Solute-induced strengthening during creep of an aged-hardened Al-Mn-Zr alloy

Abstract

We examine the precipitation and creep behavior of Al-0.5Mn-0.02Si (at.%) alloys, with and without the L1₂-forming elements Zr and Er (0.09 and 0.05 at.%, respectively), utilizing isochronal aging experiments as well as compressive and tensile creep tests performed between 275 and 400 °C. The Al-0.5Mn-0.09Zr-0.05Er-0.05Si alloy exhibits an unusually high creep resistance in the peak-aged state, which is significantly better than that observed generally in its Mn-free L1₂-strengthened counterparts; for example, the creep threshold stresses at 300 °C are 34–37 MPa, about three times higher than those in a Mn-free Al-0.11Zr-0.005Er-0.02Si alloy. Scanning transmission electron microscopy illustrates that nanoscale Al₃(Zr,Er) L1₂-precipitates are formed in the dendritic cores and micron-sized Al(Mn,Fe)Si α-precipitates in the inter-dendritic channels. Moreover, the Al(f.c.c.)-matrix remains supersaturated with randomly distributed Mn solute atoms, as determined by atom-probe tomography and electrical conductivity measurements, for months at creep temperatures. Creep experiments on the Zr- and Er-free Al-0.5Mn-0.02Si solid-solution alloy reveal a small primary creep strain, a high apparent stress exponent, n_a ~9–7, and a threshold-stress-type behavior. After ruling out other possible mechanisms, we provide evidence that the threshold stress in this precipitate-free alloy originates from dislocation/solute elastic interactions leading to a strong drag force exerted on edge dislocations, hindering their ability to climb. The relatively high creep resistance of Al-0.5Mn-0.09Zr-0.05Er-0.05Si is interpreted in terms of the synergy between this solute-induced threshold stress (SITS, from Mn in solid-solution) and the known precipitate-bypass threshold stress (from the L1₂-nanoprecipitates).

1. Introduction

Designing high-temperature aluminum alloys is made difficult by the low solubility of slow-diffusing elements in aluminum, which translates into small volume fractions of thermally-stable nanoprecipitates (0.2–0.5 % as opposed to, e.g., 30–70 % in Ni-based superalloys) [1] and thereby a low creep resistance above ~0.6 T_m (T_m is the solidus temperature of the alloy). Two of the most thermally-stable precipitates utilized in designing creep-resistant aluminum alloys are the L1₂-ordered nanoprecipitates [2–13] and cubic α-precipitates, an approximant phase [14–19]. Recent research has focused on fine-tuning both composition and precipitation kinetics in these castable aluminum alloys to achieve the highest creep and coarsening resistance possible [20,21]. Additional strengthening can be achieved through solid-solution hardening using elements with high solubilities (e.g., Mg, Zn, Li and Cu): so far, however, only modest improvements in creep resistance have been reported, which are often accompanied by an undesirable decrease in the solidus temperature. For example, a 2.2 at.% Mg addition to an Al-0.12Sc (at.%) alloy with L1₂-nanoprecipitates had no measurable effects on the creep threshold stress at 300 °C, despite a significant increase of ~ 200 MPa (33%) in the hardness at ambient temperature [22]. There is, nevertheless, some evidence indicating that certain slow-diffusing solute elements such as Fe [23], even at concentrations well below 0.1 at.% in the solid-solution, can significantly slow dislocation migration during creep. Sherby et al. have demonstrated that in a dilute binary Al–0.02Fe (at.%) alloy - homogenized near the eutectic temperature (640 °C for 40 h) and water quenched - the dislocation creep deformation is controlled by the diffusivity of Fe in aluminum: a creep activation energy, Q_creep = Q_Fe ~220 kJ.mol⁻¹ (2.28 eV) was deduced, in the temperature range 200–500 °C [23], which is much greater than the activation energy for creep of pure aluminum, Q_creep = 143–151 kJ.mol⁻¹ (1.48–1.56 eV) [24–27]. They proposed that diffusion of Fe solute atoms relaxes the opposing stress fields from subgrain boundaries and from dislocation pileups during creep, thus controlling the dislocation climb processes. Later, Sherby and Ruano [28] predicted that aluminum alloys containing slow-diffusing solute atoms, such as V, Ti, Cr and Mn have slower creep rates and higher activation energies for creep than pure aluminum, generalizing their prior experimental results on Al-Fe alloys. The slow-diffusing elements, V [29], W [21, 30], Cr [31] and Mo [32] in solid-solution within the Al(f.c.c.) matrix do not, however, significantly improve the creep rates of the L1₂-strengthened alloys. This leads to our hypothesis that only selected slow-diffusing elements, such as Fe, at certain concentrations in the Al(f.c.c.) matrix can slow the creep deformation rate in solid-solution alloys. It appears that diffusing slower than Al is not, by itself, a sufficient criterion for the solute elements to hinder effectively dislocation creep, especially at small solute concentrations.

Recently, an L1₂-strengthened Al-0.1Si-0.08Zr-0.02Sc-0.01Er alloy has been demonstrated to display a much improved creep resistance through small additions of Mn (0.25–0.5 at.%) and/or Mo (0.1 at.%). Precipitation strengthening from a second population of thermally-stable α-Al(Mn,Mo,Fe)Si precipitates, as well as solid-solution hardening by Mn and/or Mo, were proposed as the causes of the enhanced creep resistance in these alloys [33–35]. The exact underlying micro-/nano-mechanisms are, however, unknown. It is the purpose of the present study to examine the creep behavior of a related but simpler Mn-modified Al-Zr-Er alloy, without Mo and Sc additions. The Mn concentration in our alloy is ~0.5 at.%, which is slightly below the maximum solid- solubility in Al (~0.7 at.% at the eutectic temperature, ~657 °C) [36], to take into account the presence of the other alloying elements. This study centers on the role of Mn solute atoms in hindering dislocation creep processes in this high-temperature Al alloy containing L1₂ nanoprecipitates and α-Al-(Mn,Fe)Si-precipitates. The possible interactions between Mn and the L1₂-nanoprecipitates, and their influence on the mechanical properties are also examined.

