Atomistix ToolKitAtomistix ToolKitVirtual NanoLabVirtual NanoLabNanoLanguageNanoLanguage
© QuantumWise A/S 20092009-02-12Ver. 2008.10.0
Junctions Interfaces Surfaces
PrefacePreface
Many application areas in nanotechnologyare related to effects occurring at
It is critical to be able to accurately modelsuch phenomena from quantum theory
QuantumWise offers unique software tools for these systems, centered around an open platform architecture
QuantumWise offers a software platform for atomistic simulations of nanoscale systems
» Atomistix ToolKit (ATK)• State-of-the-art DFT engine for electronic
structure and transport calculations
» Virtual NanoLab (VNL)• Modern GUI for setup and analysis
» NanoLanguage• Scripting language interface to ATK• Fully integrated in VNL
IntroductionIntroduction
Current-voltage characteristics of functionalized carbon nanotubes
Contact resistance and capacitance of metal-nanotube and nanotube-nanotube contacts; Schottkybarrier
Examples of application areas (1/2)Examples of application areas (1/2)
Graphene field-effect transistor
Current-voltage characteristics of molecular junctions and tunneling devices (rectification, NDR)
Reaction paths on surfaces for catalysis
Defect states in semiconductors, nanowires, nanotubes
Examples of application areas (2/2)Examples of application areas (2/2)
Spin-dependent transport across crystalline magnetotunnel junctions
Leakage currents in MOS structuresHigh-k dielectrics, complex interfaces
Selected publications 2008Selected publications 2008--2009 with ATK2009 with ATK
Graphene» OuYang et al., Transport properties of T-shaped and crossed junctions based on graphene nanoribbons, Nanotechnology 20, 055202 (2009)» Lam et al., An ab initio study on energy gap of bilayer graphene nanoribbons with armchair edges, APL 92, 223106 (2008)» Li et al., Role of symmetry in the transport properties of graphene nanoribbons under bias, PRL 100, 206802 (2008)» Sevincli et al., Superlattice structures of graphene-based armchair nanoribbons, PRB 78, 245402 (2008)
Nanotubes» Yang et al., Negative differential resistance in single-walled SiC nanotubes, Chinese Science Bulletin 53, 3770 (2008)» Li et al., Ab initio investigations of the transport properties of Haeckelite nanotubes, J Phys Cond Mat 20, 415207 (2008) » Abadir et al., Effect of single-biomolecule adsorption on the electrical properties of short carbon nanotubes, Nano’08: 8th IEEE Conference
on Nanotechnology (2008)» Bai Ping et al., Carbon nanotube Schottky diode: an atomic perspective, Nanotechnology 19, 115203 (2008)» Compernolle et al., Conductance of a copper-nanotube bundle interface: Impact of interface geometry and wave-function interference, PRB
77, 193406 (2008)» Girard et al., Current-induced forces in conducting and semiconducting carbon nanotubes, J Phys Conf Series 100, 052063 (2008)» Lee et al., Electron transport through carbon nanotube in intramolecular heterojunctions with peptide linkages, Phys Chem Chem Phys 10,
5225 (2008)» Khazaei et al., Designing Nanogadgets by Interconnecting Carbon Nanotubes with Zinc Layers, ACS Nano 2, 939 (2008)» Xu et al., Large Negative Differential Resistance in a Molecular Junction of Carbon Nanotube and Anthracene, J Phy Chem B 112, 16891
(2008)
Spintronics (incl. graphene and nanotubes)» Sahin and Senger, First-principles calculations of spin-dependent conductance of graphene flakes, PRB 78, 205423 (2008)» Min et al., CrAs(001)/AlAs(001) heterogeneous junction as a spin current diode predicted by first-principles calculations, J. Mag & Mag Mat
321, 312 (2009)» Girard et al., Spin-Dependent Electron Transport Induced by Non-Magnetic Adatoms in Metallic Carbon Nanotubes, e-Journal of Surf Sci &
Nanotech, 6, 157 (2008)» Sekiguchi et al., Observation of a bias-dependent constrained magnetic wall in a Ni point contact, PRB 78, 225518 (2008)» Yao et al., First-principles study of transport of V doped boron nitride nanotube, Phys Lett A 372, 5609 (2008)» Zhou et al., One-dimensional iron-cyclopentadienyl sandwich molecular wire with half metallic, negative differential resistance and high-
spin filter efficiency propertie, J Am Chem Soc 130, 4023 (2008)
Selected publications 2008Selected publications 2008--2009 with ATK2009 with ATK
Molecular electronics» Akdim & Pachter, Switching behavior in pi-conjugated molecules bridging nonmetallic electrodes: A density functional theory study,
