9/12/2022
MD Simulations: Basics
Dr. Radha Raman Mishra
BITS Pilani Department of Mechanical Engineering
Pilani Campus
Outlines
• Need of simulations
• Basics of simulations
• Molecular Dynamics
Basics
Potentials
Mathematical Theory
Program structure
Software
BITS Pilani, Pilani Campus
1
9/12/2022
Why simulation???
BITS Pilani, Pilani Campus
Think about ….
Weather forecasting
Flight simulations
Sun’s magnetic field simulation
Car crash simulation Cancer cell killing using ultrasound
BITS Pilani, Pilani Campus
2
9/12/2022
Think about ….
Fight simulations
Material fracture
[Link]
Material deposition
Fire simulation
Aerodynamic simulations
BITS Pilani, Pilani Campus
Observations
Practicality of systems under consideration are –
• Difficult to implement
• Involve high costs
• Impractical
• Human life threat
• Time consuming
BITS Pilani, Pilani Campus
3
9/12/2022
Observations
Application areas
Health care
Public
Computers
domains
Applications
Military Materials
Tele-
Manufacturing communicatio
n
BITS Pilani, Pilani Campus
Need
BITS Pilani, Pilani Campus
4
9/12/2022
Computational material science (CMS)
• Involves computational tools for solving materials related
problems.
• There exist different mathematical models for investigating
problems at multiple length and time scales which help in –
understanding evolution of material structures (at different
length scales) and
how these structures effectively control material properties
• With this understanding we can select materials for specific
applications and also design advanced materials for new
applications.
BITS Pilani, Pilani Campus
Computational Materials Science
• Main characters:
atoms and
constituting electrons and nuclei
• Always follow specific rules without
exceptions.
• An electron never kills itself or other
electrons because of love.
• The same goes for nuclei, atoms, molecules,
and materials.
• They simply follow the known laws of nature: mainly classical and quantum
mechanics in terms of electromagnetic force
• Possibility: all forms of phenomena encountered in materials by tracing
down the interactions between them.
BITS Pilani, Pilani Campus
10
5
9/12/2022
Scales of modelling in CMS
Phase-field
Time Method (PFM) Finite Element Method (FEM)
Macroscale
sec Molecular Dynamics (MD)
and Monte Carlo (MC)
Density Functional
Theory (DFT)
ms
Mesoscale
ns Molecular Dynamics
ps Quantum Mechanics
fs Length
nm μm mm m
BITS Pilani, Pilani Campus
11
Multi-scale method in CMS
• Solves material problems which have important features at
multiple length scales
• Involves exchange of information between computational
tools at different length scales
• Take input from thermodynamic and kinetic databases for
making quantitative predictions.
BITS Pilani, Pilani Campus
12
6
9/12/2022
AIM of CMS
In CMS, the aims are –
understand various properties and phenomena of materials
and
achieve designing and making better materials for society
BITS Pilani, Pilani Campus
13
Goal of CMS
Modeling materials with computers that are programmed with
theories and algorithms based on –
physics,
mathematics,
chemistry,
material science, and
Computer science
BITS Pilani, Pilani Campus
14
7
9/12/2022
Examples of CMS
• Example: sintering behavior of a metal/ceramic
Approach 1: sintering furnace in a laboratory
Approach 2: molecular dynamics (MD) on atomic scale using
a computer
By changing various input conditions, the entire spectra of
data can be generated efficiently and accurately if the runs
are set up properly
BITS Pilani, Pilani Campus
15
Example of CMS
Under extreme and hostile conditions
• Material processing under extreme and hostile conditions –
high pressures, high temperatures, and
presence of toxic substances or nuclear radiation
• Materials under nuclear fusion environments: various
damages occurring in fusion materials by neutron irradiation
can be simulated without worrying about expensive equipment
and danger of radiation.