2. Experimental procedures

Two alloys were studied: Al-0.5Mn-0.09Zr-0.05Er-0.05Si and its Zr-Er-free counterpart, Al-0.5Mn-0.02Si at.%. The former alloy was prepared utilizing commercially-pure (99.9%) Al, which contains impurity levels of Fe (0.029 at.%) and Si (0.053 at.%). For the latter alloy, high-purity Al (99.99%) with much smaller concentrations of Fe (0.004 at.%) and Si (0.006 at.%) was utilized. After heating the pure Al to 900 °C in a graphite crucible in an electric resistance furnace, a series of master alloys (Al-10Mn, Al-0.6Zr, Al-1Er, at.%), were added sequentially to the melt, with a holding time of 15 min between each addition. All master alloys were preheated at 600 °C, except for the low-melting Al-12.2Si, which was added without preheating. Once fully alloyed, the melt was maintained at 900°C for 2 h with periodic stirring and then cast into a graphite mold preheated at 200 °C: the mold was placed on an ice-cooled copper platen, immediately prior to casting, to enhance directional solidification, which leads to reduced shrinkage porosity and a coarse grain microstructure. The chemical compositions of the cast alloys were determined by inductively-coupled plasma optical-emission spectroscopy (ICP-OES) at Genitest (Montreal, Canada) and are listed in Table 1. Also included in Table 1, for comparison, is the composition of a Mn-free counterpart, Al-0.09Zr-0.01Er-0.02Si alloy [32]. For the aging experiments, the as-cast 11 mm diam. ingots were cut into smaller samples, which were aged isochronally (3 h time steps of 25 °C) from 200 to 600 °C in air, terminated by water quenching. Vickers microhardness (HV) measurements were performed at ambient temperature on specimens (one per temperature) polished to a 1 μm surface finish, using a Duramin-5 microhardness tester (Struers), with a load of 200 g and a dwell time of 5 s. For each specimen, ten indentations were made over several grains, and the average values are reported with one standard deviation. Electrical conductivity measurements were performed at ambient temperature utilizing a Sigmatest 2.069 eddy current instrument (Foerster Instruments) on samples 11 mm in diam. and 2 mm thick, one per alloy. For each specimen, five measurements were performed at frequencies of 120, 240, 480, and 960 kHz and an average value with one standard deviation is reported. Nanotips for three-dimensional (3D) atom-probe tomography (APT) investigations were prepared by cutting ~0.3 × 0.3 × 10 mm³ blanks of the aged samples, followed by a two-step electropolishing technique [20]. Tomographic 3-D APT experiments were performed utilizing a picosecond laser-pulsed LEAP 5000XS tomograph (Cameca Instruments Inc., Madison, WI) at 30 K in ultrahigh vacuum (<10⁻⁸ Pa). Ultraviolet (UV) laser pulses (wavelength = 355 nm) were applied with an energy of 30 pJ per pulse and a pulse repetition rate of 500 kHz, while maintaining an average detection rate of 4%. Data analyses were performed using the program IVAS 3.8.2 (Cameca, Madison, WI). The LEAP tomographic datasets were reconstructed using the voltage history during the evaporation. The proximity histogram methodology [37] was utilized to study the compositional variations within the precipitates and the matrix, after performing background corrections to improve the accuracy of the compositional measurements. For scanning electron microscope (SEM) analyses, specimens were ground with a series of SiC papers and then polished with diamond suspensions (6–1 μm) followed by vibratory polishing with a colloidal silica solution (0.06 μm). An FEI Quanta 650 field-emission-gun SEM equipped with an Oxford INCA energy-dispersive-spectroscopy (EDS) detector was used at 5–15 kV for general microstructural investigations. For transmission electron microscope (TEM) analyses, thin slices (~300 μm) of the aged samples were cut using a low-speed diamond saw and ground with a series of SiC papers to reduce their thickness to ~ 100 μm. The specimens were then electropolished with a Struers Tenupol 5 twin-jet polisher to electron transparency, using a solution of 10% nitric acid in ethanol at −10°C. A JEOL ARM300F instrument operating at 300 kV and an FEI Osiris instrument operating at 200 kV were utilized for TEM analyses. Compressive creep experiments were performed at 275, 300, 375 and 400 (± 2)°C under step loadings in air, using cylindrical specimens (10 mm diam. and 20 mm in height) of Al-0.5Mn-0.09Zr-0.05Er-0.05Si peak-aged isochronally with and without prior homogenization (640 °C for 24 h) and Al-0.5Mn-0.02Si in the as-cast and peak-aged conditions without prior homogenization. A linear variable differential transducer (LVDT, a resolution of 10 μm) measured the deformation of these specimens placed between boron-nitride-lubricated tungsten-carbide platens and subjected to a constant load. After the establishment of a steady-state deformation rate for a given load, the minimum strain rate was recorded and the applied load was increased: the process was repeated until the total strain reached ~10% for each specimen. Tensile creep tests on Al-0.5Mn-0.09Zr-0.05Er-0.05Si were performed at 300 and 400 °C utilizing dog-bone shaped specimens with a gauge diam. of 12.7 mm and a gauge length of 50 mm, according to the ASTM E8 standard. One specimen, in the peak-aged state, was used for each stress and temperature conditions. The sample deformation was measured with two extensometers, with a resolution of 1 μm.

Table 1.

Chemical composition of experimental alloys, as determined by inductively coupled plasma optical emission spectroscopy (ICP-OES).

Alloy	Concentration (at.%)
	Mn	Zr	Er	Si	Fe	Al
Al-0.5Mn-0.09Zr-0.05Er-0.05Si	0.53	0.10	0.05	0.05	0.030	Bal.
Al-0.5Mn-0.02Si	0.51	-	-	0.02	0.004	Bal.
Al-0.09Zr-0.01Er-0.02Si [32]	-	0.09	0.01	0.02	0.004	Bal.

3. Results

3.1. Aged microstructure

3.1.1. SEM and TEM analyses of the isochronally aged alloys

The Al-0.5Mn-0.09Zr-0.05Er-0.05Si alloy exhibits coarse grains (<D> ~ 0.5–1.0 mm, Fig. S1a) similar to dilute Al-Sc-Zr and Al-Zr-Er alloys studied previously [21,30,31,38]. Upon isochronal aging of the as-cast Al-0.5Mn-0.09Zr-0.05Er-0.05Si alloy to peak microhardness (450 °C, Fig. S2) a small number density of micron-sized precipitates forms preferentially in the inter-dendritic regions as revealed by SEM analyses, Fig. 1. The STEM-HAADF observations at higher magnifications, Fig. 2, confirm the preferential formation in the inter-dendritic regions of these precipitates with a platelet (irregular polygon) morphology and with Mn, Si and Fe as the main constituent elements, determined by EDS analysis, Fig. 3 and S3. These precipitates are commonly designated as α-precipitates in aluminum alloys [14–17,36,39–41]. No attempts are made herein to determine their crystal structure, due to the weak effect of these precipitates on the mechanical properties of the alloy, which is demonstrated later. Nevertheless, it is known that their crystal structure is body-centered cubic or simple cubic depending on chemical composition (i.e., Mn:Fe ratio and the presence of trace elements, such as boron), with a large lattice parameter, 12–13 Ȧ, corresponding to a cubic approximant phase [14–17,36,39–41]. Neither Zr nor Er partitions to the α-Al(Mn,Fe)Si precipitates, determined by EDS, which indicates that the solubilities of these elements in the α-phase are small (<0.1 wt.%). In addition to the α-microprecipitates in the interdendritic regions, a high number density of L1₂-nanoprecipitates, with a mean radius of <R> ~ 2.5 nm, are observed in the intra-dendritic regions, Fig. 2b,c. The inset diffraction pattern along [001]_Al in Fig. 2c exhibits f.c.c. and L1₂ superlattice reflections, confirming the L1₂ ordering of the nanoprecipitates. Their morphology projected along the [001]_Al-direction was studied utilizing HRTEM. As displayed in Fig. 4, a nanoprecipitate with a radius of ~8.8 nm (located near an inter-dendritic region) is approximately spheroidal and faceted parallel to the {010} and {011} planes, as indicated by the white lines in Fig. 4a, and as revealed in Fig. 4b by fast Fourier transform (FFT) filtering utilizing the superlattice reflections of the L1₂-phase. Smaller nanoprecipitates (<R> ~2–3 nm), which are located at the core of the dendrites of the same sample, Fig. S4, exhibit similar near-spheroidal morphology faceted on the {100} and {110} planes. This nearly spherical faceted morphology suggests a stable growth of the nanoprecipitates during isochronal aging, which is unaffected by Mn additions: similar observations were made for the Mn-free Al–Sc [42] and Al-Sc-Zr [43] alloys aged isothermally at 300 °C, where nanoprecipitates with radii ~ 2–6 nm, were faceted on the {100} and {110} planes when viewed along the [001]_Al zone axis. Herein, HRTEM analyses of the nanoprecipitates along different zone axes were not performed; faceting of the {111} planes, which are the most-closely-packed planes with a low interfacial free-energy are also anticipated, however [42,44,45]. No interfacial misfit dislocations are observed for the nanoprecipitates studied with radii ~2–9 nm, which indicates coherency between the Al(f.c.c.)-matrix and the L1₂-nanoprecipitates as seen, for instance, in Fig. 4c: following a cube-on-cube orientation relationship, the {020} planes of the Al(f.c.c.)-matrix coincide with the {020} planes of the L1₂-nanoprecipitates.

Fig. 1.

BSE-SEM micrographs of the Al-0.5Mn-0.09Zr-0.05Er-0.05Si alloy, peak-aged isochronally (450 °C) from the as-cast state: (a) at a low magnification and (b) boxed area in Fig. 1a at a higher magnification displaying the preferential distribution of micron-sized α-Al(Mn,Fe)Si precipitates (red arrows) in the interdendritic regions around the much coarser primary eutectic Al-Er-Mn-Fe-Si precipitates (white arrows). The white dashed lines highlight the secondary dendrite arms, which are free of both types of precipitates. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 2.

High-angle annular dark-field (HAADF) scanning transmission electron micrographs displaying the microstructure of the Al-0.5Mn-0.09Zr-0.05Er-0.05Si alloy isochronally peak-aged (450 °C) from the as-cast state: (a) and (b) demonstrate the preferential formation of the α-Al(Mn,Fe)Si precipitates in the interdendritic regions. (c) an enlarged view of the intra-dendritic region in (a), displaying homogenously distributed L1₂-nanoprecipitates; no α-Al(Mn,Fe)Si precipitates are observed in these regions. The inset diffraction pattern along [001]_Al in (c) exhibits f.c.c. and L1₂-reflections. A primary Al-Er-Mn-Fe-Si precipitate with a divorced eutectic structure is also observed in the inter-dendritic region in (a).

Fig. 3.

(a) HAADF scanning transmission electron micrograph and (b) corresponding EDS elemental maps of an inter-dendritic region for an isochronally peak-aged Al-0.5Mn-0.09Zr-0.05Er-0.05Si alloy, displaying the elemental distributions of solute atoms in the submicron α-precipitates, which are enriched in Mn, Si, and Fe. A few L1₂-nanoprecipitates, enriched in Zr and Er, are also visible (red arrows). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 4.

