Identification and modeling of fatigue crack growth mechanisms in a die-cast AM50 magnesium alloy
We experimentally identiﬁed the fatigue crack growth micromechanisms operating near the limit plasticity regime in a commercial high-pressure die-cast AM50 alloy with a critical review of the available literature. An existing multistage fatigue model was modiﬁed to subsequently recognize these micromechanisms in a threefold fatigue crack growth regime. The formation of the main fatigue crack occurred almost exclusively at shrinkage pores and to a lesser extent at large Mn-rich particles. At shrinkage pores, several adjacent cracks typically incubated at the edge of ﬂat interdendritic pores, propagated along the a-Mg dendrite cells/Al-rich eutectic interface, and rapidly coalesced into a main physically small fatigue crack that advanced through the Al-rich eutectic. In the long crack regime, the crack advanced in a mixed transdendritic–interdendritic mode along persistent slip bands spreading over several tens of dendrite cells. The model predicts well the fatigue life compared to the experimental data when these observed mechanisms are accounted for.
The high strength-to-weight ratio of contemporary magnesium alloys that are suitable for ultimate weight reduction purposes in automotive and aircraft components has stimulated in the last decade substantial interest in understanding their fatigue crack growth behavior with process rationalization. Experimental observations on cast Mg alloys have accumulatively revealed that the dendrite cell size, pores, secondary phase particles, persistent slip bands and twinning in the dendrite cells considerably aﬀect the fatigue durability and crack growth mechanisms of dendritic magnesium alloys [3–7].
Our interest of studying the AM50 alloy was motivated by the universal trend to reduce the aluminum content of Mg–Al alloys in an eﬀort to increase ductility under impact and fatigue loading for automotive components. In this paper, relevant micromechanisms of fatigue crack growth in an AM50 alloy are carefully examined. Fatigue tests were carried out in a fully reversed strain control condition at strain amplitudes ranging from 0.1% to 0.7%. The microstructure before and after fatigue failure was analyzed by means of ﬁeld emission gun-scanning electron microscopy (FEG-SEM) and electron probe microanalysis (EPMA) using specimens extracted from the grip regions and gage lengths of tested samples, respectively.
An AM50-based magnesium alloy plate with dimensions of 10 cm by 15 cm by 3 mm was cast in a permanent mold by a high-pressure die-casting technique. Flat dogbone-shaped samples having a 25.4 mm gage length and 10 mm by 3 mm area were extracted for fatigue testing.
Backscattered electron micrographs at low and higher magniﬁcations of polished AM50 samples extracted from the fatigue specimens after failure are shown in Figs. 1 and 2. The microstructure comprises the following:
(1) Hexagonal a-Mg dendrite cells having a linear inter- cept average size of 10 mm. However, large dendrite cells with an average size of 50 lm appear in groups within isolated nodules (Fig. 1(a)). (2) A continuous ﬁlm approximately 2–3 mm thick deco- rating the entire network of the dendrite boundaries (Figs. 1(b) and 2(a)). X-ray mapping (Fig. 3) and quantitative wavelength dispersion spectroscopy (WDS) analyses (Fig. 4) reveal progressive depletion of Al content from the grain boundary toward the dendrite center. The Al-rich crown decorating the boundary corresponds then to the Al-rich eutectic phase that seems to be harder to polish than the bulk of the dendrite cell (Figs. 1(b) and 2(a)). (3) Within the Al-rich eutectic ﬁlm, two distinctive precipitates evolve with substantial chemical contrast.
Fatigue life modeling
We used the McDowell et al.  multistage fatigue model originally developed for a cast A356-T6 aluminum alloy to account for the microstructural eﬀects on the cyclic damage progression. The model partitions the fatigue crack growth into three regimes: crack incubation (INC), microstructurally small crack (MSC) propagation, and long crack propagation (LC). As such, the total fatigue life is calculated as the cumulative number of cycles spent in these consecutive regimes. The model correlation results are shown in Fig. 18. Following our fractographic analyses, four types of defects were listed in descending order of severity level: (1) round gas pores greater than 20 mm in diameter; (2) Mn-rich particles greater than 20 mm in diam- eter near or at the free surface or corner; (3) shrinkage pore clusters with a maximum size of the order of 500 mm in individual interdendritic lamellar voids, and (4) an Al-rich eutectic layer. In the results of Fig. 18, we assumed that the crack incubated at debonded Mn-rich particles or individual interdendritic pores. In fact, the experimental ﬁnding that a multi-nucleation phenomenon occurs at individual interdendritic shrinkage pores was not actually accounted for in the model. However, the essential eﬀect of this multi-nucleation phenomenon mainly manifests when the various small cracks coalesce into a longer crack inducing a jump in the driving force and a sharp acceleration of the crack propagation rate in the long crack regime. We believe, at present, that this crack growth acceleration can be phenomenologically accounted for by considering a larger crack-forming shrinkage pore cluster.
- ↑  Mayer H, Papakyriacou M, Zettl B, Stanzl-Tschegg SE. Int J Fatigue, 2003;25:245–56.  Horstemeyer MF, Yang N, Gall Ken, McDowell DL, Fan J, Gullett, PM. Acta Mater 2004;52:1327–36.  Gall Ken, Biallas Gerhard, Maier Hans J, Horstemeyer Mark F, McDowell David L. Mater Sci Eng A 2005;396:143–54.  Horstemeyer MF, Yang N, Gall K, McDowell D, Fan J, Gullett P., Fatigue Fract Eng Mater Struct 2002;25:1045–56.  Shih Teng-Shih, Liu Wen-Sun, Chen Yeong-Jern. Mater Sci Eng A,00202002;325:152–62.
Citation: Identification and modeling of fatigue crack growth mechanisms in a die-cast AM50 magnesium alloy, El Kadiri, H., Xue, Y., Horstemeyer, M.F., Jordon, J.B., Wang, P.T., Acta Materialia, v 54, n 19, p 5061-5076, November, 2006.