In Situ Investigations on the Ductility of an Al-Si-Mg Casting Alloy

Abstract

This paper reports the results of an investigation into the damage mechanisms that occur during straining of the aluminum alloy A356-T6 using in situ tensile experiments in a scanning electron microscope. Quantitative stereological analyses of the microstructure and in situ studies of the damage mechanisms have been used to characterize the Si particles and their eutectic distribution that ultimately controls ductility. In situ testing has revealed that, at increasing applied strain, the damage begins with brittle fractures of the Si particles due to cleavage. In this way, adjacent microcracks are created. Subsequently, the joining of the Si microcracks gives rise to the nucleation of a microcrack on the dendrite boundaries. The final damage stage involves the growth of the crack along the dendrite boundaries, until instability is reached. The investigations have shown that, by means of in situ mechanical testing, it is possible to obtain quantitative data of the damage mechanisms which are useful to predict the engineering characteristics of casting alloys.

Introduction

It is generally accepted that ductility in hypoeutectic aluminum-silicon alloys is governed by a sequence of damage events, that is, the initiation of microcracks in the silicon particles, the growth of the cracks into cavities, and finally, coalescence of the cavities due to the collapse of the resisting matrix ligaments.1–3 The first stage is usually considered critical, since the silicon particle size and morphology have the most important effect on the embrittlement of the metal.4–8 Eutectic modification and a solution heat treatment, which lead to an improvement in the size, shape, and distribution of the silicon particles, are used to avoid brittleness.

The particles in modified microstructures assume a more or less globular shape. The dendrite size, instead, mainly depends on the solidification rate, and it cannot be modified by treatments. The size and morphologic parameters of the particles vary in a relatively narrow range as a result of modification and heat treatments,9 whereas the opposite is true for the dendrite size during solidification. However, there is no general consensus on the microstructural features that actually control the tensile properties of the material, especially the ductility. Some researchers4,10 claim that the dendrite size is the main factor of influence, whereas others3,8 postulate the combined effect of silicon particles and dendrite cell sizes.

The question of the development of damage during straining and its associated microstructural parameter is not straightforward. One confusing aspect of the interpretation of damage mechanisms arises from the fact that it has been studied indirectly from tensile test results and post mortem metallographic observations. However, it is known that the preparation of metallographic samples from broken tensile specimens may result in overestimation or underestimation of the cracked Si particle percentage, as well as in a blurred picture of what happens after particle cracking. In situ SEM tensile tests10–12 have provided a clearer picture of some details of the damage process, but at the moment, except for the case where second-phase particles are evenly distributed,13,14 it is still far from complete. New in situ tensile tests are therefore necessary, in order to obtain more detailed knowledge of the damage mechanisms and to formulate quantitative descriptions of the phenomena.

This paper investigates the damage mechanisms through in situ tensile tests, using the hypoeutectic aluminum A356 alloy as the reference material, since it is one of the most widespread light alloys in automotive applications, for example, in engine support. The experimental results are used to approximately model the role of the microstructure in the ductility.

Experimental Procedures

This investigation was carried out on an A356-T6 alloy modified with Na to improve ductility through eutectic Si morphology alterations, eliminating embrittling needle-shaped particles. The chemical composition is reported in Table I where the Fe content was kept as low as possible, since it induces brittleness. Sand and chill castings were used to obtain different cooling and solidification rates. During casting, the chosen solidification times were of the order of 130 s in sand and 20 s in chill. After casting, the alloy was subjected to a T6 heat treatment. The sequences of the treatment are as follows: solubilization at 540°C for 6 h, quenching in water, natural ageing at room temperature for 24 h, and artificial ageing at 160°C for 6 h. The T6 heat treatment improves the strength, due to a precipitation of the metastable β′ particles in the matrix. Moreover, solutionizing contributes to the globulization of the Si particles, which further increases ductility.

Table I Chemical composition (mass%) of the A356 alloy casting utilized in this study

The microstructure was initially characterized by means of quantitative optical metallography. The silicon particle size, aspect ratio, and interparticle spacing were determined by means of the Quantimet image analyzer software, and were later verified in the SEM during in situ tests. For reasons that will be clarified later on, the interparticle spacing is the Si particle-to-particle distance along the boundary of the matrix dendrites. The primary phase cell size, i.e., the size of the dendrite arms, was also determined on metallographic samples. It should be recalled that the microstructure analyzed by means of metallography is a 2D approximation. However, the results of this kind of analysis are generally considered meaningful, and this makes in situ mechanical testing a powerful means of investigating the mechanical behavior of materials.