(a) High-resolution transmission electron micrograph of an L1₂-nanoprecipitate viewed along the [001] zone axis of the matrix and showing facets on the {010} and {011} planes. (b) Fast-Fourier transform (FFT) filtered image of the nanoprecipitate in (a) utilizing the L1₂ superlattice-reflections, revealing its faceted morphology (c) FFT filtered image along the matrix(f.c.c.)/L1₂ heterophase interface, showing the full coherency of the interface with {020}Al//{020}L1₂.

3.1.2. Atom-probe tomography of the L1₂-nanoprecipitates and Mn partitioning

To study the effect of Mn alloying on the precipitation of the L1₂-nanoprecipitates, APT analyses were performed on the Al-0.5Mn-0.09Zr-0.05Er-0.05Si samples aged isochronally to 300, 425 and 475 °C from the as-cast state. The partitioning behavior of Mn at these different temperatures may shed light on the nucleation, growth and coarsening of the nanoprecipitates. Fig. 5a displays the APT reconstruction of Al-0.5Mn-0.09Zr-0.05Er-0.05Si aged to 300 °C with a microhardness of ~ 350 MPa, which is almost identical to the as-cast microhardness value, Fig. S2a. Very small nanoprecipitates, with a mean radius <R> <0.5 nm, are present in this nanotip, Fig. 5b, as well as a few larger L1₂-nanoprecipitates with R ~ 1–2 nm, Fig. 5c. The average concentration profiles for these small and large nanoprecipitates are displayed in Fig. 5d–e. The small ErSi-Zr-rich nanoprecipitates, Fig. 5b,e, have distinct Si- and Er-rich core and Zr-rich shell, which are formed upon aging according to their relative intrinsic diffusivities in aluminum (D_Si>D_Er>D_Zr). The precipitates are on average richer in Zr than Er, Table 2, and are labelled as (Al,Si)₃(Zr,Er)(L1₂-structure). A small concentration of Mn, $C_{Mn}^{L{1}_2}=1.1$ at.%, is measured in these nanoprecipitates, which is about twice the average Mn concentration of the matrix, $<C_{Mn}^{\alpha -Al}>\text{=}0.495$ at.%, indicative of a weak partitioning of Mn from the supersaturated matrix to the nanoprecipitates: the partitioning ratio is $k_{Mn}^{L{1}_2/\alpha -Al}=C_{Mn}^{L{1}_2}/C_{Mn}^{\alpha -Al}\approx 2.2$, where $C_{Mn}^{L{1}_2}$ and $C_{Mn}^{\alpha -Al}$ are the concentrations of Mn (at.%) in the L1₂- and Al(f.c.c.)-phases, respectively, Table 2. The larger Er- and Si-rich nanoprecipitates are an L1₂-ordered Al₃Er phase with a high solubility for Si. The sum of the Al and Si concentrations measured in this L1₂ phase is close to ~75 at.%, consistent with Si replacing Al on its sublattice, so that the phase can be described as (Al,Si)₃(Er,Zr)(L1₂-structure). Similar observations have been reported for the (Al,Si)₃Sc(L1₂-structure)-[46–48] and (Al,Si)₃Zr(L1₂-structure)-phases [46,48]. The Zr concentration of these larger nanoprecipitates is relatively small, 3.14 at.%, as expected from the very sluggish diffusivity of Zr at 300 °C: the root-mean-square diffusion distance of Zr is ~ 0.75 nm. These (Al,Si)₃(Er,Zr)(L1₂-structure)-nanoprecipitates exhibit a higher average Mn concentration, $C_{Mn}^{L{1}_2}=2.14$ at.%, (with a partitioning ratio of $k_{Mn}^{L{1}_2/\alpha -Al}\approx 4.3$), as compared to the smaller L1₂-nanoprecipitates, Table 2. The higher Mn concentration may be associated with the Gibbs-Thomson effect (interfacial curvature influence on the chemical potential of Mn) [49–51]], and/or may reflect a higher solubility of Mn in Al₃Er(L1₂-structure), which may be an electronic effect, when compared to Al₃Zr(L1₂-structure).

Fig. 5.

(a) APT reconstructions of an Al-0.5Mn-0.09Zr-0.05Er-0.05Si alloy aged isochronally to 300 °C from the as-cast state (HV ~350 MPa, Fig. S2a), displaying small Er-Si-Zr-rich nanoprecipitates (<R> below 0.5 nm) and a few larger L1₂-(Al,Si)₃(Er) nanoprecipitates (<R> ~ 1–2 nm). (b) Elemental distributions in a 5 nm thick slice of the reconstruction in (a), with small nanoprecipitates highlighted employing circles. (c) An example of a larger L1₂-nanoprecipitate with a stoichiometry close to (Al,Si)₃Er. (d) and (e) proximity histograms computed from the nanotip displayed in (a), showing concentration profiles across the matrix/L1₂-nanoprecipitate interface for the large and small L1₂-nanoprecipitates, respectively. The gray shaded areas represent the detection limit (DL) defined as one atom per proxigram bin. The matrix/L1₂ interface (vertical dot-dash lines) is defined as the inflection point of the Al concentration profile. The error bars represent a one-sigma statistical error. Some error bars are smaller than the marker size. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Table 2.

Compositions (as measured using atom-probe tomography) of the L1₂-nanoprecipitates, matrix (far-field) and nanotips for the Al-0.5Mn-0.09Zr-0.05Er-0.05Si alloy aged isochronally to different temperatures.

Aging temperature (°C)	L12 <R>(nm)	Mean precipitate composition (at.%)					Matrix (far-field) composition (at. ppm)				Nanotip composition (at. ppm)
		Al	Zr	Er	Si2+	Mn	Zr	Er	Si2+	Mn	Zr	Er	Si2+	Mn
300	0.4 ± 0.3	74.07	13.19	8.11	3.52	1.10	841	36	116	4947	844	40	119	4950
	1.6 ± 0.1*	60.05	3.14	21.38	13.29	2.14
425	1.9 ± 0.2	71.24	27.42	0.62	0.39	0.33	955	13	219	4302	1384	18	173	4246
475	3.1 ± 0.3	72.05	26.49	0.93	0.21	0.30	410	7	38	4218	1304	31	48	4122

^*(Al,Si)₃(Er) in Fig. 5d

Fig. 6 displays the nanostructure of Al-0.5Mn-0.09Zr-0.05Er-0.05Si aged isochronally to 425 °C (slightly underaged) and 475 °C (slightly overaged). The proximity histograms, computed utilizing the dataset in Fig. 6, are displayed in Fig. 7. An average nanoprecipitate number density of N_v = 6.6 ± 0.9 × 10²² m⁻³ and a mean radius of ⟨R⟩ = 1.6 ± 0.4 nm are measured for the sample aged to 425 °C, while these values are N_v = 4.8 ± 0.3 × 10²² m⁻³ and ⟨R⟩ = 3.5 ± 0.4 nm for the sample aged to 475 °C. As displayed in Fig. 7, both underaged (425 °C) and overaged (475 °C) nanoprecipitates exhibit an Er-enriched core ($C_{Er}^{L{1}_2}~\mathrm{1.5-2}$ at.%) with Zr as the main constituent at ~27 at.%, Table 2. This core-shell structure is similar to the structure of the L1₂-nanoprecipitates in the Mn-free, Al-Zr-Er-based alloys [52–56], and is thus unaffected by the Mn addition. After aging to 425 °C, a small concentration of Mn is present in the nanoprecipitates, $C_{Mn}^{L{1}_2}=0.33$ at.% (Table 2), which is ~20% smaller than the far-field Mn concentration of the Al(f.c.c.) matrix, $<C_{Mn}^{\alpha -Al}>=0.43$ at.%, demonstrating that the Mn partitioning is reversed toward the Al(f.c.c.) matrix, $k_{Mn}^{L{1}_2/\alpha -Al}\approx 0.77$. At 475 °C, the average Mn concentration of the nanoprecipitates decreases slightly to $C_{Mn}^{L_2{1}_2}=0.30$ at.%, while the far-field Mn concentration of the matrix is ~ 0.42 at.%, leading to $k_{Mn}^{L{1}_2/\alpha -Al}\approx 0.71$.

Fig. 6.

3D-APT reconstructions of an Al-0.5Mn-0.09Zr-0.05Er-0.05Si alloy after isochronal aging from the as-cast state to 425 °C (just before peak hardness) and to 475 °C (slightly overaged), showing the core-shell L1₂-nanoprecipitates delineated with 2.5 at.% Zr plus Er iso-concentration surfaces. Only 1% of the Al atoms are displayed for clarity.