• Silicon IC Chips:
used in cell phone, smart phone, smart pad, TV, computer,
and so on
Using CMS, design better materials and develop faster,
smaller, and lighter IC chips
BITS Pilani, Pilani Campus
16
8
9/12/2022
Fundamental Interactions in CMS
• To study any effect in matter, any one of the four known
fundamental interactions are responsible –
Electromagnetic
Strong nuclear
Weak nuclear, and
Gravitational
• In materials science:
Only the electromagnetic interaction are important
Electromagnetic interactions between nuclei, electrons,
and atoms
BITS Pilani, Pilani Campus
17
Basic procedures in CMS
• Define what to calculate
• Make a model system that represents the real system properly
• Select the relevant rules (classical mechanics, quantum mechanics,
theories, algorithms, etc.)
• Select a code/module/program/package to do the job for the system.
• Run simulation, analyze the results, and refine the run under better
defined conditions
• Produce data and compare them with reported data by other relevant
studies and experiments.
Recreating a part of nature in the frame of our simulation
system in a simplified and well-controlled manner
BITS Pilani, Pilani Campus
18
9
9/12/2022
Molecular Dynamics (MD)
• MD is based upon Newton’s second law of
motion –
𝐹⃗ = 𝑚 𝑎⃗
The method using this classical mechanics
for atomic movements is called Molecular
Dynamics (MD).
• Uses algorithms to integrate the equations of motion
• Newtonian dynamics:
to compute each atom's net force and acceleration
definition of a force field or potential is required to compute the forces
• Practically, MD is the only option for big systems of more than thousands of
atoms such as for – melting nano-spheres, sintering particles, deforming
nanowires, and crack propagating blocks
BITS Pilani, Pilani Campus
19
Atomic model in MD
Approximation: an atom is basically
approximated as a sphere with point
mass at the center
Note:
• electron’s role is totally neglected
• electronic wave function is assumed to instantaneously adapt to the
current configuration of atoms
• computation becomes drastically simple
• Negative consequence: presence of electrons is the origin of
interatomic potential; is totally neglected
• potential between atoms should be generated empirically to carry out
MD.
BITS Pilani, Pilani Campus
20
10
9/12/2022
MD: Basics
Classical Mechanics System: Observations:
A jumping atom with a given • Given the following information –
energy in a potential well the present position and
velocity of an atom
• Its future and past trajectories can be
calculated precisely by using Newton’s
equations of motion.
• Energy changes during a jump are
continuous from the very bottom of the
well (zero-energy point) to the maximum.
• No jump possible: if an atom’s energy <
the potential of walls
BITS Pilani, Pilani Campus
21
MD: Basics
MD is an integration of Newton’s equations of motion over time to obtain
the time evolution of the system and thus the properties of our interests.
• Force calculation: initial positions and
velocities of every atom, and interatomic
potential
• New properties (new positions, velocities,
and so on): use previous information to
advance initial positions toward lower
energy states through a small-time interval
(called a time step, Δt)
• Equilibrium (no change in properties with
time): input new data and repeat the above
steps typically for > 1000
BITS Pilani, Pilani Campus
22
11
9/12/2022
MD: Basics
Raw data: during and after equilibration, for each or some time steps –
• atomic positions and momenta,
• energies,
• forces, and so on.
Calculation of Properties: directly or via statistical analysis, following can be
calculated:
• Basic energetics, structural and mechanical properties
• Thermal expansion coefficient, melting point, and phase diagram in terms of
pressure and volume
• Defect structure and diffusion, grain boundary structure, and sliding
• Heat capacity, free energy differences between phases, and thermal
conductivity
• Radial distribution function and diffusion coefficient for liquids
• Descriptions for processes and phenomena such as sputtering, vapor
deposition, fast plastic flow, crack growth and fast fracture, nano-indentation,
propagation of shock wave, detonation, irradiation, ion bombardment, cluster
impact, and operation of nanogear
BITS Pilani, Pilani Campus
23
MD: Limitations
• Availability of potential is limited especially for multi-
component systems,
• Accuracy of a potential is always under question
• Length scale is still not macroscopic
• Time scale is limited to nanoseconds
• No electromagnetic properties can be obtained
BITS Pilani, Pilani Campus
24
12
9/12/2022
MD: Potentials
• When atoms are near enough to feel each other, the balance between
attraction and repulsion takes place, and it is determined by so-called
interatomic potential.
