 489 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Doru M. Stefanescu on 12 February 2014.

The user has requested enhancement of the downloaded file.

Name /bam_asmint_106420/6044_002d1/Mp_1 08/09/2004 08:20AM Plate # 0 pg 1 #

Fundamentals of Solidification
Doru M. Stefanescu and Roxana Ruxanda, The University of Alabama

SOLIDIFICATION PROCESSING is one of scribe the results of computer models. An ex- ● The microscale (microstructure) is of the or-
the oldest manufacturing processes as it is the ample of a solidification mesoscale structure der of 10ⳮ6 to 10ⳮ5 m. The microscale de-
principal component of metal casting process- is given in Fig. 3. scribes the complex morphology of the solid-
ing. While solidification science evolved from
the need to better understand and further develop
casting processes, in 2004 solidification science
is at the base of many new developments that
fall out of the realm of traditional metal casting.
Solidification is, strictly speaking, the trans-
formation of liquid matter into solid matter. The
microstructure that results from solidification
may be the final one, in which case it directly
affects the mechanical properties of the product.
In other cases, heat treatment or other processes
may be used after solidification to further modify
the solidification microstructure. However, the
outcome of this additional processing will be
greatly affected by the solidification microstruc-
ture.

Length Scale of
Solidification Structures Fig. 1 Solidification length scale. Source: Ref 1

The effect of solidification on the morphology

of the matrix can be deconstructed at four dif-
ferent length scales (Ref 1) (Fig. 1):
● The macroscale (macrostructure) is of the or-
der of 10ⳮ3 to 1 m. Elements of the macro-
scale include shrinkage cavity, macrosegre-
gation, cracks, surface roughness (finish), and
casting dimensions. A typical example of a
solidification macrostructure is given in Fig.
2. Columnar grains growing inward into the
casting are seen.
● The mesoscale is of the order of 10ⳮ4 m. It
allows description of the microstructure fea-
tures at the grain level, without resolving the
intricacies of the grain structure. As seen in
Fig. 1, the solid/liquid (S/L) interface is not
sharp. Three regions can be observed: liquid,
mushy (containing both liquid and solid
grains), and solid. Mechanical properties are
affected by the solidification structure at the
mesoscale level, which is described by fea-
tures such as grain size and type (columnar
or equiaxed), the type and concentration of
chemical microsegregation, and the amount
of microshrinkage, porosity, and inclusions.
The term “mesoscale” has been introduced in
solidification science to more accurately de- Fig. 2 Macrostructure of a cast Ti-6Al-4V alloy specimen. Etchant: Keller’s reagent
Name /bam_asmint_106420/6044_002d1/Mp_2 08/09/2004 08:20AM Plate # 0 pg 2 #
2 / Metallurgy and Microstructure

ification grain. In a sound casting, mechanical solidification are discussed in terms of the the system. Under such conditions, no changes
properties depend on the solidification struc- transfer of individual atoms from the liquid occur with time. When global equilibrium exists,
ture at the microscale level. To evaluate the to the solid state. Features such as disloca- the fraction of phases can be calculated with the
influence of solidification on the properties of tions and individual atoms are observed with lever rule, and the phase diagram gives the uni-
the castings, it is necessary to know the as- electronic microscopes. An example of dis- form composition of the liquid and solid phases.
cast grain morphology (i.e., size and type, co- locations seen at nanoscale magnification is Such conditions exist only when the solidifica-
lumnar or equiaxed) and the length scale of given in Fig. 5. tion velocity is much smaller than the diffusion
the microstructure (interphase spacing, e.g., velocity. Uniform chemical potentials and tem-
As is discussed in some detail in the following
dendrite arm spacing and eutectic lamellar perature may truly appear only when solidifica-
sections, two basic phenomena must take place
spacing). The term microstructure is the clas- tion takes place over geological times.
in the liquid for solidification to occur: under-
sic term used in metallography to describe Solidification as encountered in common pro-
cooling and nucleation. If these conditions are
features observed under the microscope, as cesses does not occur at equilibrium, since dur-
met, the nuclei can grow into the new solid
seen in the electron micrograph from Fig. 4, ing solidification of most castings, both tem-
grains.
which shows the etched surface of a pure alu- perature and composition gradients exist across
minum sample (99.99 % Al). Etching re- the casting. Elementary thermodynamics dem-
moved the layers of atoms, one after the other, Undercooling onstrates that a liquid cannot solidify unless
preferentially from crystallographic planes some undercooling below the equilibrium (melt-
parallel to the faces of the elementary cube ing) temperature, Te, occurs. Five types of solid-
Global equilibrium phase diagrams are fre-
(Ref 2). ification undercooling have been identified: ki-
quently used to understand alloy behavior when
● The nanoscale (atomic scale) is of the order netic undercooling, thermal undercooling,
the alloy is cooled from the liquid state to room
of 10ⳮ9 m (nanometers) and describes the constitutional (solutal) undercooling, curvature
temperature. Global equilibrium requires uni-
atomic morphology of the S/L interface. At undercooling, and pressure undercooling.
form chemical potentials and temperature across
this scale, nucleation and growth kinetics of

Fig. 3 Cast aluminum-silicon alloy. Etchant: 60% HCl,

Fig. 5 Cellular dislocation arrangement in direct chill
30% HNO3, 5%HF, 5%H2O cast 99.0% Al. Dislocation lines in the as-cast
structure. Transmission electron micrograph. 10,000⳯.
Source: Ref 2
Fig. 7 Formation of a constitutionally undercooled re-
gion (dashed area) in the liquid next to the solid/
liquid interface because of the lower liquidus temperature
produced by the higher solute content

Fig. 6 Left corner of a phase diagram. CE, eutectic com-

position; CL, composition of liquid; CS, compo-
Fig. 4 Microstructure (electron micrograph) of a pref- sition of solid; CSM, maximum solubility in solid; T *, inter- Fig. 8 Equiaxed grain accumulation on a steel sieve in-
erentially etched surface of a pure aluminum face temperature; TL, liquidus temperature; TS, solidus serted in a solidifying Al-2%Cu alloy. Source:
sample (99.99% Al). Etchant not reported. Source: Ref 2 temperature Ref 3
Name /bam_asmint_106420/6044_002d1/Mp_3 08/09/2004 08:20AM Plate # 0 pg 3 #
Fundamentals of Solidification / 3

Nevertheless, in most cases, the overall solid- ically, for solute trapping the solidification ve- coefficient. For the case in the figure (where the
ification kinetics can be described with sufficient locity must exceed 5 m/s. equilibrium temperatures decrease with in-
accuracy by using the local equilibrium condi- Kinetic Undercooling. When a number of creased alloy composition) k  1. The slope of
tion, that is, by using the mass, energy, and spe- simplifying assumptions are introduced (pure the liquidus line is mL ⳱ dTL/dCL, where TL is
cies transport equations to express the tempera- metal, constant pressure, no thermal gradient in the liquidus temperature.
ture and composition variation within each phase the liquid, and flat S/L interface—that is, the ra- Because of solute rejection, a boundary layer,
and by using equilibrium phase diagrams to dius of curvature of the interface is r ⳱
), the richer in solute than the bulk liquid, is formed at
evaluate the temperature and composition of only undercooling driving the S/L interface is the the S/L interface. A direct consequence of this
phase boundaries, such as the S/L interface (cor- kinetic undercooling. It is a nanoscale length ef- phenomenon is that the liquidus temperature is
rections must be made for interface curvature). fect, resulting from the net difference in atoms lower next to the interface than away from it
Most phase transformations, with the exception transported from L to S and from S to L. Typi- (Fig. 7). The heat flow from the liquid to the
of massive (partitionless) and martensitic trans- cally for metals, the kinetic undercooling is very solid imposes a thermal gradient, GT, which dic-
formations can be described with the local equi- small, of the order of 0.01 to 0.05 K. tates the local temperature in the melt. If the ther-
librium condition. When the stable phase cannot When the simplifying assumptions are relaxed mal gradient is lower than the liquidus gradient,
nucleate or grow sufficiently fast (e.g., gray-to- to reflect typical solidification scenarios, upon GL (the tangent to TL at the S/L interface), which
white transition in cast iron), metastable local the solidification of a discrete volume of liquid is the case shown in Fig. 7, the temperature in
equilibrium can occur. For both stable and meta- the free energy of the liquid-solid system will the boundary layer will be lower than the equi-
stable local equilibrium, the chemical potentials increase by (Ref 1): librium liquidus temperature. Thus, a constitu-
of the components across the interface must be tionally undercooled region will result. The size
equal for the liquid and for the solid. DFv ⳱ DGT Ⳮ DGc Ⳮ DGr Ⳮ DFP (Eq 1) of the local constitutional (solutal) undercooling
However, at large undercooling the solidifi- can be calculated as a function of the local com-
cation velocity exceeds the diffusive speed of position as:
where F and G are the Helmholtz and Gibbs free
solute atoms in the liquid phase (rapid solidifi-
energy, respectively. The four right-hand terms
cation). The solute is trapped into the solid at
levels exceeding the equilibrium solubility. Typ-
are the change in free energy because of tem-
perature, composition, curvature, and pressure
(
∆Tc = TL − T * = − mL C*L − C0 ) (Eq 3)
variation, respectively. Solidification cannot oc-
where DTc is the constitutional undercooling,
cur unless each of these energies is balanced by
C*L is the composition of the liquid, C0 is the bulk
a corresponding undercooling of the system, as
composition of the alloy at the beginning of so-
discussed in this section.
lidification, and the other terms are as previously
Thermal Undercooling. If nucleation does
defined.
not occur, a pure metal can undercool under the
Curvature Undercooling. When a discrete
equilibrium temperature because of heat extrac-
liquid volume becomes solid, the newly formed
tion. The liquid is said to be thermally under-
S/L interface produces an increase in the energy
cooled by a quantity:
of the system through the interface energy as-
sociated with it. For solidification to continue,
DTT ⳱ Te ⳮ T * (Eq 2) additional undercooling is required. This under-
cooling is called the curvature undercooling.
where DTT is the thermal undercooling, Te is the Since a smaller radius (higher curvature) of the
equilibrium (melting) temperature of the flat in- solid results in a higher surface-to-volume ratio,
terface, and T * is the interface temperature. higher curvature is associated with higher un-
Constitutional (Solutal) Undercooling. dercooling. Thus:
During alloy solidification, solute is rejected by
the solid. This can be understood from the phase γ
∆Tr = Te − Ter = K = ΓK
diagram in Fig. 6. For a given temperature, T *, ∆Sf (Eq 4)
the composition of the solid, CS, is smaller than
Fig. 9 Broken dendrite branches transported in the cen-
that of the liquid, CL, in equilibrium with the where DTr is the curvature undercooling, T re is
ter of the ingot by liquid convection in an am-
monium-chloride/water system. Source: Ref 4 solid. The ratio k ⳱ CS/CL is called the partition the equilibrium temperature for a sphere of ra-

