Stabilizing nanoprecipitates in Al-Cu alloys for creep resistance at 300°C

ABSTRACT Commercial precipitation-hardened Al-Cu alloys are normally used at temperatures below its ageing temperature of ∼225°C, to avoid thermally induced precipitate coarsening and resultant softening. Making such popular Al alloys creep resistant at or above 300°C is thus challenging. Here we present a modified precipitation protocol, exploiting Sc-microalloying and a carefully designed three-step heat treatment to enhance Sc segregation at the matrix/θ′-Al2Cu precipitate interfaces. The stabilized nanoprecipitates enable an order-of-magnitude reduction in creep rate at 300°C, demonstrating the room for microstructure improvement and the potential for property elevation in traditional engineering alloys through innovative processing coupled with synergetic alloying elements. GRAPHICAL ABSTRACT IMPACT STATEMENT A carefully designed precipitation protocol produces stable nanoprecipitates in Sc-microalloyed Al-Cu alloys, rendering an order-of-magnitude elevation of creep resistance at 300°C.


Introduction
Precipitation-hardened Al alloys are arguably the most popular lightweight engineering material for structural applications at room temperature [1][2][3]. They are also highly desirable for applications at elevated temperatures (such as 300°C) [4], for example in fuselage bulkheads and wing skins in high-performance military aircrafts, automotive engines, and heat-resistant conductors in power transmission, if the microstructure of these Al alloys can remain stable under moderately hot conditions. However, the challenge is that such a desired service temperature is considerably above the age-hardening temperature, which is < ∼ 225°C for conventional Al alloys to acquire strengthening precipitates [5][6][7][8], such as the θ -Al 2 Cu in the commercial 2000-series Al alloy based on Al-Cu [8][9][10][11][12]. This classical precipitation strengthening system is the basis for a wide range of age-hardening Al alloys used in cast conditions, and will be the base alloy of our study. At service temperatures such as 300°C, thermally induced diffusion of the alloying element (Cu, etc.) becomes so rampant that the θ -Al 2 Cu precipitates coarsen rapidly, leading to severe loss in strength [13,14].
However, the formation of strengthening Al 3 Sc requires an ageing temperature of 300-350°C, which is ∼ 100°C higher than the θ -Al 2 Cu-precipitation temperature, thus coarsening and dissolving away the θ -Al 2 Cu precipitates. Owing to the rather limited solubility of Sc in Al, the Al 3 Sc population is usually too low (volume fraction, f v < 1%) to make up for the yield strength drop due to the loss of the θ -Al 2 Cu precipitates (f v > 2%) [3,14,[16][17][18][19]. Previously, several microstructural designs have been put forward in Al-Cu-based alloys, including more coarsening-resistant phase [6,7,13] via Mg/Ag co-addition and synergistic coupling among multiple precipitates [8,12]. Recently, Sc segregation at the α-Al/θ interfaces has been shown to effectively suppress the coarsening of the θ -Al 2 Cu precipitates [18,19]. However, if the ageing temperature is kept low using the normal precipitation protocol for θ -Al 2 Cu precipitates, the precipitation time allowable before over-ageing would not be sufficiently long for the vast majority of the slowly segregating Sc atoms to reach the α-Al/θ interfaces to stabilize the fast-coarsening θ -Al 2 Cu precipitates (see below). By the time the Al 3 Sc form throughout the alloy upon heat treating, the fast-appearing θ -Al 2 Cu precipitates would have already formed some time earlier and coarsened.
As such, the traditional precipitation hardening routine falls short, when it comes to exploiting the majority of Sc solutes in the supersaturated Al-Cu-Sc solution, in binding them to θ -Al 2 Cu precipitates to suppress coarsening. To come up with a solution, we have been inspired by the process originally developed for the 7000-series Al alloys to achieve an optimal balance between strength and corrosion resistance, known as the Retrogression and Re-ageing(RRA) heat treatment [20,21], and multi-step/isochronal ageing in Al-Sc based alloys [3,16]. The highlight of our strategy is to apply a regression step to re-precipitate the θ -Al 2 Cu, after redissolving (akin to solution treatment) the θ -Al 2 Cu precipitates back into the matrix at an elevated temperature, a temperature high enough for the θ -Al 2 Cu precipitates to dissolve, but low enough such that the Al 3 Sc precipitates stay put. Afterwards, the θ -Al 2 Cu precipitates are Particle size distribution of θ -Al 2 Cu precipitates, comparing different heat treatments, as well as creep test versus heating at 300°C. The precipitate sizes (L) were measured using TEM after each treatment, with the average size L aver marked for each distribution curve. The x-axis is in reduced unit, using normalized precipitate dimension L/L aver . deployed a second time at a lowered ageing temperature, re-precipitating heterogeneously and preferentially on the Al 3 Sc precipitates. We can simultaneously achieve the following three goals by manipulating the nanoprecipitates: (i) a high density of θ -Al 2 Cu precipitates of the right sizes as dislocation obstacles for strength at both room temperature and elevated temperatures, (ii) narrow size distribution of the θ -Al 2 Cu precipitates to reduce the driving force for their coarsening, and most notably (iii) affluent segregation of Sc that quickly accrues from nearby regions to effectively enclose the θ -Al 2 Cu precipitates to suppress their coarsening. We will show a much reduced steady-state creep rate at 300°C and under stresses (30-60 MPa) in uniaxial tension.
This finding opens a new opportunity for the selection of popular Al-Cu based alloys at moderately elevated temperatures.

