Volume 09 - Metallography and Microstructures Part 15 doc

Shilling The decomposition of a supersaturated solid solution typically occurs by a sequence of reactions: αo →α'' + Guinier-Preston zones Metastable →α' + β' →αeq + βeqMetastable Stab

supersaturated matrix and spatially partitions the structure into transformed and untransformed regions; it is termed discontinuous precipitation Figures 10 and 11 depict the cellular reaction propagating into the supersaturated matrix from the grain boundaries The lamellar morphology of the transformation product is clearly revealed The cellular reaction often moves into a matrix in which a less stable transition precipitate has already precipitated The residual chemical-free energy drives the reaction front, and the duplex colonies consume the initial precipitate and produce a matrix of modified composition, as shown in Fig 12 and 13

Fig 10 Cellular or discontinuous precipitation growing out uniformly from the grain boundaries in an Fe-24.8Zn

alloy aged 6 min at 600 °C (1110 °F) 2% nital 1000× (W.C Leslie)

Fig 11 Cellular colonies growing out from grain boundaries in Au-30Ni alloy aged 50 min at 425 °C (795 °F)

50 mL 5% ammonium persulfate and 50 mL 5% potassium cyanide 100× (R.D Buchheit)

Fig 12 Transmission electron micrograph showing early stages of cellular reaction in a Cu-3Ti alloy aged 104 min at 375 °C (710 °F) The cellular product consumes the fine, coherent precipitates, which are revealed by strain contrast in the matrix 57,400× (J Cornie)

Fig 13 Cellular reaction in a Cu-4Ti alloy aged 103 min at 600 °C (1110 °F) The cellular reaction produces the lamellar equilibrium phase and leads to overaging and loss of ductility Diagonal band is an annealing twin in the matrix phase 1245× (A Datta)

Reference cited in this section

1 D.A Porter and K.E Easterling, Phase Transformation in Metals and Alloys, Van Nostrand Reinhold Co.,

1981

Precipitation Sequence

In many precipitation systems and in virtually all effective commercial age-hardening alloys, the supersaturated matrix transforms along a multistage reaction path, producing one or more metastable transition precipitates before the appearance of the equilibrium phase The approach to equilibrium is controlled by the activation (nucleation) barriers separating the initial state from the states of lower free energy The transition precipitate is generally crystallographically similar to the matrix, allowing the formation of a low energy coherent interface during the nucleation process Classical

nucleation theory shows that the nucleation barrier ∆G* is proportional to 3 2

M P G v G s

σ − ∆ + , where σM-P is the interfacial

energy of the matrix-precipitate interphase interface, ∆Gv is the thermodynamic driving force per unit volume of the

nucleus (which is proportional to the undercooling), and Gs is the strain energy per unit volume associated with the

coherency strains Because the nucleation rate is proportional to exp - ∆G*/kT (k is Boltzmann's constant, and T is the

absolute temperature), the transition phase nucleates more rapidly despite the smaller driving force (∆Gv) for its formation compared to the equilibrium precipitate Transmission electron microscopy reveals coherent transition precipitates formed during aging before the formation of the equilibrium phase (Fig 14 and 15)

Fig 14 Coherent transition precipitates revealed by strain contrast (dark-field) in transmission electron

microscopy The specimen is a Cu-3.1Co alloy aged 24 h at 650 °C (1200 °F) The precipitate is a metastable fcc phase of virtually pure cobalt in the fcc matrix The particles are essentially spherical, and the "lobe" contrast is characteristic of an embedded "misfitting sphere." This strain contrast reveals the particles indirectly through their coherency strain fields 70,000× (V.A Phillips)

Fig 15 Coherent (Co,Fe)3Ti metastable precipitates in a Co-12Fe-6Ti alloy aged 10 4 min at 700 °C (1290 °F) The ordered particles are imaged in dark-field transmission electron microscopy using an L12 superlattice reflection This imaging mode reveals the actual size of the particle, because the superlattice reflection stems only from the precipitate The precipitates are aligned along the <100> directions of the matrix The foil normal

is near [100] 60,000× (J.W Shilling)

The decomposition of a supersaturated solid solution typically occurs by a sequence of reactions:

αo →α'' + Guinier-Preston zones (Metastable)

→α' + β' →αeq + βeq(Metastable) (Stable equilibrium)

(Eq 1)

where αo is the supersaturated parent phase Each step in the precipitation sequence leads to a decrease in the free energy

and represents a state of metastable or stable equilibrium The (n + 1) transition phase tends to (but not exclusively) nucleate heterogeneously at the interphase boundaries of the nth transition phase This is due to the role of the interfaces in catalyzing the nucleation process and to the reduction of the available driving force resulting from the prior precipitation

of the nth transition phase The precipitation scheme can be depicted in a free-energy composition diagram, as shown in Fig 16 The metastable phases have corresponding solvus curves determined by the common tangent construction at each temperature The metastable solvi are included in the hypothetical phase diagram of Fig 17, which shows that the solubility is less the more thermodynamically stable the phase

Fig 16 Free-energy composition diagram showing the metastable and stable equilibria in the precipitation

sequence The points of common tangency at compositions C'', C', and Ceq are points on the metastable and stable solvi at this temperature

Fig 17 Hypothetical simple phase diagram showing the locus of metastable and stable solvus curves L, liquid

Guinier-Preston (GP) zones are coherent, solute-rich clusters resulting from phase separation or precipitation within a metastable miscibility gap in the alloy system They may form by homogeneous nucleation and grow at small undercoolings or by spinodal decomposition at large undercoolings or supersaturations (see the article "Spinodal Structures" in this Volume) After GP zone formation, the appearance of a more stable phase (for example, β' in Eq 1) leads to the dissolution of the zones, as revealed in Fig 18 Each successive step replaces the less stable phase by a more stable one, lowering the free energy In Fig 19, the equilibrium phase is shown growing by a cellular reaction into a metastable Widmanstätten structure

Fig 18 GP zones (dark spots) in matrix of an Al-15Ag alloy dissolving near plates of the more stable γ' precipitates (dark lines) Transmission electron micrograph of a specimen aged 1200 h at 160 °C (320 °F) 22,500× (J.B Clark)

Fig 19 Colonies of cellular precipitation reaction growing out and consuming a metastable Widmanstätten

precipitate in an Al-18Ag alloy aged 4 h at 300 °C (570 °F) 0.5% HF 1000× (J.B Clark)

Microstructural Features

The microstructures that evolve during the aging of a supersaturated solid solution are governed by the complex interplay

of thermodynamic, kinetic, and structural factors controlling the basic processes of nucleation, growth, and coarsening The precipitation system maximizes the rate of free energy release and not the overall free energy change as it decomposes toward the state of stable equilibrium Thus, coherent transition precipitates often appear in preference to the equilibrium phase, because of more favorable nucleation kinetics Precipitation of these metastable phases generally produces uniform, fine-scale microstructures that can enhance the physical and mechanical properties of commercial alloys The location of metastable solvus curves is essential to understanding and controlling the precipitation sequences

of age-hardening systems

The distribution and morphology of the precipitate phase depend on the nature of the active nucleation sites, the compromise between surface and strain energies, and the type of interphase interface that develops between the precipitate and matrix A two-phase mixture can also evolve during precipitation through a cooperative growth mechanism similar to the cellular phase separation in eutectic and eutectoid transformations This cellular precipitation reaction often leads to the formation of the equilibrium precipitate and subsequent degradation of such properties as strength and ductility Therefore, control of this reaction can be critical to optimizing properties in age-hardenable alloys Trace element additions have been used effectively to suppress the nucleation and growth of this microconstituent