J Phys Chem C 112, 3170 (2008)» Fan et al., Theoretical investigation of the negative differential resistance in squashed C60 molecular device, APL 92, 263304 (2008)» Ford et al., Rectification in donor-acceptor molecular junctions, J Phys Cond Mat 20, 374106 (2008)» Kaun & Seideman, Conductance, contacts, and interface states in single alkanedithiol molecular junctions, PRB 77, 033414 (2008)» Li et al., Theoretical investigation on electron transport properties of a single molecular diode, J Mol Struc: THEOCHEM 867, 59 (2008)» Li et al., First-principles study of substituents effect on molecular junctions: Towards molecular rectification, Comp Mat Sci 42, 638 (2008)» Long et al., Negative differential resistance behaviors in porphyrin molecular junctions modulated with side groups, APL 92, 243303 (2008)» Okuno & Yokoyama, Theoretical study of molecular rectification in porphyrin dimer, Thin Solid Films 516, 2630 (2008)» Sen & Chakrabarti, How Important Is the Position of the Molecule between the Electrodes in Tuning Negative Differential Resistance
Behavior?, J Phys Chem C 112, 1553 (2008)» Sen & Chakrabarti, Ab Initio Density Functional Study on Negative Differential Resistance in a Fused Furan Trimer, J Phys Chem C 112, 1685
(2008)» Zhao et al., A possible anthracene-based optical molecular switch driven by a reversible photodimerization reaction, APL 93, 013113 (2008)» Zhao et al., 15,16-Dinitrile DDP/CPD as a possible solid-state optical molecular switch, Chem Phys Lett 453, 62 (2008)» Stadler et al., A theoretical view of unimolecular rectification, J Phys Cond Matt 20, 1840 (2008)» Gemma et al., Quantum Interference in Acyclic Systems: Conductance of Cross-Conjugated Molecules, J Am Chem Soc 130, 17301 (2008)
Nanowires» Liao et al., Studying the strain effect on silicon atomic wires, J Korean Phys Soc 53, 3655 (2008)» Ng et al., Geometry dependent I-V characteristics of silicon nanowires, Nano Letters 8, 3662 (2008)» Yang et al., Compositional ordering and quantum transport in Mo6S9-xIx nanowires: Ab initio calculations, PRB 77, 165426 (2008)» Zhou et al., First-principles study on the differences between the equilibrium conductance of carbon and silicon atomic wires, J Phys Cond
Mat 20, 045225 (2008)
Other» Wang et al., Excess-silver-induced bridge formation in a silver sulfide atomic switch, APL 93, 152106 (2008)» Geng et al., Ab Initio Modeling of Schottky-Barrier Height Tuning by Yttrium at Nickel Silicide/Silicon Interface, IEEE Electron Device Letters
29, 746 (2008)» Kamiya et al., Interface electronic structures of zinc oxide and metals: First-principle study, physica status solidi (a) 205 (2008)
Traditional quantum-based software packages can model either isolated molecules or periodic systems
ATK is the only commercial software that can model complex nanostructures that combine molecules with
periodic systems and macroscopic elements
GaussianDMOLTurboMol
VASPCASTEPFLAPWWien2k
Uniqueness of ATKUniqueness of ATK
Main capabilitiesMain capabilities
Spin-dependent electronic structure» Molecules (and supercells)» Periodic systems (crystals, nanowires, nanotubes)» Quasi-periodic systems (slabs)
Spin-dependent quantum transport calculations in homo- or heterogeneous two-probe systems under finite bias
» Transmission probability, conductance, I-V characteristics, transmission eigenstates, spin-torque transfer
Qualitative simulation of electrostatic gateReaction pathways & transition states via nudged elastic bands (NEB)Geometry optimization & MDParallelized
» Optimized for transport calculations» Linear scaling for transmission calculations
Virtual experimentsVirtual experiments
Compute quantities that can be compared to experiments
» Current–voltage characteristics» Co-linear spin torque transfer» Conductance» Bias-induced forces» Reaction pathways & activation energies
Analyzing quantum transport mechanismsAnalyzing quantum transport mechanisms
Conducting and non-conductingeigenchannels in a perturbedatomic wire
ATK provides several analysis tools for understanding transport mechanisms
» K-point resolved transmission» Transmission spectrum and surface DOS» Molecular projceted self-consistent Hamiltonian
eigenstates» Transmission eigenvalues and eigenchannels
LargeLarge--scalescale systemssystems
