Solitary Waves and Supersonic Reaction Front in Metastable Solids

Download as pdf or txt
Download as pdf or txt
You are on page 1of 30

SOLITARY WAVES AND SUPERSONIC REACTION FRONT IN

METASTABLE SOLIDS

Hendrik J. Viljoen and Lee L. Lauderback


Department of Chemical Engineering
University of Nebraska-Lincoln
Lincoln, NE 68588-0126
Didier Sornette
Institute of Geophysics and Planetary Sciences
and Department of Earth and Space Sciences
University of California, Los Angeles, CA 90095
and Laboratoire de Physique de la Matiere Condensee
CNRS UMR6622 and Universite des Sciences, B.P. 70, Parc Valrose
06108 Nice Cedex 2, France

Abstract: Motivated by an increasing number of remarkable experimental


observations on the role of pressure and shear stress in solid reactions, explosions and
detonations, we present a simple toy model that embodies nonlinear elasticity and
dispersion as well as chemical or phase transformation. This generalization of the
Toda Lattice provides an effective model for the description of the organization
during an abrupt transformation in a solid. One of the challenges is to capture both the
equilibrium degrees of freedom as well as to quantify the possible role of out-of-
equilibrium perturbations. In the Toda Lattice, we verify that the particle velocities
converge in distribution towards the Maxwell-Boltzmann distribution, thus allowing
us to define a bona-fide temperature. In addition, the balance between nonlinearity
and wave dispersion may create solitary waves that act as energy traps. In the
presence of reactive chemistry, we show that the trapping of the released chemical
energy in solitary waves that are excited by an initial perturbation provides a positive
feedback that enhances the reaction rate and leads to supersonic explosion front
propagation. These modes of rupture observed in our model may provide a first-order
description of ultrafast reactions of heterogeneous mixtures under mechanical
loading.

Keywords: solitons, non-equilibrium, solid phase reactions

1
1 INTRODUCTION
1.1 Experimental motivations
Diffusion transfers of mass or heat usually control front propagation
associated with chemical reactions or phase transformations. As a consequence, the
velocity of fronts is small and even negligible compared to the sound velocities of the
reactants and of the products. Typical solid-solid reactions like Ta + C → TaC or
solid-liquid reactions like 2 Al + Fe2O3 → Al2O3 + 2 Fe , characterized by extremely
high activation energies, can react in the combustion mode and these rates are
determined by the preheating of reactants by thermal conduction. The combustion
κ
front velocity is proportional to where the thermal diffusivity
τ
k O(101 )
κ= ≈ ≈ O(10 −6 ) and the characteristic reaction time
ρCp O(10 ) × O(10 )
3 3

1
τ= − E / RTad ≈ O(10 −2 ) . Therefore, the reaction front velocity is of the order of
k0 e

κ
v∝ ≈ O(10 −2 )m / s. Thus, diffusive transfer cannot explain events propagating at
τ
front velocities much faster than cm/s such as detonations or deflagrations, explosive
recrystallization, photo-induced reactions and the high-pressure heterogeneous
reactions studied by Bridgman in his pioneering work and later by Enikolopyan.
The ultrafast reaction of heterogeneous mixtures under mechanical loading is
particularly intriguing. In 1935, Bridgman reported results of combined hydrostatic
pressure and shear for a wide variety of materials [1]. Whilst most substances
underwent polymorphic transformation, some reacted rather violently. In contrast to
PbO , that decomposed quiescently to a thin film of lead, PbO2 detonated and
residue of Pb was found afterwards. Reactive mixtures produced even more violent
results: stoichiometric mixtures of Cu and S detonated at applied axial load of 2 GPa
(even without applied shear), producing CuS . More exothermic reactions like
Al / Fe2 O3 proceeded in a detonation-like manner, damaging parts of the press - this

reaction was initiated at hydrostatic pressure between 1-3 GPa, even without
application of shear.
Russian scientists actively continued the work of Bridgman. Enikolopyan and
co-workers studied many systems, both endothermic and exothermic in Bridgman

2
anvils and high-pressure extruders [2-10]. They expanded the list of compounds that
were originally investigated. Thermite mixtures of Al and Fe2O3 , pressed into discs

of thickness 4mm, reacted completely within 100 nanoseconds [7,9] (which, if


nucleated from a side would correspond to a velocity as large as 40,000 m/s). The
anvil was destroyed and the lack of plastic deformation in its fracture zones points to
a detonation. Particles were ejected from cylindrical pre-forms (samples were not
radially contained) at velocities up to 2,000 m/s [13]. Reactions were accompanied by
the emission of light, high-energy electrons, acoustic emission and gamma radiation
[5]. The experiments of Enikolopyan lend further proof to the existence of structural
collapse. In order to explain the required level of mixing for these rates, the reactions
must have been preceded by decomposition steps which consist of either a phase
change (solid to liquid or supercritical fluid) or mechanical disintegration (pseudo-
volumetric fracture) or a combination of them. The results for the thermite reaction is
particularly intriguing, since the particle sizes are so large
( 300 µm ≤ φ Al ≤ 1, 000 µm;φ Fe2 O3 ≈ 1, 000 µm ) that the system would otherwise have

difficulty to ignite and react in the normal self-heat-sustained (SHS) mode. To explain
such conversion rates on the basis of diffusion and thermal conduction only, the
reacting particles would have to be clusters of 5-6 molecules.
Fast decomposition of a metastable phase under strain or through the coupling
between chemistry and mechanical strain has also been observed in glassy
semiconductors and metals, as well as in a geological context [11, 12]. The “Prince
Rupert drops”, and more generally, tempered glasses under high strain condition,
which explode as a result of a mechanical stimulation [13,14] is a spectacular
example, which has remained a mystery for several centuries. Studies of detonation of
classical solid explosives (such as heavy metal azides which are compounds bearing
the group N3) also reveal the existence of a fast propagation regime with velocities of
the order of the sound velocity in solid matrices, before the gas explosion stage [15].
The explosive nature of recrystallization of amorphous materials has been described
by Koverda [16]. Hlavacek [17] has observed a clearly distinguishable thermal wave
when intensely milled aluminum powder transforms from amorphous (and highly
plasticized) state to polycrystalline state. Fortov and co-workers [18] applied high
current densities to thin Nb − Ti wires in a cryostat. They have measured propagation

