DURINT Review
Processing and Behavior of Nanoenergetic Materials
November 17, 2005
Aberdeen, Maryland
MOLECULAR DYNAMICS STUDIES OF
NANOPARTICLES OF ENERGETIC MATERIALS
Donald L. Thompson
Department of Chemistry
University of Missouri-Columbia
Final Review
Collaborators
Saman Alavi (Now: NRC-Ottawa)
Jerry Boatz (AFRL-Edwards)
Don Brenner (NCSU)
John Mintmire (OSU)
Ali Siavosh-Haghighi (MU)
Dan Sorescu (NETL-Pittsburgh)
Gustavo Velardez (MU)
Focus
Simulations of Nanoparticles of Energetic Materials
Model physical and chemical properties of energetic nanoparticles
Systems:
 Al and Al2O3
 Nitro and Nitramine compounds
Processes:
 Structure
 Melting
 Chemistry
Understanding the
properties of nanoparticles
& how they relate to
bulk materials
Overview
Reaction of HCl on Al2O3 (validation study)
Reactions of energetic molecules on Al and Al2O3 surfaces
Oxidation of Al nanoparticles
Structures and properties of nanoparticles:
Al, NM, RDX, CL-20
Melting of Al, nitromethane, and CL-20
Past year: Completed Melting of Al study
Current: Shapes of nanoparticles
RDX crystals grown in various solvents
Acetone
Cyclohexanone (with water)
Cyclohexanone (without water)
g –Butyrolactone
Original dissolved crystal
E. D. M. van der Heijden and R. H. B. Bouma, Cryst. Growth Des. 4, 999 (2004)
Theoretical Predictions of Shapes
The equilibrium shapes of crystals are the result of the dependence
of the interface free energy per unit area on the orientation of the
interface relative to the crystallographic axes of the bulk solid, and
the microscopic properties of solids and interfaces determine the
details of this dependence.
The shapes can be predicted, given an accurate potential, by using
Wulff construction [G. Wulff, Z. Kristallogr. 34, 449 (1901).]
Some preliminary studies of predictions of the shapes of RDX
nanoparticles…
Wulff Construction
The interfacial free energy per unit area fi(m) is plotted in a polar frame.
A radius vector is drawn in each direction m and a plane is drawn
perpendicular to it where it intersects the Wulff plot.
The envelop of the family of Wulff planes
is the shape of the crystal.
A cusp in the Wulff plot
occurs for a facet of the
corresponding orientation
of the crystal shape.
m
M. Wortis, Chemistry and Physics of Solid
surfaces VII Vol 10(7), 367-405, 1998.
Generating Initial Conditions
9x9x9
Begin with a
9x9x9 supercell
5x5x5
Rotate by angles of θ and φ,
then cut from the core a 5x5x5*
simulation supercell with various
crystallographic surfaces
* A 5x5x5 supercell contains
~1000 RDX molecules.
Simulations: DL-POLY-2.15
10000 time steps of NVT simulation which of 7000 steps are equilibration. (time steps = 0.1 fs)
Simulations
5x5x5
A series of crystals, with various surfaces,
were equilibrated in a vacuum (no boundary
conditions.
* Actually, 1x10-8 K
We take T = 0 K* so that we need only compute the interaction energy
(avoiding the difficulty of computing the entropy).
10,000 time steps of NVT simulation of which 7,000 steps are
equilibration. (time steps = 0.1 fs)
Force Field:
SRT* (intermolecular) + AMBER (intramolecular)
Approximate, but satisfies basic requirements for our purpose:
Accurate description of solid-phase properties & flexible to qualitatively
account for molecular behavior in response to surface tension.
vdw cutoff radius: 11Å
* Sorescu, Rice, and Thompson,
J. Phys. Chem. B 101, 798, 1997.
Surface free energy
To avoid the complexity of calculating DS, we determine the equilibrium shape of the
crystal at a temperature very close to 0 K (T=1x10-8K). So that the problem is reduced
to calculating the surface enthalpy of the crystal at various angles.
0
Interaction energy (kj/mol)
Surface
core
-500
Free energy
at 0 K
-1000
-1500
-2000
-2500
-3000
-3500
-4000
-15
Interaction energy (kj/mol)
0
-5
0
5
volume layers
...
-500
-10
-1000
The interaction energy
is calculated for the
molecules in the bins
-1500
-2000
-2500
-3000
-3500
-4000
-15
-10
-5
0
volume layers
5
10
15
Repeat for different
values of θ and φ.
