Vibrational energy transfer in shocked molecular crystals

Joe Hooper

Citation: The Journal of Chemical Physics 132, 014507 (2010); doi: 10.1063/1.3273212
View online: http://dx.doi.org/10.1063/1.3273212
View Table of Contents: http://scitation.aip.org/content/aip/journal/jcp/132/1?ver=pdfcov
Published by the AIP Publishing

Articles you may be interested in

Molecular dynamics simulation of shock melting of aluminum single crystal
J. Appl. Phys. 114, 093507 (2013); 10.1063/1.4819298

Molecular dynamics simulations of shock waves in oriented nitromethane single crystals

J. Chem. Phys. 134, 124506 (2011); 10.1063/1.3561397

Thermophysical properties of liquid carbon dioxide under shock compressions: Quantum molecular dynamic
simulations
J. Chem. Phys. 133, 134503 (2010); 10.1063/1.3491834

The equation of state and nonmetal–metal transition of benzene under shock compression
J. Appl. Phys. 107, 083502 (2010); 10.1063/1.3380593

Shock Induced Decomposition and Sensitivity of Energetic Materials by ReaxFF Molecular Dynamics
AIP Conf. Proc. 845, 585 (2006); 10.1063/1.2263390

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
155.33.16.124 On: Fri, 28 Nov 2014 02:17:00
 THE JOURNAL OF CHEMICAL PHYSICS 132, 014507 共2010兲

Vibrational energy transfer in shocked molecular crystals

Joe Hoopera兲
Research Department, Naval Surface Warfare Center, Indian Head, Maryland 20640, USA
共Received 20 August 2009; accepted 20 November 2009; published online 6 January 2010兲

We consider the process of establishing thermal equilibrium behind an ideal shock front in
molecular crystals and its possible role in initiating chemical reaction at high shock pressures. A new
theory of equilibration via multiphonon energy transfer is developed to treat the scattering of
shock-induced phonons into internal molecular vibrations. Simple analytic forms are derived for the
change in this energy transfer at different Hugoniot end states following shock compression. The
total time required for thermal equilibration is found to be an order of magnitude or faster than
proposed in previous work; in materials representative of explosive molecular crystals, equilibration
is predicted to occur within a few picoseconds following the passage of an ideal shock wave. Recent
molecular dynamics calculations are consistent with these time scales. The possibility of
defect-induced temperature localization due purely to nonequilibrium phonon processes is studied
by means of a simple model of the strain field around an inhomogeneity. The specific case of
immobile straight dislocations is studied, and a region of enhanced energy transfer on the order of
5 nm is found. Due to the rapid establishment of thermal equilibrium, these regions are unrelated to
the shock sensitivity of a material but may allow temperature localization at high shock pressures.
Results also suggest that if any decomposition due to molecular collisions is occurring within the
shock front itself, these collisions are not enhanced by any nonequilibrium thermal state.
© 2010 American Institute of Physics. 关doi:10.1063/1.3273212兴

I. INTRODUCTION frequency phonon modes of the crystal,10 and ultimately full

The initiation of chemical reaction in a solid material by
a shock wave is a complicated process involving many dif-
ferent physical mechanisms on a wide range of length and
time scales. The important applications of this process in
detonation physics and shock-induced chemistry have led to
a number of studies attempting to understand the transition
from mechanical impulse to chemical decomposition. The
events occurring at the microscopic scale immediately be-
hind a shock wave are unclear and remain an area of active
study.
Of particular interest to us are the fundamental physics
of thermal processes at early times behind a shock, and the
possibility raised by a number of authors that ignition 共that
is, the onset of chemical reaction兲 may be enhanced by non-
equilibrium phonon distributions at these time scales, espe-
cially near defects.1–7 Recently there is renewed interest in
studying these early time events due to advances in large
scale simulation and experimental prospects for studying
shock waves in situ with ultrafast spectroscopy.8,9 The pos-
sibility of subpicosecond optical diagnostics of shocked ex-
plosives and molecular crystals offers hope of direct confir-
mation of microscopic theories of thermal equilibration and
shock-induced chemistry.
Two primary frameworks for microscopic theories have
emerged that describe the coupling between a shock wave
and the molecules in a crystal. The first generally describes
the process as purely driven by thermal considerations. The
shock deposits thermal energy preferentially into low-
a兲
Electronic mail: joseph.p.hooper@navy.mil.
thermal equilibrium in the sample is achieved by mul-
tiphonon scattering into higher frequency modes. Molecular
dissociation occurs after sufficient energy is transferred into
the internal modes of the molecule. This initial process of
thermal equilibration immediately behind a shock front is
assumed to be complete prior to the later thermal and chemi-
cal nonequilibrium that occurs in the reaction zone of a deto-
nating explosive. These ideas were explored in early works
by Strenzwilk11 and Owens12 using analogies with ultrasonic
attenuation models. Soon afterward, Coffey and Toton1 and
Zerilli and Toton2 proposed more advanced theoretical mod-
els for multiphonon processes in a shock, suggesting
nanosecond-range time scales for equilibration and dissocia-
tion. Dlott and Fayer4 and Tokmakoff et al.7 later proposed a
model for three-phonon processes in molecular solids based
on their extensive experimental work.13–15 They proposed
that energy transfer was rate limited by scattering into a
small number of low-lying internal modes, and that final
thermal equilibrium in a system such as napthalene would be
complete in approximately 100 ps independent of shock
pressure.
Nonequilibrium molecular dynamics simulations have
suggested much faster characteristic times for this energy
transfer based on the analysis of the kinetic energy of mol-
ecules behind a shock. Strachan and Holian16 calculated
equilibration times between 0.5 and 2 ps in polyvinylidene
fluoride shocked at pressures of 10 GPa or higher. Jaramillo
and co-workers17 reported thermal equilibration times of 3–5
ps for elastic-plastic shock waves in cyclotetramethylene-
tetranitramine. Nomura et al.18 similarly observed lattice

0021-9606/2010/132共1兲/014507/14/$30.00 132, 014507-1 © 2010 American Institute of Physics

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
155.33.16.124 On: Fri, 28 Nov 2014 02:17:00
 014507-2 Joe Hooper
phonon/internal vibration equilibrium in approximately 2 ps
in their simulation of shocked cyclotrimethylene-
trinitramine. These recent computational results suggest the
need for further theoretical inquiry into the time scales nec-
essary for equilibrium behind a shock to determine if these
results can be understood in terms of multiphonon scattering.
The second approach for describing the molecule/shock
coupling has focused on mechanical interactions between
neighboring molecules directly within the shock front itself.
Dremin, Klimenko, and co-workers3,19–21 proposed a high-
pressure ignition model that focused on thermally assisted
collisional decomposition within the narrow region of a ris-
ing shock. More recent molecular dynamics simulations us-
ing reactive classical force fields have also suggested that
some manner of direct mechanochemistry may be present at
the shock front regardless of any nonequilibrium thermal
state.18,22
Here we expand on existing theoretical work and explore
vibrational energy transfer following the passage of a shock
wave to study thermal equilibration and whether it may play
a role in the early stages of initiation at high pressures. We
focus primarily on energetic molecular crystals due to wide
interest in the safety and short-time reaction kinetics of such
materials. A simple, analytic theory is introduced which
treats a multiphonon scattering process of any order, neces-
sary for systems with large phonon gaps such as the proto-
typical molecular explosive nitromethane. The approach to
equilibrium is described by a simple relaxation time picture,
which in the high temperature limit reduces to a two-
temperature model. Simple relations are developed for the
change in scattering rate and equilibration time at different
Hugoniot end states following the shock. Predicted time
scales for full thermal equilibration are an order of magni-
tude or faster than suggested by previous theories, on the
order of a few picoseconds for energetic molecular crystals.
These results agree well with the recent large-scale classical
molecular dynamics simulations discussed above, suggesting
such simulations can generally be understood in terms of
multiphonon processes. The possibility of defect-induced
temperature localization due purely to nonequilibrium pho-
non processes is studied by means of a simple physically
based model. The results suggest such a process has little
bearing on shock sensitivity, but may result in temperature
localization at very high pressures such as the detonation
limit. Finally, the implications of our theory for the initiation
of chemistry or sustained detonation in molecular crystals
are discussed in the context of previous work.
II. ENERGY TRANSFER MODEL
We begin by expressing the interaction between the lat-
tice phonons and the internal modes in terms of a simple
relaxation time model. We separate the molecular crystal
phonon system into lattice and internal modes, between
which there is a region with a low phonon density of states
共DOS兲; this general picture has been widely used in theoret-
ical treatments of molecular crystals.23,24 The lattice and in-
ternal systems are assumed to be in local thermal equilib-
rium, allowing the shocked lattice phonons to be
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
155.33.16.124 On: Fri, 28 Nov 2014 02:17:00
J. Chem. Phys. 132, 014507 共2010兲
characterized by a single local equilibrium temperature T p,
with a separate temperature Tv similarly describing higher
frequency phonons, which are assumed to be similar to iso-
lated molecular vibrations.23 In this paper we will refer to
these higher frequency phonons as internal vibrations. We
assume that the multiphonon energy transfer between the lat-
tice and internal systems is the rate limiting step in establish-
ing complete thermal equilibrium behind a shock and in pro-
viding energy to those internal modes related to shock-
induced chemistry or dissociation. Local thermal equilibrium
in the lattice and internal systems is maintained by low-order
scattering processes and intramolecular vibrational redistri-
bution, which are rapid on the time scales of higher order
multiphonon scattering.2,4 It is assumed that some number of
low-lying vibrations in the region of low phonon DOS inter-
act directly with the phonon bath. A shock wave immediately
produces a large increase in the lattice phonon occupation
numbers,10 leading to energy flow into these low-lying inter-
nal vibrations due to phonon scattering. Similar assumptions
were used previously by Dremin and Klimenko,3,20 Zerilli
and Toton,2 and by Dlott and Fayer4 to study three-phonon
processes in molecular solids, and a similar approach is used
in treating related phenomena such as the thermal equilibra-
tion between electrons and phonons after laser pulse
heating.25–27 Here we extend these concepts using a new cou-
pling between the systems and generalizing the treatment to
allow an arbitrary multiphonon process. Additional improved
results for treating the changes in this coupling under shock
compression or near simple crystalline defects are presented
in later sections.
Our first goal is to derive an expression for the flow of
energy into a single low-lying mode which interacts with the
phonon bath directly; we then assume that this mode will
equilibrate with internal modes on a time scale faster than
the mode-bath interaction. We consider a single internal pho-
non of energy ប␻0 at a temperature Tv and assume that the
lattice phonons in the system are at, at a given point in time,
in equilibrium at a temperature T p. In this work subscripts p
and v are used to differentiate between the phonon and in-
ternal vibrational systems, respectively. The internal mode
will relax to the higher shock-induced temperature with a
characteristic time ␶ through multiphonon scattering. We ex-
pand the phonon distribution of the internal mode n共␻0 , Tv兲
around the distribution it would have if it were at equilibrium
with the lattice phonons, n共␻0 , T p兲,23,28
n共␻0,Tv兲 ⬇ n共␻0,T p兲 + ␺n共␻0,T p兲关n共␻0,T p兲 + 1兴, 共1兲
where ␺ represents the deviation from the lattice phonon
system. This expansion is most appropriate at low to moder-
ate shock pressures, but we make the standard assumption
that it is permissible to write the time dependence of the
occupation of this internal phonon of energy ប␻0 within a
general relaxation time picture28
⳵ n共␻0,Tv兲 n共␻0,Tv兲 − n共␻0,T p兲
=− = Pif ␺ , 共2兲
⳵t ␶
where Pif is the transition rate per unit time of anharmonic
phonon scattering that brings the system toward full thermal
equilibrium.
 014507-3 Vibrational energy J. Chem. Phys. 132, 014507 共2010兲