3
velocities for the transformation from the superconduction to the normal conduction
phase of 10 − 12km / s.
In a different context, it has been shown that chemical waves propagate at
very low temperature [19,20] and at usual temperatures [21] at rather high velocities,
due to a coupling between chemistry and mechanical deformations. In the context of
cryochemistry of solids, there is evidence of a transition between slow and fast heat-
mechano-chemical wave modes of, possibly, gas-less detonation were [22]. This
physical phenomenon may be very important, as the fast autowave concept may help
to explain the mystery of fast chemical evolution of substances in the universe [23]. It
has also been proposed that catastrophic geo-tectonic phenomena, such as
earthquakes, may be triggered by gas-less detonation processes of phase
transformations in the earth's crust (for example, explosive decay of a metastable
glassy state of rocks to a more stable, polycrystalline phase) [24,25]. This hypothesis
of phase transformations of rocks induced by a high value of the strain may resolve a
number of difficulties with the current purely mechanical theory of earthquakes [24].
The experimental results described above strongly suggest the importance of a
coupling between chemistry and mechanical deformations. Batsanov [26] pointed out
that, with increased pressure, ionic compounds rearrange valence electron density
distributions and the Szegeti charges (i.e., the actual degree of ionization of the atoms
in ionic crystals) are reduced until a state of decomposition is reached. Gilman has
shown that shear strain changes the symmetry of a molecule or of a solid and is thus
effective in stimulating reactions, much more so than isotropic compression [27-29].
The reason is that a shear strain displaces electronic band energies in a different way,
thus leading in general to a narrowing of the band gap separating the valence from the
conduction bands. When the gap closes, the semi-conductor solid becomes metallic
which triggers a strong chemical reactivity. This process belongs to the class of
phenomena grouped under the term “mechano-chemistry” [30] and is also at work in
the motion of a dislocation by that of kink (leading to plastic deformation). Indeed,
the motion of a kink is akin to a local chemical reaction in which an embedded
“molecule” is dissociated, and then one of the product atoms joins with an atom from
another dissociation to form a new ``molecule'' [28]. Gilman has also proposed [31]
that intense strain deformation (which therefore lead to bending of atomic bonds)
occur in a very narrow zone of atomic scale that can propagate at velocities
comparable to or even higher than the velocity of sound in the initial material.

4
1.2 The limits of conventional theory of shock waves and of explosions and the
need for out-equilibrium mechano-chemistry
The initiation and propagation of shock waves have been studied for many
years and the mainstay of theoretical description is still the macroscopic Rankine-
Hugoniot equations, augmented by the Chapman-Jouguet processes [32]. This theory
has proven to be remarkably accurate despite its simplicity, but it has certain
restrictions and over the years the theory has evolved and it has seen some
modifications. For example the Chapman-Jouguet process yields a constant value for
the velocity of the reaction front with respect to the unreacted phase. Sano and
Miyamoto [33] have developed an unsteady state Rankine-Hugoniot theory and point
out the deficiencies of the steady state theories. But arguably the greatest shortcoming
of the theory to account for the above mentioned experiments is the lack of detailed
description of events that occur on the microscopic level. Erpenbeck [34] combined
molecular dynamics (MDS) and Monte Carlo methods to study diatomic exchange
reactions and showed qualitative agreement with the CJ theory. An interesting finding
of that study is the relatively long time required to reach equilibrium. This raises the
question of non-equilibrium reactions, specifically in the setting of a detonation.
White and co-workers [35] analyzed a similar system by MDS and their findings also
confirmed the validity of the CJ theory only once equilibrium is established.
Considerable effort has been focused on excitation and non-equilibrium in molecular
crystals, due their importance as secondary explosives. Dlott and Fayer [36] and Kim
and Dlott [37] showed that, during the incipient phase after shock loading, bulk
phonon modes are excited first and then energy is transferred to intra-molecular
modes through doorway modes that are most conducive to the transfer. In the case of
naphtalene, equilibrium is only reached 200ps after a 40kbar impact. Coffey [38,39]
studied the interaction of molecules with compressive waves and developed a model
for multi-phonon excitation of intra-molecular vibrational modes. These models have
proven invaluable in the interpretation of experimental results under conditions where
thermodynamics equilibrium is reached at the microscopic scale. However, as
summarized in section 1.1, there are now an increasing number of experiments which
cannot be understood within the conventional framework.
.
1.3 The two fundamental mechanisms

5
Two mechanisms have recently been identified as potential candidates for
explaining very fast phase transformation and explosions in solids: 1) the products are
mechanically stronger than the reactants leading to mechanical shock waves [25, 40,
41]; 2) the released energy is sufficiently large and its release rate is so fast that its
transfer to microscopic degrees of freedom literally boosts and propels atoms to
collide against each other leading to supersonic chemical fronts [29, 31, 25].
Courant and Friedrichs [31] have studied wave propagation of finite amplitude
in elastic-plastic materials and pointed out that shocks are not possible when the
stress-strain characteristics of plastic material is of the weakening type. Sornette [25]
adapted their one-dimensional formalism to study the opposite case in which the
products are more elastically rigid than the reactants, and the density is smaller for the
products than for the reactions, leading to a larger sound velocity for the products.
Under these rather special conditions, a shock develops which propagates at a velocity
intermediate between that the acoustic wave velocities of the reactions and products.
Consider a bar of material deformed uniformly with an initial strain everywhere along
it. Suppose that a localized perturbation or inhomogeneity produces a local
deformation larger than the initial strain at the left boundary of the bar. Qualitatively,
the density perturbation will start to advance to the right in the product phase. Since
the velocity is larger in the product phase, the largest deformations propagate the
fastest. An initial smooth disturbance will progressively steepen and a shock will
eventually form. The shock is fundamentally due to the rigidifying transformation
from reactants to products. The importance of this condition was independently
recognized by Pumir and Barelko [40,41], using a slightly different formulation. Their
framework coupling the elastic wave equation to a reaction-diffusion equation
allowed them to reveal the existence of supersonic modes of deformations with the
existence of a critical strain necessary to ignite gasless detonation by local
perturbations.
The second mechanism discussed by Gilman [29, 31] is illustrated in figure 1
(see also Ref. [25]). A one-dimensional chain is made of atoms of mass m linked to
each other by energetic links of spring constant k which, when stressed beyond a
limit, rupture by releasing a burst of energy ∆g converted into kinetic energy
transmitted to the atoms. Initially the chain of atoms is immobile. Suppose that the
first atom on the left is suddenly brought to a position that entails the rupture of the

6
first bond. This rupture releases the energy ∆g which is converted into kinetic
energies of the atom fragment that is expelled to the left and of the next atom to the
right which becomes the new left-boundary of the chain. Now, due to the impulsive
boost (∆g/m)1/2 that the boundary atom received, it will eventually stress the bond
linking it to the next atom towards its rupture threshold. When this occurs, it is
expelled by the energy that is released and the next atom forming the new boundary is
itself boosted suddenly by the amount (∆g/m)1/2. It is then clear that this leads to a
shock propagating at a velocity larger than the sound velocity equal to c=(k/m)1/2 in
the long wavelength limit, since the atoms are receiving boosts that accelerate their
motion faster than what would be the propagation by the springs with the usual
acoustic wave velocity. Taking the continuous limit, Gilman [26, 28] proposes that
the resulting supersonic shock velocity U is given by the Einstein formula U2 = c2 +
∆g.

