© 2008 Macmillan Publishers Limited. All rights reserved.

NEWS & VIEWS

nature photonics | VOL 2 | SEPTEMBER 2008 | www.nature.com/naturephotonics 529

Stefan eisebittis at the Berliner Elektronenspeicherring-Gesellschaft für Synchrotronstrahlung (BESSY), Albert-Einstein-Straße 15, 12489 Berlin, Germany.

e-mail: eisebitt@bessy.de

S ince Antonie van Leeuwenhoek’s first glimpse into the world of the very small in the seventeenth

century, the ability to obtain magnified images of microscopic entities has been crucial to many advances in all areas of natural science. Ongoing microscopy research aims to improve spatial resolution and reduce exposure time while maintaining sufficient contrast and brightness. But these figures of merit for imaging are often at odds. Nevertheless, a spatial resolution corresponding to interatomic distances in conjunction with sufficient temporal resolution to capture a ‘still image’ could be considered as the ultimate goal of microscopy. By increasing the imaging efficiency of a holographic approach compatible with ultrafast stroboscopic X-ray illumination, Stefano Marchesini and collaborators make an important step in this direction, as reported on page 560 of this issue1.

Full-field imaging is commonly achieved by placing a lens in the scattered-wave field behind the specimen (‘microscopy’) or by recording the interference of the scattered radiation with a reference wave (‘holography’). More recently, the concept of replacing the optical elements by an iterative computational process in conjunction with suitable experimental boundary conditions has also been successfully used (‘iterative phase retrieval via oversampling’)2. In all cases, illumination with extreme-UV radiation or X-rays enables high spatial resolution owing to the short wavelength.

In the oversampling approach, no light-consuming optical system is required; however, a certain loss of reliability is incurred as a result of convergence issues in the iterative algorithm. Both microscopy and holography on the other hand require

optical elements: lenses and beam splitters, respectively. A part of the incident photon flux is consumed by the optical system (in addition to absorption in the specimen) — this is the price you pay for obtaining an unambiguous image of the object. In addition to this loss, efficient optical elements are particularly hard to come by in the few-nanometre-wavelength regime, where the quality of the elements is limited by nanopatterning capabilities.

Between microscopy and iterative phase retrieval, holography represents a middle way in that the required optics can be very simple and thus aberration free. In

Fourier-transform holography — which is a particularly well suited geometry for use with short-wavelength radiation — the optics can be as simple as a pinhole in the object plane, approximating a point source that provides a reference beam3. Akin to a pinhole camera, a small pinhole produces high spatial resolution at the expense of image brightness, thus limiting the sensitivity of the imaging technique. As illustrated in Fig. 1a, the use of several pinholes can increase the imaging efficiency and hence the overall sensitivity, as each pinhole will produce a separate image of the object (plus an associated twin image)4. For each individual image

Object plane Holographic image

Figure 1 in Fourier-transform holography, each reference point in the object plane (left) will produce a pair of images of the object on reconstruction (right). a, the images can be non-overlapping if the references are spread out, requiring a large illuminated area. b, Pooling the references together enables the generation of brighter images, as many points and the object can be located closely together in a more focused illuminated spot. the use of a uniformly redundant coded reference array allows the overlapping images to be disentangled at high spatial resolution.

Introducing coded apertures to X-ray laser holography paves a route to efficient imaging at the nanoscale.

X-ray HoloGraPHy

the hole story

© 2008 Macmillan Publishers Limited. All rights reserved.

NEWS & VIEWS

530� nature photonics | VOL 2 | SEPTEMBER 2008 | www.nature.com/naturephotonics

to be separated from the next a large area must be coherently illuminated — the more pinholes, the larger the area.

By applying coded-aperture imaging concepts, already used in astronomy, to nanoscale X-ray holography, Marchesini et al. show that the efficiency can be increased by pooling many pinholes close to the sample. Furthermore, placement of the pinholes in a uniformly redundant array (URA), as shown in a very simple case in Fig. 1b, helps to augment the reference-wave intensity uniformly over a large region in reciprocal space, enabling high spatial resolution5. In this case, the resolution of the reconstructed image depends on the URA pattern, that is, on both the individual pinhole size and the spatial arrangement. The researchers demonstrate that the resolution can be enhanced beyond these limits by subsequent oversampling phasing of the dataset. Importantly, using amplitude and phase information obtained by URA-based holography increases the reliability of the iterative phase retrieval.

The underlying optical concepts used by Marchesini et al. are not new,

so why is this work important? Apart from the fact that the team mastered the nanotechnology to actually transfer the concepts to the soft-X-ray regime, their approach arrives at the beginning of a revolution in the X-ray research world. With the advent of the soft-X-ray free-electron laser facility FLASH in Hamburg6 and hard-X-ray facilities around the corner, the X-ray community is going through a transition that can only be likened to the transition from mercury arc lamps to lasers in the visible spectral range. Table-top sources based on high-harmonic generation in plasmas contribute to this revolution in the extreme-UV part of the spectrum7. Soon, for the first time since Röntgen’s discovery of X-rays in 1895, there will be X-ray sources with sufficiently high coherent flux to comfortably exploit interference and to observe nonlinear effects. The approach demonstrated by Marchesini et al. means that these sources for imaging can be taken advantage of in a way that is efficient while keeping the photon flux density on the sample high, as required, for example, for the observation of

nonlinear phenomena. More efficient imaging allows more information to be captured in a single shot and thus makes it possible to take advantage of the femtosecond pulsed nature of the new X-ray sources — as Marchesini et al. demonstrated by taking a 15-fs snapshot image of a bacterium with a resolution of 75 nm. Although not yet reaching the spatial resolution achieved by X-ray microscopy based on Fresnel zone plates8, the approach will directly benefit from further X-ray-source and nanotechnology improvements, and has great potential to become the method of choice for high-resolution femtosecond X-ray imaging. With this work, we’ve come one step closer to the ultimate X-ray vision on the nanoscale.

references1. Marchesini, S. et al. Nature Photon. 2, 560–563 (2008).2. Miao, J. W., Charalambous, P., Kirz, J. & Sayre, D. Nature

400, 342–344 (1999).3. Eisebitt, S. et al. Nature 432, 885–888 (2004).4. Schlotter, W. F. et al. Appl. Phys. Lett. 89, 163112 (2006).5. Fenimore, E. E. & Cannon, T. M. Appl. Opt. 17, 337 (1978).6. Ackermann, W. et al. Nature Photon. 1, 336–342 (2007).7. Wang, Y. et al. Nature Photon. 2, 94–98 (2008).8. Chao, W. L., Harteneck, D., Liddle, J. A., Anderson, E. H. &

Attwood, D. T. Nature 435, 1210 (2005).