Title: Computational Domains for Explorations in Nanoscience and Technology
1Computational Domains for Explorations in
Nanoscience and Technology
Shaoping Xiao, CCAD, College of Engineering,
University of Iowa Deepak Srivastava, NASA Ames
Research Center Meyya Meyyappan, NASA Ames
Research Center Jun Ni, ITS, University of Iowa
2Outline
- 1. Introduction to nanotechnology
- 2. Computational nanotechnology
- Molecular modeling Quantum mechanics, Monte
Carlo and molecular dynamics - Multiscale modeling Hierarchical and concurrent
multiscale methods - High-performance computer techniques
- 3. Applications of Computational nanotechnology
- Nanoscale materials nanotubes and nanocomposites
- Nanoscale devices nano-gears and
nano-oscillators - 4. Conclusions
3Nanotechnology
- Nanotechnology is the understanding and control
of matter at dimensions of roughly 1 to 100
nanometers, where unique phenomena enable novel
applications. - Encompassing nanoscale science, engineering, and
technology, nanotechnology involves imaging,
measuring, modeling, and manipulating matter at
this length scale.
4Nanotechnology
- National Nanotechnology Initiative (launched in
2001) - The vision of the National Nanotechnology
Initiative is a future in which the ability to
understand and control matter on the nanoscale
leads to a revolution in technology and industry
(NNI, 2004). - NNI Goals and Plans (NNI 2004)
- Maintain a world-class research and development
program aimed at realizing the full potential of
nanotechnology. - Facilitate transfer of new technologies into
products for economic growth, jobs, and other
public benefits. - Develop educational resources, a skilled
workforce, and the supporting infrastructure and
tools to advance nanotechnology. - Support responsible development of
nanotechnology.
5Nanotechnology
- Four generations of nanotechnology applications
(NNI 2004) - First generation of products (2001 - ) passive
nanostructures, illustrated by nanostructured
coatings, dispersion of nanoparticles, and bulk
materials such as nanostructured metals, polymers
and ceramics. - Second generation (2005 - ) active
nanostructures, illustrated by transistors,
amplifiers, targeted drugs and chemicals,
actuators, and adaptive structures. - Third generation (2010 - ) three-dimensional
nanosystems with heterogeneous nanocomponents
using various syntheses and assembling techniques
such as bioassembling networking at the
nanoscale and multiscale architectures. Research
focus will shift towards heterogeneous
nanostructures and supramolecular system
engineering. - Fourth generation (2015 - ) heterogeneous
molecular nanosystems, where each molecule in the
nanosystem has a specific structure and plays a
different role. Molecules will be used as devices
and from their engineered structures and
architectures will emerge fundamentally new
functions.
6Computational Nanotechnology
- COMPUTATIONAL NANOTECHNOLOGY
- Modeling and simulation are becoming vital to
designing and improving nanomaterials and
nanodevices. - One of the challenges
-
- Multi-scale
- Scale plays an important role in science and
engineering.
7Multiscale
- Biology/Bioengineering challenge (from top to
bottom) - Peter Hunter
- University of Auckland
Cell
Tissue
(10-3m)
(10-6m)
Organ
(1m)
(10-9m)
Protein
Organ systems
Atom
(10-12m)
8Multiscale
- Physical scales (from bottom to top)
Microscale (10-6m)
Mesoscale (10-3m)
Nanoscale (10-9m)
Macroscale (1m)
Molecular scale
(10-12m)
9Nano-to-Micro Scale Simulations and Modeling
Few microns
Multi-scale Modeling and Simulations
10Computational Nanotechnology
- Common computational methods
- Quantum mechanical calculations
- First principle calculation
- Ab initio
- Etc.
- Molecular methods
- Molecular dynamics/mechanics
- Monte Carlo methods
- Multiscale methods
- Hierarchical multiscale modeling
- Concurrent multiscale modeling
11Quantum mechanical calculations
- Quantum mechanical calculations
- tight binding method, the method of linear
combination of atomic orbitals. - Hatree-Fock approximation
- Density functional theory
- First principle calculations, solving
Schrodingers equation. - Etc.
