Survey
* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project
Chemical Dynamics Telluride July 2009 J. Tully Telluride July 2009 J. Tully Chemical Dynamics Time-dependent processes: i t H t t 2 classes: 1. H depends on time E. Schrödinger 2. H independent of time a. Initial state: let i (t0 ) j where t E j t t (t ) (t0 ) 2 H j Ej j (t ) exp(iE j t / ) (t0 ) 2 time-independent Telluride July 2009 J. Tully Chemical Dynamics Time-dependent processes: i t H t t 2 classes: 1. H depends on time E. Schrödinger 2. H independent of time b. Initial state: let (t0 ) c j j j where H j Ej j (t ) c j exp(iE j t / ) j j (t ) | c j | | j | 2 2 j 2 * * c c exp i ( E E ) t / k j j k k j jk All Done! time-dependent Chemical Dynamics Objective: i r, R, t H r, R r, R, t t r = electrons H Telluride July 2009 J. Tully R = nuclei 2 2 2 2 2 R r V r, R 2R H el r; R 2M 2M i 2 me H el r; R j r; R E j R j r; R Adiabatic (Born-Oppenheimer) Potential Energy Surface Chemical Dynamics Born-Oppenheimer Approximation: Max Born J. Robert Oppenheimer Telluride July 2009 J. Tully Chemical Dynamics Telluride July 2009 J. Tully Born-Oppenheimer Approximation: r , R , t j r; R j R , t Substitute into TDSE, multiply from left by j r; R , integrate over r: * i j R , t t 2 2 R E j R j R , t 2M student problem: why is this not quite right? Max Born J. Robert Oppenheimer Chemical Dynamics I. II. III. IV. V. VI. VII. VIII. IX. X. XI. XII. Quantum Dynamics Semiclassical Dynamics aside: tutorial on classical mechanics The Classical Limit via the Bohm Equations Classical Molecular Dynamics Adiabatic “on-the-fly” Dynamics Car-Parrinello Dynamics Infrequent Events aside: transition state theory and re-crossing Beyond Born Oppenheimer Ehrenfest Dynamics Surface Hopping Dynamics at Metal Surfaces Mixed Quantum-Classical Nuclear Dynamics Telluride July 2009 J. Tully I. Quantum Dynamics i j R , t t wavefunction for nuclear motion Telluride July 2009 J. Tully 2 2 R E j R j R , t 2M Born-Oppenheimer P.E.S (or diabatic P. E. S.) 1. Zero-point energy 2. Quantized energy levels 3. Tunneling 4. Phase interference (coherence) I. Quantum Dynamics i j R , t t ln(Rate) wavefunction for nuclear motion Quantum Arrhenius (classical) 1/T Telluride July 2009 J. Tully 2 2 R E j R j R , t 2M Born-Oppenheimer P.E.S I. Quantum Dynamics i j R , t t Wave packet methods: Telluride July 2009 J. Tully 2 2 E R j R, t R j 2M Initial wave function: Gaussian wave packet 1/4 2 ( x, 0) 2 exp(ik0 x) exp[( x x0 ) 2 / a 2 ] a k0 p0 initial avg momentum student problem: what is momentum distribution? I. Quantum Dynamics i j R , t t Wave packet methods: Telluride July 2009 J. Tully 2 2 E R j R, t R j 2M Initial wave function: Gaussian wave packet 1/4 2 ( x, 0) 2 exp(ik0 x) exp[( x x0 ) 2 / a 2 ] a k0 p0 initial avg momentum 1/2 | ( x, 0) |2 2 2 exp[2( x x0 ) 2 / a 2 ] a I. Quantum Dynamics Telluride July 2009 J. Tully 2 d 2 E x, t i x, t x 2 t 2 M dx Wave packet methods: Propagate wave function i[ x, t x, t ] x, t (x,t) 2 d 2 E x x, t 2 2 M dx i ( x, t ) 2 d 2 2 M dx 2 E x x, t crude! d 2 ( x, t ) ( x x, t ) ( x x, t ) 2( x, t ) dx 2 (x) 2 finite difference approximation ? Telluride July 2009 J. Tully I. Quantum Dynamics 2 d 2 E x, t i x, t x 2 t 2 M dx Wave packet methods: Propagate wave function: Fourier Transform method 1. Compute F.T. 1 ( p, t ) 2 ipx / x t dx ( , ) e R. Kosloff 2. Inverse F.T. 2 d 2 ( x, t ) 1 p2 ipx / ( , ) e p t dp 2 2M dx 2 2 M student problem: derive Eq. 2 (x,t) I. Quantum Dynamics i j R , t t