Car-Parrinello Method and Applications Moumita Saharay Jawaharlal Nehru Center for Advanced Scientific Research, Chemistry and Physics of Materials Unit,

Slides:



Advertisements
Similar presentations
Time averages and ensemble averages
Advertisements

Simulazione di Biomolecole: metodi e applicazioni giorgio colombo
Introduction to Computational Chemistry NSF Computational Nanotechnology and Molecular Engineering Pan-American Advanced Studies Institutes (PASI) Workshop.
Molecular Bonds Molecular Spectra Molecules and Solids CHAPTER 10 Molecules and Solids Johannes Diderik van der Waals (1837 – 1923) “You little molecule!”
Molecular dynamics modeling of thermal and mechanical properties Alejandro Strachan School of Materials Engineering Purdue University
Molecular Modeling of Structure and Dynamics in Fuel Cell Membranes A. Roudgar, Sudha N.P. and M.H. Eikerling Department of Chemistry, Simon Fraser University,
Transfer FAS UAS SAINT-PETERSBURG STATE UNIVERSITY COMPUTATIONAL PHYSICS Introduction Physical basis Molecular dynamics Temperature and thermostat Numerical.
Modelling of Defects DFT and complementary methods
1 A molecule of ammonia NH 3 is made up of one nitrogen and three hydrogen atoms: Coordinate bond The nitrogen atom forms three bonds and the hydrogen.
Workshop on HPC in India Chemical Dynamics in Aqueous Media Amalendu Chandra Department of Chemistry, Indian Institute of Technology Kanpur, India
Molecular Bonding Molecular Schrödinger equation
Solvation Models. Many reactions take place in solution Short-range effects Typically concentrated in the first solvation sphere Examples: H-bonds,
Introduction to Molecular Orbitals
Computational Chemistry
Computational Solid State Physics 計算物性学特論 第2回 2.Interaction between atoms and the lattice properties of crystals.
Ab initio Calculations of Interfacial Structure and Dynamics in Fuel Cell Membranes Ata Roudgar, Sudha P. Narasimachary and Michael Eikerling Department.
Quantum Mechanics Discussion. Quantum Mechanics: The Schrödinger Equation (time independent)! Hψ = Eψ A differential (operator) eigenvalue equation H.
Conical Intersections Spiridoula Matsika. The study of chemical systems is based on the separation of nuclear and electronic motion The potential energy.
The total energy as a function of collective coordinates, ,  for upright and tilted structures are plotted. Two types of frequency spectrum are calculated:
Ab initio and Classical Molecular Dynamics Simulations of Supercritical Carbon Dioxide Moumita Saharay and S. Balasubramanian Jawaharlal Nehru Center for.
The Calculation of Enthalpy and Entropy Differences??? (Housekeeping Details for the Calculation of Free Energy Differences) first edition: p
Overview of Simulations of Quantum Systems Croucher ASI, Hong Kong, December Roberto Car, Princeton University.
D G Kanhere, Center for Simulation and modeling Pune University, Mastani School- IISER Pune, July 2014.
Joo Chul Yoon with Prof. Scott T. Dunham Electrical Engineering University of Washington Molecular Dynamics Simulations.
Thermal Properties of Crystal Lattices
Crystal Lattice Vibrations: Phonons
Quantum Monte Carlo for Electronic Structure Paul Kent Group Meeting - Friday 6th June 2003.
Quantum Calculations B. Barbiellini Thematics seminar April 21,2005.
Vibrational Spectroscopy
Computational Chemistry
Algorithms and Software for Large-Scale Simulation of Reactive Systems _______________________________ Ananth Grama Coordinated Systems Lab Purdue University.
Norm-conserving pseudopotentials and basis sets in electronic structure calculations Javier Junquera Universidad de Cantabria.
The Nuts and Bolts of First-Principles Simulation Durham, 6th-13th December : DFT Plane Wave Pseudopotential versus Other Approaches CASTEP Developers’
ChE 452 Lecture 24 Reactions As Collisions 1. According To Collision Theory 2 (Equation 7.10)