Nanowires, multiwall nanotubes and complex interface structures can easily contain several hundred atoms per unit cellIn order to treat effects from defects (including doping), very large unit cells are also requiredATK can handle very large systems (over 1,000 atoms)
Si fcc
O
TwoTwo--probe systems, electrodesprobe systems, electrodes
A two-probe system consists of» left and right electrodes» a central scattering region
Each electrode is a semi-infinite periodic system
» Cleaved bulk crystal (eg. a gold [111] surface)
» Carbon nanotube, graphene, atomic chain, etc
The two electrodes can be different
» Two different metals» Two different carbon nanotubes» Metal–nanotube contact» Different spin-polarization
(magnetic tunnel junctions)
Mg OCoSi
Mn
Parallel or Anti-Parallel
Scattering regionScattering region
The scattering region can be ... anything!
» A molecule (e.g. between two metal surfaces)
» A piece of a carbon nanotube (e.g. a metal–nanotube–metal contact)
» A graphene nanoribbonzigzag/armchair contact
» A periodic structure (layered interfaces)
Interface boundary conditionsInterface boundary conditions
Periodic boundary conditions are applied in the transverse plane
» Makes it simple to model real interfaces, e.g. with surface impurities
» Vacuum padding for 1D electrodes like nanotubes and wires
HfO2
Ru Ru
Ni graphite Ni
Transport boundary conditionsTransport boundary conditions
Scattering boundary conditions apply in the transport direction
» Electrons flow through the system under the influence of an applied voltage bias
» Non-equillibrium electron distribution described by NEGF
» Electric current is due to ballistic, coherent tunneling, computed using the Landauerformalism
Spintronics and magnetismSpintronics and magnetism
In addition to the electronic current, ATK can compute the
» spin current » spin torque transfer (STT)» tunnel magneto resistance (TMR)
Co Cu Co
Transistor simulationsTransistor simulations
The ability of ATK to simulate an electrostatic gate enables calculations of transistor characteristics such as» Ambipolar characteristic» On-current» On/off current ratio» Leakage current» Sub-threshold swing
Gate is introduced as a pure shift of the central Hamiltonian elements B. Huang et al.,
APL 91, 253122 (2007)
Surface reactions and catalysisSurface reactions and catalysis
Reactant Transition State 1 Transition State 2
Product’ Transition State 3 Product
Via an open source NEB (nudged elastic bands) module, ATK can compute reaction pathways and activation energies for catalytic reactions on surfaces
Detailed list of features, Atomistix ToolKitDetailed list of features, Atomistix ToolKit
Electronic structure calculations» SIESTA-based method, plus inclusion of indirect atom
pairs for improved accuracy» Up to double-zeta double polarized (DZDP) numerical
atomic orbital basis sets, with detailed user control of basis set parameters
» Norm-conserving (Troullier-Martins) pseudopotentials provided for all elements, with possibility to use customized pseudopotentials
» LDA-PZ, GGA-PBE GGA-revPBE exchange correlation potentials (also spin-polarized)
» Fermi level smearing for improved convergence stability» Customizable Broyden/Pulay mixing for self-consistent
scheme» Monkhorst-Pack k-point sampling grids
Calculation of» Molecular spectra» Band structure of periodic structures» Eigenfunctions (molecular orbitals, Bloch functions)» Mulliken population analysis» Real-space electron density and effective potential» Forces (analytic Hellmann-Feynman)
Ion dynamics» Quasi-Newton or steepest descent method» Constraints: fixed positions» Nudged Elastic Band (NEB) transition state analysis
Transport calculations» Two-probe systems NEGF using the TranSIESTA
method (self-energy coupling to semi-infinite leads)» Green's function description of non-equilibrium
electron distribution in scattering region» Finite bias» Ability to treat heterogeneous systems through multi-
grid solution to Poisson's equation» Simulation of electrostatic gate
Transport analysis» Transmission coefficients (k-point resolved) and
transmission spectrum» Spin-dependent current and conductance/resistance» Transmission eigenvalues and real-space
eigenchannels» Surface density of states» Real-space local density of states» Fast evaluation of linear response current» Collinear spin-torque transfer» Molecular Projected Self-Consistent Hamiltonian
eigenvalues and real-space eigenfunctions
