X-ray Free Electron Lasers
A New Revolution in Structural Studies
X-ray crystallography is a crucial research technique for chemists who would like to investigate the structure of their compounds in solid state.
So far lots of Nobel prizes have been awarded for research fundamental to or applying crystallographic methods, from the early discoveries of Röntgen until the recent prizes for research on quasicrystals or implementation of theoretical methods in the refinement of macromolecules. Nevertheless, the field still undergoes a rapid development and more breakthroughs are expected, in particular in the field of X-ray free electron laser (XFEL) crystallography.
X-ray Free Electron Lasers
X-ray free electron lasers are devices that produce very short (in order of femtoseconds) pulses of very coherent X-rays . This could bring so far unavailable possibilities in structural research, providing access to time-resolved studies, following reactions during the diffraction experiment or overcoming the problem of radiation damage.
So far there are two facilities offering access to X-ray free electron lasers suitable for high-resolution structural studies, although there are many institutes with operating free electron lasers. These are LCLS (Linac Coherent Light Source) in Stanford and Spring-8 (Super Photon Ring) SACLA in Japan. A European facility will be open for users in 2017 in DESY (Hamburg), also a Swiss source is planned for 2016 .
Production of X-Ray Flashes
How do the X-ray free electron lasers produce their X-ray pulses? First bunches of electrons have to be accelerated which takes place in very long resonators where the oscillating microwaves inside transfer energy to the electrons. Each stage of this procedure is controlled very strictly in order to ensure a very well-defined regular electron beam. Then the electrons are directed to the so-called undulator which is an array of magnets with alternating north and south poles (Fig. 1) . This undulator wiggles the movement of electrons. As a result they start emitting X-rays. X-rays are faster than electrons - due to their interaction the electrons form assemblies resembling "disks" which emit very coherent radiation - this is how X-ray flashes are produced.
This phenomenon is called self-amplified spontaneous emission. The X-ray flashes are released, the electrons are diverted at the end so that they do not come out together with the flashes.
The produced beams have unique characteristics. They are much brighter than radiation produced by the most recent generation of synchrotrons. Their wavelength can be tuned from 10 to 0.5 Å. The pulses are very short, even at 1 fs, and energetic; they could cut through steel.
But they could be also used for structure determination. It was suspected that this could be done even before XFELs were constructed. Simulations were done for protein molecules, such as the model protein in crystallography, the lysozyme  showing that before the molecule is destroyed as a result of Coulomb expansion, it diffracts and the diffracted beams can be detected. Now the first XFEL was open in 2009 and the first protein structure determination with the aid of XFELs was performed in 2011 for the membrane protein Photosystem I, proving that the principle of "diffraction before destruction" can really be implemented .
So far there are two main options for sample delivery in XFELs (Fig. 2). In the first option nanocrystals are suspended in a liquid or gel matrix and continuously injected into the place where they are hit by X-ray pulses. Before destruction they diffract and the diffracted beams are detected by the detector. This is the basis of serial femtosecond crystallography where changes in the sample, for instance upon irradiation, can be monitored in a time-resolved manner. There is no control of the crystals orientation or quality.
Another approach is to use a fixed target, such as a large crystal or microcrystals embedded in a film. Different places in such a collective sample are probed with X-ray pulses, therefore the sample holder has to be moved so that the places can be chosen.
Data Collection & Analysis
The detectors in XFEL crystallography, e. g. CSPAD (silicon pixel array detector), must meet specific challenges which are the high radiation intensities and exposures at a femtosecond-time-scale at a rate of hundreds of pulses per second .
Data integration is problematic as the diffraction data are collected for many nanocrystals in random orientations. Additional source of experimental error are the instabilities in the pulse intensity and the noise from accompanying solvent. Therefore during data integration a Monte Carlo approach is adopted, assuming that these orientations are perfectly random. But first the data need to be indexed - this is done selecting reflections which nearly fulfill the Bragg's condition .
Already a couple of alternative software suites are available for treatment of these data, such as "Cheetah" or "Crystfel" . Lots of data are produced, in the amounts of hundreds of terabytes. Not all of these data are useful, so they have to be analyzed and the useful data must be selected. This is now called the "hit finding" and is a very time-consuming procedure. Typically out of about 4 millions of collected frames 400 000 are useful and 300 000 can be indexed .
Nevertheless, in spite of these problems XFEL crystallography opens new pathways for answering of still unanswered questions, concerning e. g. the structure of S-states in the Kok cycle of Photosystem II or imaging of single virus particles under physiological conditions. Some of the recent breakthroughs include obtaining radiation-damage-free structure of PSII WOC which has important implications for the understanding of the natural photosynthesis process and design of catalysts for artificial photosynthesis .
The author gratefully acknowledges Prof. Dr. Stefanie Dehnen and Dr. Klaus Harms (generous support and helpful discussions).
 Spence J. C. H. et al.: Rep. Prog. Phys. 75, 102601. (2012)
 Waldrop M. M.: Nature 505, 606 (2014)
 Neutze R. et al.: Nature 2 406, 752 (2000)
 Chapman H. N. et al.: Nature 2011, 470, 73.
 Herrmann S. et al.: Nucl. Instr. and Meth. in Phys. Res. A718, 550 (2013)
 Foucar L. et al.: Comp. Phys. Commun. 183, 2207 (2012)
 Barty A. et al.: J. Appl. Cryst. 47, 1118 (2014)
 (a) Kupitz C. et al.: Nature, 513, 261, (2014) (b) Kern J. et al.: Nature Commun. 5, 4371, (2014) (c) Suga M. et al.: Nature 517, 99 (2015)