T heory predicts 1-4 that, with an ultrashort and extremely bright coherent X-ray pulse, a single diffraction pattern may be recorded from a large macromolecule, a virus or a cell before the sample explodes and turns into a plasma. Here we report the first experimental demonstration of this principle using the FLASH soft-X-ray free-electron laser. An intense 25 fs, 4 × 10 13 W cm −2 pulse, containing 10 12 photons at 32 nm wavelength, produced a coherent diffraction pattern from a nanostructured non-periodic object, before destroying it at 60,000 K. A novel X-ray camera assured single-photon detection sensitivity by filtering out parasitic scattering and plasma radiation. The reconstructed image, obtained directly from the coherent pattern by phase retrieval through oversampling 5-9 , shows no measurable damage, and is reconstructed at the diffraction-limited resolution. A three-dimensional data set may be assembled from such images when copies of a reproducible sample are exposed to the beam one by one 10 .X-ray free-electron lasers (FELs) are expected to permit diffractive imaging at high resolutions of nanometre-to micrometre-sized objects without the need for crystalline periodicity in the sample [1][2][3][4] . Structural studies within this size domain are particularly important in materials science, biology and medicine. Radiation-induced damage and sample movement prevent the accumulation of high-resolution scattering signals for such samples in conventional experiments 11,12 . Damage is caused by energy deposited into the sample by the very probes used for imaging, for example photons, electrons or neutrons. At X-ray frequencies, inner-shell processes dominate the ionization of the sample; photoemission is followed by Auger or fluorescence emission and shake excitations. The energies of the ejected photoelectrons, Auger electrons and shake electrons differ from each other, and these electrons are released at different times, but within about ten femtoseconds, following photoabsorption 1,13 . Thermalization of the ejected electrons through collisional electron cascades is completed within 10-100 fs (refs 14,15). Heat transport, diffusion and radical reactions take place over some picoseconds to milliseconds.The effect of X-ray-induced sample damage on the recorded image or diffraction pattern could be substantially reduced, if we could collect diffraction data faster than the relevant damage processes 1,16 . This approach requires very short and very bright X-ray pulses, such as those expected from a short-wavelength FEL. However, the large amount of energy deposited into the sample by a focused FEL pulse will ultimately turn the sample into a plasma. The question is when exactly would this happen. There are no experiments with X-rays in the relevant time and intensity nature physics VOL 2 DECEMBER 2006 www.nature.com/naturephysics

The motion of atoms on interatomic potential energy surfaces is fundamental to the dynamics of liquids and solids. An accelerator-based source of femtosecond x-ray pulses allowed us to follow directly atomic displacements on an optically modified energy landscape, leading eventually to the transition from crystalline solid to disordered liquid. We show that, to first order in time, the dynamics are inertial, and we place constraints on the shape and curvature of the transition-state potential energy surface. Our measurements point toward analogies between this nonequilibrium phase transition and the short-time dynamics intrinsic to equilibrium liquids.