10
15
Crystallographic orientations of Wulff planes calculated
Wulff Plane
θ
φ
200
0°
0°
002
90°
0°
102
70°
0°
210
0°
30°
111
40°
49°
110*
0°
49°
332
150°
49°
020
0°
90°
021
30°
90°
* Blue numbered Wulff planes are not reported in Bouma and van der Heijden study.
[Cryst. Growth Des. 4, 999 (2004)]
For example, results for φ=30°
Cusps in a Wulff plot indicate surfaces with low surface energy.
The line that is perpendicular to the vector from the center
represents an equilibrium plane – a Wulff plane.
90
120
60
150
30
Cusps
180
0
0
1e+5
2e+5
3e+5
210
4e+5
5e+5
330
Wulff plane
Of a cusp
240
300
270
φ=49°
002
Black labels: Seen in
lab-grown RDX crystal
111
90
332
120
60
150
Blue labels: Not seen
in lab-grown RDX crystal
30
110
180
0
0
1e+5
2e+5
3e+5
210
4e+5
5e+5
Interaction energy
(kJ/mol)
330
Area enveloped by
Equilibrium surfaces.
240
300
270
φ=0°
200
102
002
90
120
g(q) plot
60
150
30
180
0
0
1e+5
2e+5
3e+5
210
4e+5
5e+5
330
240
300
270
Interaction energy
(kJ/mol)
φ=30°
002
90
120
60
150
210
30
180
0
0
1e+5
2e+5
3e+5
210
4e+5
5e+5
330
240
300
270
Interaction energy
(kJ/mol)
φ=49°
002
332
90
120
111
60
150
30
110
180
0
0
1e+5
2e+5
3e+5
210
4e+5
5e+5
330
240
300
270
Interaction energy (kJ/mol)
φ=90°
002
021
90
120
60
020
150
30
180
0
0
1e+5
2e+5
3e+5
210
4e+5
330
240
300
270
Interaction energy
(kJ/mol)
Shape
002
102
111
021
200
210
020
332
Oxygen
Nitrogen
Carbon
Hydrogen
002
102
332
020
200
Shape
111
Oxygen
Nitrogen
Carbon
Hydrogen
021
Conclusions/Future Work
Tentative Conclusions based on very approximate potential
In accord with experiment, we predict that the surfaces more frequently
seen in the lab grown crystals of RDX are the ones with oxygen atoms
sticking out of the surfaces.
We predict the same “large faces” as seen experimentally.
Next:
Simulations in solvents (e.g., acetone)
T>0K
Other materials, e.g., CL-20
Effects of binders
Very Brief Review
Reaction of HCl on Al2O3 (validation study)
Reactions of energetic molecules on Al and Al2O3 surfaces
Oxidation of Al nanoparticles
Structures and properties of nanoparticles:
Al, NM, RDX, CL-20
Melting of Al, nitromethane, and CL-20
Publications
 S. Alavi, D. C. Sorescu, and D. L. Thompson, “Adsorption of HCl on a Single-
Crystal -Al2O3 (0001) Surface,” J. Phys. Chem. B 107, 186-195 (2003).
 D. C. Sorescu, J. A. Boatz, and D. L. Thompson, “First-Principles Calculations of
the Adsorption of Nitromethane and 1,1-Diamino-2,2-dinitroethylene (FOX-7)
Molecules on the Al (111) Surface,” J. Phys. Chem. 107, 8953-8964 (2003).
 S. Alavi and D. L. Thompson, “A Molecular Dynamics Study of Structural and
Physical Properties of Nitromethane Nanoparticles,” J. Chem. Phys. 120, 1023110238 (2004).
 S. Alavi, G. F. Velardez, and D. L. Thompson, “Molecular Dynamics Studies of
Nanoparticles of Energetic Materials,” Materials Research Society Symposium
Proceedings 800, 329-338 (2004).
 S. Alavi, J. W. Mintmire, and D. L. Thompson, “Molecular Dynamics Simulations
of the Oxidation of Aluminum Nanoparticles,” J. Phys. Chem. B 109, 209-214
(2005).
 D. C. Sorescu, J. A. Boatz, and D. L. Thompson, “First Principles Calculations of
the Adsorption of Nitromethane and 1,1-Diamino-2,2-Dinitroethylene (FOX-7)
Molecules on Al2O3(0001) Surface,” J. Phys. Chem. B 109, 1451-1463 (2005).
