1. Introduction
Al–Mg–Si alloys (6xxx series) are characterized by medium to high strength, high fracture toughness and good resistance against corrosion and stress corrosion cracking. They also feature good weldability [
1,
2]. For these reasons, Al–Mg–Si alloys are widely used for components in various applications. Due to their good recyclability and the excellent formability of sheet materials in the T4 condition, they are particularly attractive for the automotive industry [
3,
4]. Modern automotive sheet alloys are designed to harden during the so-called paint bake cycle (PBC), where the complete car body is heated to elevated temperatures for a limited period. Both a pronounced increase in strength during the PBC and a reduced tendency for natural aging (NA) prior to the PBC must be ensured. Recently, an attempt was made to reduce the paint bake temperature in order to realize polymer-based components and save energy.
Diffusion processes and the resulting cluster formation directly after quenching are known to cause natural aging in Al–Mg–Si alloys. This produces an undesirable increase in strength, a decrease in formability, and a reduction in age hardening potential [
5,
6,
7]. To counteract these effects, a pre-aging heat treatment procedure is usually performed. This is typically carried out immediately after quenching at temperatures between 60 and 200 °C [
8]. According to Zhen and Kang [
9], pre-aging is assumed to suppress negative cluster formation during natural aging, but generates a high number density of β′′ phase nuclei, stimulating the subsequent paint bake response. An alternative method for suppressing the natural aging effect was recently presented by Pogatscher et al. [
10,
11,
12]. Their studies showed that adding Sn inhibits cluster formation and improves subsequent artificial aging. This effect is attributed to the trapping of vacancies by Sn atoms at low temperatures. At elevated temperatures, i.e., during the PBC, Sn releases the vacancies and fast non-equilibrium diffusion is enabled. This desirable Sn addition effect, called “diffusion on demand” [
10], has been confirmed by other authors [
13,
14,
15,
16].
While it is desirable to retard the kinetics of cluster formation at room temperature, the reverse is true for artificial aging. For paint bake hardening, using Zn additions in Al–Mg–Si alloys has proven to be an effective approach [
17,
18,
19,
20]. Saito et al. [
21] showed that hardness increases after ageing at 185 °C if 1 wt % Zn is added to an Al–Mg–Si alloy. They detected an increased number density of needle-shaped precipitates from the Al–Mg–Si system, but no phases containing zinc. Guo et al. [
18] carried out DSC measurements on Al–Mg–Si alloys with varying Zn content. They observed a shift of the precipitation peak at 250 °C, known as the β′′-peak, to lower temperatures when 3 wt % Zn was added. They also reported that the type of precipitate is not affected by Zn addition, i.e., no Mg–Zn precipitates were observed [
18].
In this study, the above approaches of adding Sn to retard natural aging and alloying Zn to promote aging in the paint PBC were combined in various experimental alloys to investigate hardening behavior and cluster formation during pre-aging and PBC. Four different experimental alloys were investigated: (1) standard Al–Mg–Si as reference; (2) Al–Mg–Si with Zn addition; (3) Al–Mg–Si with Sn addition; and (4) Al–Mg–Si with both Zn and Sn addition. Via hardness and atom probe tomography (APT) measurements, we characterized and evaluated the influence of the alloying elements individually and in combination.
2. Materials and Methods
The alloys were produced by gravity casting at AMAG Rolling GmbH, Ranshofen, Austria. The Al–Mg–Si base alloy with different additions of other elements were melted in an induction furnace and then cast into preheated gravity dies to obtain ingots with a size of 170 × 80 × 40 mm3. Subsequently, the ingots were homogenized in a two-step procedure at temperatures of 540 and 565 °C. Afterwards they were first hot- and then cold-rolled to 1.5 mm thick sheets. The solution heat treatment took place in an air furnace at 560 °C for 15 min, and was followed by water quenching. Subsequently, the sheets were pre-aged at a temperature of 110 °C for 5 h (denoted as the “T4P” condition). Afterwards, 2% of pre-straining was applied. Half of the samples were tested in this condition and the other half was subjected to a modified paint bake hardening treatment/cycle (PBC) at a reduced temperature of 165 °C for 20 min (“T6P” condition). The alloy with Zn and Sn additions was also overaged to the “T7” condition at 170 °C for 120 h.
The nominal compositions of the experimental alloys are listed in
Table 1. It should be noted that all alloys exhibit the same Mg and Si levels and differ only in their Zn and Sn content.
To analyze the hardening behavior during pre-aging and PBC, Brinell hardness measurements were carried out on a Diatestor 2 RC (Wolpert, Aachen, Germany). A ball diameter of 2.5 mm, a load of 306 N (31.25 kp) and a dwell time of 10 s were applied according to DIN EN ISO 6506-1. Tensile tests were also performed on an Inspekt 250 (Hegewald & Peschke, Nossen, Germany) machine according to EN ISO 6892-1 to analyze mechanical properties.
