Skip to main content

Metallic glasses with reverse aging


Ultra-stable metallic glasses have increased stability over conventional metallic glasses from their fabrication process. X-ray photon correlation spectroscopy shows that an unconventional “anti-aging” of the structural relaxation occurs when the system is driven close to the glass transition.

  • Share

A glass is a peculiar state of matter exhibiting liquid-like (e.g. structural disorder) and solid-like features (e.g. resistance to shear). Metallic glasses (MGs) are considered ideal candidates for technological applications due to their improved properties such as elasticity, hardness, and resistance to chemical corrosion when compared to their (poly-) crystalline counterparts. However, glasses are out-of-equilibrium systems and therefore they tend to spontaneously evolve with time. This phenomenon is known as physical aging and is particularly pronounced in metallic glasses, impairing their properties and therefore their usage.

Ultra-stable glasses form a new and promising family of glasses because of their extraordinary thermodynamic and kinetic stability. The ultra-stability is a direct consequence of the fabrication process, physical vapour deposition onto a substrate kept at temperatures slightly below the glass transition temperature Tg (0.7 Tg - 0.9 Tg) and at low deposition rates (about 1 nm s-1). This method enhances the atomic mobility allowing the system to reach a lower state in the potential energy landscape. The Tg of ultra-stable glasses is usually increased in comparison to conventionally-produced (rapidly-quenched) glasses of the same composition.

Macroscopic studies in ultra-stable organic glasses indicate the absence of physical aging in these systems [1,2], but no direct information on the microscopic processes has been reported so far.

X-ray photon correlation spectroscopy (XPCS) can access the microscopic relaxation of a structural glass at the atomic scale while macroscopic techniques can only measure averaged properties, which are less sensitive to aging phenomena. At ID10CS, XPCS is used to measure the microscopic dynamics in the time domain for hard and soft-condensed matter by (time) correlating coherent scattering patterns (speckles) emerging from the characteristic length scales of the system. The loss of the intensity correlation with time between q-equivalent speckles (q being the momentum transfer) is directly related to the loss of correlation with time between spatial configurations (relaxation) in the evolving internal arrangement. For structural glasses, XPCS can measure the characteristic relaxation time of the next neighbour structural arrangement corresponding to the maximum of the structure factor.

Two-time correlation functions of an as-prepared ultra-stable metallic glass along isothermal XPCS measurements upon a temperature cycle

Figure 1. Two-time correlation functions of an as-prepared ultra-stable metallic glass along isothermal XPCS measurements upon a temperature cycle.

In the case of metallic glasses, XPCS has shown that the atomic dynamics can relax completely within the experimental time window (up to104 s). This has also been observed in the ultra-stable metallic glass studied here, a Cu50Zr50 alloy, vapour deposited on a polished NaCl substrate at 0.89 Tg. Structural rearrangements were revealed on a length scale of atomic distances with a characteristic time of the order of 1000 s, even hundreds of degrees below Tg. To study the aging behaviour, the dynamics of three samples were characterised, one as prepared and two annealed a little below and above Tg (0.97 Tg and 1.03 Tg, respectively), under repeated temperature cycles of three isothermal steps of 1h duration between 0.71 Tg and 0.65 Tg. A fourth standard (quenched) sample has been measured for comparison following the same protocol. Figure 1 shows the temporal evolution of the two-time correlation function during the isothermal steps. Changes in the dynamics are signalled by different widths of the reddish area along the main diagonal. The classical behaviour of a metallic glass was revealed: a full decorrelation around 200 s at the highest temperature, a progressive slowing down at lower temperatures and finally a partial acceleration when reheating at the initial temperature without overlap with the previous measurement. Unexpectedly, the ultra-stable metallic glasses exhibited full decorrelation even hundreds of degree below Tg on a next-neighbour scale and with classical aging behaviour. However, a more striking feature appeared when the three samples were compared under the same protocol. Usually, annealing a glass close to Tg relaxes the dynamics (i.e. increases the relaxation time) since the system is able to find a configuration closer to equilibrium. But this is not the case for ultra-stable metallic glasses. Figure 2 shows that annealing the sample accelerates the dynamics, leading to an anti-aging behaviour where annealing at higher temperature produces an acceleration of the dynamics with respect to the as-prepared sample (Figure 2a). Figure 2b summarises the overall behaviour of an ultra-stable metallic glass with respect to a classical quenched metallic glass. The ultrastability of as-prepared ultra-stable metallic glasses was confirmed as the dynamics is always slower than in conventional metallic glasses over all the temperature cycle. Together with the anti-aging, this behaviour could be interpreted by considering the potential energy landscape: starting from a level with reduced atomic mobility in the bulk, annealing promotes the system to a higher energy state where relaxation can proceed only via very small steps back to a lower energy basin in the potential energy landscape.

Intensity autocorrelation functions of as-produced ultra-stable metallic glass and around Tg annealed ultra-stable metallic glass, revealing anti-aging upon annealing.

Figure 2. (a) Intensity autocorrelation functions of ultra-stable metallic glasses as-prepared and annealed around Tg. The anti-aging upon annealing is revealed by the faster decorrelation of the annealed glasses. (b) Evolution of relaxation times of as-produced and annealed ultra-stable metallic glasses and conventional metallic glasses upon cycles of isothermal XPCS measurements. The shaded region corresponds to panel (a).

In conclusion, ultra-stable metallic glasses and metallic glasses have different local structures due to the different preparation, layer-by-layer deposition vs. quenched melt, such that ultra-stable metallic glasses exist in a more thermodynamically favourable area of the potential energy landscape. Annealing ultra-stable metallic glasses close to Tg can trigger other pathways towards new minima that are less thermodynamically favourable, i.e. annealing decreases the stability of the ultra-stable metallic glass, the opposite of its effect on conventional metallic glasses where annealing increases aging effects.


Principal publication and authors
Anti-aging in ultrastable metallic glasses, M.L. Lüttich (a,b), V.M. Giordano (c), S. Le Floch (c), E. Pineda (d), F. Zontone (b), Y. Luo (a), K. Samwer (a) and B. Ruta (c,b), Phys. Rev. Lett. 120, 135504 (2018); doi: 10.1103/PhysRevLett.120.135504.
(a) I. Physikalisches Institut, Universität Göttingen (Germany)
(b) ESRF
(c) Univ. Lyon, Université Claude Bernard Lyon 1, CNRS, Institut Lumière Matière, Villeurbanne (France)
(d) Department of Physics, Universitat Politècnica de Catalunya-BarcelonaTech, Castelldefels (Spain)


[1] Yu et al., Adv. Mater 25, 129 (2013).
[2] E. Leon-Gutierrez et al., J. Phys. Chem. Lett. 1, 341 (2010).