Elsevier

Corrosion Science

Volume 53, Issue 1, January 2011, Pages 168-176
Corrosion Science

A combined neural network and mechanistic approach for the prediction of corrosion rate and yield strength of magnesium-rare earth alloys

https://doi.org/10.1016/j.corsci.2010.09.013Get rights and content

Abstract

Additions of Ce, La and Nd to Mg were made in binary, ternary and quaternary combinations up to ∼6 wt.%. This provided a dataset that was used in developing a neural network model for predicting corrosion rate and yield strength. Whilst yield strength increased with RE additions, corrosion rates also systematically increased, however, this depended on the type of RE element added and the combination of elements added (along with differences in intermetallic morphology). This work is permits an understanding of Mg–RE alloy performance, and can be exploited in Mg alloy design for predictable combinations of strength and corrosion resistance.

Research highlights

► This study presents a body of corrosion data for a set of custom alloys and displays this in multivariable space. These alloys represent the next generation of Mg alloys for auto applications. ► The data is processed using an ANN model, which makes it possible to yield a single expression for prediction of corrosion rate (and strength) as a function of any input composition (of Ce, La or Nd between 0 and 6 wt.%). ► The relative influence of the various RE elements on corrosion is assessed, with the outcome that Nd additions can offer comparable strength with minimal rise in corrosion rate. ► The morphology and solute present in the eutectic region itself (as opposed to just the intermetallic presence) was shown – for the first time – to also be a key contributor to corrosion. ► The above approach sets the foundation for rational alloy design of alloys with corrosion performance in mind.

Introduction

The replacement of Fe-based alloys with Mg alloys can lead to substantial reductions in energy usage over the life cycle of automobiles [1]. The greatest benefits are achieved with the use of cast Mg alloys, particularly those produced by high-pressure die casting (HPDC) where the excellent castability of Mg can be exploited [2]. However, two of the key challenges for the uptake of Mg alloys are: (1) the inherent poor performance of the most common Mg–Al alloys at elevated temperatures typical of those required in applications such as automotive power-trains [3], [4], and (2) the generally poor corrosion resistance of Mg alloys [5].

Mg alloys containing rare earths (RE) as the predominant alloying addition have been shown to possess superior creep resistance to other Mg alloys [6], [7], [8], [9], [10], particularly those based on the Mg–Al system; even when RE, Ca and/or Sr additions are made to Mg–Al alloys in an attempt to improve creep resistance [11]. Much of the work on creep-resistant Mg–RE alloys to date has revealed that REs behave differently, with Nd more effective than La and Ce in improving creep resistance. However, since Ce, La and Nd are the most common RE elements, typically found in misch-metal, the most cost effective alloys are to be derived from a combination of these elements, Therefore, it is necessary to have a detailed understanding of the combination of RE elements with respect to the key physical properties including corrosion rate and yield strength.

Initial studies have focused on the role of individual REs on the mechanical properties of binary Mg–RE alloys [12]. The REs investigated included Ce, La and Nd, which have different solubilities in Mg, viz. 0.23 wt.% for Ce, 0.74 wt.% for La and 3.6 wt.% for Nd. It was found that there is a strong relationship between the volume fraction of intermetallic compounds and the yield strength [13], [14]. In fact, as a first approximation, the yield strength is a linear function of the amount of RE earth content in alloys. It is expected that a similar finding would occur for alloys containing multi-component RE alloys, since Mg12RE (where RE is Ce, La or Nd) is the stable phase in these alloys. The exception to this is that in cases where Nd is the sole or overwhelmingly dominate RE addition, Mg3Nd may form instead of Mg12RE [13], [14].

In addition, we have fundamentally characterised the corrosion and electrochemical properties of binary Mg–RE alloys [15]. It was shown that the electrochemistry of such alloys is rather complex, in that the relative increase in corrosion rate that occurs with increasing alloying addition was heavily influenced by the type of the intermetallic phase. It was determined from micro-capillary electrochemical studies that the corrosion rate (icorr) was increased most remarkably with additions of Ce, followed by La and then Nd. This suggests that the corrosion of multi-component Mg–RE systems may be more complex, particularly if other impurity elements such as Fe, are also present. Fe is well known to have pronounced effect on corrosion rate of alloys, and can be very difficult to control without the addition of Mn as used in Mg–Al alloys [16].

