Four-point bending fatigue behavior of rheocast AlSi7Mg0.3 alloy: Role of the surface liquid segregation

Qing Zhang a, Stefan Jonsson b, Anders E.W. Jarfors a

aJönköping University, School of Engineering, Materials and Manufacturing, 551 11 Jönköping, SwedenbKTH Royal Institute of Technology, School of Industrial Engineering and Management, Materials Science and Engineering, SE-100 44 Stockholm, Sweden


The surface liquid segregation (SLS) layer in semisolid casting presents higher hardness than the surface of specimens cast using high-pressure die casting (HPDC). Bending fatigue tests showed that semisolid castings present better fatigue properties at higher stress, and this improvement disappears when the applied stress is lower than a critical load. This is because HPDC and SSM castings share the same surface deformation at low stress. More significant deformation is observed for HPDC castings when the stress exceeds the critical load. The presence of surface defects enlarges the difference in deformation at high stress and reduces the critical load.


Semisolid casting

Four-point fatigue test

Surface liquid segregation


1. Introduction

Numerous studies indicate that the Al-7%Si-Mg alloys have better fatigue strength when cast by the Semisolid Metal (SSM) process than cast by the conventional High-Pressure Die Casting (HPDC) process [1][2]. This improvement is attributed to a few causes. First, casting defects significantly influence fatigue strength in terms of defect size [3][4] and location [5][6]. The amount and size of solidification shrinkage and entrapped gases in SSM castings are reduced when injecting a partially solidified slurry. Besides, the significantly different microstructures also influence fatigue performance. Brochu et al. [7] proposed that in the absence of defects, the fatigue strength of aluminum alloy 357 is a function of the grain size (D) rather than the secondary dendrite arm spacing (SDAS) or the spherical diameter of the alpha phase globules. Thus, it is concluded that the fatigue strength improvement of the SSM alloy is also related to the smaller grains of the rheocast specimens. The Si content and morphology of Si-rich phases also affect the fatigue properties of Al-Si-Mg alloys by altering the threshold and pseudo-fracture toughness [8].

Reviewing the literature, most studies on the fatigue properties of SSM castings were performed with machined specimens, excluding the casting surface’s influence. Few studies have been reported on the influence of SSM casting surface [9][10]. However, one of the most distinctive characteristics of the SSM castings is the presence of a surface liquid segregation (SLS) skin with higher hardness, which ranges from 0.2 to 1 mm in thickness [11][12]. This harder surface layer leads to gradient mechanical properties across the castings, affecting the fatigue properties. Alain et al. [10] conducted the first investigation into the role of SLS on fatigue properties. They discovered that the presence of SLS leads to a reduction in average fatigue strength due to deformation incompatibility between the high silicon zone and the surrounding matrix in the SLS layer. Consequently, it was recommended to remove the SLS layers. Similarly, Santos et al. [9] reported similar results, with most cracks initiating from surface defects. Notably, the crack initiation sites reported in [9][10] shared a common feature: a distinct boundary between a significantly higher silicon zone and a lower silicon matrix. However, this particular feature was not observed in our previous work [13]. Therefore, further investigation is necessary to comprehensively understand the combined influence of SLS and surface casting defects. Furthermore, in bending fatigue load conditions, the samples bear gradient stresses throughout the section, and the maximum stress locates at the surface, amplifying the role of the SLS skin on the fatigue properties. Therefore, the effect of the SLS skin on fatigue strength needs further study.

In this study, the RheoMetalTM process was adopted for specimen preparation. Compared to other SSM processes, the RheoMetalTM process enables a short casting duration and can be included in a high-pressure die casting (HPDC) route without significant process adjustments making the process a promising alternative for industrial application [14]. On the other hand, the quick slurry-making process results in a relatively higher porosity due to insufficient stirring. In [13], the RheoMetalTM process was optimized by increasing the stirring intensity, reducing porosity significantly. Therefore, in this study, SSM specimens were cast using an optimized process, as reported in [13]. For comparison, HPDC specimens were also cast to study the role of the gradient structure.