Traditional quasistatic tensile tests were performed at strain rate of 8 × 10⁻⁴ s⁻¹ at room temperature on sand and chill cast samples. The results were also used to determine a constitutive law in the plastic domain; the Hollomon equation was chosen. The decrease in stiffness was monitored during the tests to obtain more information. Mechanical damage gives rise to a decrease in the apparent Young’s modulus E _app compared with the nominal one E ₀. The evolution of the damage parameter D = (1 − E ₀/E _app) was calculated during straining.

In situ tensile tests were carried out in the chamber of an SEM (Philips XL30 DX4i) on previously polished mini samples. The tests were carried out at strain rate of 8 × 10⁻⁴ s⁻¹, adopting a step-by-step strain increment procedure. After each 1% plastic strain increase, the test was suspended and a set of images was recorded. Subsequently, the test was resumed and another 1% strain increment was introduced. The damage characteristics were therefore analyzed at each loading step.

Results

The results of the metallography are reported in Table II. The difference between the sand and chill casting is not so pronounced as far as the average equivalent diameter D _eq, shape factor α, shape-size parameter αD _eq and interparticle spacing, s, are concerned. The real difference between the two castings is in the cell size L. It plays the role of controlling parameter in the damage sequence.

Table II Microstructural parameters of the A356 castings

The tension testing results are reported in Table III. The two castings have quite similar strength. The elongation A and the hardening exponent n are higher in the chill casting. It should be noticed that the stress–strain curves did not display necking before rupture. The damage parameter at breaking D _c is also reported in Table III. It is interesting to notice that sand and chill casting show very similar losses in stiffness at failure.

Table III Mechanical characteristics of the A356 castings

The damage evolution during straining in the SEM has the following salient characteristics:

• 1. The damage mainly develops through Si particle cracking or interface decohesion; the progress in damage with plastic strain consists of an increase in the Si cracked particle number (Figs. 1, 2a–d).

  Fig. 1

  

  Microstructure of a chill cast sample before plastic straining. The dark-gray particles are Si particles; the white ones are FeSiAl₅

  Fig. 2

  

  (a) Early-stage damage due to Si particle cracking and debonding in a chill cast sample with 5% plastic strain. Picture of the same area as Fig. 1. Faint dislocation bands are visible. (b) Damage progress due to Si particle cracking and debonding in the same area as Fig. 1. The dislocation bands are more evident than in (a). Chill cast sample with 8% plastic strain. (c) Damage progress in the same area as Fig. 1. Chill cast sample with 10% plastic strain. (d) Damage in the same area as Fig. 1 with 12% plastic strain. The image was obtained in the SEM using back-scattered electrons to minimize the blurring effect due to surface wrinkling, typical with high local strains

• 2. The stress in the particles and their breaking is mostly built up through an accumulation of dislocations at the boundaries of the dendrites (Fig. 3).

  Fig. 3

  

  Dislocation slip bands passing through the dendrites and accumulating on the Si particles at the dendrite boundaries. Si particle cracking and interfacial decohesion can be noted in this sand cast sample with 5% plastic strain

• 3. The dislocation bands give rise to stress concentrations and Si particle cracking (Fig. 4).

  Fig. 4

  

  Si particle cracking due to impingement of the dislocation slip bands in sand cast sample after 6% plastic strain

• 4. Si particle cracking causes an accumulation of damage, which is more pronounced at the dendrite boundaries (Fig. 5).

  Fig. 5

  

  Damage occurring at the dendrite boundaries (electron back-scattering image). Post mortem observation in a zone far from the global collapse region in chill cast sample with more than 11% plastic strain

• 5. The microcracks in the Si particles, having a size on the order of D _eq (Table II), coalesce at some of the dendrite boundaries and thus produce an irregular array of microcracks. These microcracks have the same size as the order of the cell size L (Fig. 6).

  Fig. 6

  

  Microcrack nucleated at a cell boundary. The direction of the tensile stress is vertical in this chill cast sample with 8% plastic strain

• 6. The joining of several microcracks at the cell boundaries creates a macrocrack, and this ultimately leads to failure (Fig. 7).

