Benefits and inconveniences of channel-specific absorption boundaries for the solution of the TDSE Luca Argenti & Eva Lindroth Albanova Universitetscentrum Stockhom Oslo 28 Apr 2010 Temporal scales in the microcosm τatomic ≈ 24 as Sub-femtosecond extreme-UV pulses by High Harmonic Generation Milan 2006 Sansone et al., Science 314, 443 (2006) Garching 2008 Goulielmakis et al., Science 320, 1314 (2008) An example: pump probe investigation of doubly excited states in helium PUMP He (1s2) + ϒxuv → He** (1Po) → He+ (1s,2s,2p) + ePROBE He** (1Po) + n ϒir → He** (1Lπ) → He+ (1s,2s,2p) + e- Detection of photofragments System confinement Reflection on the box boundary Absorbing boundaries Possible inconveniences of the absorbing boundaries reflection on the absorption potential annihilation of long-range bound satellites Time propagation Time propagation Radiationless hamiltonian Time propagation Radiationless hamiltonian Radiation-matter interaction (V-gauge) Time propagation Radiationless hamiltonian Radiation-matter interaction (V-gauge) Time propagation Radiationless hamiltonian Radiation-matter interaction (V-gauge) Time propagation Radiationless hamiltonian Radiation-matter interaction (V-gauge) Time propagation Radiationless hamiltonian Radiation-matter interaction (V-gauge) Time propagation Radiationless hamiltonian Radiation-matter interaction (V-gauge) SPLIT Time propagation Radiationless hamiltonian Radiation-matter interaction (V-gauge) SPLIT Magnus I + Krylov Time propagation Radiationless hamiltonian Radiation-matter interaction (V-gauge) SPLIT Spectral resolution Magnus I + Krylov 1s ! l 2s ! l 2p ! l!1 2p ! l+1 N= oo .. . N=3 N=2 1 N=1 L "L The Basis - Close Coupling - B-splines - L≤6 Absorption potentials vα(r) = Pα zα (r - rα)nα θ(r - rα) Pα V = Σα Vα ∀ α,β z = -( 1+ 5i ) 10-4 n=2 ro = Rbox - 100 au zα ∈ C [ Vα ,Vβ ] = 0 Rbox = 400 - 800 au Absorption potentials vα(r) = Pα zα (r - rα)nα θ(r - rα) Pα V = Σα Vα ● Complex ∀ α,β zα ∈ C [ Vα ,Vβ ] = 0 Extraction & Absorption ● Channel specific & Fully independent ● zα freely adjustable on the run z = -( 1+ 5i ) 10-4 n=2 ro = Rbox - 100 au Rbox = 400 - 800 au Effect of V on Close Coupling wave functions v(r) = z θ( r - ro ) ( r - ro )n Effect of V on Close Coupling wave functions v(r) = z θ( r - ro ) ( r - ro )n PARTIAL WAVE CHANNEL Effect of V on Close Coupling wave functions v(r) = z θ( r - ro ) ( r - ro )n PARTIAL WAVE CHANNEL LOCALIZED CHANNEL Effect of V on Close Coupling wave functions v(r) = z θ( r - ro ) ( r - ro )n PARTIAL WAVE CHANNEL LOCALIZED CHANNEL Effect of V on Close Coupling wave functions v(r) = z θ( r - ro ) ( r - ro )n PARTIAL WAVE CHANNEL LOCALIZED CHANNEL Effect of V on one-particle functions a b Effect of V on one-particle functions f a b Effect of V on one-particle functions f a c b Effect of V on one-particle functions f[a,c] a f[c,b] c b Projection on a B-spline basis 12 3 4 a i i+1 ... n n+1 ... c b Projection on a B-spline basis 12 3 4 a V Bi = 0 i i+1 ... i = 1, 2, ..., n n n+1 ... c b Projection on a B-spline basis 12 3 4 a V Bi = 0 i i+1 ... i = 1, 2, ..., n Span {Bi}i≤n ⊂ ker V n n+1 ... c b Projection on a B-spline basis 12 3 4 a V Bi = 0 i i+1 ... i = 1, 2, ..., n Span {Bi}i≤n ⊂ ker V n n+1 ... c b V Bn+j ≠ 0 j = 1, 2, ..., m Projection on a B-spline basis 12 3 4 a i i+1 ... V Bi = 0 i = 1, 2, ..., n n n+1 ... c V Bn+j ≠ 0 j = 1, 2, ..., m Span {Bi}i≤n ⊂ ker V V’ = PVP b P : projector onto Span{Bi}i≤n+m Projection on a B-spline basis 12 3 4 a i i+1 ... V Bi = 0 n n+1 ... i = 1, 2, ..., n c V Bn+j ≠ 0 j = 1, 2, ..., m Span {Bi}i≤n ⊂ ker V V’ = PVP V’ ϕ = λϕ, P : projector onto Span{Bi}i≤n+m λ≠0 b ϕ∈ Span {Bi}i≤n⊥ Projection on a B-spline basis 12 3 4 a i i+1 ... V Bi = 0 n n+1 ... i = 1, 2, ..., n c V Bn+j ≠ 0 j = 1, 2, ..., m Span {Bi}i≤n ⊂ ker V V’ = PVP V’ ϕ = λϕ, P : projector onto Span{Bi}i≤n+m λ≠0 b ϕ∈ Span {Bi}i≤n⊥ Projection on a B-spline basis 12 3 4 a i i+1 ... V Bi = 0 n n+1 ... i = 1, 2, ..., n c V Bn+j ≠ 0 j = 1, 2, ..., m Span {Bi}i≤n ⊂ ker V V’ = PVP V’ ϕ = λϕ, P : projector onto Span{Bi}i≤n+m λ≠0 b ϕ∈ Span {Bi}i≤n⊥ Construction of the orthogonal complement to ker v B-spline superposition matrix 2(k-2)-th order homogeneous linear difference equation Boundary conditions Solution of the inner part of the linear system ci M ci+1 Associated secular problem Symmetry of the characteristic polynomials λ is a root λ-1 is also a root therefore the eigenvalues are separated in three classes: λ <1 λ =1 λ >1 As a matter of fact, for B-splines, all the λs are real, negative, and λ ≠ 1 Absolute values of the coefficients ci of an eigenfunction of V on the boundary k λ 2 3.732 3 2.322 4 1.869 5 1.645 6 1.512 7 1.425 8 1.363 9 1.316 10 1.280 First summary The absorbing boundaries we have described are - easy to implement and fast to execute - localized but with a characteristic minimal range - channel specific - adjustable on the run The process under study PUMP He (1s2) + ϒxuv → He** (1Po) + → He (1s,2s,2p) + e- PROBE He** (1Po) + n ϒir → He** (1Lπ) → He+ (1s,2s,2p) + eIR PROBE ωir = 0.057 au (1.55 eV) I = 1⋅1012 W/cm2 s fwhm = 5.3 fs 0.02 0.01 E(au) 0 −0.01 −0.02 0 XUV PUMP ωxuv = 2.23 au (60.7 eV) I = 2⋅1014 W/cm2 s fwhm = 550 as 200 400 600 time (au) 800 1000 Quantum interference Cooper, Fano, Prats, Phys. Rev. Lett. 10 5168 (1963) Quantum interference What if the impinging radiation pulse were much shorter than the lifetime of doubly excited states ? Cooper, Fano, Prats, Phys. Rev. Lett. 10 5168 (1963) A single channel, single resonance example. A single channel, single resonance example. XUV pump S P D F G H I time (au) XUV pump S P D F G H I time (au) XUV pump S P D F G H I time (au) XUV pump He + ϒxuv → He+(1s) + e- (strongest) S P D F G H I time (au) XUV pump He + ϒxuv → He+(1s) + e- (strongest) He + ϒxuv → He** → He+(1s)+eHe + ϒxuv → He+(N=2) + eS P D F G H I time (au) XUV pump He + ϒxuv → He+(1s) + e- (strongest) He + ϒxuv → He** → He+(1s)+eHe + ϒxuv → He+(N=2) + eHe + 2ϒxuv → He+(1s) + eHe + 2ϒxuv → He+(N=2) + e- S P D F G H I time (au) XUV pump He + ϒxuv → He+(1s) + e- (strongest) He + ϒxuv → He** → He+(1s)+eHe + ϒxuv → He+(N=2) + eHe + 2ϒxuv → He+(1s) + eHe + 2ϒxuv → He+(N=2) + eHe + 3ϒxuv → He+(N=1,2) + eHe + 2ϒxuv → He+(1s) + e- + ϒxuv He + 2ϒxuv → He+(N=2) + e- + ϒxuv S P D F G H I time (au) XUV pump He + ϒxuv → He+(1s) + e- (strongest) He + ϒxuv → He** → He+(1s)+eHe + ϒxuv → He+(N=2) + eHe + 2ϒxuv → He+(1s) + eHe + 2ϒxuv → He+(N=2) + eHe + 3ϒxuv → He+(N=1,2) + eHe + 2ϒxuv → He+(1s) + e- + ϒxuv He + 2ϒxuv → He+(N=2) + e- + ϒxuv Four photon processes ... S P D F G H I time (au) Partial Diff. Photoelectron Spectrum 1s 2s 2p Energy (au) N=2 XUV pump -1 0 cos θ 1 1au ~ 27.2 eV Partial Diff. Photoelectron Spectrum 1s 2s 2p Energy (au) N=2 Direct ionization XUV pump -1 0 cos θ 1 1au ~ 27.2 eV Partial Diff. Photoelectron Spectrum 1s 2s 2p Energy (au) Continuous across threshold N=2 Direct ionization XUV pump -1 0 cos θ 1 1au ~ 27.2 eV Partial Diff. Photoelectron Spectrum 1s 2s 2p Energy (au) Continuous across threshold N=2 Direct ionization XUV pump -1 0 cos θ 1 1au ~ 27.2 eV Partial Diff. Photoelectron Spectrum 1s 2s 2p Energy (au) N=2 ? -1 0 cos θ 1 What happens when the first bunch of fast electrons is absorbed at the boundaries (but the other lot still isn’t)? Partial Diff. Photoelectron Spectrum 1s Zero 2s 2p The same as before Energy (au) N=2 -1 0 cos θ 1 Partial Diff. Photoelectron Spectrum 1s Zero 2s 2p The same as before Energy (au) N=2 He+(2s,2p)+e- continua are comparable to the decaying part of the DES, and yet they are neatly resolved -1 0 cos θ 1 Partial Diff. Photoelectron Spectrum 1s Zero 2s 2p The same as before Energy (au) N=2 sp+3 sp+2 -1 0 cos θ 1 Breathing sp+ mode 15 R (au) 0 15 + 1 o ([001] P ) Breathing sp+ mode 15 R (au) 0 15 + 1 o ([001] P ) Breathing sp+ mode 15 R (au) 0 15 + 1 o ([001] P ) Electron density at large radii 700 R (au) 0 Electron density at large radii 700 R (au) 0 MEASURING THE WAVE NUMBER OF THE ELECTRON BURSTS kbursts ≈ 0.078 au magenta = scale saturation Long range interference of resonances kbursts ≈ 0.078 au sp+3 E sp+2 k2 = [2(E2-Eth)]1/2 k1 = [2(E1-Eth)]1/2 Long range interference of resonances kbursts ≈ 0.078 au sp+3 E sp+2 k2 = [2(E2-Eth)]1/2 k1 = [2(E1-Eth)]1/2 Long range interference of resonances kbursts ≈ 0.078 au sp+3 E sp+2 k2 = [2(E2-Eth)]1/2 Δk ≈ 0.074 au k1 = [2(E1-Eth)]1/2 τbeat= 48.7 au (~1.2 fs) Burst-breath correlation 800 Radius (au) 600 400 200 0 700 800 900 Time (au) 1000 1100 to fit (au) 421 471 518 564 621 671 717 764 817 869 920 965 1011 1073 1118 tmax-den 422 470 518 568 622 670 717 768 818 866 920 968 1012 1068 1120 Ionization-excitation yield Pα = what is annihilated in channel α xuv on Pα(t) & P’α(t) 1.5E−3 field free propagation 1sEp 1E−3 2pEs x 1000 2sEp x 1000 5E−4 2pEd x 1000 0 0 2000 4000 time (au) 6000 Probe of the metastable wave packet dynamics with an IR pulse On the tail of the pulse 1s 2s 2p N=2 E [001]+2 + ϒIR Emerging new ang. dist.: He** + n ϒIR → He+(2s,2p) + e- [001]+2 - ϒIR [001]+2 - 2ϒIR -1 0 cos θ 1 IR PROBE 2S+1Lπ[N N m]A 1 2 1Se [100]+ 1Se [010]+ 1Po[100]1Po [010]- 1De [100]+ 1De [001]1Fo [010]1Fo [001]+ 1Ge [010]+ 1Po [001]+ 1De [010]+ Yield of the parent ions IONIZATION-EXCITATION RATES 1s XUV 2s 2s IR 2p 2p 0 1000 2000 3000 time (au) 4000 5000 6000 Yield of the parent ions IONIZATION-EXCITATION RATES 1s cannot be extrapolated XUV 2s 2s IR 2p 2p 0 1000 2000 3000 time (au) 4000 5000 6000 Yield of the parent ions IONIZATION-EXCITATION RATES 1s cannot be extrapolated XUV slowly decaying DES 2s 