Fig. 10 Nucleation and coalescence of eutectic grains in cast iron. (a) Early solidification. (b) Late solidification. (c) After solidification at room temperature. Source: Ref 5
Name /bam_asmint_106420/6044_002d1/Mp_4 08/09/2004 08:20AM Plate # 0 pg 4 #
4 / Metallurgy and Microstructure

dius r, c is the S/L interface energy, DSf is the The initial chilling action of the mold may in- terface. During growth, local perturbations (in-
entropy of fusion, K ⳱ 1/r is the curvature, and duce local solid grain formation. These grains stabilities) will form at the S/L interface. If the
C is the Gibbs-Thomson coefficient. are then carried into the bulk by fluid flow and perturbations cannot survive, the interface re-
Pressure Undercooling. An increase in the survive and grow in the undercooled liquid. This mains planar (Fig. 11a). If on the contrary they
pressure applied to the liquid, DP, will cause an is called the big bang mechanism. Experimental become stable, they will continue to develop and
increased undercooling of the liquid, called pres- evidence for this mechanism is provided in Fig. different interface morphologies can result (Fig.
sure undercooling, DTP. It can be calculated as: 8, which shows the results of experiments with 11b–d). In pure metals, the only source of grow-
a steel net inserted in the middle of a crucible ing interface instabilities is the thermal field
∆P ∆v where an Al-2%Cu alloy was allowed to solidify (thermal undercooling). In alloys, both thermal
∆TP =
∆Sf (Eq 5) (Ref 3). In the lower part, only columnar grains and solutal instabilities may grow, as a result of
are formed. In the upper part, a large number of thermal and solutal undercooling.
where Dv is the change in volume. For metals small equiaxed grains are seen in the vicinity of When the thermal gradient in the liquid at the
the pressure undercooling is rather small, of the the sieve. It is argued that these grains have orig- S/L interface is less than the liquidus tempera-
order of 10ⳮ2 K/atm. Therefore, typical pressure inated at the mold wall, then have been carried ture gradient—that is, GT  GL (see Fig. 7)—
changes for usual processes have little influence by convection currents in the middle of the ingot, the liquid at the interface (T *) is at a lower tem-
on the melting temperature and the pressure un- and have been prevented from sinking by the perature than its liquidus (TL). This liquid is
dercooling is negligible. sieve. constitutionally undercooled. Instabilities grow-
Undercooling is a basic condition required for Another possible mechanism for dynamic nu- ing in this region will become stable, because
solidification. However, for solidification to oc- cleation is the fragmentation of existing crystals they will find themselves at a temperature lower
cur nuclei must form in the liquid. through ripening and local remelting of colum- than their equilibrium temperature. They will
nar dendrites. The detached dendrite arms are continue to grow. On the contrary, if GT  GL,
carried to the center of the mold by convection the interface will remain planar (Fig. 11a).
Nucleation currents (Fig. 9) (Ref 4). If the center of the mold For small constitutional undercooling, the in-
is undercooled, these crystals act as nuclei for stabilities will only grow in the solidification di-
It is convenient to classify the types of nuclei equiaxed grains. rection (the x-direction), and a cellular interface
available in the melt as resulting from homoge- The number of grains identified by metallo- will result (Fig. 11b and c). This is shown in Fig.
neous nucleation, heterogeneous nucleation, and graphic techniques at the end of solidification is 12. The planar-to-cellular transition occurs at a
dynamic nucleation (Ref 1). not necessarily equal to the number of initial nu- gradient Gp/c. As the constitutional undercooling
Homogeneous nucleation, which implies clei. Some nuclei may redissolve because of in- increases because of the lower thermal gradient,
that growth is initiated on substrates having the sufficient undercooling, while others may grow the spacing between the cells increases, and con-
same chemistry as the solid, is not common in and coalesce, thus decreasing the final number stitutional undercooling may also occur perpen-
casting alloys. of grains. As demonstrated through liquid dicular to the growth direction (in the y-direc-
Heterogeneous nucleation is based on the quenching experiments on gray cast iron (Ref 5), tion). Instabilities will develop on the sides of
assumption that the development of the grain the final eutectic volumetric grain density was the cells, resulting in the formation of dendrites
structure occurs upon a family of substrates of found to be as much as 27% less than the max- (Fig. 11d). This is the cellular-to-dendrite tran-
chemistry different than that of the solid. Het- imum number of grains developed during solid- sition. It takes place at a temperature gradient
erogeneous nucleation occurs when substrate ification (Fig. 10). Gc/d. Both cellular and dendritic growth occur-
particles are deliberately introduced into a melt ring from the wall in the direction opposite to
to promote equiaxed grain formation. This is the heat transport can be described as columnar
common practice in liquid processing of metallic Growth and Interface Stability growth.
alloys. If constitutional undercooling is greater,
Dynamic nucleation occurs because of the The thermal and compositional fields at the S/ equiaxed grains can be nucleated in the liquid
action of convective currents within the melt. L interface determine the morphology of the in- away from the interface. The dendritic-to-

Fig. 11 The change of the morphology of the solid/liquid interface as a function of growth velocity (V) in a transparent organic system (pivalic acid, 0.076% ethanol) directionally
solidified under a thermal gradient of 2.98 K/mm. (a) Planar interface, V ⳱ 0.2 lm/s. (b) Cellular interface, V ⳱ 1.0 lm/s. (c) Cellular interface, V ⳱ 3.0 lm/s. (d) Dendritic
interface, V ⳱ 7.0 lm/s. Source: Ref 4
Name /bam_asmint_106420/6044_002d1/Mp_5 08/09/2004 08:20AM Plate # 0 pg 5 #
Fundamentals of Solidification / 5

equiaxed transition occurs at Gd/e. If the thermal Basic Solidification square centimeter. These dislocations result from
gradient is almost flat—that is, GT ⳱ 0—the Structures of Pure Metals stresses induced during growth by thermal, me-
driving force for the columnar front will be ex- chanical, and composition gradients, as well as
tremely small. A complete equiaxed structure is from entrapped particles. In addition, vacancies
expected. Pure metals can solidify with planar interfaces can condense to form small dislocation loops
All the transitions described in the previous (single crystals or polycrystals with columnar subsequent to growth (Ref 6).
paragraph are controlled by the initial composi- front) or with equiaxed structure. To obtain sin- Dislocations present in a metal crystal often
tion of the alloy, by the solidification velocity gle crystals the following methods are used (Ref polygonize into subboundaries during growth.
(V), and by the thermal gradient. The planar-to- 1): normal freezing (the boat method or the These subboundaries produce subgrains during
cellular transition during steady-state solidifica- Bridgman method), crystal pulling (Czochral- solidification. The subboundaries are elongated
tion is governed by the criterion for the onset of ski), or zone melting and zone freezing (with in the direction of the growth and can be irreg-
constitutional undercooling given by: crucible or crucibles). High-temperature gradi- ular if the material is pure, or are regular and
ents and low-solidification velocities are re- straight in a very dilute alloy in which cellular
GT m C (1 − k ) ∆T0 quired to obtain a planar interface. growth has occurred (see Fig. 1 and 2 in the ar-
<− L 0 = When casting a pure metal in a mold, the ini-
V kDL DL (Eq 6) ticle “Solidification Structures of Pure Metals”
tial shape and size of the grains are determined in this Volume). Subboundaries also are formed
by nucleation and growth during solidification, where the liquid between two slightly misori-
where k is the partition coefficient, DL is the liq-
but they may be altered significantly by grain ented dendrite arms solidifies (Ref 2). Disloca-
uid diffusivity, and DT0 is the temperature dif-
growth after solidification. Two solidification tions in subboundaries can be resolved by careful
ference between the equilibrium liquidus and
cases may be considered: starting with a melt etching. Dislocations produced by precipitation
solidus of the alloy of composition C0. If solid-
having low superheating or with a melt having and condensation of vacancies during cooling
ification is not at steady state, DT0 is substituted
high superheating. usually are in the form of small loops.
with the undercooling DT at which solidification
When the metal is poured at low superheat, Equiaxed Microstructures. After nucleation,
occurs.
nucleation will occur on the cold mold wall. A each equiaxed dendrite will grow independently
thin region of small equiaxed dendrites will be until it comes in contact with and impinges on
produced in the undercooled zone adjacent to the adjacent growing dendrites. At this stage den-
wall. Then, a thermal gradient from the melt to drite coherency has been reached, and the initial
the mold will be established and solidification shape of the grains is formed. When all the in-
will proceed inward from the mold wall with an terdendritic liquid has been solidified, the solid-
isothermal front. The grains will elongate in the ification grain boundaries are established. These
direction of heat flow, resulting in a columnar boundaries will move during subsequent cooling
polycrystalline grain structure. As growth pro- in solid state when grain growth occurs to min-
ceeds, some crystal orientations will tend to per- imize free energy.
sist at the expense of others, resulting in a pre- Although the grain boundaries in high-purity
ferred orientation texture. metals are very mobile, they may be slowed
For a high superheat, the mold walls may be down or pinned by small concentrations of im-
heated above the melting point during pouring. purities. If the initial grain size was large, grain
Then, as undercooling occurs, nucleation may growth is less likely, but there will usually be
start in the bulk liquid producing equiaxed den- some adjustment of the grain boundaries to
dritic grains. Finer equiaxed grains can be pro- lower energy configurations.
duced by promoting heterogeneous nucleation The energy of a grain boundary increases with
through the addition of nucleating agents. the orientation mismatch between the grains, as
Single Crystals. Most metal single crystals shown in Fig. 13. Small-angle boundaries (up to
have dislocation densities of about 106 to 107 per 10), consisting of dislocation arrays, have much

