Real-time and Sub-wavelength Ultrafast Coherent Diffraction Imaging in the Extreme Ultraviolet

OPEN SUBJECT AREAS: ULTRAFAST PHOTONICS HIGH-HARMONIC GENERATION MICROSCOPY IMAGING AND SENSING Received 15 September 2014 Accepted 18 November 2014 Published 8 December 2014 Correspondence and requests for materials should be addressed to M.Z. (michael. ) Real-time and Sub-wavelength Ultrafast Coherent Diffraction Imaging in the Extreme Ultraviolet M. Zürch1, J. Rothhardt2,3, S. Hädrich2,3, S. Demmler2, M. Krebs2, J. Limpert2,3, A. Tünnermann2,3,4, A. Guggenmos5,6, U. Kleineberg5,6 & C. Spielmann1,3 1 Institute of Optics and Quantum Electronics, Abbe Center of Photonics, Friedrich-Schiller-University Jena, Max-Wien-Platz 1, 07743 Jena, Germany, 2Institute of Applied Physics, Abbe Center of Photonics, Friedrich-Schiller-University Jena, Albert-Einstein-Straße 15, 07745 Jena, Germany, 3Helmholtz Institute Jena, Fröbelstieg 3, 07743 Jena, Germany, 4Fraunhofer Institute for Applied Optics and Precision Engineering, Albert-Einstein-Straße 7, 07745 Jena, Germany, 5Ludwig-Maximilians-Universität München, Am Coulombwall 1, D-85748 Garching, Germany, 6Max-Planck-Institut für Quantenoptik, Hans-Kopfermann-Str. 1, D-85748 Garching, Germany. Coherent Diffraction Imaging is a technique to study matter with nanometer-scale spatial resolution based on coherent illumination of the sample with hard X-ray, soft X-ray or extreme ultraviolet light delivered from synchrotrons or more recently X-ray Free-Electron Lasers. This robust technique simultaneously allows quantitative amplitude and phase contrast imaging. Laser-driven high harmonic generation XUV-sources allow table-top realizations. However, the low conversion efficiency of lab-based sources imposes either a large scale laser system or long exposure times, preventing many applications. Here we present a lensless imaging experiment combining a high numerical aperture (NA50.8) setup with a high average power fibre laser driven high harmonic source. The high flux and narrow-band harmonic line at 33.2 nm enables either sub-wavelength spatial resolution close to the Abbe limit (Dr50.8l) for long exposure time, or sub-70 nm imaging in less than one second. The unprecedented high spatial resolution, compactness of the setup together with the real-time capability paves the way for a plethora of applications in fundamental and life sciences. N ovel short wavelength sources combined with lensless imaging methods such as coherent diffraction imaging (CDI) and others1 paved the way for microscopy with a resolution well below the wavelength of visible light. During the last decade we have witnessed a tremendous progress in the development of coherent extreme ultraviolet (XUV) and X-ray sources and their application to high-resolution imaging2–4. In CDI one directly collects the diffracted light with a large area CCD detector close to the sample, i.e. with a high numerical aperture. It thus holds promise for further improving the spatial resolution down to the atomic level. Most of the diffraction imaging experiments are carried out at large-scale facilities such as free-electron lasers and synchrotron sources. Limited access to these sources is a severe bottleneck for many applications. Laser driven high harmonic generation (HHG) sources allowed successful implementation of CDI at the lab scale5,6. However, typical HHG sources suffer from a low conversion efficiency and thus one either needs a large laser system for single-shot performance7–9 or relatively long exposure times10–13, which again renders broad usage of this technology, e.g., high resolution imaging in life sciences, difficult. As mentioned above, an important prerequisite for high-resolution microscopy is an imaging system with a high numerical aperture (NA). However, for high NA imaging the temporal coherence of the light limits the achievable resolution in addition to the spatial coherence requirement that must be fulfilled for any NA. The smallest details of the sample can be only resolved if the diffraction pattern for large diffraction angles is recorded and evaluated. The diffraction pattern for large angles can be seen only if the coherence length is long enough, i.e. the relative bandwidth is small enough. In a recent publication Miao et al. presented an estimation of the resolution for a CDI experiment as a function of the relative bandwidth14 Dr~ SCIENTIFIC REPORTS | 4 : 7356 | DOI: 10.1038/srep07356 OaDl , l ð1Þ 1 www.nature.com/scientificreports Figure 1 | The experimental setup for high resolution XUV imaging at high average power: (a), The pulses of a fiber based CPA (FCPA) system are spectrally broadened in a xenon filled hollow-core fiber and subsequently compressed with dielectric chirped mirrors (DCM). HHG is driven with 40 W of average power into a gas nozzle backed with 5 bar krypton. Two fused silica plates with antireflection coating for the IR are mainly reflecting the XUV light. Spectral selection of the 31st harmonic at 33.2 nm is done by a pair of aluminium filters followed by two curved XUV multilayer mirrors. The curved mirrors refocus the light tightly onto the sample. The diffracted light is captured by an XUV-sensitive CCD (CAM1). A beamstop (BS) suppresses the strong central speckle. (b) Measured XUV beam profile 20 mm behind the rear focus reveals good spatial beam properties. (c) The XUV multilayer mirrors, which are hit under almost normal incidence to minimize aberrations, have an overall reflectivity of 4.6% (solid black line) for the selected single harmonic line (dotted blue line). where O is the linear oversampling ratio, which should be a little larger than two in an experimental realization, a is the typical object size and Dl/l is the relative bandwidth of the illuminating light. A closer inspection of Equ. (1) reveals a major limitation for the achievable resolution using HHG sources driven by ultrashort pulse lasers. For a high conversion efficiency the laser pulses must be of high intensity and the pulse duration must be as short as possible15. For typical Ti:sapphire laser systems the pulses are 30 fs or shorter, resulting in a rather large relative bandwidth. Consequently, the XUV light will also have a rather large relative bandwidth, which is typically on the order of Dl/l51/30. Using Equ. (1) the achievable resolution for a 3 micron large object can be estimated to be roughly 100 nm. Using longer driving pulses will lower the relative bandwidth, but at the same time reduces the conversion efficiency substantially. The lower photon flux will decrease the resolution by lowering the signal to noise ratio for a finite exposure time. Hence, the best ever reported relative spatial resolutions achieved with tabletop CDI setups do not significantly reach below two wavelengths of the illuminating light7,11,16. SCIENTIFIC REPORTS | 4 : 7356 | DOI: 10.1038/srep07356 To overcome this limitation we employ an XUV source based on a high repetition rate ytterbium-doped fiber CPA system operating at 1030 nm wavelength wit (...truncated)