The simplest phase transformation that can produce a spinodal reaction product is decomposition within a stable or

metastable miscibility gap, as shown in Fig 1 If a solid solution of composition C0 is solution treated in the single-phase

field at a temperature T0, then aged at an intermediate temperature TA (or TA'), the single-phase alloy tends to separate into

a two-phase mixture At the temperature TA, the compositions of the conjugate phases α1, and α2 under equilibrium

conditions are C1 and C2, respectively However, the supersaturated solid solution may decompose into two phases along two different reaction paths

Fig 1 Schematic showing miscibility gap in the solid state and spinodal lines (chemical and coherent)

At small undercoolings or low supersaturations (TA'), the solution is metastable; appearance of a second phase requires relatively large localized composition fluctuations This is the classical nucleation process, giving rise to "critical nuclei," which can grow spontaneously As the particles of the new phase grow by diffusion, the matrix composition adjusts

toward equilibrium At large supersaturations (TA), the solution is unstable, and the two-phase mixture gradually emerges

by the continuous growth of initially small amplitude fluctuations (see Fig 2) The rate of reaction is controlled by the rate of atomic migration and the diffusion distances involved, which depend on the scale of decomposition (undercooling) Therefore, spinodal structures refer to phase mixtures that derive from a particular kinetic process governing the initial stages of phase separation The "spinodal line" shown in Fig 1 is not a phase boundary but a demarcation indicating a difference in thermodynamic stability

Fig 2 Schematic illustrating two sequences for the formation of a two-phase mixture by diffusion processes:

nucleation and growth and spinodal decomposition (Ref 1)

Reference

1 J.W Cahn, Trans Met Soc AIME, Vol 242, 1968, p 166

Theory of Spinodal Reactions

The spinodal reaction is a spontaneous unmixing or diffusional clustering distinct from classical nucleation and growth in metastable solutions This different kinetic behavior, which does not require a nucleation step, was first described by Gibbs in his treatment of the thermodynamic stability of undercooled or supersaturated phases The spinodal line in Fig 1 indicates a limit of metastability with respect to the response of the system to compositional fluctuations The locus,

called the "chemical spinodal," is defined by the inflexion points of the isothermal free energy (G) composition curves

(∂2

G/∂C2 = 0) Within the spinodes where ∂2

G/∂C2 < 0, the supersaturated solution is unstable and spinodal decomposition can occur Spinodal decomposition or continuous phase separation involves the selective amplification of long wavelength concentration waves within the supersaturated state resulting from random fluctuations The transformation occurs homogenously throughout the alloy via the gradual buildup of regions enriched in solute, resulting

in a two-phase modulated structure The continuous amplification of a quasi-sinusoidal fluctuation depicted in Fig 2 is rather general, because this sinusoidal composition wave may be viewed as a Fourier component of an arbitrary composition variation that grows preferentially

The essential features of the spinodal process can be understood by considering this diffusional clustering as the inverse

of the homogenization of a nonuniform solid solution exhibiting a sinusoidal variation of composition with distance In

metastable solutions, the small deviations from the average concentration, C0, will decay with time according to the

equation ∆C = ∆C0 exp (-t/τ), where the relaxation time τ≈λ2

/ ∆; λ is the wavelength of the fluctuation and Dis the appropriate diffusion coefficient In a binary system D ∝ ∂2

G/∂C2, and within the spinodes ∂2

G/∂C2 < 0; that is, the curvature of the free energy-composition curve is negative Therefore, in an unstable solid solution D¨is negative, and

"uphill" diffusion occurs The amplitude of the concentration wave grows with time, that is, ∆C = ∆C0 exp (+ R(β)t), where the amplification factor R(β) is a function of the wave number β= 2π/λ The factor R(β) is a maximum for

intermediate wavelengths Long wavelength fluctuations grow sluggishly because of the large diffusion distances; short wavelength fluctuations are suppressed by the so-called gradient or surface energy of the diffuse or incipient interfaces that evolve during phase separation Therefore, the microstructure that develops during spinodal decomposition has a characteristic periodicity that is typically 2.5 to 10 nm (25 to 100 Ao ) in metallic systems

The factors controlling the spinodal reaction and resultant structures are clarified by examining the energetics of amplitude fluctuations in solid solutions The free energy of an inhomogenous solution expressed as an integral over the

small-volume, V, of the crystal can be written as:

of decomposition varies as K1/2(∆T)-1/2, where ∆T = TS - TA, in which TS is the spinodal temperature The coherency strain energy term is independent of wavelength, but can vary markedly with crystallographic direction in elastically anisotropic crystals Therefore, the dominant concentration waves will develop along elastically "soft" directions in anisotropic systems For most cubic materials, the <100> directions are preferred, although <111> waves are predicted in certain alloys, depending on the so-called anisotropy factor The strain energy can also stabilize the system against decomposition and effectively displace the spinodal curve (and the solvus), thus defining a "coherent spinodal" (Fig 1)

Periodic composition fluctuations in the decomposing solid solution cause diffraction effects known as "satellites" or

"sidebands." The fundamental reflections in reciprocal space are flanked by satellites or secondary maxima, and the distance of the satellites from the fundamental varies inversely with the wavelength of the growing concentration wave This diffuse scattering arises from the periodic variation of the lattice parameter and/or scattering factor The strain effects are negligible around the origin of reciprocal space Small-angle x-ray and neutron scattering can be used to study quantitatively the kinetics of the reaction by monitoring the changes in the intensity distribution around the direct beam due to changes in the structure factor modulations The electron diffraction pattern of a spinodally decomposed copper-titanium alloy shown in Fig 3 reveals the dominant <100> concentration waves that develop during the early stages of phase separation

Fig 3 [001] electron diffraction pattern from spinodally decomposed Cu-4Ti (wt%) alloy aged 100 min at 400

°C (750 °F) showing satellites flanking the matrix reflections (A Datta)

Microstructure

If the strain energy term in the free energy expression is negligible (small misfit) or if the elastic modulus is isotropic, the resultant microstructure will be isotropic, similar to the morphologies evolving in phase-separated glasses In Fig 4, an isotropic spinodal structure developed in a phase-separated iron-chromium-cobalt permanent magnet alloy is clearly revealed by transmission electron microscopy The two-phase mixture is interconnected in three dimensions and exhibits

no directionality The microstructure is comparable to the computer simulation of an isotropically decomposed alloy shown in Fig 5 (Ref 1) In Fig 6, the dominant composition waves have developed preferentially along the <100> matrix directions to produce an aligned modulated structure in a copper-nickel-iron alloy Because the homogenous phase separation process is relatively structure-insensitive, the spinodal product is generally uniform within the grains up to the grain boundaries, as revealed in the copper-nickel-chromium spinodal alloy shown in Fig 7