Telluride July 2009 J. Tully 2 2 E R j R, t R j 2M Angular and velocity distributions of the reaction F + H2 FH + H Experiment Theory M. Qiu et al., Science 311, 1440 (2006) I. Quantum Dynamics Problem: Scaling with Size is Prohibitive: 103N-6 (x,t) 1D 2D Telluride July 2009 J. Tully Quantum Dynamics: Path Integrals I. The Propagator i t Telluride July 2009 J. Tully (H indep of t) it (t ) exp H (t 0) H Rewrite using position representation: define: (r0 ) r0 | (r ) | r0 (r r0 ) and ( x, t ) x | e it H dr | r 0 | (t 0) dy x | e 0 it H dy x | e The contribution to the wave function at x and t from the wavefunction at y and t=0. r0 | 1 it H | y y | (t 0) | y ( y, t 0) the propagator Quantum Dynamics: Path Integrals II. Evaluating the Propagator e [e A A/ N N ] x|e it H Telluride July 2009 J. Tully |y N it iH t , with t t / N exp H exp k 1 Aside: functions of operators? functions of matrices? 1 2 1 3 e 1 A A A 2! 3! 1 1 e A B 1 A B ( A2 AB BA B 2 ) ( A3 A2 B ABA 2! 3! A Student Problem: evaluate / 4 / 4 sin / 4 / 4 Telluride July 2009 J. Tully Quantum Dynamics: Path Integrals II. Evaluating the Propagator e [e A A/ N N ] x|e it H |y N it iH t , with t t / N exp H exp k 1 iH iH exp (tk 1 tk ) exp (tk tk 1 ) iH iH (tk 1 tk ) | xk xk | exp (tk tk 1 ) dxk exp iH xN | exp t | x0 already looks like a path integral: iH dx dx x ... | exp 1 N 1 k 1 k t | xk 1 N x x0 xN t Quantum Dynamics: Path Integrals Split Operators: Trotter’s formula p2 H ( x, p ) V ( x) T V 2m (T and V do not commute) i H t iV t iT t iV t exp exp exp exp 2 2 xk | e iH t | xk 1 xk | e iT t it | xk 1 exp (Vk Vk 1 ) 2 Telluride July 2009 J. Tully Quantum Dynamics: Path Integrals Telluride July 2009 J. Tully Evaluate T-integral analytically: xk | 1 i dp p pxk ) and | exp( 2 xk | e iT t | xk 1 | xk 1 1 i dp p ' xk 1 ) | p ' 'exp( 2 m i m exp ( xk xk 1 ) 2 2 it 2t Telluride July 2009 J. Tully Quantum Dynamics: Path Integrals Put pieces back together: xN | e iH t i N mN t | x0 C dx1... dxN 1 exp ( xk xk 1 ) 2 (Vk Vk 1 ) 2N k 1 2t The Classical Action S is the integral of the Lagrangian: S L dt t m N mN 2 2 ( ) ( ) S t x x V x x Vk k k 1 k k k 1 2 N 2 t k 1 2t k 1 N remember Rig’s discussion of the classical action (T V ) dt Telluride July 2009 J. Tully Quantum Dynamics: Path Integrals Put pieces back together: xN | e iH t i N mN t | x0 C dx1... dxN 1 exp ( xk xk 1 ) 2 (Vk Vk 1 ) 2N k 1 2t The Classical Action S is the integral of the Lagrangian: S L dt (T V ) dt t m N mN 2 2 ( ) ( ) S t x x V x x Vk k k 1 k k k 1 2 N 2 t k 1 2t k 1 N Feynman Path Integral: xN | e iH t i | x0 C dx1... dxN 1 exp S ( x0 , x1 ,...xN ) Richard Feynman Telluride July 2009 J. Tully Quantum Dynamics: Path Integrals Feynman Path Integral: S(x0,x1,…) xN | e iH t i | x0 C dx1... dxN 1 exp S ( x0 , x1 ,...xN ) Far from classical path: rapidly oscillating phases cancel Near classical path: phases do not cancel path coordinates x0, x1, … least action = stationary phase = classical path Nancy Makri Aside: Quantum Statistical Mechanics: Canonical Partition Function: Q allstates n Eigenstate representation: Telluride July 2009 J. Tully e En / kT e En n n | e En | n n | e H | n Q n | e H | n Tr e H Valid for any representation n In particular, in position representation: Q dx x | e H | x Compare to path integral evaluation of propagator: Set t i and y = x (closed loop) x | e iHt / | y x | e iHt / | x x | e H | x Evaluate Q via path integral in imaginary time Aside: Quantum Statistical Mechanics: xN | e iH t i x C dx dx