But the road to this finding was not as smooth as predicted. When experimental results did not match leading computer simulations, the team’s theorists stepped in to add additional nuance to their models. This work shows how researchers combine theoretical models with experimental evidence to more reliably predict the outcome of photochemical reactions.
To make each frame of the molecular movie, the team used two precisely timed X-ray flashes at SLAC’s Linac Coherent Light Source (LCLS). The first flash dislodged one of the molecule's electrons, giving the molecule an extremely quick jolt of energy. The second flash, arriving just attoseconds later, gave researchers insight into where the electrons had moved through a technique known as X-ray absorption spectroscopy.
The key was exquisite timing control, a capability pioneered by SLAC accelerator scientists. By adjusting the delay between flashes with attosecond accuracy, researchers could choose exactly when to take each snapshot.
The team documented 10 timestamps within the first 10 femtoseconds, or 10 millionths of a billionth of a second, creating a frame-by-frame sequence of electron motion.
The First Frames: Less Than 1 Femtosecond
Within the first femtosecond of being energized by X-rays, the molecule relaxed by ejecting an electron from one of its inner shells. This process creates a low-energy electron that moves slowly enough to interact with surrounding molecules, causing radiation damage in biological systems and even breaking DNA strands. This team is the first to capture the steps of this effect, known as a Coster-Kronig decay, on its natural timescale.
Quantum Coherence: 2 to 10 Femtoseconds
After the electron was kicked out, it left behind a hole which migrated through the molecule until another electron filled it in. This fleeting motion was driven by a phenomenon known as quantum coherence.
“We believe this coherence may influence what happens downstream – the breaking and forming of chemical bonds,” Driver said. “By capturing and characterizing electronic coherence, we can better understand, and perhaps one day gain some measure of control over its effects.”
Chemistry Begins: More Than 10 Femtoseconds
At the 10-femtosecond mark, researchers saw the chemical consequences of all this electron movement: Chemical bonds began breaking and new ones started to form.
Tuning theory to reality
The experimental data revealed that actual electron behavior was more complex than models anticipated. Because modeling ultrafast molecular motion is incredibly complex, computer models typically keep atomic nuclei fixed in place and focus only on certain types of electron motion. But accurate models need to incorporate additional complexity, explained Alicia Palacios, an associate professor at the Autonomous University of Madrid, Spain.
“While these calculations are far more computationally demanding, they produce simulations that are much closer to the reality captured during experiments,” Palacios said.
The team is working on next-generation experiments that can collect even more data in less time, mapping electron movements in larger, complex molecules.
“When we collected this data in 2021, it took more than a week to collect the data we needed to make a comparison with our predictions. Now, with the LCLS superconducting accelerator and its high-repetition rates, our users are collecting better data, on a variety of molecules, in a small fraction of that time,” said James Cryan, senior author on the paper, SLAC associate professor of photon science and interim lead of the LCLS science, research and development division.
The research was conducted by a large collaboration including researchers from Stanford University; IMDEA Nanoscience and Autonomous University of Madrid, Spain; Imperial College London, UK; University of Connecticut; Charles University, Czech Republic; Ohio State University; Paul Scherrer Institute and Ecole Polytechnique Federale de Lausanne, Switzerland; Kansas State University, DOE’s Argonne National Laboratory and Lawrence Berkeley National Laboratory; Tohoku University, Japan; and University of Chicago.
This research is supported in part by the DOE Office of Science. LCLS is a DOE Office of Science user facility.