Formability and failure mechanisms of AA2024 under hot forming conditions

L. Wang1, M. Strangwood2D. Balint1, J. Lin*1 and T. A. Dean3

1Department of Mechanical Engineering, Imperial College London

SW7 2AZ, UK

2School of Metallurgy and Materials, University of Birmingham, Edgbaston,

Birmingham B15 2TT, UK

3School of Mechanical Engineering, University of Birmingham, Edgbaston,

Birmingham B15 2TT, UK

Abstract

Aluminium alloy 2024 (AA2024) is extensively used as a structural material in the aircraft industry because ofits good combination of strength and fatigue resistance. However, complex shaped components, particularly those made from sheet, are extremely difficult to form by traditional cold forming due to its low ductility at room temperature. A possible solution of this problem is to form sheet workpieces at elevated temperature. The aim of the work described in this paper is to determine the relationship between formability and temperature for AA2024 by conducting a series of isothermal tensile tests at elevated temperatures ranging from 350 to 493 °C. Ductility of AA2024 was found to increase gradually with increasing temperature up to 450°C, followed by a sharp decreasewith further increase in temperature. So-called cup tests confirmed that the formability of AA2024 is very high at a temperature of about 450 °C.Fracture surfaces and longitudinal sections of formed samples were examined by scanning electron microscope. It was found thatfracture occurred in three different modes depending upon the temperature, and the sharp decrease in ductility when the temperature exceeds 450°C was caused by softening of grain boundaries by solute enrichment (at higher heating rates liquation may be involved) and softening of the matrix around inclusion particles.

Keywords: AA2024; Fracture mechanisms, Hot forming, Formability

* Corresponding author. Tel.: +44 020 7594 7082; fax: +44 020 7594 7017;

Email address: (J. Lin).

  1. Introduction

Aluminium alloy 2024 (AA2024) is widely used for structural applications in the aerospace industry, due to its good combination of strength and fatigue resistance. Age hardening markedly affects the microstructure and mechanical behaviour of AA2024. After appropriate heat treatment, finely dispersed precipitates are obtained and a high strength to weight ratio is achieved. Low ductility, however, is typically a limitation of AA2024,making complex-shaped parts difficult to form using traditional cold forming processes.Hence some thin-walled and complex-shaped partsaremachinedfrom solid blocks of metal, which can incur up to 90% material wastage for some applications, with corresponding cost and energy wastages.

In recent years, efforts have been made to improve the strengthand ductility of AA2024 [1-9]by developing advanced processing routes.For example, following severe plastic deformation (i.e.ECAP [1, 4, 5], cryo-rolling [3] or high-pressure torsion processes[9]) and suitable ageing, highly dispersed nanometre-sized precipitates can be obtained. Due to the low density of dislocationspresent after ageing, and the strong pinning effect and aggregation of precipitates to dislocations, an elongationof 18% can be achieved at room temperature[3]. However, the use of extreme strain levels is limited to very small samples, and is currently not a suitable alternative for bulk materials processing. On the other hand, pre-homogenisation treatment [7, 8] might be an effective way of improving the ductility of AA2024, which, however, sacrifices the material strength. On the contrary, the material strength could be enhanced by adding SiCp into AA2024, but the ductility of the resulting composite has been found to decrease [6].

As for most metallic alloys, the ductility of AA2024 increases with increasing deformation temperature[6]. However, hot stamping is seldom used for forming complex-shaped parts because the desired microstructure (which has been fixed in the sheet prior to forming) tends to be destroyed at high temperatures, thereby reducing the mechanical performance.Furthermore, if heat treatment to restore strength is carried out after stamping, the parts tend to lose their shape due to thermal distortion.

Recently, a new technique,Hot Forming and cold-die Quenching (HFQ),has been developed [10]. The basic idea of this novel process is to (i) heat the sheet metal to its solution heat treatment (SHT) temperature, at whichits ductility is expected to be maximal, and (ii) simultaneously form and quench the sheet by using cold dies. After forming, the workpiece material is held within the dies for roughly 5 to 6 seconds in order to reduce its temperature rapidly to approximately 100 °C and freeze the microstructure as a supersaturated solid solution (SSSS). If a heat-treatable, aluminium-based alloy is used, the part can then be aged to obtain full strength. The feasibility of this novel process has been demonstrated for AA6XXX in [11]. This technique has gained favour over traditional forming methods because it producesparts with high formability, negligible springback, rapid processing and efficacious mechanical properties.