Fig. 7.

Proximity histograms of an Al-0.5Mn-0.09Zr-0.05Er-0.05Si alloy after isochronal aging from the as-cast state to: (a) 425 °C (just before peak hardness) and (b) 475 °C (slightly overaged) computed from the datasets displayed in Fig. 6, utilizing the 2.5 at.% Zr plus Er isoconcentration surfaces. The detection limits (DL), defined as one atom per proxigram bin, are too small to be visible on the proximity histograms. The matrix/L1₂ interface (vertical dot-dash red lines) is defined as the inflection point of the Al concentration profile. The mild enrichment of Mn at the Al(f.c.c.)/L1₂ heterophase interface (peak concentration ~ 0.62 at.%) at 425 and 475 °C may be due to the rejection of Mn towards the matrix by the growing nanoprecipitates. At longer aging times, the Mn concentration peak at the interface is anticipated to diminish, driven by minimization of the free energy of the alloy, as observed in an Al-Zr-Sc-Mn-Mo-Si alloy [33] aged isothermally at 400 °C for 11 days.

3.2. Creep behavior at 275–400 °C

Compressive creep tests were performed on Al-0.5Mn-0.09Zr-0.05Er-0.05Si, peak-aged isochronally with and without prior homogenization (640 °C for 24 h) and Al-0.5Mn-0.02Si in the as-cast and peak-aged states. Fig. 8 displays a plot of the minimum creep rate, ${\dot{\epsilon }}_m$, as a function of applied stress, σ, on a double-logarithmic scale for these alloys. Prolonged tensile creep tests lasting 500 h, Fig. S5, were also performed at 300 and 400 °C at low stresses (15–25 MPa) to test the thermal stability of Al-0.5Mn-0.09Zr-0.05Er-0.05Si peak-aged isochronally from the as-cast state. The minimum creep rates obtained from these tensile tests are included in Fig. 8a. All the compressive creep tests, Fig. 8a, are performed in a stress range corresponding to dislocation creep, except for the tests performed at 375 °C in a stress range corresponding to both diffusional and dislocation creep. The apparent stress exponents, $n_a\left(=\partial ln{\dot{\epsilon }}_m/\partial ln\sigma \right)$, in the dislocation creep regimes are significantly higher than that for pure Al (n = 4.4) [57] and they vary with stress (n_a ~ 20 − 35), which is indicative of a threshold stress, σ_th, below which dislocation creep is inhibited [25]. In the L1₂-strengthened aluminum alloys, threshold stresses are attributed to the elastic interaction of dislocations with the strain fields around the coherent nanoprecipitates during the climb bypass process [58,59]. The threshold stresses vary with both precipitate radius and matrix/precipitate lattice parameter mismatch, as reported for several aluminum alloys strengthened with coherent L1₂ nanoprecipitates [1,31,60]. In the presence of a threshold stress, σ_th, the minimum creep rate, ${\dot{\epsilon }}_m$, is expressed through a modified power-law equation [25]:

$${\dot{\epsilon }}_m=A{\left(\sigma -{\sigma }_{th}\right)}^n$$ (1)

where σ is the applied stress, n is the stress exponent for the Al matrix, and $A=K\exp \left(-\frac{Q}{kT}\right)$ contains a constant, K, and the creep activation energy, Q, with kT having its standard meaning. The threshold stresses, Table 3, are estimated employing a best-fit procedure [61]. Our results indicate an unexpectedly high creep resistance for the peak-aged Al-0.5Mn-0.09Zr-0.05Er-0.05Si (with or without homogenization, Section S3), when compared to the Mn-free counterpart alloys, Table 3; for instance, the creep threshold stress at 300 °C is ~ 3 times greater than that of the Mn-free Al-0.11Zr-0.005Er-0.02Si alloy. A similar effect has been reported due to Mn (or Mn plus Mo) additions to a Sc-containing Al-Zr-Sc-Er-Si alloy [35], Fig. S6. This enormous enhancement in creep resistance and the underlying micro-/nano-mechanisms are discussed in the next section, where we examine the isolated influences of different microstructural features on the creep resistance of Al-0.5Mn-0.09Zr-0.05Er-0.05Si by comparing it with the creep resistance of the L1₂-free Al-0.5Mn-0.02Si reference alloy, Fig. 8b, which exhibits much reduced creep rates (by ~ 6 orders of magnitude) in the both as-cast and aged states when compared to pure Al at 300 °C.

Fig. 8.

Double-logarithmic plot of minimum compressive creep strain rate vs. applied stress: (a) at 275, 300, 375 and 400 °C for the Al-0.5Mn-0.09Zr-0.05Er-0.05Si alloy, peak-aged isochronally with and without prior homogenization at 640 °C for 24 h. A transition from diffusional creep at smaller applied stresses to dislocation creep at larger applied stresses is evident at strain rates of ~ 10⁻⁹ s⁻¹ at 375 °C. The tensile creep data from Fig. S5 are also included. (b) at 300 °C for the Al-0.5Mn-0.02Si alloy in the as-cast and peak-aged (475 °C) conditions. The shaded areas represent calculated diffusional creep rates for pure aluminum (Coble plus Nabarro-Herring creep, D = 700 ± 200 μm) utilizing data in Ref. [57].

Table 3.

Creep threshold stresses estimated utilizing the procedure given in [61]. IC and IT indicate isochronal and isothermal aging, respectively.

Alloy	Heat treatment	Creep temperature (°C)	Threshold stress (MPa)	Ref.
Mn-containing alloys:
Al-0.5Mn-0.09Zr-0.05Er-0.05Si	IC peak-aged	275	37	This work
	“	300	34	”
	Homog. & IC peak-aged	“	37	”
	IC peak-aged	375	27	”
	“	400	20	“
	Homog. & IC peak-aged	“	22	“
Al-0.5Mn-0.02Si	IC aged to 475 °C	300	11	“
	As-cast	300	10	“
Mn-free alloys:
Al-0.11Zr-0.005Er-0.02Si	IC peak-aged	300	11	[21]
Al-0.06Zr-0.05Sc-0.01Er-0.04Si	IT peak-aged at 400 °C	400	9	[111]
	IT over-aged at 400 °C	“	14	“
Al-0.06Sc-0.02Zr-0.005Er-0.09Si	IT peak-aged at 400 °C	“	21	[112]
Al-0.08Zr	IC peak-aged	“	7	[113]

4. Discussion

4.1. Solidification and preferential phase formation upon aging

Localized precipitation of two populations of precipitates – α-Al(Mn,Fe)Si microprecipitates in the interdendritic regions and L1₂-nanoprecipitates in the intradendritic regions – is observed in peak-aged Al-0.5Mn-0.09Zr-0.05Er-0.05Si, Figs. 1–2, which is consistent with micro-segregation-induced inhomogeneities in the distribution of solute atoms within the as-cast microstructure on the scale of the secondary dendrite arm spacing, ~10–20 μm. In Fig. 9 we present thermodynamic calculations for this alloy, based on the CALPHAD method [62], displaying the compositional variations within the as-cast microstructure, utilizing ThermoCalc [63], Al-DATA-V6 database [64] and a Scheil model [65]. Negligible diffusion in Al(f.c.c.) and complete mixing in the liquid phase are assumed. Fig. 9 displays the calculated composition of the Al(f.c.c.) matrix as a function of the mole fraction of solid Al(f.c.c.), f_Al, as the dendrites grow outward from the nucleation sites (dendrite cores, f_Al = 0). The first solid (f_Al = 0) starts to form at ~ 660 °C, which is depleted in Mn (0.35 at.%) and enriched in Zr (0.30 at.%) with respect to the bulk alloy composition (0.50 at.% Mn, 0.09 at.% Zr). As the solidification front of each dendrite progresses toward the interdendritic regions with decreasing temperature, the Mn concentration in solid Al(f.c.c.) increases, reaching a peak value of ~ 0.75 at.% (f_Al ~ 0.95), beyond which a sharp decrease in the Mn concentration in Al(f.c.c) occurs due to the formation of primary α-precipitates. The enrichment of Mn is significant for f_Al > ~ 0.7. The Si and Fe enrichments (with the equilibrium solid–liquid partitioning coefficients, $k_o^{FeorSi}<1$[65]) of the interdendritic regions are predicted by the Scheil calculations, Fig. 9. Erbium is excluded from the calculations due to a lack of information in the database. It is, however, expected to behave similarly to Mn, Fe and Si $\left(k_o^{Er}<1\right)$, segregating toward the interdendritic regions. Fig. 9 predicts that only a small fraction of the Al(f.c.c.) matrix, estimated to be ~10%, in the interdendritic regions, is significantly enriched with Mn, Si and Fe, which explains the preferential formation of the α-Al(Mn,Fe)Si precipitates in these regions upon aging, Figs. 1–2. Scheil calculations demonstrate that, away from the interdendritic regions, the Si and Fe concentrations are very small, and the Mn supersaturation is smaller. Consequently, in these regions, Al(f.c.c.) is in local thermodynamic equilibrium with the Al₆Mn or Al₁₂Mn phases [66,67] and the metastable L1₂-phase. The Al₆Mn and Al₁₂Mn phases are, however, known to be extremely difficult to nucleate in the Al-Mn system, which is attributed to the high activation energy for the nucleation of these precipitates, ranging from 1.3 to 1.8 eV [36,68,69]; no significant phase transformation is reported below 327 °C and the nose of the time-temperature-transformation (TTT) curve is in the temperature range 427–627 °C [36]. It was reported [70] that even in a highly supersaturated Al-2.3Mn-1.3Mg-0.5Sc-0.2Zr-0.05Si (at.%) alloy processed with selective laser melting (SLM), Mn tends to remain in solid-solution within the Al(f.c.c.) matrix after 5 h of aging at 300 °C, as determined by APT analyses, demonstrating the high thermal stability of the Al-Mn solid-solution even for very high supersaturations.