1.4 Our modeling strategy using the reactive Toda lattice


The understanding of the conditions under which these two mechanisms will
be active in real materials is at a rudimentary stage, not to speak of their possible
interplay. We need models that allow us to explore the relationship between non-
equilibrium processes at the microscopic scale, the chemical reactions and the
possible development of supersonic shocks and explosions. In this goal, we propose
to use a mesoscopic approach describing how the microscopic processes self-organize
into dynamic macroscopic structures of patterns and waves. The fundamental
problem, and therefore the challenge, in any mesoscopic model lies in the
compromise between scale (macroscopic limit) and detail (microscopic limit). The
compromise proposed here is based on the Toda lattice [42], which is a one-
dimensional system of entities interacting through an interaction potential limited to
nearest neighbors. Using a coarse-grained discrete lattice allows us to perform
simulations on large systems at times large compared to all the relevant times scales
so that the characteristics of the self-organization behavior can be studied in details.
The advantage of the Toda lattice are multifold. First, it recovers linear
elasticity and the non-dispersive acoustic waves (phonons) in the large wavelength
and small amplitude limits. Its non-linear potential at large deformation leads to
mode-coupling which, as we shall briefly discuss, gives naturally the Maxwell-

7
Boltzmann distribution of energies. Thus, with a purely deterministic dynamics
(Newton’s equations) on a minimal model, we have a basic thermodynamics that we
can then enrich to study chemical reactions coupled with mechanical strain. The Toda
lattice has dispersion at non-vanishing frequencies, which is adequate to capture the
effect of the microscopic atomic structure as well as any possible mesoscopic
organization. The Toda lattice has been extremely well studied in the literature as a
remarkably simple system exhibiting stable localized collective excitations, called
solitons. Solitons are a particular set of a more general class of solitary waves or
moving discrete breathers found in many lattice systems with both dispersion and
anharmonic interactions [43]: the dispersion tends to disperse the modes while the
non-linearity tends to concentrate them. The resulting behavior is a localized coherent
nonlinear wave. We stress that our choice of the Toda potential does not restrict our
conclusions as similar solitary waves can be observed for a large class on non-linear
potentials. In particular, we have checked that Taylor expansions of the exponential
terms in the Toda potential truncated at different orders do not change our results at
the qualitative level. We think that the result reported below are robust to a
generalization of the Toda potential. Indeed, recently, the existence of solitons has
been established in anharmonic lattices for a large class of interatomic potentials
[48,49].
We modify the classic Toda model to include metastable states and the
possibility for a phase transformation or chemical reaction. Specifically, beyond a
certain strain threshold, the potential felt by the particles of the lattice is modified to
represent a change of phase from the reactants (initial lattice) to the products. In
contrast to regular dispersive waves, solitons act as energy traps because they create a
dynamic state where the local energy flux points in the direction of wave propagation.
As a consequence, the potential energy, which is released during the reaction as
kinetic energy, can be trapped within the soliton, enhancing its localization and its
velocity. This process occurs out-of-equilibrium, i.e., without equilibration with other
degrees of freedom and can thus focus energy to extremely high levels. In our
knowledge, a Toda lattice with chemical reaction has not been studied before.
More generally, the Toda lattice should also shed light on several issues in
ultra-fast solid phase reactions that are currently not understood. When the shock
wave travels through a particle, either fracture or melting must occur, because the
high reaction rates can only be explained by the generation of large surface areas.

8
Toda [42] mentions that the Toda lattice has a property called “chopping
phenomenon”. When a soliton, which is a compressive pulse, is reflected at a free
boundary a temporary tensile pulse develops that could be causal in spalling or
fracture near the free boundary. Since the tensile soliton is unstable it degenerates into
a series of pulses and there is a rapid transfer of energy from the soliton to thermal
vibrations of the lattice.
The paper is organized as follows. In section 2, the classic Toda lattice
(without reaction) is introduced and the lattice equilibrium is investigated. Local
departures from equilibrium are demonstrated for solitons in inert lattices (with no
metastable states) and the problem of non-equilibrium is discussed. In section 3, the
reactive lattice is defined and studied. The model is easily generalized to an axi-
symmetrical 2-D geometry. Some of the interesting results involve the boundary
conditions and wave inversions lead to hot spots and spalling. Section 4 is a
discussion of the results. The connection is made between this model and photo-
induced first order phase transition models.

2 THE CLASSIC TODA LATTICE (TL) MODEL


2.1 Definition
The Toda lattice is a model of a one-dimensional chain of atoms. Consider a
one-dimensional lattice consisting of N particles. Each particle is described by a point
of mass mn that only interacts with neighboring masses. Hence, heterogeneities,

discontinuities, pores and perfectly isotropic states are all defined by the pair-wise
values of the interaction potential and masses. The displacement of the n th mass from
its equilibrium position is yn (t ) . The relative displacement is defined as

rn = yn +1 − yn . The lattice motion is described by the following canonical equations:

d 2 rn
mn = φ ′(rn +1 ) − 2φ ′(rn ) + φ ′(rn −1 ) (1)
dt 2
where φ ′ denotes the first derivative of the potential function with respect to the
relative displacement. In his search for an integrable lattice that also exhibits realistic
mechanical behavior, Toda used a recursive formulae to find periodic and single
soliton solutions. As a result, the potential function
b − ar
φ (r ) = e + br , ( ab > 0 ) (2)
a