- Computationally Intensive, O(N4)
- Up to 3000 atoms
-
12Quantum mechanical calculations
Electron Density Profiles
Electronic Transport Y-junction CNTs
13Molecular Dynamics
- Molecular Dynamics
- Based on the Newtonian classical dynamics
- Atoms are viewed as mass points
- Equations of motion can be derived from classical
Lagrangian or Hamiltonian mechanics - Method has received widespread attention since
the 1970 - Liquids
- Defects in crystals
- Fracture
- Surface
- Friction
- others
14Molecular Dynamics
Atomistic Simulation
(without consideration of external forces)
Periodic Boundary Condition
15Molecular Dynamics
- Molecular mechanics potential
i
j
k
j and k are no-bonded
16Molecular Dynamics
- Molecular dynamics simulation
-
(NSF summer institute, Northwestern Unviersity,
2004)
17Molecular Dynamics
- Macroscopic properties
- Can be evaluated based on atomic positions and
velocities
18Molecular Dynamics
- Temperature regulation
- Velocity scaling
- Langevin dynamics
- Berendsen thermostat
- Nose-Hoover thermostat
19Molecular Dynamics
- MD simulations are very important
- nano-to-mesoscale characterization
- of materials
- Typical 10-100,000 atoms
- 10- Groups few million atoms
- 1 group billion atom simulations
- Issues Mechanics,
- Thermal, Chemical,
- Reactivity, Phase diagrams,
- Thermal and Mechanical response
- Vibrations, placement, fatigue
- Plasticity etc
Carbon Nanotube Mechanics
Si Nanowire Mechanics
20Molecular Dynamics
- Why molecular dynamics?
- It is consistent all results are derived from a
classical interatomic potential with a few
parameters - It is predictive equilibrium structures,
reaction transition states, and dynamical
averages are obtained - It is cheap up to billions of atoms for nano
seconds can be simulated - Its downsides
- Quantum bonding and reactive response are hard to
build in - Poorly chosen input potentials produce garbage
outputs - Time step is essentially locked to molecular
vibrations
21Monte Carlo Method
- Monte Carlo Method
- A simple example Evaluation of
-
the number of total sampling the number of hits
in the shaded area
22Monte Carlo Method
- Monte Carlo Method
- Importance sampling
- Partition function
- Distribution function
- Metropolis method
- Set up a Markov chain with transition matrix
- Transition matrix consists of the probability of
transition from one state to another state - Using the transition matrix to control the flow
of the simulations
23Monte Carlo Method
- Monte Carlo simulation in a canonical ensemble
- Constant parameters atom number( ),
- Volume( ), and temperature( )
- Atom displacement
- Randomly select an atom
- Randomly choose a position for the selected atom
- Compute the potential energy change due to
moving the atom to the new position - Accepting criterion
24Monte Carlo Method
- Why Monte Carl method?
- It is consistent all results are derived from a
classical interatomic potential with a few
parameters - It is predictive equilibrium structures,
reaction transition states, and dynamical
averages are obtained - It is cheap
- Random process simulations crystal interface
growth, chemical gradients - Its downsides
- Quantum bonding and reactive response are hard to
build in - Poorly chosen input potentials produce garbage
outputs - Effective importance sampling required
25Monte Carlo Method
- Monte Carlo method versus molecular dynamics
- For some systems in equilibrium state, such as
system in canonical ensemble, both molecular
dynamics and Monte Carlo method can work well - Grand canonical ensemble simulation is easier to
implement with Monte Carlo method than molecular
dynamics - Molecular dynamics allows studying the
time-dependent phenomena
26Kinetic Monte Carlo Method
- Kinetic Monte Carlo method
- The Monte Carlo method, implemented with the
standard Metropolis method, which is powerful for
phase-space explorations, fails to represent the
time evolution of the system. Therefore, it is
mainly used for equilibrium description - The Kinetic Monte Carlo method provides a tool to
simulate a stochastic process with unambiguous
time relationship between Monte Carlo steps and
real-time steps. -
27Monte Carlo Method
- Kinetic Monte Carlo (KMC) method
- Advantages
- KMC simulates dynamics of the system in and out
of equilibrium with a firm correspondence to real
time - At each time step, the system can be recreated.