The relaxation time ␶ is related to the transition rate via III. RATE OF UPCONVERSION
the expression28 A. Model Hamiltonian

Pif In keeping with the relaxation time model used in Sec.

␶−1 = . 共3兲
n共␻0,T p兲关n共␻0,T p兲 + 1兴
The change in energy of the lattice phonon system due to
energy transfer into this mode is then
⳵ Ep ⳵ T p ប␻0关n共␻0,Tv兲 − n共␻0,T p兲兴
= Cvp共T p兲 = ,
⳵t ⳵t ␶共T p兲
where Cvp is the heat capacity at constant volume for the
lattice phonons.
In previous work it was assumed that a single vibrational
mode was initially excited by the bath, and all higher lying
modes were excited subsequently by lower order phonon
processes involving this initial mode.4 In general, however,
experiments have suggested that there will be a number of
modes interacting directly with the bath, particularly for
large molecules.29,30 The total energy transfer should prop-
erly include a sum over all such relevant modes. Addition-
ally, at typical shock conditions we may use the high tem-
perature limit of the Bose–Einstein distribution, n
⬇ kBT / ប␻. With these additions the total rate of energy
change can be written as
II, we seek a simple approach in our development of the
transition rate ␶−1 that is grounded in a microscopic Hamil-
tonian but can be expressed in analytic form. We begin by
introducing a Hamiltonian which describes the coupling be-
tween a shock-induced bath of lattice phonons and an inter-
nal vibration. Our approach follows closely that used in clas-
sic works by Nitzan and co-workers24,31,32 and more recently
共4兲
by Skinner, Hsu, and Egorov33–37 in their treatments of vi-
brational relaxation.
The internal vibration is treated as a simple two-level
system described by the states 兩0典 and 兩1典. We then define the
operators
兩1典具1兩 = p,
兩0典具1兩 = q, 共8兲
兩1典具0兩 = q† .
The Hamiltonian consists of three terms, describing the
internal low-lying vibration, the bath of phonons induced by
the shock, and their interaction
H = Hvib + Hshock + Vint . 共9兲

⳵ Tp kB关Tv − T p兴 The 兩0典 state of the internal oscillator is taken to have zero
Cvp共T p兲 =兺 , 共5兲 energy, and the internal vibration term is
⳵t i ␶i共T p兲
Hvib = ប␻0 p, 共10兲
where the sum is over the relevant modes in the low phonon
where ␻0 is the angular frequency of the vibration. The lat-
DOS region. The energy flow out of the lattice phonon sys-
tice phonons are treated using the standard harmonic expres-
tem must equal the flow into the internal modes, and so we
sion
have a corresponding equation for the internal system
Hshock = 兺 ប␻k共b†k bk + 1/2兲. 共11兲
⳵ Tv − kB关Tv − T p兴
=兺
k
Cvv共Tv兲 . 共6兲
⳵t i ␶i共T p兲
The interaction between the two systems is written as
Equations 共5兲 and 共6兲 represent a simple two- Vint = ប␻I共q + q†兲f共Q兲, 共12兲
temperature model analogous to that used for electron-
phonon equilibration25 and for two-phonon absorption by where f共Q兲 is a function of the scalar collective coordinate
Dlott and Fayer,4 although we note that the latter employed
different expressions for treating energy transfer.4,7 Q = 兺 sk共b†k + bk兲. 共13兲
k
The heat capacities that appear in Eqs. 共5兲 and 共6兲 can be
treated in a simple way. The lattice phonons can be modeled Nitzan and Jortner31 introduced a similar phonon/internal vi-
using the Debye approximation for specific heat, but in all bration coupling in their classic work on vibrational relax-
cases of interest when dealing with shock waves we are well ation. Here the sk are dimensionless coefficients describing
above the acoustic Debye temperature TD = ប␻D / kB and the the coupling between the given internal mode and the pho-
classical heat capacities may be used. The heat capacity of non bath. Our simple Hamiltonian ignores any dephasing
the internal vibrations is expressed as a sum over dispersion- effects.
less oscillators, Multiphonon processes are treated by expanding the
function f共Q兲,
共ប␻i兲2exp共ប␻i/kBTv兲/共kBT2v兲
Cv共Tv兲 = 兺 . 共7兲 ⬁
f 共m兲共0兲Qm
i 共exp共ប␻i/kBTv兲 − 1兲2 f共Q兲 = 兺
m=1 m!
. 共14兲
The key step now is to develop an appropriate theory for
the characteristic multiphonon scattering time ␶ and its varia- The interaction term for a process involving scattering of m
tion under shock loading. lattice phonons is then
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
155.33.16.124 On: Fri, 28 Nov 2014 02:17:00
 014507-4 Joe Hooper J. Chem. Phys. 132, 014507 共2010兲

共m兲 ប ␻ I f mQ m the coupling constants sk, setting the value of f m to unity.

Vint = 共q + q†兲, 共15兲 Second, anticipating the use of Debye-type expressions for
m!
the lattice DOS, we assume ␻I / ␻D = c, where ␻D is the De-
where ␻I sets the energy scale of the coupling. bye frequency and c is a constant and is also absorbed into
the coupling coefficients. The weighted phonon DOS is then
B. Vibrational upconversion rates normalized as

The rate for a process in which m phonons combine to

excite an internal mode of angular frequency ␻0 is given by
the usual Golden-rule formula
冕 0
⬁
g共␻兲 = ␣ , 共23兲

2␲ where ␣ is a dimensionless constant which represents an

Pif = 兺 pi兩具1, f兩Vint共m兲兩0,i典兩2␦共EF − EI兲,
ប i,f
共16兲
overall strength of the coupling between the lattice and in-
ternal systems.
where the lattice phonon state is 兩i典 = 兩兵nik其典 and pi is the Bolt- To proceed, we need an expression for the phonon DOS;
zmann distribution describing the lattice phonons, in accord with our desire to obtain analytic formulas for the
exp共− ␤Ei兲 rate, the natural choice is a simple Debye model describing
pi = , 共17兲 the lattice phonons. Enforcing the normalization condition,
兺iexp共− ␤Ei兲
we write the DOS as
where Ei = 兺k共nik + 1 / 2兲ប␻k. The matrix element for an m or-