Experimental
Al-0.3 wt.%Sc and Al-2.5 wt.%Cu-X wt.%Sc (X = 0, 0.3) alloys were melted and cast in an argon stream, see composition in Table S1. Microstructures of the alloys were characterized using transmission electron microscopy(TEM), high-resolution TEM(HRTEM) and atom probe tomography(APT). Due to space limitation, experimental details are presented in Supplemental Information.

Step I (AA treatment)
The first and beginning step of our processing protocol, see Figure 1(a), after the solution treatment, was artificial ageing(AA) at 250°C, akin to the traditional T6 treatment for conventional Al alloy. This Step I forms θ -Al 2 Cu precipitates with an average size (half plate length) L = 330 nm, see Figure 2(a-b) and Table S2, smaller than in the Sc-free Al-Cu alloy under the same AA condition (L = 539 nm). Figure 1(b) compares the size distribution of θ -Al 2 Cu precipitates after different treatments (and tests), and indicates that the size distribution of θ -Al 2 Cu precipitates is the most uniform after applying our new precipitation protocol (see results later). A high density of Sc-enriched heterogeneities(precursors of Al 3 Sc) were evident within the matrix(see Figure 2(d), referred to as 'Sc clusters' [18]), with core Sc concentration similar to Al 3 Sc in previous investigations [15,17], see APT results in Figure 3(a). No coherency strain(Ashby-Brown contrast [16]) or lattice change was observed under TEM and HRTEM, nor diffraction spots indicative of Al 3 Sc precipitates in the SAED pattern, see Figures 2(b) and 4(a). This is in contrast with the well-developed Al 3 Sc particles that have easily identifiable crystal structure (Figure 4(b)). However, some weak diffraction patterns were observed in the FFT images, see Figure 4(a), possibly originating from some already formed Al 3 Sc precipitate that are very small in size. This is in accord with the slow Al 3 Sc transformation at 250°C [17]. The proxigram was displayed in Figure 3(b) to show the solute concentration across an α-Al/θ interface in the Al-Cu-Sc AA alloy. The solid vertical line represents the position of the coherent interface between the Al matrix and the θ -Al 2 Cu precipitate. A peak Sc concentration of about 0.8 at.% is detected at the α-Al/θ interface, about 10 times greater than in the matrix [18,19]. The interfacial Sc segregation layer is about 1 ∼ 2 nm wide (FWHM of the concentration profile). However, the Al-Cu-Sc AA alloy still displays a rather low creep resistance ( Figure 5) because Sc atoms segregated at the matrix/precipitate interfaces constitute of only about 2.6% of the total Sc atoms added in the alloy, see Table S3. Nearly 90% Sc atoms remain far away from the θ -Al 2 Cu precipitates, as in numerous locations the Sc atoms are accrued into precursors of Al 3 Sc not in the close proximity of θ -Al 2 Cu precipitates. Sc diffusion is so slow such that during AA only a small number of Sc atoms manage to reach and deposit at the matrix/precipitate interfaces. As such, AA alone is insufficient for effective blocking of precipitate coarsening for satisfactory creep resistance.