 S. Alavi and D. L. Thompson, “Molecular Dynamics Simulations of the Melting of
Aluminum Nanoparticles,” J. Phys. Chem. B, in press.
Nitromethane on Al2O3
Minimum energy reaction pathway for dissociation NM leading to
adsorbed OH and CH2NO2
Calculations performed using VASP
Nitromethane on Al
N-O bond broken, Al-O and Al-N bonds formed
D. C. Sorescu, J. A. Boatz, and D. L. Thompson, “First-Principles Calculations of the Adsorption
of Nitromethane and 1,1-Diamino-2,2-dinitroethylene (FOX-7) Molecules on the Al (111) Surface,”
J. Phys. Chem. 107, 8953-8964 (2003).
Calculations performed using VASP
Aluminum Nanoparticles
• Streitz-Mintmire potential. More flexible than other model potentials
used in metal nanoparticle simulations
* Simulated annealing
* NVT simulation
* T = 250 K
* Δt = 2 fs
* 400 ps simulation time
• Characterization of structures
• Magic number effects
• Determination of melting points
* Potential energy plots  bistabilty
* Lindemann Index, 
• Charge distribution in the nanoparticles; implications on reactivity
Melting of “Non-Magic Number”
Aluminum Nanoparticles
Melting of “Magic Number”
Aluminum Nanoparticles
Lindemann Index

2

N ( N  1) i  j
rij2  rij
t
rij
t
2
t
Melting Point as a Function of Aluminum Nanoparticle Size
• Melting point determined from the Lindemann Index
• Melting range determined from the potential energy curves
Magic number nanoparticles
Other nanoparticles
Average Charge Distribution in Al Nanoparticles
+0.025
+0.051 (2nd shell,
corners)
0.29
0.004 (2nd
shell)
13 atoms
0.065
(1st shell)
+0.018
+0.031
0.22
+0.017
55 atoms
19 atoms
+0.038
(core atom)
Conclusions: Al Nanoparticles
• Show magic number behavior
• Some small metallic nanoparticles differ from their
Lennard-Jones analogs
• Small nanoparticles show bistability between solid and liquid
phases at intermediate temperatures
• Atoms in the nanoparticles have non-uniform charge distributions
and may show different reactivities at various surface sites
for different particle sizes
Nitromethane nanoparticles
•
Nanoparticles with 32 to 480 nitromethane molecules
•
Characterization of structure
•
Energetics of the nanoparticle
 enthalpy of melting
 enthalpy of vaporization
•
Determination of melting point
for different sized nanoparticles
 density
 diffusion coefficient
 Lindemann index
Nitromethane nanoparticles
480 molecules
After 50 ps runs
240 molecules
96 molecules
170 K
115 K
“solid”
230 K
In solid nanoparticles, dipolar
forces maintain the ordered
structure in the core
“liquid”
250 K
Do not appear to show
magic number structures,
or we didn’t find them.
Melting range and temperature with nanoparticle size: Nitromethane
S. Alavi and D. L. Thompson, “A Molecular Dynamics Study of Structural and
Physical Properties of Nitromethane Nanoparticles,”
J. Chem. Phys. 120, 10231-10238 (2004).
Nitromethane nanoparticles
• The structure is dominated by dipole forces
• We did not discover magic number clusters
• Melting point varies smoothly with nanoparticle size
Simulation of CL-20 nanoparticles
DL_POLY MD program
• Fixed molecular structures
• Sorescu, Rice, and Thompson potential
(Buckingham + Coulombic)
• Annealed and non-annealed nanoparticles
• Time step = 2 fs
• 100 ps equilibration
• 200 ps runs
Nanoparticles of CL-20 or HNIW
(2,4,6,8,10,12-hexanitrohexaazaisowurtzitane)
Simulations on CL-20 nanoparticles
•
•
•
Characterization of structure
 density
 dipole-dipole correlations
 surface dipole alignments
 surface functional group alignments
Energetics of the nanoparticle
 enthalpy of vaporization
Surface coating (next stage)
S. Alavi, G. F. Velardez, and D. L. Thompson,
“Molecular Dynamics Studies of Nanoparticles of Energetic Materials,”
Materials Research Society Symposium Proceedings 800, 329-338 (2004).
bulk solid CL-20
Open Sorescu et al.
Solid: present study
Densities of CL-20 Nanoparticles
48-molecule
non-annealed
88-molecule
annealed
48-molecule
annealed
Snapshots of CL-20 Nanoparticles
48-molecule
non-annealed
48-molecule
annealed
88-molecule
annealed