This allows KMC to model dynamics on large time
scales - Time scale can change automatically
- Disadvantages
- KMC gives a coarse-grained picture of time
evolution - Calculating rates are independent of KMC method
- Need most efficient data structures for each
specific problem
28Computational Nanotechnology
- Limitations of Atomistic (MD) simulation
- Limitations in length scale
- Computationally intensive, limited to small
systems - Rigid/periodic conditions lead to nonphysical
behavior - Cause unrealistic wave reflections (iron
deposition) - Present unrealistic constraints to dislocation
nucleation and movement (nanoindentation) - Artificially dissipate system energy to maintain
constant temperature - Limitations in time scale
- Incapable of simulating material behavior at two
different time scales simultaneously - Incapable of simulating material properties over
large time scales
29Multiscale Methods
- Multiscale methods
- Hierarchical multiscale modeling
- Ortiz and Philips of Caltech
- Klein of Sandia National Laboratory
- Etc.
- Concurrent multiscale modeling
- MAAD, Abraham and co-workers at IBM
- Bridging scale, Liu and co-workers, Northwestern
University - Bridging domain coupling method
- Belytschko (Northwestern University) and Xiao (U
of Iowa) - Etc.
30Hierarchical Multiscale Methods
- Hierarchical Modeling
- The continuum approximation is used to model a
large group of atoms - (Cauchy-Born rule)
- The major drawback is that it is only static
without temperature effect
Arroyo and Belytschko (2002)
Tadmor et al. (1996)
31Concurrent Multiscale Methods
- Concurrent Modeling
- An appropriate model is solved at each length
scale simultaneously while smooth coupling is
implemented between different scales - continuum mechanics for macro elastic media
- molecular dynamics for large groups of atoms
- quantum mechanics for bonds breaking
- The major drawback is the spurious wave
reflection at the interface of two domains with
different dispersive characteristics, as well as
the need in handshake regions of scaling the
computational mesh down to atomistic scales.
32Concurrent Multiscale Methods
- MAAD (Macroscopic, Atomistic and Ab initio
Dynamics) - Quantum mechanics semiempirical tight binding
(TB) - Atomistic statistical mechanics molecular
dynamics (MD) - Continuum mechanics finite element method (FE)
F. F. Abraham et al. Spanning the Length Scales
in Dynamic Simulation Computers in Physics 1998
2538-546
33Concurrent Multiscale Methods
nonphysical spurious wave reflection
34Concurrent Multiscale Methods
- Multiscale issues
- numerical issues
- the finite elements are meshed down to the
atomistic scale in handshake regions. - Time step is governed by the smallest element in
the mesh. Time step will be too small for
continuum region and many time steps will be
wasted. - physical issue
- Pathological wave reflection. The wavelength
emitted by MD region is considerably smaller than
that which can be captured by the continuum FE
region. Wave reflection occurs at the interface
between the MD and FE regions.
