Crystallization Instability in Glass-Forming Mixtures

Abstract

Understanding the mechanisms by which crystal nuclei form is crucial for many phenomena such as gaining control over crystallization in glass-forming materials or accurately modeling rheological behavior of magma flows. The microscopic nature of such nuclei, however, makes their understanding extremely hard in experiments, while computer simulations have hitherto been hampered by short timescales and small system sizes. Here we use highly efficient graphics processing unit simulation techniques to address these challenges. The larger systems we access reveal a general nucleation mechanism in mixtures. In particular, we find that the supercooled liquid of a prized atomistic model glass former (Kob-Andersen model) is inherently unstable to crystallization, i.e., that nucleation is unavoidable on the structural relaxation timescale, for system sizes of 10 000 particles and larger. This is due to compositional fluctuations leading to regions composed of one species that are larger than the critical nucleus of that species, which rapidly crystallize. We argue that this mechanism provides a minimum rate of nucleation in mixtures in general, and show that the same mechanism pertains to the metallic glass former copper zirconium (CuZr).

Unveiling the medium-range order in glass models and its role in glass formation

“…Previous studies observed that the crystallization of the KA system starts with the formation of pure-A nuclei [29], while in the crystallized MKA system, we also observe a pure-A crystal and an AB B2 region, as can be seen from Fig. 1(f).…”

“…S1 in the Supplemental Material [37]). In comparison, it has been shown in the literature that the nucleus in KA is in pure A [29], suggesting that the nucleation in KA is also aided by the composition change. Nevertheless, at higher B concentrations (>40%), the B2 nucleus in KA can form instead [41,42].…”

“…1(c). Previous studies have revealed that the KA system is also a poor glass former at large system size [29]; it is thus necessary to check whether the difference in GFA between the KA and MKA systems is caused by the finite size effect. For this purpose, we perform MD simulations for both systems with N = 16 000 and the cooling rate being 1 × 10 −5 and 1 × 10 −7 .…”

“…Therefore, it is the pure-A crystal with marginal energy gain serving as the nucleus in KA, because B particles are trapped in the pentagon-rich MRO cages. Hence in larger systems that facilitate the composition fluctuation, the crystallization of KA becomes easier [29], while in the MKA system, the lack of MRO allows the atoms to rearrange easily. Under annealing, once a region with an A/B ratio close to 1 has been formed during compositional fluctuation, it can easily become a B2 nucleus with the largest free energy gain such that the crystallization of the MKA system is much easier than that of KA.…”

Excess-entropy scaling in supercooled binary mixtures

“…The value of the wave vector q is that of the first peak of the static structure factor. The 2:1 composition is a much better glass former than the standard 4:1 composition 72 – 74 , thus giving access to a wider dynamical range. The supercooled dynamics goes three to four decades below T MCT where the standard 4:1 composition would crystallize.…”