Multiscale Modeling and Simulation

First-Principles Based Models, Methods, and Experiments on Reversible Shear Thickening Fluids for Drilling, Stimulation, Conformance Control, and EOR Operations

Understanding Plasma Formation from Hypervelocity Impact (HVI)

Our role in this effort has been on the development of first-principles-based multiscale methods to understand the dynamics of materials during the extreme conditions of hypervelocity impact (HVI), including the characterization of pressure and temperature dependent material transformations from ground state, through warm-dense regimes and up to plasma phases. A critical goal for HVI modeling is to capture the non-adiabatic effects on material properties that result from medium to large electronic excitations, and to understand the particle interactions that take place between high-energy electrons, nuclei, ions and larger species during HVI. The warm-dense regime is particularly challenging since it lays in a "computational no-mans land" between the cold matter described well by ground-state quantum mechanics (QM) methods (e.g. density functional theory, DFT) and the hot matter described well by classical plasma models (e.g. Particle-in-Cell, PIC). We have developed unique methods to address these issues via mixed quantum-classical theory, which extend the limited capabilities of QM calculations into large-scale dynamics (millions of atoms) of systems at finite (and large) temperatures/pressures. In addition to this fundamental role, our group provides first-principles predicted properties to parameterize atomistically-informed continuum methods developed by other groups in the Caltech-PSAAP effort (e.g. OTM).

Our contributions to date have included the development of reaxFF reactive force fields for accurately modeling the reaction processes and phases of different materials of interest to NNSA (e.g. Nylon, Al, Ta, Fe), Tight Binding (TB) models to implicitly correct reaxFF trajectories for high-energy electronic effects, as well as the development of a non-adiabatic quantum-classical wave-packet dynamics method called the electron force field (eFF) for modeling the explicit dynamics of electronically excited states (Figure to the left shows the general flow for HVI simulations using our methods). eFF has been demonstrated and validated for ground state and highly excited system dynamics [1,3,4,5,7,9-11].

Using ReaxFF we have modeled the effect of reaction processes, including bond breaking and formation and phase transformations, in the extreme conditions of temperature and pressure that occur during the HVI (<10km/s), for materials including Ta, Fe, Al, and Nylon. To incorporate electronic structure effects in ReaxFF we have tested the use of a TB approach that restricts the basis sets to the minimal set needed to capture the excitations in the bands of interest, i.e. it focuses only on the bands near the Fermi energy for a particular molecular system, to avoid the cost of a complete basis set[13]. The key benefit of this first-principles parameterized TB approach is that it can scale to systems with tens of millions of atoms, while the description remains at the atomic level, with the position of each atom described explicitly as well as the electronic density at the atomic site.

At higher velocities (>10km/s) we have shown the applicability of eFF to understand the electronic effects on material properties, including cascaded electronic ionization, non-adiabatic dissociation,, increased conductivity, among other phenomena.  We have validated the eFF approach for low-Z elements (H, He, Li, Be, B, and C) using full electron representations via floating spherical Gaussian wavefunctions and to overcome the limitations of spherical Gaussians in describing p and higher order angular momentum orbitals, multiple bonds and lone pairs in eFF, we introduced first-principles-based effective core potentials (ECP), which describes the valence electrons in the vicinity of core electrons combined as a single pseudo core particle. As a result atoms C, N, O, Na, Al, and Si are now described with accurate bonding energy and geometries in the eFF model. This results in proper description of multiple bonds (sigma-pi rep.) and lone pairs (open shell rep.), but further efforts are necessary for accurate representation of conjugation and hydrogen bonding. To reach the length- and time-scales required for studying materials under extreme conditions, e.g. in Rayleigh-Taylor instabilities, we have made our parallel implementation of eFF available as a user package in the Sandia code LAMMPS.

We have applied eFF on large-scale, long-term dynamics problems, including: static and dynamic shock (figure on the left shows a piston driven at HV into a lithium slab, depicting electron excitations in the compressed front as red spheres and a movie of two lithium slabs colliding at HV) Hugoniots for Lithium [7,11], poly(ethylene) [1], material interfacial instabilities, high-pressure EOS for Aluminum and Nylon [results now in use by the Ortiz group in the OTM hypervelocity impacts of Nylon on Aluminum], brittle fracture dynamics (Movie shows a Si crystal delaminating from uniaxial mechanical loading - electronic energies are color-coded) of silicon[3], and hypervelocity impact of different molecular species and clathrates [5].