冦 冧
der process is 3␣␻2
0 ⱕ ␻ ⱕ ␻D
ប␻I f m g共␻兲 = ␻D3 共24兲
共m兲
具1, f兩Vint 兩0,i典 = 兺 sk ¯ skm具f兩bkm ¯ bk1兩i典,
m! k1¯km 1 0 ␻ ⬎ ␻D .
共18兲 We now make two further assumptions, following
Egorov and Skinner,35 to obtain tractable analytic expres-
where we have ignored any phonon emission processes.
sions. The first is that the occupation number n共␻兲 varies
Conservation of energy requires that EF − EI = ប␻0 − ប␻k1 ¯
more slowly with frequency than the weighted DOS, and can
−ប␻km.
thus be moved outside the integral in Eq. 共22兲. The second is
Having assumed that the phonon bath is in local equilib- that we assume the most likely process involves m phonons
rium, we take the thermal average of the phonon creation and of the same frequency; this allows us to evaluate the occu-
annihilation operators in the matrix element, giving a rate pation functions at ␻i = ␻0 / m for i = 1 , . . . , m. The rate then
2␲ becomes
Pif = 共␻I f m兲2 兺 sk2 ¯ sk2 关n共␻k1兲 ¯ n共␻km兲兴
m! k1¯km
1 m
共 ␻ D兲 2
⫻ ␦共␻k1 + ¯ + ␻kn − ␻0兲, 共19兲 Pif = 关n共␻o/m兲兴m
m!
where
n共␻兲 = 共e␤ប␻ − 1兲−1 . 共20兲
⫻ 冕 ⬁

−⬁
冉 冋冕
dte−i␻0t exp m ln
␻D

0
d␻g共␻兲ei␻t 册冊 . 共25兲

We next define a phonon DOS with units of inverse fre- Let

quency, weighted by the dimensionless expansion coeffi-
cients
g共␻兲 = 兺 s2k ␦共␻ − ␻k兲. 共21兲
Gm = 冕 0
␻D
d␻g共␻兲ei␻t . 共26兲
k

The rate can now be written in a general way using the Substituting in the weighted DOS and integrating, we have
multiphonon weighted DOS as
3 ␣ e i␻Dt
Pif =
2␲
m!
共 ␻ I f m兲 2 冕
0
⬁
d␻1g共␻1兲n共␻1兲¯
Gm =
共␻Dt兲3
关2i + 2␻Dt − i共␻Dt兲2 − 2ie−i␻Dt兴. 共27兲

To obtain analytic expressions, we can approximate the

冕 ⬁
0 d␻mg共␻m兲n共␻m兲␦共␻1 + ¯ + ␻m − ␻0兲. 共22兲
integral in Eq. 共25兲 by assuming that m is large and using the
saddle point method. We expand the argument of the second
exponential in Eq. 共25兲 and keep terms only up to second
The extremely limited experimental data on ultrafast pro- order,
cesses in shocked molecular solids make it difficult to fix all
parameters in the rate expression at this time. Previous work 3i␻Dt 3共␻Dt兲2
on vibrational relaxation has generally assumed a functional ln Gm ⬇ ln共␣兲 + − . 共28兲
4 160
form f共Q兲 = exp共−␣Q兲,32 suggesting that our f m is of the form
共−␣兲m. Thus, to simplify the expressions we absorb f m into The rate expression then becomes
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
155.33.16.124 On: Fri, 28 Nov 2014 02:17:00
 014507-5 Vibrational energy J. Chem. Phys. 132, 014507 共2010兲

Pif =
共 ␻ D兲 2
m!
关␣n共␻o/m兲兴m 冕 冉冋
⬁

−⬁
exp it
3m␻D
4
− ␻0 册冊 100

⫻ exp 冉 − 3m共␻Dt兲
160
2
冊 dt. 共29兲
10
-1 Nitromethane
ω0 = 657 cm-1
The assumption that m is large is most appropriate for crys-

τ/τ0
tals with small molecules where a large energy gap exists
between the lattice and internal modes. However, Egorov
and Skinner35 show, using some simple forms of the phonon 10-2
DOS, that the saddle point approximation is quite good even
for m = 2.
Performing the integration, the total rate of energy up-
conversion is then 10-3
1 1.2 1.4 1.6 1.8 2 2.2 2.4

Pif =
4␻D
m!
冑 10␲
3m
关␣n共␻o/m兲兴m
ωD / ω0D

FIG. 1. The change in the characteristic relaxation time ␶ as a function of

⫻ exp 冉 − 5共3␻Dm − 4␻0兲2

6m␻D
2 冊 . 共30兲
increasing Debye frequency. Results are shown for the rate expression using
all possible multiphonon processes 关dotted line, Eq. 共30兲兴 as well as a reso-
nant approximation 关solid line, Eq. 共32兲兴.

This is a general analytic expression for the rate of energy

transfer into the internal modes of a molecular crystal behind
a shock front and can be used directly by summing over all
orders of m. However, we can simplify this rate even further
by considering only the resonant contributions, which occur
when ␻0 = 3m␻D / 4. Plugging this into Eq. 共30兲 yields
We now treat the parameter ␣ as a fit parameter and fix
its value by requiring that the rate produce the experimental
relaxation constant ␶ at ambient conditions, which can be
estimated from experimental phonon lifetimes as discussed
below. All other components of the rate equation can be
taken from quantum mechanical calculations or from experi-

Pif =
4␻D
m!
冑 10␲
3m
关␣n共␻o/m兲兴m . 共31兲
mental spectroscopy results.

IV. HUGONIOT DEPENDENCE

We let m = 4␻0 / 3␻D and approximate the factorial m!
using Stirling’s approximation, ln m ! ⬇ m ln m − m We next consider the change in the energy transfer when
+ ln冑2␲m. Substituting these and simplifying, we have the system has been dynamically compressed by an ideal

冑冋 册
shock. Shock compression of a material leaves the system at
␬
4␻D 5 ␣e a point on the Hugoniot curve, representing the allowable
Pif = n共␻0/␬,T p兲 , 共32兲
␬ 3 ␬ end states following a shock transition from a specific initial
state.38 Here we use only the principal Hugoniot, for which
where the initial state is undeformed and unstressed. The Hugoniot,
combined with additional thermodynamic variables, yields
4␻0
␬= , 共33兲 information on the compression of the material following a
3␻D shock of a given pressure and the macroscopic temperature
increase due to the shock wave. In our case, we are primarily
and e is the standard mathematical constant. This simplified
interested in the change in lattice phonon frequencies due to
rate expression depends only on the lattice phonons as char-
the material compression and the change in phonon occupa-
acterized by the Debye frequency ␻D, the internal mode fre-
tion numbers due to the temperature jump.
quency ␻0, and the parameter ␣. The connection with the
We first consider the pressure shift of the lattice phonon
relaxation time ␶ discussed above is via Eq. 共3兲. The resonant
modes. Our use of a Debye-type spectrum for the phonon
assumption is a very good approximation of the full rate for
DOS allows us to consider the pressure dependence of the
the systems and pressure ranges considered in this work.
rate by considering changes in macroscopic lattice quantities.
Figure 1 shows the Debye frequency dependence of the re-
The change in Debye frequency under compression can be
laxation time ␶ using ␻D 0
= 160 cm−1 and ␻0 = 657 cm−1, cor-
related to the change in the thermodynamic Grüneisen pa-
responding to the parameters for the NO2 bending mode in
rameter ␥ through the equation23,38
nitromethane. Results are shown using Eq. 共30兲 summed
over all m 共dotted line兲 and using the resonant approximation
in Eq. 共32兲 共solid line兲. The full rate expression shows small
oscillations due to different multiphonon channels becoming
␻D共v兲 = ␻D0 exp − 冋冕 v
␥ 共 v ⬘兲
v0 v ⬘
册
dv⬘ , 共34兲

dominant as the phonon gap ␻0 − ␻D changes. We note that at where ␻D 0

is the Debye frequency at ambient conditions, v0 is
the Chapman–Jouget detonation pressure in nitromethane, the initial specific volume v0 = 1 / ␳0, and v is the specific
the Debye frequency is only shifted by a factor of 1.32 in our volume following shock compression. A general relation for
calculations 共see below for further discussion兲. the compression dependence of ␥ for molecular crystals is
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
155.33.16.124 On: Fri, 28 Nov 2014 02:17:00
 014507-6 Joe Hooper J. Chem. Phys. 132, 014507 共2010兲

280 TABLE I. Parameters used for nitromethane in this study 共Ref. 59兲.

260 ␳ 1.13 g / cm3

Debye frequency (cm-1) A 1647 m/s
RDX b 1.5785
240
Specific heat cv 1736 J/共kg K兲
Grüneisen ␥ 0.68
220
Nitromethane
200 This relation, combined with the Rankine–Hugoniot re-
lations for a steady shock, allows us to write the pressure at
180 a Hugoniot state in terms of the compression ␩ as38

160 ␳ 0A 2␩
0 2 4 6 8 10 PH = . 共38兲
共1 − b␩兲2
Hugoniot pressure (GPa)

