Ever since the discovery of X-rays, the foremost application of radiography was to reveal the hidden inner structures of objects not transparent to visible light. The spatial resolution of radiographs can now be increased, since the size of the source becomes smaller. Strictly speaking, a small source used at a large distance is the ideal case. At the ESRF, this leads to extending beamlines to 150 m in the case of medical (beamline ID17) and topographic (beamline ID19) imaging. Even at this distance, synchrotron radiation from insertion devices possesses a very high flux.

While it was originally intended to use radiography and tomography with the usual amplitude contrast, another type of contrast due to coherence (phase contrast) turned up almost as a surprise.

When waves originate from a point source, they spread out as undistorted spherical wavefronts and produce a uniform illumination on a film or fluorescent screen. A thin, almost non-absorbing object in the path of such a wave introduces slight local distortions of the wavefront depending on the differences of the index of refraction in different regions of the object.

In soft human tissue, where almost no amplitude contrast can be seen between muscle and fat, the small difference in the refractive index distorts the wavefront. Right behind the object, nevertheless, no contrast will be observed. Only when the wavefront is allowed to propagate over roughly half a metre, this distortion of the wavefront transforms into an intensity pattern which mainly enhances the borderlines between materials with different refractive indices. This may lead to a second important branch of radiography for problems up to now not accessible for absorption imaging.

Examples of coherent imaging and other imaging techniques are presented in this section as well as in the section "Life Sciences".


Phase radiography and topography to characterise defects in quasicrystals


Quasicrystals (QCs) are a new class of solids which are ordered but non-periodic, exhibiting symmetries forbidden in ordinary crystals. Despite non-periodicity, QCs produce sharp Bragg diffraction peaks, which can be indexed in a 6-dimensional periodic hyperspace. The study of QC defects is an important and controversial topic. It has been proposed, theoretically, that the strain field around QC defects can be considered as the sum of two components, each refered to a 3D subspace: the physical space (phonon strain) and the perpendicular one (phason strain). But the actual extent of the strain field in each subspace and the behaviour of the defects under mechanical and thermal treatments remain to be determined.

The availability of high quality QCs, as evidenced by the fine diffraction peaks (~ 1') they produce, allowed the X-ray topographic (Bragg diffraction imaging) investigation of their defects. Entangled loop-shaped defects (LSDs) were observed by this technique, in icosahedral single QCs of both AlPdMn and AlCuFe alloys either immediately after growth or after further annealing. These experiments showed that LSDs increase in density and size (up to a mean size of a few hundred microns) in annealed grains. The characterisation of these LSDs appeared to be complicated both because they are not isolated and because of the double component of the strain field. Furthermore we never observed LSDs at a lower level of size, by examining the same grains by electron microscopy.

We have demonstrated that important progress in the understanding of LSDs can be expected by combining two X-ray imaging techniques which take advantage of the unique features of ESRF sources: phase radiography and X-ray topography. The transverse coherence of such third-generation synchrotron radiation sources makes it possible to image features ("phase objects") that affect the phase of a transmitted X-ray beam, just using free space propagation (Fresnel diffraction) [1].

Moreover the small source size and angular divergence (0.1-1 microradian) improve the resolution of topographic images ( 1 µm) and the high flux allows images to be recorded in monochromatic beam Bragg diffraction, very far from the Bragg peak (weak beam), with reasonable exposure times, for the investigation of the defect core. Weak beam images show that the LSDs are associated with elongated, very misoriented (up to 6') regions [2,3].

In monochromatic beam Bragg diffraction imaging, the beam transmitted through the sample shows the defects that, through the strain field in the crystal or quasicrystal, lead to diffraction ("topographic") contrast, in reverse because more diffracted intensity means depleting the transmitted beam.

At the beamline ID19, it was possible to simultaneously record phase radiography and X-ray diffraction images using the experimental set-up of Figure 1.

In this experimentally very simple approach, the sample is illuminated by a monochromatic beam, and a position-sensitive detector (high-resolution film or camera) is placed at an appreciable distance (~ 0.5 m) downstream.

Figure 2 shows a simple phase radiograph, obtained in transmission with no strong Bragg reflection excited, of an as-grown AlPdMn grain. Figure 3 represents the combination of phase radiography and reverse Bragg-diffraction contrast obtained by setting an annealed grain for a strong reflection, and Figure 4 the corresponding Bragg-diffraction image ("topograph").