9
was proposed. Application of this choice in eq. (1) gives the Toda lattice (TL). Note
that the integrability of the Toda lattice makes this system (1) special because it
exhibits by construction an infinite number of invariants [42,43]. The TL equations
can also be written in terms of the displacement:
d 2 yn
m 2
= b[e − a ( y n − y n −1 ) − e − a ( y n +1 − y n ) ] . (3)
dt
While the TL model is 1D and mesoscopic in nature, we still want to use parameters
that are realistic. As an illustration, we use the properties of Al to determine the
o
parameters in eq.(3). If a = 1 / λ , where λ is the athermal lattice constant = 4.5 Α and
the cold longitudinal sound velocity is c0 = 6,420m / s , it follows that

b = 4.1 × 10 −9 N . The model is written in non-dimensional form: t → tc0 / λ


and y → y / λ . Later, we will look at the effect of chemistry in a simplistic manner by
changing the parameters of the potential function (more details in section 2.3). If we
call a ' and b' some new parameter values due to the chemical reactions that will be
introduced in section 2.3, we will define the dimensionless parameters α = a ' / a ,
β = b ' / b . Thus, α = β = 1 are the values used to model the standard (non-reactive)
lattice. The non-dimensional TL equation becomes

d 2 yn
2
= β [e −α ( y n − y n −1 ) − e −α ( y n +1 − y n ) ] . (4)
dt

2.2 Energy partitioning in the Toda Lattice


2.2.1 The Maxwell-Boltzmann distribution
If the TL is a fair description of a chain of atoms, one would expect that in the
absence of external forcing, the lattice should approach a thermodynamic equilibrium
state. Oscillators that are initially more excited should exchange energy with
neighbors so that in the end the Maxwell-Boltzmann distribution is recovered. It turns
out that the TL model is not ergodic (due to the integrability property), but that does
not exclude energy sharing. This state is described by the Maxwell-Boltzmann
distribution for one translational degree of freedom. Indeed, since the total energy of
the system is conserved (in the absence of dissipation), this corresponds to the micro-
canonical ensemble and the relevant observables are the energy of the microscopic
degrees of freedom.

10
For the sake of brevity, let us define the instantaneous velocity of the n th oscillator as
c and let us decompose it as the sum of a drift or wave component w and of a random
fluctuating component v. When the oscillations are resolved on the time scale
associated with the Debye frequency (which is basically the period of a thermal
vibration), the question arises: how do we discriminate between motion associated
with thermal fluctuation and motion associated with an event like a wave that
propagates through the lattice? The oscillator does not discriminate, it only responds
to the immediate force, be it of thermal or mechanical (wave) origin. The numerical
results of Fig.2 pertains to a situation where no external force was applied, in other
words, we did not have a drift component w and the velocity of each oscillator was
taken to be v. The stochasticity was introduced through the initial conditions of the
oscillators; the initial conditions were zero velocity for all oscillators and random
displacement between [-W, +W] where W is an input parameter for the system. The
larger W, the larger the input energy. The Toda Lattice was then integrated for
random initial displacement and zero velocity and the stochastic component of the
velocity was equal to the total velocity.

If the chain is in equilibrium, it has a Maxwell-Boltzmann distribution parameterized


by T . This means that the probability to find a state of energy E is e-E/kT, where k is
the Boltzmann constant (whose sole use is to convert a temperature scale into an
energy scale) and T is the (temperature) parameter quantified the degree of excitation
or disorder of the vibrations within the lattice. Using the form (2) for the potential
energy, the Maxwell-Boltzmann distribution reads

1
−2 γ [ Φ + v 2 ] / T mc02
f = Ae 2
with γ = (units K), (5)
2k
where we have used the expression of the total energy of a given oscillator (a spring
+ mass) as the sum of the potential energy stored in the spring and the kinetic energy
associated with the stochastic component of the mass velocity.
The dimensionless potential function is defined as
β −α ( y n + 1 − y n )
Φ= [e − 1] + β [ yn +1 − yn ] . (6)
α
This distribution (5) should be calculated by counting the number of oscillators in a
given energy bin, For this, we integrate the non-dimensional TL equation (4) (with

11
α = β =1 since chemistry is not considered here) and construct the histogram of the
energy of individual oscillators using the cumulative statistics over all the elements in
the lattice and over time. Assigning random initial displacements of the lattice points
uniformly distributed in [ −W , W ], the density distribution of the oscillators over
energy space has the form shown in Fig.2 where it is compared with the Maxwell-
Boltzmann distribution. W is thus the characteristic scale of the energy put initially
inside the TL.Fig.2 has been constructed with the choice W = 0.081 which is best
matched by the Maxwell-Boltzmann distribution at the temperature T = 297 K . We
verify by launching several runs that T is proportional to W, indicating that higher
input energy results in higher temperature. This is an important test of our model. In
addition, we kept the boundaries free in these numerical experiments and observed
thermal expansion. It is actually quite interesting that the linear expansion that occurs
compares quite favorably with the linear (volumetric/3) expansion coefficient of
aluminum (recall that the a and b values of the potential correspond toaluminum).

To further test for the relevance of the Maxwell-Boltzmann distribution, we have also
constructed the distribution over a single oscillator (for instance the 50th oscillator)
and by summing the statistics over a time interval ∆t. As ∆t increases, we verify that
the energy distribution of a single oscillator is also well-described by the Maxwell-
Boltzmann distribution. This result implies that the stochastic component of a typical
oscillator velocity added to its potential exhibit an approximate ergodicity property.

2.2.1 Non-equilibrium configurations


The quest for equipartition of energy and for the Maxwell-Boltzmann statistics from
nonlinear dynamics was first initiated by Fermi, Pasta and Ulam who failed [44].
More recent works (see for instance [45]) have shown the subtlety of this problem. In
their pioneering work, Fermi, Pasta, and Ulam revealed that even in strongly
nonlinear one-dimensional classical lattices, recurrences of the initial state prevented
the equipartition of energy and consequent thermalization. The questions following
from this study involve the interrelations between equipartition of energy (Is there
equipartition? In which modes?), local thermal equilibrium (Does the system reach a
well-defined temperature locally? If so, what is it?), and transport of energy/heat
(Does the system obey Fourier's heat law? If not, what is the nature of the abnormal

12
transport?). The surprising result of Fermi, Pasta and Ulam has now been understood:
under general conditions for classical many-body lattice Hamiltonians in one
dimension, it has been shown that total momentum conservation implies anomalous
transport in the sense of the divergence of the Kubo expression for the coefficient of
thermal conductivity [46]. The anomalous transport is thus a specific feature of one-
dimensional systems. Thus, our verification of an approximate Maxwell-Boltzmann
distribution does not prevent the existence of anomalous transport or propagation
properties as we discuss in the sequel.

