Nucleation Processes in Recrystallization: History and Current Status Roger Doherty Dept of Materials Science & Engineering Drexel University.

Nucleation Processes in Recrystallization:

History and Current Status

Roger Doherty

Dept of Materials Science & Engineering

Drexel University

Topics• Importance of Recrystallization• Two Types of Annealing Processes - Gibbs Types

I and II• Nucleation in Recrystallization is different from

that in Phase Transformations - crystal structure remains. Selective growth.

• Current state of knowledge and challenges: Qualitative - descriptive success, but need to be quantitative and so predictive.

Some Personal Points• Studied Recrystallization since graduate research

1960-63. Whenever new techniques became available: Kossel (1970) EBSD (1985) OIM (1992), X-Ray Synchrotron methods …

• 1963-> studied solidification, (ingots, dendritic growth, spray forming) and then solid state precipitation and coarsening.

• Industrial Emphasis. Interaction of processes. • Reviewed both areas - inspired by Jack Christian

and mentored by Robert Cahn

Importance of Recrystallization

• 90% + of metals are “wrought products” Cast and thermo-mechanically worked to shape: Plate, Sheet, Wire, Forgings…

• Strain hardened - dislocation density increased by up to 103 x. ES = Gb2.

• Annealed - to soften the metal and to change its microstructure: Grain size, shape and texture.

Some Examples• Controlled Rolling of High Strength Low Alloy

Steels. Doubling of YS of Structural Steels and lowering of Ductile to Brittle Transition Temperature - by fine ferrite grain size. NbC type ppts.(fcc) in hot rolled Austenite inhibits recrystallization. from thin grains.

• - fiber, <111> // Rolling Plane, in recrystallized steels for deep drawing - auto bodies etc.

Ideal texture, also <111>, not yet produced in aluminum sheet ! Major challenge.

• Cube texture in recrystallized can alloy sheet: 3004: Al - 1%Mg - 1%Mn - 0.4%Fe. “Pressure Vessel for 10¢”.

• Mn and Fe - Hard particles that clean the wall ironing rings. Oxide causes damage to can wall.

• Mg promotes strain hardening (inhibits dynamic recovery) Higher strength and resists necking.

• 85 + % cold rolling for strength. But this rolling gives “- fiber” texture that forms 45°

“Ears” in drawn can. Now, controlled hot rolling and recrystallization to give high Cube texture,{001}<100> that has offsetting, 90°, Ears.

Origin of this cube RX texture ? Major dispute.

Earing after Cold Roll vs Earing as RX’d (CubeTexture)

Can incorrectly processed to have low cube texture (3mm) prior to cold rolling (300µm), drawn and redrawn with wall ironing (≈ 100µm).

> 5% 45° (to RD) “Ears” from the strong -fiber.

( Coherent Al3Zr dispersoids play a role in this) Improves fracture toughness of plate - tensile loaded

along plate. Why ?? Thin, unrecrystallized “pancaked”grains. So most grain boundaries, GB, parallel to plate -

GB embrittled by precipitates. However cracks that are normal to plane, and tensile

loads, find it difficult to grow. \ (Weak interfaces in fiber composites give improved

toughness)

Inhibition of Recrystallization during solution treatment of hot rolled 7xxx alloys.

• Unrecrystallized 7xxx thick plate, GB ppts formed both on slow quenching and on aging. These ppts embrittle the GB, but cause crack deflection under tensile loading in the plane of the plate.

Annealing of Deformed Metals• Recovery - dislocation rearrangement to form low

angle (sub) grain boundaries, LAGB. Very limited sub grain coarsening. LAGBs are very immobile. Why ?

• Recrystallization - “Nucleation” and growth of new, low dislocation density, grains. Migrating GB are always high angle ( > 15/20°)

• Grain Coarsening - reduction of GB area. Usually “Normal” with uniform size distribution, but sometimes “Abnormal” Growth of one or two grains into a grain structure that is anchored (by particles)

Gibbs: Two Transformations• Type I - “Heterogeneous” “Discontinuous”

“Nucleation and Growth” - Transformation is complete within defined regions separated by a boundary from untransformed region. Most Common

• Type II “Homogeneous” “Continuous” Transformation partially complete every where. As in Spinodal Decomposition

Annealing processes

• Recrystallization and Abnormal Grain Coarsening (AGC) are Type I - “Nucleation and Growth”.

AGC is sometimes called “Secondary Recrystallization”

• Recovery and Normal Grain Coarsening (NGC) are Type II - Homogeneous.

Recrystallization of Aluminum, 40% compressed (normalto image) and annealed.Nucleation and Growth.

Numbered new grains in a large grain fragmented into two regions, A and B, misoriented by 40°. White grains have near A and dark grains have near B orientations.(Bellier by Kossel - 1971)

Recovery. Aluminum deformed 10% and annealed in situ.a) Deformed b) Annealed for 2mins at 250°C. HVEM.

Humphreys and Hatherly (1995)

Normal Grain Coarsening in Ti

Abnormal Grain Coarsening in Al-Al3Fe

Abnormal subgrain growth. “Nucleation”

in Aluminum 60% compressed and briefly (2mins)

annealed at 250°C.

HAGB between 1 and 13,14,15. LAGB

between 1 and 6,7,11,12.

Faivre and Doherty (1979)

Aluminum 20% compressed and annealed. New grains 1, 2, 3at HAGB.

LAGB with regions 4-9 in lower grain.

“Strain Induced Boundary Migration”

Bellier 1971

Why Abnormal not Normal Coarsening?