In AA2024, copper is the most effective strengthening constituent. The age hardening effect can be increased by increasing copper content up to 6 wt % [12]. Magnesium is used in combination with copper to accelerate the age hardening effect at room temperature. The equilibrium precipitate phases for this system are mainly CuMgAl2 (S phase) and CuAl2 (θ phase) [13], although CuAl2 is less often observed in the alloy compared to CuMgAl2. Both of these phases are largely soluble during SHT. In addition, manganese has a marked strengthening effect, because it can influence material properties through the formation of intermetallics that provide Zener drag and limit grain size. The intermetallic compounds (IMCs) in AA2024 are very complicated [14-16], and the shape, size and chemicalcomposition of the IMCsvary remarkably, which are dictated by the processing route [16]. A range of IMC compositions have been reported for AA2024, such as Al20(Cu,Fe,Mn)5Si(Al8Fe2Si), periphery, Al10(Cu,Mg), (Al,Cu)93(Fe,Mn)5(Mg,Si)2, Al3(Cu,Fe,Mn), Al2Cu (θ phase), CuMgAl2 (S phase) and Al7Cu2Fe [15]. These are normally formed during the casting or homogenization stages of processing, i.e. high temperature stages; these are normally not soluble during SHT.

A number of studies have been carried out on the fracture mechanisms of AA2024 at room temperature [17-20]and under superplastic or creep age forming conditions [8, 9, 19, 21],i.e.high temperature and low strain rate conditions. However, little has been published on the fracture mechanisms of AA2024 under hot forming conditions, i.e. high temperature and high strain rate conditions. The objective of the present study is to address the above issue; to determine the conditions for which maximum ductility arises, and to investigate the flow behaviour and fracture mechanisms of AA2024 when subjected to HFQ.

  1. Experimental procedure
  2. Materials

The composition of the AA2024[1] used in the current research is listed in Table 1. The 2 mm thick AA2024 sheets were in a T3 condition (solution heat treated, quenched and stress-relieved by cold stretching).

Table 1. Chemical composition of AA2024 (wt %) [12].

Si / Fe / Cu / Mn / Mg / Zn / Al
0.5 / 0.5 / 3.8-4.9 / 0.3-0.9 / 1.2-1.8 / 0.25 / balance

2.2.Testingprogramme

Gleeblethermomechanicaltesting

Isothermal tensile tests were conducted on a Gleeble 3800 thermomechanical simulator, which can heat a specimen by direct resistance heating at a rate as high as 10,000 °C/s.On the other hand, the specimen was held by two high thermal conductivity grips clamped by two jaws, each with an embedded water-cooling system. This makes the Gleeble 3800 also capable of high cooling rates. In addition, a pair of thermocouples was welded on the specimen to provide signals for the accurate feedback control of the specimen temperatures. The combined effect of the efficient heating and cooling system and the accurate thermal control system ensures that the temperature of the specimen can be controlledaccurately.

Due tothe possibility of overheating, specimens were first heated to a temperature 25°C lower than the target temperature at a heating rate of 50°C/s, then were further heated to the target temperature at a rate of 5°C/s. Isothermal tensile tests to failure were performed as soon as the target temperature was reached, at a constant strain rate of 1s-1, representative of typical strain rates in hot forming processes. Table 2 lists the isothermal tensile tests that were conducted in the present study. Additional tests were conducted at 350 and 450°C to verify the repeatability of the tests; good agreement from one test to another (mean deviation less than 5%) of both ductility and flow stress was achieved.

Table 2. Isothermal Gleeble 3800 tensile tests (bold indicates select temperatures where repeatability of the tests was verified).

Temperature (°C)
Strain Rate
1 s-1 / 350 / 360 / 370 / 380 / 390 / 400 / 410 / 420
430 / 440 / 450 / 460 / 470 / 480 / 487 / 493

Formability testing

Formability tests (so-called cup tests) were performed on a 25 tonne ESH high-speed (ram speeds up to 5 m/s) press. The formability test rig was designed to be a portable, integrated structure onto which either a spherical head or flat head punch could be mounted. A high speed camera and prism were employed during the tests to record the diameter evolution of the central hole. Fig. 1 shows the formability test rig in place on the ESH high-speed press.In Table 3, the dimensions of the test piece and punches, and main process parameters for the forming tests are listed. A central hole, 16mm in diameter, was cut in the centre of the test pieces, for the qualitative evaluation of plastic deformation and verification of modelling results.

Microstructure examination

Precipitate and inclusion populations have a strong effect on the ductility of AA2024. A Hitachi S3400-N scanning electron microscope (SEM) equipped with a Gatan H1002 heating stage(750 °C peak temperature) was used to examine the microstructure of AA2024; thearea close to the fracture surface was examined in order to reveal the fracture mechanisms at elevated temperatures. The operating voltage was 15 kV. Longitudinal sections, as well as the fracture surfaces, of the samples deformed at 350, 450 and 493 °C were examined. Precipitate evolution and inclusion distribution were examined at a temperature of 493°C using a heating stage installed in the SEM for in-situ observations.All the samples were polished prior to SEM examination.