Fig. 12 Correlation between the thermal gradient at the

interface and the interface morphology. Fig. 13 Definition of orientation mismatch between grains and variation of energy associated with the boundary as a
Source: Ref 1 function of the tilt angle
Name /bam_asmint_106420/6044_002d1/Mp_6 08/09/2004 08:20AM Plate # 0 pg 6 #
6 / Metallurgy and Microstructure

lower energies than large-angle boundaries. The other, as shown in Fig. 14. In Fig. 15, the junc- may be completed by some other process, such
dislocation arrays in tilt and twist boundaries are tion of a small-angle boundary (low energy) with as a eutectic (Fig. 16b), peritectic (Fig. 16c), or
illustrated in Fig. 11 and 12 in the article “So- two large-angle boundaries (high energy) is monotectic reaction (Fig. 16d). The details of the
lidification Structures of Pure Metals” in this shown. The ratios of the boundary energies can solidification of the solid solution are important
Volume. be calculated from the angles. in determining the final microstructure and thus
Most boundaries between grains will be large- Coincidence boundaries usually have lower the properties.
angle boundaries and will have more than 10 energies than those of large-angle boundaries. A As shown in Fig. 6, the liquid has a different
misorientation. Except twin or coincidence boundary that separates twins (two crystals ori- composition (CL) than the solid (CS) from which
boundaries, these boundaries have roughly the ented in a special crystallographic arrangement it is forming (in the figure CL  CS). Conse-
same energy, so when equilibrated at the junction in which they are related by mirror symmetry) quences of this phenomenon are the occurrence
of three grains they form 120 angles with each has low energy if it is coherent, that is, if it lies of constitutional undercooling and segregation.
in the mirror plane. Twin boundaries can be pro- Constitutional undercooling is instrumental in
duced during crystal growth or annealing. Small- destabilizing the S/L interface and promoting in-
angle boundaries and individual dislocations are terface morphologies different than planar. As
usually present within the grains of a polycrystal, inferred by Eq 6, there is a critical solute content
just as in a single crystal (Ref 6). (C0) of the alloy for a given GT/V ratio combi-
nation, at which the interface becomes unstable.
This can be presented graphically as shown in
Solidification Structures of Fig. 17, where the line for Eq 6 indicates the
Solid Solutions planar-to-cellular transition. As the GT/V ratio
continues to decreases (or C0 to increase), the S/
Alloys consist of a base metal, such as iron, L interface becomes increasingly unstable with
aluminum, copper, or nickel, to which other ele- successive formation of a columnar dendritic and
ments (solutes) are added to yield desired prop- then equiaxed dendritic structure.
erties. The simplest liquid-to-solid transforma- The formation of the equiaxed dendritic struc-
tion (solidification) occurs when the liquid ture requires bulk nucleation. In the absence of
solution transforms into a solid solution (Fig. bulk nucleation, the columnar front will continue
16a). However, for many alloys solidification to grow, or, alternatively, at high undercoolings

Fig. 14 Secondary electron micrograph of grain

boundaries in polycrystalline iron. Angles at
the indicated triple junction of the grain boundaries: 105,
125, and 130

Fig. 15 Secondary electron micrograph of low-energy

and high-energy grain boundaries in polycrys- Fig. 16 Binary phase diagrams. L, L1, and L2 are liquid solutions. ␣ and b are solid solutions. (a) Complete solid
talline iron. Angles at the indicated triple junction of the solubility. (b) Partial solid solubility with eutectic reaction. (c) Partial solid solubility with peritectic reaction.
grain boundaries: 92, 125, and 145 (junction at arrow). (d) Partial solid solubility with monotectic reaction
Name /bam_asmint_106420/6044_002d1/Mp_7 08/09/2004 08:20AM Plate # 0 pg 7 #
Fundamentals of Solidification / 7

amorphous solidification may occur (i.e., solid- V is much smaller than the ratio DT/DL, the in- plex, three-dimensional structures, plane-section
ification without crystallization, also termed terface becomes columnar dendritic, as shown in micrographs must be interpreted carefully. Fig-
glass formation). Fig. 11(d). The regions between the cells and ure 20 shows the structure of cobalt dendrites
Planar Interfaces. Planar growth of alloys dendrites, which are still liquid in the micro- when the material surrounding the dendrites has
can usually be achieved only in crystal growth graphs, are enriched in solute and produce mi- been removed by selective deep etching. Figure
furnaces at high-temperature gradients and low crosegregation at the end of solidification. 21 shows the same structure observed in a plane
solidification velocities. For example, for planar Cellular structures are most often observed in section. Parts of a single dendrite often appear
solidification of an alloy with DT ⳱ 5 K and GT dilute alloys where DT is small. Cellular struc- disconnected when viewed in a plane section.
⳱ 100 K/cm, the maximum allowable solidifi- tures can take on three characteristic morpholo- The high degree of segregation present in this
cation velocity calculated with Eq 6 is 2 lm/s. gies when viewed in a metallographic section structure is revealed by the elemental mapping
Figure 18 shows a dilute tin-cadmium alloy transverse to the growth direction: nodes (Fig. of cobalt and copper, as shown in Fig. 22.
solidified at high-temperature gradient and low- 19a), elongated cells (Fig. 19b), and hexagonal Effect of Crystallographic Orientation. Den-
solidification speed to achieve planar growth cells (Fig. 19c). The hexagonal cellular structure drites are single grains that have preferred
(Ref 7). Planar growth of doped semiconductors is observed much more frequently than the node growth directions. The morphology of a colum-
constitutes the basis of an entire industry. How- or elongated cell structures (Ref 7). nar dendrite is influenced by the orientation of
ever, most metallurgical alloys solidify with non- Dendritic Structures. The dendritic mor- the grain with respect to that of heat extraction,
planar interfaces. phology is the most commonly observed solidi- as shown in Fig. 23, where the heat-extraction
Cellular Structures. When constitutional un- fication structure. Because dendrites are com- direction is upward (Ref 8).
dercooling occurs, the S/L interface morphology Influence of the Type of Phase Diagram. The
becomes cellular or dendritic. For conditions of nature of the material as represented by the type
growth where the GT/V ratio is only slightly of phase diagram will also influence the dendritic
smaller than the ratio DT/DL, the interface is cel- structures. If the phase diagram shows complete
lular, as shown in Fig. 11(c) for a transparent solid solubility (Fig. 16a), the structure will be
organic “alloy” that solidifies like a metallic al- single phase, containing only dendrites. If the
loy. For conditions of growth where the ratio GT/ phase diagram contains a eutectic (Fig. 16b), the
interdendritic regions will be composed of the
two-phase eutectic. Figure 24 presents a low-
magnification microstructure of a Ni-25Cu
(at.%) alloy. The microstructure appears to be
fully dendritic. However, at higher magnification
a large volume fraction of interdendritic eutectic
is apparent (Fig. 25). For alloys where primary
solidification is followed by a peritectic reaction,
the microstructure depends strongly on solid dif-
fusion rates. When this diffusion is slow, the
dendrites are coated by the peritectic phase.
Effect of Constitutional Undercooling. As
shown in Fig. 17, as the amount of solute in-
creases, or as the GT/V ratio decreases, a cellular-
Fig. 18 Directionally solidified Sn-0.6Cd alloy. Section to-dendritic solidification occurs. This is because
parallel to the growth direction shows a the constitutional undercooling is large. An ex-
quenched planar solid/liquid interface. G ⳱ 320 K/cm, V
⳱ 0.85 lm/s, DT ⳱ 5.7 K. Etchant: 5 mL HNO3, 95 mL
ample of the cellular-to-dendritic transition pro-
Fig. 17 Transition to different interface morphologies
duced by higher solidification velocity is pre-
as a function of the temperature-gradient/solid- lactic acid. Magnification: 80⳯. Courtesy of C. Brady.
ification-velocity ratio (GT/V) and solute concentration (C0) Source: Ref 7 sented in Fig. 26 (Ref 9).

