Available online at www.sciencedirect.

com

Acta Materialia 60 (2012) 3643–3654

www.elsevier.com/locate/actamat

Effect of Er additions on ambient and high-temperature strength

of precipitation-strengthened Al–Zr–Sc–Si alloys
Christopher Booth-Morrison a, David N. Seidman a,b, David C. Dunand a,⇑
a
Department of Materials Science and Engineering, Northwestern University, Evanston, IL 60208-3108, USA
b
Northwestern University Center for Atom-Probe Tomography (NUCAPT), Evanston, IL 60208-3108, USA

Received 25 July 2011; received in revised form 14 February 2012; accepted 16 February 2012
Available online 7 April 2012

Abstract

The effect of substituting 0.01 at.% Er for Sc in an Al–0.06Zr–0.06Sc–0.04Si (at.%) alloy subjected to a two-stage aging treatment
(4 h/300 °C and 8 h/425 °C) is assessed to determine the viability of dilute Al–Si–Zr–Sc–Er alloys for creep applications. Upon aging,
coherent, 2–3 nm radius, L12-ordered, trialuminide precipitates are created, consisting of an Er- and Sc-enriched core and a Zr-enriched
shell; Si partitions to the precipitates without preference for the core or the shell. The Er substitution significantly improves the resistance
of the alloy to dislocation creep at 400 °C, increasing the threshold stress from 7 to 10 MPa. Upon further aging under an applied stress
for 1045 h at 400 °C, the precipitates grow modestly to a radius of 5–10 nm, and the threshold stress increases further to 14 MPa. These
chemical and size effects on the threshold stress are in qualitative agreement with the predictions of a recent model, which considers the
attractive interaction force between mismatching, coherent precipitates and dislocations that climb over them. Micron-size, intra- and
intergranular, blocky Al3Er precipitates are also present, indicating that the solid solubility of Er in Al is exceeded, leading to a
finer-grained microstructure, which results in diffusional creep at low stresses.
Ó 2012 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.

Keywords: Aluminum alloys; Precipitation; Scandium; Zirconium; Erbium

1. Introduction [1]. This core/double-shell structure is formed as solute ele-

ments precipitate sequentially, according to their intrinsic
Dilute Al–Zr–Sc–Er alloys have recently been diffusivities D in the aluminum matrix, where DEr >
demonstrated to offer an attractive combination of high- DSc > DZr. In the absence of Er, dilute Al–Sc–Zr alloys
temperature creep resistance, coarsening resistance and form coarsening-resistant L12-ordered precipitates, which
affordability, the latter because they are castable and consist of a Sc-enriched core surrounded by a Zr-enriched
heat-treatable [1]. These alloys are excellent candidates to shell [1–15]. The substitution of some of the Sc with the
replace much heavier (but less costly) iron-based alloys much lower-cost Er has proved effective in improving the
and heavier (and much costlier) titanium alloys for auto- creep resistance of Al–Sc alloys at 300 °C, while maintain-
motive and aerospace parts exposed to intermediate tem- ing their coarsening resistance [16–18]. Erbium is the least
peratures (250–400 °C) and low stresses (<20 MPa). With expensive of the heavy rare-earth elements (RE) that form
appropriate aging treatments, Al–Zr–Sc–Er alloys form L12 precipitates in aluminum [19], and Al3Er (L12) has the
coherent, spheroidal, L12-ordered precipitates that remain largest lattice parameter misfit with Al (d = +4.08% at
coarsening resistant for >64 days at 400 °C, owing to a room temperature) [20], thereby improving creep resistance
structure consisting of an Er-enriched core surrounded by by enhancing elastic interactions with dislocations
a Sc-enriched inner shell and a Zr-enriched outer shell [17,21,22].
It was shown previously that an alloy composition of Al–
⇑ Corresponding author. 0.06Zr–0.04Sc–0.02Er (at.%) provides a desirable combina-
E-mail address: dunand@northwestern.edu (D.C. Dunand). tion of strength, coarsening resistance and affordability [1].

1359-6454/$36.00 Ó 2012 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.actamat.2012.02.030
3644 C. Booth-Morrison et al. / Acta Materialia 60 (2012) 3643–3654

The Al–Zr–Sc–Er alloys in this previous study were, how- emission spectroscopy (DCPMS) (ATI Wah Chang,
ever, found to have intragranular Al3Zr flakes retained Albany, OR). The impurity Fe concentration was 0.006
from the melt as a result of incomplete dissolution of the at.% for both alloys.
Al–Zr master alloy, and intergranular Al3Er (L12) precipi- The cast alloys were homogenized in air at 640 °C for
tates formed during solidification and/or homogenization, 72 h and then water quenched to ambient temperature. A
because the limited matrix solubility of Er was exceeded two-stage aging treatment of 4 h at 300 °C followed by
[1]. These particles lowered the effective solute concentra- 8 h at 425 °C, denoted in the following as peak-aging,
tion of the alloys, reducing the precipitate volume fraction, was employed to achieve peak strength and coarsening
and limiting alloy strength. These issues are addressed here resistance, as explained above. The second-stage tempera-
by induction-melting of the alloys to ensure complete disso- ture of 425 °C was selected so that the final aging temper-
lution of Al3Zr present in the master alloy, and by employ- ature was higher than the creep testing temperature of
ing a lower Er concentration of 0.01 at.% to compare with 400 °C.
the base Al–0.06Zr–0.06Sc–0.04Si alloy. Silicon was added
intentionally to accelerate the precipitation kinetics of Sc, so 2.2. Microstructure observations
that peak strength from Al3Sc (L12) precipitates at 300 °C
could be achieved after only 4 h [23]. A two-stage aging The microstructures of samples polished to a 1 lm sur-
treatment of 24 h at 300 °C, followed by 8 h at 400 °C, face finish were imaged by SEM using a Hitachi S3400N-
was previously shown to provide peak strength in Al– II microscope equipped with an Oxford Instruments
0.06Zr–0.06Sc and Al–0.06Zr–0.04Sc–0.02Er (at.%), where INCAx-act detector for energy-dispersive X-ray spectros-
Si was not controlled, but present naturally at <0.005 at.% copy (EDS). Polished specimens were then etched for
[1]. 30 s, using Keller’s reagent to reveal their grain boundaries.
The focus here is on the mechanical properties of Al– Vickers microhardness measurements were performed on a
0.06Zr–0.06Sc–0.04Si and Al–0.06Zr–(0.05Sc–0.01Er)– Duramin-5 microhardness tester (Struers), using a 200 g
0.04Si (at.%). at ambient and elevated temperature to load applied for 5 s on samples polished to a 1 lm surface
determine the viability of these modified compositions finish. Fifteen indentations were made per specimen across
(additional Si and reduced Er compared with Al–0.06Zr– several grains.
0.04Sc–0.02Er (at.%) [1]) in high-temperature, low-stress, Specimens for three-dimensional local-electrode atom-
applications. The effect of substituting 0.01 at.% Er for probe (LEAP) tomography were prepared by cutting
Sc on the microstructure, precipitate nanostructure and blanks with a diamond saw to dimensions of 0.35 
creep resistance of the alloys following aging to peak ambi- 0.35  10 mm3. These were electropolished at 8–20 V DC
ent temperature strength is studied by scanning electron using a solution of 10% perchloric acid in acetic acid, fol-
microscopy (SEM), atom-probe tomography (APT), lowed by a solution of 2% perchloric acid in butoxyethanol
microhardness and creep testing. at room temperature. Pulsed-voltage APT was performed
with a LEAP 4000X Si X tomograph (Cameca, Madison,
2. Experimental procedures WI) [24–29] at a specimen temperature of 35 K, employing
a pulse repetition rate of 250 kHz, a pulse fraction of 20%
2.1. Alloy compositions and processing and an evaporation rate of 0.04 ions per pulse. LEAP
tomographic data were analyzed with the IVAS 3.4.1 soft-
Alloys with nominal compositions of Al–0.06Zr–0.06Sc– ware program (Cameca). The matrix/precipitate hetero-
0.04Si and Al–0.06Zr–(0.05Sc–0.01Er)–0.04Si (at.%) (Al– phase interfaces were delineated with Al isoconcentration
0.2Zr–0.1 Sc–0.04Si and Al–0.2Zr–(0.08Sc–0.06Er)–0.04Si surfaces, and compositional profiles were obtained by the
wt.%) were inductively melted to a temperature of 900 °C proximity histogram (proxigram) methodology [30,31].
from 99.99 at.% pure Al, 99.995 at.% Si, and Al–0.96 The measurement errors for all quantities were calculated
at.% Sc, Al–3 at.% Zr and Al–78 at.% Er master alloys. based on counting statistics and standard error propaga-
The two alloys, referred to in this paper as Er-free and tion techniques [32].
Er-containing alloys, were cast into a cast-iron mold pre- Previous attempts to measure Si concentrations in Al by
heated to 200 °C. Their compositions in the as-cast state LEAP tomography resulted in measured values that are
are given in Table 1, as measured by direct current plasma smaller than both the expected nominal value and the value