It is indeed possible to perturb a small region of the lattice in such a way that
these nodes depart from the equilibrium distribution and this perturbation propagates
with conservation of form and of energy. If the chain contains many nodes, the non-
equilibrium state of the small number of nodes perturbed in this coherent mode
(namely forming a soliton) will not significantly alter the overall distribution. The
transit time of a soliton over a given oscillator is very small compared to the time
scale over which equilibrium is achieved at the scale of a single oscillator. The transit
of a soliton is thus fundamentally a non-equilibrium process. To recognize this fact is
essential for our investigation of the coupling with a chemical reaction performed
below.
Indeed, when a shock wave propagates through the lattice, the distribution of
the energies of the oscillators in the shock zone is perturbed away from its Boltzmann
distribution. Depending on the magnitude of the shock (ranging from a sound wave to
a detonation), the degree of deviation from equilibrium could vary between
insignificant to complete. Quantifying this degree of deviation from equilibrium
constitutes one of the major dilemmas of shock theory: within the shock zone, non-
equilibrium could exist, temperature could become meaningless and a macroscopic
description of chemistry with Arrhenius kinetics becomes nonsensical. Our
mesoscopic approach allows us to investigate precisely this regime and the interplay
between the equilibrated degrees of freedom and the out-of-equilibrium impulses.
To study the effect of an external perturbation, consider a force that is applied
at the first node:
πt
FL = M sin( ), t ≤ t I and FL = 0, for t ≥ t I . (7)
tI

13
This impulse corresponds to a single arch starting from and returning to 0. The motion
of the first node is then described by
d 2 y0
2
= β [1 − e −α ( y1 − y 0 ) ] + FL . (8)
dt
The last node could be either fixed or unbounded. When the force is small, the
deviation from equilibrium is slight. Application of a stronger force at the first node
will cause the creation of a solitary wave leading to a strong deviation from
equilibrium as illustrated in a phase diagram for the 70th oscillator shown in Fig.3.
The conditions and parameter values for this example are listed in Table I as case 1.
The oscillator has been in equilibrium before the first arrival of the shock wave as can
be seen from the chaotic trajectory in phase space shown in Fig. 3. When this
oscillator is shaken by the passage of the shock, it rapidly moves away from its
equilibrium state to a novel state, a new position in phase space where it settled
chaotically and equilibrates over time. Due to the finiteness of the system used in the
simulations, the shock will reach the boundary of the lattice and be reflected at the
last node (we use a fixed boundary condition at the extremity n = 200 and a free
boundary at the other extremity). When the 70th oscillator is again shaken by the
reflected solitary wave, it undergoes a second translation. Between these strong
perturbations, the oscillators can reach equilibrium provided the time intervals
between the solitary wave perturbations are much longer than a vibrational period. In
this example, the 70th oscillator started with a temperature of 297 K before the
solitary wave arrived the first time and finished with the temperature of 328 K after
the solitary wave was completely dissipated and final equilibrium was restored
globally in the lattice. Before this final equilibrium was reached, the oscillator has
been displaced several times by the weakening multiply reflected solitary wave.
In the next example (case 2), the duration t f of the applied force is extended

ten times such that it becomes slightly larger than twice the dimensionless natural

period of oscillation of the oscillators, obtained by expanding eq. (4) to linear
2αβ

order and neglecting dispersion (Einstein approximation. The strain (relative


displacement) profiles along the lattice of cases 1 and 2 are shown in Fig. 4 at
t = 100 . There is a distinct difference between the two profiles. In case 1, the solitary
wave travels as a single perturbation through the medium with velocity of Mach 1. In

14
case 2, the initial perturbation has split into two solitary waves. The leading one has
velocity of Mach 1.08 and the second wave travels at Mach 1.02. As a consequence,
as time progresses, the distance between the two waves of case 2 increases. The
solitary waves conserve their form as they pass through each other, consistent with
the original studies of Zabusky and Kruskal.[47], defining them as “solitons” At the
free boundary, the solitons are destabilized and their energy is transferred to the
lattice. The energy near the free boundary thus increases after the soliton has been
reflected and a ‘hot spot’ is generated.
The system was integrated until equilibrium at which the temperature finally
reached system after equilibrium is 517 K in case 2 compared to 328 K in case 1. The
higher temperature of case 2 is expected, since considerable more energy has been
injected into the system. An interesting observation is the creation of multiple waves
whose number is given by the integer of t f / 5 . For t f less than 5, no solitary structure

was observed, only strong concentrated acoustic waves propagating at Mach 1 and
dissipating within the lattice at they propagate. When the (dimensionless) impact time
are 5, 10 and 20, one, two and four solitons are formed respectively. Another
interesting result is the effect of impact strength. When M decreases, (regardless of
impact time), the perturbation does not split into separate solitons. Instead, it travels
as a single perturbation at Mach 1 through the lattice. Although this critical value of
M has not been determined precisely, it is clear that a minimum strength of impact is
required to create the soliton. When the impact strength is small, the system is
adequately described by a linear system (linearization of the exponential terms) and
the velocity of any wave is determined by material properties alone. When the impact
strength increases, the nonlinear interaction becomes prominent and the wave speed is
no longer determined by material properties alone, but also by the magnitude of the
impact. This reflects the fundamental property of solitons and more generally of
solitary waves to result from the competition between dispersion (which tends to have
waves at different frequencies propagate at different velocities) and nonlinearity
(which tends to concentrate waves into sharp shocks).
The stability of solitons and the instability of dark solitons are thoroughly
investigated in the literature (cf. [42]). Solitons keep energy focused in a small region
of the chain and the energy does not disperse to the rest of the chain - a curiously
stable, albeit non-equilibrium state. This is a clear indication that energy transfer in

15
the soliton must occur exclusively in the direction of propagation. The energy flux is
only directed in the direction of propagation. This property is essential to understand
our results reported below of the coupling between the mechanical deformations
associated with solitary waves and chemical reactions occurring in the lattice. Indeed,
if chemical energy is released in the soliton and if it retains its coherent non-
dispersive localized structure, it is possible to focus energy within this solitary wave
to very high levels.

3 CHEMICAL REACTION ON THE TODA LATTICE


The term `chemical energy` is used in a loose sense. We envision the following
experimental situation in which a mechanical system can suddenly undergo a phase
transformation or a chemical reaction when its local mechanical deformation reaches
a threshold. This can occur, for instance, according the mechanisms of Batsanov [26]
and Gilman [27-29] in which the distortion of the lattice by shear strain moves the
electronic bands, leading eventually to a closing of the band gap (metallization) and
therefore to a sudden strong chemical reactivity. In our system, we shall account for
the existence of a phase transformation or of a chemical reaction by introducing a
potential energy source when the relative displacement (strain) rn=yn+1-yn at some
point in the lattice reaches a critical threshold r c. Thus, when two nodes are
sufficiently compressed so that the relative displacement rn becomes smaller than the
threshold r, we assume that the initial system (reactants) is transformed into a new
phase (products) characterized by different parameters for the potential function
defined in the right-hand-side of eq.(4), as shown in Table 1. Note that the
transformation only occurs and the `chemical energy’ is released only when the
material is in the compressed state. In this description, the chemical energy could
refer to the energy difference between phases, to an amorphous-crystalline
transformation or a chemical transformation. Chemical reaction between different
species necessarily involves the issue of mixing, which is not addressed in the present
study.
Some complication may occur as a chemical reaction brings additional time
scales

16
into the system. Here, we assume that the time scale for the release of chemical
energy is much shorter than the time scales involved in the wave propagations. It can
thus be considered to be a discontinuous step in the Hamiltonian of the system.
To investigate the interplay between mechanical waves and chemical energy
release at a phenomenological level, the interaction potential is irreversibly changed
into the following expression
λ −λ f
βf −α f ( y n + 1 − y n + )
Φf = [e λ
− 1] + β f [ yn +1 − yn ] , when rn becomes smaller than rc. (9)
αf

The lattice constant λ f of the product state could be smaller or larger than λ . Three

different cases are investigated, the first case (case 3 in table I) describes a product
that is more dense (i.e. λ f < λ ), the second case (case 4 in table I) describes a

material with larger lattice constant ( λ f > λ )and smaller sound velocity and the third
case (case 5 in table I) describes a product with larger lattice constant and larger
sound velocity than the initial material. The parameters are listed in Table I. (The
interaction potentials before and after the reaction are made continuous at r = rc by

adding a constant to Φ f ).

