Among the most important developments that are pursued at the ESRF is the ability to take full advantage of the coherence properties of the X-ray beam. New imaging techniques have appeared in the forefront, which have been improved year after year. In-line holography, for example, has switched from purely qualitative information to more complete (quantitative) information, showing an excellent spatial resolution and a very high signal-to-noise level thanks to the FRELON camera developed at the ESRF. The optics are also continually improved, allowing microfocusing of the beam. Microfluorescence and microspectroscopy and other microprobe techniques find new applications in semiconductor structures, biology, geology, archeology and environmental studies.


High resolution imaging microspectroscopy

X-ray fluorescence microscopy and microspectroscopy are finding applications to problems in bulk material analysis, semiconductor structures, biology, geology and environmental studies. They can provide information on the spatial distribution, oxidation state, chemical environment, and chemical transformations of trace elements. ID22 started USM operation at the end of 1997 but the microfluorescence set-up was commissioned only recently. Among the various experiments performed using the micro-focused beams, two microspectroscopy applications have been selected.

The first deals with the analysis of fluid inclusions trapped in quartz for earth sciences studies. These inclusions are 20-30 µm gas/liquid/solid filled bubbles formed inside a crystal, which carry information on the crystal genesis and as such are unique witnesses of the paleo-circulation of fluids as well as their interactions with crust-forming rocks. Figure 109 shows the result of mapping out the As concentration; results on Au and Br were also recorded. The fixed-exit Kohzu monochromator delivered beams of energies E = 12, 15 and 22 keV onto the microfluorescence set-up located in the experimental hutch EH1. The versatile layout presented in Figure 110 comprises an FZP focusing lens, a pinhole assembly, a precision sample XYZ stage and an X-ray imaging device - medium/high resolution CCD camera. The beamspot was about 0.5 x 8 µm except at E = 22 keV, where the front end Compound Refractive Lenses (CRL) were used in combination with a 15 µm pinhole.

The current research interest concerns the presence of metals of economic or environmental interest such as Ag, Pb and U. Results obtained tracking out Au presence are already being processed and will be presented shortly [1].

The second part of thestudy involved the corrosion analysis of Roman glasses from the Jordan valley. This is of great interest to the archeological community. These measurements are pioneering steps in the non-destructive microstructure analysis of archeological artifacts and represent the highest resolution micro-Xanes measurements ever performed. The knowledge of the structure and concentration gradients as well as oxidation states inside the corrosion layers is necessary for the development of a protection and restoration strategy. In these samples, after prolonged exposure to moisture, corrosion layers of mm-size appear and propagate in a hemispherical shape. Leached-out glass allows the penetration of various elements into the µm-sized cracks, forming new separated layers from the original substrate. Maps 1 and 2 from Figure 111 are derived at two energies that were identified in the Xanes spectra as exhibiting mostly MN IV+ and II+ contributions. By weighed subtraction, map 3 is obtained, which is mostly Mn IV+, revealing the predominance of MnO2 in the crevices between corrosion shells [2].

[1] P. Philippot (a), B. Menez (a), M. Drakopoulos (b), A. Simionovici (b), I. Snigireva (b), A. Snigirev (b), submitted to Geol. Chem.
[2] K. Janssens (c), A. Simionovici (b), A. Rindby (d), B. Salbu (e), T. Krekling (e), A. Snigirev (b), accepted in EDXRS-98.

(a) Labo. de Pétrologie, Univ. Paris VI (France)
(b) ESRF
(c) University of Antwerp (Belgium)
(d) Chalmers Institute (Sweden)
(e) University of Ås (Norway)



Quantitative coherent imaging

Phase-sensitive imaging through propagation or Fresnel diffraction is an experimentally simple tool to image light materials with better contrast and resolution than classical absorption radiography. The variations in X-ray optical path length correspond to variations in thickness and/or electronic density.

The majority of the work up to now has used the edge-detection regime to image the phase discontinuities in the object. The phase jumps are useful, but the extracted information is qualitative. The complete information entangled in the defocused images can be used to retrieve the phase of the wave exiting the object, i.e. the projection of the electronic density in the sample. This "holographic" reconstruction provides a projection of the object with substantially improved spatial resolution and with a straightforward interpretation.

When combined with the geometry of computed tomography, phase imaging reveals the three-dimensional structure of the object. Algorithms developed for absorption tomography were used first, as a make-shift solution, for the phase-sensitive case. This gave rise to artefacts and did not reveal quantitatively the structure of the object. The combination of a "holographic" and the classical "tomographic" reconstruction allows a complete determination of the distribution of the electron density in the sample.

The basis of the reconstruction algorithm is that the in-line holograms depend strongly on the propagation distance D between the sample and the detector, with each distance optimally sensitive to one length scale of the object. The phase modulation is retrieved from a series of in-line holograms by back-propagating the useful information to the exit of the object. The single-phase modulation map that best fits all the observed Fresnel diffraction patterns is then computed. This method, based on algorithms developed for high-resolution electron microscopy, can handle any number of distances simultaneously. It is direct in a first approach, allowing on-line treatment of the holograms. The initial solution can then be optimised with a fast iterative procedure.