Table 3. Dimensions of the testpiece and HFQ punches, and the main process parameters.

Testpiece width×length×thickness(mm) / 170×170×2
Central hole diameter (mm) / 16
Hemispherical head punch (HH) diameter (mm) / 80
Flat head punch (FH) diameter (mm) / 80
Punch temperature (°C) / 20 (room temperature)
Ram speed (mm/s) / 170 / 486 / 486
HH Punch displacement (mm) / 26±3 / 26±3 / 36±5
FH Punch displacement (mm) / 21±2 / 21±2 / -
Initial testpiece temperature (°C) / 493±5 / 500±5
Soaking time (hours) / 0 / 1
Forming temperature (°C) / 450±10 / 493±10
  1. Results
  2. Dependence of ductility on temperature

Asthe alloyishot formed in the HFQ process, the evolution of the tensile ductility at elevated temperatures, especially around the SHT temperature, is of particular importance. In the present research, a series of tensile tests, shown in Table 2, were performed on a Gleeble 3800 thermomechanical simulator at temperatures ranging from 350 to 493°C. Fig. 2 shows the results obtained from the tensile tests. It was found that the maximum failure strain of AA2024 is in excess of 1.1 (equivalent to an elongation of 200%), occurring at 450°C, hence suggesting a temperature window for high formability of AA2024. This behaviour is consistent with increased thermal activation leading to greater atomic mobility, dislocation motion and recovery. However, at 493°C, a typical SHT temperature for AA2024, the ductility was very poor, only about one-tenth of the peak value. Such a steep decrease in ductility is quite different from other heat treatable aluminium alloys, such as AA6XXX[22]. Usually, the ductility of aluminium alloys at the SHT temperature should be improved, as the amount of precipitates in the matrix is significantly decreased relative to lower temperatures, hence the chance of void nucleation should be significantly diminished [17] which would improve the ductilityof the material. In the Gleeble tests, melting of precipitates and inclusions (see section 3.4) appears to cause the drop in ductility at temperatures in excess of 450°C. Fig. 3 shows the stress versus strain relations of AA2024 at different temperatures. Significant softening due to increasing temperature can be observed, and the flow stress decreases with rising temperature from approximately 200 MPa at 350°C to about 50 MPa at 493 °C.

3.2.Dependence of formability on temperature

Temperature played adominant role in the formability tests of AA2024,as it did in the ductility tests. Fig. 4 shows the effect of temperature on formability. Formability at 450°C was high (see Fig. 4a), as expected from the high ductility observed at that temperature in tensile testing.Significantthinningwas observed, especially in the central hole area (Table 4), indicating that severe plastic deformation occurred during the tests.As the forming temperature approached the SHT temperature (493 °C), the AA2024 samples showed extremely poor formability and exhibited cracking with low levels of ductility (Fig. 4b). Thickness measurements showed little or no localised plastic deformation or thinning (average thickness is 1.95 mm), suggesting that cracking occurred early in the forming operation. The lack of plasticity associated with cracking is consistent with the low ductility noted for tensile tests at these temperatures.

Table 4.Wall thickness distributions for the samples shown in Figs. 5aI and 5bI.

Distance from the centre (mm) / 10 / 12 / 14 / 18 / 20 / 24 / 30 / 34 / 35 / 38 / 40 / 42
Wall thickness (mm) of sample Fig. 5aI / 1.67 / 1.84 / 1.83 / 1.88 / 1.8 / 1.77 / 1.28 / 1.24 / 1.2 / 1.44 / 1.5 / 1.71
Distance from the centre (mm) / 16 / 18.5 / 20.6 / 26.6 / 32.5 / 36 / 40 / 42
Wall thickness (mm) of sample Fig. 5bI or 4a / 1.32 / 1.37 / 1.44 / 1.53 / 1.7 / 1.83 / 1.86 / 1.9

3.3.Dependence of formability on forming velocity

Fig. 5 shows the effect of deformation speed on formability. For the hemispherical head punch, the material flow was influenced significantly by forming velocity. For low punch speeds, localised plastic deformation occurred circumferentially,loweron the test piece relative to the hole, with a mild necking zone visible on the side-wall(see Fig. 5aI). On the other hand, at high speeds, a considerable increase in the diameter of the central hole was observed(Fig. 5bI), i.e. plastic deformation predominatedin the vicinity of the central hole, rather than around the circumference, which indicates a change in the location of localised plastic deformation during the high speed HFQ tests.