The interpretation of ternary and quaternary data is inherently challenging in an analytical sense, owing to the difficulty in accurately capturing the evolution of properties over a range of compositions (nominally 0–6 wt.%) for several different alloying elements. As a result, in order to manage the data in a holistic framework, we present the development of an artificial neural network (ANN) model for capturing the property evolution as a function of RE additions. One of the strengths of such a tool is that it can identify the key factors that affect a property and point researchers in the direction of the key mechanistic factors that affect that property, particularly for larger data sets.

In this paper, we extensively build on the fundamental knowledge gained from investigation of binary Mg–RE alloys to describe the evolution of corrosion rate and yield strength in ternary and quaternary Mg–RE alloys. These alloys represent a more realistic case for tangible automotive alloys, and such data has not been systematically presented before. It is posited that by understanding the interdependence between the evolution of strength and corrosion rate, more rational alloy development will occur, whilst also revealing fundamental aspects of Mg alloy corrosion.

Section snippets

Alloy preparation

The Mg–RE alloys were produced from commercial purity Mg and RE metals, Ce, La and Nd. In addition to binary alloys, combinations of two and three of Ce, La and Nd, were added to Mg to make ternary and quaternary alloys. Whilst the binary alloys were made over a range of RE contents up to a nominal 6 wt.%, the ternary and quaternary alloys were made such that the nominal RE total was ∼4 wt.%. The precise and relevant compositions of the alloys tested in this study are given in Appendix 1. In all

Typical microstructure of Mg–RE alloys

Increasing the total RE content generally increased the amount of intermetallic phase or eutectic observed within the alloys. Backscattered electron imaging was able to provide good contrast between the intermetallic phase, eutectic region, and the α-Mg grains. Typical microstructures observed in this work are seen in Fig. 1. These images are a selection from representative alloys that typify a binary, ternary and quaternary microstructure (with approximately ∼3.5 wt.% total RE addition). Whilst

Discussion

The data herein has shown that RE additions to Mg raise the yield strength; however simultaneously raise the corrosion rate. The individual RE elements were similar in the ability to increase strength, with La followed by Ce showing the greatest effect. The corrosion response was complex in that the respective REs had a different impact. We note from combining microstructural analysis and ANN modelling, that Ce is most detrimental for corrosion (increasing corrosion rate the fastest) and Nd is

Conclusion

The work presented herein has revealed a large body of previously unreported data pertaining to yield strength and icorr of Mg–RE alloys. The breadth of data lends itself to integration into an ANN model framework that can harness the empirical data into a scientific tool. This allows for prediction and interpolation of both yield strength and icorr values for alloys with any combination of Ce, La or Nd in a given composition window (up to 6 wt.% alloy loading by combination of the three RE

Acknowledgements

The CAST Co-operative Research Centre was established under, and is funded in part by, the Australian Governments Co-operative Research Centres Scheme. The Australian Research Council (through the Centre of Excellence for Design in Light Metals) and the Victorian Government (through the Victorian Facility for Light Metals Surface Technology) are gratefully acknowledged. Maya Gershenzon and Andrew Yob are thanked for producing the samples. We thank the Monash Centre for Electron Microscopy for

References (20)

  • M. Hakamada et al.

    Energy

    (2007)
  • I.P. Moreno et al.

    Scripta Mater.

    (2001)
  • S.M. Zhu et al.

    Scripta Mater.

    (2008)
  • T.L. Chia et al.

    Intermetallics

    (2009)
  • N. Birbilis et al.

    Corros. Sci.

    (2009)
  • M.K. Cavanaugh et al.

    Corros. Sci.

    (2010)
  • J.F. King

    Mater. Sci. Technol.

    (2007)
  • P. Humble

    Mater. Forum

    (1997)
  • A. Luo

    Int. Mater. Rev.

    (2004)
There are more references available in the full text version of this article.

Cited by (0)

1

These authors contributed equally.

View full text