Fig 4 Transmission electron micrograph of isotropic spinodal structure developed in Fe-28.5Cr-10.6Co (wt%)

alloy aged 4 h at 600 °C (1110 °F) Contrast derives mainly from structure-factor differences 225,000× (A Zeltser)

Fig 5 Computer simulation of an isotropically decomposed microstructure (J.W Cahn and M.K Miller)

Fig 6 Spinodal microstructure in a 51.5Cu-33.5Ni-15Fe (at.%) alloy aged 15 min at 775 °C (1425 °F) revealed

by transmission electron microscopy Foil normal is approximately [001], and the alignment along the <100> matrix directions is apparent The wavelength of the modulated structure is approximately 25 nm (250 Ao ) 70,000× (G Thomas)

Fig 7 Transmission electron micrograph of spinodal microstructure developed in a 66.3Cu-30Ni-2.8Cr (wt%)

alloy during slow cooling from 950 °C (1740 °F) The microstructure is homogenous up to the grain boundary indicated by the arrow 35,000× (F.A Badia)

Atomic ordering and spinodal clustering can occur concomitantly in a precipitation system (see Ref 2 for a review of ordering and spinodal decomposition) In these systems, a supersaturated phase spinodally decomposes into two phases, one or both of which are ordered A transmission electron micrograph of a spinodally decomposed iron-beryllium alloy is shown in Fig 8, and a corresponding field-ion micrograph is shown in Fig 9 The brightly imaged phase in the electron micrograph (Fig 8) is the ordered phase (B2 superstructure), whereas the brightly imaged phase in the field-ion micrograph (Fig 9) is the iron-rich disordered phase The microstructure is periodic and aligned along the "soft" <100> directions

Fig 8 Spinodal structure aligned along <100> directions of decomposed Fe-25Be (at.%) alloy aged 2 h at 400

°C (750 °F) The bright phase is the Be-enriched ordered B2 structure revealed by clark-field imaging using a superlattice reflection; the dark phase is the Fe-rich disordered (or weakly ordered) transformation product The TEM foil normal is approximately [001] 200,000× (M.G Burke)

Fig 9 Field-ion micrograph of spinodally decomposed Fe-25Be (at.%) alloy aged 20 min at 400 °C (750 °F)

The axis of the needle-like specimen is [001] The iron-rich phase images brightly because of the different contrast mechanism operating in the field-ion microscope 375,000× (M.K Miller)

The spinodal mechanism provides an important mode of transformation, producing uniform, fine-scale, two-phase mixtures that can enhance the physical and mechanical properties of commercial alloys Spinodal decomposition has been particularly useful in the production of permanent magnet materials, because the morphologies favor high coercivities The structure can be optimized by thermomechanical processing, step aging, and magnetic aging Continuous phase separation or spinodal decomposition appears to be important in the classic Alnicos and copper-nickel-iron alloys, as well

as in the newly developed iron-chromium-cobalt materials

References cited in this section

1 J.W Cahn, Trans Met Soc AIME, Vol 242, 1968, p 166

2 W.A Soffa and D.E Laughlin, in Solid-Solid Phase Transformations, Proceedings of an International Conference, H.I Aaronson, D.E Laughlin, R.F Sekerka, and C.M Wayman, Ed., AIME, Warrendale, PA,

1982, p 159

Massive Transformation Structures

T.B Massalski, Professor of Metallurgical Engineering, Materials Science and Physics, Carnegie-Mellon University

Introduction

MASSIVE TRANSFORMATION may change crystal structure during heating or cooling if no change of composition occurs and if the rate of heating or cooling is rapid enough to allow only a limited amount of diffusion Massive transformation appears to proceed primarily by a noncooperative (ran-dom) transfer of atoms across the interfaces between the parent and product phases The details of the atomic movements at a transformation interface are not well understood, but the process does not appear to involve shearlike movements

Massive transformations are thermally activated and exhibit nucleation and growth characteristics The kinetics of these transformations are controlled primarily by interface diffusion and other interface features such as lack of coherence between the parent and product phases Growth of the product phase during massive transformations thus occurs mainly

by displacement of incoherent (high-energy) boundaries, often at speeds up to 10 to 20 mm/s (0.39 to 0.79 in./s) In these transformations, no simple orientation relationships are known to exist between the parent and product phases

The microstructure in a specimen that has undergone a massive transformation often exhibits massive patches of grains that have irregular boundaries These patches are surrounded by a mixture of planar and curving boundaries

Massive Transformations

The phase relations that are necessary for massive transformation to occur are illustrated in Fig 1 For alloys (Fig 1b to d), the two different crystal structures must be simple and stable or metastable at the same composition, but at different temperatures These conditions are also satisfied during allotropic transformations in pure metals (see Fig 1), which may occur by massive transformation Details of typical massive transformations for pure metals and binary systems are given

in Table 1

Table 1 Typical massive transformations

Temperature during quenching at which transformation occurs (a)

Alloy system or metal Amount of solute at which

transformation occurs (a) , at.%

Change in crystal structure (b)

(a) Values listed are approximate

(b) bcc, body-centered cubic; fcc, face-centered cubic; hcp, hexagonal close-packed

Fig 1 Schematic phase diagrams for (a) a pure metal and (b to d) three types of alloys that may undergo

massive transformations Critical compositions are indicated by the dashed vertical lines

Allotropy and Congruent Points. The compositions in Fig 1(a) and (b) correspond, respectively, to a pure metal that can exist in more than one allotropic form and an alloy in which the two phase fields touch at a congruent point (such

as the bcc and hcp phases in aluminum-silver at 24.5 at.% Ag) In these instances, the critical composition line does not cross a two-phase field, and thus a possible massive transformation is not interfered with by another transformation that might require long-range diffusion and solute partitioning Therefore, in such alloys, or in others for which the two-phase field is suitably narrow, a massive transformation is likely

The resulting microstructure is characteristic of a process of random and rapid growth (Fig 2) Microstructures of this type have been observed in iron, low-carbon steels, and low-nickel steels

Fig 2 Fe-0.002C alloy quenched in iced brine from 1000 °C (1830 °F) Microstructure, which resulted from a

massive transformation, shows ferrite grains with irregular boundaries Etchant: 2% nital 350×

Two-Phase Fields. Massive transformations that correspond to the compositions in Fig 1(c) and (d) usually occur in brass alloys of the β type as a result of decomposition of the high-temperature bcc phase Three competing transformations that may occur in alloys of this type upon cooling through the two-phase fields are (1) growth of Widmanstätten precipitates of the equilibrium phases adjoining the β phase, (2) equilibrium decomposition into two phases, and (3) a bainitic transformation At lower temperatures, the formation of martensite also may occur Therefore, the products of these other transformations frequently may be present together with the massive phase in the resulting microstructure