FIG. 2. The change in the Debye frequency along the shock Hugoniot for The parameters used for these two materials are listed in
nitromethane and cyclotrimethylene-trinitramine 共RDX兲. Tables I and II.
The shock will also increase the temperature in the ma-
terial; the rate for a multiphonon process will, in general,
still an outstanding question in shock physics. One option,
have a strong dependence on temperature due to terms of the
used previously by Zerilli and Toton,2 is to calculate a new
type n共␻ , T兲m. The final temperature of the shocked molecu-
Debye frequency based on an experimental knowledge of the
lar crystal after full equilibration of the phonon systems can
thermodynamic Grüneisen parameter along the entire shock
be calculated from macroscopic thermodynamic consider-
Hugoniot. However, such information is only available for a
ations. We begin with the Hugoniot relation
handful of materials.
We instead use the standard assumption in shock physics
eH − e0 = PH共v0 − v兲/2, 共39兲
that ␥ is independent of temperature and that its volume
dependence is given by the simple relation38 where eH is the specific internal energy at a Hugoniot point
␥ 0v described by a pressure PH and a specific volume v. e0 and
␥共v兲 = , 共35兲 v0 are the corresponding quantities for the unshocked mate-
v0
rial. Then
where ␥0 is the thermodynamic value at standard conditions,
␥0 = 3␣T共cs0兲2 / c p, where ␣T is the thermal expansion coeffi- 共v0 − v兲 PH
deH = dP − dv , 共40兲
cient, cs0 is the bulk sound speed at ambient conditions, and 2 2
c p is the specific heat capacity at constant pressure. Properly,
␥0 should be an average of the mode Grüneisen parameters and combining this with the fundamental thermodynamic re-
for the acoustic modes and lower lying lattice optical lation, we can write
phonons; for the current work we use the standard ␥0 aver-
aged over all modes, which is the only quantity widely avail- 共v0 − v兲 PH c v␥
TdS = dP + dv = cvdT + Tdv , 共41兲
able for molecular systems. If the pressure-induced fre- 2 2 v
quency shifts of the internal modes are minimal as we have
assumed, the thermodynamic ␥0 will provide an adequate where cv is the specific heat capacity of the material and ␥ is
representation of the behavior of the low-frequency modes. again the Grüneisen parameter. The final relation in Eq. 共41兲
The pressure dependence of the Debye frequency with this is a fundamental thermodynamic identity. At a specific
assumed form is then Hugoniot point H, we have

␻D共␩兲 = ␻D0 exp共␥0␩兲,

where ␩ = 1 − v / v0. For the systems of interest here the De-

共36兲
cv 冏 冏
dT
dv H
+
v
=
2
冏 冏
cvT␥ 共v0 − v兲 dP
dv H
+
PH
2
dv . 共42兲

bye frequency is generally on the order ␻D 0

Assuming that both ␥ / v and cv are constants and solving for
−1 13,29
= 150– 200 cm . temperature using an integrating factor, we have
The change in Debye frequency under shock compres-
sion is depicted in Fig. 2 for two molecular systems, ni-
TABLE II. Parameters used for RDX in this study 共Refs. 58, 60, and 61兲.
tromethane and cyclotrimethylene-trinitramine 共RDX兲. For
convenience we have plotted the data as a function of Hugo- ␳ 1.81 g / cm3
niot pressure, making the common assumption that there is a A 3110 m/s
linear relationship between the shock velocity US and the b 1.41
particle velocity behind the front u p,38 Specific heat cv 1077 J/共kg K兲
Grüneisen ␥ 1.103
Us = A + bu p . 共37兲
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
155.33.16.124 On: Fri, 28 Nov 2014 02:17:00
 014507-7 Vibrational energy J. Chem. Phys. 132, 014507 共2010兲

TH共v兲 = T0e␥0/v0共v0−v f 兲 + e共−␥0/v0兲v f ⫻ 冕v0

vf
e␥0v f /v0 f共v兲
cv
dv , ⌫共0兲 =
1
2␲cTl
, 共49兲

共43兲 where c is the speed of light and Tl is the phonon lifetime

due to three-phonon decay processes. Dephasing and upcon-
where version processes produce no linewidth at zero temperature.

f共v兲 =
共v0 − v兲 dP
2 dv
冏 冏 H
+
PH
2
dv . 共44兲
The temperature dependence of the population decay contri-
bution is
⌫d = ⌫共0兲关n共␻1,T兲 + n共␻2,T兲 + 1兴 共50兲
Again assuming a linear shock-velocity/particle-velocity
relationship and using Eq. 共38兲 to evaluate f共v兲, we have the for a three-phonon process involving decay into lower lying
final expression for the temperature at a Hugoniot point,39 modes ␻1 and ␻2. The decay contribution to the phonon

冕 ␩f
lifetime at nonzero temperature is then simply Ta
A2be␥0␩ ␩2 = 1 / 2␲c⌫d.
T H共 ␩ 兲 = T 0e ␥0␩ + e −␥0␩d ␩ , 共45兲
cv 0 共1 − b␩兲3 Experimental studies by Ye and co-workers40 and by
Fayer and co-workers41 have studied the Raman bandwidths
where ␩ f is the final compression after the shock, A and b are for some characteristic phonons and internal vibrations of
the Hugoniot coefficients in Eq. 共37兲, T0 is the initial tem- systems such as RDX and HMX. The low-temperature life-
perature, and cv is the specific heat capacity at constant vol- times due to three-phonon processes involving decay into
ume. two lower frequency phonons are generally on the order of
We assume, as is widely done in shock 2.5–5 ps, with the average around 4 ps for both systems.
physics,1,4,7,10,19,21 that the low-frequency lattice phonon Other molecular crystals such as anthracene and napthalene
modes initially accommodate all the thermal energy after the are observed to have similar lifetimes.42
passage of the shock. The heat balance equation for a given We are interested in the characteristic time of upconver-
molecule immediately after a shock is then sion at ambient conditions ␶共300 K兲, which can be estimated
from Ta using detailed balance as ␶ = Ta exp共␤ប␻0兲. This re-
冕 T0
Tp
CvpdT = 冕 TH共␩兲

T0
关Cvp + Cvv共T兲兴dT, 共46兲
sults in a characteristic upconversion time of close to 3 ps,
and for the large molecular systems considered here 共RDX
and ␣-HMX兲, we will assume that ␶共300 K兲 is exactly 3 ps
where T0, the initial temperature, is assumed to be 300 K in for all modes. We then use this value to fix the parameter ␣
all our calculations. If we assume that the lattice phonon heat for each individual mode that energy is transferred into.
capacity can be treated using the classical values, the initial Nitromethane represents an interesting case whose prop-
phonon temperature is erties are closest to our theoretical assumptions; it also has
the advantage of being the most widely studied system in
T p共 ␩ 兲 = T 0 + 冕 TH共␩兲