A 1 mm thick piece of polystyrene foam, with negligible attenuation, was used as a test. 16 images were recorded in the unfocused X-ray beam of ID19, monochromatized at 18 keV, for distances D ranging from 0.01 to 0.91 m, with a high-resolution detector based on a transparent luminescent screen (A. Koch) and the Frelon CCD camera (J.C. Labiche).

Figure 112a shows some of the recorded images. The contrast and width of the interference fringes increase strongly with D. The reconstructed false color phase map of Figure 112b, shows that the wider structures correspond to stronger phase modulation, indicating that the skeleton has rod-like structures. The magnified view of Figure 113a reveals the excellent spatial resolution obtained, down to the level of the pixel size of 1.9 µm. Figure 113b shows the phase profile obtained along the arrow of Figure 113a. It shows the quantitative character of the reconstruction and the very low noise level provided by the use of a large number of images.

P. Cloetens (a, b), W. Ludwig (a), D. Van Dijck (b), J. Baruchel (a), J.P. Guigay (a,c), M. Schlenker (c), submitted for publication.

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



Small-angle scattering with coherent X-rays: Speckle statistics and photon correlation spectroscopy

One of the most prominent properties of the newest synchrotron sources is their large coherent X-ray flux. At the ID10A branch of the Troika beamline coherent X-rays have been used to measure the equilibrium dynamics of disordered systems using the technique of X-ray photon correlation spectroscopy (XPCS). When a disordered sample is illuminated coherently, it produces a highly modulated &laqno;speckle» diffraction pattern, which is related to the exact spatial arrangement of the illuminated material. As the sample evolves in time, the intensity at a point in the pattern evolves with the characteristic timescale of the dynamics of the sample for that wavevector.

In contrast to a fully coherent laser source, synchrotron radiation at today's sources is only partially coherent and it is of outmost importance to develop methods of characterizing the coherence of the X-ray beam produced. One way is to understand the statistics of the speckle pattern produced by a static sample. Since the scattering does not vary in time one may take a series of measurements, and by treating the images statistically the exact nature of the sample is not important. Information is extracted about the degree of coherence of the incident beam, as well as such parameters as the source and incident aperture sizes, the detector resolution and the energy bandpass. Since the static sample can be mounted exactly as the samples to be studied dynamically, the entire measurement system can be evaluated just as it will be used in the dynamic XPCS experiment.

One important parameter is the longitudinal coherence length 1 = x /, which is a measure of the temporal coherence of the beam. Because XPCS measurements are typically flux-limited, due to the extreme filtering necessary to produce a coherent beam, it is essential to understand the effects of changing the energy bandpass. Broadening the energy width decreases the longitudinal coherence length but also improves the flux. It must be understood, however, how this affects the ability to perform XPCS in a practical way. Two experimental set-ups have been evaluated by studying the wavevector dependence of the intensity-intensity correlation function of static speckle patterns. In one case, a Si(111) symmetric Bragg monochromator gives a bandpass of / = 1x10-4. A bandpass of 1x10-2 is achieved by using a simple Si mirror to isolate the third harmonic of the undulator spectrum to produce a "pink" beam.

Figure 114 (a) shows the layout of ID10A schematically as used to produce and characterize the nominally coherent X-rays. The 8 keV X-rays from the undulator source pass primary slits at 27 m, which serve to reduce the effective source size in order to increase the transverse coherence length at the sample. After passing either the monochromator or mirror at 45 m, a pinhole aperture of 12 µm diameter selects a nominally coherent beam. The sample is placed roughly 12 cm after the pinhole, with a guard slit mounted just in front to cut out stray scattering. The small-angle scattering pattern of the sample is measured by a directly-illuminated CCD camera some 2 m away. The resolution of such a camera is limited simply by the pixel size of 22.5 µm.

Portions of the speckle patterns from a porous glass sample (tradename Vycor) are shown in Figure 114 (b) in the monochromatic case and Figure 114 (c) for the "pink" beam case. The average scattering in the two cases is the same, showing a peak due to the pore-pore correlations of the glass at Q = 0.024 Å-1. However there is a striking difference in the nature of the speckle pattern produced in the two cases. This is due to the wider bandpass of the "pink" beam, which smears out the speckles in the radial direction due to the larger uncertainty in the wavelength.

One experimentally relevant way to quantify the difference between the two speckle patterns is to find the contrast as a function of the wavevector transfer Q, where contrast is simply defined as the mean-squared deviations of the intensities in a given region of the image after normalizing by the circularly averaged scattering pattern and correcting for Poisson statistics. Figure 115 shows this comparison between monochromatic and wide-bandpass beam. The wider bandpass has the effect of reducing the contrast as Q grows larger. Models of the expected results for the particular experimental configurations are shown by the solid lines, confirming that quantitative predictions of the contrast may be made. This is important experimentally, since the contrast of a static speckle pattern is the same as that expected while doing a dynamic experiment which measures the normalized time autocorrelation function. Too low a value of the contrast will make the experiment difficult, since a much longer acquisition time will be needed to get a significant measure of the dynamics. Although the &laqno;pink» beam configuration gives roughly two orders of magnitude more flux, its utility for doing photon correlation spectroscopy will be limited to a smaller Q range due to the reduction of the contrast.