In addition to our force fields effort in PSAAP, we have developed new methods to improve the accuracy of DFT-QM and incorporated corrections to describe the London dispersions responsible for long-range van der Waals attractions in our ReaxFF engine[12].

The first-principles-based approach provides increased confidence in our simulation results, but to estimate the level of accuracy we confront our results against available experimental data and use both results, from computations and experiments, to estimate the level of uncertainty involved using the Uncertainty Quantification (UQ) formalisms (Owhadi and Ortiz) and code (Aivazis) under development by other teams in the Caltech PSAAP.

1. Theofanis, P.L, Jaramillo-Botero, A., Goddard, W.A. III, Mattsson, T., Thompson, A. "Electron dynamics of shocked polyethylene crystal", Accepted in PRB, March, 2012
2. Qi, A., Zybin, S., Goddard, W.A., III, Jaramillo-Botero, A. "Elucidation of the dynamics of hot spot initiation and chemical reactions at interfaces of highly shocked materials" Accepted in Phys. Rev. B, August 2011.
3. Theofanis, P.L, Jaramillo-Botero, A., Goddard, W.A. III, "Non-adiabatic study of dynamic electronic effects during brittle fracture in silicon", Accepted in PRL, December, 2011
4. Jaramillo-Botero, A. "Modeling and simulation of large-scale reactive systems in extreme conditions", invited talk HPC User Forum, San Diego, September 7, 2011.
5. Jaramillo-Botero, A., Cheng, MJ, Beegle, L., Hodyss, R., Goddard, WA III. "Predicting the molecular composition of Enceladus' south pole plume after hypervelocity impact with the Cassini orbiter", Journal article in preparation.  Data presented at the 42nd Lunar and Planetary Science Conference, March 7-11, 2011, Texas.  
6. Jaramillo-Botero, A., J. Tahir-Kheli, P. von Allmen, and W.A. Goddard, Multiscale, multiparadigm modeling for nano systems characterization and design, in Handbook of nanoscience, engineering, and technology, W.A. Goddard, et al., Editors. 2012, CRC Press, Taylor & Francis Group.
7. Jaramillo-Botero, A., J.T. Su, A. Qi, and W.A. Goddard, Large-scale, long-term nonadiabatic electron molecular dynamics for describing material properties and phenomena in extreme environments. Journal of Computational Chemistry, 2011. n/a. doi: 10.1002/jcc.21637.
8. Theofanis, P.L., A. Jaramillo-Botero, and W.A. Goddard, Non-Adiabatic Study of Dynamic Electronic Effects During Brittle Fracture of Silicon. Submitted to PRL, 2011.
9. Jaramillo-Botero, A., J.T. Su, and W.A. Goddard, pEFF: the parallel electron force field for large-scale, long-term nonadiabatic excited electron dynamics under LAMMPS, 2009, http://lammps.sandia.gov/movies.html - eff: Pasadena.
10. Jaramillo-Botero, A., M.J. Cheng, V. Cvicek, L.W. Beegle, R. Hodyss, and W.A. Goddard. First-principles-based reactive atomistic simulations to understand the effects of molecular hypervelocity impact on Cassini's Ion and Neuratl Mass spectrometer. in 42nd Lunar and Planetary Science Conference. 2011. The woodlands, Texas.
11. Kim, H. and W.A. Goddard, The PBE-lg method of improved London Dispersion for elements up to Lr (103). To be published in Journal of Physical Chemistry Letters, 2011.
12. L. Liu, Y. Liu, S. Zybin, H. Sun and W.A. Goddard III, "ReaxFF-lg: Correction of the ReaxFF Reactive Force Field for London Dispersion, with Applications to the Equations of State for Energetic Materials", Journal of Physical Chemistry A, 115 (40). pp. 11016-11022. ISSN 1089-5639
13. Jaramillo-Botero, A., Tahir-Kheli, J., von Allmen, P., Goddard WA III, "Multiscale, multiparadigm modeling for nano systems characterization and design", CRC handbook of Nanoscience, Engineering and Technology, 3rd ed. Chapter 29. In press (May 2012).