T0
Cvv共T兲 + Cvp
Cvp共T兲
dT, 共47兲
terms of experimental spectroscopy. The phonon gap is large,
as would be expected for a simpler molecule. Generally a
four-phonon or higher process is required for activating those
where TH is given by Eq. 共45兲. lower-lying vibrational modes which experiment indicates
The relaxation time, taking into account the Hugoniot interact directly with the bath of phonons.15,30 Opposing this
pressure dependence and the dependence on the time- is a smaller internal heat capacity, thus requiring less overall
dependent lattice phonon temperature, is energy transfer to achieve full thermal equilibrium. Chen and
co-workers14 previously studied the upconversion process
␶−1 =
4 ␻ D共 ␩ 兲
␬
冑 5 关␣en共␻0/␬,T p共␩兲兲/␬兴␬
3 n共␻0,T p共␩兲兲关n共␻0,T p共␩兲兲 + 1兴
, 共48兲
following a rapid 40 K jump in the lattice temperature using
a near-IR heating pulse. No shock was produced by the heat-
ing. Picosecond Raman spectroscopy was simultaneously
where ␬ = 4␻0 / 3␻D共␩兲. used to monitor the temperature of the internal modes based
Using this Hugoniot dependent rate, we can now solve on their anti-Stokes scattering intensities. The NO2 symmet-
for the time required for thermal equilibrium using Eqs. 共5兲 ric bending mode at 657 cm−1 came to full equilibrium with
and 共6兲. To fix the parameter ␣, we use the experimental time the heated bath in approximately 150 ps.
constant for multiphonon upconversion ␶ taken at ambient As a comparison to this experimental work, a calculation
conditions. For nitromethane, we can use the direct up- of a rapid temperature jump in nitromethane using Eqs. 共5兲
pumping study of Chen and co-workers,14 which reveals a and 共6兲 is shown in Fig. 3. The initial phonon temperature is
time constant of approximately 40 ps for the 657 cm−1 vi- set at 340 K, identical to the experimental heat pulse. The
bration. The other material parameters used for nitromethane heat pulse was shown to result in only a mild pressure in-
are given in Table I. crease 共on the order of 50 MPa兲, which has a negligible
For the larger molecular systems considered here, no effect on the rate. The characteristic transfer time is thus
such direct information exists and thus we estimate ␶ in the solely dependent on the temperature variation in the phonon
following way. The Raman bandwidth at zero temperature, occupation terms. Our calculation shows that thermal equi-
⌫共0兲, can be written as40 librium is essentially complete in 175 ps, in good agreement
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
155.33.16.124 On: Fri, 28 Nov 2014 02:17:00
 014507-8 Joe Hooper
340
Lattice phonons
335 Vibrations
330
Temperature (K)
325
320
315
Nitromethane
310 -1
ω0 = 657 cm
305
300
0 50 100 150 200
Time (ps)
FIG. 3. Equilibration of the 657 cm−1 mode in nitromethane after a small
temperature and pressure jump similar to that observed by Chen and co-
workers 共Ref. 14兲.
with the experimental result. We note that all multiphonon
rate theories will have a general Tm-type dependence, so this
agreement with experiment is more a calibration rather than
a true test of the theory.
We note that our choice of a Debye distribution implic-
itly assumes a solid-state system, whereas typically shock
Hugoniot data for nitromethane are for the liquid form. A
liquid phase could be treated more explicitly via a phonon
DOS expression describing instantaneous normal modes
with a gradual decline in the DOS at higher frequencies.
However, as our primary interest is to treat anharmonic in-
teractions in molecular solids and maintain simple analytic
expressions, a more detailed consideration of liquid-type
DOS forms is saved for later investigation and we retain the
Debye form.
A more interesting calculation for nitromethane is to ex-
amine how the thermal equilibration process changes under
shock compression rather than a simple temperature jump. At
this point we must move beyond what is available in terms of
existing experiments, and going forward our main compari-
son will be to large-scale molecular dynamics simulations.
We now consider the thermal equilibration of the system at
higher shock Hugoniot pressures via energy transfer into the
480, 607, and 657 cm−1 modes. Recent experimental work
examining vibrational relaxation in nitromethane has sug-
gested that all these lowest-lying vibrational modes interact
directly with the bath system.30 The characteristic time ␶ at
ambient conditions is assumed to be 40 ps for all modes.
The internal vibrational system of nitromethane consists
of 14 modes, whose frequencies we have assumed to be
those observed experimentally by Deak et al.15 The Debye
frequency at ambient conditions is 160 cm−1. The lattice
phonon system consists of the translations and rotations of a
single molecule, plus the low-frequency methyl rotor
mode.15 In Fig. 4 we show the time required for thermal
equilibration of these lattice and vibrational systems in the
range of shock pressures between 0 and 20 GPa, which is
approximately the magnitude of the von Neumann spike
pressure in an established nitromethane detonation wave 共the
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
155.33.16.124 On: Fri, 28 Nov 2014 02:17:00
J. Chem. Phys. 132, 014507 共2010兲
40
35
30
Equilibration time (ps)
Nitromethane
25
20
15
10
5
0
0 5 10 15 20
Hugoniot pressure (GPa)
FIG. 4. Equilibration of shocked nitromethane as a function of Hugoniot
pressure up to the von Neumann detonation spike pressure.
Chapman–Jouget pressure at the rear of the reaction zone is
approximately 12.5 GPa兲.43 Equilibration is assumed to be
complete when there is less than a 0.5% difference in the
temperatures of the two systems.
At very low shock pressures, there is only a small tem-
perature jump and a negligible change in the lattice Debye
frequency. This results in a very small range where the
equilibration time increases slightly with pressure, although
this is partly an artifact of our method of assigning full
equilibration. Beyond this small range, however, the com-
bined effects of a significant blueshift in ␻D and the shock-
induced temperature rise result in an approximately, but not
exactly, exponential decrease in the time required for thermal
equilibrium to be established.
Our theory indicates than even on the rapid time scales
of a detonation front the system will be in thermal equilib-
rium very rapidly, on the order of a picosecond at high shock
pressures. Thus, as discussed above, we anticipate that the
multiphonon equilibration processes discussed in this paper
are complete prior to any later thermal or chemical nonequi-
librium conditions that will develop in the reaction zone of
an explosive.6 We note also that for nitromethane 共generally
studied in its liquid form in shock experiments兲, a
nanosecond-scale induction time is observed between the
leading shock in the cellular detonation front and the onset of
significant exothermic chemical reaction.44,43 This induction
time in homogeneous gaseous and liquid explosives, during
which slow chemical decomposition processes are taking
place, occurs after the shock Hugoniot temperature is estab-
lished and is also unrelated to the solid-state phenomenon
considered here.
We next consider systems with much larger polyatomic
molecules, which are representative of important materials in
the areas of explosives, shock-induced chemistry, and safety
concerns. We begin with cyclotrimethylene-trinitramine
共commonly referred to as RDX兲, a common high explosive
that requires significant energy input to initiate chemical re-
action. Recent large-scale classical molecular dynamics cal-
culations by Nomura et al.18 tracked the rotational and vibra-
tional kinetic energies of solid RDX during a simulated
 014507-9 Vibrational energy
800
Lattice phonons
750 Vibrations
700
650 RDX
Temperature (K)
PH = 8.0 GPa
600
550
500
450
400
350
300
0 1 2 3 4 5
Time (ps)
FIG. 5. Equilibration of RDX behind an 8.0 GPa shock. The calculated
equilibration time of 3.5 ps is consistent with the value of ⬃2 ps suggested
by the simulations of Nomura and co-workers 共Ref. 18兲.
shock. At a shock pressure of approximately 8.0 GPa, the
rotational and vibrational temperatures estimated from the
energies level out at equilibrium values of approximately 2
ps after the arrival of the shock wave.
For our calculation, the Debye frequency is set to ␻D
= 200 cm−1, and we consider the transfer into those modes
between ␻D and 2␻D. The energy transfer proceeds through
six modes in this range at 204.0, 307.9, 351.5, 351.8, 379.7,
and 381.1 cm−1. All these modes are in the range of two-
phonon processes. The internal vibrations of RDX are taken
from a standard ab initio isolated molecule calculation using
the B3LYP functional with a 6-31Gⴱⴱ basis set in the pack-
age GAUSSIAN03. We note that we have also included all
modes with frequencies below 200 cm−1 as part of the lattice
system and are factored into its heat capacity. In the solid,
such frequencies generally correspond to lattice optical
phonons involving complex motions involving, for example,
the entire cyclic ring.45 The inclusion of these modes results
in a larger specific heat than that considered in previous
work, and thus a lower initial lattice temperature. The param-
eters used for RDX are given in Table II.
In Fig. 5 we show the results of our theory for the energy
transfer into RDX at the same shock pressure 共8.0 GPa兲 as
presented by Nomura et al. We calculate that full thermal
equilibrium will be reached at approximately 3.5 ps, consis-
tent with the molecular dynamics values.
We next consider the ␣ phase of HMX for direct com-
parison to recent nonreactive classical molecular dynamics
simulations by Jaramillo et al.,17 which explicitly considered
time scales for thermal equilibration behind a shock wave.
Separating the lattice and rovibrational systems using the
method of Strachan and Holian,16 local spatially resolved
temperatures were estimated from the kinetic energy of the
molecules.
The phonon and internal frequencies for our calculation
are determined from a full solid-state calculation of ␣-HMX
using the planewave code ABINIT.46 Norm-conserving
Troullier–Martins47 pseudopotentials are used and the calcu-
lation employs the Perdew–Burke–Ernzerhof functional48
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
155.33.16.124 On: Fri, 28 Nov 2014 02:17:00
J. Chem. Phys. 132, 014507 共2010兲
TABLE III. Parameters used for ␣-HMX in this study 共Refs. 17, 60, and
61兲.
␳ 1.81 g / cm3
A 4000 m/s
b 1.25
Specific heat cv 1100 J/共kg K兲
Grüneisen ␥ 1.10
with a planewave cutoff of 80 Ry. The atomic coordinates are
optimized until the maximum forces between atoms are less
than 5.0⫻ 10−5 eV/ Å. Frequencies used are taken at the ⌫
point in the Brillouin zone. The solution of the coupled en-
ergy transfer equations 关Eqs. 共5兲 and 共6兲兴 for ␣-HMX thus
represents multiphonon transfer in the entire solid-state unit
cell. We sum over 16 calculated modes between ␻D and 2␻D
using the material and thermodynamic values given in Table
III. The variation in equilibration times with Hugoniot pres-
sure is shown in Fig. 6, along with similar data for RDX.
The Hugoniot dependence of the RDX and ␣-HMX
equilibration times shows a similar trend as that seen in ni-
tromethane. At low pressures there is a range where the in-
creasing temperature due to the shock results in slightly in-
creasing equilibration times. At higher pressures, however,
the temperature jump and shift in Debye frequency dominate
and the systems approach roughly constant equilibration
times of approximately 3.1 ps for RDX and 3.7 ps for
␣-HMX. Jaramillo et al. observe full thermal equilibrium on
the order of 3–5 ps for the leading elastic waves in shocks in
the range of 3–5 GPa. They also consider a high-pressure
overdriven plastic shock close to 26 GPa, for which equili-
bration occurs in approximately 4 ps, consistent with the
asymptotic limit we observe here. We note that these simu-
lations do not project the molecular motion onto the normal
modes of the crystal; rather, they simply track local and cen-
ter of mass kinetic energies to assign lattice and internal
temperatures. Overall the agreement between our theory and
these large-scale classical simulations is quite good, suggest-
ing that the simulations can be generally understood in terms
of equilibration via multiphonon scattering processes. Our
8
7
Equilibration time (ps)
6
α-HMX
5
4
RDX
3
2
0 5 10 15 20
Hugoniot pressure (GPa)
FIG. 6. Equilibration of RDX and ␣-HMX as a function of Hugoniot
pressure.
 014507-10 Joe Hooper
theory predicts an overall equilibration time on the order of
picoseconds for shocks at or below the detonation pressure in
large polyatomic explosive molecular crystals such as RDX
or HMX.
V. RATE ENHANCEMENT NEAR A DEFECT
We next consider a simple model for defects to examine
if they might have any meaningful effect on the thermal
equilibration of the system or initiation of chemical reaction.
It has long been known that inhomogeneities play a critical
role in shock-induced chemistry by producing regions of lo-
calized high temperatures, often referred to as “hot spots.” In
solid molecular explosives, chemical decomposition gener-
ally requires such defects to produce temperature localization
and begin reaction.44,38 Shock insult at pressures above the
Hugoniot elastic limit of molecular crystals will result in a
plastic wave in solid molecular crystals; such waves will
induce inelastic mechanical deformation and defects even in
an otherwise ideal crystal.38 The simple treatment of plastic
defects that follow allows us to consider some aspects of
intrinsic inhomogeneities in the material as well as those
arising from plastic compression waves.
There have been a number of proposals that nonequilib-
rium phonon states near a defect might give rise to locally
high temperatures leading to chemical reaction.1,3,4,10,20 It
was hypothesized by Coffey and Toton1 and later by Dlott
and Fayer4 that, where a region near some defect to have an
increased rate of energy transfer into internal modes, there
might be a significant local temperature overshoot or an in-
creased rate of bond dissociation in this spatially localized
region.4 Here our goal is to develop a physically based model
for a characteristic defect and investigate this concept within
the framework of the energy transfer equations developed
above. Specifically, we consider the case of immobile
straight dislocations, which might be seeded by a plastic
wave or present in a crystal due to the crystallization process.
A detailed description of lattice and phonon band structure
changes in the small region near a dislocation core is beyond
the scope of this paper, but it is instructive to pursue a simple
analytical model for the mesoscale region around a disloca-
tion and consider its role in the initial establishment of ther-
mal equilibrium behind a shock front.
We assume that we have a material which has been com-
pressed by an ideal shock wave. We now superimpose the
strain field due to an ideal straight dislocation in the material.
The region is assumed to be small compared with the bulk
material, such that the only effect of the elastic strain field is
a shift in the lattice phonon frequencies.23 The change in
sound speed induced by the dislocation strain field is as-
sumed to be23,49
c␧共r, ␩兲 = ␮cs共␩兲,
where c␧ is the modified sound velocity, cs is the bulk sound
speed in the defect-free shocked material, and ␮ is given by
the relation
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
155.33.16.124 On: Fri, 28 Nov 2014 02:17:00
J. Chem. Phys. 132, 014507 共2010兲
冋
␮ = 1 − ␥共␩兲 兺 ␧ij共r兲 ,
ij
册 共52兲
where ␥ is the Grüneisen parameter and ␧ is the Cauchy
strain tensor arising from the dislocation at the point r.
We assume that for these small strains, the change in
Debye frequency which characterizes the lattice phonon sys-
tem is similar to the strain-dependent change in the long
wavelength acoustic phonon frequencies. That is, if ␻D共␧兲
= c␧qD共␧兲, where qD is the Debye radius,23 the overall change
in ␻D near the dislocation is then
␮ ␻ D共 ␩ 兲
␻D共␩,␧兲 = . 共53兲
共1 + Tr ␧兲1/3
If the strain were a pure dilatation, this change in sound
velocity and Debye frequency is the same that would arise in
Grüneisen’s law of thermal expansion.23,28 It is, in that case,
essentially equivalent to adding a further perturbation which
shifts the frequencies of the phonon states by a factor of 共1
− ␥ Tr ␧兲.49 In general there will be dilatational and shear
components of the strain around most inhomogeneities of
interest. Thus we have used a simplistic summation over the
full strain tensor, similar to the approach used by Klemens49
and Ziman23 in classic works discussing the scattering of
phonons off elastic strain fields with off-diagonal compo-
nents. For an edge dislocation, the term in the denominator
represents the change in qD due to the additional dilatational
strain arising from the dislocation.
We also ignore changes in the internal frequency ␻0; this
is likely invalid within the small dislocation core region, but
we assume that in the elastic strain field outside the core the
internal frequencies will not be significantly altered. We as-
sume that the small dislocation region has no effect on the
overall heat capacity of the lattice system and that the initial
lattice temperature is identical to that in the defect-free ma-
terial.
We consider a two-dimensional slice through a disloca-
tion line in RDX, with the sense of both screw and edge
dislocations directed out of the page.50 The Poisson’s ratio
for RDX is taken to be 0.3.51 The Burgers vector b for both
screw and edge dislocations is assumed to be 1.32 nm based
on experimental indentation work,52 and a region of diameter
2b around the dislocation core is ignored due to nonlinear
effects.50
The strain equations for an ideal, sessile screw disloca-
tion are50
␧xx = ␧yy = ␧zz = ␧xy = 0, 共54兲
b y
␧xz = ␧zx = − , 共55兲
4␲ x + y 2
2
where b is the Burgers vector,
共51兲
b x
␧yz = ␧zy = . 共56兲
4␲ x + y 2
2
The relevant strain tensor for an edge dislocation is50
 014507-11 Vibrational energy J. Chem. Phys. 132, 014507 共2010兲