2. Experimental materials and methods

2.1. Materials and specimens

The material used in this study was the AlSi7Mg0.3 alloy. The chemical composition of the alloy, measured with a Spectro Max CCD LMXM3 optical emission spectrometer, is listed in Table 1.

Table 1. Chemical composition of AlSi7Mg0.3 alloy [wt.%].


Specimens for fatigue tests were cast to the geometry shown in Fig. 1 (a). The alloy was melted using a resistance furnace with a 200-kg capacity and held at 700 ℃. SSM castings were cast using the RheoMetalTM process. More details on the SSM casting process can be found in [13]. In comparison, HPDC castings were also cast using the same values for the standard casting process parameters. For all tests, the specimens are under as-cast conditions.

Tensile specimens with a thickness of 1 mm were cut from the fatigue specimens at 0, 0.6 and 1.3 mm depth from the surface by CNC machining. All samples were extracted from the reduced parallel section. Before testing, the samples were mounted and ground to reduce the thickness to 0.4 mm, with their centres located at 0.2, 0.8 and 1.5 mm below the surface. For each depth, three specimens were tested. The final geometry of the tensile specimens is shown in Fig. 1 (b).

2.2. Micro-hardness measurements and tensile tests

The variation of mechanical properties from the surface to the centre was evaluated through the microhardness profile for both SSM and HPDC specimens. In addition, a tensile test was performed with the 0.4 mm specimens extracted from different depth.

The micro-Vickers hardness measurements were performed according to SS-EN ISO 6507-1:2008 standard in an E. Leitz Wetzlar hardness testing equipment. First, the measurement was performed on the specimens’ surface after lightly polishing using 6 µm, 3 µm and 1 µm diamond paste, corresponding to the depth of 0. Then the cross-section of the specimens was mounted and mechanically ground using 220 and 320 grit papers, followed by polishing using 6 µm, 3 µm and 1 µm diamond paste. The microhardness on the cross-section was measured up to a depth of 1500 µm, with a measurement taken at increments of 50 µm from the surface until 300 µm and increments of 100 µm from 300 to 1500 µm. A load of 50 g was applied for a duration of 10 s. To avoid interferences between two contiguous indentations, for the first 300 µm, the measurements were taken along inclined lines from the surface (45°). Ten measurements were taken at each depth. A schematic illustration of the indentation locations is shown in Fig. 2.

Tensile tests were performed at a miniature tensile/compression stage (Kammrath Weiss GmbH) with a capacity of 5 kN. The cross-head motion was set to a constant speed of 5 µm/s. A clip-on extensometer was used to determine the elongation.

2.3. Fatigue tests

To investigate the fatigue performance, four-point bending fatigue tests were carried out with an MTS LandMarkTM servo-hydraulic machine at a frequency of 10 Hz with a stress ratio of R = -1. All tests were carried out at room temperature. Tests finished with the final fracture of the specimen or after 2 × 106 cycles without failure. The choice of 2 × 106 cycles as the run-out criterion is determined based on the practical application. Crack growth was monitored by a direct current potential drop (DCPD) method. During the test, a constant direct current was passed through the specimen, and the potential drop between two grips close to the two outer fixtures, as shown in Fig. 3 was measured and logged. At a constant direct current, the potential drop is proportional to the resistance of the specimen, which is determined by crack initiation and propagation. After the test, the logged potential was analyzed for the determination of cycles for crack initiation. As shown in Fig. 3, the thickness of the fixture cannot be considered negligible due to the short distances between the different parts. In addition, the bent spring plate under loading will also apply a specific moment to the specimens. Therefore, it is difficult to calculate the nominal maximum stress applied to the specimens using the beam theory of engineering mechanics used in [15], as the equivalent distance between the fixture is impossible to determine. In this study, a strain gauge (MICRO-MEASUREMENTS EA-13-062EN-350) was used to measure the maximum strains at the specimens’ surfaces. Then the stress amplitude was calculated from the tensile test result of the samples extracted from the surface (centres at 0.2 mm).