Holes with dodecahedron shapes located in the volume of the sample are strikingly observed, for the first time by a non-destructive technique, both on the phase-alone and combined images. Their natural faces are strongly delineated by Fresnel fringes. They are of various sizes but are all oriented in the same way, reflecting the symmetry of the quasicrystal, in the as-grown grain. The hole shape is modified under annealing. Inclusions, lamella-shaped and with lower density than the quasicrystalline matrix, form mostly around the holes during annealings.

These results show that the LSDs observed on the topographs can be directly related to the combination of holes and precipitates inducing strains within the quasicrystalline matrix [4]. Besides, by recording phase radiograph for two angular positions of the sample, it is possible to determine the hole distribution in the volume of the grain. These measurements, together with the observation of hole modification and precipitate formation after annealing, are important steps towards the understanding of the growth mechanism and defect diffusion in QCs.



[1] P. Cloetens (a,b), M. Pateyron-Salomé (a,c), J.Y. Buffière (d), G. Peix (e), J. Baruchel (a), F. Peyrin (a), M. Schlenker (f), J. Appl. Phys. 81 (1997) 5878.
[2] E. Reinier (g), J. Gastaldi (g), L. Mancini (a), J. Härtwig (a), J. Baruchel (a), N. Baluc (h), Proceedings of the "6th International Conference on Quasicrystals", Tokyo (Japan), 1997 (in press).
[3] E. Reinier (g), L. Mancini (a), J. Gastaldi (g), N. Baluc (h), J. Härtwig (a), J. Baruchel (a), submitted to Physica B.
[4] L. Mancini (a), E. Reinier (g), P. Cloetens (a, b), J. Gastaldi (g), J. Härtwig (a), M. Schlenker (f), J. Baruchel (a), submitted to Phil. Mag.

(a) ESRF
(b) EMAT-RUCA, Antwerp (Belgium)
(c) CREATIS, Lyon (France)
(d) GEMPPM, INSA, Lyon (France)
(e) CNDRI, INSA, Lyon (France)
(f) Lab. Louis Néel, CNRS, Grenoble (France)
(g) CRMC2 - CNRS, Marseille (France)
(h) EPFL, Institut de Génie Atomique, Lausanne (Switzerland)




Diffraction imaging investigation of magneto-acoustic effects

An alternative way to excite ultrasonic waves in crystals, without any contact which could disturb the phenomenon, is possible when dealing with magnetic crystals. The vibration of the magnetic moments (magnons) can lead, through the magneto-elastic coupling, to lattice waves (phonons). This is the magneto-acoustic effect. A good candidate to investigate this effect is iron borate, which is available as a very perfect crystal and displays easy plane weak ferromagnetism and a strong magneto-acoustic coupling. Synchrotron radiation topography is a unique tool to visualise the periodic distortion produced in this way within the sample.

Ultrasonic waves are introduced in an FeBO3 (111) platelet-shaped crystal using a very simple device schematised in Figure 5. A constant field Ho (~ 103 A/m), allows the domains to be removed and an a.c. magnetic field Hac is varied in the range of 1-2.5 MHz in order to find the frequencies where the induced vibration excites a resonance of the crystal. The weak ferromagnetic component aligns with the a.c. field because of the vanishingly small anisotropy within the (111) magnetically easy plane. This magnetic vibration produces, through the magneto-elastic coupling, the propagation of elastic waves and the occurrence of standing waves.

Figure 6 shows the white beam topographs obtained on ID19 at a resonance (~ 1.3 Mhz) when varying the crystal-to-film distance. The standing-waves-related contrast is hardly visible for the shortest distance (Figure 6a), is very sharp around 45 cm (Figures 6b and 6c), and the image broadens when the distance is further increased (Figure 6d). This corresponds to a focusing effect (Figure 7), the two extreme positions of the vibration being mainly imaged because they correspond to zero acceleration (change of sign of the velocity).

A model was produced which accounts for the presented results and the monochromatic beam experiments performed on the same sample. These calculations of the intensity as a function of the position on the detector include spatial and time integrations at the sample level and allow a precise determination of the details of the vibration [1].

