Ultrafast dynamics of the lowest-lying neutral states in carbon dioxide

We present a study of the ultrafast dissociation dynamics of the lowest-lying electronic excited states in ${\mathrm{CO}}_2$ by using ultraviolet (UV) and extreme-ultraviolet (XUV) pulses from high-order harmonic generation. We observe two primary dissociation channels: a direct dissociation channel along the ${}^1{\mathrm{\Pi }}_g$ electronically excited manifold, and a second channel which results from the mixing of electronic states. The direct dissociation channel is found to have a lifetime which is shorter than our experimental resolution, whereas the second channel has a significantly longer lifetime of nearly 200 fs. In this long-lived channel we observe a beating of the vibrational populations with a period of $\sim 133$ fs.

Figure 1

A two-photon pump (4.77 eV per photon, blue arrow) is used to excite the ${}^1{\mathrm{\Pi }}_g$ state. As population of ${}^1{\mathrm{\Pi }}_g$ begins to dissociate, a percentage of it transfers to other states through nonadiabatic transitions. These other states keep the molecule near the Franck–Condon region longer than is possible from the direct dissociation of the ${}^1{\mathrm{\Pi }}_g$, allowing the probe (11.1 and 14.3 eV photons, magenta arrow) to generate more ${\mathrm{CO}}^{+}$. Figure not to scale.

Figure 2

Total kinetic-energy release distribution as a function of pump-probe delay. The data plotted are ${\mathrm{CO}}^{+}$, but the KER values are scaled to represent the total kinetic-energy release of the molecule. Five probe-early times (about 1 picosecond early) were averaged and subtracted by leaving only enhancements due to the pump-probe effect. The intensity is on a common logarithmic scale. Probe arrives early for negative time delays.

Figure 3

(a) Two KER slices of the distribution are taken and fit with a bi-exponential. The two slices are representative of the two KER regions—0 to 1.2 eV and 1.2 to 4.37 eV—as discussed in the text. The data is scaled to have the same peak-to-probe–late ratio, and then offset to have same probe-late values in order to show shape differences. The quoted decay times are for the slow exponential. (b) Many slices of the KER distribution. Same fit used as in panel (a) with KER slices that are 0.5 eV wide and evenly spaced throughout the distribution, leaving no gaps. The slow decay times are plotted as a function of the center energy of the slice. The error bars represent absolute error obtained from the fit and increase (as a fraction of the time constant) at higher KER.

Figure 4

(a) Five energy slices of the same size used in Fig. 3 showing the deviation from a bi-exponential behavior. A clear secondary peak appears between 1.5 and 3 eV. (b) The same slice as presented in Fig. 3 in black but now fit with a single oscillating exponential instead of a bi-exponential. $T$ is the period of the oscillations and $\tau$ is the exponential lifetime.