The study of the statistics of static speckle patterns is a pre-requisite for using the beam to measure the dynamics of time-evolving systems. This has already been achieved for a variety of samples, including colloidal suspensions of particles and polymeric micelles for example. One interesting case is that of a suspension of magnetic particles, or ferrofluid [2]. The question of how the dynamics of such a system will change when an external magnetic field is applied is still under debate. Figure 116 shows the small-angle scattering from a ferrofluid sample measured on beamline ID2. In zero-applied field (a) the scattering is isotropic. When a field is applied horizontally perpendicular to the incident beam, the scattering develops a highly anisotropic character. The strong scattering in the vertical direction near the origin indicates the orientational ordering of agglomerated ferrofluidic particles. Dynamic XPCS studies at ID10A reveal evidence for both agglomerations under field and anisotropic dynamics.

[1] D.L. Abernathy (a), G. Grübel (a), S. Brauer (b), I. McNulty (b), G.B. Stephenson (b), S.G.J. Mochrie (c), A.R. Sandy (c), N. Mulders (d) and M. Sutton(e), J. Synchrotron Rad. 5, 37-47 (1998).
[2] G. Grübel (a), D.L. Abernathy (a), J. Lal (a), O. Diat (a), to be published.

(a) ESRF
(b) Argonne National Lab (USA)
(c) MIT (USA)
(d) University of Delaware (USA)
(e) McGill University



Turbo-XAS: Sequential acquisition using the dispersive focussing optics

A new experimental technique for time-resolved X-ray absorption studies in the sub-second range has been successfully tested on the Energy Dispersive XAS beamline (ID24). It consists in a sequential acquisition of energy points using the dispersive optics scheme installed on the beamline. Turbo-XAS (T-XAS) takes full advantage of the properties of third generation sources, overcoming many of the problems encountered in the classical Energy Dispersive XAS mode, based on parallel acquisition using position-sensitive detectors. The new technique benefits from the basic assets of the dispersive set-up, i.e. the absence of movement of the polychromator and the extremely small and stable horizontal focal spot, but also features advantages which are typical of double-crystal scanning monochromators set-ups such as the simultaneous recording of I0 and I1 and the possibility to perform fluorescence and electron detection.

The experience gained on ID24 since the beginning of its operation has shown that dispersive set-ups for X-ray absorption spectroscopy at third generation sources do in fact profit from the extremely small source size, yielding an unprecedented brilliance (1012 photons/s measured on a 20 µm FWHM polychromatic focus spot). But new problems have appeared, due to the much higher sensitivity to source instabilities, amplified by coherence-related phenomena and small-angle scattering from the sample. Non-perfect optics and Be windows introduce phase contrast structures (or "speckles") which appear as irregular fringes on the image formed on the position sensitive detector. In addition, powder samples introduce small-angle X-ray scattering (SAXS) which also affects the image.

In order to overcome these difficulties, a new scheme has been developed for acquiring time-resolved absorption spectra. In T-XAS, a narrow (~30 - 50 µm) slit is scanned through the polychromatic fan of radiation downstream the crystal, selecting a monochromatic beam, the intensity of which is simultaneously recorded before and after the sample by two ionisation chambers.

The simultaneous recording of I0 and I1 is the essential new feature with respect to D-XAS, leading to an experimental set-up which is intrinsically less sensitive to beam instabilities. Moreover, SAXS from powders does not affect energy resolution since the energy-position correlation is absent in this scheme.

The simple mechanical movement involved (linear translation of the slit) leads to a high time resolution. Preliminary XANES and EXAFS spectra have been recorded in ~ 100 ms and 500 ms respectively. In addition, T-XAS profits from the small and stable horizontal focal point and high-energy scale stability and reproducibility of the dispersive set-up thanks to the de-coupling of the monochromatization and of the focusing action.

The main applications of the new technique envisaged on ID24 are in the field of "slow" varying dynamical processes (on the second timescale), mainly in the transmission mode, but also using the reflection mode or fluorescence/electron yield mode, and studies of samples in extreme conditions of pressure and temperature.

The first application of the new acquisition mode has been devoted to the study of the evolution of undercooled Ge (Figures 117 and 118). Such an investigation is of fundamental interest due to the interplay between the covalent and metallic nature of the bondings which characterize the solid and liquid phases of Ge respectively. The sample consists of fine Ge powder dispersed in a BN matrix: due to SAXS from the powder, it was not possible to perform this experiment using the conventional parallel data acquisition mode.

S. Pascarelli (a), T. Neisius (a), S. DePanfilis (a), to be published.

(a) ESRF.



