Ultrafast Dynamics of Neutral Superexcited Oxygen: A Direct Measurement of the Competition between Autoionization and Predissociation

Using ultrafast extreme ultraviolet pulses, we performed a direct measurement of the relaxation dynamics of neutral superexcited states corresponding to the $nl{\sigma }_g(c{}^4\Sigma _u^{-})$ Rydberg series of ${\mathrm{O}}_2$. An extreme ultraviolet attosecond pulse train was used to create a temporally localized Rydberg wave packet and the ensuing electronic and nuclear dynamics were probed using a time delayed femtosecond near-infrared pulse. We investigated the competing predissociation and autoionization mechanisms in superexcited oxygen molecules and found that autoionization is dominant for the low $n$ Rydberg states. We measured an autoionization lifetime of $92\pm 6\mathrm{fs}$ and $180\pm 10\mathrm{fs}$ for the $(5s,4d){\sigma }_g$ and $(6s,5d){\sigma }_g$ Rydberg state groups, respectively. We also determine that the disputed neutral dissociation lifetime for the $\nu =0$ vibrational level of the Rydberg series is $1100\pm 100\mathrm{fs}$.

Figure 1

An ultrashort XUV pulse excites a $2s{\sigma }_u$ electron to form the $nl{\sigma }_g(c{}^4\Sigma _u^{-})$ Rydberg state of ${\mathrm{O}}_2$. Fast predissociation of these highly excited molecules into neutral fragments can occur through nonadiabatic interactions. Since these neutral states are embedded in an ionization continuum, autoionization can also occur through spontaneous ejection of an electron. The neutral dissociation and autoionization thus form competing decay mechanisms.

Figure 2

(a) A schematic representation of the relaxation lifetime measurement. The $5s{\sigma }_g$ Rydberg state is populated by the XUV APT and decays rapidly due to autoionization and predissociation. We interrupt this decay process by removing the Rydberg electron with a time delayed IR pulse to populate the $(c{}^4\Sigma _u^{-}){\mathrm{O}}_2^{+}$ state and produce measurable ions with well-defined kinetic energies corresponding to $c{}^4\Sigma _u^{-}\to L1$ and $c{}^4\Sigma _u^{-}\to L2$ channels. (b) The XUV excitation spectrum for the 15th harmonic used in the lifetime measurements overlaid with the $n(s,d){\sigma }_g$ Rydberg assignments. (c) The ion kinetic energy spectrum with the XUV pulse alone (lower curve, blue) as well as in the simultaneous presence of the $\mathrm{XUV}+\mathrm{IR}$ probing pulse (upper curve, red). Inset shows the full velocity map of ions with (right half) and without (left half) the IR pulse.

Figure 3

(a) A sketch showing ${\tau }_a$ and ${\tau }_d$ scaling with effective quantum number $n^{*}$. (b) The subplots illustrate the simulated delay dependence of the ion yield in three regimes (solid green curves) which differ in the relative contribution of predissociation and autoionization mechanisms for a fixed total decay lifetime of 100 fs. Dashed red curves depict the Gaussian pump-probe cross-correlation profile representing additional direct $\mathrm{XUV}+\mathrm{IR}$ excitation processes.

Figure 4

(a) Experimental ion yield measured in the $c{}^4\Sigma _u^{-}\to L1$ channel. The fit obtained for time delays $\ge 70\mathrm{fs}$ using the model discussed in the text is shown as a solid green line (a zoomed in view of the fitting region is shown in the inset). (b) Experimental ion yield measured in the $c{}^4\Sigma _u^{-}\to L1$ channel when the harmonic spectrum is tuned to be resonant with the high $n$ Rydberg states. In this region, neutral dissociation dominates resulting in the expected flat delay dependence of the ion yield.