Fig. 19 Sn-0.05Pb alloy; liquid decanted to reveal structures of solid/liquid interfaces. (a) Oblique illumination reveals nodes at the solid/liquid interface. (b) Elongated cells at the
solid/liquid interface under bright-field illumination. (c) Fully developed hexagonal cells at the solid/liquid interface (bright-field illumination). Unpolished, unetched.
Magnification: 150⳯. Source: Ref 7
Name /bam_asmint_106420/6044_002d1/Mp_8 08/09/2004 08:20AM Plate # 0 pg 8 #
8 / Metallurgy and Microstructure

Fig. 20 Secondary electron micrograph of Cu-10Co Fig. 21 Secondary electron micrograph of Cu-10Co
(at.%) alloy casting. Etchant not reported. Mag- (at.%) alloy casting. As-polished. Magnifica-
nification: 150⳯. Source: Ref 7 tion: 400⳯. Source: Ref 7

Fig. 24 Ni-25Cu (at.%) alloy. Etchant: 70 mL HNO3

and 30 mL H2O. Magnification: 10⳯. Source:
Ref 7

Fig. 22 Compositional map of the area shown in Fig. 21. (a) Bright cobalt-rich dendrite image formed by Co-K␣
wavelength x-rays emission. Source: Ref 7

Fig. 25 12 Ni-25Cu (at.%) alloy. Etchant: 70 mL HNO3

Fig. 23 Effect of crystalline anisotropy on interface shape in directional growth (growth velocity of 35 lm/s) of direc- and 30 mL H2O. Magnification: 175⳯. Cour-
tional-solidification growth patterns in thin films of the CBr4–8 mol % C2Cl6 alloy). Source: Ref 8 tesy of C. Brady. Source: Ref 7
Name /bam_asmint_106420/6044_002d1/Mp_9 08/09/2004 08:20AM Plate # 0 pg 9 #
Fundamentals of Solidification / 9

Figure 12 indicates that for rather steep ther- ning of solidification when there is no solid to CSM ⳱ 5.65% Cu when fS ⳱ 0.91. The re-
mal gradients, columnar dendrites will form, formed, the fraction solid is fS ⳱ 0. The first maining fraction of the alloy (1 ⳮ 0.91 ⳱ 0.09)
while for shallow gradients, equiaxed dendrite solid to form will have the composition kC0  solidifies as an interdendritic eutectic with an av-
will solidify. This is illustrated in Fig. 27, which C0. Assuming no diffusion in solid and liquid of erage composition of 33% Cu. The predictions
shows the computer-processed images of a tita- uniform composition, the composition of the of the Gulliver-Scheil equation should be used
nium alloy ingot. A columnar-to-equiaxed tran- solid will continue to increase as solidification carefully. Generally, the equation tends to un-
sition is also shown. progresses. It will soon be higher than C0, and derestimate slightly the composition at the center
Effect of Solidification Velocity. As empha- then it will reach the maximum solubility in of a dendrite and to overestimate the volume
sized previously, solidification velocity is, to- solid, CSM, and then the eutectic composition, fraction of eutectic (for k  1). Also, if extensive
gether with the temperature gradient, the most CE, according to the phase diagram in Fig. 6. The fluid flow exists through the dendritic or
important variable affecting microstructure tran- liquid still available will now solidify as eutectic. “mushy” zone, the average composition of the
sitions. As shown in Fig. 28, the change in so- The composition of the solid (the thick line in solidified castings may be significantly altered in
lidification velocity may determine a planar S/L Fig. 31) as well as the amount of eutectic at the regions that are large compared to the dendrite
interface to become cellular and then dendritic end of solidification, fE, can be calculated with scale. This is called macrosegregation. Details
(Ref 10). In addition, the morphology of the the Gulliver-Scheil equation: regarding this subject and the field of solidifi-
equiaxed dendrites (branching and tip radius) de- cation can be found in Ref 1.
pends significantly on the cooling rate and/or un- CS ⳱ kC0(1 ⳮ fS)kⳮ1 (Eq 9) Rapid Solidification. The microstructural
dercooling. The effect of solidification velocity length scale of solidified alloys generally de-
over a wide range of velocities can be under- Using this approach, it can be calculated that creases as the rate of heat extraction (cooling
stood from Fig. 29. At very small velocities the for a 4.5%Cu-Al alloy the first solid to form (the rate) increases. The term rapid solidification is
dendrite tip radius is very large, even infinity, in centers of dendrites) will have a composition of normally applied to casting processes in which
which case a planar interface is obtained. As the 0.77% Cu and that the composition will increase the liquid cooling rate exceeds 100 K/s (Ref 7).
velocity increases, the radius decreases and the This definition is rather vague because different
morphology changes from planar to globular/ alloys respond very differently to high rates of
cellular, then to regular equiaxed dendritic. Fur- cooling. Also, some microstructures observed in
ther increase in solidification velocity in the rapidly solidified alloys can be achieved by slow
range of rapid solidification determines a tran-
sition from fully branched to globular/cellular
dendrites, and finally again to planar interface
(absolute stability). A typical example illustrat-
ing the influence of cooling rate on the mor-
phology of equiaxed dendrites of an Al-7Si alloy
is given in Fig. 30 (Ref 11).
The solidification time scale also influences
the secondary dendrite arm spacing (SDAS). The
SDAS is the distance between adjacent branches
growing from the main dendritic arm. It is di-
rectly related to certain mechanical properties. It
is generally accepted that the SDAS is a function
of the local solidification time, tf , described by:

SDAS = µ 0 ⋅ tf1 3 (Eq 7)

where l0 is a material-specific constant (coars-

ening constant). Extensive experimental data on
secondary arm spacing have also been reported
to fit a SDAS-cooling rate equation (Ref 12):

SDAS ⳱ l1 • (Ṫ)ⳮ0.34Ⳳ0.02 (Eq 8)

where l1 is a material specific constant, and Ṫ is

the cooling rate.
Solute Redistribution and Microsegrega-
tion in Dendritic Solidification. Rejection of
solute from the solid during solidification that is
responsible for the formation of the solutal
boundary layer (see Fig. 7) produces composi-
tional nonuniformity across the dendrite during
solidification called microsegregation. To under-
stand the mechanism of formation of microse-
gregation, consider the volume element extend- Fig. 26 Transverse microstructures of Pb-2.2Sb (wt%)
ing from the axis of the dendrite arm to the edge alloy single-crystal samples grown along [100]
of the final dendrite (at the end of solidification) at GT ⳱ 164 K/cm. Etchant: 60 mL acetic acid and 40 mL Fig. 27 Macrostructure of a Ti-6Al-4V ingot. (a) Lon-
hydrogen peroxide (30%). (a) Deep cells grown at 2.5 lm/ gitudinal section with coarse equiaxed grains
shown in Fig. 31. The thick line in the lower part s. Arrow indicates tip splitting. (b) Completely dendritic in the center (light), columnar grains (gray), and fine
of the figure represents the composition change structure. Dendrites grown at 10 lm/s (same scale). equiaxed grains on the surface. (b) Cross section with re-
in the solid during solidification. At the begin- Adapted from Ref 9 verse coloration
Name /bam_asmint_106420/6044_002d1/Mp_10 08/09/2004 08:20AM Plate # 0 pg 10 #
10 / Metallurgy and Microstructure