400
Lattice phonons
390 Vibrations
Defect
380
370

FIG. 7. Scattering rate enhancement near a screw dislocation as a multiple
of the rate relative to that in a defect-free shocked material. The circular area
at the origin represents the dislocation core and is ignored due to nonlinear
effects.
− b y共1 − 2␯兲
Tr ␧ = ␧xx + ␧yy = , 共57兲
2␲ 共1 − ␯兲共x2 + y 2兲
where ␯ is Poisson’s ratio. The off-diagonal components are
b x共x2 − y 2兲
␧xy = ␧yx = . 共58兲
4␲ 共1 − ␯兲共x2 + y 2兲2
The enhancement factor of the scattering rate ␶−1 arising
from these dislocation strain fields is shown in Fig. 7 for a
screw dislocation and in Fig. 8 for an edge dislocation. A
value of unity corresponds to a rate of energy transfer ␶−1
identical to that of the defect-free material. For both species
of dislocations, there is a region on the order of 5 nm in
which the rate is enhanced by a factor of 1.5 or greater. We
note that there is also a region in the two-dimensional slice in
which the rate is reduced due to larger positive strain values
in Eq. 共52兲. As one approaches the center of the dislocation
core, there is a divergent behavior due to the failure of the
simple linear elastic description of the dislocation; for plot-
ting purposes we have capped this small divergent region at
a rate enhancement factor of 3. The circular region around
the origin corresponds to the ignored area very close to the
core.
10 3 A variety of factors have not been considered in this
2.5
5 ond time scales of thermal equilibrium. We have also ignored
2 changes in the internal phonon system due to the presence of
y (nm)
Temperature (K)
360
350
340
330
320
310
300
0 1 2 3 4 5 6 7 8
Time (ps)
FIG. 9. Temperature equilibration in RDX shocked at 2.5 GPa, approxi-
mately the shock pressure at which sustained chemical reaction will occur in
granular mixtures of this material. The temperature rise in a small area near
a dislocation in which the rate is double that of the bulk of the solid is
plotted.
Further from the dislocation core, where linear elasticity
provides a good description of the strain, there is a modest
enhancement of the rate of a similar value as that hypoth-
esized by Dlott and Fayer.4,7 While our simple model, thus,
does suggest a physical basis for such an enhancement near
defects, in light of the rapid establishment of thermal equi-
librium, this region would not result in a significant over-
shoot of local temperature for pressures within the shock to
detonation transition regime.
In Fig. 9 we show the thermal equilibration process for
RDX shocked at a value of 2.5 GPa, which is generally close
to the threshold at which the shock will transition to a deto-
nation in common experimental mixtures. We additionally
consider a region close to a dislocation in which the rate is
enhanced by a factor of 2, but whose size is so small it has
no effect on the equilibration properties of the main lattice
and internal systems. An additional equation similar to Eq.
共6兲 is solved for the local temperature near the defect. As
expected, the enhanced region only briefly reaches a slightly
higher temperature than the main internal phonon system,
eventually reaching equilibrium on a similar time scale as the
main internal system.
simple analysis. We have ignored any motion of the disloca-
tion line, assuming that it is essentially static on the picosec-
the dislocation; in reality, local vibrational modes may form