Physics “Advanced Electronic Structure” Density Functional Molecular Dynamics Contents: 1. Methods of Molecular Dynamics 2. Car-Parrinello Method.
Rosa Ramirez ( Université d’Evry ) Shuangliang Zhao ( ENS Paris) Classical Density Functional Theory of Solvation in Molecular Solvents Daniel Borgis Département.
1.Solvation Models and 2. Combined QM / MM Methods See review article on Solvation by Cramer and Truhlar: Chem. Rev. 99, (1999)
Water layer Protein Layer Copper center: QM Layer Computing Redox Potentials of Type-1 Copper Sites Using Combined Quantum Mechanical/Molecular Mechanical.
Chemical Reaction on the Born-Oppenheimer surface and beyond ISSP Osamu Sugino FADFT WORKSHOP 26 th July.
Molecular Dynamics simulations
Ab initio molecular dynamics via the Car-Parrinello method: Basic ideas, theory and algorithms Mark E. Tuckerman Dept. of Chemistry and Courant Institute.
Molecular simulation methods Ab-initio methods (Few approximations but slow) DFT CPMD Electron and nuclei treated explicitly. Classical atomistic methods.
7. Lecture SS 2005Optimization, Energy Landscapes, Protein Folding1 V7: Diffusional association of proteins and Brownian dynamics simulations Brownian.
Algorithms and Software for Large-Scale Simulation of Reactive Systems _______________________________ Metin Aktulga, Sagar Pandit, Alejandro Strachan,
Ab initio path integrals and applications of AIMD to problems of aqueous ion solvation and transport Mark E. Tuckerman Dept. of Chemistry and Courant Institute.
Fundamentals of DFT R. Wentzcovitch U of Minnesota VLab Tutorial Hohemberg-Kohn and Kohn-Sham theorems Self-consistency cycle Extensions of DFT.
Atoms in Combination: The Chemical Bond Chapter 10 Great Idea: Atoms bind together in chemical reactions by the rearrangement of electrons.
Quantum Mechanics/ Molecular Mechanics (QM/MM) Todd J. Martinez.
Role of Theory Model and understand catalytic processes at the electronic/atomistic level. This involves proposing atomic structures, suggesting reaction.
How do you build a good Hamiltonian for CEID? Andrew Horsfield, Lorenzo Stella, Andrew Fisher.
Advanced methods of molecular dynamics 1.Monte Carlo methods 2.Free energy calculations 3.Ab initio molecular dynamics 4.Quantum molecular dynamics 5.Trajectory.
Computational Physics (Lecture 11) PHY4061. Variation quantum Monte Carlo the approximate solution of the Hamiltonian Time Independent many-body Schrodinger’s.
Chapter 6 Biology The Chemistry of Life. 6.1 Elements Elements are substances that can’t be broken down into simpler substances Elements are substances.
Comp. Mat. Science School Electrons in Materials Density Functional Theory Richard M. Martin Electron density in La 2 CuO 4 - difference from sum.
Molecular Bonding Molecular Schrödinger equation
SPS1. Obtain, evaluate, and communicate information from the Periodic Table to explain the relative properties of elements based on patterns of atomic.
Solid state physics Lecture 3: chemical bonding Prof. Dr. U. Pietsch.
Maintaining Adiabaticity in Car-Parrinello Molecular Dynamics
Srinivasan S. Iyengar Department of Chemistry, Indiana University
Atomistic simulations of contact physics Alejandro Strachan Materials Engineering PRISM, Fall 2007.
Atomistic materials simulations at The DoE NNSA/PSAAP PRISM Center
Algorithms and Software for Large-Scale Simulation of Reactive Systems
Electronic Structure and First Principles Theory
Prof. Sanjay. V. Khare Department of Physics and Astronomy,
Masoud Aryanpour & Varun Rai
龙讯教程 How to simulate ultrafast dynamics using rt-TDDFT in Pwmat?
Molecular simulation methods
Algorithms and Software for Large-Scale Simulation of Reactive Systems
Car Parrinello Molecular Dynamics
The Atomic-scale Structure of the SiO2-Si(100) Interface
Presentation transcript:

Car-Parrinello Method and Applications Moumita Saharay Jawaharlal Nehru Center for Advanced Scientific Research, Chemistry and Physics of Materials Unit, Bangalore.