cooling when large liquid undercooling is ure 32(a) shows a transverse section of a fine minum-manganese alloys (Ref 15). This transi-
achieved prior to nucleation (Ref 13). cellular structure of the silver-rich phase in Ag- tion from an intermetallic to an aluminum solid
Techniques usually used to produce rapidly 15%Cu alloy (Ref 14). In this figure, most of the solution as the primary phase can be understood
solidified alloys are melt spinning, planar flow intercellular regions are filled with the copper- by a careful examination of the kinetics of the
casting, or melt extraction, which produce thin rich phase, not the eutectic of silver and copper competitive nucleation and growth of the inter-
(25 to 100 lm) ribbon, tape, sheet, or fiber; as the Gulliver-Scheil equation would predict. metallic and ␣-aluminum solid solution (Ref 16).
atomization, which produces powder (10 to Figure 32(b) shows a longitudinal view of a cel- In some cases, an intermetallic that is not
200 lm); and surface melting and resolidifica- lular solidification structure. In Fig. 32(c), the given on the equilibrium phase diagram may
tion, which produce thin surface layers. These alloy has solidified with a planar interface to pro- compete with ␣-aluminum. In aluminum-iron al-
methods may be considered casting techniques duce a microsegregation-free alloy. The fine par- loys, a metastable phase, Al6Fe, rather than the
where at least one physical dimension of the final ticles are the result of a solid-state precipitation. stable phase, Al3Fe, can form under some rapid-
product is small. Consolidation is used to yield A common occurrence in some rapidly solid- solidification conditions. This situation is anal-
large products from rapidly solidified alloys. ified alloys is a change in the identity of the pri- ogous to the appearance of cementite rather than
This consolidation often alters the solidification mary solidification phase from that observed for graphite in some cast irons. The use of metasta-
microstructure in final products. However, as slow solidification (Ref 7). Many examples are ble phase diagrams to assist in the interpretation
with ordinary castings, many features of the so- found in hypereutectic aluminum alloys contain- of rapidly solidified microstructures is described
lidification structure can remain in the final prod- ing transition elements such as iron, manganese, in Ref 17. As explained in the section “Under-
uct (Ref 7). or chromium. If the alloy is hypereutectic, cooling” in this article, other rapidly solidified
The effect of high cooling rates on the dendrite slowly cooled castings will contain intermetal- alloys have microsegregation-free structures
morphology can be understood from Fig. 29. It lics such as Al3Fe or Al6Mn as the primary (or formed by a liquid-solid transformation similar
is seen that as the cooling rate increases in the first) phase to solidify. However, under rapid- to a massive solid-solid transformation (parti-
rapid-solidification range, the tip radius in- solidification conditions the primary phase in tionless or diffusionless transformation). The liq-
creases. This also means that branching de- these alloys is the aluminum solid solution, usu- uid transforms to solid without a change in com-
creases and the equiaxed dendrite will become ally found in a cellular structure with an inter- position. The ratio of the solid composition at
globular/cellular. Typical examples of the evo- metallic in the intercellular regions. Figures 33 the interface to the liquid composition is CS/CL
lution of the microstructure as a function of the and 34 show globular/cellular structures of the ⳱ 1, rather than the equilibrium partition coef-
solidification velocity are given in Fig. 32. Fig- ␣-aluminum solid solution in hypereutectic alu- ficient. Velocities required to produce partition-
less solidification must exceed 5 m/s (40 to 400
in./s). The phase diagrams obviously do not ap-
ply in this situation.
Figures 35 and 36 show optical and transmis-
sion electron micrographs, respectively, of a sin-
gle-phase microsegregation-free solid solution
of a silver-copper alloy of eutectic composition
(28% Cu) that was formed by partitionless solid-
ification (Ref 18). The alloy is not only free of
microsegregation, but also has a solid solubility
of copper in silver far in excess of the equilib-
rium solubility limit (9% Cu).
Rapidly solidified alloy powders exhibit a
broad spectrum of solidification structures, de-
pending on alloy composition and solidification
conditions. Figure 37 shows single powder par-
ticles of stainless steel with dendritic or cellular
structure (Ref 19). The size of the particles is
less than 25 lm. Different degrees of under-

Fig. 28 Change in interface morphology of a succinonitrile-4% acetone solution when increasing the solidification
velocity from 0 to 3.4 lm/s at a temperature gradient of 6.7 K/mm. (a) 50 s. (b) 55 s. (c) 65 s. (d) 80 s. (e) 135 Fig. 29 Correlation between solidification velocity and
s. (f ) 740 s. Magnification: 30⳯. Source: Ref 10 dendrite tip radius
Name /bam_asmint_106420/6044_002d1/Mp_11 08/09/2004 08:20AM Plate # 0 pg 11 #
Fundamentals of Solidification / 11

Fig. 30 The microstructures of an Al-7Si alloy at various cooling rates. (a) 0.3 C/s. (b) 3 C/s. (c) 10 C/s. (d) 30 C/s. Source: Ref 11

cooling prior to nucleation for particles of almost ● Cooperative growth: The two phases of the Operating Compositional Range. From the
the same size determine the type of structure. eutectic grow together as a diffusion couple. eutectic phase diagram it appears that a eutectic
The dendritic structure radiates from a point on ● Divorced growth: The two phases of the eu- structure can be obtained only when the com-
the surface where nucleation has occurred. The tectic grow separately; there is no direct ex- position is exactly eutectic. Nevertheless, both
scale of the structure is relatively uniform across change of solute between the two solid phases experiments and theory show that, depending on
the powder particle. and no trijunction. the growth conditions, eutectic microstructures
Other rapidly solidified powders often show can be obtained at off-eutectic compositions.
significant microstructural variations across in- Cooperative eutectics can be further classified Such conditions include a sufficiently steep gra-
dividual powder particles. Initial growth of the based on the ratio between the fractions of the dient or slow solidification velocity during di-
solid may occur very rapidly in a partitionless two phases of the eutectic, f␣ and fb, and on the rectional solidification. This is possible because
manner. The interface velocity decreases as the morphology of the S/L interface (Ref 21) as the eutectic grows faster than the dendrites, since
S/L interface crosses the particle, because of the shown in Fig. 39. The nondimensional entropy diffusion-coupled growth is much faster than
release of the latent heat of fusion and warming of fusion, DSf/R, where R is the gas constant, is isolated dendritic growth. Accordingly, even in
of the powder particle, and the solidification used to distinguish between faceted and non- off-eutectic compositions, the eutectic may out-
front becomes cellular. Figure 38 shows a two- faceted morphologies. grow the individual dendrites, resulting in a
zone microstructure observed in a larger diam- Alloys such as Pb-Sn and Al-Al2Cu (Fig. 40a) purely eutectic microstructure. On the other
eter (10 mm) hypereutectic Al-8%Fe powder (Ref 22), where there are approximately equal hand, at high growth velocities, dendrites can be
particle. A thin foil was prepared by electropol- volume fractions of nonfaceted phases, solidify found in alloys of eutectic compositions.
ishing a 3 mm (0.12 in.) diam green powder as regular, lamellar eutectics. If one of the phases An analysis of the possible solidification mi-
compact (Ref 20). In this alloy, the zone to the is nonfaceted the morphology becomes irregular, crostructure of a binary alloy can be made based
left where nucleation occurs contains a very fine because the faceted phase grows preferentially on the growth velocities of the competing phases
cellular structure, and the zone to the right con- in a direction determined by specific atomic (Ref 1). On the phase diagram in Fig. 41(a), the
tains a coarse cellular structure of ␣-aluminum planes. Also, one solid phase projects into the three shaded regions extending under the eutec-
with Al6Fe between the cells resulting in the de- liquid far in advance of the other solid phase. A tic invariant form a coupled zone. This is a so-
crease on interface velocity because of recales- typical example is the Mg-Mg2Sn eutectic lidification velocity dependent composition re-
cence. Larger powder particles of the alloy fre- shown in Fig. 40(b). gion in which the eutectic grows more rapidly,
quently do not undercool significantly before the When the volume fraction of one phase is sig- or at a lower undercooling, than the ␣ or b den-
start of solidification and contain Al3Fe as the nificantly lower than that of the other (typically
primary phase. 0.28), a fibrous structure will result (for ex-
ample, the Ni-NbC eutectic (Ref 23) shown in
Fig. 40c). This is a result of the tendency of the
Solidification system to minimize its interfacial energy by se-
Structures of Eutectics lecting the morphology that is associated with
the smallest interfacial area. Fibers have smaller
Eutectics are alloys that have a fixed compo- interfacial area than lamellae. However, when
sition in terms of species A and B and solidify the minor phase is faceted, a lamellar structure
as two-phase solids (␣ Ⳮ b). The phase diagram may form even at a very low volume fraction,
in Fig. 16(b) shows a binary eutectic invariant because specific planes may have the lowest in-
point at temperature TE and composition CE. At terfacial energy. The minor phase will then grow
this point, two solid phases, ␣ and b, solidify such as to expose these planes even when la-
simultaneously from the liquid, L. The eutectic mellae rather than fibers are formed. The two
reaction can be written as: L r ␣ Ⳮ b. As many commercially most significant eutectics, alumi-
as four phases have been observed to grow si- num-silicon and iron-graphite fall into this cate-
multaneously from the melt. However, most gory. Note that in the iron-graphite eutectic, the
technologically useful eutectic alloys consist of fraction graphite is fGr ⳱ 0.07. The iron-graphite
two phases. The particular morphology of the eutectic can be either cooperative, irregular, as is
eutectic is a function of processing conditions the case for lamellar graphite cast iron, or di-
and of the nature of the two phases. vorced, as for spheroidal graphite cast iron (Fig.
Classification of Eutectics. Many eutectic 40d). In this last case, at the beginning of solid- Fig. 31 Model of microsegregation occurring in the
arm of a growing dendrite. phase diagram. CE,
classifications have been proposed, based on dif- ification the two phases, graphite and austenite eutectic composition; CSM, maximum solubility in solid; fs,
ferent criteria. A first classification of eutectics dendrites, grow independently from the liquid solid fraction; fE, amount of eutectic critical; C0, critical
based on their growth mechanism is: without establishing a diffusion couple. solute concentration; K, partition coefficient
Name /bam_asmint_106420/6044_002d1/Mp_12 08/09/2004 08:20AM Plate # 0 pg 12 #
12 / Metallurgy and Microstructure

Fig. 33 Globular structure of ␣-aluminum in a melt-

spun Al-12%Mn alloy. Small particles of a dif-
ferent phase decorate the cell walls. Thin-foil transmission
electron micrograph (TEM). Electropolished at ⳮ30 C
(ⳮ20 F) in 950 mL methanol, 50 mL HClO4, and 15 mL Fig. 35 Columnar grains of single-phase solid solution
of melt-spun Ag-28%Cu alloy (eutectic com-
HNO3. Magnification: 16,000⳯. Source: Ref 15 position). Section of full ribbon cross section. Chill (wheel)
side is at bottom. Etched with: 20 mL NH4OH, 10 mL 3%
H2O2, 10 mL H2O (used fresh). Magnification: 2000⳯.
Source: Ref 18