35Concurrent Multiscale Methods
- Bridging domain coupling method
Total Hamiltonian
Constraints
S. P. Xiao and T. Belytschko A bridging domain
method for coupling continuum with molecular
dynamics, Computer Methods in Applied Mechanics
and Engineering, 2004 1931645-1669
36Concurrent Multiscale Methods
- Bridging domain coupling method
Bridging domain coupling method can almost
eliminate the high frequency wave reflection
37Computational Nanotechnology
- Bridging domain coupling method multiple time
steps
continuum model
molecular model
38Concurrent Multiscale Methods
- Bridging domain coupling method
- Crack branching
39High-performance computing
- High-performance computing
- Molecular dynamics with high-performance
computing - Billions of atoms can be simulated
- Still has limitations on length and time scales
- A cubic volume of 10-3 µm3 contains billions of
atoms - A nano second needs millions of time steps
One of solutions Multiscale methods with
high-performance computing techniques
40High-performance computing
- High-performance computing
- The bridging domain coupling method is one of the
best candidates which can be extended as a
high-performance computing enhanced multiscale
method
molecular
A bridging domain coupling model
continuum
At each time step, equations of motion are solved
separately in each domain for nodal/atomic
velocities
molecular
continuum
Then, nodal/atomic velocities in the bridging
domain are corrected
41High-performance computing
- Grid-based bridging domain multiscale method
- Multiple-length-scale model
Macroscale Linear finite element
methods Microscale Meshfree particle
methods Nanoscale Molecular dynamics
42High-performance computing
- Grid-based bridging domain multiscale method
- Multiple-time-scale model
43High-performance computing
- Grid-based bridging domain multiscale method
- Domain Decomposition
44High-performance computing
- Grid-based bridging domain multiscale method
- Domain Communication
45High-performance computing
- Grid-based bridging domain multiscale method
- Nano-middleware
46Computational Nanotechnology
- Grid-based bridging domain multiscale method
- Flow chart
47Computational Nanotechnology
- Applications of computational nanotechnology
- Nanoscale materials
- Carbon nanotubes
- Nanotube reinforced nanocomposites
- Nanoscale devices
- Nanoscale oscillator
48Computational Nanotechnology
Iijima et. al. (1996)
Iijima, S. et.al. Nature (1991)
Yakobson et. al. (1996)
Poncharal et. al. (1999)
49Nanoscale Materials
10,10 armchair nanotube
Graphene sheet
R
zigzag nanotube n,0
armchair nanotube n,n
chiral nanotube m,n
50Nanoscale Materials
- Potential functions for carbon-carbon bonds
1. Second-generation Reactive empirical bond
order (Proposed by Brenner et al.)
2. Modified Morse (Proposed by Belytschko and
Xiao)
3. Full Periodic Table Force Field(Proposed by
Rappe et al.)
51Nanoscale Materials
- Mechanical properties of carbon nanotube
force
stress
strain
Youngs modulus
Poissons ratio
52Nanoscale Materials
- Mechanical properties of carbon nanotube
53Nanoscale Materials
- Mechanical properties of carbon nanotube
(a) Youngs modulus (b) Poissons
ratio Size-dependent mechanical properties of
SWNTs
54Nanoscale Materials
- Experiments of carbon nanotube fracture
Yu et. al. Strength and Breaking Mechanism of
Multiwalled Carbon Nanotubes Under Tensile Load,
Science, 2000
55Nanoscale Materials
- Fracture of carbon nanotube
- (Belytschko, Ruoff, Schatz and Xiao)
- Fracture of nanotubes is of great interest
because - A nanotube was considered a perfect molecule --
an opportunity to study fracture in an idealized
setting - Nanotubes were first predicted to have very high
strength which is 300 GPa (Yakobson et al. 1997) - The low strength is obtained from the experiments
and it is in range of 11 to 63 GPa (Yu et al.
2000)
56Nanoscale Materials
- Fracture of carbon nanotube
20,0 zigzag nanotube
57Nanoscale Materials
- Fracture of carbon nanotube
Failure strength is dependent on chirality
16,8 chiral nanotube
16,4 chiral nanotube
58Nanoscale Materials
- Fracture of carbon nanotube
Failure strain
Failure stress
30
300GPa
Theoretical studies (Brenner) Yakobson et
al.(1997)
170GPa
Kinetic theory study Samsonidze et al.(2002)
17
15.7
93.5GPa
Numerical simulations Belytschko and Xiao (2002)
213
1163GPa
Experimental results Yu et al.(2000)
59Nanoscale Materials
- Fracture of carbon nanotube
Frequency of occurrence in the experiments (Yu et
al, 2000)
(We do not believe that the scatter is due to
experimental error)
(GPa)
Nanotube fracture is governed by defects
60Nanoscale Materials
- Fracture of carbon nanotube
- Defects in the nanotubes
- Chemical Defects
- Oxidization (Mawhinney et al, 2000)
- Chemical vapor deposition (CVD) (Hafner et al,
1998) - Topological Defects
- Stone-wales defects (5/7/7/5 dislocation)
(Yakobson, 1998) - Vacancy Defects
- Irradiation effects (impacted by high energy
electrons) such as in SEM environment. (Banhart,
1999) - Vacancies in the original nanotube shell.