Outline Difference between MD and ab initio MD Why to use ab initio MD ? Born-Oppenheimer Molecular Dynamics Car-Parrinello Molecular Dynamics Applications of CPMD Disadvantages of CPMD Other methods Conclusions

DFT, MD, and CPMD Properties of liquids/fluids depend a lot on configurational entropy MD with improved empirical potentials DFT calculation of a frozen liquid configuration Configurational Entropy part of the free energy will be missing in that case. Ab initio MD offers a path that mixes the goodness of both MD and of DFT AIMD is expensive.

Molecular simulations  Classical MD Hardwired potential No electronic degrees of freedom No chemical reaction Accessible length scale ~100 Å Accessible time scale ~ 10 ns  Ab initio MD On-the-fly potential Electronic degrees of freedom Formation and breaking of bonds Accessible length scale ~ 20 Å Accessible time scale ~ 10 ps

Livermore’s Nova LaserSandia National Laboratories Z accelerators A short intense shock caused the hydrogen to form a hot plasma and become a conducting metal The experiments found different compressibilities which could affect the equation of state of hydrogen and its isotope Quantum simulations could give the proper reasons for different results Conditions of the Nova and Z flyer were different : Time scales of the pulse were different

Why ab Initio MD ? Chemical processes Poorly known inter atomic interactions e.g. at high Pressure and/or Temperature Properties depending explicitly on electronic states ; IR spectra, Raman scattering, and NMR chemical shift Bonding properties of complex systems

Born-Oppenheimer approximation Electronic motion and nuclear motion can be separated due to huge difference in mass Different time scale for electronic and ionic motion Fast electrons have enough time to readjust and follow the slow ions

Born-Oppenheimer MD Electron quantum adiabatic evolution and classical ionic dynamics Effective Hamiltonian : H o I → Ionic k.e. and ion-ion interaction 2 nd term → Free energy of an inhomogeneous electron gas in the presence of fixed ions at positions (R I ) Electronic ground state – electron density ρ(r) – F({R I }) min Born-Oppenheimer Potential Energy Surface

Minimization to BO potential surface E{ρ(r)} ρ(r)ρ 0 (r)

Born-Oppenheimer MD Forces on the ions due to electrons in ground state Ionic Potential Energy Ψ i (r) one particle electron wave function 1 st → Electronic k.e. ; 2 nd → Electrostatic Hartree term 3 rd → integral of LDA exchange and correlation energy density ε xc 4 th → Electron-Ion pseudopotential interaction ; 5 th → Ion-Ion interaction

Born-Oppenheimer MD Electronic density; f i → occupation number E eI → Electron-Ion coupling term includes local and nonlocal components Kohn-Sham Hamiltonian operator Time evolution of electronic variables Time dependence of H ks ← slow ionic evolution given by Newton’s equations U ks = minimum of E ks w.r.t. ψ i -

Merits and Demerits of BOMD Advantages Disadvantages True electronic Adiabatic Evolution on the BO surface Need to solve the self- consistent electronic-structure problem at each time step Minimization algorithms require ~ 10 iterations to converge to the BO forces Poorly converged electronic minimization → damping of the ionic motion Computationally demanding procedure

Car-Parrinello MD CP Lagrangian Ψ i → classical fields μ → mass like parameter [1 Hartree x 1 atu 2 ] 4 th → orthonormality of the wavefunctions Constraints on the KS orbitals are holonomic No dissipation

Choice of μ Folkmar Bornemann and Christof Schutte demonstrate If the gap between occupied and unoccupied states = 0 (Insulators and semiconductors) (Metals) Fictitious kinetic energy of the electrons grow without control Use electronic thermostat μ must be small → small integration time step μ ~ 400 au, time step ~ 0.096x s

CP Equations of motion Equations of motion from L cp : Ionic time evolution Electronic time evolution Constraint equation Boundary conditions

