Distribution of plates׳ sizes tell the thermal history in a simulated martensitic-like phase transition

https://doi.org/10.1016/j.ssc.2015.04.016Get rights and content

Highlights

  • A model is proposed to explain the thermal memory effect.

  • The mechanism is based on incomplete thermal circles affecting the plates׳ sizes distribution.

  • The sizes distribution should be calorimetrically readable.

Abstract

A phenomenological 2D model, simulating the martensitic transformation, is built upon existing experimental observations that the size of the formed plates – in direct transformation – decreases as the temperature is lowered; then they transform back in reversed order. As such, if a reverse transformation is incomplete (arrested), the subsequent direct one will show anomalously a large number of big size plates – old plus newly formed – but consequentially a depletion of intermediate sizes, due to geometrical constraints, phenomenon that generates thermal memory.

Introduction

The temperature memory effect, as manifested by shape memory alloys, remarkably shows that a thermodynamic system can remember not only the sense of a transformation, as in magnetic hysteresis, but also quasi-precise temperatures at which certain actions were performed in the past, such as the interruption of a previous phase transition. Naturally, the subject attracted a number of experimental efforts (see, e.g. [1], [2], [3], [4], [5], [6], [7], [8], [9], [10]) towards a clear description, but surprisingly few theoretical scenarios have been proposed (see e.g. [2], [8], [9] ), an unanimously accepted phenomenological model being so far rather elusive.

Succinctly, the temperature memory effect (also called “thermal arrest”) consists in the system remembering one – or several – previous incomplete phase transition(s), as described in the following four-steps process: at step I, a complete direct phase transition (parent to product) is performed, then at step II the reverse (product to parent) transformation is “arrested” – stopped before completion – at a temperature “TA”. Next, at step III, another complete direct transformation is performed and finally at step IV a complete reverse one. In this final reverse transformation, the calorimetric signal shows a dip at a temperature close to TA, at which the transformation was previously arrested at step II.

Efforts are on-going towards understanding of the thermal memory effect. Previous theoretical works suggest that the release of stress from the untransformed martensite can increase its transformation back temperature, which results in a separated second peak [2], [7], [8]. Rodriguez-Aseguinolaza et al. [9] propose a more quantitative model also relaying on the redistribution of stress amongst martensite fractions and the parent matrix after one or multiple incomplete reverse transformations.

In this paper we propose a different mechanism possibly relevant for the thermal memory effect, namely that incomplete thermal circles influence the distribution of the product phase plates׳ sizes, making the latter a “witness” of the thermal history – which should be readable by a calorimetric scan.

The idea is rather simple and is based on the existing microscopic evidence that the martensitic transformation takes place with formation of plates, which do not grow indefinitely, but rather stop at a certain maximum size; then, the phase transition proceeds by formation of other plates. There is also evidence that this intrinsic maximum size decreases with temperature (see, e.g. [11]). Keeping in mind that the size of a formed plate possesses such an intrinsic limitation, it is a simple next step to assume that geometrical constraints may impose supplementary limitations: the new plates can be limited in growth by the existing puzzle of neighboring plates. As such, the distribution of plates׳ sizes becomes history dependent and it will be shown that this property can generate thermal memory.

The outline of the paper is as follows: in Section 2, the general assumptions of the model are described, then a numerical illustration is given in 3 Numerical simulations assuming isothermal direct transformation, 4 Reading thermal memory by reverse transformation (toy model illustration); Section 5 discusses briefly the relevance and limitations of the calculations, and Section 6 concludes the paper.

Section snippets

Model

In this section we present the basic assumptions within a simplified phenomenological model for a solid state phase transition inspired by certain properties of the martensitic transformation. The basic features of the proposed model are:

  • (i)

    Finite size plates: The product phase, once nucleated, grows in the form of plates only up to some intrinsic maximum size where the growth stops. Then, the phase transition proceeds by formation of new plates (rather than further growth of existing ones). In

Numerical simulations assuming isothermal direct transformation

In our numerical simulation, the product phase will be assumed to nucleate in random positions in the parent matrix and to grow in the form of squares which gradually fill the available surface. If a square, during its growing process, encounters neighboring (already existing) squares, or the system borders, it will stop growing, remaining at that respective size.