2.4. Characterization

Three failed specimens after the fatigue test (at the maximum, medium and minimum stress levels, respectively) were selected from SSM and HPDC specimens for metallographic analysis. Each half-specimen was cut orthogonally to the longitudinal direction and around 5 mm from the fracture surface. The cross sections of the specimens were mechanically ground using 220 and 320 grit papers, followed by polishing using 6 µm, 3 µm and 1 µm diamond paste. To analyze the pores, all cross sections (6 in total for each casting condition) were observed using an Olympus DSX1000 microscope without etching. The pores in the whole cross-section (the total area for each section was 60 mm2) were analyzed using ImageJ software. For SLS layers and microstructure characterization, the cross sections were etched using 10% NaOH solutions and then observed with an Olympus GX71F microscope. The fracture surfaces of failed specimens were analyzed with scanning electron microscopy (JEOL JSM-7001F SEM) equipped with energy-dispersive X-ray spectroscopy (EDS) to identify the crack-initiating defects and measure their size.

3. Experimental results

3.1. Microstructure

Fig. 4 (a, b) show the typical microstructure of the SSM and HPDC castings after a light etch. A typical SLS was observed in the SSM castings, while no distinct SLS was present in the HPDC castings. This segregation surface is expected in SSM castings due to the migration of solid particles in the semisolid slurry. The flow-related lift mechanism is widely accepted to describe this type of segregation [16][17][18]. According to the flow-related lift mechanism, the solid particles are forced to migrate from the wall to the centre due to the higher velocity difference between the particle and the liquid near the wall during the filling process. Due to the concentrated pre-solid particles in the SSM specimens, a steeper density gradient of the eutectic phase was formed. On the contrary, a relatively even distribution of the eutectic phase was found in the HPDC castings, as shown in Fig. 4 (a).

As for the pores, more and larger pores were presented in the HPDC specimens compared to the SSM specimens, as shown in Fig. 4 (c, d). The histogram plots of pore size distribution in the cross-section are shown in Fig. 5 It is evident in Fig. 5 that the majority of the pores found in the HPDC and SSM castings are below 30 μm, and that the number of pores decreases with increasing pore size. However, for each size range, the number of pores in the HPDC specimens is always greater than in the SSM specimens, indicating fewer and smaller pores in the SSM specimens.

3.2. Mechanical properties

Fig. 6 (a) shows the hardness profiles from the casting surface to the centre of the SSM and HPDC specimens. For both curves, fluctuations were observed, accompanied by a significant scatter of the measurements at each depth. This may be a result of the distribution of particles and the presence of pores. For distances to the surface smaller than 200 μm, the SSM specimens present less scatter due to the formation of the SLS layer, within which very small amounts of particles exist, consequently forming a homogeneous feature, as shown in Fig. 4 (b). In contrast, the casting surface of the HPDC specimens is characterized by a mixture of crystals and a surrounding eutectic phase, resulting in low hardness when the indentations hit the crystals. With a depth beyond 200 μm, the SSM castings show more significant scatter and fluctuations. Since the solid particles concentrate in the centre zone, relatively larger eutectic regions with higher hardness can be formed.

Considering the scatter and fluctuations, to determine the tendency of the hardness variation, a moving average of the measurements was obtained using 150 μm as the averaging interval, as shown in Fig. 6 (b). A sharp decrease in hardness from the surface to the centre is observed in SSM specimens, which can be attributed to the segregation in SSM. In the HPDC specimens, no such reduction is found. However, when passing 900 μm in depth there is an onset of a gentle decrease which is related to the higher porosity in the centre region. In addition, the two curves cross at ∼300 µm depth, corresponding well to the SLS layer in SSM specimens, as shown in Fig. 4 (b).