Fig. 9.

Calculations of the solute atom distributions within the Al(f.c.c.) matrix of an Al-0.5Mn-0.09Zr-0.05Si alloy during non-equilibrium solidification, utilizing ThermoCalc [63], Al-DATA-V6 database [64], and a Scheil model. Erbium, which is excluded from the calculations due to a lack of information in the database, is expected to behave similarly to Mn due to $k_o^{Er}<1$. Approximate locations of the inter- and intra-dendritic regions are indicated just above the bottom abscissa. In the Al-Zr system, Al(f.c.c.) is assumed to be in equilibrium with the metastable Al₃Zr(L1₂-structure)-phase, consistent with our experimental observations. Note that the Fe solid-solubility in Al(f.c.c.) is much smaller than the Fe concentration in Al-0.5Mn-0.09Zr-0.05Si.

Our isochronal aging experiments of the Zr- and Er-free reference alloy, Al-0.5Mn-0.02Si, section S2, confirms that the partial decomposition of the solid-solution commences above 425 °C. Microstructural analyses of Al-0.5Mn-0.02Si aged isochronally to 475 °C, Fig. S7, reveals that, similar to the peak-aged Al-0.5Mn-0.09Zr-0.05Er-0.05Si, precipitation of the α-phase is limited to the inter-dendritic regions, where micron-sized α-precipitates are observed. It is noted that in our Al-0.5Mn-0.09Zr-0.05Er-0.05Si alloy, the presence of L1₂-nanoprecipitates does not assist the formation of Mn-rich precipitates through a heterogeneous nucleation mechanism; the S/TEM (Fig. 2 and S8) and APT (Fig. 6) analyses reveal no Mn-rich precipitates in the intradendritic regions, indicating that the Al(f.c.c.)/L1₂ heterophase interfaces do not provide a potent nucleation site for the Mn-rich precipitates. This is because both the L1₂ and Al(f.c.c.) phases are cubic and their lattice parameter mismatch is small (~0.05–0.4%) [53].

4.2. Manganese partitioning behavior and its effect on the L1₂-precipitation

a. Mn solubility in the L1₂-nanoprecipitates:

Fig. 10a displays, for three isochronal aging temperatures, the Mn concentration profiles within the L1₂-nanoprecipitates of Al-0.5Mn-0.09Zr-0.05Er-0.05Si, computed from the datasets in Figs. 5–6. The concentration profile at 300 °C is for ~300 small nanoprecipitates, Fig. 5e: the few larger (Al,Si)₃Er(L1₂) nanoprecipitates, Fig. 5d, are excluded from the calculations due to their small number density. Concentration profiles in Fig. 10a indicate that the partitioning of Mn to the L1₂-nanoprecipitates for aging at 300 °C is reversed toward the Al(f.c.c.) matrix at higher aging temperatures, 425 and 475 °C. Conducive to this evolution are: (i) the increased solubility of Mn in the Al(f.c.c.) matrix, Fig. 10b, at high temperatures; and (ii) the anticipated smaller solubility of Mn in the Al₃Zr(L1₂-structure) phase, the bulk of the nanoprecipitates at above 425 °C (when compared to Al₃Er(L1₂-structure), Fig. 5e). The nanoprecipiates formed at 300 °C exhibit a distinct core enriched with Er plus Si; and a Zr plus Er rich shell, Section 3.1.2. Because of the small Mn concentrations measured in the L1₂-nanoprecipitates, particularly in those near the peak-aged conditions, 425 and 475 °C (Table 2), and because of the negligible influence of ~0.4 at.% Mn on the interplanar spacings of Al (determined by synchrotron X-ray diffraction analyses, Table S1), the lattice parameter mismatch between the L1₂-nanoprecipitates and the Al(f.c.c.) matrix is not anticipated to be affected significantly by the Mn additions. Consequently, no significant changes are anticipated in the creep threshold stresses, controlled by the matrix/precipitate lattice parameter mismatch [59,71], due to the Mn modifications of the L1₂-nanoprecipitates. This expectation is based on several other L1₂-strengthened alloys microalloyed with non-L1₂ forming transition metals, such as W [21,30], Cr [31], Mo [32], V [29], Ta and Nb [72], for which the lattice parameter changes of the L1₂-phase with these additions are minimal due to their small solubility in the L1₂-phase.

Fig. 10.

(a) Evolution of the Mn concentration profile within the L1₂-nanoprecipitate of as-cast Al-0.5Mn-0.09Zr-0.05Er-0.05Si upon isochronal aging to 300 °C (well before peak hardness), 425 °C (just before peak hardness) and 475 °C (slightly overaged). The concentration profiles are computed using the datasets in Figs. 5–6 and the proximity histogram methodology. (b) Temperature dependence of the Mn concentration in the Al(f.c.c.) matrix (far-field), $C_{Mn}^{\alpha -Al}$ determined by APT; equilibrium solid solubility of Mn in α-Al, C_o [64,114]; and supersaturation, calculated as $\mathrm{\Delta }\text{C}=C_{Mn}^{\alpha -Al}-{\text{C}}_{\text{o}}$.

b. Mn influence on nucleation and coarsening of L1₂-nanoprecipitates:

Despite Mn diffusing much faster than Zr in aluminum (D_Mn = 6.2 × 10⁻¹⁹ and D_Zr = 1.2 × 10⁻²⁰ m².s⁻¹ at 400 °C [1]), its concentration in the nanoprecipitates at the early stages of nucleation/growth (300 °C, Table 2, Fig. 5) is much smaller than Zr, suggesting a small thermodynamic driving force for its partitioning to the L1₂-phase or co-precipitation with Er, Si and Zr. Furthermore, no Mn clustering or short-range ordering is observed in any of the specimens aged to 300–475 °C by APT or TEM, which could have assisted the nucleation of the nanoprecipitates through a heterogeneous nucleation mechanism as described in Ref. [73]. We will further discuss the distribution of Mn solute atoms in the Al(f.c.c.) matrix later in the text. Thus, it is highly unlikely that Mn can alter significantly the nucleation current (the number of nuclei formed per unit volume per unit time) of the L1₂-nanoprecipitates or their thermodynamic stability. This hypothesis is supported by the nearly identical age-hardening response of the Mn-free and Mn-modified alloys, Fig. S2, demonstrating that the number density and the volume fraction of the L1₂-nanoprecipitates are unaffected by Mn additions. It is noteworthy that the coarsening kinetics of the nanoprecipitates are also unaffected by Mn, as confirmed by the identical isochronal peak temperatures of the Mn-free and Mn-modified alloys, Fig. S2.