Table 1
Compositions (at.%) of the alloys investigated, as measured by DCPMS and LEAP tomography of peak-aged alloys; measurement uncertainty is given in
parenthesis after the significant digit to which it applies.
Nominal Measured alloy composition (DCPMS) Measured alloy composition (LEAP)
Si Zr Sc Er Si2+ Zr Sc Er
Al–0.06Zr–0.06Sc–0.04Si 0.036(1) 0.062(1) 0.059(1) – 0.0211 (3) 0.0441 (5) 0.0583 (5) –
Al–0.06Zr–(0.05Sc–0.01Er)–0.04Si 0.033(1) 0.056(1) 0.046(1) 0.011(1) 0.0347 (3) 0.0412 (5) 0.0434 (5) 0.0044 (5)
 C. Booth-Morrison et al. / Acta Materialia 60 (2012) 3643–3654 3645

measured by DCPMS [33,34]. LEAP tomography mea- primary precipitates were not observed in the Er-free alloy,
sured Si concentrations in a similar study of Al–Sc–Li– indicating that the solubility limit of the Er-containing
Yb–Si alloys [34] were shown to be incorrect because of alloy was exceeded during solidification and heat-treat-
an artifact in the spatial detection of Si, which segregates ment, as observed previously in Al–0.06Zr–0.04Sc–0.02Er
at low-index crystallographic poles, owing to surface (at.%) [1]. The addition of Sc and Zr has thus significantly
migration of the Si atoms during pulsed evaporation. The decreased the 0.046 at.% solubility of Er in a binary Al–Er
Si atoms are then preferentially retained with respect to alloy [35]. The Er concentration, as measured by LEAP
Al until the atomic terraces evaporate completely. For tomography of the peak-aged Er-containing alloy, is
the LEAP tomographic operating conditions employed, 0.0044 ± 0.0005 at.%. Thus, less than half the nominal
which were the same as those used in Ref. [34], Si evapo- value of 0.01 at.% Er is available for nanoscale precipitates
rates exclusively as 28Si2+, whose peak in the mass spec- formed on aging, while the remainder is present in the
trum lies in the decay tail of the 27Al2+ peak, further coarser primary Al3Er precipitates. The Er-containing
reducing the accuracy of the concentration measurement. alloy also contains submicron intragranular Al3Er precipi-
As in previous research [34], the Si2+ concentration is tates, Fig. 1c, which is probably a result of microsegrega-
reported, and is measured to be less than both the nominal tion during solidification. The first solid to form in dilute
and DCPMS measured values (Table 1). Al–Zr–Sc–Er alloys is enriched in Zr, resulting in a micro-
structure consisting of Zr-enriched dendrites surrounded
2.3. Creep experiments by Sc and Er-enriched interdendritic regions. Previous
research on as-cast arc-melted Al–0.06Zr–0.06Sc and Al–
Constant load compressive creep experiments were per- 0.1Zr–0.1Sc (at.%) alloys revealed microsegregation of
formed at 400 ± 1 °C on cylindrical samples with diameter both Sc and Zr using linear composition profiles obtained
10 mm and height 20 mm. The samples were heated in a with quantitative electron-probe microanalysis [36,37]. In
three-zone furnace, and the temperature was verified by a summary, the presence of Al3Er primary precipitates
thermocouple placed within 1 cm of the specimen. The refines the grain size and reduces the effective Er concentra-
samples were placed between boron nitride-lubricated alu- tion available for strengthening nanoscale precipitation. In
mina platens and subjected to uniaxial compression by Ni the following, the nominal compositions are used to label
superalloy rams in a compression creep frame using dead the alloys.
loads. Sample displacement was monitored with a linear
variable displacement transducer with a resolution of 3.2. Nanostructure of peak-aged alloys
6 lm, resulting in a minimum measurable strain increment
of 3  104. When a measurable steady-state displacement The nanostructures of the Er-free and Er-containing
rate was achieved for a suitable duration, the applied load alloys, after aging isothermally for 4 h at 300 °C and 8 h
was increased. Thus, a single specimen yielded minimum at 425 °C, are compared employing LEAP tomography in
creep rates for a series of increasing stress levels, at the Fig. 2. The spheroidal precipitates in the Er-free alloy con-
end of which the strain did not exceed 11%. Strain rates sist of a Sc-enriched core surrounded by a Zr-enriched shell
at a given load were obtained by measuring the slope of (Fig. 3). The precipitates have an average radius of
the strain vs time plot in the secondary or steady-state 2.4 ± 0.5 nm, a number density of 2.5 ± 0.5  1022 m3
creep regime. and a volume fraction of 0.259 ± 0.007% (Table 2). The
spheroidal precipitates in the Er-containing alloy consist
3. Results of a core enriched in both Er and Sc, surrounded by a
Zr-enriched shell with an average radius hRi of 2.3 ± 0.5
3.1. Microstructure nm, a number density Nv of 2.0 ± 0.3  1022 m3 and a
volume fraction / of 0.280 ± 0.006%. Silicon partitions
The microstructures of the peak-aged Er-free and Er- to the precipitate phase and shows no preference for the
containing alloys are displayed in Fig. 1a and b, respec- precipitate core or shell in either alloy.
tively. The grains in both alloys are elongated radially The precipitate and matrix compositions of the two
along the cooling direction, with smaller grains at the cen- alloys (Table 3) demonstrate that all alloying additions
ter of the billet, as expected for cast alloys. The Er-contain- (Si, Zr, Sc and Er) partition to the precipitate phase. The
ing alloy has smaller grains than the Er-free alloy, with a matrix of the Er-containing alloy is more depleted of solute,
larger grain density of 2.1 ± 0.2 compared with 0.5 ± 0.1 with a composition of 107 ± 12 at.ppm Zr, 32 ± 4 at.ppm
grains mm2, as determined by counting grains in the billet Sc and 7 ± 4 at.ppm Er, than that of the Er-free alloy,
cross sections. The finer grain structure in the Er-contain- with a composition of 153 ± 28 at.ppm Zr, 89 ± 14 at.ppm
ing alloy is due to intergranular Al3Er precipitates with Sc; the negative Er concentration is a result of a background
trace amounts of Sc and Zr, with diameters of 2 lm, vis- correction and should be interpreted as 0, within measure-
ible in Fig. 1c, and with compositions verified by semi- ment accuracy. Given the difficulties in measuring accu-
quantitative EDS. These particles inhibit grain growth rately the concentration of Si with LEAP tomography,
after solidification and/or during homogenization. Such and given that Si does not affect the precipitated volume
3646 C. Booth-Morrison et al. / Acta Materialia 60 (2012) 3643–3654

