Start-to-end simulation of single-particle imaging using ultra-short pulses at the European X-ray Free-Electron Laser

Carsten Fortmann-Grote, А. В. Бузмаков, Zoltán Jurek, N. Duane Loh, Liubov Samoylova, Robin Santra, E.a. Schneidmiller, T. Tschentscher, S. D. Yakubov, Chun Hong Yoon, Michael V. Yurkov, Beata Ziaja, Adrian P. Mancuso⋈IUCrJ 2017 · DOI

TL;DR

This paper presents a comprehensive simulation of single-particle imaging (SPI) experiments at the European X-ray Free-Electron Laser (XFEL), focusing on how X-ray pulse duration impacts the quality of structural information obtained. They found that for molecules of a few thousand atoms, pulse durations between 3 and 9 femtoseconds (fs) are optimal, with 3 fs pulses containing insufficient photons for reliable reconstruction.

What Problem Does This Paper Address?

Single-particle imaging (SPI) using X-ray Free-Electron Lasers (XFELs) holds immense promise for determining the atomic-resolution structures of non-crystalline biomolecules. The key advantage of XFELs is their ability to deliver ultra-short, high-intensity pulses that can capture diffraction data before the sample is destroyed by radiation damage. However, achieving atomic resolution requires a delicate balance: shorter pulses minimize damage but also reduce the number of photons, leading to noisy diffraction patterns. Longer pulses provide more photons but increase damage. This paper tackles the critical question of finding the "sweet spot" for X-ray pulse duration in SPI experiments to maximize structural information quality, considering both radiation damage and photon statistics.

Methods

The authors employ a sophisticated "start-to-end" simulation pipeline that models the entire SPI process. This includes:

  • XFEL Source Simulation: Using precomputed data from the European XFEL, they simulate the temporal structure and photon statistics of SASE (Self-Amplified Spontaneous Emission) X-ray pulses at 4.96 keV, with durations of 3, 9, and 30 fs.
  • Wave Propagation: The simulated X-ray pulses are propagated through a beamline and focusing optics to the sample position.
  • Molecular Dynamics (MD) and Radiation Damage Simulation: A molecular dynamics scheme (using XMDYN and XATOM codes) models the interaction of the X-ray pulse with a sample molecule (the protein 2nip). This captures electronic transitions, Auger decay, and subsequent ionization, leading to atomic displacements and plasma expansion.
  • Diffraction Pattern Generation: For each simulated trajectory, they calculate the instantaneous scattering intensity and integrate it over the pulse duration to generate 2D diffraction patterns. These patterns are then subjected to Poisson noise to mimic experimental conditions.
  • Orientation and Phasing: The simulated diffraction patterns are fed into the expand-maximize-compress (EMC) algorithm to reconstruct a 3D electron density map. The consistency of these reconstructions across multiple runs is assessed.

What's novel here is the integration of these steps into a comprehensive workflow, allowing for a realistic assessment of how pulse duration affects the entire SPI process, from photon generation to structural reconstruction.

Datasets

The primary "dataset" used in this study is simulated. They utilize precomputed XFEL pulse data from the European XFEL's Photon Pulses Database (XPD). For the sample, they use the structure of the two-nitrogenase iron protein (2nip), a small molecule previously studied in similar simulations. This choice was pragmatic due to computational cost, but they acknowledge it's not a typical target for current XFEL SPI experiments.

Key Insights and Ideas

  • Pulse Duration is Crucial: The study unequivocally demonstrates that X-ray pulse duration is a critical parameter for successful SPI, directly impacting the signal-to-noise ratio and the interpretability of diffraction data.
  • 3 fs is Too Short: While 3 fs pulses minimize radiation damage, they contain too few photons to generate diffraction patterns with sufficient signal-to-noise for reliable orientation and reconstruction algorithms. The resulting data quality is poor.
  • 9 fs Offers a Sweet Spot: A 9 fs pulse duration strikes a good balance. It significantly reduces radiation damage compared to 30 fs pulses while providing enough photons for robust orientation and reconstruction, leading to higher quality structural data.
  • Scattering Efficiency vs. Photon Flux: The paper highlights a trade-off: shorter pulses lead to higher scattering efficiency (less ionization means more coherent scattering), but longer pulses have a higher overall photon flux, which can compensate for increased ionization if the scattering efficiency doesn't drop too drastically.
  • Coefficient of Variation as a Quality Metric: The coefficient of variation (σ₁) is effectively used to quantify the consistency of the reconstructed 3D orientation data, serving as a reliable indicator of data quality and the achievable resolution.

Key Results

  • Optimal Pulse Duration: For the 2nip molecule at 5 keV, the optimal pulse duration for SPI is found to be between 3 and 9 fs.
  • 3 fs vs. 9 fs: The number of detected photons per diffraction pattern (Nph,det) for 3 fs pulses is reduced by more than a factor of three compared to 9 fs and 30 fs pulses.
  • Radiation Damage: At 3 fs, ionization is kept below 30% even for heavier atoms, and atomic displacement is negligible (<0.1 Å). At 9 fs, displacement is still minimal, while at 30 fs, displacement can exceed 10 Å.
  • Data Consistency: The coefficient of variation (σ₁) analysis shows that 3 fs data leads to inconsistent orientation results, while 9 fs data yields significantly more robust and consistent reconstructions, especially at higher resolutions.

Limitations and Future Work

  • Small Molecule Size: The study acknowledges that the 2nip molecule is small (~7 nm) and not a typical target for current XFEL SPI experiments. Extrapolating these findings to larger, more realistic biomolecules is speculative due to potential non-linearities in radiation damage and plasma expansion.
  • Single Algorithm Focus: The simulations used a fixed set of algorithms (e.g., EMC for orientation). The authors note that comparing different algorithms for radiation damage modeling and orientation recovery is an important open question.
  • Solvent Effects: The simulations ignored the presence of solvent, which can have a complex influence on both radiation damage and background scattering.
  • Future Work: The authors plan to extend their simulations to larger particles and different photon energies to build a more comprehensive simulation database. This will aid in developing analytical models for radiation damage and determining minimum photon requirements for SPI. They also plan to incorporate realistic detector response models.

Summary updated

Abstract

Single-particle imaging with X-ray free-electron lasers (XFELs) has the potential to provide structural information at atomic resolution for non-crystalline biomolecules. This potential exists because ultra-short intense pulses can produce interpretable diffraction data notwithstanding radiation damage. This paper explores the impact of pulse duration on the interpretability of diffraction data using comprehensive and realistic simulations of an imaging experiment at the European X-ray Free-Electron Laser. It is found that the optimal pulse duration for molecules with a few thousand atoms at 5 keV lies between 3 and 9 fs.
BibTeX
@inproceedings{fortmanngrote2017start,
  author = {Carsten Fortmann-Grote and А. В. Бузмаков and Zoltán Jurek and N. Duane Loh and Liubov Samoylova and Robin Santra and E.A. Schneidmiller and T. Tschentscher and S. D. Yakubov and Chun Hong Yoon and Michael V. Yurkov and Beata Ziaja and Adrian P. Mancuso⋈},
  title = {{Start-to-end simulation of single-particle imaging using ultra-short pulses at the European X-ray Free-Electron Laser}},
  booktitle = {IUCrJ},
  year = {2017},
  doi = {10.1107/s2052252517009496},
}
Figure