An example of a partial massive transformation of the bcc β phase in a specimen of Cu-19.3Al (at.%) that had initially consisted of an equilibrium two-phase mixture of α and β phases is shown in Fig 3 The massive transformation of β during cooling was arrested by the formation of martensite The irregular boundaries of the massive patches reveal the pattern of random growth (Ref 1)

Fig 3 Cu-19.3Al (at.%) alloy quenched in ice water from 900 °C (1650 °F) Irregularly shaped patches (light),

resulting from β-to-α massive transformation, are visible in a background of equilibrium α grains in martensite

5 g FeCl3, 15 mL HCl, 60 mL ethanol 135× (Ref 1)

When only a partial massive transformation has occurred on cooling, the remaining parent matrix is sometimes retained in

a metastable state In Fig 4, growth of the massive α phase has occurred at the boundaries of and inside the parent (bcc) grains Patches of the massive phase extend on both sides of a prior parent grain boundary, indicating that the growth of massive phase was unaffected by orientation relationships between the parent and product phases The lack of any simple orientation relationship across interfaces between massive and parent phases was confirmed further by electron diffraction studies (Ref 3) In a similar manner, the massive hcp phase grains in a copper-gallium-germanium alloy (Fig 5) cross the prior β (bcc) boundaries (Ref 4)

Fig 4 Cu-37.8Zn (at.%) alloy after a partial massive transformation Massive α phase (dark, mottled) has formed at the boundaries of and inside the parent grains of β phase Etchant: Same as Fig 3 40× (Ref 2)

Fig 5 Cu-18.4Ga-5Ge (at.%) alloy, quenched Massive hcp phase grains containing γ precipitation (rosettes) that formed in the prior-β phase before the massive transformation occurred 5 g FeCl3, 15 mL HCl, 60 mL ethanol 600× (Ref 4)

Feathery structures, a common transformation product in copper-gallium and silver-cadmium alloys, result from the formation of duplex fcc-hcp massive grains, each associated with a twin on the hcp (1011) plane A characteristic feature

of the feathery growth is that layers of fcc α terminal solid solution and hcp intermediate phase form alternately and share

a common close-packed plane as the plane of contact (Ref 5) Figure 6 shows that the feathery grains are again able to cross prior-β grain boundaries The duplex growth occurs through a two-dimensional nucleation and growth of close-packed planes (Ref 6) Figure 7, which shows a tip of an advancing duplex massive grain, reveals that the lamellae of alpha and the hcp phase originate at dislocations on the twin plane

Fig 6 Cu-21.5Ga (at.%) alloy quenched from β structure (temperature above 775 °C (1425 °F) Twinned

feathery grains formed by massive transformation, cross prior grain boundaries (arrows) revealed by αprecipitation Etchant: Same as Fig 5 250×

Fig 7 Thin-foil transmission electron micrograph of Cu-21.5Ga (at.%) alloy Tip of a feather unit shows fine

lamellae of α and hcp phases that originated at dislocations on twin plane (arrow) 66,000×

Growth Without Conventional Nucleation. Figure 8 illustrates the result of massive transformation in the β phase

of an initially two-phase hcp-bcc alloy The transformation of bcc beta to the hcp phase is accomplished by the growth of the original hcp phase into bcc β without formation of new incoherent nuclei The massive hcp phase has the orientation

of the equilibrium hcp phase, but with a different composition (Ref 7)

Fig 8 Cu-21Ga-1.5Ge alloy, quenched from a two-phase structure at 650 °C (1200 °F) During quenching, bcc

β phase at the grain boundaries transformed to massive hcp phase One hcp-bcc boundary was active at I, two were active at II, and three were active at III 1 g FeCl3,10 mL HCl, 100 mL H2O 250×

Growth of Single Crystals. Because a large driving force is required to nucleate incoherent grains, a single massive grain, once it is nucleated, may consume all prior parent grains if a temperature gradient is moved through the specimen

in a controlled manner as the boundary of the massive grain advances Figure 9 illustrates a single crystal that was produced by massive transformation of the β phase in a silver-aluminum alloy of congruent composition

Fig 9 Ag-24.5Al (at.%) alloy, consisting of a single crystal of hcp phase The crystal was formed from a

polycrystalline specimen of bcc phase by massive transformation during controlled cooling Etchant: Same as Fig 8 6× (J.H Perepezko)

Recent Developments. Detailed crystallographic orientation relationship studies and theoretical considerations (Ref 8) suggest that in almost all cases of nucleation during massive transformations some form of a rational, or nearly rational, orientation relationship appears necessary at the nucleation stage to reduce the activation energy for nucleation of the massive phase The concept of crystallographic relationships remains somewhat unsettled and has been reviewed recently at an AIME symposium on massive transformations At this symposium, the feasibility of the occurrence of a

metastable, compositionally invariant massive transformation in two-phase fields in phase diagrams was reexamined, and the current status of various distinguishing features of the massive transformation mode was also reviewed

References cited in this section

1 T.B Massalski, A.J Perkins, and J Jaklovsky, Extension of Solid Solubility During Massive

Transformations, Met Trans., Vol 3, 1972, p 687

2 T.B Massalski, Massive Transformations, in Phase Transformations, American Society for Metals, 1970

3 E.B Hawbolt and T.B Massalski, Observations Concerning the β→αm Massive Transformation in Cu-Zn

Alloys, Met Trans., Vol 1, 1970, p 2315

4 T.B Massalski, The Mode and Morphology of Massive Transformations in Cu-Ga, Cu-Zn, Cu-Zn-Ga and

Cu-Ga-Ge Alloys, Acta Metall., Vol 6, 1958, p 243

5 G.A Sargent, L Delaey, and T.B Massalski, Formation of "Feathery" Structures During Massive

Transformation in Cu-Ga Alloys, Acta Metall., Vol 16, 1968, p 723

6 H Gleiter and T.B Massalski, Atomistic Model for the Growth of Feathery Structures in Duplex Massive

Transformations, Acta Metall., Vol 18, 1970, p 649

7 A.J Perkins and T.B Massalski, Observations on the β→ζ m Massive Transformation in Two-Phase (β + ζ )

Alloys of the Cu-Ga-Ge System, Met Trans., Vol 2, 1971, p 2701

8 "The Massive Transformation," American Society of Metallurgical, Mining, and Petroleum Engineers

Symposium, published in Met Trans A, Vol 15, 1984, p 410

Table 1 Eutectoid transformations in nonferrous and ferrous alloys

Eutectoid temperature

Low-temperature phases and crystal structures

δ(deformed gamma brass)

Lamellar pearlite; granular pearlite

γ'-fcc (interstitial N) Lamellar pearlite; granular pearlite

Fe-O 23.3 O 560 1040 Wüstite cubic (NaCl) α-bcc

56 Zn 675 1247 β-cubic (CsCl) β1-tetragonal (CuAu)

γ(gamma brass) Lamellar pearlite

Fig 2 Eutectoid region of the Fe-Fe3C phase diagram

Fig 3 Nonlamellar eutectoid structure in a Cu-27Sn alloy Electrolytic etchant: 1% CrO3 150×