S x x x | 0 1... N 1 exp ( 0 , 1 ,... N ) S where Telluride July 2009 J. Tully t mN 2 ( ) x x Vk k 1 2t k N k 1 N now want: i x | e H | x C dx1... dxN 1 exp S where path is specified by x, x1, …xN-1, x mN ( xk xk 1 ) 2 S (t i ) i Vk i S R N k 1 2 N where SR is real Necklace of Classical Beads Telluride July 2009 J. Tully Aside: Quantum Statistical Mechanics: Necklace of Classical Beads i x | e H | x C dx1... dxN 1 exp S where path is specified by x, x1, …xN-1, x mN ( xk xk 1 ) 2 S (t i ) i Vk i S R N k 1 2 N Q C dx where SR is real S R (path) exp all loop paths potential energy function for a classical mechanical “ring polymer” decaying: no phase cancellation problems x6 x5 x x4 x3 x1 x2 Aside: Quantum Statistical Mechanics: How does a purely classical mechanical formalism introduce quantum effects like zero-point energy and tunneling? zero-point energy tunneling Telluride July 2009 J. Tully Telluride July 2009 J. Tully Quantum Dynamics: Path Integrals CMD: Centroid Variable Molecular Dynamics x6 RPMD: Ring Polymer Molecular Dynamics x6 x5 x x4 x5 x x4 x3 x1 x3 x1 x2 x2 solves scaling problem but: short times only Greg Voth David Manolopoulos Chemical Dynamics I. II. III. IV. V. VI. VII. VIII. IX. X. XI. XII. Quantum Dynamics (Semiclassical Dynamics) aside: tutorial on classical mechanics The Classical Limit via the Bohm Equations Classical Molecular Dynamics Adiabatic “on-the-fly” Dynamics Car-Parrinello Dynamics Infrequent Events aside: transition state theory and re-crossing Beyond Born Oppenheimer Ehrenfest Dynamics Surface Hopping Dynamics at Metal Surfaces Mixed Quantum-Classical Nuclear Dynamics Telluride July 2009 J. Tully II. Semiclassical Dynamics Telluride July 2009 J. Tully Semiclassical Path Integral Approach: S(x0,x1,…) xN | e iH t i | x0 C dx1... dxN 1 exp S ( x0 , x1 ,...xN ) Far from classical path: rapidly oscillating phases cancel Near classical path: phases do not cancel path coordinates x0, x1, … least action = stationary phase = classical path Bill Miller Telluride July 2009 J. Tully II. Semiclassical Dynamics Frozen Gaussian: 1/4 2 ( x, 0) 2 exp(ik0 x) exp[( x x0 ) 2 / a 2 ] a fixed Center of Gaussian, x0, follows classical trajectory Advantages: 1. Simple 2. Includes delocalization of wavepacket 3. Includes quantum interference effects (add amplitudes, not probabilities) Rick Heller II. Semiclassical Dynamics Frozen Gaussian (E. J. Heller) 1/ 4 2 ( x, 0) 2 exp(ik0 x) exp[( x x0 ) 2 / a 2 ] a Center of Gaussian, x0, follows classical trajectory Disadvantages: Telluride July 2009 J. Tully Chemical Dynamics Classical Molecular Dynamics: Aside: brief tutorial on classical mechanics Telluride July 2009 J. Tully Chemical Dynamics I. II. III. IV. V. VI. VII. VIII. IX. X. XI. XII. Quantum Dynamics Semiclassical Dynamics aside: tutorial on classical mechanics The Classical Limit via the Bohm Equations Classical Molecular Dynamics Adiabatic “on-the-fly” Dynamics Car-Parrinello Dynamics Infrequent Events aside: transition state theory and re-crossing Beyond Born Oppenheimer Ehrenfest Dynamics Surface Hopping Dynamics at Metal Surfaces Mixed Quantum-Classical Nuclear Dynamics Telluride July 2009 J. Tully Tutorial on Classical Mechanics Newton’s equations: F ma mq p Force is the gradient of the potential: F V (q) q mq p V (q) q Newton’s Equations Telluride July 2009 J. Tully Tutorial on Classical Mechanics Telluride July 2009 J. Tully Hamilton’s equations: Define total energy = Hamiltonian, H: q H ( p, q) p p H (q, p) q if