For the past several decades, we have been able to directly probe the motion of atoms that is associated with chemical transformations and which occurs on the femtosecond (10−15-s) timescale. However, studying the inner workings of atoms and molecules on the electronic timescale1, 2, 3, 4 has become possible only with the recent development of isolated attosecond (10−18-s) laser pulses5. Such pulses have been used to investigate atomic photoexcitation and photoionization6, 7 and electron dynamics in solids8, and in molecules could help explore the prompt charge redistribution and localization that accompany photoexcitation processes. In recent work, the dissociative ionization of H2 and D2 was monitored on femtosecond timescales9 and controlled using few-cycle near-infrared laser pulses10. Here we report a molecular attosecond pump–probe experiment based on that work: H2 and D2 are dissociatively ionized by a sequence comprising an isolated attosecond ultraviolet pulse and an intense few-cycle infrared pulse, and a localization of the electronic charge distribution within the molecule is measured that depends—with attosecond time resolution—on the delay between the pump and probe pulses. The localization occurs by means of two mechanisms, where the infrared laser influences the photoionization or the dissociation of the molecular ion. In the first case, charge localization arises from quantum mechanical interference involving autoionizing states and the laser-altered wavefunction of the departing electron. In the second case, charge localization arises owing to laser-driven population transfer between different electronic states of the molecular ion. These results establish attosecond pump–probe strategies as a powerful tool for investigating the complex molecular dynamics that result from the coupling between electronic and nuclear motions beyond the usual Born–Oppenheimer approximation.
Figures at a glance
Figure 1: Dissociative ionization of hydrogen by an EUV–infrared pulse sequence. 
a, Photoexcitation of neutral hydrogen leads to the excitation of the Q1 (red) and Q2 (blue) doubly excited states and ionization to the 1sσg and 2pσu states, which can be followed by dissociation. R, internuclear distance; a.u., atomic units (0.529 Å). b, c, Experimental D+ (b) and calculated H+ (c) kinetic energy distributions with (from top to bottom) only the isolated attosecond laser pulse present, with only the few-cycle infrared (IR) laser pulse present and for two delays between the EUV and infrared pulses. d, e, Experimental D+ (d) and calculated H+ (e) kinetic energy distributions as functions of the delay between the attosecond pulse and the infrared pulse. Colour scale shows fragment yield in arbitrary units (d) and calculated probabilities (e).
Figure 2: Asymmetry in EUV–infrared dissociative ionization of hydrogen. 
a, Experimentally measured asymmetry parameter (colour scale) for the formation of D+ ions in two-colour (EUV–infrared) dissociative ionization of D2, as a function of the fragment kinetic energy, Ek, and the EUV–infrared delay. A fragment asymmetry is observed that oscillates as a function of the EUV–infrared delay and that strongly depends on the kinetic energy. b, Calculated asymmetry parameter for the formation of H+ ions in two-colour EUV–infrared dissociative ionization of H2, as a function of the fragment kinetic energy, Ek, and the EUV–infrared delay, obtained using the close-coupling method described in the text. c, Same as in a, but for H+ ions.
Figure 3: Mechanisms that lead to asymmetry in EUV–infrared dissociative ionization. 
a, Asymmetry caused by the interference of a wave packet launched in the 2pσu state by direct EUV ionization or rapid ionization of the Q11Σu+ doubly-excited states by the infrared pulse and a wave packet in the 1sσg state resulting from autoionization of the Q11Σu+ states. Blue arrows indicate the effect of the EUV pulse and red arrows that of the infrared pulse; purple lines and arrows signify dynamics that is intrinsic to the molecule. b, Close-coupling calculations in which direct photoexcitation to the 1sσg state has been excluded, supporting the notion that the Q1 autoionizing states have an important role in the localization dynamics. c, Asymmetry caused by the interference of a wave packet that is launched in the 2pσu state by direct EUV ionization and a wave packet in the 1sσg state that results from stimulated emission during the dissociation process. d, Time-dependent asymmetry from a two-level calculation in which the wavefunction of the dissociating molecule is considered to be a coherent superposition of the 1sσg and 2pσu states.
right