References

1 C.W Spencer and D.J Mack, Decomposition of Austenite by Diffusional Processes, V.F Zackay and H.I

Aaronson, Ed., Interscience John Wiley & Sons, 1962, p 549-606

2 N Ridley, in Phase Transformations in Ferrous Alloys, A.R Marder and J.I Goldstein, Ed., TMS/AIME,

Warrendale, PA, 1984, p 201-236

Pearlite Microstructures

Figure 4 depicts the individual constituents of pearlite A nodule nucleates at a grain boundary, triple point, grain corner,

or surface and grows radially until impingement occurs with surrounding nodules (Ref 4) Individual colonies are present inside the nodule, each nodule having an orientation relationship with the parent austenite grain Inside the colonies a complex microstructure forms, consisting of alternating parallel lamellae of the two product phases (ferrite and cementite) Figure 5 illustrates the typical microstructure of a partially transformed eutectoid steel in which the morphology of the nodules is apparent

Fig 4 Microstructural features of the pearlite eutectoid transformation (Ref 3)

Fig 5 Cross-sectional views of the microstructure of pearlite nodules in partially transformed hot-stage Fe-0.8C

specimens showing nodule formation Picral 220× (Ref 3)

Figure 6 shows the colonies of pearlite with faults and imperfections present in the lamellar structure Changes in orientation are observed at the boundaries between the colonies These changes occur because of faults in the structure Figure 7, a transmission electron micrograph, shows the cementite lamellae (dark rods) stopping at a cell boundary Also noted is the occasional bending of the lamellae achieved by a series of growth steps The growth of the cementite-ferrite interface, which is very sensitive to crystallographic relationships, occurs by a ledge-type mechanism (Ref 5) Figure 8 illustrates the ledges in cementite lamellae

Fig 6 Faults found in the colonies of pearlite in an Fe-0.8C specimen Picral 200×

Fig 7 A transmission electron micrograph of a ferrite cell interrupting the growth or bending the cementite in a

thin-foil Fe-0.8C specimen 17,250× (Ref 6)

Fig 8 A surface replica transmission electron micrograph showing growth steps on the cementite lamellae in an

Fe-0.8C specimen Picral 8000× (Ref 6)

References cited in this section

3 A.R Marder, in Phase Transformations in Ferrous Alloys, A.R Marder and J.I Goldstein, Ed., TMS/AIME,

Warrendale, PA, 1984, p 201-236

4 B.L Bramfitt and A.R Marder, Met Trans., Vol 4, 1973, p 2291-2301

5 S.A Hackney and G.J Shiflet, in Phase Transformations in Ferrous Alloys, A.R Marder and J.I Goldstein,

Ed., TMS/AIME, Warrendale, PA, 1984, p 237-242

6 B.L Bramfitt and A.R Marder, Metallography, Vol 6, 1974, p 483-495

Crystal Orientation

Pearlite colonies grow as if they were two interpenetrating single crystals Measurements of the orientation relationships

of the phases of ferrous pearlite colonies have been determined through transmission electron microscopy:

Pearlite Nucleation

Pearlite nucleation is heterogeneous and is generally restricted to the austenite grain boundaries and surface sites Saturation of these sites generally occurs within 20 to 25% of the total transformation time, followed by growth of the nodules until impingement Pearlite nucleates in a eutectoid alloy by the cementite or the ferrite nucleating on the austenite grain boundary (Fig 9a) This nucleus will form an orientation relationship with the prior-austenite grain (γ1) to lower the energy barrier If the first nucleus to form is cementite, the area surrounding this nucleus will be depleted of carbon, which enhances the formation of ferrite As the ferrite forms, the carbon is rejected into the surrounding matrix, further encouraging cementite to form

Fig 9 Nucleation and growth of pearlite (a) On a "clean" austenite grain boundary (1) Cementite nucleates on

grain boundary with coherent interface and orientation relationship with γ2 (2) α nucleates adjacent to cementite with a coherent interface and orientation relationship with γ1 In addition, this produces an orientation relationship between the cementite and ferrite (3) The nucleation process repeats sideways while incoherent interfaces grow into γ2 (4) New plates can also form by a branching mechanism (b) When a proeutectoid phase (cementite or ferrite) already exists on an austenite boundary, pearlite will nucleate and grow on the incoherent side, resulting in a different orientation relationship between the cementite and ferrite (Ref 7)

Hyper- and hypoeutectoid alloys decompose similarly The hyper- or hypoeutectoid composition causes proeutectoid ferrite (in a hypoeutectoid composition) or cementite (in a hypereutectoid composition) to form before the pearlite transformation In a hypereutectoid steel, for example, ferrite nucleates on the proeutectoid cementite (Fig 9b) and forms

an orientation relationship with the cementite A similar growth process occurs for the hypoeutectoid steel, with proeutectoid ferrite forming initially at prior-austenite grain boundaries

Reference cited in this section

7 D.A Porter and K.E Easterling, Phase Transformations in Metals and Alloys, Van Nostrand Reinhold, 1981,

p 331

Pearlite Growth

The growth rate of pearlite changes as a function of the time, transformation temperature, and prior-austenite grain size For a given temperature and austenite grain size, the transformation rate occurs in three stages As shown in Fig 10, at any given temperature the volume fraction of pearlite at any given time, f (t), fits an S-shaped or sigmoidal curve

Initially, the transformation rate is quite low and depends on site saturation As more nodules develop, the rate of transformation increases Finally, the nodules impinge, and the rate of transformation again slows as the microstructure gradually approaches complete transformation

Fig 10 Calculated fraction austenite transformed to pearlite as a function of time for the parameters shown f(t)

= 1 - exp [- πNG3t4/3], where f(t) is the volume fraction pearlite formed at any given time, t, at a given temperature, N is the nucleation rate of the pearlite colonies, and G is the rate at which the colonies grow into

the austenite (Ref 8)

The temperature at which the austenite is transformed also affects the pearlite growth rate Lowering the temperature increases the driving force for nucleation, which increases the transformation rate Figure 11 illustrates a C-curve relationship between transformation temperature and growth rate Finally, decreasing the austenite grain size will increase the number of nucleation sites More nuclei growing into the austenite decrease the time for transformation and increase the transformation rate

Fig 11 Variation in pearlite growth rate with transformation temperature (Ref 9)

References cited in this section

8 G Krauss, Principles of Heat Treatment of Steel, American Society for Metals, 1980

9 A.R Marder and B.L Bramfitt, Met Trans A, Vol 6, 1975, p 2009-2014

Interlamellar Spacing

Interlamellar spacing is a strong function of the transformation temperature Lower temperatures will result in a finer lamellar structure Figure 12 illustrates the relationship between the reciprocal of the interlamellar spacing and transformation temperature

Fig 12 Effect of transformation temperature on the reciprocal of interlamellar spacing (Ref 9)

Reference cited in this section

9 A.R Marder and B.L Bramfitt, Met Trans A, Vol 6, 1975, p 2009-2014

Alloying Effects

Substitutional alloying elements added to the Fe-C system affect all the transformation parameters The transformation temperature (Fig 13) and the eutectoid carbon content (Fig 14) are significantly altered Furthermore, alloying additions can significantly decrease the pearlite growth rate because of the partitioning of these elements between the ferrite and cementite (Fig 15)