Tensile curves of SSM and HPDC specimens taken from different depths are shown in Fig. 7. An increase in the ultimate tensile strength (UTS) and yield strength (YS0.2) from the centre to the surface of the fatigue specimens was obtained in both SSM and HPDC specimens. In more detail, the SSM and HPDC specimens at the centre (1.5 mm depth) show the lowest UTS and YS0.2, likely due to the concentrated α-Al particles in the centre of the fatigue specimens. Furthermore, compared to the HPDC specimen, the strength of the SSM specimen at the centre was lower due to the higher solid fraction. The surface SSM sample (0.2 mm depth) presents a higher UTS and YS0.2 strength than the corresponding HPDC sample. For the specimens located at intermediate depth (0.8 mm depth) and surface (0.2 mm depth), the variation for HPDC was smaller than for SSM, indicating a steeper microstructure gradient in SSM. In addition, SSM specimens show an overall much higher elongation. The better ductility can be attributed to the fewer and smaller pores in SSM, as shown in Figs. 4 (c, d) and 5, especially in the case of these thin specimens, which amplifies the sensitivity to the defect size. In addition, the correlation between the yield strength of each layer and the weighted average hardness value within the corresponding layer was plotted in Fig. 8. A linear relationship was concluded, which agrees well with the findings in [19][20][21][22].

3.3. Fatigue properties

The S-N curves of SSM and HPDC specimens are shown in Fig. 9. At higher stress amplitude, the SSM specimens present a higher fatigue strength than HPDC specimens. For example, the HPDC and SSM specimens, tested at 160 MPa, failed at 3.5 × 104 and 8 × 104 cycles, respectively, corresponding to two times increase in total lifespan. However, this enhanced fatigue performance reduces with decreasing stress amplitude. It can be seen that the difference in fatigue properties between HPDC and SSM specimens disappears when the stress amplitude is reduced to 100 MPa.

Fig. 10 shows the result of DCPD measurement for both SSM and HPDC castings. It is clear that SSM specimens present a higher fatigue life than HPDC specimens at the same stress level, as shown in Fig. 10 (a) and (b). Depending on the load, a sharp increase of the DCPD is observed at 5000–15,000 cycles, followed by a steady stage. This increase, in the beginning, is the result of the gradual increase of the load (it takes time for the machine to reach the target load) and, therefore, is a consequence of the gradually increased dislocation accumulation under repetitive load [23]. As seen in Fig. 10 (c), which represents a run-out test, the curve fluctuates between 0.01 V and 0.025 V. The interval between the two peaks corresponds to 24 h, indicating that the fluctuations are likely because of the temperature difference between day and night. Therefore, the value of DCPD = 0.025 V was chosen as the criterion for the transition of the cracks from crack initiation to the crack propagation stage. Based on this criterion, the proportion of the cycles for crack initiation to the total fatigue cycles was calculated, as shown in Fig. 11. For HPDC and SSM castings, the crack initiation stage accounts for 90 %, dominating the fatigue life. Furthermore, it is noteworthy that different from the enhanced fatigue property, which depends on the applied stress amplitude, the proportion of the crack initiation stage is independent of the loading conditions. This indicates that the increased stress will accelerate the crack initiation and crack propagation process to the same extent.

3.4. Fractography

Fig. 12(a) shows a typical fracture surface, which can be divided into three regions: (1) fatigue crack initiation region A, (2) fatigue crack propagation region B and (3) fracture region C. Fig. 12 (b) shows the typical crack initiation site of fatigue tested SSM and HPDC specimens. The fracture surface analysis shows that all fatigue cracks initiate from surface defects. Fig. 12 (c) and (d) exhibit striation in crack initiation and propagation regions, respectively. These parallel fatigue striations are the main micro-characteristic of fatigue fracture, and their directions are perpendicular to the crack growth direction, as indicated by the arrows. It can be noted that there is a clear difference in the striation spacing in regions A and B, Fig. 12 (c) and (d). From the figures, it could be calculated that the fatigue striation spacing in region A is roughly 0.5–0.6 µm/cycle, while, in region B, the average value is about 0.9–1 µm/cycle. The striation spacing is the distance of the crack growth per cycle [24][25]. Therefore, the crack growth rate in the crack propagation region is much higher than in the region close to the crack initiation site, which agrees well with the DCPD measurement shown in Section 3.3. The morphology of the final fracture surface, region C, is shown in Fig. 12 (e). It can be seen that only dimples exist, indicating severe plastic deformation.