It has to be noted that synchrotron radiation topography not only allows the visualisation of the magnetoacoustic standing waves, but also the measurement of the sound velocity and, through the recording of images as a function of distance, the determination of the amplitude, shape and polarisation of the magneto-acoustic vibrations, which are not easy to characterise otherwise.


[1] I. Matsouli (a,b), V. Kvardakov (a,c), J. Espeso (a,d), L. Chabert (a), J. Baruchel (a), to be published.

(a) ESRF
(b) Univ. of Warwick (UK)
(c) Kurchatov Institute, Moscow (Russia)
(d) Univ. de Cantabria, Santander (Spain)



Diffraction imaging with coherent high-energy X-rays

With the appearance of third generation synchrotrons, the conventional imaging techniques such as topography and tomography gain new features closely connected to the advanced properties of the source, namely very small source size, highly collimated beam and a large distance to sample. In the high-energy domain of X-rays where absorption by matter decreases, these techniques, based on the determination of an amplitude modulation of diffracted X-rays, suffer from insufficient sensitivity or spatial resolution. Hopefully, this can be overcome by means of using highly coherent X-rays. First of all, they allow one to vary the sample-to-image distance, still keeping the spatial resolution of the image as high as about 1 µm. Secondly, high resolution images at relatively large distance hold information about the phase changes in the reflected beam through the diffraction or interference phenomenon in vacuum. By this means, phase contrast topography becomes an excellent technique to study objects whose features do not emerge in the amplitude contrast. Here we would like to present some experimental results on the grazing incidence surface diffraction imaging and crystal topography with coherent high energy X-rays (10 keV < E < 30 keV). The experiments were conducted at the ESRF Optics beamline BM5. They consisted in recording an intensity distribution in the beam a) specularly reflected by the rough surface of an X-ray mirror, and b) Bragg-diffracted by crystals with defects situated in the surface layers.


Topographic characterisation of a mirror surface with Å-resolution

The experiments performed demonstrated that X-ray mirrors, which have been using the very best potentialities of modern technology, give rise to a speckle structure in a reflected X-ray beam. The observed speckle structure was recorded with a high-resolution CCD camera, as well as by means of the cross-scanning of a pinhole coupled with a conventional detector (Figure 8). The theoretical analysis of a partially coherent X-ray beam, scattered by a moderately rough mirror surface under total reflection conditions, shows that one can evaluate the main surface parameters simply by processing a single image obtained in coherent X-rays. In the approximation of a point source, and for the classical exponential and Gaussian functions of relief profile, the image contrast rms can be evaluated as

From the intensity measurements shown in Figure 9 ( i = 0.071) one can immediately deduce m = 2.75 Å, which is in good agreement with the values obtained by means of other techniques, namely a 1.8 Å rms finish, as measured in a standard way in the ESRF metrology laboratory, and 4 Å as measured with the diffuse scattering technique. In a more comprehensive way, the proper treatment of the images taken at various distances solves the inverse problem of surface profile retrieval. This means that one can obtain statistical characteristics not only of surface roughness, but of a particular relief profile.


New possibility in diffraction topography for crystals

The samples under investigation had different degrees of perfection: from the crystals with strongly deformed areas, such as InP/InGaA structures grown by selective area epitaxy in-between oxide strips, up to almost perfect substrates with very weak and localised deformation fields in the near surface layer caused by thin oxide film. In contrast to specular reflection, Bragg diffraction occurs in the bulk. Thus, the deformations localised in a small volume or oriented perpendicular to the diffraction plane are invisible in the images taken right after a sample (in Figure 10, Ko, Kh denote directions of incident and reflected beams respectively, b denotes the deformation gradient). Figures 11a and 11b show the topographs from oxide strip taken close to the sample and at some distance.

Along with the experiments, a theoretical analysis of simple types of defects in perfect crystal substrate was performed, which was intended to reveal the high sensitivity of the method while using coherent X-rays. The plot in Figure 12 represents a computer simulation of the image intensity.

For example, for Si(111) reflection, it can be shown that the minimum oxide strip thickness that can be resolved by coherent X-ray topography with 5% intensity contrast is about 20 nm. In this case, semi-infinite oxide film on the substrate yields in the deformations with mean value d/d ~ 2 x 10-6. An estimation of the minimum oxide strip width with a thickness of 0.1 µm that can be resolved at the image plane is about 0.15 µm.