Hellmann-Feynman Theorem If Ψ is an exact eigenfunction of a Hamiltonian H, and E is the corresponding energy eigenvalue : λ is any parameter occurring in H For an approximate wavefunction Ψ For an exact Ψ

Force on Ions G I → constraint force + G I When, ψi is an eigenfunction Force on the ions due to electronic configuration, when electronic wavefunction is an eigen function is zero

Constants of motion Vibrational density of states of electronic degrees of freedom Comparison with the highest frequency phonon mode of nuclear subsystem

Constants of motion

Merits and Demerits of CPMD  Advantages Fast dynamics compared to BOMD No need to perform the quenching of electronic wave function at each time step  Disadvantages Dynamics is different from the adiabatic evolution on BO surface Forces on ions are different from the BO forces Ground state Ψ i ≡ Ψ ks i → good agreement with the BOMD

Velocity Verlet algorithm for CPMD.

References  R. Car and M. Parrinello; Phys. Rev. Lett. 55 (22), 2471 (1985)  D. Marx, J. Hutter;  F. Buda et. al; Phys. Rev. A 44 (10), 6334 (1991)  D.K. Remler, P.A. Madden; Mol. Phys. 70 (6), 921 (1990)  B.M. Deb; Rev. Mod. Phys. 45 (1), 22 (1973)  M. Parrinello; Comp. Chemistry 22, (2000)  M.C. Payne et. al; Rev. Mod. Phys. 64 (4), 1045 (1992)

CPMD CPMD code is available at Code developers : Michele Parrinello, Jurg Hutter, D. Marx, P. Focher, M. Tuckerman, W. Andreoni, A. Curioni, E. Fois, U. Roethlisberger, P. Giannozzi, T. Deutsch, A. Alavi, D. Sebastiani, A. Laio, J. VandeVondele, A. Seitsonen, S. Billeter and others PWscf (Plane Wave Self Consistent field) PINY-MD

Applications

Autoionization in Liquid Water Chandler, Parrinello et. al Science 2001, 291, 2121 pH determination of water by CPMD Intact water molecules dissociate → OH - + H 3 O + Rare event ~ 10 hours >>>> fs Transition state separation between the charges ~ 6Å Proposed theory → Autoionization occurs due to specific solvent structure and hydrogen bond pattern at transition state Diffusion of ions from this transition state Role of solvent structure in autoionization Diffusion of ions Microsecond motion of a system as it crosses transition state can not be resolved experimentally pH = - log [H + ]

Nature of proton transfer in water Grotthuss’s idea : Proton has very high mobility in liquid water which is due to the rearrangement of bonds through a long chain of water molecule; effective motion of proton than the real movement + +

Charge separation Chandler, Parrinello et. al Science 2001, 291, Dissociation: Fluctuation in solvent electric field ; cleavage of OH bond 2 H 3 O+ moves by proton transfer within 30 fs 34 Conduction of proton through H-bond network 60 fs 5 Crucial fluctuations carries system to transition state ; breaking of H-bond : 30 fs 6 NO fast ion recombination

Order parameter for autoionization Fluctuations that control routes for proton : No. of hydrogen bond connecting the ions : ℓ ℓ = 2 ; recombination occurs within 100 fs reactant ℓ = 0 ; product ℓ ≥ 3 Critical ion separation is 6 Å At ℓ = 2, sometimes reactant basin ; Thus ℓ is not the only order parameter Potential of proton in H-bonded wire → fluctuation q → configuration description ; q = 1 neutral ; q = 0 charge separated ΔE = E[r(1) – r(0)] → solvent preference for separated ions over neutral molecules

Potential of protons in hydrogen bonded wires connecting H 3 O+ and OH- ions Chandler, Parrinello et. al Science 2001, 291, 2121 Neutral state, bond destabilizing electric field has not appeared Electric field starts to appear ; metastable state w.r.t. proton motion ; 2kcal/mol higher than neutral state Field fluctuations increase ; stable charge separated state ; 20kcal/mol more stable