Uranium sensitive tomography with synchrotron radiation

Element-sensitive tomography has many potential applications in materials science and engineering. In the case of uranium-containing minerals, the distribution in the ore is of interest as well as the ability to follow up the diffusion of nuclear waste in the storage material.

A standard procedure for tomography is to measure the attenuation of radiation through a sample for several orientations. The recorded patterns are then used to reconstruct the three-dimensional attenuation distribution for the whole sample. By scanning a sample using photons having two different energies, one just below and one just above the energy of the K-edge of the element under study, the method becomes element-sensitive because the difference images reveal only this element. If the elements to be discovered become very heavy, like actinides, the photon energies approach the gamma-ray domain. The main problem is then to obtain a monochromatic photon beam of sufficiently high energy (E > 100 keV) and intensity. The advantage at this high energy is a big penetration depth allowing the use of thick samples.

Earlier work on heavy-element detection relied on crystal diffraction of bremsstrahlung to provide the incident beam and semiconductor detectors for the detection. Because of limited source intensity and detector sensitivity, these studies yielded only integrated two-dimensional distribution maps [1]. The tremendous improvements in synchrotron radiation as provided at the high-energy beamline ID15 allowed the transfer of the technique to photon beams with intensities that are several orders of magnitudes higher than hitherto.

A first experiment on uranium tomography was performed using synchrotron radiation around 115 keV [2]. Two asymmetrically-cut 5 mm thick bent Si crystals working in Laue mode provided the monochromatic beam. At the sample position, a reasonable homogeneity of the beam was obtained over a 5 mm high and 20 mm wide surface. Two energies close to the K-edge of uranium (114.6 keV and 116.6 keV with an energy bandwidth of about 0.25 keV) were chosen to perform the tomographies. The intensity of the beam at the sample was of the order of 108 photons/s/mm2. Various natural and artificial samples containing uranium were scanned. The attenuation of the beam in large samples up to 2 x 2 x 4 cm3 was measured using a luminescent screen consisting of an 80 microns thick Gd2O2S:Tb layer. The screen was followed by a mirror and CCD camera. The spatial resolution of the detection system was approximately 50 microns and is limited at this high energy by light scattering in the powder phosphor screen. The samples were chosen such that the uranium concentration, distribution and granularity were very different, allowing the assessment of the possibilities of the method. Each sample was scanned at 200 different angles. Because of the wide variety of samples, the exposure time ranged from 25 ms to 2 sec.

The quality of the data was sufficient in terms of statistics to analyse them using the filtered back-projection reconstruction technique. Examples of two reconstructed samples are given in Figure 119 in which the distribution of uranium in an artificial and a natural sample are shown. One sees clearly the inhomogenities produced when filling the holes of the artificial sample. In Figure 120 the uranium content is superposed on a photograph of the scanned stone showing the layered deposit of uranium in the natural sample.

In conclusion, it was shown that element-sensitive tomographies could be extended to the heaviest natural element. Therefore in principle the distriubtion of any element can now be reconstructed at third generation synchrotron radiation sources.

[1] M. Bertschy (a), J. Jolie (a), W. Mondelaers (b), Applied Physics, A62 (1996) 437.
[2] Th. Materna (a), J. Jolie (a), W. Mondelaers (b), B. Masschaele (b), V. Honkimaki (c), A. Koch (c), Th. Tschentscher (c), to be published.

(a) Université de Fribourg (Switzerland)
(b) Rijksuniversiteit Gent (Belgium)
(c) ESRF.


A high-energy micro-probe

Due to the penetration power, neutron diffraction is the standard non-destructive technique used to characterize the residual stress in structural bulk samples as they are often used in engineering applications. A gauge volume of about 1 mm3 is required due to the low brilliance of neutron sources. High-energy synchrotron radiation can penetrate several millimeters into important metals (e.g. the absorption length in steel at 80 keV is 2.2 mm) and the high brilliance allows for micro-focusing. A spatial resolution on the micro-meter scale can be reached and it is therefore possible to detect the local residual stress and structure of typical micro-units (grains, inclusions, cracks ).

Several techniques have been developed to achieve micro-focusing of synchrotron radiation in the classical, crystallographic energy range below 30 keV. However, most of these techniques become inefficient at high energies due to the salient features of high-energy X-rays: small Bragg angles, large penetration power and extremely slight deviation of the refractive index from unity. Therefore, a novel technique has been developed which utilizes the reflection properties of bent, asymmetric Laue-crystals [1].

The focusing geometries of thin, bent crystals are well known. In general, micro-focusing cannot be achieved due to the beam divergences caused by diffraction and due to crystal thickness effects. It turns out that both contributions can be made to compensate for one another by a correct choice of the asymmetric cut of bent Laue crystals. The diffraction geometry, as realized on the high-energy beamline ID15 is shown in Figure 121. A focal spot size of 1.2 µm was observed (Figure 122). The tails are due only to the thermal diffuse scattering within the crystal. The gain factor, the ratio of the flux through an aperture of the spot size with and without the optics, was about 300.