Fig 13 Effect of percentage of substitutional alloying elements on the temperature of the eutectoid

transformation point in steel See also Fig 14 (Ref 10)

Fig 14 Effect of percentage of substitutional alloying elements on the carbon content of the eutectoid

transformation point in steel See also Fig 13 (Ref 10)

Fig 15 Pearlite growth rate of Fe-C-X alloys (Ref 11)

In an equilibrium alloy Fe-C-X, in which X is a substitutional element, decomposition of austenite to pearlite will occur in two ways First, substitutional elements will diffuse more slowly than carbon, which decreases the transformation rate This type of reaction partitions the substitutional element X between the two phases The carbide-forming elements, such

as chromium and molybdenum, will concentrate in the carbide Ferrite-stabilizing elements silicon, for example will concentrate in the ferrite Figure 16 depicts the concentration of chromium, manganese, and silicon in the ferrite and cementite The second type of decomposition in an Fe-C-X alloy occurs when the X alloy does not undergo any long-range diffusion The rate of the reaction, therefore, is controlled solely by the diffusion of carbon For example, nickel additions will stabilize the austenite to lower temperatures, causing high undercoolings; and preventing partitioning

Fig 16 Schematic distribution of the alloy elements across the microstructure (Ref 12)

References cited in this section

10 E.C Bain and H.W Paxton, Alloying Elements in Steel, American Society for Metals, 1962, p 112

11 A.R Marder and B.L Bramfitt, Met Trans A, Vol 7, 1976, p 902-906

12 P.R Williams, M.K Miller, P.A Beavan, and G.D.W Smith, in Phase Transformations, Vol 2, The

institution of Metallurgists, London, 1979, p 11.98-11 11.100

In many alloy steels, the bainitic temperature range is separate from the pearlite range, and a bay occurs in the isothermal time-temperature transformation (TTT) diagram between the two reactions Below the Bf (bainite finish) temperature, fully bainitic structures can be achieved, but at higher temperatures isothermal bainitic transformation may stop before complete decomposition of the austenite a phenomenon known as incomplete reaction The extent of decomposition is a

function of steel composition and reaction temperature and decreases as the isothermal reaction temperature is increased

to the point at which zero percentage transformation defines the Bs (bainitic start) temperature, above which bainite does not form (Fig 1) The Bs temperature corresponds approximately to the temperature of the bay in the TTT curve

Fig 1 Total dilation (proportional to the degree of completion of reaction) versus transformation temperature

during isothermal formation of bainite in 4340 steel (Ref 1)

Classical ferrous bainite consists of a nonlamellar aggregate of lath- or plate-shaped ferrite grains with carbide precipitation within the ferrite grains or in the interlath regions However, in some steels (for example, steels containing significant silicon content), carbide precipitation can be suppressed completely, although a lathlike ferritic product forms

in a manner identical in morphology and kinetics to the formation of classical upper bainitic ferrite Such special free structures are also generally referred to as bainitic

carbide-Several nonferrous alloys also produce microstructures similar to bainite in steels, either as nonlamellar aggregates of two phases or, frequently, with only one lathlike phase precipitating from the parent matrix, similar to the silicon steels mentioned above These nonferrous structures frequently are described as bainitic, although the implication that their formation occurs by a mechanism similar to that in steels is often disputed

Another important characteristic of bainite in ferrous and nonferrous alloys is that the formation of the bainitic ferrite plates results in surface relief that is indicative of a shape change accompanied by a significant shear component similar

to that found in martensite plates (Fig 2 and 3)

Fig 2 Hot-stage micrograph of surface relief from formation of upper bainite in nickel steel (0.15% C, 9.0%

Ni), austenitized, transformed at 400 °C (750 °F) Prepolished, not etched 360× (Ref 1)

Fig 3 Optical micrograph (left) and matching interferogram (right) of surface relief from formation of bainitic

plates in Cu-44.1Zn alloy, solution treated at 830 °C (1525 °F) for 2 min, quenched into 10% aqueous sodium hydroxide at 0 °C (32 °F), heat treated at 520 °C (970 °F) for 6 min Electropolished in ortho-H3PO4, not etched 3000× (Ref 2)

Numerous terms have been adopted to describe the more recognizable microstructures that frequently result from steels that are transformed in the bainitic temperature range Well-documented differences in the distribution of carbides formed

in the upper and lower portions of the temperature range, as well as evidence of different reaction kinetics, have led to the distinct classifications of upper and lower bainite These terms are generally adopted to describe the classical forms of steel microstructures

In hypereutectoid steels, the carbide phase can apparently precipitate first, and the resulting initial microstructural unit has been termed "inverse" bainite The bainite terminology also has been extended further to describe the more complex structures that are formed in alloy steels or that are formed under special experimental conditions

Upper bainite in hypoeutectoid steels comprises an aggregate of ferrite laths that usually are formed in parallel groups

to yield plate-shaped regions, often described as sheaves (Fig 4 and 5) Nucleation occurs predominantly at prior austenite grain boundaries The individual laths, or ferrite subunits, have similar orientations within a sheaf and usually are separated by low-angle grain boundaries (Ref 3, 4) These laths generally adopt an orientation relationship with the parent austenite (Ref 5, 6, 7), as given by Kurdjumov-Sachs and Nishiyama-Wassermann (Table 1)

Table 1 Orientation relationships for bainitic structures

Fig 4 Upper bainite in a 4360 steel specimen that was austenitized, isothermally transformed at 495 °C (925

°F), and quenched Picral 750×

Fig 5 Thin-foil transmission electron micrograph illustrating substructure of upper bainite plates in a 2340

steel, austenitized at 1095 °C (2000 °F) and isothermally transformed at 540 °C (1000 °F) for 15 h 6000× (Ref 3)

A common habit plane has not been identified, and {111}γ, {223}γ, {569}γ, and {0.37, 0.66, 0.65}γ have been reported (Ref 5, 6, 7, 15, 16), although there is greater agreement that the long direction of the laths follows <111>α The orientation relationship and habit plane may thus be irrational Decreasing the transformation temperature or increasing the carbon content decreases the widths of the individual laths and increases the amount of carbide precipitation The laths also have a high dislocation density, which increases with decreasing transformation temperature

Cementite (Fe3C θ-phase) usually is precipitated in the interlath regions and, at higher carbon contents, often forms nearly complete carbide films between the parallel ferrite laths to give the microstructure an almost lamellar appearance (Fig 6) The cementite spacing is generally larger than that in pearlite formed at the same temperature and results in different etching characteristics, thus enabling the two structures to be distinguished by optical (light) microscopy

Fig 6 Replica electron micrograph of upper bainitic microstructure in a high-carbon hypoeutectoid steel

(Fe-0.61C-0.53Mn-0.36Si-0.53Mo-0.0023B), austenitized and isothermally transformed at 500 °C (930 °F) 12,500× (Ref 4)