  Fig. 7

  

  Coalescence of microcracks before global breaking. Post mortem observation in a zone far from the global collapse region in a sand cast sample having more than 8% plastic strain

Discussion

The experimental data allow one to establish the microstructural parameters that are relevant for the ductility of the material. On the basis of the metallographic analyses, it is evident that the shape-size parameter αD _eq and the cell size L are both involved in determining the damage of the material. According to Caceres et al.,3,15 the stress buildup in a particle can be estimated by applying dispersion hardening theory. At this stage, the size scale is on the order of D _eq, i.e., some micrometers. An increase in strain leads to Si particle breaking, which is consistent with the in situ results. From the same results (Figs. 4, 5, 8), it is easy to infer that the different aspect ratios of the particles induce a difference in particle stress between the two casting methods. However, the value of α is similar in sand and chill casting (Table II), whereas the difference in cell size is considerable. Therefore, there is a tendency for damage to occur in sand casting. In fact, Caceres et al.3 found this result experimentally, and it was later confirmed by Doglione et al.10

Fig. 8

Opening of fracture faces due to further straining after microcrack nucleation in sand cast sample after 8% plastic strain

The in situ results (Fig. 3) illustrate that the damage of Si particles is important primarily at the cell boundaries, and corresponds to the incubation stage of failure, in agreement with the results of Yeh and Liu.16 The damage accumulation is not homogeneous, because of the complex effect of the 3D spatial arrangement of the eutectic: some dendrites are more stressed than others, and the progress of the particle breaking along the cell boundaries is somewhat dispersed. Moreover, after Si microcracking, cavity growth does not occur in the usual sense: an increase in strain induces microcracks to open without propagation in the surrounding matrix (Fig. 8). Upon further straining, the ligament between the microcracks may become unstable and collapse (Fig. 9).

Fig. 9

Onset of plastic collapse of the ligament between microcracks in Si particles of chill cast sample containing 10% plastic strain

The data obtained from the in situ tests offer a quantitative explanation of the damage progress, and they are also indispensable to predict the specific effect of the microstructure. In this respect, the damage development depicted by the SEM images suggests the application of Thomason’s model of ductile fracture17 to interpret the sequence of the observed phenomena. The incipient limit load σ _l,l, which is necessary for the plastic collapse of the ligament between two microcracks in adjacent Si particles (Fig. 9), depends above all on the geometry of the system during straining. The geometric variable is the distance h between the faces of the microcrack in Si. It should be noted that h = h(ε _p), whereas D _eq and s are fixed, as reported in Table II. Therefore, the scale length of the phenomenon is D _eq or s, on the order of some micrometers. σ _l,l should be considered as a local load limit between particles that does not concern the global specimen. The plastic collapse stress of the ligament, σ _l,l reported here, after modifying Thomason’s original model, is

$$ \frac{{\sigma_{{{\text{l}},{\text{l}}}} }}{{R_{0.2} }} = \left[ {\frac{\alpha \left( n \right)}{{\left( {\frac{{h\left( {\varepsilon_{\text{p}} } \right)}}{{s \cdot e^{{ - 0.5 \cdot \varepsilon_{\text{p}} }} - D_{\text{eq}} }}} \right)^{2} }} + \frac{1.2}{{\sqrt {\frac{{D_{\text{eq}} }}{{s \cdot e^{{ - 0.5 \cdot \varepsilon_{\text{p}} }} }}} }}} \right] \cdot \left[ {1 - \frac{\pi }{4}\left( {\frac{{D_{\text{eq}} }}{{s \cdot e^{{ - 0.5\varepsilon_{\text{p}} }} + D_{\text{eq}} }}} \right)^{2} } \right], $$

(1)

where α(n) is a function of the hardening exponent published by Pardoen and Hutchinson18 and R _0.2 is the yield strength of the matrix.

At this point, and on the basis of the sequence of events revealed by the experimental results, a change of scale is made, considering the collapse of a dendritic cell boundary. The relevant microstructural scale length is now the cell size L, of the order of some tens of micrometers. The fraction β of the microcracks naturally increases during straining, i.e., β = β(ε _p). In a later paper, Thomason19 considered the situation of variable size and spacing of the voids, while in the present work, Si microcracks are considered. He concluded that some of the ligaments reach the limit load of the local collapse in advance, but they remain suspended until the limit load is reached in the neighboring ligaments. In Thomason’s geometric system, global collapse occurs at this point, while in the present case, the in situ observations suggest that the situation may be transformed to the cell boundary. At this stage, the global collapse only concerns a dispersion of favorable individual boundaries (Fig. 6). Thus, Eq. 1 is able to describe the collapse of cell boundary if s is substituted with L, D _eq is substituted with βD _eq, h(ε _p) is substituted with the h _max value reached at the moment of the ligament collapse, and finally the ligament collapse stress σ _l,l is substituted with the collapse stress of the cell boundary σ _c,l. It should be emphasized that β(ε _p) and h _max were determined by means of the in situ experiments.