Step II (Regression)
During Step II, a regression treatment at elevated temperature of 450°C (Figure 1(a)), the θ -Al 2 Cu precipitates formed during the lower-temperature AA treatment would decompose, letting the Cu atoms dissolve back into the matrix (which was depleted to only ∼ 0.2 at.% Cu after Step I), to get set to re-do their distribution in a subsequent step that takes them to the numerous Al 3 Sc precipitates that form during this Step II. Now nano-sized coherent Al 3 Sc precipitates are created at a high density from the Sc-enriched heterogeneities (Figure 2(d)). Figure 2(e-f) present a schematic and a TEM image, respectively, to show the Al 3 Sc precipitates at a high number density. Corresponding 3D-APT images were  and uniformly distribute in the matrix. A fraction of them will serve as potential preferred sites for nucleating copious θ -Al 2 Cu precipitates later on in Stage III, see Figure 2(l) and Figure S1.
We now point out that one cannot omit Step I and directly go to Step II. The necessity of Step I is because the lower-temperature AA pulls the Sc solutes out of its original uniform distribution in the matrix to become spreadout and populous precursors of Al 3 Sc in the matrix. This happens because during AA, at the large undercooling the driving force is high for the resultant Sc-enriched heterogeneities to be small but numerous. During Step II at 450°C, the majority of the Sc-enriched heterogeneities rapidly grow to nano-sized Al 3 Sc precipitates. In contrast, if one goes straightly to 450°C without Step I, the nucleation rate of Al 3 Sc is low due to smaller undercooling such that the particles are lower in number density but grow to fairly large sizes, jeopardizing the targeted copious heterogenous θ -Al 2 Cu precipitation in Step III (see next paragraph). This is why the Step I is a prerequisite for the ensuing steps.

Step III (Re-ageing)
We are then in the position to reform the θ -Al 2 Cu precipitates in Step III, a re-ageing treatment at 250°C. As seen in the sketch and TEM image in Figure 2(i-j), the re-precipitated θ -Al 2 Cu plates have a number density more than twice that of the AA treatment, because of the high density of (Sc-)heterogeneous nucleation sites sowed in Step II, see Figure 2(l) and Figure S1. In detail, a clear orientation relationship can be detected when the θ -Al 2 Cu heterogeneously nucleate on the Al 3 Sc [8,12,19]. Such nucleation assistance is efficient, whereas during AA there are only occasional clusters with neither clear crystal structure nor orientation  [9] or reinforced with 15 vol.%SiC [25] (pink diamonds and purple triangles, respectively). relationship (Figure 4(a)). The distributed θ -Al 2 Cu precipitates with a high density accomplish our goal (i) to uniformly spread out abundant precipitates for strengthening.
We next demonstrate the very important realization of goal (iii), i.e. quick and ample Sc segregation to intercept the coarsening of these θ -Al 2 Cu precipitates. The Sc segregation at α-Al/θ interfaces is remarkably enhanced after Step III. Figure 2(k) show a representative element maps cross a α-Al/θ interface with an average peak Sc concentration of 2.35 ± 0.71 at.%, which is about three times of that in the AA condition, see Figure 3(b and c). The interfacial energy as driving force for precipitate coarsening [22,23] is further reduced due to the greater Gibbsian interfacial excess [15,24], rel Sc , induced by enhanced interfacial Sc segregation (see SI for more discussions). In the meantime, kinetically this interfacial Sc aggregation is likely to slow down the Cu atom diffusion cross the interfaces, delaying the dissolution or growth of the precipitates. The overall result is a fine θ -Al 2 Cu precipitate size at ∼ 250 nm after Step III, see Figure 1(b), 25% smaller than after AA treatment. More importantly, counting together the interface Sc (12.8%) and Al 3 Sc particles (66.0%), overall ∼ 80% of the total Sc atoms in this alloy have been exploited to influence the strength and creep resistance of the alloy (see Table S3). During Step III, what makes the Sc enrichment at the α-Al/θ interfaces far more pronounced and rapid than in AA is the Sc supply from our intentionally distributed Sc-enriched heterogeneities transforming into Al 3 Sc particles that are now either in contact with the θ -Al 2 Cu precipitates or very close by. To begin with, the θ -Al 2 Cu precipitates initiate preferentially at some of these Sc-enriched locations, where the Sc is already available. When many θ -Al 2 Cu precipitates appear, the very small ones coarsen into a larger one, such as the examples in Figure S2. Previous reports [19] have demonstrated that an increased Cu content leads to the decreasing volume fraction of Al 3 Sc dispersoids. The Sc attached to many tiny precipitates migrate to and join the larger Sc patches at the α-Al/θ interface. As such, Al 3 Sc precipitate-depleted zones(PDZs) form both ahead and around the growing θ -Al 2 Cu precipitates, as marked in Figure 2(j). All these constitute fast supply of Sc atoms towards the matrix/precipitate interfaces, quickly enclosing the θ -Al 2 Cu precipitate and inhibiting its further coarsening. The eventual Sc segregated to the α-Al/θ interfaces amounts to ∼ 13% of the total Sc atoms in the alloy, almost quintupling that achieved after the AA treatment. This sizable fraction is due to the more but smaller θ -Al 2 Cu precipitates compared with the AA case, and the high interfacial Sc concentration (2.35 ± 0.71 at.%), which on average is three times that of the AA case.