Al-0.04 Si-0.06 Zr-0.06 Sc Al-0.04 Si-0.06 Zr-0.05 Sc-0.01 Er

(a) (b) (c)

5 µm

Intragranular
Al3,Er

Intergranular
Al3Er

5 mm 5 mm 100 µm

(d) (e) (f)

Intergranular
Al3 Er

5 mm 5 mm 100 µm

(g) (h)

5 mm 100 µm

Fig. 1. Optical and SEM micrographs of cross sections for (a) Al–0.06Zr–0.06Sc–0.04Si and (b) Al–0.06Zr–(0.05Sc–0.01Er)–0.04Si, following a two-stage
peak-aging treatment (4 h/300 °C and 8 h/425 °C). The Er-containing alloy has smaller grains, with a grain density of 2.1 ± 0.2 grains mm2, owing to the
presence of (c) intergranular Al3Er precipitates with trace amounts of Sc and Zr, as verified by semi-quantitative EDS. The Er-free alloy has
0.5 ± 0.1 grains mm2 and no intergranular precipitates. After creep at 400 °C, the grain size of Al–0.06Zr–0.06Sc–0.04Si is (d) 0.6 ± 0.1 grains mm2.
The grains in the Er-containing alloy after creep (e) have recrystallized, resulting in an increase in the grain density to 3.6 ± 0.2 grains mm2, and the
intergranular and intragranular Al3Er precipitates remain (f). Grain boundary void formation and coarsening of the intragranular Al3Er precipitates are
evident after creep at 400 °C for 1045 h, (g) and (h). The grains are larger compared with the Er-containing sample crept for 123 h, with a smaller grain
density of 3.1 ± 0.2 compared with 3.6 ± 0.2 grains mm2.

fraction because it substitutes on the Al sublattice site of the values may be evidence of early-stage clustering or precip-
ordered trialuminides [23,38], the analysis focuses on the itation, possibly as a result of the addition of Si, which has
precipitate-forming elements (Zr, Sc and Er), which provide been shown to accelerate precipitate nucleation in an Al–
ambient and elevated temperature precipitation 0.06Zr–0.06Sc (at.%) alloy aged at 300 °C [23]. After
strengthening. homogenization and peak aging, the microhardness values
of the present alloys increase to 627 ± 10 and 606 ± 20
3.3. Room and elevated temperature mechanical properties MPa, respectively. To determine the operating coherent
precipitate strengthening mechanism in the alloys, the
3.3.1. Peak-aged condition strength increments for order strengthening (Drord), coher-
The as-cast microhardness values of the Er-free and Er- ency and modulus strengthening (Drcoh + Drmod) or
containing alloys are 256 ± 4 and 270 ± 8 MPa, respec- strengthening by the Orowan bypass mechanism (DrOr)
tively. These microhardness values are larger than those are calculated using the equations given in the Appendix,
of previous as-cast dilute Al–Sc–X alloys, with comparable and are listed in Table 4. The predicted strengthening incre-
solute contents, of 210–240 MPa. The larger microhardness ments are compared with the measured strength increment,
 C. Booth-Morrison et al. / Acta Materialia 60 (2012) 3643–3654 3647

(a) Al-0.04 Si-0.06 Zr-0.06 Sc (b) Al-0.04 Si-0.06 Zr-0.05 Sc-0.01 Er

50 nm

Fig. 2. APT reconstructions of alloys after a two-stage peak-aging treatment (4 h/300 °C and 8 h/425 °C): (a) Al–0.06Zr–0.06Sc–0.04Si (with two different
specimens to improve precipitate counting statistics); (b) Al–0.06Zr–(0.05Sc–0.01Er)–0.04Si. Scandium atoms are displayed in red, Zr atoms in green, Er
atoms in blue, illustrating the core–shell structure. Aluminum and Si atoms are omitted for clarity. (For interpretation of the references to color in this
figure legend, the reader is referred to the web version of this article.)

estimated as DHV/3 [39], where DHV is the increase in increments was previously reported for Al–0.06 at.% Sc
microhardness from the as-quenched state to the peak-aged alloys with and without additions of Zr, Ti, Gd, Tb, Dy,
state. The strengthening increment from precipitate shear- Ho, Er, Tm, Yb, Lu and Li [18,34,40,42–46].
ing is taken as the larger of (a) the sum of modulus Fig. 4 displays the minimum compressive strain rate vs
strengthening and coherency strengthening or (b) the order uniaxial compressive stress at 400 °C for the two alloys
strengthening. This is because the mechanisms in (a) and tested in the peak-aged condition. The apparent stress
(b) occur sequentially as dislocations move towards the exponent for dislocation climb-controlled creep for the
matrix/precipitate interface and shear the precipitates. Er-free alloy (measured over the range 7–13 MPa) is
Shearing and Orowan bypass occur in parallel (i.e., disloca- 16 ± 1, which is significantly greater than that of 4.4
tions either shear or bypass the precipitates), so strengthen- expected for Al [47]. Larger than expected stress exponents
ing is dictated by the mechanism that requires the smallest were previously measured in other Al–Sc-based alloys
stress. The critical radius at which the deformation mecha- [17,40,41,45,48–51] and are indicative of a threshold stress
nism changes from precipitate shearing to an Orowan for creep, below which dislocation creep is not measureable
bypass mechanism in the Al–Sc system is 1.5–2.0 nm in in laboratory time-frames. To determine the threshold
an Al–0.18 at.% Sc alloy [40,41]. Accordingly, for the two stress for dislocation creep rdisloc
th , the dislocation creep data
peak-aged alloys studied here, with average precipitate are analyzed with a modified version of the Mukherjee–
radii of 2.3 ± 0.5 and 2.4 ± 0.5 nm, the Orowan bypass Bird–Dorn creep equation:
mechanism is predicted to be operative, as indicated by  
numerical results (Table 4). Note that good agreement disloc n Q
e_ ¼ Aðr  rth Þ exp ð1Þ
between the measured and predicted Orowan strength kBT
3648 C. Booth-Morrison et al. / Acta Materialia 60 (2012) 3643–3654