4.3. Creep behavior

Our prior investigations on the micro-additions of slow-diffusing transition metals, e.g., V [29], W [21,30], Cr [31] and Mo [32] revealed minimal effects of these additions on the creep resistance of the L1₂-strengthened Al alloys containing Sc and/or Zr (i.e., ~ ± 2–4 MPa difference in σ_th at 300–400 °C). Results from the compressive creep experiments in the present study indicate, however, that the Mn-modified Al-0.5Mn-0.09Zr-0.05Er-0.05Si alloy exhibits a remarkable increase in creep resistance (with σ_th in the range 35–37 MPa at 300 °C), when compared to its Mn-free L1₂-strengthened counterpart (with a σ_th of ~ 11 MPa): this is a greater than 3-fold increase in σ_th with a modest (an inexpensive) addition of 0.5 at.% Mn, Fig. 11a and Table 3. The main question in this context is whether this increase in creep resistance imparted by Mn is due to the strengthening of the L1₂-free interdendritic regions by the preferential formation of the α-Al(Mn,Fe)Si precipitates, Figs. 1–2, given that we observe no significant changes in the L1₂-nanoprecipitates with Mn additions, neither in their structure, composition or morphology nor in their number density, volume fraction or coarsening resistance, Section 4.2. Although no significant hardening at room temperature was obtained by the small number density of relatively coarse α-Al(Mn,Fe)Si microprecipitates in Al-0.5Mn-0.02Si, Figs. S2,S7, these precipitates, even at low number densities, have been shown to be effective strengtheners at high temperatures in different alloys without the L1₂-nanoprecipitates [14–19,40,74]. The situation for our peak-aged Al-0.5Mn-0.09Zr-0.05Er-0.05Si is, however, complex because of the significant amount of Mn remaining in solid-solution within the Al(f.c.c) matrix (~0.42 at.%), which is much above its solubility limit, Fig. 10b: the supersaturation (ΔC) of the Al(f.c.c) matrix with Mn at 450 and 300 °C is ΔC ~ 0.30 and ~0.45 at.%, respectively. Although thermodynamics favor the decomposition of the solid-solution during creep testing, Mn remains mainly within the matrix at the creep temperatures (275–400 °C) for tests lasting over two months, due to the extremely sluggish precipitation kinetics of Mn in Al, Section 4.1, which is corroborated by identical values of conductivity of the specimens measured before and after creep testing, and by the stable steady-state creep regimes during prolonged (500–1600 h) tensile creep tests at 300 and 400 °C, Figs. S5 and S9. To date, whether and how Mn in solid-solution plays a significant role in reducing the creep deformation rates of Al-0.5Mn-0.09Zr-0.05Er-0.05Si with a complex microstructure remains unclear. Prior research on the creep deformation of the Mn-containing alloys has attributed their creep strength primarily to the Mn-containing α-precipitates and the role of Mn in solid-solution is not typically addressed [75–78]. To obtain a clear understanding of the underlying creep mechanisms of our Mn-modified alloys, we replot, Fig. 11a, the compressive creep data at 300 °C for Al-0.5Mn-0.09Zr-0.05Er-0.05Si peak-aged isochronally from the as-cast state, Fig. 8. Also, for comparison, we include creep data for a Mn-free, L1₂-strengthened Al-0.11Zr-0.005Er-0.02Si alloy, peak-aged isochronally, from the archival literature [21] as well as our coarse-grained, L1₂-free, Al-0.5Mn-0.02Si reference alloy, crept in two different conditions (Fig. 8b): (i) as-cast, as a single-phase solid-solution alloy, Fig. S1b; and (ii) aged isochronally to 475 °C, Fig. S2, to form α-Al(Mn,Fe)Si precipitates preferentially in the interdendritic regions, Fig. S7, similar to the peak-aged Al-0.5Mn-0.09Zr-0.05Er-0.05Si, but without the L1₂-nanoprecipitates. In Fig. 11b, creep data for Al-0.5Mn-0.09Zr-0.05Er-0.05Si at 400 °C are compared with the available data in the archival literature, again displaying the exceptional creep resistance of our alloy at ~ 70% of its solidus temperature. By analyzing the data in Fig. 11a, we seek to isolate the individual effects of three different strengthening mechanisms on the overall creep resistance of Al-0.5Mn-0.09Zr-0.05Er-0.05Si: (i) Mn in solid-solution; (ii) micron-sized, inter-dendritic α-Al(Mn,Fe)Si precipitates; and (iii) nano-sized, intradendritic L1₂-precipitates. We first discuss the two L1₂-free Al-0.5Mn-0.02Si alloys with coarse grains (D ~ 0.7–1.0 mm), crept at 300 °C, Fig. 11a.

Fig. 11.

Double-logarithmic plot of minimum compressive creep strain rate vs. applied stress at: (a) 300 °C and (b) 400 °C, for the Al-0.5Mn-0.09Zr-0.05Er-0.05Si alloy, peak-aged isochronally from the as-cast state, an as-cast Al-0.5Mn-0.02Si solid-solution alloy and a peak-aged Al-0.5Mn-0.02Si alloy. Also included for comparison are creep data from archival literature: (a) at 300 °C, a Sc-free alloy (Al-0.11Zr-0.005Er peak-aged isochronally [21]); (b) at 400 °C, two Sc-containing alloys (Al-0.06Zr-0.05Sc-0.01Er-0.04Si peak- and over-aged [111] and Al-0.06Sc-0.02Zr-0.005Er-0.09Si peak-aged [112]), and a Sc-free Al-0.08Zr alloy [113]. The shaded areas represent calculated diffusional creep rates for pure aluminum (Coble plus Nabarro-Herring creep, D = 700 ± 200 μm) utilizing data in Ref. [57]. The best-fit lines follow Eq. 1, with threshold stress values marked above the abscissa. All the compositions are in at.%.

4.3.1. Effect of Mn as solute atoms within Al(f.c.c.)

At 300 °C, very low creep rates (~10−10s−1) were measured for the as-cast Al-0.5Mn-0.02Si solid-solution alloy at ~ < 13.5 MPa, which are 6–7 orders of magnitude smaller than those for pure aluminum, Fig. 11a. The primary creep deformation at these stresses was also negligible, (e.g. < 0.05% at 9.5 MPa), Fig. S9. The creep rates for stresses below ~13.5 MPa appear to be of the order of the diffusional creep rates of pure Al, for the same grain diam., 0.7 ± 0.2 mm, shown as a shaded area, Fig. 11a. This alloy exhibits an apparent stress exponent of na ~ 9–7 over three orders of magnitude of strain rates (from ~10−10 to ~10−7 s−1), which is significantly greater than the stress exponents observed in pure Al undergoing dislocation creep (n ~ 4.4), or in concentrated solid-solution alloys: Class II, glide-controlled alloys, e.g., Al-Mg [79–81] (n~3, with no significant primary creep deformation), or Class I, climb-controlled alloys, e.g., Al-Zn and Al-Li [81–83] (n~5, with large primary creep behavior). A transition from n ~ 5 to n ~ 7 (n+2) prior to a power-law breakdown, has been reported at low temperatures (~ half the absolute melting point of aluminum, 0.5Tm = 467 K, or 194 °C), when core diffusion becomes significant [57,80,81,84]. This transition, however, occurs at high stresses [80] and relies on the existence of subgrains with a relatively small diameter for a dominant core diffusion mechanism, which requires significant (> ~5 %) primary-creep deformation [85–87]. This is inconsistent with the small primary-creep strains measured for our solid-solution Al-0.5Mn-0.02Si at 300 °C, Fig. S9, and with a higher testing temperature. The creep characteristics described above – i.e., na ~ 9–7 and negligible primary creep – indicate that dislocation-controlled creep is reduced dramatically at stresses below ~13.5 MPa, where creep rates are on the order of those expected through diffusional creep. That is, a threshold stress for dislocation creep exists at stresses < ~13.5 MPa. Using the least squares fit procedure, Ref. [61], described above with n = 4.4, the creep data in Fig. 11a can be fitted to Eq. 1, yielding an estimated threshold stress of ~10 MPa. The origin of this threshold stress is discussed in Section 4.3.3.

4.3.2. Effect of Mn as interdendritic α-Al(Mn,Fe)Si precipitates

We now consider the peak-aged Al-0.5Mn-0.02Si alloy, which contains micron-size, inter-dendritic α-Al(Mn,Fe)Si precipitates, Fig. S7, in a matrix that remains supersaturated with Mn: ~0.4 at.%, estimated from the conductivity measurements, Fig. S2b, utilizing Nordheim’s rule [88,89]. In the aged state, the ternary Al-0.5Mn-0.02Si alloy exhibits slightly higher creep rates (by a factor ~ 3) and apparent stress exponent (n_a ~10 vs. n_a ~9–7), compared to the as-cast solid-solution alloy, Fig. 11a. A threshold-stress-type behavior is noted, as illustrated by the non-linear fit exhibited in Fig. 11a using Eq. (1). Comparing the creep characteristics of the as-cast and the aged Al-0.5Mn-0.02Si, reveals that interdendritic α-Al(Mn,Fe)Si precipitates have only a small influence on the threshold stress (a negligible net increase of ~ 1 MPa), but lead to slightly higher creep rates at a given stress above the threshold value (e.g., above ~10⁻⁸ s⁻¹). We attribute this small decrease in creep resistance of the aged Al-0.5Mn-0.02Si at strain rates > ~10⁻⁸ s⁻¹ to a decrease in the Mn concentration of its Al(f.c.c.) matrix (~0.4 at.%, as opposed to ~0.5 at.% in the as-cast alloy). Thus, the net effect of peak-aging the Al-0.5Mn-0.02Si alloy (slightly reducing the Mn concentration of the matrix while forming a small number density of the interdendritic α-precipitates) on its creep threshold stress at 300 °C is small, which indicates that the contribution of the α-Al(Mn,Fe)Si-precipitates to the threshold stress is small, and just enough to compensate for the slightly reduced Mn concentration of the matrix. By extension, we anticipate negligible effects of the interdendritic α-Al(Mn,Fe)Si-precipitates on the creep resistance of the peak-aged Al-0.5Mn-0.09Zr-0.05Er-0.05Si (certainly not explaining the 3-fold increase in threshold stress), as their mean radius, number density, volume fraction and distribution are anticipated to be similar to those in the aged Al-0.5Mn-0.02Si alloy.