While the 2D model is justified by the plates-like growth of martensite germs (also, the alloys can be obtained as thin films, etc.),

Reading thermal memory by reverse transformation (toy model illustration)

Performing reverse transformations is a key part in the arrest experiments. In this section we assume a phenomenological transformation rate to simulate DSC curves for the reverse transformation.

The simplest way to introduce the property (iii) is to assume that the interface energy has an elastic component, which can actually be responsible for initiation of the reverse product-parent transition. Let us denote the elastic fraction of the border energy with 4Lσ (necessarily, one must have σσ

Relevance and limitations of the model

A number of experimental papers (see, e.g. [10], [11]) present microscopic evidence that, when increasing the temperature, the martensite plates transform back to austenite in reversed order of their formation. More general, even if some alloys undergo morphologic changes during cooling, the reverse transformation takes place through the same stages as the direct one – but in reverse order [27]. Also in the paper [11], chosen as an example, the author comments on the fact that during the direct

Conclusions

We propose a simple phenomenological model for a solid state phase transition, inspired by some experimentally observed features of the martensitic transformation. Then it is shown that the model naturally exhibits thermal memory.

First, the key assumptions are described, namely that the direct phase transition takes place by formation of finite plates, whose sizes are temperature dependent (with bigger plates formed first) and also – possibly – geometrically restricted by existing surrounding

Acknowledgments

We acknowledge support from the Romanian Ministry of National Education, Grant PN-II-ID-PCE-2012-4-0516 and Core program 45N/2009.

References (28)

  • K. Madangopal et al.

    Acta Metall. Mater.

    (1994)
  • T. Kurita et al.

    J. Alloys Compd.

    (2005)
  • X.M. He et al.

    J. Alloys Compd.

    (2006)
  • Z.G. Wang et al.

    Mater. Lett.

    (2005)
  • Z. Wang et al.

    Int. J. Smart Nano Mater.

    (2011)
  • J. Rodriguez-Aseguinolaza et al.

    Acta Mater.

    (2008)
  • N. Jost

    Mater. Sci. Eng. A

    (1999)
  • J. Liu et al.

    Solid State Commun.

    (1996)
  • A. Banerjee et al.

    Solid State Commun.

    (2011)
  • T. Fukuda et al.

    Scr. Mater.

    (2013)
  • Y.-H. Lee et al.

    Scr. Mater.

    (2011)
  • D. Salas et al.

    Scr. Mater.

    (2014)
  • J. Ortin et al.

    Acta Metall.

    (1988)
  • K. Madangopal

    Scr. Mater.

    (2005)
  • Cited by (10)

    • Magnetic and magnetostrictive properties of the ternary Fe<inf>67.5</inf>Pd<inf>30.5</inf>Ga<inf>2</inf> ferromagnetic shape memory ribbons

      2020, Journal of Physics and Chemistry of Solids
      Citation Excerpt :

      %) over the Ni2MnGa-based Heusler alloy, allowing this material to be used for biomedical applications such as stents or pumps [5]. The Fe-Pd alloy with the atomic composition around 70-30% undergoes by cooling the thermo-elastic and reversible martensitic transformation (MT) from face-centered cubic (f.c.c.) to face-centered tetragonal (f.c.t.) structure [6,7]. A non-thermoelastic and therefore irreversible martensitic transformation (IMT), f.c.t.-b.c.t. (body-centered tetragonal) may be evidenced on further cooling [6,8].

    • Thermal memory fading by heating to a lower temperature: Experimental data on polycrystalline NiFeGa ribbons and 2D statistical model predictions

      2017, Solid State Communications
      Citation Excerpt :

      However, the SMA proved that they can remember not only shapes, but also temperatures at which the austenite-martensite phase transition was stopped before completion in the past (see, e.g. [1–7]). Unlike the shape memory effect, the thermal memory effect (TME) in SMA is less understood, proposed scenarios including the redistribution of accumulated stress [2–4], or geometrical constrictions effects [8]. Within a brief definition, TME means that if the martensite-austenite transformation is stopped before completion (or “arrested”, at a temperature TA), then after the sample is cooled back into the martensite state, if one performs one final, complete martensite-austenite transformation, a specific dip will appear in the calorimetric signal at a temperature close to TA, proving that the alloy has temperature memory.

    • Magnetoelastic properties in polycrystalline Fe-Pd based ferromagnetic shape memory alloys

      2019, IOP Conference Series: Materials Science and Engineering
    View all citing articles on Scopus
    View full text