alloy and using n = 4.4 yields a threshold stress of 6.5 ±

0.5 MPa for the Er-free alloy (Table 5).
The Er-containing alloy exhibits two creep regimes
(Fig. 4), which are identified as diffusional creep at lower
stresses and dislocation climb controlled creep at higher
stresses. The grain size in the Er-containing alloy is suffi-
ciently small for diffusional creep to occur at 400 °C, which
is a homologous temperature of 0.72 Tm (where Tm is the
absolute melting point of Al, 934 K). The apparent stress
exponent for diffusional creep, as measured between 8 and
9 MPa, is 6 ± 1, indicative of a threshold stress for diffu-
sional creep rdiff
th . This is calculated from a modified version
of the Mukherjee–Bird–Dorn creep equation identical to
Eq. (1), though the values of A and Q are specific to diffu-
sional creep, and n = 1. The resulting value of rdiff th of
6.9 ± 0.8 MPa has a significant uncertainty, beyond the
error bar of 0.8 MPa, given that only three data points were
available for the best-fit procedure. At higher stresses, dislo-
cation creep becomes the dominant deformation mecha-
nism in the Er-containing alloy, with an apparent stress
exponent of 20 ± 2 (11–15 MPa), indicating a threshold
stress for dislocation creep. Fitting the dislocation creep
data to Eq. (1) yields a threshold stress of 9.3 ± 0.7 MPa.
The microstructures of the two alloys following creep
testing at 400 °C are displayed in Fig. 1d and e, respec-
Fig. 3. Average concentration profiles (proximity histograms) across the
matrix/precipitate interface for alloys after a two-stage peak-aging tively. After creep at 400 °C, the grains in the Er-free alloy
treatment (4 h/300 °C and 8 h/425 °C): (a) Al–0.06Zr–0.06Sc–0.04Si; (b) (Fig. 1d) appear unchanged with 0.6 ± 0.1 grains mm2,
Al–0.06Zr–(0.05Sc–0.01Er)–0.04Si. Precipitates exhibit a Zr-rich shell and compared with the 0.5 ± 0.1 grains mm2 before creep
a core enriched in Sc (and Er), and Si partitions to the precipitates without (Fig. 1a). The grains in the Er-containing alloy following
preference for the core or the shell. The inset images in (a) and (b) are APT
creep (Fig. 1e) have undergone recrystallization, resulting
reconstructions of a representative precipitate from each alloy with same
color scheme as in Fig. 2. (For interpretation of the references to color in in an increase in the grain density to 3.6 ± 0.2 grains mm2
this figure legend, the reader is referred to the web version of this article.) from the pre-creep value of 2.1 ± 0.2 (Fig. 1b). The inter-
granular Al3Er precipitates remain following creep
(Fig. 1f).
Table 2
Precipitate number density Nv, mean radius hRi, volume fraction / and
3.3.2. Over-aged condition
Vickers microhardness HV of the alloys studied.
To collect more data in the diffusional creep regime of
Alloy Nv hRi / HV
the Er-containing alloy, a second series of creep experi-
(1022 m3) (nm) (%) (MPa)
ments was performed at 400 °C on another peak-aged sam-
Al–0.06Zr–0.06Sc– 2.5 ± 0.5 2.4 ± 0.5 0.259 ± 0.007 627 ± 10
ple, beginning at a lower applied stress of 6 MPa.
0.04Si
Al–0.06Zr–(0.05Sc– 2.0 ± 0.3 2.3 ± 0.5 0.280 ± 0.006 606 ± 20 Compressive creep data were collected over 325 h for four
0.01Er)–0.04Si stresses ranging from 6 to 8.5 MPa, which yielded a nearly
(peak-ageda) constant strain rate of 1.2 ± 0.2  108 s1, where the error
Al–0.06Zr–(0.05Sc– 0.2 ± 0.1 5–10 – 436 ± 10 is the standard deviation of the four resulting strain rates.
0.01Er)–0.04Si
A constant strain rate for increasing applied stress is indic-
(over-agedb)
a
ative of an evolving microstructure, that is, grain growth
4 h/300 °C and 8 h/425 °C.
b during the creep test. Since the rate of diffusional creep at
Peak-aged and 1045 h/400 °C.
a given stress decreases with increasing grain size, grain
growth can account for the nearly constant strain rate mea-
where A is a constant, r is the applied stress, and n and Q sured between 6 and 8.5 MPa.
are the stress exponent and activation energy for diffusion The applied stress was then removed, and the sample
in pure aluminum. The thresholdpstress ffiffi is obtained from a was held in the creep frame for 48 h at 400 °C to allow
n
linearpleast-squares
ffiffi pffiffi regression of _
e vs. r, with a weight of for a full recovery of the dislocation microstructure. Creep
1=r2 n e_ , where r n e_ is the uncertainty in the nth root of the testing of the sample, by then at 400 °C for 373 h
strain rate. The value of rdisloc
th is calculated by dividing the (15.5 days), and labeled in the following as ‘over-aged’,
value of the y-axis intercept of the linear regression analysis was then resumed, beginning at a stress of 6 MPa and
by its slope. Applying Eq. (1) to the data for the Er-free lasting 672 h (28 days), most of it spent below 13 MPa.
 C. Booth-Morrison et al. / Acta Materialia 60 (2012) 3643–3654 3649

Table 3
Compositions of the precipitates and matrix in peak-aged alloys (4 h/300 °C and 8 h/425 °C).
Alloy Precipitate composition (at.%) Matrix composition (at.ppm)
2+a
Zr Si Sc Er Si2+a Zr Sc Er
Al–0.06Zr–0.06Sc–0.04Si 4.23 ± 0.15 0.35 ± 0.04 25.29 ± 0.31 – 163 ± 24 153 ± 28 89 ± 14 –
Al–0.06Zr–(0.05Sc–0.01Er)–0.04Si 3.31 ± 0.08 1.75 ± 0.06 21.28 ± 0.17 2.43 ± 0.07 294 ± 12 107 ± 12 32 ± 4 7 ± 4
a
The reported Si concentrations are inaccurate, owing to difficulties in quantifying Si in Al with the LEAP tomographic technique (Section 2.2).

Table 4
Experimental (DHV/3) and calculated strength increments (Eqs. (A1)–(A4)) for peak-aged alloys.
Alloy DHV/3 Drord (MPa) Drcoh + Drmod (MPa) DrOr (MPa)
(MPa) (Eq. (A1)) (Eqs. A2, A3) (Eq. (A4))
Al–0.06Zr–0.06Sc–0.04Si 124 ± 4 120 ± 1 151 ± 8 134 ± 39
Al–0.06Zr–(0.05Sc–0.01Er)–0.04Si (peak-ageda) 112 ± 7 124 ± 1 166 ± 9 143 ± 44
Al–0.06Zr–(0.05Sc–0.01Er)–0.04Si (over-agedb) 55 ± 10 124 ± 1 215–273 52–86
a
4 h/300 °C and 8 h/425 °C.
b
Peak-aged and 1045 h/400 °C.

The results of this series of tests on the over-aged sample

are shown in Fig. 5, and compared with those obtained
for the peak-aged alloy. For all measured stresses, the creep
rates of the over-aged Er-containing alloy are lower than in
the peak-aged condition, in some cases by about three
orders of magnitude. In the dislocation creep regime at
high stresses (14–18 MPa), an apparent stress exponent of
29 ± 2 is again indicative of a threshold stress, which is
determined as 13.9 ± 1.6 MPa. In the diffusional creep
regime at low stresses (6–11 MPa), the apparent stress
exponent is 2.5 ± 0.2, and the threshold stress is 4.5 ±
0.8 MPa. A transition region between diffusional and dislo-
cation creep between 11 and 13 MPa is observed, which
was not present in the peak-aged sample.
The microstructure of the over-aged alloy after a total of
1045 h (43.5 days) in the creep frame at 400 °C is displayed
in Fig. 1g. There is evidence of void formation at the grain
boundaries, and of significant coarsening of the intragran-
ular Al3Er precipitates compared with the peak-aged state
(Fig. 1b). The formation of voids may be due to tensile
stresses developing perpendicular to the applied compres-
Fig. 4. Double logarithmic plot of minimum creep rate vs. applied stress
for compressive creep experiments at 400 °C for: (a) Al–0.06Zr–0.06Sc– sive load, resulting from slight barreling of the sample dur-
0.04Si; (b) Al–0.06Zr–(0.05Sc–0.01Er)–0.04Si after a two-stage peak-aging ing compressive creep testing. It is likely that these voids
treatment (4 h/300 °C and 8 h/425 °C). Best fit lines to Eq. (1) are plotted, formed after considerable strain had accumulated in the
and resulting threshold stresses for diffusional rdiff
th and dislocation rdisloc
th sample, and they may thus affect the last few creep data
creep are shown, where applicable.
points measured at the highest stresses, resulting in higher

