Excitation Dynamics in Quantum Dots Oleg Prezhdo U. Washington, Seattle Warwick – August 27, 2009 Outline ¾ Time-Domain Density Functional Theory & Nonadiabatic Molecular Dynamics • • • Quantum backreaction, surface hopping Zero-point energy Decoherence/dephasing ¾ Excitation Dynamics in Quantum Dots • • • Phonon bottleneck Generation of multiple excitons Dephasing, linewidths, biexciton fission Adiabatic vs. Nonadiabatic MD Nonadiabatic MD: Coupling between potential surfaces opens channels for system to change electronic states. electrons treated quantum-mechanically e e transition allowed e e e e nuclei treated classically weak coupling strong coupling Time-Domain DFT for Nonadiabatic Molecular Dynamics Electron density derives from Kohn-Sham orbitals ρ (x ) = ∑ ϕ p ( x ) 2 Ψ = ϕ p (x1 , t )ϕ q (x2 , t )Kϕ v (x N , t ) p SD DFT functional H depends on nuclear evolution R(t ) Variational principle gives ih ∂ϕ p ( x, t ) ∂t = Hϕ p ( x, t ) p = 1,2 K α α Orbitals are expanded in adiabatic KS basis ϕ p ( x, t ) = ∑ c p (t )χ ( x ) α H ( x; R (t ))χ ( x; R (t )) = ε (R (t ))χ ( x; R (t )) α α α • ⎛ • → →⎞ ih cα = ∑ c β ⎜ ε β δ αβ − ih χ α ∇ R χ β ⋅ R ⎟ ⎜ ⎟ β ⎝ ⎠ Open Theoretical Questions 9 How to couple quantum and classical dynamics? quantum influence on classical trajectory 9 Can one do better than classical mechanics for nuclear motion? zero-point motion, tunneling, branching, loss of coherence Nuclear Evolution: Ehrenfest • 2 MR Total energy of E = + V (R (t )) + Trx ρ ( x )H ( x; R (t )) electrons and nuclei tot 2 dEtot =0 is conserved dt time-dependent Hellmann-Feynman theorem gives Newton equation •• → → M R = − ∇ R V − Trx ρ ( x ) ∇ R H ( x; R (t )) quantum force Nuclear Evolution: Surface Hopping a.k.a., quantum-master equation with time-dependent transition rates: - non-perturbative - correct short time dynamics Trajectory branching: Tully, JCP 93, 1061 (1990); Within TDDFT: Craig, Duncan, Prezhdo PRL 95, 163001 (2005) Detailed balance: Parahdekar, Tully JCP 122, 094102 (2005) Quantum-Classical Lie Bracket O. V. Prezhdo, V. V. Kisil Phys. Rev. A 56 162 (1997) O. V. Prezhdo J. Chem. Phys. 124 201104 (2006) quantum commutator + classical Poisson bracket problems with Jacobi identity: Quantized Hamilton Dynamics O. V. Prezhdo, Yu. V. Pereverzev J. Chem. Phys. 113, 6557 (2000) O. V. Prezhdo Theor. Chem. Acc. 116, 206 (2006) q2 q3 + V= 2 3 d<q> dt but d < q2 > dt = < p >; d<p> dt = - < q > - < q2 > < q2 > ≠ < q > < q > = < pq + qp > ≡ 2 < pq >s the infinite hierarchy is terminated by a closure d < pq >s dt and = < p2 > - < q2 > - < q3 > < q3 > ≈ 3 < q2 > < q > − 2 < q >3 Harmonic Oscillator in Mapped QHD-2 hbar mass hbar mass Metastable Cubic Potential in Mapped QHD-2 hbar mass hbar mass Double-Slit Potential in Mapped QHD-2 potential seen by a narrow wavepacket V HqL + potential seen by a wide wavepacket 1 2 VH2L HqL s2 Schrodinger Cat and Decoherence | B0 > alive atom cat System - radioactive atom; Bath - cat In Nanomaterials System - electrons, spins; Bath - phonons dead Franck-Condon Factor and Decoherence | B0 > Σ B2 = B1 B2 ei E 1 – E 2 t/h 2 δ E1 – E 2 B1 t B2 t dt Bath (vibrational) wave functions diverge This affects evolution of (electronic) system Decoherence and Surface Hopping ρ = B ρ S–B B Reduced density matrix: ρ 11 ρ 12 ρ 21 ρ 22 → ρ 11 ρ 12 B2 B1 ρ 21 B1 B2 ρ 22 ρ12 →0 on decoherence time scale hopping probability 2 P12 ~ρ12 2 With decoherence: P 12 = T 12 + T 12 + ... 2 Without decoherence P12 = T 12 + T 12 + ... Decoherence makes transitions less likely 0.1 2 + 0.1 2 < 0.1 + 0.1 2 (quantum Zeno effect) Quantum Dot Solar Cells Open Questions in Nanoscale Materials 9 Quantum confinement effects on excitation dynamics: To what extent are there bottlenecks? 