Hamilton’s Equations H ( p, q ) T ( p ) V ( q ) q p 2 / 2m V ( q ) p/m p V (q) q William Rowan Hamilton. Tutorial on Classical Mechanics Telluride July 2009 J. Tully Lagrangian Mechanics: Define Lagrangian L: L (q, q ) T (q ) V (q) d L L dt q q if Euler-Lagrange Equations L (q, q ) T (q ) V (q) mq 1 2 mq 2 V (q) p V (q) q Joseph-Louis Lagrange Tutorial on Classical Mechanics Telluride July 2009 J. Tully Hamilton-Jacobi Equation: Define Hamilton’s Principle Function S: S p dq S = -H t S q, q Hamilton-Jacobi Equations S S H q, q t q p V (q) q 0 Karl Gustav Jacob Jacobi Chemical Dynamics I. II. III. IV. V. VI. VII. VIII. IX. X. XI. XII. Quantum Dynamics Semiclassical Dynamics aside: tutorial on classical mechanics The Classical Limit via the Bohm Equations Classical Molecular Dynamics Adiabatic “on-the-fly” Dynamics Car-Parrinello Dynamics Infrequent Events aside: transition state theory and re-crossing Beyond Born Oppenheimer Ehrenfest Dynamics Surface Hopping Dynamics at Metal Surfaces Mixed Quantum-Classical Nuclear Dynamics Telluride July 2009 J. Tully III. Classical Limit via Bohm Eqs. i j R , t t j R, t 2 2 R E j R j R , t 2M i Aj R , t exp[ S j (R , t )] (1) (2) Substitute (2) into (1) and separate real and imaginary parts: Sj Aj 2 2 1 R A j 2 [ R S j ] E j R Aj 2M 2M 1 2[ R Aj ] [ R S j ] Aj 2R S j 2M (3) (4) Telluride July 2009 J. Tully III. Classical Limit via Bohm Eqs. Telluride July 2009 J. Tully Compare Eq. (3) with Hamilton-Jacobi Equation: S = -H t Sj S q, q where S = q p 2 2 1 R A j 2 [ R S j ] E j R Aj 2M 2M (3) the “quantum potential” 0: Sj 1 [ R S j ]2 E j R 2M David Bohm Chemical Dynamics I. II. III. IV. V. VI. VII. VIII. IX. X. XI. XII. Quantum Dynamics Semiclassical Dynamics aside: tutorial on classical mechanics The Classical Limit via the Bohm Equations Classical Molecular Dynamics Adiabatic “on-the-fly” Dynamics Car-Parrinello Dynamics Infrequent Events aside: transition state theory and re-crossing Beyond Born Oppenheimer Ehrenfest Dynamics Surface Hopping Dynamics at Metal Surfaces Mixed Quantum-Classical Nuclear Dynamics Telluride July 2009 J. Tully IV. Classical Molecular Dynamics Basic Assumptions: 1. Born-Oppenheimer Approximation 2. Classical mechanical nuclear motion Unavoidable Additional Approximations: 1. Approximate potential energy surface 2. Incomplete sampling 3. (often) Extrapolate to longer timescales 4. Too few atoms 5. Continuum solvent and other shortcuts Telluride July 2009 J. Tully IV. Classical Molecular Dynamics Numerical propagation of equations of motion: Assume we know x(t) and v(t) at some time t. How do we find x(t+) and v(t+) at later time t+ Taylor’s series: dx d 2x 2 d 3x 3 x(t ) x(t ) ... 2 3 dt 2dt 3!dt 1 d 3x 3 2 x(t ) v(t ) a(t ) ... 3 2 3!dt 1 d 3x 3 2 ... x(t ) x(t ) v(t ) a (t ) 3 2 3!dt Add where x(t ) 2 x(t ) x(t ) a(t ) 2 O( 4 ) Ma (t ) E ( x) x Verlet algorithm Telluride July 2009 J. Tully IV. Classical Molecular Dynamics Telluride July 2009 J. Tully Choice of time step : The highest frequency (fastest) Motions are stiff vibrational modes t << 1/ t+ e.g, O-H stretch: 3700 cm-1 = 1x1014 s-1 Test: Energy Conservation remember Rig: “weak test” Telluride July 2009 J. Tully H + ClF HF + Cl Telluride July 2009 J. Tully IV. Classical Molecular Dynamics Inelastic scattering of NO molecules from a gold surface 0.0 2.0 time (ps) 4.0 IV. Classical Molecular Dynamics Protein in water do we need all of these water molecules? dielectric continuum approximation Telluride July 2009 J. Tully