Table 5
Comparison of the experimental normalized threshold stresses for dislocation creep rdisloc
th /DrOr measured on peak-aged alloys with those predicted by a
threshold stress model for Al–Sc-based alloys [22], which accounts for average precipitate radius hRi and unconstrained lattice parameter misfit d; all
values are given for a temperature of 400 °C.
Alloy Creep time at hRi (nm) d (%) rdisloc
th Experimental DrOr Eq. (A1) rdisloc
th /DrOr rdisloc
th /DrOr
400 °C (h) (MPa) (MPa) Experimental (%) Modeled [22] (%)
Al–0.06Zr–0.06Sc–0.04Si 130 2.4 ± 0.5 0.55 6.5 ± 0.5 111 ± 34 5.9 ± 1.8 17
Al–0.06Zr–(0.05Sc–0.01Er)–0.04Si 123 2.3 ± 0.5 0.82 9.3 ± 0.7 119 ± 35 7.8 ± 2.4 29
3650 C. Booth-Morrison et al. / Acta Materialia 60 (2012) 3643–3654

(Fig. 3). A previous study on an Al–0.06Zr–0.04Sc–0.02Er

(at.%) alloy yielded precipitates with a structure consisting
of an Er-enriched core surrounded by a Sc-enriched inner
shell and a Zr-enriched outer shell (core/double-shell struc-
ture) after either isothermal aging at 400 °C or isochronal
aging to 450 °C in stages of 25 °C h1 [1]. The core/dou-
ble-shell structure was shown by first-principles calcula-
tions to be the most energetically favorable precipitate
structure for dilute Al–Zr–Sc–Er alloys [52], and to form
by sequential precipitation of solute elements according
to their intrinsic diffusivities D, where DEr > DSc > DZr at
400 °C [1]. No diffusion data are available for Er in Al at
400 °C, although the evidence of accelerated precipitation
kinetics in the Al–Zr–Sc system due to addition of Er sug-
gests that the diffusivity of Er in Al is larger than that of Sc
at 400 °C [1]. The lack of a core/double-shell structure in
the present Er-containing alloy must therefore be a result
of either the different aging treatment used or the addition
of Si. The first-stage aging treatment of 4 h at 300 °C is
Fig. 5. Double logarithmic plot of minimum creep rate vs. applied stress designed to precipitate Er and Sc out of solution as rapidly
for compressive creep experiments at 400 °C for Al–0.06Zr–(0.05Sc– as possible to minimize both processing time and precipi-
0.01Er)–0.04Si (a) after a two-stage peak-aging treatment (4 h/300 °C and tate coarsening. The diffusivity of Er at 300 °C, DEr =
8 h/425 °C) and (b) after subsequent exposure at 400 °C for 325 h at 4 ± 2  1019 m2 s1, as determined from Al3Er (L12)
applied stresses ranging from 6 to 8.5 MPa. Best fit lines to Eq. (1) are
plotted, and resulting threshold stresses for diffusional rdiff and disloca-
coarsening experiments in a binary Al–0.045 at.% Er alloy
th
tion, rdisloc
th creep are shown, where applicable. [35], is 4 ± 2 times greater than that of Sc in binary Al–
Sc, DSc = 9.1  1020 m2 s1 at 300 °C [53]. The addition
of Si to the base alloy studied here may have accelerated
than expected strain rates. The over-aged sample exhibits a the precipitation kinetics of Sc, as it has been shown to
microhardness of 436 ± 10 MPa, following 1075 h of creep do in an Al–0.06Si–0.06Zr–0.06Sc alloy at 300 °C [23],
at 400 °C, which is, as anticipated, below the peak-aged thereby decreasing the difference between DEr and DSc,
value of 606 ± 20 MPa. and resulting in co-precipitation of Er and Sc into Al3(S-
The grains are slightly larger in the Er-containing alloy c,Er) precipitates of uniform composition. The subsequent
that was exposed for 1045 h at 400 °C, with a larger grain aging at 425 °C then creates a Zr-rich Al3(Sc,Zr) shell. A
density of 3.1 ± 0.2 grains mm2, compared with the 3.6 ± precipitate structure consisting of a co-precipitated Yb-
0.2 grains mm2 from the Er-containing sample exposed and Sc-enriched core, with evidence at the matrix/precipi-
for 123 h. LEAP tomographic analysis of the crept material tate interface of Zr-enrichment (which was deemed to not
(reconstructions not shown) revealed a number density of constitute a shell), was previously found for an Al–
precipitates of 2 ± 1  1021 m3, where the high degree of 0.02Zr–0.06Sc–0.02Yb (at.%) alloy aged for 1536 h at
error is because only five precipitates were detected in a 300 °C [54]. This is in contrast to the Yb-enriched core/
50 million atom data set, all of which were only partially Sc-enriched shell structure observed in an Al–0.06Sc–
bound by the tip volume. Given the poor precipitate statis- 0.005 Yb (at.%) alloy aged at 300 °C [35,44]. The precipi-
tics, detailed compositional and structural analyses are not tate structure observed in Al–Zr–Sc–RE alloys, be it
possible, though the precipitate radius was estimated from core/shell or core/double-shell, is clearly sensitive to the
the LEAP tomographic reconstruction to be 5–10 nm. aging treatments and alloying additions employed, a sub-
Assuming that the volume fraction of precipitates is con- ject that warrants serious future investigations.
stant for the peak-aged and over-aged sample, and using After aging, the volume fraction of precipitates is larger
the measured number density of 2 ± 1  1021 m3, a radius and the total matrix solute concentration smaller in the Er-
of 6–9 nm is calculated for the spheroidal precipitates, in containing alloy compared with the Er-free alloy (Tables 2
good agreement with the above estimate. and 3). This is evidence for increased precipitation kinetics
due to the addition of Er and was observed previously in
4. Discussion dilute Al–Zr–Sc alloys and attributed to differences in the
intrinsic diffusion coefficients (DEr > DSc > DZr) [1]. Similar
4.1. Precipitate structure effects have been observed in previous studies of Al–0.06Sc
at.% alloys with 0–0.02 at.% RE element additions of Er,
Precipitates with a structure consisting of an Er- and Sc- Yb, Gd, Ho, Lu, Tm or Tb [18,35,42,44,46].
enriched core surrounded by a Zr-enriched shell (core/shell Finally, it is emphasized that the degree of coarsening
structure) are formed in the peak-aged Er-containing alloy observed in the crept Er-containing alloy after 44 days at
 C. Booth-Morrison et al. / Acta Materialia 60 (2012) 3643–3654 3651