Nature of the hydrated excess proton in liquid water Two proposed theories : 1. Formation of H 9 O 4 + (by Eigen) 2. Formation of H 5 O 2 + (by Zundel) Charge migration happens in a few picoseconds Tuckermann, Parrinello et. al J. Chem. Phys. 1997, 275, H9O4+H9O4+ H5O2+H5O2+ + Hydrogen bonds in solvation shells of the ions break and reform and the local environment reorders Ab initio calculations show that transport of H+ and OH- are significantly different

Proton transport Tuckermann, Parrinello et. al Nature 1997, 275, 817 Proton diffusion does not occur via hydrodynamic Stokes diffusion of a rigid complex Continual interconversion between the covalent and hydrogen bonds

Proton transport δ = RO aH - RO bH + OaOa ObOb H For small δ ; equal sharing of excess proton → Zundel’s H 5 O 2 + For large δ ; threefold coordinated H3O+ → Eigen’s H 9 O 4 + Tuckermann, Parrinello et. al Nature 1997, 275, 817 ΔF(ν) = -k B T ln [ ∫ dR OO P(R OO,ν) ] Free energy : H 5 O 2 + : at δ = 0 ± 0.05Å, Roo ~ Å ΔF < 0.15 kcal/mol, thermal energy = 0.59 kcal/mol Numerous unclassified situations exists in between these two limiting structures

Breaking bonds by mechanical stress Frank et. al J. Am. Chem. Soc. 2002, 124, 3402 Reactions induced by mechanical stress in PEG 1. Formation of ions corresponds to heterolytic bond cleavage 2. Motion of electrons during the reaction Polymer is expanded with AFM tip Unconstrained reactions can not be observed by classical MD Quantum chemical approaches are more powerful in describing the general chemical reactivity of complex systems

H H C2 C C C O1 O2 H H HH HH H H O H H Solvent Small piece of PEG in water Breaking bonds by mechanical stress Method ΔE (C-O) kcal/mol ΔE (C-C) kcal/mol BLYP Exp Radicaloid bond breaking After equilibration, distance between O1 and O2 was increased continuously by au/time Reaction started at 250 K ; C2O1 ~ 3.2 Å

Snapshots of the reaction mechanisms O O H O H H O H H O H + - O H O H O H O H H O H O O H O H H O H H O H O O H O H H O H H O H O O H O H H O H H O H + - O O H O H O H H O H H 250 K 320 K Frank et. al J. Am. Chem. Soc. 2002, 124, 3402

Hydrogen bond driven chemical reaction Parrinello et. al J. Am. Chem. Soc. 2004, 126, 6280 Beckmann rearrangement of Cyclohexanone Oxime into ε-Caprolactam in SCW SCW accelerates and make selective synthetic organic reactions System description : CPMD simulation, BLYP exchange correlation MT norm conserving pseudo potential Plane wave cut-off 70 Ry, Nose-Hoover thermostat T = 673K, 300K 64 H2O + 1 solute, 18 ps analysis + 11 ps equil. Disrupted hydrogen bond network of SCW alters the solvation of O and N

Proton attack on the Cyclohexanone Oxime Parrinello et. al J. Am. Chem. Soc. 2004, 126, 6280

Problems  Computationally costly  Can not simulate slow chemical processes that take place beyond time scales of 10 ps  Inaccuracy in the assumption of exchange and correlation potential Limitation in the number of atoms and time scale of simulation Inaccurate van der Waals forces, height of the transition energy barrier  BOMD not applicable for photochemistry; transition between different electronic energy levels

Other methods  QM/MM – quantum mechanics / molecular mechanics Classical MD AIMD e.g. catalytic part in enzyme  Path-sampling approach combined with ab-initio MD for slow chemical processes  Metadynamics, for slow processes

Conclusions CPMD : nuclear and electronic degrees of freedom Interaction potential is evaluated on-the-fly Bond formation and breaking is accessible in CPMD : direct access to the chemistry of materials Transferability over different phases of matter CPMD is computationally expensive

Acknowledgement Prof. S. Balasubramanian Dr. M. Krishnan, Bhargava, Sheeba, Saswati

THANK YOU