Observation of site-selective chemical bond changes via ultrafast chemical shifts

The study investigates whether ultrafast, site-selective changes in a molecule’s chemical environment can be initiated and tracked in real time using x-ray photoelectron spectroscopy (XPS). Chemical transformations driven by photoexcitation can occur within a few femtoseconds and involve coupled electron-nuclear dynamics. While XPS is a powerful probe of local chemical environments via core-level binding energy shifts, most prior studies have been static, and time-resolved investigations have lacked sufficient temporal resolution immediately after excitation. The authors implement a hetero-site x-ray pump/x-ray probe scheme on CO to monitor transient chemical shifts that reflect electron rearrangement, Auger decay, and bond breaking dynamics, providing atomic-site specificity and femtosecond temporal resolution. This approach complements x-ray absorption pump–probe methods and aims to capture fundamental electron and nuclear dynamics underpinning chemical bond changes.

Prior work established XPS as a sensitive probe of local chemical environments, but primarily in static contexts. Time-resolved inner-shell photoelectron spectroscopy and ultrafast XPS studies have begun to access UV-excited state dynamics, though higher temporal resolution was needed to track transient chemical shifts right after excitation. X-ray absorption-based pump–probe techniques have advanced rapidly, probing unoccupied states with femtosecond resolution. However, directly following local chemical shifts of core-electron binding energies on the natural femtosecond timescale, and with site specificity, remained an open challenge. The present work addresses this gap by combining a site-selective resonant x-ray pump with a time-delayed x-ray probe to measure transient C1s binding energies in CO following O1s excitation and subsequent Auger decay.

Experimentally, a site-selective resonant x-ray pump excites the O 1s electron of CO into the unoccupied 2π* orbital, creating a neutral core-excited state CO(1s−1)2π*. This state decays within a few femtoseconds via Auger processes, producing CO+ cationic states (both dissociative and bound). A time-delayed x-ray probe ionizes the C 1s electron, and the photoelectron kinetic energy is measured to obtain C1s binding energies and corresponding chemical shifts. XPS spectra are processed and normalized; difference spectra are formed by subtracting late-time signals to isolate transient features. Free-electron laser (FEL) parameters set the spectral resolution (~3.5 eV) and temporal resolution (≤10 fs). Time delays were scanned from −5 to 40 fs. The pump–probe employed SASE x-ray pulses, with data analysis addressing timing jitter and pulse-to-pulse fluctuations. Theoretical modeling treats coupled electron and nuclear motion at the quantum level, computing time-resolved XPS spectra for core-excited and Auger states along evolving nuclear wavepackets. Electronic structure methods include state-averaged CASSCF for low-lying states and configuration interaction approaches for higher-lying satellite core-hole states, with analysis of Mulliken charge redistribution and potential energy curves. Spectral transients are quantified by integrating over energy windows of 3.5 eV centered on relevant peaks.

• The core-excited neutral CO(1s−1)2π* state exhibits a calculated chemical shift of approximately −2.4 eV relative to the ground state C1s signal and a short lifetime of ~4.2 fs. The corresponding transient in the experiment rises at 0 fs and decays within <10 fs, consistent with rapid Auger depletion.
• Experimental C1s spectra of CO+ Auger states show substantially increased binding energies spanning ~305–325 eV, with two main peaks near ~310 eV and ~317 eV (theory ~320 eV). The theoretical spectrum overestimates the shifts by ~3 eV, within expected excited-state calculation uncertainties.
• Temporal behavior distinguishes Auger channels: the lower binding energy peak (~310 eV) rises with a delay (~5 fs) and shows a residual long-lived component (>40 fs) attributed to bound Auger states; the higher energy peak (~317–320 eV) exhibits a further delayed onset and persists, consistent with dissociative channels that grow as the core-excited population decays.
• Theory and experiment agree on population flow from the core-excited state to Auger channels on a few-femtosecond timescale, reflected in the transfer of spectral intensity from lower to higher binding energy features.
• Electron–nuclear correlation underlies the observed chemical shifts: the core-excited state shows bond elongation by ~0.14 Å and electron density withdrawal toward O due to screening of the core hole; in dissociative Auger (Φ) states, the valence hole migrates and becomes trapped at the C site at large internuclear distances, increasing the C1s binding energy by ~11 eV relative to 0 fs.
• Branching: about 86% of Auger channels are dissociative; the dominant dissociative Φ and bound Π₁ channels contribute ~17.9% and ~8.9% of the total Auger yield, respectively.
• Mulliken charge analysis tracks local charge redistribution: early-time sharing of the valence hole evolves into localization at C during dissociation, correlating with the time-dependent chemical shift.

The results demonstrate that site-selective x-ray pumping combined with time-resolved XPS can directly track ultrafast, local chemical environment changes in molecules. By correlating transient C1s chemical shifts with calculated electron density redistribution, Mulliken charges, and nuclear wavepacket motion, the study connects observed spectral features to fundamental processes: core-hole screening, Auger decay, charge migration, and bond dissociation. The prompt appearance and rapid decay of the core-excited signature confirm femtosecond core-hole dynamics, while the delayed, persistent high-binding-energy C1s features reveal dissociation pathways and hole trapping at carbon. The agreement between experimental transients and quantum dynamical simulations validates the interpretation of spectral changes as a readout of population transfer among electronically and geometrically evolving states. This site-specific chemical shift tracking provides a powerful probe of energy and charge transport at their natural femtosecond timescales and complements x-ray absorption approaches by directly sensing occupied-state (core-level) binding energy changes.

This work introduces and validates a hetero-site, x-ray pump/x-ray probe XPS methodology to observe site-selective chemical bond changes and electron dynamics on few-femtosecond timescales, demonstrated on CO. The approach captures the evolution from core excitation through Auger decay to dissociation, revealing transient chemical shifts, charge redistribution, and hole trapping at specific atomic sites. The technique’s generality and local sensitivity establish a route to investigate ultrafast charge and energy transport in more complex systems, with prospects for extension to attosecond temporal resolution and larger polyatomic molecules. Future efforts could focus on improving spectral and temporal resolution, mitigating FEL timing jitter, refining theoretical treatments of highly excited core-hole states, and applying the method to site-specific charge migration in complex molecules and materials.

• Experimental limitations include finite spectral resolution (~3.5 eV) and temporal resolution (≤10 fs), as well as timing jitter and variable SASE pulse structures that broaden transients and complicate precise wavepacket characterization.
• Background subtraction and limited statistics affect the clarity of some transient features, particularly the decay behavior of the lower-energy Auger peak.
• Theoretical spectra for excited Auger states show ~3 eV overestimation of chemical shifts, reflecting challenges in describing strong electron correlation in core-hole and satellite states. Highly excited satellite states could not be treated with fully relaxed state-averaged CASSCF due to computational cost and were handled at a CI level, introducing additional uncertainty.
• Differences between transform-limited pulses used in modeling and SASE pulses used experimentally lead to discrepancies in the predicted time-dependent energy shifts versus predominantly intensity changes observed.