400 °C was not observed in a previous alloy with a similar creep in dilute Al–Sc-based alloys increases with increasing
composition (Al–0.06Zr–0.04Sc–0.02Er (at.%)) aged iso- lattice parameter mismatch between the matrix and precip-
thermally for a longer time, 64 days, at 400 °C. In the pre- itates [51]. These findings agree with the predictions of
vious study, the precipitate radius of 3.7 ± 0.3 nm after models that attribute the threshold stress to elastic interac-
0.5 h at 400 °C remained unchanged within uncertainly tions between dislocations and the coherent precipitates
after 64 days, with a value of 3.8 ± 0.4 nm. The increased with lattice parameter- and shear modulus-mismatches
coarsening behavior observed here is probably the result with the matrix that they bypass by dislocation climb
of either the applied stress during creep, as observed previ- [21,22]. The constrained lattice parameter mismatches at
ously in Al alloys [55–57], and/or the addition of Si, which 400 °C are calculated to be 0.55% and 0.82% for the Er-free
has been demonstrated to accelerate precipitation kinetics and Er-containing alloys, respectively, by linear interpola-
[23,33,38]. tion of lattice parameter data for L12-ordered Al3(Sc, Er)
and Al3(Sc, Zr), [62,63], using the experimentally measured
4.2. Creep behavior precipitate compositions, and adjusted for thermal expan-
sion of Al3(Sc, Er) and Al3(Sc, Zr) [64] and Al [65]. The
4.2.1. Diffusional creep impact of Si on the lattice parameter is neglected, owing
Diffusional creep is observed in the peak-aged Er-con- to the difficulties in measuring accurately the Si concentra-
taining alloy, where primary Al3Er precipitates result in a tion from LEAP tomography (Section 2.2), and the lack of
fine-grained microstructure after casting and aging experimental data on the lattice parameter of (Al,Si)3Sc.
(Fig. 1b and c). Diffusional creep was not reported in pre- The threshold stresses expected for the alloys studied
vious creep studies of cast dilute Al–Sc-based alloys, where here are calculated employing a recent model by Krug
the grain sizes were on the order of 1–2 mm [17,40,41, and Dunand [22], which attributes the existence of a thresh-
45,48–51], such as those in the Er-free alloy studied here. old stress to an attractive interaction force between precip-
A threshold stress for diffusional creep of 6.9 ± 0.8 MPa itates and the dislocations that climb over them. This
is measured in the peak-aged Er-containing alloy, and is model builds on the general climb model developed by
most likely due to the intergranular primary Al3Er precip- Rösler and Arzt [66] for creep in alloys strengthened
itates (Fig. 1c). Threshold stresses for diffusional creep in by coherent, unshearable precipitates, and its extension by
alloys have been attributed to limitations in the ability of Marquis and Dunand [21] to include elastic interactions
grain boundaries to act as perfect sources and sinks for between dislocations and precipitates due to lattice param-
vacancies, owing to the presence of intergranular second- eter and shear modulus mismatches between the matrix
phase particles [58–60]. These particles were modeled to and precipitates. In the model developed by Krug and
impede the movement of grain boundary dislocations, lim- Dunand [22], as dislocations climb over a precipitate by
iting vacancy condensation on the boundary and control- vacancy-mediated diffusion, they become trapped close to
ling the rate of diffusional creep. It is emphasized that the apex of a precipitate by elastic stresses induced in the
diffusional creep can be eliminated in the Er-containing matrix due to a lattice parameter mismatch with it. The
alloy by reducing the Er concentration to 0.005 at.%, the application of an externally applied stress, the threshold
effective alloy composition measured by LEAP tomogra- stress, produces a shear force that allows the dislocation
phy (Fig. 1), eliminating intergranular Al3Er precipitates to escape and dislocation creep to continue. This model
and increasing the alloy grain size. Recrystallization occurs predicts that the threshold stress, when normalized to the
only in the crept Er-containing alloy, which has signifi- Orowan stress (the stress at which dislocations bypass the
cantly finer grains than those of the Er-free alloy, and thus precipitates by looping), increases with increasing precipi-
more energy stored in the metal, and a larger driving force tate radius and/or increasing lattice parameter misfit. These
for recrystallization [61]. predictions are validated qualitatively by experimental
The decrease in the rates of diffusional creep in the over- creep results in dilute Al–Sc-based alloys at 300 °C
aged sample (Fig. 5) is expected due to grain growth that [17,40,41,45,48–51], as summarized in Ref. [51]. Using the
occurred during over-aging, which lasted 325 h under above model [22], normalized threshold stresses of 0.17
applied stresses ranging from 6 to 8.5 MPa. The decrease and 0.29 (that is, threshold stresses of 17% and 29% of
in the threshold stress for diffusional creep from 6.9 ± 0.8 the Orowan stress) are predicted for the peak-aged Er-free
to 4.5 ± 0.8 after over-aging is anticipated from the coars- and Er-containing alloys. The calculated values are larger
ening of the intergranular Al3Er precipitates, decreasing than those measured experimentally of 0.059 ± 0.018 and
the number of obstacles to grain boundary sliding. 0.078 ± 0.024, respectively (Table 5). An overestimate of
the threshold stresses by a similar factor of 2–3 was also
4.2.2. Dislocation creep observed earlier [22] for threshold stresses in Al–Sc, Al–
Fig. 4 shows that the resistance to dislocation creep of Sc–Li and Al–Sc–Yb alloys at 300 °C, and was argued to
the alloy containing 0.01 at.% Er is significantly greater be because of the geometry of individual precipitates, the
than that of the Er-free alloy, with the same average precip- overall precipitate arrangement, and the fact that the
itate radius, within experimental error (Table 2). The shapes of the climbing dislocation are more complex than
experimentally measured threshold stress for dislocation accounted for by the simple analytical model. Note that
3652 C. Booth-Morrison et al. / Acta Materialia 60 (2012) 3643–3654

the predicted Orowan stress at 400 °C (Table 5) is smaller that the precipitate radius and composition in the previous
than the value at room temperature (Table 4), owing to study were comparable with those measured here. A thresh-
the temperature dependence of the shear modulus of Al. old stress of 10 ± 3 MPa was measured at 400 °C for an Al–
The increase in the threshold stress for dislocation creep 0.06Sc–0.06 Ti (at.%) alloy with an average precipitate
in the Er-containing alloy from the peak- to over-aged state radius of 8.3 ± 2.9 nm and a lattice parameter mismatch
can be explained by the observed increase in the precipitate of 0.72%. The threshold stress in this Al–Sc–Ti alloy is larger
radius from 2.3 ± 0.5 nm for the peak-aged state to 5– than that measured here for the Er-containing alloy (Fig. 6),
10 nm for the over-aged alloy. The normalized threshold probably because of the larger precipitates, and is also larger
stress for the Er-containing alloy increases by a factor of than that of the Er-free alloy, because both the precipitate
2–3 from 0.08 ± 0.02 to 0.16–0.27, and the radius increases radius and lattice parameter misfit are larger (Table 5).
by a factor of 2–4 from 2.3 ± 0.5 to 5–10 nm (Fig. 6). The Finally, threshold stresses of 6–7 MPa were measured at
same nearly linear increase in the measured threshold stress 400 °C for Al–0.1 Zr and Al–0.1 Zr–0.1 Ti (at.%) alloys iso-
for dislocation creep with increasing radius was found in thermally aged for 100 h at 400 °C. The precipitate volume
many other dilute Al–Sc-based alloys at 300 °C fraction and average radius in these alloys are not well
[17,40,41,45,48–51], as summarized in Ref. [51], in qualita- defined, owing to solute segregation to the dendrites, and
tive agreement with the predictions of the model by Krug significant intergranular precipitation [50], preventing fur-
and Dunand [22]. ther comparison.