Case 3(cf. Table I): λ f = 0.9λ

Never does a soliton form in this case when chemistry is active. The initial
perturbation is strong enough to initiate reaction. The wave propagates supersonically
through the medium, but it appears random, like a thermal wave. However, the
leading part of the wave is always compressive. But this compressive state weakens
and about midway through the lattice, the conversion criterion is not met. The
conversion halts and an ordinary acoustic wave continues to propagate through the
remainder of the lattice with no further conversion. This is surprising as solitons do
form in absence of chemistry. Indeed, with the values of (α f , β f ) =(1.5;3.0) – but no

reaction (by setting the initial and final values equal to (α f , β f ) (then rc becomes

irrelevant) and setting λ f = λ – we found solitons for perturbations M and t I values as

listed in Table I. For (α f , β f ) =(0.9;,0.8) and (α f , β f ) =(1.1;1.5), we also found

solitons. These solitons differed in the time to develop and their relative positions
with respect to each other. In summary, we can state that for all ``non-reactive’’ cases

17
and for (α f , β f ) as listed in the table, solitons exist. Comparing this result with the

reactive systems, we can conclude that the instantaneous transformation of the lattice
parameters by chemical reaction can destroy the soliton structure.
The relative displacement is shown in Fig.5a-b at t=34 and t=110. Fig. 5a shows a
typical profile of the displacement field in the regime when the conversion is still
active, i.e., the criterion for phase transformation is met during the wave propagation.
Fig. 5b shows a profile when the conversion has ceased to be active as the energy is
not sufficiently concentrated to reach the threshold of conversion. The lattice position
where the conversion ceases depends on the initial impact energy. For example, when
the simulation is repeated for M=0.4 the reaction stops earlier.

Case 4 (cf. Table I): λ f = 1.1λ with a smaller sound velocity


A soliton is not observed for this case and the situation is very similar to the previous
case. The strain profile in Fig. 6 shows an expansion of the lattice behind the wave
front. Therefore, the conversion is interrupted soon after onset because the
compression is not strong enough. It is clear that the energy that is released in the
reaction does not contribute towards sustaining and strengthening of the compression
in the leading part of the wave. Beyond the point where the criterion is no longer met,
the wave propagates as an acoustic wave through the system.

Case 5 (cf. Table I): λ f = 1.1λ with a larger sound velocity

A solitary wave is formed in the reactive lattice and it appears to have all the
properties of a soliton. Of course, it is not a genuine soliton since its front separates
two different phases, with the products left behind it and the reactants in front of it.
All the chemical potential energy that is released during the reaction remains trapped
in the solitary wave, which travels supersonically. Both the compression and particle
velocities in the wave increase, thus the released chemical energy is stored as kinetic
and elastic potential energy.
Note that, after its reflection from the fixed boundary at n = 200 , the solitary wave
has “consumed” all the reactions and the system is made entirely of the new phase or
products. In this new phase, it becomes a bona fide soliton, traveling unaffected
through the other waves in the product material. No further chemical transformations
occur and the soliton now travels with constant speed.

18
4 DISCUSSION
We have presented a numerical study that suggests that solitary waves may play
an important role in explosive supersonic reactions and conversions. For this, we
started from the classic Toda Lattice (TD), extending it beyond its use as a
phenomenological model for processes occurring at the atomic level. Since most of
experimental observations are performed at the meso- and macro-scopic levels, we
use the TL as a simple toy model which embodies the minimal number of essential
ingredients of the problem (non-linear elasticity+dispersion) to explore the different
possible regimes. We stress that the phenomenology documented and summarized in
this paper for the TL is qualitatively in agreement with that obtained from a
continuous description based on the Boussinesq equation that can be further
transformed into the Korteweg de Vries equation. We have then extended the TL to
include chemistry in order to describe a reactive material in non-equilibrium situation,
with applications to explosions and detonations as well as very fast phase conversion.
The TL equations are deterministic but the oscillators clearly adopt the
Boltzmann distribution, thus providing the correct analog of a thermodynamics. The
equilibrium state has been calculated by the probability of finding a given velocity
state space both using an ensemble statistics over a large number of oscillators at a
fixed moment in time and by cumulative the statistics over a long period of time for a
single oscillator (effective ergodicity property). We have clearly documented that the
oscillators approach the Boltzmann distribution at long times, that the state variables
are defined and they can be matched with the continuum model.
When perturbations as in the shock wave occur on time scales too short to
allow the oscillators to reach the Boltzmann distribution, one needs a model that takes
suitably into account both equilibrium and out-of-equilibrum states and their
interactions. In this sense, the TL is ideally suited. We have in addition build-in
chemistry by modifying the Hamiltonian at reactant interfaces to a double well
potential: when a compressive stress threshold is reached, reactants combine
irreversibly to give products that are modeled by a different interaction potential.
Chemical energy is released when the mechanical vibrations resulting from both the
possible existence of a solitary wave and of the rest of the vibrational background
forces a node to reach the conversion threshold and thus crosses the energy barrier.