To analyze the clustering behavior at the atomic scale, APT measurements were conducted. For this purpose, small 15 × 15 mm² samples were ground to a thickness of 0.5 mm. Then, small rods of 15 × 15 × 0.5 mm
3 were cut from these sheets and electropolished using a standard two-step method to form a small tip. First, we performed “rough polishing” with an electrolyte consisting of 25% perchloric in 75% acetic acid. Then, we carried out “fine polishing” using a solution of 2% perchloric acid in butanol. The measurements were conducted on a LEAPTM 4000X HR (Cameca, Madison, WI, USA) in voltage mode at a temperature of 40 K. This system has a detection efficiency of ~37%, meaning only 37% of the atoms arriving at the detector are amplified. We used a pulse fraction of 20% and a detection rate of 1%. Data reconstruction was carried out using IVASTM 3.6.8 software from Cameca. To minimize influences from crystallographic artefacts, we limited our analysis to cubes with an identical volume of 30 × 30 × 100 nm
3 which lay away from major crystallographic poles. The clustering of the atoms of interest was determined by an algorithm based on Voronoi tessellation of the solute atoms and Delaunay triangulation to test the random distribution of the solid solution (as described elsewhere [
22,
23,
24]). The data were analyzed using custom scripts in MATLAB R2016a from MathWorks. 3D atom maps, cluster size, Mg/Mg + Si distribution, and proxigrams were used to visualize the results of the APT in this study. The 3D atom maps show a 3D picture of the most important atoms: Si, Mg, Zn, and Sn. The Mg/Mg + Si distribution is deployed to display detailed information about the Mg and Si ratio within the clusters. The proxigrams illustrate the element content in the cluster, starting in the center of a cluster. For this analysis, the data of all identical cluster types are averaged, and the clusters are regarded as spheres. Of course, this procedure does not account for the exact geometry of the clusters, but it reflects for the differences between the various conditions and alloys. The mean cluster size of the Mg–Si co-clusters indicates how many Mg and Si atoms participate in the local elemental enrichment.
It is important to note that for reasons of simplification and ease of readability, all detected aggregates were referred to as clusters according to the cluster detection algorithm used, even if different types of precursor of the β phase might be present.
4. Discussion
The presented results provide a fairly clear picture of how the modifying elements Sn, Zn, and Sn + Zn influence the early stages of precipitate formation and, thus, the mechanical properties as represented by hardness and yield strength values of Al–Mg–Si alloys. A summary of the most relevant results of the ATP cluster analysis is presented in
Figure 7. For both the T4P and T6P conditions,
Figure 7a shows the mean cluster size of the Mg–Si co-clusters, and
Figure 7b displays the number density of clusters in these states. Sn obviously retards cluster growth in condition T4P but accelerates it during PBC. On the other hand, the addition of Zn generates an increased number density of clusters independent of the presence of Sn.
Before discussing the elemental influences on cluster size and cluster number density, we will take a closer look at the relationship between cluster formation and yield strength. Hardening is controlled by interactions between dislocations and clusters. An appropriate approach to describing the flow stress increase by the presence of clusters delivers the “dispersed barrier model” [
27,
28],
where
M is the Taylor factor,
µ the shear modulus,
b the Burgers vector,
d and
N the cluster obstacle size and cluster number density, and α the cluster barrier strength. We now take the liberty of simplification and set the mean cluster size determined by APT as obstacle size d. For the various alloys in conditions T4P and T6P, we calculate the
values listed in
Table 6 and illustrate the correlation to the yield strength values in
Figure 8.
In
Figure 8, two special features are easy to recognize. Firstly, the yield stress values of both T4P and T6P states correlate quite well with
, which supports the applicability of the hardening model. Secondly, the T6P data (black) lie predominantly above the T4P data (grey), indicating a greater hardening effect on the part of the T6P clusters, i.e., the cluster barrier strength α in Equation (1) is significantly higher for T6P clusters than for T4P clusters. The greater barrier strength of the T6P clusters is not surprising because these clusters were transformed into a more mature state by the PBC treatment.
We now discuss the influence of Sn, Zn, and Sn + Zn on the cluster parameters.
Alloy Sn shows a reduced number density of Mg–Si co-clusters and a smaller mean cluster size compared to the Reference. We attribute these effects to the vacancy binding effect of Sn, as often described in the literature [
10,
11,
12,
29]. At the T4P temperature, substitutional diffusion is inhibited. At the PBC temperature of 165 °C, however, Sn releases the trapped vacancies, generating rapid cluster growth.
With the addition of Zn in
Alloy Zn, there is an increase in both mean cluster size and especially number density. Zinc does not form clusters by itself, but stimulates the nucleation and growth of Mg–Si co-clusters. These findings agree well with those in the literature [
17,
30,
31,
32]. Zn has an attractive vacancy binding energy [
33,
34]) and it can thus be assumed that when Zn atoms are present, more vacancies survive the quenching procedure. In contrast to Sn addition, however, the critical temperature for releasing the trapped vacancies is already exceeded during the T4P treatment at 110 °C, and rapid nucleation and growth can take place. Only a slight enrichment of Zn in the clusters was detected, which might be related to the Mg–Zn formation enthalpy (−6.1 kJ/mol), calculated by Wolverton [
35]. Guo et al. [
31] postulate that the Mg atoms can diffuse more easily to the clusters when Zn is present, perhaps explaining not only the faster nucleation and growth of the clusters but also the increased concentration of Mg atoms within the clusters (mean Mg/Si ratio ≈ 1.7). In accordance with our findings, Yuan et al. [
30] also detected no Mg–Zn phases in Al–Mg–Si–Cu–Zn–Mn. They also observed an enhanced nucleation process and a higher number density of β′′ precipitates. Ding et al. [
17] measured zinc containing phases after excessive artificial aging (170 °C for 120 h or 200 °C for 24 h). In our study, however, no η-phase or their precursors were detected, not even in overaged T7 condition after long-term artificial aging (
Figure 6).
Alloy Zn + Sn shows a synergetic effect of Zn and Sn. The obtained number density and mean cluster size of the Mg–Si co-clusters are reduced compared with those of Alloy Zn in T4P. Here, too, adding Sn also retards the formation of clusters. However, the mean cluster size and the number density have higher values than those of Alloy Sn due to the abovementioned contribution of Zn to the nucleation and growth of clusters.