4.2.3. Comparison with other castable, creep-resistant Al 5. Conclusions

alloys
Dilute Al–Sc, Al–Sc–RE and Al–Sc–Li alloys are not suf- The effect of substituting 0.01 at.% Er for Sc in an Al–
ficiently coarsening resistant to be considered for creep 0.06Zr–0.06Sc–0.04Si (at.%) alloy subjected to a two-stage
applications at 400 °C, but they have excellent creep resis- peak-aging treatment of 4 h at 300 °C and 8 h at 425 °C
tance between 300 and 350 °C [17,34,40,41,45,48–51]. Com- was assessed to determine the viability of dilute Al–Zr(–
parisons of the creep behavior at 400 °C between the present Sc/Er)–Si alloys for high-temperature, low-stress, applica-
alloys and several highly coarsening-resistant Al–Sc–Zr, Al– tions. The following conclusions are reached.
Sc–Ti, Al–Zr and Al–Zr–Ti alloys, however, can be made. A
7 MPa threshold stress for dislocation creep was determined 1. The effective Er concentration of the Al–0.06Zr–
at 400 °C for an Al–0.06Zr–0.06Sc (at.%) alloy aged isochro- (0.05Sc–0.01Er)–0.04Si (at.%) alloys was diminished by
nally to 400 °C, for which neither precipitate radius nor lat- a factor of 2, owing to the formation of inter- and
tice parameter data are available. This threshold stress is in intragranular primary Al3Er (L12) precipitates. The
excellent agreement with the value of 6.5 ± 0.5 MPa mea- Er-containing alloy has finer grains, with a larger grain
sured here for the peak-aged Er-free alloy with a similar density of 2.1 ± 0.2 grains mm2 compared with
composition (Al–0.06Zr–0.06Sc–0.04Si (at.%)), indicating 0.5 ± 0.1 grains mm2 for the Er-free alloy, owing to
the intergranular Al3Er precipitates that inhibit grain
growth.
2. Coherent, spheroidal, L12-ordered, nanometric, trialu-
minide precipitates consisting of an Er- and Sc-enriched
core and a Zr-enriched outer shell were formed in the
Er-containing alloy, providing excellent creep and coars-
ening resistance.
3. At ambient temperature, the Orowan bypass mechanism
is predicted to be operative for both Er-free and Er-con-
taining alloys, which have average precipitate radii of
2.4 ± 0.5 and 2.3 ± 0.5 nm, respectively, after peak
aging.
4. The addition of 0.01 at.% Er significantly improves the
resistance of the alloys to dislocation creep at 400 °C,
increasing the threshold stress rdisloc
th from 7 to 9 MPa.
Increasing the precipitate radius from 2 to 5–10 nm fur-
ther increases the threshold stress of the Er-containing
alloy to 14 MPa. Both results agree with experimental
observations of increasing values of rdisloc
th with increas-
ing lattice parameter misfit and increasing precipitate
Fig. 6. Plot of threshold stresses for dislocation creep at 400 °C
normalized to the Orowan stress vs average precipitate radius hRi for radius in dilute Al–Sc–RE-based alloys. These findings
the alloys studied here. The result for an Al–0.06Sc–0.06Ti (at.%) alloy is also agree with the qualitative predictions of a model
also shown [45]. that attributes the existence of a threshold stress in
 C. Booth-Morrison et al. / Acta Materialia 60 (2012) 3643–3654 3653

 1=2
Al–Sc-based alloys to an attractive interaction force 3=2 hRi/
between precipitates and the dislocations that climb over Drcoh ¼ Mae ðGeÞ ðA2Þ
0:5Gb
them [22].
where ae = 2.6 is a constant [67], hRi is the mean precipitate
5. Diffusional creep is observed in the Er-containing alloy
radius, G = 25.4 GPa [47] is the shear modulus of Al, and e
at 400 °C due to the finer-grained microstructure that
is the constrained lattice parameter mismatch obtained
results from intergranular Al3Er (L12) precipitates.
using Vegard’s law and the precipitate compositions mea-
Decreasing the alloy concentration of Er to 0.005
sured here.
at.% will eliminate these intergranular precipitates,
Strengthening by modulus mismatch is given by
resulting in grain sizes on the order of 1–2 mm and dis-
 1=2  ð3m2 1Þ
appearance of diffusional creep at laboratory time 3=2 2/ hRi
scales, while maintaining the improved resistance to dis- Drmod ¼ 0:0055MðDGÞ b ðA3Þ
Gb2 b
location creep imparted by the nanoscale, coherent,
where DG = 42.5 GPa is the shear modulus mismatch be-
spheroidal, L12-ordered precipitates.
tween the matrix and the precipitates [73], and m is a con-
stant, taken to be 0.85 [67].
Acknowledgements Finally, the increase in yield strength for the Orowan
bypass mechanism is given by
This research was sponsored by the Ford–Boeing–
0:4 Gb lnð2rbÞ
Northwestern University Alliance (81132882). APT was DrOr ¼ M pffiffiffiffiffiffiffiffiffiffiffi ðA4Þ
performed at the Northwestern University Center for p 1v k
Atom-probe Tomography (NUCAPT). The LEAP tomog- where m = 0.34 is the Poisson’s ratio of the matrix [69], and
raphy system was purchased and upgraded with funding k is the edge-to-edge inter-precipitate distance, which is ta-
from NSF-MRI (DMR-0420532) and ONR-DURIP ken to be the square lattice spacing in parallel planes [74]:
(N00014-0400798, N00014-0610539 and N00014-0910781) "  #
1=2
3p
grants. The authors also gratefully acknowledge the Initia- k¼  1:64 hRi ðA5Þ
tive for Sustainability and Energy at Northwestern (ISEN) 4/
for grants to upgrade the capabilities of NUCAPT. They
thank Prof. P. Sanders, Mr N. Johnson and Mr P. Quimby References
(Michigan Technological University) for casting the alloys,
Dr M. Krug (NU) for many useful discussions and for [1] Booth-Morrison C, Dunand DC, Seidman DN. Acta Mater
modeling the normalized threshold stress, Mr M. Diaz 2011;59:7029.
(NU) for his help with SEM and Dr D. Isheim for his assis- [2] Elagin VI, Zakharov VV, Pavlenko SG, Rostova TD. Phys Met Phys
Metall 1985;60:97.
tance with LEAP tomography. Dr J. Boileau (Ford), Dr B.
[3] Tolley A, Radmilovic V, Dahmen U. Scr Mater 2005;52:621.
Ghaffari (Ford), Mr C. Huskamp (Boeing) and Dr K.K. [4] Forbord B, Lefebvre W, Danoix F, Hallem H, Marthinsen K. Scr
Sankaran (Boeing) are thanked for many useful Mater 2004;51:333.
discussions. [5] Clouet E, Lae L, Epicier T, Lefebvre W, Nastar M, Deschamps A.
Nat Mater 2006;5:482.
[6] Clouet E, Nastar M, Sigli C. Phys Rev B 2004;69:064109.
Appendix A [7] Fuller CB, Murray JL, Seidman DN. Acta Mater 2005;53:5401.
[8] Fuller CB, Seidman DN. Acta Mater 2005;53:5415.
The strength increments for order strengthening (Drord), [9] Toropova LS, Eskin D, Kharakterova M, Dobatkina T. Advanced
aluminum alloys containing scandium: structure and properties.
coherency and modulus strengthening (Drcoh + Drmod),
Amsterdam: Gordon & Breach Science; 1998.
and strengthening by the Orowan bypass mechanism [10] Clouet E, Nastar M, Barbu A, Sigli C, Martin G. Adv Eng Mater
(DrOr), are calculated employing the methodology 2006;8:1228.
described [67,68], as performed previously [37,40,42,43]. [11] Clouet E, Nastar M, Barbu A, Sigli C, Martin G. In: Howe JM,
First, the contribution to the yield strength from order Laughlin DE, Lee JK, Dahmen U, Soffa WA, editors. Solid-solid
phase transformations in inorganic materials. Warrendale, PA: TMS;
strengthening is given by
2005. p. 1.
 1=2 [12] Lefebvre W, Danoix F, Hallem H, Forbord B, Bostel A, Marthinsen
cAPB 3p/
Drord ¼ 0:81M ðA1Þ K. J Alloys Compd 2009;470:107.
2b 8 [13] Clouet E, Barbu A, Lae L, Martin G. Acta Mater 2005;53:2313.
[14] Hallem H, Lefebvre W, Forbord B, Danoix F, Marthinsen K. Mater
where M = 3.06 is the mean matrix orientation factor for
Sci Eng A 2006;A421:154.
aluminum [69], b = 0.286 nm is the magnitude of the ma- [15] Royset J, Ryum N. Int Mater Rev 2005;50:19.
trix’s Burgers vector [47], / is the volume fraction of pre- [16] Karnesky RA, Dunand DC, Seidman DN. Acta Mater 2009;57:4022.
cipitates, and cABP = 0.5 J m2 is an average value of the [17] Karnesky RA, Seidman DN, Dunand DC. Mater Sci Forum
Al3Sc anti-phase boundary (APB) energy for the (1 1 1) 2006;519–521:1035.
[18] Karnesky RA, van Dalen ME, Dunand DC, Seidman DN. Scr Mater
plane [70–72].
2006;55:437.
The increase is yield strength due to coherency strength- [19] Hedrick JB. US Geological Survey 2007 minerals yearbook: rare
ening is given by earths. Reston, VA: USGS; 2007.
3654 C. Booth-Morrison et al. / Acta Materialia 60 (2012) 3643–3654