19
We have observed that the released chemical energy initially contributes to the kinetic
energy of the oscillator. Due to friction (as in the Langevin equation), this energy
tends to be transformed over time into vibrational energy or heat in the transition zone
behind the shock wave. If solitary waves can be nucleated by the initial conditions,
we have shown that they act as traps of energy: in the presence of on-going chemical
energy release, the energy piles up within the trapping solitary wave, providing a
positive feedback which enhances the strength and velocity of the solitary wave. The
trapping property of solitons and of solitary waves result from the fact that the energy
flux within them points forwards in the wake. In contrast, the energy flux that points
backwards in the wake of ordinary dispersive waves resulting into a negative
feedback: chemical reactions thus tend to dampen out and stop in absence of solitary
waves.
When chemistry is included in the lattice, a non-steady state solitary wave is
only observed for the case where the product expands with respect to the reactant and
the system has higher sound velocity. It is not certain that this is a universal result as
there seems to be subtle interply between the formation of solitary waves and the
reaction chemistry that can either be stored and enhance the strength of the soliton or
on the contrary destroy it. The situation thus seems more complex and rich that
obtained previously in [25] and [40,41]. As long as reactants are available, it converts
the material and traps the chemical energy as elastic potential energy and the balance
as kinetic energy. Within the spatial zone in which the chemical reaction actually
takes place, the lattice is highly compressed and particles have large particle
velocities. These velocities belong to the drift component and not the random (or
thermal) component of the total velocity. The chemical transformation is athermal as
the system is far from equilibrium. Furthermore, the chemical energy is not directly
released as thermal energy. No energy spills over into thermal energy as long as free
boundaries or internal defects are absent. At defects and free boundaries, conversion
to thermal energy occurs and eventually leads to thermal equilibrium at long times.
When the product constitutes a contracted state with respect to the reactant, a
supersonic wave is observed directly after impact. But it is not self-sustaining and
disappears, leaving only an acoustic wave. A similar observation is made for a
product with larger lattice constant and lower sound velocity. There results seem to
hold true in all the simulations we have performed until now which explored a
significant part but nevertheless non-exhaustive fraction of the phase diagram.

20
It is an ideal model to study non-equilibrium chemistry and lattice motions. The
limitations of this model must also be recognized, especially its one-dimensionality
and interaction is limited to nearest neighbors. Although 3-D MDS are
computationally costly, they should be used to verify the TL findings. This study is
underway.

21
References
[1] Bridgman, P., ``Effects of High Shearing Stress Combined with High Hydrostatic
Pressure’’, J. Phys. Rev. 48 825 (1935).
[2] Enikolopyan, N.S., ``Ultrahigh Molecular Mobility in Solids’’, Doklady Akad.
NAUK 283 612 (1985).
[3] Enikolopyan, N.S., A.A. Mkhitaryan and A.S. Karagezyan, ``Superfast
Decomposition Reactions in Solid Materials’’, Doklady Akad. NAUK 288 436
(1985).
[4] Enikolopyan, N.S., ``Super-fast Chemical Reactions in Solids'', Russian J. Phys.
Chem. 63 1261 (1989).
[5] Enikolopyan, N.S., A.A. Mhkitaryan, A.S. Karagezyan and A.A. Khzardzhyan,
``Critical Events in the Explosion of Solids under High Pressure’’, Doklady Akad.
NAUK 292 121 (1987).
[6] Enikolopyan, N.S., V.B. Vol’eva, A.A. Khzardzhyan and V.V. Ershov,
``Explosive Chemical Reactions in Solids’’, Doklady Akad. NAUK 292 177 (1987).
[7] Enikolopyan, N.S., A.A. Khzardzhyan, E.E. Gasparyan and V.B. Vol’eva,
``Kinetics of Explosive Chemical Reactions in Solids’’, Doklady Akad. NAUK 294
567 (1987).
[8] Enikolopyan, N.S. and A.A. Mkhitaryan, ``Low-Temperature Detonation
Reactions in Solids’’, Doklady Akad. NAUK 309 900 (1989).
[9] Enikolopyan, N.S., ``Detonation, Solid-Phase Chemical Reaction’’, Doklady
Akad. NAUK 302 862 (1988).
[10] Enikolopyan, N.S., A.I. Aleksandrov, E.E. Gasparyan, V.I. Shelobkov and A.A.
Mkhitaryan, “Direct Conversion of Chemical Energy into Mechanical without
Thermalization”, Doklady Akad. NAUK 319 612 (1991).
[11] Shklovskii, V.A., ``Thermal instability of the phase transition front on decay of
'frozen' metastable states’’, Zh. Eksp. i Teor. Fiz. 82 536-543 (1982).
[12] Meade, C. and R. Jeanloz, ``Acoustic emissions and shear instabilities during
phase transformations in Si and Ge at ultrahigh pressures, Nature 339 616-618 (1989).
[13] Olcott, J.S., Science 140 1189 (1963).
[14] Polyakova, T.V., I.M. Fomin, D.N. Karpinskii, ``Fracture features of Prince Rupert
drops’’, Akad. Nauk 311 431-433 (1993).

22
[15] Barelko, V.V., S.M. Ryabykh, K.Sh. Karabukaev, ``Non-gas detonation in
processes of explosive decomposition of heavy metals azides’’, Sov. J. Chem. Phys.
12 377 (1994).
[16] Koverda, V.P., ``Motion of a crystallization wave in an amorphous medium
containing crystal nuclei’’, Tech. Phys. 39 255 (1994).
[17] Pranda, P; Prandova, K; Hlavacek, V., “Particle size and reactivity of aluminum
powders”, Combustion Science and Technology 156 81-96 (2000).
[18] Fortov, V.E. , V.L. Gershenkroi, V.L. Dorofeev and Yu. D. Kuroedov,``
Detection of Electrodetonational Modes of Normal-Phase Propagation in
Superconductors with High Current Density’’ , Doklady Phys. 43 484 (1998).
[19] Barelko, V.V., I.M. Barkalov, V.I. Goldanskii, D.P. Kiryukhin, A.M. Zanin,
Adv. Chem. Phys. 74 339 (1988).
[20] Barelko, V.V., I.M. Barkalov, V.I. Goldanskii, A.M. Zanin and D.P. Kiryukhin,
Russ. ``High speed autowave transformation regimes in low-temperature chemistry of
solid states’’, Chem. Rev. 59 205 (1990)
[21] Enikolopyan, N.S., Doklady Akad. NAUK 283 897 (1985).
[22] Kiryukhin, D.P., P.S. Mozhaev and V.V. Barelko, ``Detonation mechanism of
autowave phenomena in cryochemical solid phase processes’’, Khimicheskaya Fizika
(russian) 11 264-268 (1992).
[23] Barelko, V.V., I.M. Barkalov, V.I. Goldanskii, D.P. Kiryukhin andA.M. Zanin,
Adv. Chem. Phys. 74 339 (1988).
[24] Sornette, D., ``Earthquakes: from chemical alteration to mechanical rupture’’,
Physics Reports 313 238-292 (1999).
[25] Sornette, D., ``Mechano-chemistry: an hypothesis for shallow earthquakes’’, in
``Earthquake thermodynamics and phase transformations in the Earth’s interior, R.
Teisseyre and E. Majewski (eds), International Geophysics Series, Vol. 76,
(Cambridge University Press, 2000), pp. 329-366.
[26] Batsanov, S.S., ``The Metallization of Inorganic Substances under Pressure’’,
Russ. J. Inorg. Chem. 36 1265 (1991).
[27] Gilman, J.J., ``Insulator-metal transitions at microindentations’’, J. Mater. Res. 7
535-538 (1992).
[28] Gilman, J.J., ``Shear-induced metallization’’, Philos. Mag. B 67 207-214 (1993).
[29] Gilman, J.J., ``Mechanism of shear-induced metallization’’, Czechoslovak J.
Physics 45 913-919 (1995).