The micro-focused high-energy beam was used to detect residual strain gradients in a structural Cu/Ni multilayer as a function of the distance from the sample surface z [2]. Transmission geometry could be exploited where the beam is parallel to the surface and scanned across the sample. Thus, the z-dependence is probed directly as compared to reflection geometries which yield only cumulative information. Steep gradients were detected at the buried Ni-Cu interfaces (Figure 123). The results underline the unique properties of micro-focused high-energy synchrotron radiation as a local probe of the micro-structure in bulk samples.

[1] C. Schulze (a), U. Lienert (b), M. Hanfland (b), M. Lorenzen (b) and F. Zontone (b), J. Synchrotron Rad. 5 (1998) 77-81
[2] U. Lienert (b), C. Schulze (a), V. Honkimäki (b), T. Tschentscher (c), S. Garbe (d), A. Hignette (a), A. Horsewell (d), M. Lingham (a), H.F. Poulsen (d), N.B. Thomsen (e) and E. Ziegler (a), J. Synchrotron Rad. 5 (1998) 226-231.

(a) Swiss Light Source, Villingen (Switzerland)
(b) ESRF
(c) HASYLAB, Hamburg (Germany)
(d) Ris national laboratory (Denmark)
(e) Danfoss A/S, Nordborg (Denmark)


Refractive optics: from cylindrical to parabolic compound refractive lenses

Recently, it was shown that refractive optics are well adapted to undulator beams at third-generation SR sources. The advantages of the compound refractive lenses (CRL) may be summarized as follows: they are very robust and small (5 to 10 cm in length), they are easy to align and to operate, they are less sensitive to surface roughens than X-ray mirrors and they can withstand a high heatload. Some are even installed in the primary undulator beam. Although drilled holes are easy to fabricate, cylindrical CRL suffer from a number of inconveniences. In the meantime a new type of CRL with a rotational parabolic profile has been developed in collaboration with a group at the Zweites Physikalisches Institut der RWTH in Aachen. Figure 124 shows a sketch of a parabolic CRL and a cross-section through one lens. The lenses are manufactured in Aachen by a precision pressing procedure and are tested for performance at the ESRF beamline ID22. As for cylindrical CRL, low Z lens material is a pre-requisite for achieving good transmission and thus a good gain. It is also necessary to keep small-angle scattering at a low level in order not to blur the focal spot. A number of materials have been tested. Mixtures of amorphous and crystalline phases are not acceptable. Finally, the material must not deteriorate in the X-ray beam. Most plastics and many insulators are inappropriate because of their liability to radiation damage. Beryllium, boron, magnesium, aluminium among others are good candidates for lens material.

The advantages of the new lenses are the following:

- The parabolic CRL focus in both directions.

- They have virtually no spherical aberration which results in a minimum focal size.

- The transmission is larger than in the case of crossed cylindrical holes, because the number of drilled holes is 2N whereas the number of parabolic lenses is only N for the same focal length. This is particularly useful when aluminium is used as a lens material where the material in the bridges between the holes is a limiting factor of transmission. In addition, the thickness of the walls between adjacent holes can be reduced for concave lenses with rotational parabolic profile:

- The surface finish of pressed lenses is inherently better than that of drilled holes.

- The large depth of field of the CRL allows for the transmission of an energy band pass of 2%, a value typical for a harmonic of an ESRF undulator.

Figure 125 and 126 show the intensity distribution in the focal plane at 1 m from the lens and at a photon energy of 20 keV. The spot size is 1.6 µm vertical multiplied by 14 µm horizontal FWHM.

The fields of application of the new lenses are in micro-diffraction, micro-fluorescence, imaging in phase-contrast, X-ray holography and tomography. Micro-diffraction and micro-fluorescence profit primarily from the small spot size. First experiments are under way. The small image of the source makes the parabolic CRL an ideal candidate for an X-ray source with high lateral coherence length. It may be used in coherent scattering as e.g. in speckle spectroscopy. Imaging will profit in two ways from the parabolic CRL. On the one hand, the divergent beam generates a magnified image. This is very useful in view of the fact that the detector pixel size is limited at present to about a micron. On the other hand, a laterally coherent source opens interesting possibilities in imaging in phase contrast.

A first imaging experiment with a parabolic CRL was performed at 16 keV. Figure 127 shows the set-up and an image of a free-standing gold mesh with 15 µm pitch size and 3 µm bar size which was used as a test object. The X-ray image, optically enlarged by the CRL, was recorded by a high-resolution X-ray detector. Small deviations from the ideal parabolic lens profile result in some pincushion distortion of the image. Nevertheless, the high contrast of the full-field image of the mesh clearly demonstrates the great potential of the refractive optics in high-energy and high-resolution X-ray microscopy and diffraction techniques.

[1] A. Snigirev (a), V. Kohn (b), I. Snigireva (a), A. Souvorov (a), B. Lengeler (c), Applied Optics 37 653-662 (1998).
[2] B. Lengeler (c), J. Tuemmler (c), A. Snigirev (a), I. Snigireva (a), C. Raven (a), J. of applied Physics, V84, 5855-5861 (1998).
[3] B. Lengeler (c), M. Richwin (c), C. Schroer (c), J. Tuemmler (c), M. Drakopoulos (a), A. Snigirev (a), I. Snigireva (a), to be published.