Thus, it appears that Mn, when in the solid-solution, imparts a major increase in the creep resistance of Al. In fact, as shown in Fig. 11a, the as-cast solid-solution Al-0.5Mn-0.02Si exhibits a higher creep-resistance than the L1₂-strengthened, Mn-free Al-0.11Zr-0.005Er-0.02Si alloy in the peak-aged state at 300 °C, while at ambient temperatures Al-0.5Mn-0.02Si (HV ~300 MPa) is much weaker than the peak-aged Al-0.11Zr-0.005Er-0.02Si (HV ~500 MPa). A fundamental question is how does Mn in solid-solution impart such extraordinary creep resistance to Al(f.c.c.). One possible explanation could be the clustering or short-range-ordering (SRO) of Mn atoms within the Al(f.c.c.) matrix, due to the supersaturation of the matrix with Mn, specifically at the creep temperatures. The clusters or SRO regions, if they exist, can pin dislocations, as reported for many different alloy systems [90–94], and hinder climb processes, leading to a higher creep resistance and possibly a threshold stress below which the density of mobile dislocations is negligible. Our further APT analyses, Fig. 12, provides, however, evidence that this hypothesis is invalid. Fig. 12 displays the Mn-centered partial radial distribution functions (PRDFs) within the Al(f.c.c.) matrix of a peak-aged (450 °C) Al-0.5Mn-0.09Zr-0.05Er-0.05Si before and after creep testing at 300 °C for one month under stresses ranging from 35 to 50 MPa. The Mn-Mn PRDF at a given radial distance is defined as the average concentration distribution of Mn around a given Mn solute, normalized to the overall concentration of Mn in the matrix [95]. By definition, the PRDF values of unity describe perfectly random distributions. The L1₂-nanoprecipitates, as well as the crystallographic poles and zone lines running throughout the APT volumes, are excluded from the calculation to limit the analyses to the Al(f.c.c.) matrix and avoid artifacts associated with the surface migration of atoms during evaporation. In both samples, before and after creep testing, the Mn-Mn PRDF values are close to unity, indicating that no significant clustering of the Mn solute atoms occurs in these highly supersaturated matrices before or during creep testing at 300 °C, Fig. 12a–b. We note a weak negative (repulsive) Mn-Mn correlation at the first nearest neighbor (NN) distances, in both cases, which is consistent with repulsive Mn-Mn binding energies at first NN distances (~ 0.35 eV at 0 K) [96], which makes Mn clusters in Al thermodynamically unstable. The matrix composition of this alloy, Table 2, is close to the composition of the Al-0.5Mn-0.02Si solid-solution alloy, thus similar Mn distributions are expected within the matrix of both alloys. From these analyses, we deduce, with a high degree of certainty, that the randomly-distributed Mn solute atoms in the Al(f.c.c.) matrix impart a very high creep resistance to aluminum.

Fig. 12.

Partial radial distribution functions (PRDFs) with respect to the Mn solute atoms within the Al(f.c.c.) matrix of an Al-0.5Mn-0.09Zr-0.05Er-0.05Si alloy. (a) Aged isochronally to 450 °C (peak microhardness) from the as-cast state and (b) after additional creep deformation at 300 °C for one month under stresses ranging from 35 to 50 MPa (cooled under stress). The analyses reveal a lack of significant clustering of Mn solute atoms in both cases as demonstrated by the Mn-Mn PRDF values close to unity. We note a weak negative (repulsive) Mn-Mn correlation at the first nearest neighbor (NN), in both cases.

4.3.3. Creep mechanisms

As discussed earlier, the creep characteristics of the Al-0.5Mn-0.02Si solid-solution alloy are not fully consistent with any of the known creep mechanisms of random solid-solution alloys. We also ruled out the possibility of Mn clustering and dynamic precipitation (i.e., precipitation during creep deformation), which could potentially lead to a threshold-stress-type behavior. Nonetheless, there seems to be a plausible connection between the creep behavior of the Al-0.5Mn-0.02Si solid-solution alloy and those of an extremely dilute Al-0.02Fe (at.%) alloy [23,28] discussed earlier. In both cases, apparent stress exponents higher than 3–5 (n_a ~9–7 for Al-0.5Mn-0.02Si and n_a ~ 5–25 for Al-0.02Fe) were observed and some type of dislocation-climb-controlled mechanism is inferred to be active, in which slow-diffusing solute atoms play a key role. The creep mechanism proposed by Sherby et al. [23,28] for the Al-0.02Fe alloy, which is based on a conceptual model for solute diffusion along subgrain boundaries, may not be applicable for Al-0.5Mn-0.02Si, as it relies on the formation of subgrains, which require significant primary creep strain (~ 20% in case of Al-0.02Fe [23]) before the establishment of the steady-state creep. This is contrary to the extremely small primary creep measured in our Al-0.5Mn-0.02Si (~0.05%, Fig. S9) and Al-0.5Mn-0.09Zr-0.05Er-0.05Si alloys (Fig. S5) and cannot explain convincingly the threshold-stress-type behavior observed in the former alloys. These characteristics suggest strongly, however, that individual dislocations are pinned by the Mn solute atoms at the onset of creep deformation in Al-0.5Mn-0.02Si, which is consistent with the existence of a threshold stress and thereby an apparent stress exponent greater than n = 4.4. It is well-known that dragging of a Cottrell atmosphere by edge dislocations plays an important role not only in the glide motion of dislocations, Weertman [97–99] for Class II alloys (n ~3) but also in their climb as proposed by Takeuchi and Argon [100,101]. It is reasonable to assume that the hinderance of climb by solute atoms depends on: (i) the existence of a solute atmosphere, which relies on a strong interaction between the solute atoms and the edge segment of individual dislocations; and (ii) a smaller diffusivity of solute atoms than the matrix atoms. In the absence of strong dislocation/solute interactions, slow-diffusing solute atoms may not have a significant influence on the dislocation mobility and thereby creep rate. This important argument is often not discussed in creep theories of solid-solution aluminum alloys, perhaps because there are few elements which fulfill all three conditions: (i) strong interactions with edge dislocations; (ii) diffusion that is much slower than self-diffusion in Al; and (iii) sizable solubility in Al.

Dislocation pinning by Mn solute atoms for Al-0.5Mn-0.02Si is anticipated to be Cottrell-type elastic interactions [102], rather than Suzuki-type chemical interactions with stacking faults [103], as density functional theory calculations have shown that Mn segregation to stacking faults in Al is unfavorable energetically [104]. We note that the Snoek-type (stress-induced short-range ordering) interactions [105], although unlikely, cannot be ruled out entirely and require further investigations. Our hypothesis that Mn solute atoms interact strongly with climbing edge dislocations, slowing the creep rates, is strengthened by a first-principles study performed by Vannarat et al. [106]. Fig. 13, redrawn with some modifications from this study [106], displays Mn, Fe and Si concentrations enhancements around an edge dislocation in Al. Also included in Fig. 13 are data for Zr, estimated here employing a similar approach to that in Ref. [106]. Each shaded circle represents approximately the region around the dislocation core where the concentration of each solute species is greater than 10 times the far-field concentration in the Al(f.c.c.) matrix. The corresponding calculated interaction energies between these solute atoms and an edge dislocation are presented in Table S2. The strong segregation of Mn and Fe atoms to the compressive area above the dislocation core are predicted due to their high interaction energies with dislocations, Table S2 (0.54 and 0.49 eV, respectively). Nonetheless, in our Mn-containing alloys, specifically the peak-aged Al-0.5Mn-0.09Zr-0.05Er-0.05Si and Al-0.5Mn-0.02Si, the atmosphere of Fe atoms around the dislocations is deemed negligible, due to its very small concentration in the Al(f.c.c) matrix (< ~50 at. ppm). The Mn atmosphere is, however, pronounced, which is consistent with our creep results; a relatively large region with a diameter of ~1.5 nm above the dislocation core is predicted with Mn concentrations greater than ~4 at.% (a ten-fold increase in their far-field value of 0.4 at.%). Fig. 13 also demonstrates that the Zr and Si atmospheres are much smaller and hence unlikely to exert significant drag forces on dislocations during creep, compounded by the small interaction energies of these elements with an edge dislocation and their low concentrations in the matrix.

