Visualizing ultrafast chemical dynamics with X-rays

We live in a world bathed in light. Light drives photosynthesis and is responsible for the oxygen environment that enables the diversity of life found on our planet. Photochemistry provides the means to harness light energy for productive function through the movement of charge, a change in molecular shape, or the cleavage of a bond. Photochemistry also cleaves DNA and produces damaging free radicals. Photochemical processes—both harnessed and destructive—are unavoidable. Ultrafast transient spectroscopies in the UV-visible and infrared regions of the spectrum have allowed the detailed probing and characterization of many photochemical processes occurring on a wide range of timescales. These probes, however, provide only indirect insight into the electronic and structural dynamics of photochemical transformations. The development of X-ray free electron lasers (XFELs) and tabletop high harmonic sources have now provided the tools required to interrogate directly the electronic and structural changes that take place on the very fastest timescales of chemistry. These measurements provide insight into photochemistry, but also more generally into the coupling of electronic and nuclear degrees of freedom in chemistry. It is becoming possible to visualize chemical reactions with both electronic and atomic resolution on timescales from attoseconds to seconds.

In PNAS, Bacellar et al. (1) combine femtosecond X-ray absorption (XAS) and X-ray emission (XES) spectroscopies to address an outstanding question in heme protein photophysics: Does the decay of photoexcited ferric heme involve a cascade of iron-centered spin states or is it characterized more simply by subpicosecond internal conversion to the ground state (2–5)? This question is addressed by Bacellar et al. on a room temperature protein sample, at modest concentration, with ultrafast time resolution, and exploiting the sensitivity of X-rays to both electronic and structural dynamics. Fe K-edge XES probes the 2p→1s (K_α) and 3p→1s (K_β) transitions of the excited metal center and is sensitive to oxidation and spin state. The time-resolved XES data, contrary to the expectations of many, suggest that in ferric cytochrome c (cyt c), a spin cascade produces a low-energy high-spin metal-centered iron state (MC) prior to ground state recovery. This state is populated on a timescale of 650 fs and has a lifetime of ∼10 ps, similar to that observed in UV-visible transient absorption studies of Fe(III) tetraphenylporphyrin chloride (3). Changes in the Fe K-edge X-ray absorption near edge structure (XANES) permit the identification of a doming motion coupled to the spin cascade, providing insight into the structure–function relationship in an important class of heme proteins. Doming has now been implicated in a wide range of heme proteins, often involving states at energies low enough to be thermally accessible and thus with the potential to influence function under a variety of different conditions.

This is the latest in a series of recently published studies demonstrating the potential of ultrafast X-ray spectroscopies in chemistry. A nice overview, current through mid-2018, is given by Wernet (6), but this is a rapidly evolving field with continuing new developments. K-edge XES provides a sensitive probe of the element-specific electronic configuration and spin state of transition metals as exploited by Bacellar et al. in their study of cyt c (1). Kunnus and coworkers (7, 8) have demonstrated that it can also be sensitive to structural changes in a given electronic state through the coupling of electronic and nuclear dynamics. Pronounced coherent oscillations were observed in the time-resolved K_α and K_β XES signal of an iron carbene complex [Fe(2,6-bis(3-methyl-imidazole-1-ylidine)-pyridine)₂]²⁺ following photoexcitation. These were mapped directly to bond length changes in the excited metal-centered state through comparison with simultaneous X-ray scattering measurements (8). While the oscillations report on a vibrational motion in the molecule, the appearance of these motions in the core orbital X-ray emission reports on the coupling of vibrational and electronic degrees of freedom. Valence-to-core (VtC) emission, although much weaker than K_α and K_β emission, provides a more general probe of structural evolution in molecular systems. Changes in bond lengths and bond angles can be encoded into the difference between the ground and excited state spectra as a function of time. The dependence of VtC emission on bond length was recently demonstrated for another class of iron complexes, [Fe(CN)_2n(2,2′-bipyridine)_3-n]⁻²ⁿ⁺² n = 1, 2, 3, in solution (9). The integrated VtC emission correlates with average Fe–ligand bond lengths highlighting evolution in the structure on timescales of less than 500 fs.