First-Principles, Multiscale Modeling of Fire-Fighting Aqueous Film Forming Foams (AFFF)

Aqueous Film Forming Foam (AFFF) is the prime fire-fighting agent used for liquid hydrocarbon fires, it works by forming a water film beneath the foam that cools the liquid fuel and stops the formation of flammable vapors. This provides dramatic fire knockdown, an important factor in crash rescue fire fighting. The Air Force (AF) uses this fire-fighting agent in all Aircraft Rescue and Fire Fighting vehicles to respond to aircraft fires. The AF Research Laboratory is interested in determining and clearly understanding the unique processes of this agent for extinguishing burning liquid hydrocarbons.

This research is meant to understand and elucidate fundamental structures and properties of fire-fighting AFFFs, derived from the interfacial systems in Newton-Black Films (Figure to the right shows a molecular model of a perfluorinated surfactant-based NBF, and figure above shows a graph of its calculated disjoining pressure isotherm as a measure of film stability), by leveraging, extending and applying existing first-principles-based multiscale theory, methods and computational tools developed at the Materials and Process Simulation Center, at the California Institute of Technology. These methods and tools will provide accurate, atomistically resolved, information about AFFFs composition and behavior to enable Air Force Research Laboratory scientists to predict and optimize quantities (e.g. reduce volumes) and compositions (e.g. reduce toxicity via reduction/replacement of fluorinated surfactants) of AFFF precursors for performance enhancement.

Influence of Hydrogen Surface Coverage and Diffusion on Silicon Growth Processes

Silicon (Si) is the most widely used semiconductor material in industry. It is fundamental for the production of semiconductor microelectronic devices as well as to other areas of application, including, the development of novel high efficiency thin films for solar cells. Most of these applications require starting with an atomically smooth Silicon surface which is conventionally achieved using a Chemical Vapor Deposition (CVD) process to perform epitaxial crystal growth of Si. A model image of the growth process that takes place during the CVD is shown to the right.

In order to understand and optimize the multiscale properties and phenomena that lead to a smooth semiconductor thin film surface it is crucial to develop predictive models capable of expressing the chemical-physical processes involved in crystal growth within a CVD reactor.

This study involves developing first-principles-based multiscale methods and tools to elucidate the influence of adsorbed hydrogen on growing silicon surfaces. It is known that the presence of hydrogen adsorbed on the surface can alter the growth regime, leading either to a desired smooth crystal surface or to an undesired atomically rough surface morphology. In order to understand the material properties and phenomena of interest it is imperative to account for critical events that take place at the atomistic scale, in particular chemical reactions, and their effect on structure evolution over into the mesoscale length and time scales.  Accurately modeling chemical reactions would ordinarily require quantum chemical calculations that take place in the femtosecond time-scales, unfortunately, these are limited to a few hundred atoms (< 1,000 at most) and understanding the larger scale structural changes over time would require modeling systems with thousands of atoms over microsecond time-scales.  

In order to circumvent these limitations we are using a multiscale-multiparadigm approach which integrates novel first-principles-based reactive force field methods (reaxFF), to determine the appropriate kinetic rates during diffusion (see movie shown to the left) and adsorption processes found under experimental conditions of temperature and pressure, and kinetic Monte Carlo methods to complete an accurate mesoscale predictive atomistically-resolved description of the growth process.

Understanding and Optimizing Portland Cement Properties During Hydration

Soon after mixing (Portland) cement with water, the aluminate phase (C₃A) mineral in the cement clinker reacts strongly with the water to form an aluminate-rich gel that in turn reacts with sulfate in solution to form nano-crystals of ettringite. This phase of hydration is strongly exothermic and it is often known as the Aft phase ("Alumina, Ferric oxide, tri-sulfate" or Al2O3 – Fe2O3 – tri). The alite and belite minerals in the cement start to hydrate after the Aft phase has initiated, forming calcium silicate hydrate and calcium hydroxide, and leading to an increase in concrete strength. During a period of heat evolution that lasts for hours, the cement grains react from the surface inwards, breaking the anhydrous particles into smaller grains.  A Ferrite (C₄AF) hydration reaction phase also starts quickly as water is added, but slows down as a byproduct iron hydroxide gel layer forms on the surface of the ferrite to inhibit further reactions.