In recrystallization, after recovery to sub grain structure, most boundaries are LAGB, << 10-15°. These are not mobile. Cottrell 1953. Frequently confirmed qualitatively and semi-quantitatively.

V = M GV M is the mobility. Why are these LAGB immobile ? Later Only a sub-grain with a HAGB can grow.

With very heavy strain ( > 10), 80% HAGB. Then annealing gives NGC, the Type II process.

Abnormal vs Normal Coarsening

AGC of grains: Particles (Al3Fe) anchor GB and inhibit NGC. However at low particle density, and small grain size,alloys with particles, often show AGC.

Al-Mn ( Beck et al 1948) Al-Cu ( Calvet and Renon 1960). AGC just below Solvus

But how/why do giant grains have mobile GB ?

Nucleation in RX different from that in PT

• Driving pressure GV much smaller, J/mole not kJ/mole.

• Comparable interfacial energy = 0.33J/m2.• But “new” grain structure (fcc crystal) is present.

So it does not have to be grown, atom by atom, as in nucleation theory.

L-> S, .• Just selected “abnormal” growth of a sub grain.• Obvious in AGC, grains do not “nucleate” they

are already there - 10+ even 100+µm!

Nucleation Theory in RX. r* = 2 / GV (1) Still applies.

With = 0.35 and GV = 0.25 MJ/m3.

r* ≈ 3 µm. ( VM = 10-5 m3/mole) Grains, with HAGB, smaller than this r, vanish. Atom by atom build up to the critical nucleus:

G* = 16 π 3 / GV2 (2)

n* = NV exp {- G*/ kT) (3) In RX, eqs. 2 & 3 are irrelevant ! All the potential nuclei are there already. So, no need for them to be formed from zero size.

Nucleation in RX

• Cahn (1948) Beck (1948) from cells/subgrains• Cottrell (1953) LAGB immobile so only at

HAGB. Cahn-Beck-Cottrell model• Hu (1959) Walter & Dunn (1959) Nucleation in

Misoriented Transition Bands in Fe-Si,single crystals. Coalescence of sub-grains ?

• Sussex (1969-79) Kossel and Kikuchi studies. CBC fully supported. Now the orthodox view.

• Humphreys and Hatherly RX Textbook (1995)• Problem Solved ? Not yet:-

Problem I • Recent observations - Poulsen et al (Riso)• X-Ray Synchrotron studies find new grain

orientations in deformed Al - annealed in situ that, though close to the observed deformed sub-grain orientations, were not found before annealing.

• Experimental error or new mechanism? (Large sub grain rotation) ??

• Scientific conservatism ?

Problem II

• Need for Predictive Model• Kocks ( 1994) “From a given microstructure and

known deformation and annealing condition can we predict the recrystallized microstructure needed.” ?????

• We need the nucleation density, the sites and orientations of the new grains, and their growth rates.

• Shuey (1998) “How to model RX ?” & “The answer that we can’t - is not acceptable” !

• Why so difficult?

Predictive Model of Nucleation ?• Abnormal growth of subgrains - so we may need

not just average features of the deformed microstructure but its heterogeneities - Transition bands etc.

• Cottrell “Strain hardening was the first problem tackled by dislocation theory and may be the last one to be solved” (TEM ?)

• Polycrystalline metal plasticity is a very active area of mechanical/materials study…. (OIM)

• Kalidindi, Doherty et al. at Drexel (Proc Roy Soc A460 (2004) 1935-56)

Problems for Predictive Model-cont. • Modeling “nucleation” as abnormal growth of sub

grains needs:• Detailed local microstructure - currently by

observation (TEM, OIM, with automated serial sectioning or statistical methods, ideally by 3D X-Ray..). In longer term by plasticity modeling.

• Energy (OK) and mobility (??) of low, medium and high angle grain boundaries !

• Transient “solute drag”. Solute drag theories currently steady state - New GB are clean but pick up solute. Necker. HP copper, 90%, grains grew rapidly to 15µm then almost stop growing.

Problems for Predictive Model-cont

• Orientation Pinning (Juul-Jensen (1995) Doherty et al.(1995) ). In heavily rolled metals bands of deformed orientation - in fcc (- fiber) 90% by volume. So often “- fiber” new grains, growing in the direction normal to rolling plane, meet deformed bands with similar orientation, so reforming LAGB that are immobile.

• Shown to be important in RX texture modeling. Oriented Nucleation or Growth of Cube texture.

Doherty et al Mater Sci & Eng 257A (1998) 18-36

Why are LAGB immobile?• Arrays of dislocations - so to move they must

glide and climb/cross slip. The later two require thermal activation. For climb: self diffusion. (Viswanathan and Bauer 1973)

• Motion of dislocations must cause a shape change. A problem for an array of sub grains

• 3D X-ray study of recovery. (Gundlach et al. Scripta M. 50 (2004) 477-81). Indications of strain controlled sub boundary migration. (a) Initial rapid coarsening then near stability. (b) Limited individual growth - but a big cell shrank and small ones grew ! Local elastic strains that relax quickly ?? Very speculative !

Conclusions

• “Nucleation” different in RX than in PT• Growth of selected sub-grains (1 in 106)• Need HAGB ( local misorientation) and

size advantage. Role of Coalescence ?• Vital role of as-deformed microstructure• Currently by microscopy - need for

plasticity modeling• More insight needed on GB properties.• After 45 years still a fascinating subject.

Jack Nutting (1978) “ Recrystallization a great subject for

academic research - there will always be enough problems to last until retirement!”

Jack Christian (1985) “Retirement just meant that Oxford

University stopped paying me”