(a) ESRF
(b) Kurchatov Institute, Moscow (Russia)
(c) Physikalisches Institute, RWTH Aachen (Germany)


X-ray diffraction by a piezoelectric crystal modulated by surface acoustic waves

An X-ray beam falling on a crystal excited by surface acoustic waves (lateral Rayleigh wave) is diffracted by the periodic modulation formed in the near surface region. Constructive interferences may occur in the reflected field resulting in diffraction satellites located around the specular reflected beam. They are found at angular directions given by the grating equation:



with o the incident angle, m the emergent angle of the mth order satellite, the X-ray wavelength, the acoustic wavelength.

The most frequently employed method of exciting high-amplitude SAW is to use a piezoelectric substrate on which an interdigital acoustic transducer has been deposited by photolithography. In the present case, the piezoelectric substrate is a LiNbO3 monocrystal (300 oriented) with a surface roughness of the order of 5 Å rms. The interdigital transducer has a resonance frequency of 288 MHz and emits a 12 µm wavelength wave propagating along the surface at a velocity of 3500 ms-1. The acoustic amplitude can reach 15 Å at the surface.

The rise of satellite peaks around the specular reflected beam has already been studied in several diffraction situations such as total external reflection and Bragg diffraction on a multilayer deposited on the vibrating surface. First tests were made on the optics beamline (BM5) to study the effects of surface acoustic wave propagation on the Bragg diffraction of a piezoelectric crystal.

Figure 128 is a sketch of the principle of the experiment. A monochromatic beam (12 keV) collimated by slits (1 x 0.05 mm) diffracts on the 300 planes of the piezoelectric crystal. The propagation of the travelling lateral acoustic wave acts as a dynamical grating, providing many diffraction orders around the specular diffracted peak (0 order). This can be observed in Figure 129 showing rocking curves of the crystal for

various acoustic amplitudes. An X-ray beam chopper working in the MHz range and based on this acoustic modulation has been tested. The basic idea of this chopper consists of emitting acoustic pulses in such a way that they reach the beam footprint simultaneously with the X-ray pulses to be chosen. About 10% of each of these pulses are diffracted through the low order satellites. All the other X-ray pulses hit a flat surface and are thus entirely diffracted in the specular direction. For a beam-width of 50 µm and for 110 ns long acoustic pulses, it has been possible to separate about 40 closed X-ray pulses from each superbunch made of 660 pulses in the 2/3 filling mode (see Figure 130). The main advantage of this electro-acoustic chopper is the high flexibility in the selection frequency of the X-ray pulses, which can vary from the Hz to the MHz range. The background intensity is nevertheless much too high and should be decreased by improving the crystal quality.

R. Tucoulou (a), D. V. Roshchupkin (b), I. A. Schelokov (b), O. Mathon (a), M. Brunel (c), to be published.

(a) ESRF
(b) Laboratory of X-ray acousto-optics, Institute of Microelectronics Technology, Chernogolovka, (Russia)
(c) Laboratoire de Cristallographie, CNRS, Grenoble (France)


Diffraction imaging with coherent high-energy X-rays

The problem of phase determination is widely encountered in physics. In simple terms it can be said that the phase is a purely imaginary part of matter, and is, usually, undetectable in reality. In X-ray researches, knowledge of the phase of scattered radiation may tell scientists much about properties of a sample under investigation. For instance, if one knows the phase of the X-rays reflected from a crystal with lattice deformation, it is possible to retrieve the profile of this deformation uniquely from the measured intensity. In practice, a very sophisticated triple-crystal arrangement is necessary to perform such measurements. Then, to retrieve the deformation depth profile, a computer-aided trial-and-error method is usually used which does not always give a unique solution.

The new generation synchrotron radiation facilities gave rise to the advanced experimental techniques where the phase variation of scattered X-rays can be detected in a straightforward manner. These techniques are based on the phenomenon of interference, and are known overall as phase-contrast imaging. Progress has been made possible due to the high spatial coherence of the delivered radiation which can be directly associated with the small X-ray source size and the large distance to the experiment unit.