[1] A. Souvorov (a), I. Snigireva (a), A. Snigirev (a), X-TOP'96, 80 (1996).
[2] A. Souvorov (a), I. Snigireva (a), A. Snigirev (a), Proc. SPIE, vol. 3113, 476 (1997).
[3] S. Kuznetsov (b), A. Souvorov (a), I. Snigireva (a), A. Snigirev (a), to be published.

(a) ESRF
(b) Kurchatov Institute (Russia)




X-ray tomography with micrometer spatial resolution

Obtaining three-dimensional information on a microscopic scale of a sample without destroying it has long been of high interest in many areas of research. Phase-contrast tomography with coherent high energy X-rays is a very promising microprobe for various applications, due to the high sensitivity to borders and interfaces in a sample. The main limitation was until recently the limited spatial resolution of on-line X-ray detectors of 5-8 µm. The resolution of CCD-based X-ray detectors was improved considerably to 1.6 µm at the ESRF by using transparent crystalline luminescent screens in order to convert the incoming X-rays to detectable visible light.

In a beam with a wavelength of about 1 Å, and at a distance from 2 to 10 cm, transparent objects show sharp contrast in a small region ­ comparable with the detector resolution ­ around edges and borders in the sample (outline imaging mode). Thus, the outline image ressembles the features of the sample in a good approximation, with contrast enhancement at edges and interfaces in the object. Combining phase-contrast imaging with a microtomography set-up enables one to obtain 3D information on interfaces and borders in the sample with micrometer resolution (Figure 13).

The applicability of high-resolution tomography is demonstrated by imaging a spider fang (sample courtesy of A. Thompson). In Figure 14 a projection through the fang is shown at 20 keV at a distance of 6 cm. To reconstruct the internal structure in 3D, 500 images were recorded at the ID22 undulator beamline over 180 degrees of angle of projection and the exposure time was 500 msec. In the cross-sections (Figure 15) the inner channel is visible as well as a bunch of very thin (Ø ~ 1 µm) fibres or tubes around the inner channel. From reconstructing 60 slices, a 3D rendering of the fang was made (Figure 16).

Various applications can be seen in materials science and biology, where samples have to be imaged in a natural state without dyeing or drying, and where three-dimensional information is needed with spatial resolution down to a micrometer.


[1] C. Raven (a), A. Koch (a), and A. Snigirev (a), CellVision, 4:157, 1997.
[2] A. Koch (a), C. Raven (a), P. Spanne (a) and A. Snigirev (a). to be published in J. Opt. Soc. Am.
[3] C. Raven (a), A. Snigirev (a), A. Koch (a), I. Snigireva (a), V. Kohn (b), SPIE San Diego, 1997.

(a) ESRF
(b) Kurchatov Institute (Russia)




High-resolution phase-contrast microscopy with an X-ray waveguide


Contrast, in X-ray imaging techniques such as radiography and tomography, is due to differences in the absorption coefficient of the different elements in the sample. For low density materials, such as polymeric or biological samples, the sample absorption often becomes too small to give a detectable contrast. An alternative is the so-called phase contrast technique which is based on the phase modulation of a coherent beam induced by an object. While this technique is sensitive to organic matter, its main drawback is that it is limited in its classical, lens-less free application by the detector resolution to about 1 micrometer.

On the Microfocus beamline (ID13) a new scheme of phase-contrast radiography is based on the use of a coherent and divergent submicrometer beam exiting an X-ray waveguide. The X-ray waveguide is a thin film resonator of a low absorbing material enclosed between two metal layers with smaller refractive index. The beam leaves the waveguide with a vertical size which can be as small as 0.13 micrometer and a divergence of 1 mrad. The dimension and divergence of the source together with the geometry of the experiment induce a magnification effect which allows a submicrometer resolution to be obtained with a standard detector.

In the simple set-up (Figure 17) the magnification can be controlled by changing the position of the sample if the distance from the detector to the waveguide is fixed. Because of the magnification, no high-resolution detector is needed. This scheme works at present only in one dimension.