9 Electron-vibrational relaxation (heating): Which phonons are involved and why? Biexcitons in PbSe Quantum Dots Shaller, Klimov PRL 92 186601 (2004); Ellingson, Beard, Johnson, Yu, Micic, Nozik, Shabaev, Efros, NanoLett 5 865 (2005) Biexciton creation yield Exciton to biexciton time under 0.25ps Electron-Phonon Relaxation in PbSe Quantum Dots Schaller, Pietryga, Goupalov, Petruska, Ivanov, Klimov PRL 95 196401 (2005) No phonon bottleneck. Times are similar to biexciton creation times Larger dots relax more slowly ?! Structural relaxation of PbSe Quantum Dots 32 atoms Pb16Se16 d=0.9nm Bulk, T=0 K Relaxed, T=0 K Heated, T=300 K 136 atoms Pb68Se68 d=1.3nm Bulk, T=0 K Relaxed, T=0 K Heated, T=300 K Even a very small PbSe quantum dot preserves its bulk topology Orbitals and Density of States (LUMO) (LUMO) (LUMO+1) (LUMO+2) (LUMO+3) (LUMO+1) (LUMO+2) (LUMO+3) states mix; sp3 hybridization (?) DOS of Pb68Se68 Time, ps DOS of Pb16Se16 3 2 1 electronic states hole states -2 -1 0 E-Ef, eV 1 2 Absorption Spectra Pb16Se16 oscillator strength Energy, eV Pb68Se68 oscillator strength Energy, eV Population of 4 states near to threshold Comparison of Relaxation Times agree with experiment Similar relaxation times for electrons and holes hole states (VB) Biexciton creation is faster than relaxation Larger dot relaxes more slowly electron states (CB) due to weaker NA coupling Time, fs Active Phonon Modes Electron-Phonon Coupling: Lower frequency acoustic modes are more active than optical modes Raman Spectrum of 3-nm PbS QD [Acc Chem. Res. 2000, 33, 773-780] Phonon Bottleneck for 1P Electron in CdSe Quantum Dots Pandey, Guyot-Sionnest, Science 322 929 (2008) Many factors must be avoided. 1P to 1S electron relaxes within ~1ns Phonon Bottleneck for 1P Electron in Cd Se Quantum Dots big 1P–1S electron gap ZnS shell does not change electronic structure of CdSe core Geometric and Electronic Structure of Ge and Si Clusters DOS is symmetric, slightly higher for electrons Relaxation in Ge and Si Clusters Electrons relax much faster than holes ! (despite nearly symmetric DOS) Active Phonon Modes Low frequency modes are active for both electrons and holes However, high frequency modes are active only for electrons Proposed Multiplication Mechanisms Inverse Auger Dephasing Direct Excitation Hartree-Fock Band Structure 1. Small dots represent large dot DOS 2. Huge one-electron gap 3. Symmetric vs. asymmetric DOS 4. Secondary gaps in PbSe DOS SAC-CI Spectra and Fraction of Multiple Excitons fraction of multiple excitons spectra PbSe CdSe CdSe spectra agree with experiment JACS 128, 629 (2006) 1. Sharp onset of multiple excitons 2. Above threshold: double excitons in PbSe; single, double and superpositions in CdSe New Experimental Data on Multiple Exciton Generation “Ideal” Klimov et al. Bawendi et al. Bulk Apparent increase in Static decay due to ionization, etc. Calculations for Charged PbSe Dots Conduction band transitions overwhelm MEs Much higher ME threshold Phonon-Induced Pure-Dephasing Times, fs 300K/100K ME Fission – much slower Smaller dots – faster dephasing Lower Temperature – slower dephasing Summary for Quantum Dots 9 No bottleneck due to small gaps Bottleneck only at lowest energy PbSe 9 Smaller dots relax faster, coupling over DOS 9 Acoustic, not optical modes are active 9 All three MEG mechanisms are important: inverse Auger, dephasing and direct (PbSe) 9 Charged QDs show much lower ME yields 9 Surface ligands are important 9 Dephasing: 10fs lumin, MEG, 100fs MEF CdSe/ZnS General Questions 9 Quantum confinement effects on excitation dynamics: To what extent are there bottlenecks? • → → 9 Electron-vibrational relaxation (heating): α ih χ ∇ R χ β ⋅ R Which phonons are involved and why? Dots NAMD/TDDFT TDDFT: Phys. Rev. Lett. 95 163001 (2005) Bracket: Phys. Rev. A 56, 162 (1997) J. Chem. Phys. Rapid. 124, 201104 (2006) Nano Lett. 6 2295 (2006) Decoher: J. Chem. Phys. 111 8366 (1999) Nano Lett. 7 3260 (2007) Phys. Rev. Lett. 85, 4413 (2000) J.Phys.Chem.C 111 4871 (2007) Bohmian: Phys. Rev. Lett. 86 3215 (2001) J.Phys.Chem.C 112 7800 (2008) Rev. Comp. Chem. in press J. Phys.Chem.C 112 18291 (2008) QHD: J. Chem. Phys. 113, 6557 (2000); Chem.Phys. Lett. 458 113 (2008) 116, 4450 (2002); 116, 8704 (2002); J.Photochem.-Photobiol.A 190, 342 (2008) 117, 2995 (2002); 120 11209 (2004); Pure&Appl.Chem. 80 1433 (2008) 121 10967 (2004); 122 234109 (2005); Chem.Phys.Lett. FRONTIER 460 1 (2008) 126 204108 (2007); 129, 144104 (2008) ACS-Nano 3 93 (2009) Chem. Phys. Lett. 346, 463 (2001); 378, Phys. Rev. B 79 235306 (2009) 533(2003) J. Mol. Struct. 630 45 (2003) J. Phys. Chem. C 113 12617 (2009) Adv.Top.Theor.Chem.Phys. 12B 339 (2003) Dalton Trans. in press Theor. Chem. Acc. 116, 206 (2006) ACS-Nano in press J. Phys. Chem. A 111, 10212 (2007) Acc. Chem. Research, submitted J. Phys. Soc. Jap. 77 044001 (2008)