Stabilized nanoprecipitates contribute to much reduced creep rates
As seen in Figure 1(b) and Table S2, our three-step treatment makes the θ -Al 2 Cu precipitates far more stable than conventional AA counterpart due to improved Sc availability and segregation. Estimates using thermodynamic calculations in SI suggest a ∼ 65% reduction in the interfacial energy due to interfacial Sc segregation, reducing precipitate coarsening [22,23] in the Al-Cu-Sc RR alloy, in comparison with the ∼ 24% in its AA counterpart. For plate-like θ -Al 2 Cu precipitates to grow, the interface-controlled thickening [23] fed by the flux of Cu from the matrix to the migrating ledge now necessarily requires the redistribution of Sc. During the creep test, the average size of θ -Al 2 Cu precipitates was in fact almost unchanged, remaining at about 250 nm at 300°C/30 MPa for ∼ 220 h, see Figures 1(b) and 5(a). TEM examinations also reveal that the half length of θ -Al 2 Cu actually decreased to ∼ 180 nm after creep test for 350 h, as a result of splitting in some of the θ -Al 2 Cu precipitates, with only a slight drop of volume fraction f v from 2.13 ± 0.19% to 1.97 ± 0.10%.
The secondary stage of the Al-Cu-Sc RR alloy now holds on for as long as ∼ 350 h, a 10-fold rise in endurance time than its AA counterpart, see Figure 5(a) at 30 MPa. Figure 5(b) summarizes the creep rate data in the tensile stress range of 30-60 MPa; the present Al-Cu-Sc RR alloy sustains a creep rate several orders of magnitude slower when compared with binary Al-Cu or Al-Sc alloys, and Al-Cu based alloys and composites. Additional discussions are presented in SI on precipitate stability, creep mechanism (Fig. S4) and hightemperature tensile properties. For the latter, regular tensile test of the Al-Cu-Sc RR alloy shows a yield strength of 127 ± 12 MPa at 300°C, which is much higher than that of the Al-Cu AA (59 ± 5 MPa) and Al-Sc AA (34 ± 2 MPa) alloys, and close to some commercial alloys (i.e. 7075, 2219 Al). This short-term strength of the present Al-Cu-Sc RR alloy is nevertheless lower than the ∼ 200 MPa yield strength at ∼ 300°C for Al-Cu alloys with high Cu content, such as Al-6.3Cu-0.5Mg-0.5Ag-0.5Mn-0.2Zr (their long-term microsctructure stability and creep rates at 300°C are not available for a direct comparison) [6,13].
We note that in addition to θ -Al 2 Cu (protected by Sc segregation), important contribution to strengthening also arises from the Al 3 Sc particles (see an estimate in the SI). Our RR protocol combines these two contributing factors and exploits both types of precipitates.

Conclusions
In summary, for Al-Cu based alloys we have modified the precipitation protocol to boost the thermal stability of strengthening precipitates at preferred sizes for creep resistance at 300°C under relatively high tensile stresses. Specifically, our deliberate integration of three heat-treatment steps, taking advantage of the interaction and synergy between the main alloying element Cu and minor alloying element Sc, generates dispersion of θ -Al 2 Cu precipitates with a combination of high density, uniform size distribution, and efficient use of Sc sources to quickly collect Sc to the α-Al/θ interfaces to stabilize the θ -Al 2 Cu precipitates at desirable sizes to delay runaway coarsening. The mechanistic insight gained through this work is that one can strategically exploit two different alloying elements to deliver a coexisting and coupled (θ -Al 2 Cu precipitates + Al 3 Sc particles) nanoprecipitate microstructure that performs better in terms of creep resistance. These unconventional microstructures give rise to order-of-magnitude reduction in creep rate, over previous Al-Cu based alloys. Our discovery extends the repertoire of cast Al-Cu alloys for potential use at service temperatures as high as ∼ 300°C.