Finally, applying Eq. 1 to the sand and chill casting separately, it is possible to calculate the σ _c,l(sand)/σ _c,l(chill) ratio of the plastic collapse of the cell boundary for the two cases. The results are illustrated in Fig. 10. It is evident that the ratio below unity means that cell boundary ligament collapse will occur first in sand casting. The microstructural parameters that play a role are the particle diameter D _eq, the interparticle spacing s, and the cell size L. Such quantities are not independent of one another. In an Na (or Sr)-modified and T6 heat-treated alloy, they are determined above all by the solidification rate. As already stated, modification and heat treatments may lead to a certain uniformity in αD _eq and s, whereas L cannot be altered. In conclusion, the tension tests and the calculations carried out by applying Eq. 1 to the quantitative data of β(ε _p) and h _m, resulting from in situ tests, show that αD _eq and s have only a moderate influence on ductility. However, the present in situ results, along with other experimental data published in literature on Al-Si modified alloys, indicate L to be the main parameter that controls the ductility of this alloy.

Fig. 10

Ratio between the sand and chill casting limit loads for a cell boundary

The final step necessary to link microstructure and ductility in a quantitative approach is to consider the last damage mechanism. The behavior of the mesocrack resulting from the joining of microcracks at the cell boundaries is studied to interpret the failure of the sample. Another scale change is necessary: initially the characteristic length is some hundreds of micrometers. The in situ observations highlight the role of the coalescence of some of the cell boundary microcracks (Fig. 7). At a certain point, the resulting crack has propagated enough to reach instability and cause specimen breakage. Having a size on the order of a millimeter, its instability can be modeled by elastoplastic fracture mechanics theory.20 The aim is to obtain a quantitative relationship between the microstructural parameters, in this case L, and ductility, expressed as A. Elongation is, in fact, generally considered an adequate parameter to describe the ductility of a metal. Values of A% are very often used to judge the suitability of castings for specified service conditions.

It is not within the scope of the present paper to give a detailed description of the development of the model. The model modifies the empirical relationships between the fracture toughness K _Ic and L reported by Fan21 and Vorren et al.,22 and then introduces the details of the damage process obtained from the in situ experiments into the equations. The final equation is

$$ A = \exp \left[ {\left( {k_{5} - k_{6} L} \right)^{{^{{\frac{2}{1 + n}}} }} } \right] - 1. $$

(2)

The empirical constants k ₅ and k ₆ are material constants that depend on microstructural parameters, such as αD _eq and s. Their values can be obtained simply by best-fitting the experimental data. Equation 2 expresses elongation as a function of cell size L, which is approximately equivalent to the size of the dendrite secondary arms, and the hardening exponent n of the Hollomon law determined by means of tensile tests. L is the dominating variable, because it is very sensitive to the casting process. For practical use, L can easily be obtained through metallographic analyses.

Equation 2 has been related to Spear and Gardner’s experimental data.23 The result is plotted in Fig. 11, where the data are quite scattered, as is usual in casting products. In fact, apart from experimental errors, casting defects such as porosity, dross, and oxide skins may also be present, thus giving rise to pronounced variability of the results. Nevertheless, the fitting in Fig. 11 is quite good for casting products. In summary, the modeling of ductility, based on the microstructural parameters (here the cell size L) and quantitative details of the damage mechanisms has proven successful. It offers the ability to predict the ductility of castings where the alloy has been modified with Na or Sr and then subsequently heat treated.

Fig. 11

Experimental data of Spear and Gardner23 (small rhombs) and of the present study (thick circles). The fitting line from Eq. 2 is also shown

Conclusions

This study focused on the opportunity to use in situ tensile testing to obtain insight into the microstructural damage mechanisms that underlie the mechanical behavior of materials. Here, A356-T6 Al-Si-Mg alloy casting was chosen as a material model.

The in situ tension tests, together with microstructural characterization and conventional tension testing, made it possible to study the mechanics of the microstructural damage of the material during straining. It has been shown that the in situ tension test is able to span different scale lengths, and offers the possibility of identifying the specific role of the microstructural parameters and of quantitatively describing the material damage progress during straining. As a consequence, it is possible to obtain the quantitative data necessary to construct a model that is capable of predicting the ductility of a cast alloy. The present model is of a multiscale type, in order to fully exploit the data obtained in the in situ experiments.

In the author’s opinion, in situ tension testing should be used more widely to fill the gaps in knowledge on the mechanical behavior of materials.