Fig. 13.

Solute redistribution around an edge dislocation in Al, re-drawn for Mn, Fe, and Si, with some modifications, from Ref. [106], where a combined quantum mechanical and continuum isotropic elasticity approach was employed; data for Zr is estimated here utilizing a similar approach. The dislocation core at (0,0) is represented by a dashed red circle (radius ~ the magnitude of the Burgers vector) with the extra-half plane on the Y > 0 side. Each circle represents approximately the region around the dislocation where the concentration of each solute species is greater than 10 times the far-field concentration of the Al(f.c.c.) matrix, whose value is given in parentheses for the peak-aged Al-0.5Mn-0.09Zr-0.05Er-0.05Si alloy. It is assumed that within the dislocation core, where the continuum model is invalid, the solute concentration is the same as the concentration just outside the core. Strong Fe and Mn segregation at edge dislocations is predicted, while the segregation of Zr and especially Si, is anticipated to be weaker. All the quantum mechanical calculations are performed at 0 K.

4.3.4. Solute-induced threshold stress

These results give strong support for the view that certain slow-diffusing solute atoms such as Mn, at moderately low concentrations (e.g., ~ 0.4–0.5 at.%), can have a major effect on the creep resistance of aluminum alloys, which does not appear to be addressed in the current Al creep literature. The slow-diffusing solute atoms that interact weakly with dislocations, such as Zr, are anticipated to have negligible influences on the creep rate. These elements, if present at sufficiently high concentrations, may indeed reduce the glide-controlled creep rate by increasing the critical resolved shear stress [107,108]. These hypotheses, along with supporting evidence and its implications for alloy design, will be further discussed elsewhere [109]. Although the solubilities of the slow-diffusing elements in Al (most of the transition metals) are small [1], rapid cooling techniques (e.g., selective laser melting) can achieve an order-of-magnitude higher concentration of solute atoms in Al. These highly supersaturated solid-solution alloys can be thermally stable at elevated temperatures [70]. Thus, a solute-induced threshold stress (SITS), which we propose herein for our cast Al-Mn solid-solution alloys, can be expected to be an even more important phenomenon in such alloys, and critical to the interpretation of their creep behavior. Our results indicate that the L1₂-strengthened alloys can be made even more creep-resistant with Mn in solid-solution within the Al(f.c.c.) matrix, as demonstrated for the Al-0.5Mn-0.09Zr-0.05Er-0.05Si alloy, mainly because of: (i) a lack of deleterious effects of Mn on the coarsening resistance or strengthening effectiveness of the L1₂-nanoprecipitates at high temperatures, Fig. S2; (ii) a high thermal stability of the Al-Mn solid-solutions at temperatures below 425 °C, Figs. S2b,10b; (iii) a relatively high solid-solubility of Mn in Al (~0.7 at.% as compared to the other slow-diffusing elements in Al); and (iv) strong interaction energies of Mn solute atoms with edge dislocations, Fig. 13. In the peak-aged condition for Al-0.5Mn-0.09Zr-0.05Er-0.05Si (σ_th = 34–37 MPa) the threshold stresses may, however, contain two contributions: (i) from the L1₂-nanoprecipitates in the intradendritic volumes (estimated to be equal to the threshold stress of the peak-aged Mn-free Al-0.11Zr-0.005Er-0.02Si, σ_th ~ 11 MPa); and; (ii) from the solute-pining effect described above (SITS) together with a small contribution from the interdendritic α-Al(Mn-Fe)Si-precipitates (estimated to be that of the peak-aged Al-0.5Mn-0.02Si, σ_th ~ 11 MPa). Adding these contributions yields 22 MPa, which is smaller than the observed threshold stress of Al-0.5Mn-0.09Zr-0.05Er-0.05Si (σ_th = 34–37 MPa). Such a synergetic effect is expected as the dynamics of the dislocation climb processes in Al-0.5Mn-0.09Zr-0.05Er-0.05Si is complex, involving dislocations surmounting the L1₂-nanoprecipitates via a climb mechanism, while dragging their Mn atmosphere. Of the possible different origins of this synergy one can cite changes in the dislocation line energy due to Mn segregation, which can alter the contribution to the overall threshold stress caused by the L1₂ nanoprecipitates. The elastic interactions between the L1₂-nanoprecipitates and dislocations with and without a Mn atmosphere are expected to differ as well. In view of our ignorance of these complex processes, a detailed analysis of their kinetics is left for future studies.

The solute-induced threshold stress is expected to diminish gradually with increasing creep temperature as the excess concentration of solutes atoms around the dislocation cores decreases: segregation decreases with increasing temperature due to the entropic term in the Gibbs free energy of segregation [110]. All creep experiments on the Al-0.5Mn-0.02Si solid-solution alloy were conducted at 300 °C. The creep experiments performed on the peak-aged Al-0.5Mn-0.09Zr-0.05Er-0.05Si alloy at 400 °C, however, confirm the reduced solute-induced threshold stress, demonstrated by the smaller differences between the threshold stress of Al-0.5Mn-0.09Zr-0.05Er-0.05Si and all the Mn-free alloys at 400 °C, Fig. 11b (Δσ_th ~25 MPa at 300 °C, vs. ~2–15 MPa at 400 °C).

5. Summary and conclusions

The microstructure and creep properties of a cast Al-0.5Mn-0.09Zr-0.05Er-0.05Si alloy and its Zr- and Er-free counterpart, Al-0.5Mn-0.02Si (at.%) are studied and compared to those of Mn-free L1₂-strengthened alloys. The SEM and STEM analyses, Figs. 1–2, reveal that in the peak-aged Al-0.5Mn-0.09Zr-0.05Er-0.05Si, nanoscale Al₃(Zr,Er)(L1₂-structure) precipitates are formed in the dendrite cores, surrounded by micron-sized α-Al(Mn,Fe)Si-precipitates in the interdendritic channels. Electrical conductivity and APT measurements demonstrate that, despite the formation of inter-dendritic α-Al(Mn,Fe)Si precipitates, most of the Mn remains randomly distributed in solid-solution in the Al(f.c.c.) matrix of the peak-aged Al-0.5Mn-0.09Zr-0.05Er-0.05Si and Al-0.5Mn-0.02Si alloys at concentrations above its solubility limit. The influence of Mn on the nucleation, growth and coarsening of the L1₂-nanoprecipitates is insignificant, and the aging behavior is identical for the Mn-modified Al-0.5Mn-0.09Zr-0.05Er-0.05Si alloy and its Mn-free counterpart, Al-0.09Zr-0.01Er-0.02Si, Fig. 2S. A pronounced increase in creep resistance is observed, however, in the peak-aged Al-0.5Mn-0.09Zr-0.05Er-0.05Si alloy due to the presence of Mn in solid-solution. The following main conclusions are reached:

1. As-cast, precipitate-free Al-0.5Mn-0.02Si exhibits creep characteristics: apparent stress exponent n_a ~ 9–7, negligible primary creep and creep slower than pure Al by 6 to 7 orders of magnitude, which are consistent with the presence of a threshold stress, Fig. 8b. This threshold stress is attributed to interactions of Mn solute atoms with edge dislocations, leading to strong drag forces exerted on them during the climb processes. We denote this phenomenon a “solute-induced threshold stress (SITS)”.

2. The significantly enhanced creep resistance of the L1₂-strengthened Al-0.5Mn-0.09Zr-0.05Er-0.05Si alloy as compared to the Mn-free L1₂-strengthened counterparts, Fig. 11, is attributed to Mn-solute-induced retardation of climb processes over the precipitates: i.e., a synergy between Mn-induced drag on dislocations and climb-bypass threshold stresses.

3. The interdendritic α-Al(Mn,Fe)Si-microprecipitates formed upon aging of the Al-0.5Mn-0.02Si alloy depletes partially Mn from the solid-solution and provides no significant net increase in the threshold stress. And by extension, the influence of the interdendritic α-precipitates on the creep resistance of the Al-0.5Mn-0.09Zr-0.05Er-0.05Si alloy is also anticipated to be small.