The influence of the loads on the number of crack initiation sites on the fracture surface was also analyzed, as shown in Fig. 13. As seen, only one crack initiation site was found in both HPDC and SSM specimens when the applied loading was lower than 109 MPa. More than one crack initiation site was observed when the applied stress exceeds 109 MPa, and the number increases with increasing stress amplitude. Furthermore, at a given stress, the number of crack initiation sites in HPDC specimens was higher than that in SSM specimens.

4. Discussion

The fracture surface analysis shows that all cracks initiate from surface defects for SSM and HPDC specimens, as shown in Fig. 13 (b) and (c). Furthermore, the DCPD results show that the crack initiation stage accounts for 90 % of the total fatigue life, indicating that surface defects play a critical role in fatigue properties. In addition to the surface defects, the difference in the surface strength between the HPDC and SSM specimens can also affect the fatigue strength. Therefore, in this study, we will focus on the influence of the defects and the casting surface on the fatigue properties.

4.1. Statistical analysis of the defects

To investigate the role of defects, a statistical analysis of the defects present in the fracture surface was performed, as shown in Fig. 14. As seen, the critical defect size (along the dashed red line in Fig. 14) for crack initiation on the surface decreases with increasing applied stress and consequently smaller surface defects can also initiate the cracks and more crack initiation sites were found, as shown in Fig. 13. Whereas the largest surface crack initiation defects are almost the same, around 200 µm (only one exception for HPDC and SSM specimens, respectively, where the largest surface defects were ∼500 µm), independent from stress amplitude and the type of casting process. Therefore, it is reasonable to conclude that the maximum surface defect size in HPDC and SSM specimens is the same in a statistical meaning.

In addition to the surface defects, larger internal defects are also observed in the fracture surface. The internal defects size can be one order higher than for the surface defects, as shown in Fig. 14. However, no crack initiation from the internal defects was observed, indicating a less detrimental effect on the fatigue properties. This is because, under the bending load, the maximum stress is located in the surface region with decreasing stress from the surface towards the centre, where the stress can be estimated as 0 MPa. In this study, most internal defects are concentrated in the central region, as shown in Fig. 12 (a); consequently, much bigger (approximately 10 times) internal defects can be tolerated during the bending fatigue test.

4.2. Influence of the SLS layer on the fatigue performance

As shown in Fig. 6Fig. 7, gradient properties were formed in both SSM and HPDC specimens. However, due to the pre-existing particles, which accelerated particle migration in SSM casting, distinct SLS layers and steeper gradients were formed in SSM specimens. Consequently, the SSM specimens presented a harder surface and higher tensile strength at the surface.

The correlation between fatigue strength and the intrinsic mechanical properties, including hardness, yield strength and tensile strength, have been widely reported in the literature [26][27][28]. The most interesting observation is that fatigue strength increases with increasing yield or tensile strength for materials with relatively low strengths. However, a reduction in fatigue strength can be observed with further increasing tensile strength [28][29]. This nonsynchronous correlation was attributed to the microstructure characteristics, such as ageing state [30] and the dual-phase microstructure [29]. Therefore, the influence of the strength on the fatigue properties needs to be further analyzed based on the microstructure.

In the current work, the surface of SSM and HPDC specimens consists of Al-particles and a eutectic phase, as shown in Fig. 15 (a, b). Compared to HPDC specimens, the proportion of soft α-Al particles near the surface is much lower in SSM specimens, resulting in higher yield strength of the SSM surface. The yield strength of the HPDC and SSM casting surface was approximated as 166 MPa and 198 MPa, respectively, based on the linear correlation between hardness and yield strength shown in Fig. 8 and the hardness profile shown in Fig. 6.

At relatively high-stress amplitude, for instance, 160 MPa, the applied stress is still 30 MPa lower than the yield strength of the surface of SSM specimens. In contrast, the surface of HPDC specimens approached the yield strength (166 MPa), resulting in a higher strain at the surface, as shown in Fig. 16. Due to higher strain at the surface, the α-Al particles in HPDC specimens suffer from severe deformation and form slip bands. With the increase in the number of cycles, the dislocations accumulate in the α-Al particles, whereas the harder eutectic phase will inhibit the dislocation slip due to the high resistance and high misorientation between these two phases [31]. Consequently, stress concentration generates at the boundaries between the α-Al particles and the eutectic phase, where microcracks are easy to initiate, as shown in Fig. 15 (c) [31]. While, no similar microcracks are found in SSM specimens due to the much less deformation, as shown in Fig. 15 (d).This fatigue mechanism is commonly reported in the dual-phase microstructure [32][33].