23
[30] Gilman, J.J., “Mechanochemistry”, Science 274 65-65 (1996).
[31] Gilman, J.J., “Chemical reactions at detonation fronts in solids”, Philos. Mag. B
71 1057-1068 (1995).
[32] Courant, R. and K.O. Friedrichs, Supersonic Flow and Shock Waves, Interscience
Publishers, Inc. New York (1948).
[33] Sano, Y. and I. Miyamoto ``Shock Jump Equations for Unsteady Wave Fronts of
Finite Rise Time’’, J. Appl. Phys. 84 6606 (1998).
[34] Erpenbeck, J.J. ``Molecular Dynamics of Detonation. I. Equation of state and
Hugoniot Curve for a Simple Reactive Fluid’’, Phys. Rev. A 46 6406 (1992).
[35] Mintmire, J.W., D.H. Robertson and C.T. White, ``Molecular-dynamics
simulationsof void collapse in shocked model-molecular solids’’, Phys. Rev. B 49
14859 (1998).
[36] Dlott, D.D. and M.D. Fayer, ``Shocked molecular solids: Vibrational up
pumping, defect hot spot formation, and the onset of chemistry’’, J. Chem. Phys. 92
3798 (1990).
[37] Kim, H. and D.D. Dlott, ``Theory of ultrahot molecular solids: Vibrational
cooling and shock-induced multiphonon up pumping in crystalline naphtalene’’, J.
Chem. Phys. 93 1695 (1990).
[38] Coffey, C.S., ``Energy localization in rapidly deforming crystalline solids’’,
Phys.Rev. B 32 5335 (1985).
[39] Coffey, C.S. and E.T. Toton, ``A microscopic theory of compressive wave-
induced reactions in solid explosives’’, J. Chem. Phys. 76 949 (1982).
[40] Pumir, A. and Barelko, V.V., Detonation type waves in phase (chemical)
transformation processes in condensed matter, Eur. Phys. J. 10 379-383 (1999).
[41] Pumir, A. and Barelko, V.V., “Propagation and ignition of fast gasless detonation
waves of phase or chemical transformation in condensed matter”, Eur. Phys. J. 16
137-145 (2000).
[42] Toda, M., ``Theory of Nonlinear Lattices ‘’ (Springer Series in Sold-State
Sciences 20) (Berlin: Springer, 1981)
[43] Flach, S. and Willis, C.R., “Discrete breathers”, Physics Reports 295 182-264
(1998).
[44] Fermi, E., Pasta, I.R. and Ulam, S.M. Los Alamos National Laboratory Report
No. LA-1940, (unpublished, 1955)

24
[45] Livi,R.,Pettini, M., Ruffo, S., Sparpaglione, M. and Vulpiani, A., “Equipartition
threshold in nonlinear large Hamiltonian systems: the Fermi-Pasta-Ulam model”,
Phys. Rev. A 31 1039-1045 (1985).
[46] Prosen, T. and Campbell, D.K., “Momentum Conservation Implies Anomalous
Energy Transport in 1D Classical Lattices”, Phys. Rev. Lett. 84, 2857-2860 (2000)
[47] Zabusky, N. and M.D. Kruskal, ``Interaction of Solitons in a Collisionless
Plasma and the Recurrence of Initial States’’ Phys. Rev. Lett. 15 240 (1965).
[48] G. Friesecke, J.A.D. Wattis, Existence theorem for solitary waves on lattices,
Commun. Math. Phys. 161, 391-418 (1994).
[49] J. Szeftel, P. Laurent-Gengoux, E. Ilisca and M. Hebbache, Properties of one-
dimensional anharmonic lattice solitons, Physica A 288, 225-243 (2000).

25
TABLE I
M is the maximum amplitude of the force impulse applied at one extremity of the
chain. tI is the duration of the forcing applied at the boundary which creates a train of
solitary waves disturbing the equilibrium. The parameter λ (resp. λf) is the lattice
constant of the reactants (resp. the products if there is a transformation). When both
are equal (i.e., λ/λf=1), there is no chemical reaction and the simulations corresponds
to the classic Toda lattice. The parameters α f and βf are the ratio of the two
parameters defining the Toda potential taken for the reactions over those taken for the
products. The values αf = βf =1 corresponds to the absence of any chemistry, i.e. to
the standard Toda lattice. rc is the threshold for compression at which the phase
transformation or chemical reaction is triggered.

Case No. M tI λf /λ rc αf βf

1: no 0.5 1. 1. _ 1. 1.
reaction
2: no 0.5 10. 1. _ 1. 1.
reaction
3 0.5 10. 0.9 -0.3 1.5 3.0
4 0.5 10. 1.1 -0.2 0.9 0.8
5 0.5 10. 1.1 -0.2 1.1 1.5

26
Figure 1: Cartoon of a one-dimensional chain made of blocks linked to each other by
energetic links which, when stressed beyond a given deformation threshold, rupture
by releasing a burst of energy converted into kinetic energy transmitted to the blocks.
The figure shows two successive bond ruptures that lead to velocity boosts to the
ejected fragments on the left and to the boundary blocks.

27
Figure 2: Maxwell-Boltzmann distribution (- -) and TL distribution for a chain of 201
oscillators integrated over a total reduced time equal to 500.

Figure 3: Phase diagram for the 70th oscillator showing velocity as a function of
displacement from the equilibrium position before (left chaotic region) and after (right
chaotic region) a strong impact of the oscillator by a solitary wave.

0.15

0.1

0.05

0
-0.1 0 0.1 0.2 0.3 0.4 0.5

-0.05

-0.1

28
Figure 4: Strain profiles at time t=100 of case 1 (a) and case 2 (b) showing
respectively 1 and 2 solitons propagating over a noisy “thermally” equilibrated
background.

0.15 0.1

0.10 0.0

-0.1
0.05
-0.2
-0.00
-0.3
-0.05
-0.4
-0.10 -0.5

-0.15 -0.6

-0.7
200 175 150 125 100 75 50 25 0 200 175 150 125 100 75 50 25

X (a) X (b)

Figure 5: Strain profiles for case3, at a) t=34 and b) t=110.

(a) (b)

29
Figure 6: Strain profiles for case 4.

Figure 7: Strain profiles for case 5 at a) t=60, b) t=160

(a) (b)

30

You might also like