Silicon-containing steels are unique in that carbides do not form and the separately nucleated ferrite subunits are then divided by films of carbon-enriched retained austenite (Fig 7) (Ref 3, 17) The retained austenite can then be observed to contain a distribution of planar faults (Fig 8), analyzed to be deformation twins, probably produced to accommodate the transformation stresses The density of the twins increases upon subsequent deformation of the structure (Fig 9)

Fig 7 Thin-foil transmission electron micrograph of upper bainitic ferrite with interwoven laths of retained

austenite (gray phase) in a high-silicon steel (Fe-0.43C-3.0Mn-2.12Si), austenitized at 1200 °C (2190 °F) for 5 min and isothermally transformed at 350 °C (660 °F) for 205 min 22,000× (Ref 17)

Fig 8 Thin-foil transmission electron micrograph (dark-field image) of upper bainitic, retained austenite

illustrating faulted structure in a high-silicon steel (Fe-0.43C-3.0Mn-2.12Si) austenitized at 1200 °C (2190 °F) for 5 min and isothermally transformed at 350 °C (660 °F) for 205 min 20,000× (Ref 17)

Fig 9 Thin-foil transmission electron micrograph illustrating deformation twinning in retained austenite of

upper bainite microstructure in a high-silicon steel (Fe-0.4C-4.1.5N-2.01Si), austenitized at 950 °C (2120 °F) for 15 min and isothermally transformed at 400 °C (750 °F) for 1 h Specimen deformed to fracture 45,000× (Ref 18)

The stability of the retained austenite allows examination of the bainitic ferrite/austenite interface, and high-resolution images using weak-beam techniques indicate the presence of a closely spaced distribution of linear defects in the interface (Fig 10) Tempering these silicon steels at elevated temperatures leads to decomposition of the retained austenite and the precipitation of carbides, thus yielding structures more equivalent to classical upper bainite (Fig 11)

Fig 10 Thin-foil transmission electron micrograph (weak beam image) showing evidence of a distribution of

linear defects (see arrows) spaced 10 to 15 nm (100 to 150 Ao ) in the bainite/ferrite interface of a high-silicon steel (Fe-0.4C-4.15Ni-2.01Si), austenitized at 1100 °C (2010 °F) for 30 min and isothermally transformed at

370 °C (700 °F) for 8 h 580,000× (Ref 19)

Fig 11 Thin-foil transmission electron micrograph of a tempered upper bainite microstructure in a high-silicon

steel (Fe-0.43C-3.0Mn-2.12Si) that was austenitized at 1200 °C (2190 °F) for 5 min, isothermally transformed

350 °C (660 °F) for 205 min, and tempered at 500 °C (930 °F) for 120 min 37,000× (Ref 17)

The orientation relationships found between cementite and the bainitic ferrite are commonly those of Bagaryatski or Isaichev (see Table 1) and are consistent with carbide precipitation from the parent austenite The habit plane and long direction reported are (101)θ and [010]θ, respectively (Ref 20, 21), The orientation relationship between cementite and austenite is generally assumed to be that given by Pitsch see Table 1

Lower bainite. The transition between upper and lower bainite usually is reported to vary from approximately 550 °C (1020 °F) at low carbon contents to approximately 350 °C (660 °F) at 0.8% C As shown in Fig 12, it is paralleled by similar variations in the Bs and Ms (martensite start) temperatures (Ref 4, 22, 23)

Fig 12 Lower bainite start temperature (data from Ref 4) in relation to the Bs and Ms temperatures (Ref 23)

Lower bainite consists of heavily dislocated ferritic plates, rather than laths (Ref 24), and although the evidence is less substantial, the plates are probably comprised of a cluster of smaller ferrite subunits (Fig 13) as in upper bainite (Ref 3, 17) Nucleation of the plates occurs from prior austenite grain boundaries or from previously formed plates (Fig 14)

Fig 13 Thin-foil transmission electron micrograph showing the morphology at the tip of a lower bainitic ferrite

plate in a high-silicon steel (Fe-0.43C-3.0Mn-2.12Si), austenitized 1200 °C (2190 °F) for 5 min and isothermally transformed at 255 °C (495 °F) for 10 min 19,000× (Ref 17)

Fig 14 Lower bainite in a 4360 steel specimen, austentized, isothermally transformed at 300 °C (570 °F), and

quenched The matrix is untempered martensite Picral 500× (Ref 1)

The orientation relationship between lower bainite plates and parent austenite is close to Kurdjumov-Sachs or Wassermann (see Table 1) (Ref 17, 20, 24) Different habit planes of {496}γ and {254}γ have been reported (Ref 15, 24), from which it may be concluded that the orientation relationship and habit plane, as in the case of upper bainite, are likely

Nishiyama-to be irrational

Refinement of the plate structure at low temperature makes it difficult to differentiate between lower bainite and either upper bainite or tempered martensite using optical microscopy However, the most characteristic metallographic difference between classical upper and lower bainite is the distribution of carbides, which is readily apparent using electron microscopy In the lower bainitic microstructure, carbide precipitates are located within the ferrite plates (Fig 15), rather than between plates

Fig 15 Thin-foil transmission electron micrograph of lower bainite in 4360 steel that was austenitized,

isothermally transformed at 300 °C (570 °F), and quenched The dark bands in the bainite are cementite 15,000× (Ref 1)

The lathlike carbides typically adopt a unique habit plane variant in the ferrite, usually oriented at a characteristic angle of approximately 60° to the long axis of the bainitic plate This feature is in contrast to tempered martensitic structures, in which more than one variant is always observed Furthermore, lower bainitic ferrite does not contain transformation twins that are characteristic of martensite in medium- and high-carbon steels Lower bainitic carbides are identified as cementite

or ε-carbide (Fe2.4C) Generally, ε-carbide forms first and is then succeeded by cementite, as in the tempering of martensitic steels, provided that the carbon concentration is sufficiently high to overcome the energetically more favorable conditions for carbon atom segregation to dislocations (estimated to be ~0.55% C for bainitic structures) (Ref 23)

It has been shown (Ref 25) that ε-carbide occurs in lower bainite with an orientation relationship to the ferrite close to that proposed by Jack (see Table 1) Cementite has been reported (Ref 4, 15, 20, 25) to adopt the Bagaryatski or Isaichev orientation relationships, with a habit plane of (201)θ, or (213)α suggested in the case of the Isaichev relationship Jack and Bagaryatski orientation relationships are found for ε-carbide and cementite precipitates, respectively, in tempered martensite, in which the carbides have formed directly from supersaturated ferrite However, both relationships can be interpreted in terms of precipitation from austenite Moreover, the observation of a unique variant of the orientation relationship and a linear dispersion of the carbides within the plates (Fig 16) has been interpreted as evidence for precipitation in contact with parent austenite at the bainitic plate interface (Ref 14) However, a rational orientation relationship of:

(011)θ P{011}α [122]θ P<100>α

with up to four different variants of the cementite {011}α habit plane (Fig 17) has been found in a silicon-manganese steel (Ref 17) Because this unique orientation relationship cannot be combined with the Kurdjumov-Sachs relation to yield the anticipated three-phase α-γ-θ, relationship, it is consistent only with direct precipitation from supersaturated ferrite

