Multiscale Modeling of Theta ' Precipitation in Al-Cu Binary Alloys
Multiscale Modeling of Theta ' Precipitation in Al-Cu Binary Alloys
Multiscale Modeling of Theta ' Precipitation in Al-Cu Binary Alloys
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
V. Vaithyanathan et al. / Acta Materialia 52 (2004) 2973–2987 2979<br />
<strong>Al</strong> 2 <strong>Cu</strong> (θ')/<strong>Al</strong> Supercell Energetics<br />
Fig. 5. Relaxed supercells from the first-pr<strong>in</strong>ciples <strong>in</strong>terfacial energy<br />
calculations <strong>of</strong> coherent (1 0 0) and semi-coherent (0 0 1) <strong>in</strong>terfaces <strong>of</strong><br />
h 0 -<strong>Al</strong> 2 <strong>Cu</strong> <strong>in</strong> fcc <strong>Al</strong> solid solution [10]. Dashed l<strong>in</strong>es <strong>in</strong>dicate the<br />
1a h<br />
0 ¼ 1a <strong>Al</strong> and 2c h<br />
0 ¼ 3a <strong>Al</strong> relationships <strong>of</strong> the coherent and semicoherent<br />
<strong>in</strong>terfaces, respectively.<br />
and a semi-coherent rim. We construct supercells consistent<br />
with the observed orientation relations between<br />
h 0 and the <strong>Al</strong> matrix: ð001Þ h<br />
0kf001g <strong>Al</strong> and ½010Š h<br />
0<br />
k½010Š <strong>Al</strong><br />
[4]. Representative cells show<strong>in</strong>g the <strong>in</strong>terfacial<br />
structures are given <strong>in</strong> Fig. 5. <strong>Al</strong>though the present<br />
phase-field model is 2D, we note that the first-pr<strong>in</strong>ciples<br />
calculations used to generate the various energetics <strong>of</strong><br />
the h 0 /<strong>Al</strong> system are fully three-dimensional. The coherent<br />
and semi-coherent <strong>in</strong>terfaces possess very different<br />
<strong>in</strong>terfacial structures. Due to lattice-misfit arguments<br />
(see below), the semi-coherent <strong>in</strong>terface structure is<br />
found to have a 2 h 0 to 3 <strong>Al</strong> SS unit cell arrangement. This<br />
configuration was proposed by Stobbs and Purdy [8]<br />
and confirmed by their TEM stra<strong>in</strong> field observations<br />
around the <strong>in</strong>terface.<br />
From Fig. 6, we see that the calculated T ¼ 0 K <strong>in</strong>terfacial<br />
energies from first-pr<strong>in</strong>ciples LDA calculations<br />
<strong>of</strong> the coherent and semi-coherent <strong>in</strong>terfaces are 190 and<br />
600 mJ/m 2 , respectively. GGA calculations, give slightly<br />
lower values <strong>of</strong> 170 and 520 mJ/m 2 , respectively. (The<br />
LDA numbers given here are slightly smaller than previous<br />
values published <strong>in</strong> [10] due to a more careful<br />
consideration <strong>of</strong> k-po<strong>in</strong>t convergence.) From our convergence<br />
studies, we estimate an uncerta<strong>in</strong>ty <strong>in</strong> these<br />
calculated <strong>in</strong>terfacial energies on the order <strong>of</strong> 5–10% due<br />
to supercell size. Interest<strong>in</strong>gly, for both LDA and GGA,<br />
the <strong>in</strong>terfacial anisotropy between semi-coherent and<br />
coherent <strong>in</strong>terfaces is consistently around 3. In addition<br />
to the isolated <strong>in</strong>terface energies, one can obta<strong>in</strong> some<br />
<strong>in</strong>dication <strong>of</strong> the <strong>in</strong>terface-<strong>in</strong>terface <strong>in</strong>teractions from<br />
the energies <strong>in</strong> Fig. 6 for relatively small supercells. For<br />
the coherent <strong>in</strong>terface, the small-cell energies fall above<br />