Figure 18 shows interference fringes recorded from the edge of a 12 micrometer thick nylon fiber, placed at a distance of 4.1 mm from the waveguide. The interference fringes can be quantitatively simulated as shown in Figure 19. In this case a magnification of 241 has been reached, but magnifications as high as about 800 have been demonstrated and further progress can be expected from an improvement of the set-up. In principle the method could be extended into two dimensions which will be a challenge for future developments.


A. Cedola (a, b), P. Cloetens (b, c), S. Di Fonzo (d), W. Jark (d), S. Lagomarsino (e), G. Soullié (d) and C. Riekel (b), Appl. Phys. Lett., in press.

a) INFM and IESS-CNR Rome (Italy)
c) EMAT, University of Antwerp (RUCA) (Belgium)
e) IESS - CNR (Italy)




The Talbot effect revisited

"... On removing the lens a little further from the grating, the bands gradually changed their colours, and became alternately blue and yellow. When the lens was a little more removed, the bands again became red and green. And this change continued to take place for an indefinite number of times, as the distance between the lens and grating increased...."
From: "Facts relating to Optical Science. No IV. By H.F. Talbot, Esq., F.R.S.". The London and Edinburgh Philosophical Magazine and Journal of Science (Third Series), 9, 401-407 (1836)


Although synchrotron radiation images are not colourful (without the help of a computer), it nevertheless seemed worthwhile to re-visit the Talbot effect, using hard X-rays: this effect provides an efficient way of quantitatively assessing the coherence of the X-ray beam.

In 1836, Talbot discovered the self-replication (periodic as a function of the defocusing distance) of periodic objects under coherent illumination. This effect, later explained by Lord Rayleigh, is now considered one of the most spectacular manifestations of Fresnel diffraction. In the last few years, it has raised new interest and found applications in visible light optics.

The Fresnel diffraction approach can be used in a straightforward way for the X-ray case, with the extra simplification that, without lenses to produce an image, the only strictly "in focus" position is at the specimen itself. It says that, under plane wave (completely coherent) illumination, the intensity distribution at a distance D downstream of a periodic object, with spatial period a along the x direction, should reproduce exactly at distances that are integer multiples of the Talbot distance DT = 2a2/, where is the wavelength of the radiation used. It also says that, at distances that are a rational fraction (M/N) of DT , i.e. when D = (M/N)DT, images simply related to the object should appear. Formally, the amplitude distribution D(x) can be expressed in terms of the transmittance T(x) of the object as the superposition, with proper coefficients, of N replicas of the object laterally shifted by a multiple of a/N. For example, increasing the distance by DT/2 should just shift the pattern by a/2.

The experiment has been performed at ID19 using optical gratings, with 2,000 or 4,000 LPI (line pairs per inch). These plastic gratings behave as pure phase objects for the X-ray energy (18 keV) used. The experimental set-up was extremely simple: the monochromatic beam from a perfect silicon monochromator went through the grating, and a high-resolution film was placed at various distances downstream. As shown on Figure 20a, there is indeed no contrast at all when the film is set right after the sample (a phase object gives no variation in intensity until Fresnel diffraction has taken over to produce interference between different parts of the beam). Contrast then builds up as the film is set further from the specimen (increasing D), decreases again to zero for D = DT/2, and reappears as D is further increased. The intensity profiles of Figure 20b show this in more detail.

Going over to a more quantitative view, it is observed that, when the amplitude of the first Fourier component of the periodic intensity distribution is plotted versus distance (Figure 21), the behaviour is not strictly periodic. This can be related to the deviation from perfect coherence of the beam and is indeed a good measure of the lateral coherence of the beam, i.e. of its angular divergence. The value obtained is 1.4 µrad larger than expected from the size of the source and its distance from the sample. It is believed that the broadening is due to vibration of the monochromator through the cooling system. An analysis of the process involved indicates that this method may be a very good one for this kind of measurement. In particular, it requires at least two pictures (a couple of minutes of exposure time), but involves the response of the detector at one spatial frequency only, thus minimising the response and noise problems.


P. Cloetens (a, b), J.P. Guigay (a, c), C. De Martino (a), J. Baruchel (a), and M. Schlenker (c), Opt. Lett. 22 (1997).

(a) ESRF
(b) EMAT, University of Antwerp, (Belgium)
(c) Laboratoire Louis Néel - CNRS, Grenoble (France)