Ring-diffusion mediated homogeneous melting in the superheating regime
Xian-Ming Bai* and Mo Li
†
School of Materials Science and Engineering, Georgia Institute of Technology, Atlanta, Georgia 30332-0245, USA
Received 11 January 2008; revised manuscript received 3 March 2008; published 18 April 2008
Homogeneous melting in the superheating regime is investigated by using molecular dynamics simulation of
a Lennard-Jones model system. We show that the commonly observed catastrophic melting at the superheating
limit is caused by fast heating rate. By keeping the system isothermally at temperatures below the superheating
limit, we observe intense self-diffusion motions as the precursor of melting. The highly correlated atomic
motions are related to the self-diffusion loops or rings. Two types of loops are observed, closed loop and open
loop, where the latter is directly related to the homogeneous nucleation of the liquid phase. Homogeneous
melting occurs when the number density of diffusion loops reaches a critical value. Our results suggest that
homogeneous melting in the superheating regime is a first-order thermodynamic phase transition triggered by
the self-diffusion loops when the kinetic constraint imposed by heating rate is lessened.
DOI: 10.1103/PhysRevB.77.134109 PACS numbers: 64.70.D-, 61.72.Bb, 66.10.C-, 87.10.Tf
I. INTRODUCTION
Melting is commonly observed but also one of the least
understood phenomena in nature. When the temperature
reaches the melting point, heterogeneous melting usually oc-
curs at defect sites in solids, such as surfaces, grain bound-
aries, and interfaces.
1
The heterogeneous melting can be ex-
plained satisfactorily through the thermodynamic relation
among the interface energies. Namely, the sum of the newly
formed solid-liquid and liquid-vapor interface energies is
less than or equal to the solid-vapor interface energy, or
sv
sl
+
lv
, where s, l, v represent solid, liquid, and vapor
phases, respectively.
1
The decrease in the total interfacial
free energies is the cause for the heterogeneous melting.
However, how a liquid nucleates in a perfect i.e., surface-
free and defect-free crystal remains an open question to
date. Specifically, the detailed microscopic mechanisms and
the thermodynamic relations of this seemingly simple phe-
nomenon still remain unanswered, despite extensive studies
made in the past century.
2–21
The difficulties originated mainly from the presence of
surfaces. Since all materials have surfaces and the preemp-
tive heterogeneous surface melting is always dominant, ho-
mogeneous melting could not occur without superheating the
solids.
4,7,8
Nevertheless, in several cases, superheating was
indeed achieved when special techniques were used to sup-
press the surface melting. For example, Daeges et al.
9
coated
a silver sphere with a thin layer of gold. Since the lattice
constants of gold and silver are very close, coating or pref-
erably epitaxially growing gold is tantamount to removing
the free surface of the silver particle. Because the melting
temperature of gold is higher than that of silver, the coated
silver core would melt earlier than the gold shell. As a result,
up to 25 K of superheating above the normal melting point
in the silver was observed for a time period of about 1 min.
Superheating can also be achieved by using picosecond
pulsed laser irradiation techniques.
10–14
The laser irradiation
heats the materials internally and therefore can initiate melt-
ing from the bulk, or homogeneously. Substantial superheat-
ing about 20% above the normal melting point has been
observed from these experiments. However, melting is ex-
tremely fast in these experiments, typically, in just a few
nanoseconds. Thus, many kinetic and thermodynamic param-
eters could not be precisely measured or controlled. Conse-
quently, it is difficult to investigate the detailed microscopic
mechanisms of homogeneous melting in the superheating
regime from these experiments. Recently, computer
simulations
15–19,21
have been used for investigating homoge-
neous melting. By employing periodic boundary conditions,
one can “remove” surfaces, so superheating can be achieved
easily. However, the heating rates are extremely high in most
simulations,
15–19,21
usually on the order of 10
11
–10
13
K / s.
Under such high heating rates, melting is always observed to
occur catastrophically. Due to the short time window e.g., a
few hundred femptoseconds available in the catastrophic
melting, many vital kinetic and thermodynamic properties
are suppressed. As a result, the detailed microscopic mecha-
nism and kinetic behavior related to homogeneous melting
are either missed or only partially accessible.
15–19,21
In this paper, we investigate the microscopic mechanism
of homogeneous melting in the superheating regime by using
molecular dynamics simulation. The baseline of this study
begins at the heating rate effects on melting. We show that
while the fast heating rate leads to catastrophic melting at
the superheating limit, slow heating rate can lead to more
detailed observations of the thermodynamic and, in particu-
lar, the kinetic processes at temperatures below the super-
heating limit. To this end, an isothermal heating method is
implemented see Sec II A for the explanation. Detailed in-
formation will be examined at fixed temperatures before and
after melting occurs by monitoring various structural and
defect characterization quantities, thermodynamic properties,
kinetic behaviors, and atomic movements. In particular, we
focus on the correlated atomic motions under the isothermal
heating condition and their relation to local disorder or liquid
nucleus formation. Our results reveal that strong diffusive
atomic motions occur in a quiescent period before melting
occurs, which have been missed in the fast heating process.
The highly correlated motions consist of both closed and
open loops of self-diffusing atoms. It is the open loops that
eventually lead to the formation of liquid nuclei. The ther-
modynamic and kinetic properties obtained from this work
allow us to probe into the atomic mechanisms of homoge-
PHYSICAL REVIEW B 77, 134109 2008
1098-0121/2008/7713/13410913 ©2008 The American Physical Society 134109-1