Experiments

The experiments were carried out in our laboratory at the University of Gothenburg and at the synchrotron radiation facility MAX IV in Lund using time-of-flight photoelectron–photoelectron (TOF-PEPECO) and photoelectron–photoelectron–photoion–photoion–coincidence (TOF-PEPEPIPICO) set-ups based on magnetic bottles, as described previously21,22. In Gothenburg, the TOF-PEPECO measurements employed a unique 5.6 m instrument consisting of a conical _∼_1 T magnet in the light–matter interaction region coupled to a solenoid running along the entire flight tube and producing a magnetic field of a few mT. The sample was introduced into the vacuum chamber through a stainless-steel needle, forming an effusive jet in the interaction region. A pulsed helium discharge lamp provided radiation at a repetition rate of about 4.4 kHz, with the photon energy selected using an ellipsoidal grating with a groove density of 1100 lines/mm. This enabled measurements at the discrete photon energies 21.2, 40.8, and 48.4 eV (He I-α, He II-α, and He II-β, respectively) in a small interaction volume. The 5.6 m instrument has a collection–detection efficiency of ∼ 40 % and a nominal resolving power E/Δ E of ∼ 120. At the synchrotron radiation facility MAX IV, operated in single bunch mode, the beamline FlexPES23 was utilized. This beamline is equipped with a mechanical chopper system to reduce the photon repetiton rate from 3.125 MHz to about 94 kHz to allow sufficient time for all the electrons to reach the detector before the next radiation pulse acts. The photon energy used in the measurements was 56.0 eV and the data were collected with a 2.2 m magnetic bottle with a resolving power of E/Δ E ∼ 50, while otherwise being very similar to the 5.6 m instrument.

The TOF-PEPEPIPICO measurements were performed in Gothenburg using the 2.2 m instrument in which the conical magnet was replaced by a hollow ring magnet of reduced field strength. A repeller and pulser plate were installed such that suitable voltages were applied ∼50 ns after ionization to guide the nascent ions into the 0.12 m ion flight tube in the direction opposite to the electron flight tube, triggered once the electrons had left the interaction region as described in Ref22. The potentials guiding the ions were adjusted according to Wiley–McLaren conditions to achieve time focussing for mass resolution and linearity between time deviations and initial ion momentum, allowing the deduction of kinetic energy releases24. Replacing the conical magnet with a hollow ring magnet and using a shorter flight tube reduces the nominal resolving power of the electrons to E/Δ E = 20 and yields a collection–detection efficiency of approximately 20 % for ions. The photon energy was selected using an ellipsoidal grating with a groove density of 550 lines/mm and the repetition rate was about 4.4 kHz. To ensure that electrons and ions recorded within the same time window originate from the same ionization event, the ionization rate is ideally kept below 2 %. However, because of comparatively low vapor pressure of ozone at the working temperature of 195 K and thus relatively high background gas contribution in the apparatus, the count rates were chosen to be slightly higher than ideal, leading to some unwanted coincidences. In both the electron-only and electron–ion configurations, a small voltage (<1 V) was applied across the interaction region to enable collection of low–kinetic-energy electrons.

Ozone was generated using a commercially available ozoniser from C-Lasky and adsorbed onto non-indicating silica gel beads in a U-tube maintained at 195 K using a dry-ice–isopropanol slush bath. The ozone-loaded U-tube was transferred at 195 K to the spectrometer and kept at the same temperature, desorbing from the beads at an acceptable rate. The gas handling system allows pumping the gas and simultaneously introducing it into the spectrometer, providing an ozone sample with only small amounts of \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\mathrm {O_2}$$\end{document}, in addition to a very minute contamination of air5. To monitor sample purity throughout the measurements, valence single-ionization spectra were taken periodically in all the acquisitions, and additional ion spectra were collected in the case of TOF-PEPEPIPICO measurements. The main contaminants identified were helium, water, molecular oxygen, and molecular nitrogen, where the ionized oxygen molecule can be distinguished from dissociated ozone by peak width from kinetic energy release.