Femtosecond XAS measurements provide a more direct probe of element-specific molecular structure. In a recent experiment, femtosecond XANES at the Cu K-edge was used to probe the evolution of coherent vibrational dynamics in a prototypical copper complex, [Cu(2,9-dimethyl-1,10-phenanthroline)₂]⁺ (10). The vibrational motion is again encoded in oscillations in the measured difference signal. Femtosecond XANES is also capable of monitoring coherent wave packet motion in the absence of obvious oscillations in the signal. A Co K-edge XANES measurement on coenzyme B₁₂ in solution (5′-deoxyadenosylcobalamin) demonstrated the potential of this method (Fig. 1). The linear polarization of the optical excitation laser and the X-ray pulses were used to identify two sequential stages of ring expansion along the short and long axes of the corrin ring in the optically bright state within the first 70 fs, followed by axial expansion out of the bright state into the dark excited state minimum on a 200-fs timescale (11). Comparison with broadband UV-visible transient absorption measurements correlates the axial bonds elongation with the loss of fluorescence as the population moves out of the bright state. New valence electronic transitions appear when the molecule relaxes to the dark excited state minimum energy structure. This is not a kinetic process, but rather provides evidence for wave packet evolution on the excited state potential energy surface. Although coherent oscillations are observed in the data, these report on motions orthogonal to the reaction coordinate rather than on the reaction coordinate itself.

Fig. 1.

Schematic cartoon of the kind of structural evolution revealed in polarized femtosecond XANES measurements on coenzyme B₁₂ (11). This can be correlated with electronic dynamics revealed in femtosecond XES measurements and optical absorption measurements as well as structural evolution revealed in femtosecond X-ray scattering measurements to develop a complete picture of the coupled electronic and structural dynamics of a photoexcited molecule.

XANES and XES measurements provide much insight into dynamics, but extended X-ray absorption fine structure (EXAFS) will provide the opportunity for quantitative measurements of bond lengths as a function of time. A proof-of-principle measurement of Fe K-edge EXAFS on [Fe(2,2′:6′,2′′-terpyridine)₂]²⁺ was able to track structural changes in a femtosecond intermediate (∼100-fs lifetime) between the initially excited MLCT state and the long-lived MC intermediate (12). Time-resolved EXAFS measurements in the hard X-ray region will become almost routine as high-repetition rate XFELs come online over the next several years.

Most of these XAS and XES measurements were performed on moderately concentrated solutions (>10 mM) available in large volume. In contrast, the cytochrome c measurement by Bacellar et al. (1) was performed on a ∼4 mM protein sample, and the measurements on B₁₂ coenzymes and analogs were performed at 5 mM or less (11). Even lower concentrations and smaller quantities can be achieved with increased XFEL repetition rates and with drop-on-demand sample delivery. A time-resolved XANES measurement on a synthetic alkynyl B₁₂ analog was performed using <4 mM solution in 55-μm droplets synchronized to the 120-Hz repetition rate of the XFEL (13). Only ∼45 μL of solution was exposed to X-rays during this experiment. Measurements with parallel and perpendicular polarizations for the optical and X-ray pulses permit the assignment of the excited state structural changes to axial and equatorial motions. Comparisons with simulated difference spectra demonstrate that the excited state has an extended Co–N axial bond but the Co–C bond is locked at the ground state bond length, providing direct experimental insight into the mechanism for the photostability of this compound (13, 14).

The accomplishments of Bacellar et al. and the others outlined above are impressive, although there remain some technical challenges to overcome. The K-edge of common first row transition metals falls in an experimentally convenient energy range. The measurements can be performed in air or in helium environments, and common solvents provide little or no interference with the measurements. In contrast, the L-edge of these transition metals, and the K-edges of chemically important elements such as carbon, oxygen, and nitrogen, fall in the soft–X-ray region. For application to chemical dynamics, effective measurements in this region will require the capacity for in-vacuum measurements on very thin liquid samples. High-repetition rate soft X-ray XFEL sources, such as LCLS-II with repetition rates up to 1 MHz and the ChemRIXS end station, both currently under construction, will permit the extension of ultrafast time-resolved X-ray measurements in liquid solution to soft X-ray energies. This will enable direct observation of orbital evolution in polyene isomerization reactions, the potential of which has already been demonstrated in the gas phase (15). It will also enable the direct correlation of metal and ligand dynamics with metal–ligand orbital mixing (16), using element-specific probes of both the transition metal and its ligands.

With the proliferation of ultrafast XFEL sources around the world, the future is indeed bright. Ultrafast X-ray measurements will provide a window on chemical reaction dynamics through the combination of scattering, absorption, and emission measurements—often on the same sample at the same time—with both electronic and atomic resolution on timescales from attoseconds to seconds.