download report - Sapienza
download report - Sapienza
download report - Sapienza
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
Scientific Report 2007-2009<br />
Theoretical physics<br />
T13. The glassy state<br />
The glass is state of matter characterized by a very<br />
large viscosity. Consequently relaxational processes in<br />
a glass are extremely slow and the dynamics of the<br />
glass constituents takes place on a broad spectrum of<br />
timescales. The coexistence of fast and very slow processes<br />
makes the glass dynamics very interesting from<br />
the physical point of view, but also very difficult to treat<br />
analytically.<br />
Experimental facts and analytical theories about the<br />
glass transition and the aging dynamics of glas-formers<br />
have been reviewed in a recent book [1] written by a<br />
member of our group. Thermodynamics of glasses poses<br />
interesting questions still largely unanswered, e.g., the<br />
existence of a thermodynamical phase transition to a<br />
glass state, lowering the temperature or increasing the<br />
density. Such a transition is predicted by mean field approximations<br />
and is called a random first order transition<br />
(RFOT), but its existence in finite dimensional systems<br />
is still a matter of debate.<br />
It is well known that the free-energy barriers between<br />
states, that diverge in mean field approximation,<br />
are large but non-diverging in finite-dimensional models.<br />
Still, how much of the mean field scenario is maintained<br />
in low-dimensional models is unclear. Recently we have<br />
studied in Ref. [2] a one-dimensional version of the Derrida’s<br />
Random Energy Model (REM). The REM, being<br />
a long range model, has a clear RFOT. In our 1D model<br />
we have introduced a length (proportional to the system<br />
size, as in the Kac limit) such that interactions are REMlike<br />
on smaller scales. We indeed find a limiting value<br />
for this crossover length between the REM-like and the<br />
1D behavior, but corrections with respect to the mean<br />
field approximation are huge and would make hard to<br />
find the crossover length in actual glassy models.<br />
Our group has, as well, dedicated quite a large effort<br />
in recent years on the study of glasses of hard spheres. A<br />
system of monodisperse hard spheres is maybe the simplest<br />
showing most of the glass phenomenology and can<br />
be thus considered as a prototypical model. Moreover,<br />
amorphous packings have attracted a lot of interest as<br />
theoretical models for glasses, because for polydisperse<br />
colloids and granular materials the crystalline state is not<br />
obtained in experiments. We have reviewed in Ref. [3]<br />
most of the recent results on systems of hard spheres<br />
obtained with the replica method.<br />
At a first sight it could look strange to use the replica<br />
method, invented to average out the disorder, in systems<br />
with no disorder at all. But a closer look will reveal that<br />
a dense system of hard spheres is likely to be in one of<br />
the many amorphous packing configurations. Even if the<br />
original model has no disorder at all, the configurations<br />
dominating the high density phase are very many as if<br />
they were generated from a disordered Hamiltonian and<br />
the replica method is a natural tool to deal with such<br />
complexity.<br />
Figure 1: The replicated potential for estimating the number<br />
of states in a glassy phase.<br />
In this context the replica method works more or less<br />
as follows. Given a reference equilibrium configuration,<br />
one can construct many replicated configurations, that<br />
interact with a small coupling term with the reference<br />
one. In the inset of Figure 1 we show with red dashed<br />
circle the replicas of the central molecule, which are free<br />
to evolve, but with a coupling ε with respect to the reference<br />
configuration. The main question is what happens<br />
when ε is sent to zero after the thermodynamical limit.<br />
In this limit, under mean field approximations, one can<br />
infer the existence of more than one state from the computation<br />
of the so-called replicated potential (which is<br />
shown with full lines in Figure 1). Actually in Figure 1<br />
we are showing the entropy of configurations having a<br />
certain overlap q with the reference configuration.<br />
In Ref. [4] we have also extended our theory of<br />
amorphous packings of hard spheres to binary mixtures<br />
and more generally to multicomponent systems. The<br />
theory is based on the assumption that amorphous<br />
packings produced by typical experimental or numerical<br />
protocols can be identified with the infinite pressure<br />
limit of long-lived metastable glassy states. We test<br />
this assumption against numerical and experimental<br />
data and show that the theory correctly reproduces<br />
the variation with mixture composition of structural<br />
observables, such as the total packing fraction and the<br />
partial coordination numbers.<br />
References<br />
1. L. Leuzzi et al., Thermodynamics of the glassy state,<br />
Taylor & Francis (2007).<br />
2. Franz et al., J. Phys. A41, 324011 (2008).<br />
3. G. Parisi et al., J. Stat. Mech., P03026 (2009).<br />
4. I. Biazzo et al., Phys. Rev. Lett. 102, 195701 (2009).<br />
Authors<br />
G. Parisi, E. Marinari, F. Ricci-Tersenghi, L. Leuzzi 3 , I.<br />
Biazzo, F. Caltagirone<br />
<strong>Sapienza</strong> Università di Roma 36 Dipartimento di Fisica