the l<strong>in</strong>e extracted from large cells, thus <strong>in</strong>dicat<strong>in</strong>g a repulsion<br />
between these <strong>in</strong>terfaces at short distances. For<br />
Formation Energy (meV/atom)<br />
140<br />
120<br />
100<br />
80<br />
60<br />
40<br />
20<br />
SEMICOHERENT (θ' rim)<br />
LDA: γ = 600 mJ/m 2<br />
GGA: γ = 520 mJ/m 2<br />
COHERENT (θ' broad face)<br />
LDA: γ = 190 mJ/m 2<br />
GGA: γ = 170 mJ/m 2<br />
0<br />
0.00 0.02 0.04 0.06 0.08 0.10 0.12<br />
1/N<br />
Fig. 6. First-pr<strong>in</strong>ciples (VASP) formation energies <strong>of</strong> <strong>Al</strong>/h 0 N-atom<br />
supercells as a function <strong>of</strong> 1=N for both <strong>in</strong>terfaces shown <strong>in</strong> Fig. 5.<br />
Energetics are shown for both LDA (filled symbols) and GGA (empty<br />
symbols) calculations. The energies <strong>of</strong> the large-cell calculations are fit<br />
to straight l<strong>in</strong>es, and the <strong>in</strong>terfacial energies (r) are extracted from the<br />
slopes, 2rA, <strong>of</strong> these l<strong>in</strong>es [see Eq. (18)].<br />
the semi-coherent <strong>in</strong>terface, the opposite is true, i.e.,<br />
these <strong>in</strong>terfaces tend to attract one another at short<br />
separations.<br />
We can contrast our first-pr<strong>in</strong>ciples calculated <strong>in</strong>terfacial<br />
anisotropy <strong>of</strong> 3 with a previous estimate by<br />
Aaronson and Laird [28] <strong>of</strong> 12. Our first-pr<strong>in</strong>ciples<br />
calculations <strong>in</strong>clude more physical contributions, and<br />
hence, are more predictive and certa<strong>in</strong>ly more accurate<br />
than the previous highly simplified estimate [28].<br />
Therefore, we assert that the currently most reliable<br />
value <strong>of</strong> the <strong>in</strong>terfacial anisotropy for the h 0 /<strong>Al</strong> system is<br />
3. This is noteworthy, s<strong>in</strong>ce the anisotropy estimate <strong>of</strong><br />
12 has been widely used <strong>in</strong> the literature as a prediction<br />
<strong>of</strong> the equilibrium aspect ratio <strong>of</strong> h 0 . These estimates <strong>of</strong><br />
the equilibrium aspect ratio are flawed not only because<br />
we have shown the more accurate <strong>in</strong>terfacial anisotropy<br />
is 3, but also there exists a strong stra<strong>in</strong> anisotropy<br />
contribution <strong>in</strong> this system (discussed <strong>in</strong> the next section),<br />
which can significantly alter the equilibrium aspect<br />
ratio <strong>of</strong> h 0 precipitates.<br />
We note that the <strong>in</strong>terfacial energy anisotropy obta<strong>in</strong>ed<br />
from first-pr<strong>in</strong>ciples is obta<strong>in</strong>ed at T ¼ 0 K and<br />
for a completely sharp <strong>in</strong>terface (Fig. 5). It would be <strong>of</strong><br />
considerable <strong>in</strong>terest to know how this anisotropy<br />
changes with temperature. At f<strong>in</strong>ite temperature, the<br />
<strong>in</strong>terfaces will naturally be diffuse to some extent and<br />
hence, configurational degrees <strong>of</strong> freedom will alter the<br />
<strong>in</strong>dividual <strong>in</strong>terfacial energy values. <strong>Al</strong>so the vibrational<br />
entropy at the <strong>in</strong>terface should be considered <strong>in</strong> a<br />
complete description <strong>of</strong> the temperature-dependence <strong>of</strong><br />
the <strong>in</strong>terfacial free energies. Future work along these<br />
l<strong>in</strong>es would be most <strong>in</strong>terest<strong>in</strong>g.