Figs. 5aII and 5bII show the deformation characteristics using a flat head punch instead of a hemispherical punch. In both the low and high speed tests, an enlarged central hole can be observed, suggesting that plastic deformation took place primarily inthevicinity of the central hole in both tests. Again, the high-speed test exhibited a larger final hole diameter (D=28.3mm; Fig. 5bII) than the low speed test(D=27.2mm; Fig. 5aII), indicating that more plastic deformation took place around the central hole area under high speed forming. The implication is that an optimum speed exists between the low and high speed limits such that plastic deformation is most evenly distributed, and localises minimally as a neck or a concentration around open features, both of which could ultimately lead to ductile failure during forming.

3.4.Damage mechanisms

Fig. 6aI shows the fracture surface of the tensile sample deformed at 350°C. It has a dimpled cup-and-cone appearance, indicating that microvoid nucleation, growth and coalescence was the dominant fracture mechanism, which is similar to the fracture mechanism of AA2024 at room temperature [17, 19, 20]. Fig. 6aII indicates the region adjacent to the fracture area on a longitudinal section of the deformed sample showing the path of the fracture and intermetallic particles (white). This section shows that secondary crack branching from the primary fracture surface to nearby large inclusions occurred, but inclusions further away from the primary fracture surface did not show evidence of decohering or voiding. This suggests that the primary crack nucleated perpendicular to the loading direction and propagated across the section without initiation of further cracks from inclusions or other sites occurring readily.

Fig. 6bIshows the fracture surface of the sample deformed at 450°C. The features of this fracture surface are much different than those of the sample deformed at 350 C, which appears as fine ductile features superimposed on a background of elongated surface features and resembles an intergranular type of failure. Heating the sample to 450 C will have resulted in an initial tendency for the original T3 condition to age with ”, ’ and S’ forming prior to the equilibrium  and S (these stages will also have taken place on heating to 350 C). Rapid heating will restrict the formation of strengthening precipitates until temperatures when dissolution again becomes favourable. This results in most of the Cu and Mg being in solid solution during deformation at 450 C. During elevated temperature deformation, thermally activated diffusion of alloying elements will be taking place, enhanced by the fast diffusion paths offered by the excess dislocations present from the initial T3 treatment and those introduced during deformation. Segregation of alloying element atoms to defect sites such as dislocations and, more importantly, grain boundaries would be expected through this treatment although high-resolution transmission electron microscopy (TEM) would be needed to confirm this behaviour. Any precipitates formed would be too fine for SEM resolution and so the bright particles observed in Figs. 6aII and 6bII are from Mn-, Fe- and Si-rich intermetallic phases formed largely in the melt. These will be present with a range of compositions, sizes and, hence, thermal stabilities, but would generally be hard and undeformable within the aluminium matrix. As seen from a comparison of Figs 6aII and 6bII, there is an apparent decrease in the volume fraction of inclusions from 6aII to 6bII, but, given the inhomogeneous distribution of inclusion particles, this falls within the range of anticipated scatter in inclusion volume fraction. Likewise, there is no major difference in the range of sizes and aspect ratios between 350 and 450 C, indicating neither dissolution nor plastic deformation of these particles took place. The continued non-deformable nature of the inclusion particles would explain the greater tendency to form voids (associated with large particles or clumps of particles) further from the main crack (Fig. 6bII). The greater thermal softening of the matrix at 450 C compared with 350 C would result in greater strain concentration in the matrix around the inclusion particles and so localised necking would occur, giving rise to ductile voids. With a greater difference in strength between the particle and the matrix, smaller particles would give rise to voiding resulting in the observed larger number of smaller ductile dimples seen in Fig. 6bI compared with Fig. 6aI. The presence of the inclusions explains the fine-scale ductile voiding, but does not explain the background surface features in Fig. 6bI. However, a secondary crack (marked ‘X’) is seen in Fig. 6bII, which is not associated with voiding around particles and follows a curved boundary-like path. This suggests a role for grain boundary failure processes at elevated temperatures.

Fig. 6cI shows the fracture surface at 493°C, which is characterised by the absence of ductile dimples and an entirely intergranular fracture path. Although increasing temperature would facilitate voiding, changes at the matrix grain boundaries result in a much lower resistance to cracking at these sites. Hence, at this temperature extensive secondary cracking occurs below the main crack (see Fig. 6cII). Fig. 6cII appears to show the absence of inclusions, but this is just a spatial variation; although some of the less stable Fe- and Si-bearing inclusions would be expected to dissolve (limited due to the low time of exposure to elevated temperatures), the majority of Mn-rich inclusions do not start to dissolve below 500 C. The behaviour exhibited in the fractographic study necessitated a more detailed, in-situ SEM examination of the behaviour of inclusions and grain boundaries.