When the applied stress is reduced to a relatively lower value, for example, 100 MPa, the applied stress is blow the yield strength of both HPDC and SSM surface, as shown in Fig. 7 (Fig. 7 shows the tensile test result, and the yield strength of the casting surface is higher than 100 MPa. Taking the cyclic hardening effect, the yield strength of specimens in the fatigue test will be further increased and higher than 100 MPa). Thus, the surface of both HPDC and SSM specimens mainly suffers from an elastic deformation. As shown in Fig. 7, Young’s modulus for the HPDC and SSM surfaces are the same. Therefore, the HPDC and SSM specimens present the same surface deformation, confirmed by the measured surface strain during the fatigue test, as shown in Fig. 16. As a result, the role of SLS layer on the improved fatigue property disappears.

4.3. Coordinated effect of surface defects and the SLS layer on the fatigue performance

Notably, in section 4.2, the effect of a surface defect was not considered. The cracks are easy to initiate around the surface defects because surface defects can locally generate stress [34][35] and strain [36] concentrations. In this work, the comparable defect size in HPDC and SSM specimens will result in the same local stress concentration in HPDC and SSM specimens. However, as shown in Fig. 7, the surface strain response to increasing stress differs for HPDC and SSM specimens. As a result, the stress concentration can further enlarge the difference in local strain around surface defects between HPDC and SSM, as shown in Fig. 17. Besides, due to the presence of surface defects, the critical load where the SLS layer starts to affect the value of strain around the defects reduces. Therefore, the surface strain curves separate around 120 MPa while the difference in fatigue property disappears at somewhat lower stress, around 100 MPa, as shown in Fig. 16Fig. 9. Furthermore, when the applied stress is lower than the critical load, HPDC and SSM share the same correlation between strain and stress, resulting a similar fatigue performance. Once the applied stress exceeds the critical load, HPDC specimens will surfer larger deformation and the presence of surface defects can aggravate the deformation locally and thus further deteriorate the fatigue performance.

5. Conclusion

This study investigated the effect of SLS and pores on the four-point bending fatigue behavior of an AlSi7Mg0.3 alloy. Some significant result and conclusion are summarized as follow:

  • 1.A typical harder SLS layer of ̴200 μm thickness was formed in SSM specimens of 3 mm thickness. Compared to the HPDC specimens, a steeper gradient structure was formed in the SSM specimens.
  • 2.For HPDC and SSM specimens, all cracks that lead to final failure initiate from surface defects and the crack initiation period occupies up to 90% of the total fatigue life. Surface defects dominate the fatigue properties, while they are not the cause of the improvement of the fatigue performance of SSM castings due to similar size ranges. Approximately 10 times larger internal defects can be tolerated compared to surface defects.
  • 3.SSM castings show better fatigue properties at higher applied stress due to the higher yield strength of the SLS layer, which results in less surface deformation. In comparison, this improvement disappears at lower stress amplitude, where the elastic deformation dominates, resulting in the same surface deformation for HPDC and SSM specimens.
  • 4.The presence of surface defects can magnify the role of SLS layer in high-stress load and also reduce the critical applied load where the SLS layer starts to play a role.

CRediT authorship contribution statement

Qing Zhang: Conceptualization, Methodology, Investigation, Writing – original draft. Stefan Jonsson: Supervision. Anders E.W. Jarfors: Conceptualization, Methodology, Supervision, Funding acquisition.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.


This work was funded by VINNOVA under the ReCKA project (Grant No. 2018-02831). The authors are thankful for the support from Scania CV, Volvo Car Corporation and Comptech AB. Jacob Steggo and Vasile Lucian Diaconu are also acknowledged for their assistance with the fatigue test and casting experiment.

Data availability

Data will be made available on request.