Fig 16 Thin-foil transmission electron micrograph of cementite distribution within a lower bainite plate in a

plain-carbon steel (0.69% C), austenitized at 1200 °C (2190 °F) for 3 min and isothermally transformed at 300

°C (570 °F) The habit plane of the bainite plate is nearly parallel to the foil surface 9500× (Ref 22)

Fig 17 Thin-foil transmission electron micrograph (dark-field image) illuminating several cementite variants

precipitated intragranularly within a single plate of lower bainite in a high-silicon steel 2.12Si), austenitized at 1200 °C (2190 °F) for 5 min and isothermally transformed at 300 °C (570 °F) for 2 min 32,000× (Ref 17)

(Fe-0.43C-3.0Mn-References cited in this section

1 R.F Hehemann, Ferrous and Non-Ferrous Bainitic Structures, in Metals Handbook, Vol 8, 8th ed., Metallography, Structures and Phase Diagrams, American Society for Metals, 1973, p 194-196

3 J.M Oblak and R.F Hehemann, Structure and Growth of Widmanstätten Ferrite and Bainite, in

Transformation and Hardenability in Steels, Climax Molybdenum Co., Ann Arbor, MI, 1967, p 15-30

4 F.B Pickering, The Structure and Properties of Bainite in Steels, in Transformation and Hardenability in Steels, Climax Molybdenum Co., Ann Arbor, MI, 1967, p 109-129

5 G.V Smith and R.F Mehl, Lattice Relationships in Decomposition of Austenite to Pearlite, Bainite and

Martensite, Trans AIME, Vol 150, 1942, p 211-226

6 A.T Davenport "The Crystallography of Upper Bainite," Republic Steel Corp Research Center, Cleveland, Feb 1974

7 B.P.J Sandvik, The Bainite Reaction in Fe-Si-C Alloys: The Primary Stage, Met Trans A, Vol 13, 1982, p

777-787

8 G Kurdjumov and G Sachs, Z Physik, Vol 64, 1930, p 325-343

9 Z Nishiyama, Sci Rep Tôhoku Univ., Vol 23, 1934, p 637-664

10 G Wassermann, Arch Eisenhüttenwes., Vol 6, 1933, p 347-351

11 Yu A Bagaryatski, Doklady Akad Nauk SSSR, Vol 73, 1950, p 1161-1164

12 I.V Isaichev, Z Tekhn Fiziki, Vol 17, 1947, p 835-838

13 W Pitsch, Acta Metall., Vol 10, 1962, p 897-900

14 K.H Jack, J Iron Steel Inst., Vol 169, 1951, p 26-36

15 Y Ohmori, The Crystallography of the Lower Bainite Transformation in a Plain Carbon Steel, Trans Iron

Steel Inst Jpn., Vol 11, 1971, p 95-101

16 S Hoekstra, A Check of the I.P.S Theory with the Aid of an Accurate Determination of Habit Planes and

Orientation Relationships in Bainitic Steels, Acta Metall., Vol 28, 1980, p 507-517

17 H.K.D.H Bhadeshia and D.V Edmonds, Bainite Transformation in Silicon Steel, Met Trans A, Vol 10,

1979, p 895-907

18 V.T.T Miihkinen, Mechanical Properties of High Strength Bainitic Steels Containing Retained Austenite, doctoral dissertation, University of Oxford, 1983

19 S Cowley, private communication, University of Oxford, 1984

20 D.N, Shackleton and P.M Kelly, The Crystallography of Cementite Precipitation in the Bainite

Transformation, Acta Metall., Vol 15, 1967, p 979-992

21 Y Ohmori and R.W.K Honeycombe, The Isothermal Transformation of Plain Carbon Austenite, Trans Iron Steel Inst Jpn (Suppl.), Vol 11, 1971, p 1160-1164

22 R.F Mehl, The Physics of Hardenability The Mechanism and the Rate of the Decomposition of Austenite,

in Hardenability of Alloy Steels, American Society for Metals, 1939, p 1-54

23 H.K.D.H Bhadeshia, The Lower Bainite Transformation and the Significance of Carbide Precipitation,

Acta Metall., Vol 28, 1980, p 1103-1114

24 G.R Srinivasan and C.M Wayman, The Crystallography of the Bainite Transformation, Acta Metall., Vol

16, 1968, p 609-636

25 D.-H Huang and G Thomas, Metallography of Bainitic Transformation in Silicon Containing Steels, Met Trans A, Vol 8, 1977, p 1661-1674

Extensions of Bainite Terminology

Inverse Bainite. Bainitic structures are also produced in hypereutectoid steels In these materials, however, the carbide phase can nucleate first, thus leading to differences in the overall appearance of the microstructure compared to typical ferrous bainite (Ref 4, 26, 27) The initial cementite precipitates as a lath or plate, which then becomes engulfed by a sheath of ferrite This formation (Fig 18), also known as inverse bainite, then acts as the nucleus for adjacent austenite decomposition to produce larger ferrite laths and smaller cementite particles by a more classical bainitic reaction (Fig 19)

Fig 18 Replica electron micrograph showing the microstructural unit of inverse bainite comprising a single

cementite plate sheathed with ferrite in an Fe-1.34C alloy, austenitized at 1200 °C (2190 °F) for 15 min and isothermally transformed at 600 °C (1110 °F) for 2 s 17,000× (Ref 27)

Fig 19 Replica electron micrograph showing the evolution of a normal bainitic structure from initially formed

units of inverse bainite in an Fe-1.34C alloy, austenitized 1200 °C (2190 °F) for 15 min and isothermally transformed at 550 °C (1020 °F) for 7 s 7000× (Ref 27)

The initial structural unit of inverse bainite may also change to normal bainite during lengthening The limited ability of the inverse bainite unit to reproduce itself, due to the greater volume fraction and higher growth velocity of the ferrite regions, ensures that it generally occupies only a relatively small fraction of the total microstructure A significant proportion of the normal bainite fraction of the microstructure nucleates independently of the inverse bainite structure

Granular bainite refers to granular structures comprising classical bainite mixed with relatively coarse grains of polygonal and massive ferrite and regions of martensite and retained austenite (Fig 20) These granular structures are observed only in low- or medium-alloy steels and are most often produced by continuous cooling rather than isothermal treatment (Ref 28)

Fig 20 Replica electron micrograph of mixed microstructure with the original austenite grain boundaries

delineated by irregularly shaped particles, identified as two-phase austenite-martensite, but occasionally also containing ferrite and carbide Specimen was low-alloy steel (Fe-0.2C-1Cr-0.5Mo), austenitized and transformed during continuous cooling at 185 °C (335 °F) per minute No magnification given (Ref 28)

Columnar bainite refers to nonacicular ferritic grains containing cementite precipitates observed in medium-carbon steels partially transformed in the bainitic temperature range under very high pressures (Fig 21) However, similar structures have also been reported to occur in higher carbon steels (Ref 29)