Fig. 32 Microstructures of silver-copper alloys electron

beam melted and resolidified at different ve-
locities. (a) Ag-15Cu alloy resolidified at 0.025 m/s (1 in./
s). Globular microsegregation pattern. Magnification:
32,000⳯. (b) Ag-15Cu alloy resolidified at 0.3 m/s (12 in./
s). Cellular microsegregation pattern revealed by disloca-
tion networks along cell walls. Magnification: 18,000⳯.
(c) Ag-15Cu alloy resolidified at approximately 0.6 m/s (24 Fig. 34 Elongated cellular structure in a melt-spun Al-
in./s). The cellular structure is absent, and the solid pro- 15%Mn alloy. The contrast between some cells Fig. 36 Same alloy as in Fig. 35. Three grains observed
duced is uniform in composition except for fine copper indicates crystallographic misorientation (subgrains). Thin- in the thin foil parallel to the chill surface pre-
precipitates formed during solid-state cooling. Magnifica- foil transmission electron micrograph. Electropolished at pared by ion milling Transmission electron micrograph.
tion: 87,000⳯. Thin-foil transmission electron micrograph ⳮ30 C (ⳮ22 F) in 950 mL methanol, 50 mL HClO4, and Magnification: 100,000⳯. Courtesy of D. Shechtman.
prepared by ion milling. Source: Ref 14 15 mL HNO3. Magnification: 13,000⳯. Source: Ref 15 Source: Ref 7
Name /bam_asmint_106420/6044_002d1/Mp_13 08/09/2004 08:20AM Plate # 0 pg 13 #
Fundamentals of Solidification / 13

Fig. 37 Scanning electron micrographs of atomized droplets of martensitic stainless steel. (a) and (b) Dendritic structures. (c) and (d) Cellular structures. Source: Ref 19

drites. For regular eutectics the coupled zone is plane was normal to the direction of solidifica- phorus and the formation of a low-melting-point
symmetric. Note that the widening of the cou- tion) and the fanlike arrangement (where the me- phosphide eutectic.
pled zone near the eutectic temperature is ob- tallographic plane was parallel to the direction Grain boundaries are not easily observable in
served only in directional solidification (DS), of solidification). eutectics solidifying with a planar interface. For
where the thermal gradient is positive. If the undercooling is high, equiaxed eutectic example, the right side of Fig. 46 has the ap-
For small undercooling the S/L interface is grains may form even during directional solidi- pearance of a specimen containing two grains:
planar (see Fig. 42). As shown on the right side fication. A typical example is given in Fig. 10(c) the lamellae in one grain are at an angle of about
of Fig. 41(a), even for a hypereutectic alloy so- for a gray cast iron. The boundaries of the grains 40 to the lamellae in the other grain. At the left
lidifying at small undercooling, the eutectic has are outlined because of the segregation of phos- of the micrograph, however, the two grains
the highest growth velocity and a planar, coupled
eutectic is produced. At higher undercooling, the
b phase will have higher growth velocity, and a
eutectic-dendritic structure will result. At even
higher undercooling, the eutectic velocity will
again become the highest. However, because of
the undercooling, a planar structure is not pos-
sible, and equiaxed coupled growth will result.
If one of the eutectic phases is faceted, the
growth of this phase and consequently that of the
eutectic is slowed down. Dendrites of the other
phase may grow faster at a given undercooling
than the eutectic, even for the eutectic compo-
sition. Consequently, purely eutectic microstruc-
tures can be obtained only at hypereutectic com-
positions. This is exemplified in Fig. 41(b), for
the case of faceted b phase. An asymmetric cou-
pled zone results.
From Fig. 41(a) it can be noted that as the
undercooling increases, the microstructure of the
eutectic changes from planar to cellular, den-
dritic, and then equiaxed. When the alloy solid-
ifies with a cellular rather than a planar interface,
eutectic colonies are formed. Figure 43 illus-
trates the effect of increased solidification veloc-
ity (higher undercooling) on the eutectic mor-
phology (Ref 25). Alternating light regions of
(Cr,Mo) plates, and dark regions of NiAl plates
are seen at both growth speeds. As the solidifi-
cation velocity increases, the microstructure
changes from planar to fanlike arrangement col-
onies.
A section normal to the direction of solidifi-
cation of a colony structure is presented in Fig.
44. Note the honeycomb pattern of the colonies
(Ref 24). Dark and light layers in each colony
are Mg2Al3 phase and Al, respectively. Colony
structures are observed also in non-DS speci-
mens (Fig. 45). The section shows both the Fig. 38 Thin-foil transmission electron micrograph of vacuum-atomized Al-8Fe powder. The green powder compact
was electropolished at ⳮ30 C (ⳮ22 F) in 950 mL methanol, 50 mL HClO4, and 15 mL HNO3. Magnification:
honeycomb pattern (where the metallographic 6300⳯. Courtesy of W.J. Boettinger, L. Bendersky, and J.G. Early. Source: Ref 7
Name /bam_asmint_106420/6044_002d1/Mp_14 08/09/2004 08:20AM Plate # 0 pg 14 #
14 / Metallurgy and Microstructure

merge into one. In other eutectics, particularly

those in which the volume fraction of one phase
is f  0.01 to 0.02, grains of the predominant
phase will be visible.
Coupled two-phase eutectic structures also
can be produced whose amounts and individual
phase chemistries can be varied by solidification
of compositions that lay on or near eutectic
troughs in the liquidus surface of ternary alloys
(see Fig. 47) (Ref 26).
Caution must be exercised during metallo-
graphic analysis of eutectic microstructures. Se-
rial sectioning, phase-extraction techniques, and
examination of two sections meeting at a com-
mon edge have revealed that the microstructures
of most eutectics cannot be considered aggre-
gates of many simply shaped, discrete particles
of one phase embedded in a matrix of the other
phase or phases. Rather, if the three-dimensional
shapes of all phases are examined, the apparently
individual particles of each phase are typically
found interconnected in a topologically complex
arrangement. Additional examples are provided
for specific alloys.
Length Scale of Eutectics. The length scale
of the eutectic strongly affects the mechanical
Fig. 39 Types of cooperative eutectics. Source: Ref 21 and physical properties of the eutectic aggregate.
For cooperative eutectics, the length scale is
given by the lamellar spacing (interlamellar or
interfiber) spacing, which is affected by solidi-
fication velocity, thermal gradients, undercool-
ing, atomic bonding, relative amounts, crystal-
lographic factors, interfacial energies, impurity
content, and alloy composition. The lamellar
spacing, k, and the solidification velocity are re-
lated by the simple equation k2V ⳱ constant.
The effect of solidification velocity is illustrated
in Fig. 48. It is seen that the spacing of irregular
eutectics is significantly larger than that of reg-
ular eutectics.
The adjustment in the eutectic spacing during
growth occurs through faults. Two types of faults
are shown in Fig. 49. Figure 49(a) shows a no-
net fault in which the number of lamellae on both
sides of the fault is the same. Figure 49(b) shows
a net fault in which one side of the fault has one
more lamellae than the other side. This fault is
analogous to an extended dislocation in that the
number of lamellae above and below the fault
differ by one (Ref 27).
For equiaxed eutectics, the length scale may
include grain size in addition to lamellar spacing.
Metallographic identification of the grain size is
alloy specific.

Solidification
Structures of Peritectics

Peritectic solidification is very common in the

solidification of metallic alloys. Many techni-
cally important alloy systems such as steels, cop-
per alloys, rare-earth permanent magnets, and
high TC superconductors (TC is the temperature
Fig. 40 Eutectic microstructures. (a) Regular nonfaceted/nonfaceted eutectic (Al-Al2Cu). (b) Irregular faceted/nonfa-
above which superconductive behavior is lost)
ceted eutectic (Mg-Mg2Sn). The dark phase is the faceted Mg2Sn. (c) Rod faceted/nonfaceted eutectic (Ni-
NbC). (d) Divorced eutectic (Fe-spheroidal graphite) display peritectic reactions in the regions of their
Name /bam_asmint_106420/6044_002d1/Mp_15 08/09/2004 08:20AM Plate # 0 pg 15 #
Fundamentals of Solidification / 15