An elementary analysis and experimental results demonstrate an extreme sensitivity of the topography with coherent X-rays to crystal lattice distortions. Lattice deformation with a gradient vector perpendicular to the diffraction plane does not show up in the "near-field" image. However, it produces pronounced phase-contrast images at some distance from the sample. The sensitivity of the method can in some cases overcome ultra-high plane wave topography and can achieve d/d ~10-6 - 10-8. The combination of phase and diffraction imaging led to spectacular results in the studies with periodically-poled lithium niobate crystals [1]. These crystals exhibit large optical non-linearities and are widely used for electro-optic and acousto-optic applications as well as for optical frequency conversion. Because of complex interactions during domain-inversion processing, this is accompanied by lattice distortions across the domain walls. These distortions split the diffracted wavefront of a coherent X-ray beam, giving rise to a pattern of interference that reflects an underlying pattern of lattice distortions. The experiment was performed at the Optics Beamline (BM5) which is on a bending magnet. The sample was a LiNbO3 crystal with a thickness of 200 µm, and a domain-inverted structure of period 30 µm. Figure 131 shows schematically the experimental set-up, and a series of images at different sample-to-film distances, in the 006 symmetric Bragg geometry. The fringe patterns revealed represent the lattice deformations in the domain-inverted regions. Because of the high sensitivity to deformation, the method can be effectively used for the investigation of defects and microstructure in extensive bulk and thin-film materials such as are used in optoelectronics and photonics. Use of the phase information obtained from imaging is expected to advance the understanding of the various post-growth modifications that have been extensively used in the fabrication of novel devices.

By these means, previously very difficult laboratory experiments become trivial when using the third generation synchrotron source, and give extremely useful insights.

Z.H. Hu (a), P.A. Thomas (a), A. Snigirev (b), I. Snigireva (b), A. Souvorov (b), P.G.R. Smith (c), G.W. Ross (c) & S. Teat (d), Nature 392, 690-693 (1998).
(a) Department of Physics, University of Warwick (UK)
(b) ESRF
(c) Optoelectronics Research Centre, University of Southampton (UK)
(d) Daresbury Laboratory (UK)



New developments in X-ray optics

Besides the optics developments associated with beamlines reported earlier in this issue, highlights of recent progress achieved in the area of X-ray optics will be presented below:



Sagittal focusing by bent crystals

Sagittal focusing takes place in the plane perpendicular to the diffraction plane which permits to collect more flux on the sample without affecting the energy resolution. To achieve variable curvature, i.e. dynamical focusing, precise bending devices have to be designed and built. At the ESRF, a mature technique based on flexure hinges machined in a steel monolith [1] was modified and applied to the purpose of sagittal microfocusing. Figure 132 shows this device: a silicon crystal with a 6 mm wide groove is mounted on top of two supports that are part of the same monolithic structure. A stepping motor drives a micrometer screw that bends a leaf spring situated below the helical spring. This changes the distance between the ends of the two lever arms attached to these supports that are tilted symmetrically about axes along the two groove edges lying in the neutral fiber of the thin crystal part. In this way two equal moments of opposite direction are applied to the crystal and bend it to a circular cylinder. The radius of curvature can be varied from infinity (flat crystal) until 0.5 m.

Tests of this device equipped with a Si (111) crystal were carried out at 8 keV on the Optics Beamline BM5. To achieve a minimum focal line width, vertical focusing geometry was chosen. The bender was positioned at 40 m distance from the source and the crystal was bent to 0.58 m radius. The knife edge scan across the focus and its derivative (see Figure 133) show an experimental width of 4.4 µm for an incident beam height of 200 µm. This result represents a record for sagittal crystal focusing. The gain of 27 with respect to a slit of the same width was almost the theoretical one. By improving the crystal thickness uniformity, undulator beams of several mm width could be focused down to a few microns with gains exceeding two orders of magnitude. More details are given in [2]. It should be mentioned here that a different type of bending device was developed at the Swiss Light Source in collaboration with the ESRF [3]: a 6 x 1 mm2 beam (horizontal x vertical) was horizontally focused to a line of 20 µm x 1 mm with an efficiency of 89.6%, corresponding essentially to theory, and the gain exceeded 200.



Diamond crystals

Diamond crystals have the advantages that simple watercooling is sufficient even for very high heat flux and that the beam transmitted through the monochromator can be used on other stations downstream because the absorption is small. This serial beam multiplexing is used on the Troika (ID10) and Quadriga (ID14) beamlines of the ESRF. Of course, the diamonds must be of outstanding quality and very thin. The former condition requires a perfect starting material, the latter adequate preparation techniques. In the framework of an ongoing collaboration, Professor Sellschop (Johannesburg) provided us with several excellent 111-oriented plates that are between 67 µm and 175 µm thick and all larger than 30 mm2. X-ray topographs revealed a very low defect density in most of the platelets. Diffraction profiles recorded on the 67 µm thick and 10 x 5.2 mm2 wide crystal showed uniform reflection properties across the crystal with an excess width of less than 1 arcsec. The peak reflectivity varies between 60% and 70% and is distributed very regularly too (see Figure 134). Moreover, we studied the surface quality with optical interferometry. The surface microroughness of only 0.6 nm over 1.3 mm and the flatness of 180 nm over the whole area (rms values) are excellent figures. This outstanding crystal is the best we have ever seen.

For various applications it is important that the beam can be focused and stays perfectly parallel to the white incident beam when changing the energy. This is normally achieved by a double crystal fixed exit set-up where both crystals are identical. The first crystal being diamond, the second should then be a bent, sagittally focusing diamond crystal - a scheme that has not been realized until now. We tried to bend the 67 µm thick diamond crystal using the bender shown in Figure 132. The crystal was either directly attached to the bending device, or soldered into a frame carefully machined into a silicon crystal that was then mounted and curved down to 0.5 m. With the latter method we were able to focus a 500 µm wide beam down to about 15 µm. Details are given in [4].