• Atoms will eventually settle down at the minimum potential states at the
equilibrium distances
• The sum of all forces acting on an atom, F, is
where: a is acceleration, v is velocity, t
is time, r is position, p is momentum
• Isolated system for MD simulations: If the total energy E is constant in time
(dE/dt = 0):, F is related to the negative gradient of potential with respect to
position
where U is the potential.
BITS Pilani, Pilani Campus
25
MD: Potential Calculation Method
Empirical Potential Semi – Empirical
Ab-initio MD
Functions Potential Functions
• Functional form • Calculate the • Direct QM
for the potential electronic wave calculation of
• fitting the function for fixed electronic
parameters to atomic positions structure
experimental from QM
data • Ex - Car-Parrinello
• Ex - EAM, Glue using plane-wave
• Ex - Lennard- Model, Tersoff psuedopotential
Jones, Morse,
Born-Mayer
BITS Pilani, Pilani Campus
26
13
9/12/2022
MD: Pair Potential
• For each particle: N-1 pairs
• For N particles: N(N-1) pairs
• Pair potential calculation: 𝛼 O(N )
U U1 (ri ) U 2 (ri , r j ) U 3 (ri , r j , rk ) ...
i i, j i , j ,k
BITS Pilani, Pilani Campus
27
MD: Pair Potential – Leonard – Jones Potential
Considers only the two-atom interactions
𝜎 𝜎
𝑈 𝑟 = 4𝜀 −
𝑟 𝑟
Express the atomic interactions of noble gases, spherical molecules, and secondary bonds
very well.
Ex - Argon
BITS Pilani, Pilani Campus
28
14
9/12/2022
MD: Pair Potential: Embedded Atom Method (EAM)
The EAM considers both contributions from the pair potential and the embedding
energy that approximates the missing N-atom effect in a pair potential.
𝑈 = 𝑈 𝑟 + 𝐹 𝜌
Gives an accurate result of FCC elements and alloys.
Ex - Aluminium
BITS Pilani, Pilani Campus
29
MD: Time steps
• For the numerical integration of simulated trajectories, we
need to determine the timestep for each unit of this time step
in which we do the integration.
• Too long Δt : Energy is not conserved.
• Δr/Δt < 1/20 of the nearest atom distance.
• In practice Δt < 4 sec.
• MD is limited to < 100 ns
BITS Pilani, Pilani Campus
30
15
9/12/2022
MD: Ensembles
An ensemble is the collection of possible systems that belongs
to the different microscopic states but with an identical
macroscopic or thermodynamic state.
Ensemble Fixed variables
Microcanonical N, V, E
Canonical N, V, T
Isobaric N, P, T
Isothermal
Grand Canonical μ, V, T
BITS Pilani, Pilani Campus
31
Periodic Boundary Conditions
create images of the
simulation box:
duplication in all directions
(x, y
and z).
if an atom is moving out of
boundary, it comes from the
other side.
BITS Pilani, Pilani Campus
32
16
9/12/2022
MD: Basic Steps for Simulation
• Starting configuration:
Atomic positions (x,y,z)
density, mass, charge, ….
• Initial velocities: depend on temperature
• Boundary conditions (PBC)
• Set the appropriate potential:
available and supported potentials
depend on the system to simulate
• Set the appropriate time step: should be short (order of 1fs).
• Set the temperature and pressure control:
define the thermodynamic ensemble (NVT, NPT, NVE, …).
• Fix run time and customize the output: depend on the software
(thermo, dump).
BITS Pilani, Pilani Campus
33
MD: Simulation Process
𝑟 𝑡 ,𝑉 𝑡
𝐹 𝑟
MD steps/ Loop over time
𝑟 𝑡 → 𝑟 𝑡 + ∆𝑡
𝑉 𝑡 → 𝑉 𝑡 + ∆𝑡
Thermodynamic
properties
MD trajectories
𝑡 = 𝑡 + ∆𝑡
𝑡>𝑡
BITS Pilani, Pilani Campus
34
17
9/12/2022
Software Tools for MDS
LAMMPS:
[Link]
NAMD:
[Link]
GROMACS: [Link]
Visualization Tool VMD OVITO, Avogadro,
and VESTA
BITS Pilani, Pilani Campus
35
Thank You
For further interaction:
[Link]@[Link]
36
18