phase diagrams where phase and microstructure by in situ dynamic observation of the progress Consequently, a cycle is set up leading to the
selection play an important role for the process- of peritectic reactions and transformations of layered microstructure (Ref 31).
ing and the properties of the material. Basically, iron-carbon alloys made with a combination of Many researchers have attempted to produce
the peritectic solidification means that at the per- a confocal scanning laser microscope and an in- microstructure selection diagrams for peritectics
itectic temperature TP, a solid phase b of peri- frared image furnace (Ref 28). in an attempt to clarify phase and microstructure
tectic composition CP solidifies from a mixture Simultaneous growth of two phases in the selection. Taking into account nucleation and
of liquid L and solid phase ␣. The peritectic so- form of oriented fibers and lamellae has been growth of the phases, Hunziker et al. (Ref 32)
lidification can be written as: L Ⳮ ␣ r b. A observed in some peritectic alloys when the have proposed a criterion to determine the phase
phase diagram with peritectic solidification is composition was on the tie-line of the two solid as a function of alloy composition and growth
presented in Fig. 16(c). The different reactions phases and the GT/V ratio was close to the limit conditions. A microstructure selection diagram
occurring along the solidus lines, corresponding of constitutional undercooling for the stable for peritectic alloys close to the limit of consti-
to various compositions (white and gray regions) phase having the smaller distribution coefficient tutional undercooling was developed (see Fig.
are also shown. (Ref 29). Figure 55 shows such a structure for 57). The lines 1 through 7 on the diagram were
Two different mechanisms are involved in an iron-nickel alloy. calculated. It is seen that the diagram allows pre-
peritectic solidification, namely peritectic reac- Banded structures have been observed in per- diction of planar front, cellular, dendritic, and
tion and peritectic transformation. These mech- itectic alloys at low growth rates. An example is band solidification.
anisms are shown in Fig. 50. In a peritectic re- provided in Fig. 56 for lead-bismuth alloys (Ref Fluid flow can further complicate the possible
action, all three phases (␣, b, and liquid) are in 30). The formation of bands is explained by nu- microstructures. Up to five different structures
contact with each other. In the peritectic trans- cleation and growth of the second phase during have been identified, as summarized in Fig. 58,
formation, the liquid and the primary ␣ phase the initial transient of planar growth of the pri- where the solidification direction is upward
are isolated by the b phase. The transformation mary phase and vice versa. This occurs because (Ref 33).
takes place by long-range diffusion through the the liquid at and ahead of the growing interface
secondary b phase. A variety of microstructures is constitutionally undercooled with respect to
can result from peritectic solidification, mostly the other phase. As the second phase nucleates Solidification
depending on the GT/V ratio and nucleation con- and grows ahead of the primary phase, the for- Structures of Monotectics
ditions. The possible structures include cellular, mer phase cannot reach the steady state. Simi-
plane-front, bands, eutecticlike structures. larly, the primary phase nucleates again during Monotectic solidification occurs in alloys
Examples of such microstructures are given in the transient growth regime of the second phase, where the liquid separates into two distinct liquid
Fig. 51 to 54. The pictures have been obtained preventing it from reaching the steady state. phases of different composition during cooling.
These alloys have limited commercial applica-
tions. However, one could mention lead-contain-
ing copper alloys and the possible fabrication of
thin microfilters has been proposed as possible
application.
A dome-shaped region within which the two
liquids mix and coexist is seen on the phase di-
agram (Fig. 16d). The maximum temperature of
this dome, Tc, is called the critical (or consolute)
temperature. At the monotectic temperature, Tm,
the following monotectic reaction occurs:

L1 ⳱ ␣ Ⳮ L2 (Eq 10)

Fig. 42 Regular cadmium-tin eutectic in a casting di-

rectionally solidified from bottom to top, then
rapidly cooled to preserve the shape of the solid/liquid in-
terface. Section parallel to solidification direction. Etchant:
Fig. 41 Coupled eutectic zones. (a) Symmetric coupled zone (regular eutectics. (b) Asymmetric coupled zone (irreg- aqueous solution of FeCl3. Magnification: 210⳯. Source:
ular eutectics). Source: Ref 1 Ref 24
Name /bam_asmint_106420/6044_002d1/Mp_16 08/09/2004 08:20AM Plate # 0 pg 16 #
16 / Metallurgy and Microstructure

on one side and by the critical velocity of push-

ing-to-engulfment transition (Vcr) on the other
(Fig. 62). When the solidification velocity is
smaller than Vcr a banded structure may result.
An example of such a structure is provided in
Fig. 63 (Ref 37). It is suggested the L2 phase,
which precipitates at the solid/liquid interface,
piles up and covers the S/L interface. This pro-
duces a lead-rich layer and increases the under-
cooling of the L1/L2 interface with respect to the
monotectic temperature. Then, nucleation of the
␣Cu phase occurs on the lead-rich layer. The
temperature at the growth front is also returned
to the monotectic temperature. The repetition of
this process will result in the banded structure.
This process is shown in Fig. 64. It should be
Fig. 43 Microstructure of directionally solidified Ni-33Al-31Cr-3Mo rods. The arrows indicate growth directions (lon-
noted that some fibrous structure might form
gitudinal section). (a) 12.7 mm/h. (b) 25.4 mm/h. Etchant not reported. Source: Ref 25
even in the case of banding as seen in Fig. 65.
For high-dome alloys, the lowest energy exists
Monotectic alloys can be classified based on the velocity, Vcr, L2 is incorporated with formation when an ␣/L2 interface exists. Consequently, ␣
difference Tc ⳮ Tm as high-dome alloys (high Tc of an irregular fibrous composite (Fig. 59b). A and L2 will grow together (L2 wets ␣), resulting
ⳮ Tm of hundreds of C) and low-dome alloys detailed discussion on particle engulfment and in a regular (uniform) fibrous composite. The k2
(low Tc ⳮ Tm of tens of C). They can also be pushing by solidifying interfaces is given in Ref • V relationship is about two orders of magnitude
classified based on the Tm/Tc ratio. High-dome 1. This mechanism has been observed experi- larger for irregular than for regular monotectic
alloys have Tm/Tc  0.9, while low-dome alloys mentally. composites (with the exception of the aluminum-
have Tm/Tc  0.9. Figure 60 shows fibrous growth in the succi- bismuth alloy) and about one order of magnitude
The morphology of the microstructure pro- nonitrile-20% ethanol system when the velocity higher for regular monotectic composites than
duced during directional solidification is a func- is above the critical velocity for particle engulf- for regular eutectics. The differences come from
tion of the density difference between the two ment (Ref 35). Similar microstructures were ob- the controlling mechanism. For irregular fibrous
liquids, and of the wetting between L2 and ␣. As served in metallic system such as Cu-60%Pb and eutectics, the controlling mechanism is the push-
theorized by Cahn (Ref 34), for low-dome alloys Bi-50%Ga alloy as shown in Fig. 61 for a cop- ing-engulfment transition, which is a function of
the phases ␣ and L2 are separated by L1 as shown per-lead alloy (Ref 36). Note that the velocity in solidification velocity and surface energy. For
in Fig. 59. At low growth rate, L2 particles are this last microstructure is very high. regular fibrous monotectics, the spacing is con-
pushed by the S/L interface (Fig. 59a). If the The range of existence of the fibrous compos- trolled by surface energy. For eutectics, the spac-
solidification velocity increases above a critical ite is limited by the constitutional undercooling ing is controlled by diffusion (Ref 1).

Fig. 44 Colony structure of a directionally solidified la- Fig. 45 Colony structure of a non-directionally solidi- Fig. 46 Directionally solidified CuAl2-Al lamellar eu-
mellar eutectic. Etchant not reported. Magni- fied CuAl2-Al lamellar eutectic. As-polished. tectic. As-polished. Magnification: 180⳯.
fication: 200⳯. Source: Ref 24 Magnification: 250⳯. Source: Ref 24 Source: Ref 24
Name /bam_asmint_106420/6044_002d1/Mp_17 08/09/2004 08:20AM Plate # 0 pg 17 #
Fundamentals of Solidification / 17

Fig. 47 Monovariant liquidus troughs (dashed lines) on a Ni-Al-Mo isothermal section at 1300 C (2370 F) with varying “eutecticlike” structure produced by compositional
adjustment. Source: Ref 26

Fig. 49 Cross sections (a) and (b) of a directionally so-

lidified lead-cadmium eutectic showing the
Fig. 48 Comparison of the k ⳮ V correlation for eu- presence of faults in the lamellae. Etchant not reported. Fig. 50 Mechanisms of peritectic solidification.
tectics and eutectoids. Source: Ref 1, 24 Source: Ref 27 Source: Ref 1
Name /bam_asmint_106420/6044_002d1/Mp_18 08/09/2004 08:20AM Plate # 0 pg 18 #
18 / Metallurgy and Microstructure

Fig. 53 Peritectic reaction and transformation of Fe-

Fig. 51 Planar growth of d crystals in Fe-0.14C alloy Fig. 52 Cellular growth of d crystals in Fe-0.14% C al- 0.14C alloy during solidification and at 1768
melt (GT ⳱ 22 K/mm, V ⳱ 4.3 lm/s), same loy melt (GT ⳱ 4.3 K/mm, V ⳱ 2.5 lm/s). (a) K (GT ⳱ 4.3 K/mm, cooling rate ⳱ 20 K/min). (a) 0 s. (b)
1
scale. (a) At 0 s. (b) At 60 s. Source: Ref 28 At 0 s. (b) At 20 s. Source: Ref 28 ⁄30 s. (c) 2 s. Source: Ref 28
Name /bam_asmint_106420/6044_002d1/Mp_19 08/09/2004 08:20AM Plate # 0 pg 19 #
Fundamentals of Solidification / 19

Fig. 54 Peritectic reaction and transformation of Fe-0.42C alloy during isothermal holding at 1765 K (same scale). (a)
0 s. (b) 0.2 s. (c) 3 s. (d) 7 s. Source: Ref 28

Fig. 55 Quenched solid/liquid interface of simultaneous two-phase growth in peritectic iron-nickel alloy. Source:
Ref 29
Name /bam_asmint_106420/6044_002d1/Mp_20 08/09/2004 08:20AM Plate # 0 pg 20 #
20 / Metallurgy and Microstructure