Work-flow control» Store and restore the state of a calculation for
deferred analysis, restart, or initialization of a new calculation via the self-consistent density matrix
Fully programmable object-oriented Python scripting interface to ATK» Transparent – the Python syntax is simple yet powerful, with good structural
overview» Modularized» Extendable – integrate your own or
third-party algorithms with ATK
Tying the platform together» Integrated with the GUI, Virtual NanoLab» Wrapping of third-party codes
Interpreted language» Can be used interactively» Comes with “batteries included”
(scientific modules)» Cross-platform compatible scripts» All performance-critical parts of ATK are written in C++/Fortran
Online user Forum for tips, tricks, script resources, etc
NanoLanguageNanoLanguage
© Frank Stajano
NanoLanguage offersNanoLanguage offers
Control of parameters» Loop over numerical parameters» Set up parameterized geometries
Control of flow» Automatically converge in accuracy parameters» Decision-making scripts
Access to data in native Python format» Customized analysis» Export data in any desired format
Extendability» Implement new algorithms, backed by ATK» Wrap other software packages» Anything you otherwise can do in Python, applied to quantum
chemistry!
VNLVNL
Interactive builders for » Molecules
» Crystals
» Nanotubes
» Two-probe systems (general and specialized, like MTJs)
3D visualization of structures and results» Electron density, potentials, etc, as isosurfaces or contour plots
» Superimpose plots on geometry
» Export plots as images in common formats
Interactive generation and export of NanoLanguage scripts» Structures and geometries
» Calculations and analysis
Internal NanoLanguage interpreter» Import user-defined geometry scripts for visualization or calculation setup
» Built-in script editor
Local execution of NanoLanguage scripts via drag-and-drop
MoleculesMolecules
Build complex molecules with a few mouse-clicks with the Molecular Builder
» Bond angles and lengths are automatically adjusted according to specified hybridization and bond order
» Fine-tune the geometry manually» Hydrogen added automatically» Attach pre-built side-groups from the
Crystal CupboardCalculate and plot molecular spectraVisualize molecular wave functions and electron densities in the Nanoscope
CrystalsCrystals
Browse the Crystal Cupboard for over 500 pre-built crystal templatesBuild your own crystal from scratch in the Bulk Builder
» Create slab models» Make supercells & slab models
Calculate and plot the spin-dependent band structureVisualize Bloch states
Nanotubes & GrapheneNanotubes & Graphene
Build nanotubes (C, Si, Au, …) in the Nanotube Grower with a few clicksUse scripting coupled with the GUI to set up
» boron-nitride nanotubes» graphene nanoribbons
Introduce defects or ad-atoms and study the influence on the band structureCreate nanotube or graphene two-probe systems to study current/voltage characteristics, spin-polarized transport, etc
TwoTwo--probe systemsprobe systems
Build two-probe systems easily in the Atomic Manipulator
» Cleave any crystal by specifying the Miller indices and surface cell
» Align and rotate entire molecules in the scattering region
» Control over individual atom positions, delete/add atoms
Also support for 1D electrodes» Study transport properties of
functionalized nanotubes, etc
Magnetic Tunnel Junction BuilderMagnetic Tunnel Junction Builder
Instantly generate advanced MTJ two-probe geometriesSupport for any FeCo type materials (bcc [100] surfaces)Detailed control over all layer separationsBucklingFine-tune or introduce defects in the Atomic Manipulator
O Vacancy
Fe MgO Fe
NanoLanguage in VNLNanoLanguage in VNL
VNL is designed to help the user learn and master NanoLanguageBuilders generate atomic geometries represented as NanoLanguage scripts
» Drag-and-drop to other instruments in VNL
» Drag-and-drop to editor (internal or external) for direct editing
» Use generated scripts as templates for advanced geometries
Verify and troubleshoot custom code» Define custom geometries externally in
script, then drop code directly on Nanoscope for visualization
Drop scripts on Job Manager to execute script in ATK
» Runs in dedicated thread for optimal performance
» Restricted to local hostDrop geometry on NanoLanguage Scripter to set up the calculation
» See next slide!