61Nanoscale Materials
- Fracture of carbon nanotube Vacancy defects
62Nanoscale Materials
- Fracture of carbon nanotube Vacancy defects in
320,0 nanotubes
Failure strain()
Failure stress (GPa)
N-atom defect
15.7
93.5
0
8.00
64.1
2
6.00
50.3
4
4.95
42.1
6
4.35
36.9
8
3.20
27.7
16
2.30
20.2
32
63Nanoscale Materials
- Fracture of carbon nanotube
64Nanoscale Materials
- Multiscale modeling for nanotube fracture
65Nanoscale Materials
- Nanocomposites
- Extraordinary properties
- High stiffness High strength Low density.
- Inclusions
- Nanotubes Nanoplates Nanoclay Nanoparticles
Etc. - Matrix
- Polymer Ceramics Metals.
- Key issues
- Dispersion of CNTs Alignment of CNTs Load
transfer.
66Nanoscale Materials
- carbon nanotube reinforced aluminum composites
Carbon Nanotube E 1000GPa Aluminum
E 70GPa
T. Kuzumaki et al. (1998) Experiment Tensile
test pieces L 15 mm ,D 3 mm Tensile strength
was increased by 100 TEM observations have
shown (i ) The composites are not damaged during
the composite preparation (ii) No reaction
products at the nanotube/Al interface are visible
67Nanoscale Materials
- carbon nanotube reinforced aluminum composites
Infinite Domain
O
O
RVE 1
Both cases implement periodic boundary
conditions on x,y,z
68Nanoscale Materials
- carbon nanotube reinforced aluminum composites
Strain
It shows that the discontinuous CNT can not
improve the strength of composites. Therefore,
the following study mainly focus on the
continuous CNT.
69Nanoscale Materials
- carbon nanotube reinforced aluminum composites
Axial load 001
70Nanoscale Materials
- carbon nanotube reinforced aluminum composites
Volume Change Schema Tension on Z direction
x
71Nanoscale Devices
- Nanoscale devices
- Nanoelectronics
- Molecular wires
- Diodes
- Field-effect transistor
- Etc.
- Nanotube-based devices
- Nanotweezers
- Nanocantilever
- Nanogears
- Etc.
72Nanoscale Materials
73Nanoscale Materials
- Nanotube-based oscillators
- High frequency up to 100 GHz
74Nanoscale Materials
- Nanotube-based oscillators
Frequency prediction
75Nanoscale Materials
- Nanotube-based oscillators
76Nanoscale Materials
- Nanotube-based oscillators temperature effects
Effecitve interlayer friction 100K, 0.0158 pN
per atom 300K, 0.0398 pN per atom
1000K, 0.1008 pN per atom
77Nanoscale Materials
- Nanotube-based oscillators a NanoElectroMechanica
l System(NEMS)
10 GHz
25 GHz
78Computational Nanotechnology
- Conclusions
- Computational nanotechnology plays an important
role during the development of nanotechnology - Although molecular methods are powerful in
illustrating complex physical phenomena, they
have limitations on length and time scales - Currently, multiscale modeling and simulation is
a key research area in computational
nanotechnology - The combination of multiscale methods and
high-performance computing technique, such as
Grid computing, will have potential in
nanotechnology
79Nanotechnology at The University of Iowa
80NanoEngineering_at_Iowa
81Nanotechnology at NASA Moors Law
82Nanotechnology at NASA Miniaturization
Advanced miniaturization, a key thrust area to
enable new science and exploration
missions - Ultrasmall sensors, power sources,
communication, navigation, and propulsion
systems with very low mass, volume and
power consumption are needed Revolutions
in electronics and computing will allow
reconfigurable, autonomous, thinking
spacecraft Nanotechnology presents a whole new
spectrum of opportunities to build device
components and systems for entirely new space
architectures
Micromachines Microrovers Micromanufacturing