Fig. 56 Cross sections of directionally solidified Pb-33Bi (at.%) alloys (100 mm in initial melt length) at GT ⳱ 2.7 ⳯
104 K/m. (a) V ⳱ 1.4 lm/s. (b) V ⳱ 0.83 lm/s. (c) V ⳱ 0.56 lm/s. The black region is the ␣ phase, and the
white one is the b phase. Source: Ref 30
Name /bam_asmint_106420/6044_002d1/Mp_21 08/09/2004 08:20AM Plate # 0 pg 21 #
Fundamentals of Solidification / 21

1793

1791
Temperature, K

C1
Cδ
δ Cγ

1789

1787
Cδ Cγ
(a)

1.4 × 109
Bands
2 I
1.2 × 109 δ plane front

109 Mixed bands

II γ plane front
1
G/V, sK/m2

8 × 108
3
6 × 108

δ cells/dendrites
4 × 108 III
6 4

IV
2 × 108
7 γ cells/dendrites

0
3 3.5 4 4.5 C1 5 5.5
Composition, at.% Ni
(b)

Fig. 57 Microstructural selection diagram. (a) Iron-nickel phase diagram in the vicinity of the peritectic plateau. (b)
Predicted phase/microstructure selection map assuming negligible nucleation undercooling for both d and c
phases. The calculated lines are numbered 1 through 7. Source: Ref 32
Name /bam_asmint_106420/6044_002d1/Mp_22 08/09/2004 08:20AM Plate # 0 pg 22 #
22 / Metallurgy and Microstructure

Fig. 58 Fluid-flow controlled microstructures in peritectic alloys. Solidification direction is upward. (a) Discrete bands of the two phases. (b) Partial bands or islands of one phase
in the matrix of the other phase. (c) Single primary to peritectic phase transition. (d) Simultaneous growth of the two phases with a planar solid/liquid interface. (e) Dispersed
phases due to nucleation ahead of the interface. (f ) Oscillating continuous treelike structures of the primary phase surrounded by peritectic. Source: Ref 33

Fig. 59 Monotectic solidification for low-dome alloys. Fig. 60 Growth front in succinonitrile-20wt% ethanol, Fig. 61 Microstructure of a Cu-70Pb alloy solidified at
(a) Low growth velocity. (b) High growth ve- showing incorporation of ethanol droplets. V V ⳱ 778 lm/s and G ⳱ 12 K/mm. Solidifica-
locity. Source: Ref 1 ⳱ 0.27 lm/s, G ⳱ 4.8 K/mm. Source: Ref 35 tion direction right-to-left. Source: Ref 36
Name /bam_asmint_106420/6044_002d1/Mp_23 08/09/2004 08:20AM Plate # 0 pg 23 #
Fundamentals of Solidification / 23

Fig. 62 Restriction on composite growth imposed by

the critical velocity for the pushing engulfment
transition and by constitutional undercooling. Source: Ref 1

Fig. 63 Microstructure of upward directional solidifi-

cation of a Cu-37.7Pb alloy in longitudinal sec-
tion. V ⳱ 4.4 lm/s. Source: Ref 37

Fig. 65 The solid/liquid interface covered with coa-

Fig. 64 Forming mechanism of the banded structure of copper-lead alloy in upward directional solidification. Source: lesced L2 phase. Cu-35.4Pb alloy, upward di-
Ref 37 rectional solidification, V ⳱ 2.2 lm/s. Source: Ref 37
Name /bam_asmint_106420/6044_002d1/Mp_24 08/09/2004 08:20AM Plate # 0 pg 24 #
24 / Metallurgy and Microstructure

REFERENCES 12. D. Bouchard and J.S. Kirkaldy, Metall. Ma- 25. F.D. Lemkey and R.W. Kraft, Solidification
ter. Trans., Vol 28B, 1997, p 651 Structures of Eutectic Alloys, Metallogra-
1. D.M. Stefanescu, Science and Engineering 13. J.H. Perepezko and J.J. Paike, Rapidly So- phy and Microstructures, Vol 9, ASM Hand-
of Casting Solidification, Kluwer Academic, lidified Amorphous and Crystalline Alloys, book, 1985, p 618–622
2002 B.H. Kear, B.C. Giessen, and M. Cohen, 26. M.G. Day and A. Hellawell, Proc. R. Soc.,
2. D.G. Altenpohl, Aluminum: Technology, Ed., North Holland, 1982, p 49 Vol A305, 1968, p 479–491
Applications, and Environment. A Profile of 14. W.J. Boettinger, D. Shechtman, R.J. Schae- 27. R. Trivedi, J.T. Mason, J.D. Verhoeven, and
a Modern Metal. Aluminum from Within— fer, and F.S. Biancaniello, Metall. Trans. A, W. Kurz, Metall. Mater. Trans. A, Vol 22A,
the Sixth Edition, The Aluminum Associa- Vol 15, 1984, p 55 1991, p 2523–2533
tion and the Minerals, Metals, and Materials 15. D. Shechtman, R.J. Schaefer, and F.S. Bian- 28. H. Shibata, Y. Arai, and T. Emi, Metall. Ma-
Society, 1999 caniello, Metall. Trans. A, Vol 15, 1984, p ter. Trans. B, Vol 31B, 2000, p 981–991
3. A. Ohno, Solidification. The Separation 1987 29. M. Vandyoussefi, H.W. Kerr, and W. Kurz,
Theory and Its Practical Applications, 16. I.R. Hughes and H. Jones, J. Mater. Sci., Vol Acta Mater., Vol 48, 2000, p 2297–2306
Springer-Verlag, Berlin, 1987 11, 1976, p 1781 30. K. Tokieda, H. Yasuda, and I. Ohnaka, Ma-
4. R. Trivedi and W. Kurz, Solidification of 17. J.H. Perepezko and W.J. Boettinger, Mater. ter. Sci. Eng., Vol A262, 1999, p 238–245
Single-Phase Alloys, Casting, Vol 15, ASM Res. Soc. Symp. Proc., Vol 19, 1983, p 223 31. R. Trivedi, Metall. Trans. A, Vol 26A, 1995,
Handbook, D.M. Stefanescu, Ed., ASM In- 18. W.J. Boettinger, Rapidly Solidified Amor- p 1583
ternational, 1988, p 114 phous and Crystalline Alloys, B.H. Kear, 32. O. Hunziker, M. Vandyoussefi, and W. Kurz,
5. M. Schenck, referred to in Ref 3 B.C. Giessen, and M. Cohen, Ed., North Acta Mater., Vol 46 (No. 18), 1998, p 6325–
6. K.A. Jackson, Solidification Structures of Holland, 1982, p 15 6336
Pure Metals, Metallography and Micro- 19. N.H. Pryds and A.S. Pedersen, Metall. Ma- 33. R. Trivedi, A. Karma, T.S. Lo, P. Mazumder,
structures, Vol 9, ASM Handbook, ASM In- ter. Trans. A, Vol 33A, 2002, p 3755–3761 M. Plapp, and J.S. Park, 1998 Zermatt
ternational, 1985, p 607 20. D. Shechtman and E. Gutmanas, Prakt. Me- Workshop on “Solidification Microstruc-
7. W.J. Boettinger, Metall. Mater. Trans., Vol tallogr., Vol 18, 1981, p 587 tures,” M. Rappaz and R. Trivedi, Ed., CD.
5, 1974, p 2026 21. W. Kurz and D.J. Fisher, Fundamentals of Swiss Federal Institute of Technology Lau-
8. S. Akamatsu, G. Faivre, and Th. Ihle, Phys. Solidification, 3rd ed., Trans Tech Publica- sanne, Switzerland, 1998
Rev., Vol E 51, 1995, p 4751–4773 tions, Switzerland, 1989 34. J.W. Cahn, Metall. Trans., Vol 10A, 1979, p
9. S.N. Tewari, Y.-H. Weng, G.L. Ding, and R. 22. P. Magnin and W. Kurz, Solidification of 119–121
Trivedi, Metall. Mater. Trans. A, Vol 33A, Eutectics, Casting, Vol 15, ASM Handbook, 35. R.N. Grugel, T.A. Lagrasso, and A. Hellaw-
2002, p 1229–1243 D.M. Stefanescu, Ed., ASM International, ell, Metall. Trans., Vol 15A, 1984, p 1003–
10. R. Trivedi and K. Somboonsuk, Mater. Sci. 1988, p 119 1010
Eng., Vol 65, 1984, p 65–74 23. M. McLean, Directionally Solidified Mate- 36. B.K. Dhindaw, D.M. Stefanescu, A.K.
11. C.P. Hong and M.F. Zhu, The Science of rials for High Temperature Service, The Singh, and P.A. Curreri, Metall. Trans., Vol
Casting and Solidification, D.M. Stefa- Metals Society, London, 1983 19A, 1988, p 2839
nescu, R. Ruxanda, M. Tierean, and C. Ser- 24. S.V. Raj, I.E. Locci, J.A. Salem, and R.J. 37. I. Aoi, M. Ishino, M. Yoshida, H. Fukunaga,
ban, Ed., Editura Lux Libris, Brasov, Ro- Pawlik, Metall. Mater. Trans. A, Vol 33A, and H. Nakae, J. Cryst. Growth, Vol 222,
mania, 2001, p 110s–118s 2002, p 597–612 2001, p 806–815

View publication stats