0 1.5 around the inhomogeneity, further increasing the rate of

transfer. However, given the rapid establishment of thermal
1
equilibrium in the ideal system, the overall rate enhancement
-5
would have to be substantial to observe a significant tem-
0.5
perature overshoot leading to chemical reactions at pressures
-10 0 of 0–5 GPa. Such pressures are representative of the regime
-10 -5 0 5 10 where a shock will ultimately make a transition to a self-
x (nm) sustained detonation in materials such as RDX and ␣-HMX.
We note that even an enhancement factor of 20 for the shock
FIG. 8. Scattering rate enhancement near an edge dislocation. parameters used in Fig. 9 would only result in a maximum
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
155.33.16.124 On: Fri, 28 Nov 2014 02:17:00
 014507-12 Joe Hooper
temperature jump of about 80 K, far below temperatures
needed to begin chemistry. While this process likely has no
bearing on the shock to detonation transition in explosive
molecular crystals, at much higher shock pressures it may
result in local temperature enhancements; further discussion
on this point is given below.
VI. DISCUSSION
A key result of our study is that the time required for
typical molecular crystals to be at full thermal equilibrium is
expected to be on the order of picoseconds, rather than hun-
dreds of picoseconds3,20,6,4 or nanoseconds2 as previously
suggested. Even nitromethane, with its relatively large pho-
non gap, is expected to equilibrate rapidly under shock com-
pression 共see Fig. 4兲. Recent large-scale molecular dynamics
simulations are in good agreement with these time scales, as
discussed above.
We now consider some implications of this more rapid
equilibration with regard to shock-induced chemistry in mo-
lecular crystals. We first note that the onset of chemistry at
lower shock pressures is due almost entirely to crystalline
defects, localized material failure, friction between crystal-
line grains, and so on. For granular, heterogeneous mixtures
of molecular crystals 共the most widely studied case experi-
mentally兲 these effects will generally overwhelm any intrin-
sic chemical initiation induced by nonequilibrium phonon
states at the shock front. Thus the cases of most relevance to
our theory are those of homogeneous explosives, ideal solid
molecular crystals 共such as studied by single crystal shock
experiments or simulated by molecular dynamics兲, and the
high pressure or detonation limit. In these regimes nonequi-
librium phonon states have been postulated to play an impor-
tant or even the primary role in shock-induced molecular
decomposition.
We first consider the model proposed in a number of
works by Dremin, Klimenko, and co-workers.3,19–21 The
shock energy is assumed to be entirely accommodated by the
acoustic modes and possibly the rotational modes, whose
effective temperatures can reach as high as 1–3 eV. We have
made a similar approximation in this work, and such a par-
titioning of temperature is routinely observed in nonequilib-
rium molecular dynamics simulations.6,17 The large thermal
kinetic energy in these modes is then assumed to lead to
collision-type interactions directly behind or within the
shock front itself, resulting in bimolecular energy transfer
and even ionization. High energy ions, radicals, and elec-
trons produced by these collisions then lead to very efficient
decomposition of the remaining neutral molecules in the
solid. For detonating materials, these ion/neutral collisions
are proposed to be the primary mechanism for bringing the
system from its initial shock-induced spike pressure to the
Chapman–Jouget point, where chemical decomposition is
complete and the system is in thermal and chemical
equilibrium.44
First, we note that these theories generally did not con-
sider the large number of low-lying lattice optical phonons
present in systems with large polyatomic molecules. The
very rapid energy conversion between these modes following
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
155.33.16.124 On: Fri, 28 Nov 2014 02:17:00
J. Chem. Phys. 132, 014507 共2010兲
3000
Lattice phonons
Vibrations
2500 Defect
2000
Temperature (K)
1500
1000
500
0
0 1 2 3 4 5
Time (ps)
FIG. 10. Temperature enhancement near a dislocation in RDX shocked to
20.0 GPa with a hypothetical strain field increasing the energy transfer rate
by a factor of 5.
the shock will result in a much lower effective temperature
for the lattice modes 共see Fig. 5兲. Second, our theoretical
results indicate that the intramolecular multiphonon pathway
for deexcitation of the initially populated lattice modes is
very efficient. Within 1 ps, the kinetic energy of these excited
modes would be considerably reduced from the peak phonon
temperature by multiphonon processes alone 共see Fig. 10兲, a
process which is rapid even on the time scales of the shock
front. Overall our work suggests that the equilibration pro-
cess is likely too efficient for such thermally excited colli-
sions suggested in previous work to be significant.
The second proposed theory has already been discussed
in Sec. V, namely, that small regions near inhomogeneities
have enhanced energy coupling and thus acquire a higher
localized temperature as the phonons come into equilibrium
with the vibrational system. We have considered a simple
model for defect enhancement in molecular crystals, which
does suggest such an enhancement may be present near
simple inhomogeneities such as straight dislocations. At low
or moderate pressures, such a process likely plays no role in
the initial onset of chemical reaction 共see Fig. 9兲, as the rate
enhancement required to produce a large temperature over-
shoot would be extremely large. Dislocations may of course
play other roles in temperature localization, which are unre-
lated to the issue of nonequilibrium phonon states.53
At high pressures the temperature overshoot near a de-
fect is more dramatic, but still greatly limited by the rapid
equilibration. RDX shocked at 20 GPa results in a final sys-
tem temperature of approximately 1070 K; as seen in Fig.
10, a region whose multiphonon rate is increased by a factor
of 5 due to strain reaches a temperature of 1620 K, but still
decays to the final equilibrium temperature in approximately
2.5 ps. In the limit of a steady detonation wave where lead-
ing pressures at the front might be as high as 60 GPa, it is
possible that this mechanism may contribute to temperature
localization or faster kinetics than would be seen by purely
thermal excitation. This would depend on system-specific de-
tails regarding the initial population of defects or the time
scales on which a plastic shock could generate inelastic de-
 014507-13 Vibrational energy
formations. Ultrafast vibrational spectroscopy of overdriven
plastic shock waves in molecular crystals may offer a direct
experimental method to probe this possible early-time tem-
perature localization.
A number of proposals have discussed the possibility of
electronic excitations at the shock front, although there does
not appear to be a consensus as to the physical mechanism
supplying the excitation energy. Vertical electronic excitation
by the excited lattice phonon states, as suggested in a sepa-
rate work by Dremin,3 would require an ⬃30 phonon process
even if the highest occupied molecular orbital-lowest unoc-
cupied molecular orbital gap in a molecule was lowered con-
siderably to 1 eV by the shock or by shock-induced plastic
flow. The rapid transfer of energy from the excited phonon
states, which shares energy over all modes, would dominate
the remote possibility of a thermally induced electronic ex-
citation. This rapid dissipation would also mitigate the pos-
sibility of collision-induced ionization, as discussed above.
The conductivity rise commonly observed in explosives is
more likely simply associated with species produced at later
times after the onset of significant chemical decomposition,
especially solid carbon.54
More recently, some reactive classical molecular dynam-
ics simulations have suggested that within the extremely nar-
row region of the shock front itself, molecular collisions take
place which are sufficient to induce bimolecular reactions
and decomposition.22 This suggestion differs from that of
Dremin and co-workers21 in that it is purely a collisional
interaction, in which the excited thermal state of the mol-
ecule plays no part. Landerville et al.22 observed decompo-
sition in quantum molecular dynamics simulations for two
RDX molecules colliding at relative velocities between 4 and
6 km/s, although further investigation is needed to determine
how decomposition barriers might change in a solid matrix
of RDX. A purely mechanical interaction of this nature is
outside the scope of this work, but if such collisional decom-
position does in fact occur, our theory suggests that it must
indeed be essentially independent of thermal excitation.
We also note that Fried and Ruggerio5 suggested a cor-
relation between the quantum mechanical rate of two-phonon
scattering processes into an internal vibration at 425 cm−1 in
energetic molecular crystals and the drop hammer sensitivity
test for chemical reaction. This correlation has recently been
expanded upon by Ye and Koshi.55 Qualitatively, a higher
phonon scattering rate was correlated with a lower impact
threshold for chemical reaction 共that is, a more sensitive ma-
terial兲. We first note that the drop weight test is not a true
measure of shock-induced chemistry; normally such tests in-
volve very small pressures and chemical reaction commonly
occurs in small regions of significant plastic flow or me-
chanical failure in the sample.56 Frequently this is in areas of
intense shear strain such as at the edge of the impactor. Cor-
relations with more precise tests of shock-induced chemistry
such as the large-scale gap test57 are limited; in fact, the gap
test data may be inversely correlated with a drop hammer
impact test in some cases.58 We also note that materials with
a higher phonon/internal mode scattering rate would gener-
ally be expected to have lower thermal conductivity; this
would affect properties such as adiabatic shear band or crack
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
155.33.16.124 On: Fri, 28 Nov 2014 02:17:00
J. Chem. Phys. 132, 014507 共2010兲
formation as well as the stability of defect-related hot spots.
The correlation observed in previous work may have more to
do with these properties rather than intrinsic issues of shock-
induced reactions. This correlation is intriguing, but for these
reasons we have not attempted to draw a similar comparison
between our theory of shock-induced vibrational energy
transfer and drop hammer impact testing.
VII. CONCLUSIONS
In summary, we have developed a new theory of thermal
equilibration via multiphonon energy transfer following
shock compression of a molecular crystal. Our results sug-
gest that the process of this energy transfer is very rapid, on
the order of 1–5 ps for solids such as RDX and ␣-HMX.
Recent classical molecular dynamics simulations of shocks
are in good agreement with these time scales, suggesting that
their calculated results can be interpreted in terms of mul-
tiphonon scattering processes. The rapid time scales on
which the system thermalizes do not completely exclude
collision-type mechanisms for chemical decomposition
within a shock front, but do suggest that nonequilibrium ther-
mal excitation will not significantly assist any collisional en-
ergy transfer. Using a simple physical model for straight dis-
locations, we find that such defects may result in some
temperature localization purely due to nonequilibrium pho-
non states, but that the localization is minimal except at high
shock pressures and the duration of the enhancement is on
the order of 2 ps or less. While our theory suggests that
multiphonon scattering of thermal energy is thus unlikely to
play a role in determining low-pressure shock sensitivity of
energetic materials, at high pressures it may be possible to
produce local temperature enhancements due purely to non-
equilibrium processes. Ultrafast spectroscopy of molecular
crystals, especially using laser-driven shocks, is greatly de-
sirable for further clarification of the early physical processes
behind large amplitude shock waves.
ACKNOWLEDGMENTS
The author would like to acknowledge Jerry Forbes for
many useful discussions on shock and detonation physics,
Frank Zerilli for calculations on ␣-HMX and useful discus-
sion, and Harold Sandusky for insight into shock sensitivity.
This work was supported by the Office of Naval Research
funded Indian Head Division In-House Laboratory Indepen-
dent Research program under the direction of Dr. Alfred G.
Stern.
1
C. S. Coffey and E. T. Toton, J. Chem. Phys. 76, 949 共1982兲.
2
F. J. Zerilli and E. T. Toton, Phys. Rev. B 29, 5891 共1984兲.
3
A. N. Dremin, Towards Detonation Theory 共Springer-Verlag, Berlin,
1999兲.
4
D. D. Dlott and M. D. Fayer, J. Chem. Phys. 92, 3798 共1990兲.
5
L. E. Fried and A. J. Ruggerio, J. Phys. Chem. 98, 9786 共1994兲.
6
C. M. Tarver, L. E. Fried, A. J. Ruggerio, and D. F. Calef, in Proceedings
of the Tenth Symposium 共International兲 on Detonation, 1993 共Office of
Naval Research, Arlington兲, p. 3.
7
A. Tokmakoff, M. D. Fayer, and D. D. Dlott, J. Phys. Chem. 97, 1901
共1993兲.
8
C. A. Bolme, S. D. McGrane, D. S. Moore, and D. J. Funk, J. Appl. Phys.
102, 033513 共2007兲.
9
D. S. Moore, D. J. Funk, J. H. Reho, G. L. Fisher, S. D. McGrance, R. L.
 014507-14 Joe Hooper J. Chem. Phys. 132, 014507 共2010兲