NanoLanguage ScripterNanoLanguage Scripter
Use the NanoLanguage Scripter to generate complete calculation scripts, ready for running, in just a few stepsSupports
» All ATK keywords, including I/O» All analysis options» Geometry optimization
Exports nicely formatted scripts, ready-to-run even in parallel
» Use generated scripts as templates for more advanced coding
PerformancePerformance
State-of-the-art parallel DFT code» MPI parallelization for k-points and energy points» OpenMP for threading on multi-core CPUs» Optimized code & libraries provide
excellent performance and scaling
Examples:» MgO bulk supercell with 1,024 atoms
converges in 16 hours on a single node, using 3 Gb memory (DZP basis set)
» Au/Si/Au two-probe with 1,166 atoms, 24 hours on 10 nodes, 2 Gb (SZ basis set)
System TSUBAME grid cluster
Computer Sun Fire X4600 with Infiniband interconnects
CPU Dual-core AMD Opteron 880 @ 2.4GHz 8CPUs (16 cores) + 32 Gb RAM / node
OS SUSE Linux Enterprise Server 9 SP 3
ModelElectrodes: Au(100)Molecule: Si nanowire
Basis set DoubleZetaPolarized
Exchange-correlation LDA-PZ
Mesh cut-off 150 Ry
k-point sampling (SCF) (5, 5, 500)
k-point sampling(DOS, Transmission)
(10, 10)
Model: Si nanowire between Au surfaces
Parallel scalingParallel scaling
Supercluster test, ATK 2008.10Self-consistent calculation scales well up to 32 nodesAlmost linear scaling for transmission/current calculations up to >100 nodes
Test system
0
32
64
96
128
0 32 64 96 128
CPU Cores
Spee
d-up
SCFAnalysisTotalLinear
0
40
80
120
160
200
1 4 8 16 32 64 128CPU Cores
Tim
e (h
ours
) Total time = SCF + Analysis
Parallel scaling, detailed resultsParallel scaling, detailed results
Calculation time Speed-up
SCF Total SCF Total
1 67h52m 106h51m 174h43m
46h56m
22h56m
14h09m
8h32m
5h59m
1 1
4h41m
1
3.7
7.6
12.3
20.5
29.2
37.3
3.5
8.1
10.7
14.7
17.0
18.8
19h27m
8h24m
6h19m
4h37m
4h00m
3h37m
CPU cores Physical
propertiesPhysical
properties
4 27h29m 3.9
8 14h32m 7.4
16 7h50m 13.6
32 3h55m 27.3
64 1h59m 53.9
128 1h04m 100.2
Supported platformsSupported platforms
Atomistix ToolKit» Linux
• 32-bit version, optimized for i686• 64-bit version, optimized for x86_64• Parallelized using MPICH2 (statically linked)
» Windows• 32-bit version for XP, Vista, 2000
Virtual NanoLab» 32-bit version for Linux/Windows i686» Also runs on x86_64
FLEXlm license manager» Node-locked or floating licenses
ATK and VNL run on most Linux distributions, with official support for
• RedHat Enterprise Linux (4.x, 5.x)• SUSE Linux Enterprise (9.x, 10.x)
BEST CHOICE FOR HIGH PERFORMANCE
Want to learn more?Want to learn more?Please visit our websitePlease visit our website
http://http://www.quantumwise.comwww.quantumwise.comto download a to download a FREEFREE trial versiontrial version