2s IR 2p 2p 0 1000 2000 3000 time (au) 4000 5000 6000 Yield of the parent ions IONIZATION-EXCITATION RATES 1s cannot be extrapolated XUV slowly decaying DES 2s 2s IR slow electrons (right above threshold) 2p 2p 0 1000 2000 3000 time (au) 4000 5000 6000 Yield of the parent ions IONIZATION-EXCITATION RATES 1s cannot be extrapolated XUV slowly decaying DES 2s 2s IR slow electrons (right above threshold) 2p 2p 0 1000 2000 3000 time (au) 4000 5000 absorption of metastable satellites in the wrong channel 6000 Yield of the parent ions IONIZATION-EXCITATION RATES 1s cannot be extrapolated XUV slowly decaying DES 2s 2s IR slow electrons (right above threshold) 2p 2p 0 1000 2000 3000 time (au) 4000 5000 absorption of metastable satellites in the wrong channel 6000 polarization effects resulting in wrong channel assignement Corrections to the yields for “large” times dP/dt, for all channels, converges to zero very slowly. In the 1s channel, because of the long lived doubly excited states. In the N=2 channels, because of the accumulation of density at threshold (extremely slow electrons). Moreover, it is dangerous! As soon as the slowest electrons above threshold get to the boundary, also the transiently bound electrons with long trajectories get there and get absorbed. Effects of the parent ion polarization on yields The IR field polarizes the 1s core, and thus it causes instantaneously part of the 1s ionization rate to be recored as pertaining to the 2p channel 2s 2p 1s + 2p c(t) For an IR with I=1012 W/cm2 the interference turns out to be negligible. If necessary, though, it can be singled out time (au) Quantum beats in the ionization-excitation yields Quantum beats in the ionization-excitation yields 2s/2p yields are out of phase by o ~60 Buildup of the photoelectron angular distribution -0.425 2s 2p Energy (au) -0.5 N=2 Buildup of the photoelectron angular distribution -0.425 2s 2p Energy (au) -0.5 N=2 Interference fringes at time delays differing by half IR period ! t=15.53 fs after XUV !t=16.87 fs 2s Energy (au) −0.425 −0.5 −1 2p 0 cos Argenti, Lindroth 1 " arXiv:1004.1266v1 [physics.atom-ph] 8 Apr 2010 Conclusions - incomplete absorption - when metastable states and thresholds are involved, the convergence may be very slow - reflection of low energy components - absorption of long-range satellites - channel partition possibly compromised by external fields - the angular and energy distribution of the absorbed electrons is lost - low energy components can be accelerated - propagation can be stopped before low energy components and long-range satellites reach the absorbing region and the rest be extracted by projection onto scattering states - the effect of external fields can be singled out - the loss of information in one channel does not influence the other channels The Computational Atomic Physics Group of Stockholm http://www.atom.physto.se/~genkin/Grouphomepage/group.html Group leader: Eva Lindroth (lindroth@physto.se) Luca Argenti (argenti@physto.se) Jakob Bengtsson (jakbe@physto.se) Michael Genkin (genkin@physto.se) Erik Waltersson (walter@physto.se) Luca Argenti Jakob Bengtsson Eva Lindroth Michael Genkin Erik Waltersson !ank y"