Rabie, and K. T. Gahagan, AIP Conf. Proc. 620, 1333 共2002兲. 40

S. Ye, K. Tonokura, and M. Koshi, Chem. Phys. 293, 1 共2003兲.
10
H. Eyring, Science 199, 740 共1978兲. 41
C. M. Aubuchon, K. D. Rector, W. Holmes, and M. D. Fayer, Chem.
11
D. F. Strenzwilk and P. D. Yedinak, Phys. Rev. B 14, 4704 共1976兲. Phys. Lett. 299, 84 共1999兲.
12
F. J. Owens, Theor. Chim. Acta 55, 319 共1980兲. 42
J. R. Hill, E. L. Chronister, T. C. Chang, H. Kim, J. C. Postlewaite, and
13
D. D. Dlott, Annu. Rev. Phys. Chem. 37, 157 共1986兲. D. D. Dlott, J. Chem. Phys. 88, 949 共1988兲.
14
S. Chen, W. A. Tolbert, and D. D. Dlott, J. Phys. Chem. 98, 7759 共1994兲. 43
S. Sheffield, R. Engelke, and R. R. Alcon, in Proceedings of the Ninth
15
J.C. Deak, L.K. Iwaki, and D.D. Dlott, J. Phys. Chem. A 103, 971 International Detonation Symposium, 1989 共Office of Naval Research,
共1999兲. Arlington兲, p. 39.
16
A. Strachan and B. L. Holian, Phys. Rev. Lett. 94, 014301 共2005兲. 44
W. Fickett and W. C. Davis, Detonation 共University of California Press,
17
E. Jaramillo, T. D. Sewell, and A. Strachan, Phys. Rev. B 76, 064112 Berkeley, 1979兲.
共2007兲. 45
J. Hooper, E. Mitchell, C. Konek, and J. Wilkinson, Chem. Phys. Lett.
18
K. Nomura, R. K. Kalia, A. Nakano, P. Vashishta, A. van Duin, and W. A. 467, 309 共2009兲.
Goddard, Phys. Rev. Lett. 99, 148303 共2007兲. 46
X. Gonze, J.-M. Beuken, R. Caracas, F. Detraux, M. Fuchs, G.-M. Rig-
19
A. N. Dremin and K. K. Shedov, in Proceedings of the Sixth Symposium nanese, L. Sindic, M. Verstraete, G. Zerah, F. Jollet, M. Torrent, A. Roy,
on Detonation, 1976 共Office of Naval Research, Arlington兲, p. 29. M. Mikami, Ph. Ghosez, J.-Y. Raty, and D. C. Allan, Comput. Mater. Sci.
20
V. Y. Klimenko and A. N. Dremin, Sandia National Laboratories Techni- 25, 478 共2002兲.
47
cal Report No. RS 3180, 1981. N. Troullier and J. L. Martins, Phys. Rev. B 43, 1993 共1991兲.
21 48
A. N. Dremin, V. Klimenko, O. N. Davidova, and T. A. Zoludeva, in J. Perdew, K. Burke, and M. Ernzerhof, Phys. Rev. Lett. 77, 3865 共1996兲.
Proceedings of the Ninth Symposium 共International兲 on Detonation, 1989 49
P. G. Klemens, Proc. Phys. Soc., London, Sect. A 68, 1113 共1955兲.
共Office of Naval Research, Arlington兲, p. 725. 50
J. P. Hirth and J. Lothe, Theory of Dislocations 共Krieger, Malabar,
22
A. Landerville, I. I. Oleynik, M. A. Kozhushner, and C. T. White, AIP Florida, 1992兲.
Conf. Proc. 955, 447 共2007兲. 51
T. D. Sewell and C. M. Bennett, J. Appl. Phys. 88, 88 共2000兲.
23
J. M. Ziman, Electrons and Phonons 共Oxford University Press, New 52
H. G. Gallagher, P. J. Halfpenny, J. C. Miller, and J. N. Sherwood, Phi-
York, 1967兲. los. Trans. R. Soc. London, Ser. A 339, 293 共1992兲.
24
A. Nitzan, S. Mukamel, and J. Jortner, J. Chem. Phys. 60, 3929 共1974兲. 53
R. W. Armstrong, C. S. Coffey, and W. L. Elban, Acta Metall. 30, 2111
25
P. B. Allen, Phys. Rev. Lett. 59, 1460 共1987兲. 共1982兲.
26
T. Q. Qiu and C. L. Tien, ASME Trans. J. Heat Transfer 115, 835 共1993兲. 54
B. Hayes, in Proceedings of the Fourth Symposium 共International兲 on
27
L. Jiang and H. Tsai, ASME Trans. J. Heat Transfer 127, 1167 共2005兲. Detonation, 1965 共Office of Naval Research, Arlington兲, p. 595.
28
G. P. Srivastava, Physics of Phonons 共Taylor & Francis, London, 1990兲. 55
S. Ye and M. Koshi, J. Phys. Chem. B 110, 18515 共2006兲.
29
K. L. McNesby and C. S. Coffey, J. Phys. Chem. B 101, 3097 共1997兲. 56
C. S. Coffey and V. F. D. Vost, Propellants, Explos., Pyrotech. 20, 105
30
S. Shigeto, Y. Pang, Y. Fang, and D. Dlott, J. Phys. Chem. B 112, 232 共1995兲.
共2008兲. 57
D. Price, A. R. Clairmont, and J. O. Erkman, Naval Ordnance Laboratory
31
A. Nitzan and J. Jortner, Mol. Phys. 25, 713 共1973兲. Technical Report No. NOLTR 74-40, 1974.
32
A. Nitzan, S. Mukamel, and J. Jortner, J. Chem. Phys. 63, 200 共1975兲. 58
T. N. Hall and J. R. Holden, Naval Surface Warfare Center Technical
33
D. Hsu and J. L. Skinner, J. Chem. Phys. 81, 1604 共1984兲. Report No. NSWC MP 88-116, 1988.
34
D. Hsu and J. L. Skinner, J. Chem. Phys. 81, 5471 共1984兲. 59
A. Gonor and I. Hooton, in Proceedings of the 13th International Deto-
35
S. A. Egorov and J. L. Skinner, J. Chem. Phys. 103, 1533 共1995兲. nation Symposium, 2006 共Office of Naval Research, Arlington兲, p. 1297.
36
S. A. Egorov and J. L. Skinner, J. Chem. Phys. 105, 10153 共1996兲. 60
B. M. Dobratz and P. C. Crawford, Properties of Chemical Explosives
37
S. A. Egorov and J. L. Skinner, J. Chem. Phys. 106, 1034 共1997兲. and Explosive Simulants 共Lawrence Livermore National Laboratory, Liv-
38
L. Davison, Fundamentals of Shock Wave Propagation in Solids ermore, CA, 1985兲, UCRL-52997.
共Springer-Verlag, Berlin, 2008兲. 61
S. P. Marsh, LASL Shock Hugoniot Data 共University of California Press,
39
J. M. Walsh and R. H. Christian, Phys. Rev. 97, 1544 共1955兲. Berkeley, 1980兲.

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
155.33.16.124 On: Fri, 28 Nov 2014 02:17:00