Atomistic Studies of Fundamental Properties and Processes: CSAFE

1
Atomistic Studies of Fundamental Properties and
Processes C-SAFE LANL T-14 Interactions

• Thomas D. (Tommy) Sewell
• Theoretical Division
• Explosives and Organic Materials Group
• Los Alamos National Laboratory

2
Motivation

• Part of an overall upscaling/homogenization
  modeling methodology
• Upscaling atomistics?mesoscale
• Physical quantities in (P,T) space
• Increasingly, phenomenology such as plastic
  dissipation
• Homogenization mesoscale?macroscale
• Spatio-temporal distributions of stress,
  temperature, etc.
• the tails of the distributions are what matter
  for initiation
• Basis for statistical mechanics-based
  constitutive description
• Guided by needs of, interactions with, mesoscale
  modelers
• C-SAFE, LANL T-14 share many common interests
• Synergistic interactions have resulted
• Sewell with GSmith and Voth
• Bardenhagen with McMurtry and co-workers

3
Philosophy slide

• Simulations Let the (end) user beware
• Is the potential-energy surface validated?
• Are the methods sound for the problem?
• Are the accessible length and time scales
  compatible with the property or phenomenon of
  interest?
• Are the results likely to be highly corrupted by
  pesky quantum effects?
• Was care taken to ensure convergence?
• Experiments Dont accept uncritically numbers
  from the handbooks as well-determined fact
• Must also pay attention to thermodynamic state
  dependencies of important quantities

4
Comparison of HMX isotherm and hugoniot data

• Red Yoo Cynn, 3rd-order Birch-Murnaghan fit
• Black Olinger et al., Us-Up fit
• Grey diamonds hugoniot, solvent-pressed
• Grey circles hugoniot, single crystal

JCP 119, 7417 (2003)
MD
Combust. Theor. Mod. 6, 103 (2002)
QC
5
Coefficients of Thermal Expansion of HMX
private communication (2004)
b
?300-413 0.00027 K-1
a
?300-420 0.00020 K-1
?300-450 0.00013 K-1
d
Energetic Matls. Pt. 1, Politzer Murray
Eds. (Elsevier, Boston, 2003), Ch. 10, p. 279
6
HMX thermal conductivity specific heat
J. Energetic Matls. 17, 1 (1999)
Chem. Phys. Letters 324, 64 (2000).
7
Melt curve for HMX

• The melt curve, TmTm(P) is unknown

Depending on the form used for the melt
curve, one predicts that HMX will melt in a 19
GPa shock, or not at all! But see PRL 92,
235702 (2004)
TmTm0bP
TmTm0(1aDV/V0)
Shock T, Hayes EOS
Combust. Theor. Mod. 6, 103 (2002)
8
These things matter at the mesoscale
At the 2004 Shock Modeling Conference, the need
for good, unreactive EOS information was
repeatedly emphasized.
J. Reaugh, UCRL-ID-150388 (2002)
R. Menikoff, SCCM03
If one cannot predict temperature, stress, and
their spatio-temporal distributions, s/he cannot
credibly predict initiation.
9
and we are increasingly convinced that

• Given a validated force field, atomistic
  simulation
• will often yield predictions, for appropriate
  quantities, that are within experimental
  uncertainties
• can be used (with care) to obtain information in
  thermodynamic regimes where experiments are
  difficult, if not impossible, to perform
• The C-SAFE/GSmith and LANL/T-14 collaboration is
  developing methods and specific predictions for
  many properties of interest
• This research is gaining increasingly
  high-profile attention within the mesoscale and
  homogenization communities at Los Alamos

10
Two Examples of LANL-Centered Atomistic Studies
That Have Profited From C-SAFE/LANL
Collaborations
11
PETN cold curve, PBE/6-31G(3-D, all-electron,
density functional quantum chemistry)
E P vs- compression
Linear compressibilities

• Quite good agreement with experiment
• Not necessarily expected, especially at low
  pressure see J. Phys. Chem. B 108, 13100 (2004)
• Phys. Rev. B 69, 035116 (2004)
• Analgous studies of b-HMX underway (much more
  complicated system)

12
MD Simulations of shocks in crystalline HE
13
Cutout of snapshot from large simulation(cell
length along long axis is 0.15 micron)
Challenge is to identify, characterize, and
quantify the phenomena
14
Summary

• We are pushing hard to gain as much information
  as possible from atomistic simulations
• In many cases, atomistic simulation predictions
  are within experimental uncertainties
• We are now at a point where one can credibly use
  large-scale MD to study shocks and quasi-static
  loading to understand dissipation in complicated
  molecular crystals
• The LANL/T-14 C-SAFE/GSmith-Voth collaborations
  have been, and continue to be, of high value in
  advancing our capabilities and confidence in
  these areas