[20] Zalutskaya OI, Ryabov VR, Zalutskii II. Dopov Akad Nauk Ukr [47] Frost H, Ashby M. Deformation mechanism maps. Oxford: Pergamon
RSR, Ser A 1969;31:255. Press; 1982.
[21] Marquis EA, Dunand DC. Scr Mater 2002;47:503. [48] Fuller CB, Seidman DN, Dunand DC. Scr Mater 1999;40:691.
[22] Krug ME, Dunand DC. Acta Mater 2011;59:5125. [49] Marquis EA, Seidman DN, Dunand DC. Acta Mater 2003;51:4751.
[23] Booth-Morrison C, Mao Z, Wolverton C, Dunand DC, Seidman DN. [50] Knipling KE, Dunand DC. Scr Mater 2008;59:387.
Acta Mater, in preparation. [51] Krug ME, Seidman D. N., Dunand DC. Acta Mater, submitted for
[24] Bajikar SS, Larson DJ, Kelly TF, Camus PP. Ultramicroscopy publication.
1996;65:119. [52] Mao Z, Booth-Morrison C, Seidman DN, Wolverton C. Acta Mater,
[25] Kelly TF, Camus PP, Larson DJ, Holzman LM, Bajikar SS. in preparation.
Ultramicroscopy 1996;62:29. [53] Fujikawa S. Diffus Defect Data, Pt A 1997;143–147:115.
[26] Kelly TF, Larson DJ. Mater Charact 2000;44:59. [54] van Dalen ME, Gyger T, Dunand DC, Seidman DN. Acta Mater
[27] Seidman DN. Rev Sci Instrum 2007;78:030901/1. 2011;59:7615.
[28] Seidman DN. Annu Rev Mater Res 2007;37:127. [55] Murken J, Girard O, Hohner R, Skrotzki B, Eggeler G. Mater Sci
[29] Kelly TF, Miller MK. Rev Sci Instrum 2007;78:031101/1. Forum 2000;331–337:1507.
[30] Hellman OC, Vandenbroucke JA, Rusing J, Isheim D, Seidman DN. [56] Nakajima T, Takeda M, Endo T. Mater Sci Eng A 2004;A387–
Microsc Microanal 2000;6:437. A389:670.
[31] Hellman OC, Blatz du Rivage J, Seidman DN. Ultramicroscopy [57] Skrotzki B, Murken J. Lightweight alloys for aerospace application,
2003;95:199. proceedings of [a] symposium held at the TMS annual meeting, New
[32] Parratt LG. Probability and experimental errors in science. New Orleans, LA, United States; February 12–14, 2001, 2001:52.
York: John Wiley; 1966. [58] Harris JE. Metal Sci J 1973;7:1.
[33] Beeri O, Dunand DC, Seidman DN. Mater Sci Eng A [59] Ashby MF. Scr. Met. 1969;3:837.
2010;A527:3501. [60] Burton B. Mater Sci Eng 1972;10:9.
[34] Krug ME, Dunand DC, Seidman DN. Acta Mater 2011;59:1700. [61] Shewmon PG. Transformation in metals. New York: McGraw-Hill;
[35] van Dalen ME, Karnesky RA, Cabotaje JR, Dunand DC, Seidman 1969.
DN. Acta Mater 2009;57:4081. [62] Harada Y, Dunand DC. Intermetallics 2008;17:17.
[36] Knipling KE, Dunand DC, Seidman DN. Acta Mater 2011;59:943. [63] Harada Y, Dunand DC. Mater Sci Eng A 2002;A329–A331:686.
[37] Knipling KE, Karnesky RA, Lee CP, Dunand DC, Seidman DN. [64] Harada Y, Dunand DC. Scr Mater 2003;48:219.
Acta Mater 2010;58:5184. [65] Royset J, Ryum N. Scr Mater 2005;52:1275.
[38] Du G, Deng J, Wang Y, Yan D, Rong L. Scr Mater 2009;61:532. [66] Rösler J, Arzt E. Acta Metall 1988;36:1043.
[39] Tabor D. Brit J Appl Phys 1956;7:159. [67] Ardell AJ. Metall Trans A 1985;16A:2131.
[40] Seidman DN, Marquis EA, Dunand DC. Acta Mater 2002;50:4021. [68] Argon A. Strengthening mechanisms in crystal plastic-
[41] Marquis EA, Seidman DN, Dunand DC. Acta Mater 2003;51:285. ity. Oxford: Oxford University Press; 2007.
[42] Krug ME, Werber A, Dunand DC, Seidman DN. Acta Mater [69] Meyers M, Chawla K. Mechanical metallurgy: principles and
2009;58:134. applications. Paramus, NJ: Prentice-Hall; 1984.
[43] Fuller CB, Seidman DN, Dunand DC. Acta Mater 2003;51:4803. [70] Fu CJ. Mater Res 1990;5:971.
[44] van Dalen ME, Dunand David C, Seidman David N. J Mater Sci [71] George E, Pope D, Fu C, Schneibel J. ISIJ Int 1991;31:1063.
2006;41:78814. [72] Fukunaga K, Shouji T, Miura Y. Mat Sci Eng 1997;A239:202.
[45] van Dalen ME, Seidman DN, Dunand DC. Acta Mater 2008;56:4369. [73] Hyland R, Stiffler R. Scr Metall Mater 1991;25:473.
[46] Karnesky RA. Ph.D. thesis, Northwestern University, Evanston, IL; [74] Nembach E. Particle strengthening of metals and alloys. New York,
2007. NY: John Wiley; 1997.