Microfocusing with Kirkpatrick-Baez geometry

An alternative method to achieve microfocusing is to use curved mirrors. For strong demagnification spherical aberrations are too big and shaping of the mirrors into an elliptical figure with µrad precision is mandatory. We have shown that this can be done by using the bender described in [1] where two actuators permit to apply two different moments on both ends of the plate to be bent. Furthermore, a procedure has been developed to get the desired shape in a few minutes once the system is installed in the X-ray beam [5]. This procedure becomes invaluable if one wants to achieve double focusing by two crossed curved mirrors in Kirkpatrick-Baez geometry that is shown in Figure 135. Eight degrees of freedom (four moments, two tilts and two translations) must be optimized which is very cumbersome when performed manually.

The method consists in scanning a slit in front of one of the mirrors the other being translated out of the beam, and to record the center of the beam at the sample position with sub-µm precision using a CCD camera. Assuming a local linearity between actuator displacement and the residual slope error, the measured spot size is minimized by a matrix inversion. A maximum of three iteration results in a performance better than 0.5 µrad, comparable to the manufacturing slope error of the mirrors. The same procedure is repeated for the other mirror and then the beam is doubly focused to a couple of microns in both directions. The incident beam cross section was 800 x 800 µm and the gain with respect to a slit was five orders of magnitude. This is amongst the best performances ever reported.



Graded multilayer developments

Multilayers are X-ray optical elements that cover the wide gap of spectral resolution between mirrors providing a simple high energy cut-off and perfect single crystals offering medium to very high resolution. By optimizing layer pairs we are now capable of sputtering multilayers with reflectivities exceeding 80% over a wide range of energies. In contrast to crystals their lattice spacing and its variation, both normal and parallel to the reflecting surface, can be tailored in a wide range to accurately match experimental needs. This is of particular importance for focusing applications: the ideal geometrical figure for focusing is an ellipse and when using diffraction, the Bragg optics should have a gradient both along the reflecting lattice planes and normal to them. It is essential to have multilayer manufacturing techniques at hand that are capable to produce these gradients with a precision of a few 0.1%: this can be presently achieved with our machine.

A first example of a focusing application to high energy X-ray optics demonstrated the huge gain of flux that can be obtained with a broadband multilayer by focusing the beam from 800 µm to 4 µm [6]. As a second example, a W/B4C multilayer with 50 double-layers was manufactured whose thickness varies from 3.5 nm to 6.0 nm over a length of 24 cm [7]. At the thinnest position, the W layer was only 1.4 nm or ten atomic diameters thick! The measured average peak reflectivity was as high as 76% and varied between 55% at one end and 79% at the other which is only 10% less than the theoretical value, assuming the unavoidable substrate roughness. The 10 mm thick pyrex substrate was given a special lateral shape as shown in Figure 136 where the thickness variation of the bilayer (), and the B4C and W layers can be seen, too. When clamping the thick part and pushing on the tip of this substrate it bends to a parabola. The focusing efficiency of this device was checked on BM5 at 8 keV: a 2 mm wide beam was focused to 7 µm and the flux gain exceeded 200.



Time structure control

A rapidly increasing number of experiments study the time dependence of processes in physics, chemistry, biology and other areas such as materials science. Synchrotron storage rings generate a time structure of X-ray beams consisting of pulses with a spacing of a few microseconds and a duration between between 50 and 150 picoseconds, depending on the filling pattern of the machine. Obviously, the filling pattern is a matter of compromise between the needs of the individual beamlines and this leads to a reduced global efficiency of the synchrotron radiation facility. Therefore, it would be beneficial to equip each beamline with devices capable to tailor the time structure matching the experiment. On the other hand, new and very interesting research areas could be accessed by increasing the time resolution to 1 picosecond or below which cannot be reached by the machine. In a first attempt to investigate the possibilities of pulse compression with techniques analogous to those applied in the visible, a series of experiments was conducted [8] and a theoretical study performed [9] to search for a chirp of synchrotron radiation, i.e. an eventual time - energy - direction correlation of the photons in a beam. The X-ray pulse shape was recorded with fast detectors as a function of the angle above and below the orbit plane and of the X-ray energy from 8 keV to 30 keV. No significant effects were observed that could be directly used for pulse compression, but some more subtle trends appeared after signal processing by fast Fourier transforms that need further explanations.

To start developing X-ray optical elements for the control of the time structure a prototype of an optical chopper was built [10]. This device is based on a vibrating crystal permitting fast X-ray beam deflection that could be used for other applications, e.g. to feed pulses into delay lines to achieve pulse superposition. When combined with a second identical but stationary crystal, the doubly reflected beam can be periodically interrupted, because the two angular reflection curves are scanned one across the other. The opening time is a function of the scanning speed that is determined by the angular reflection width of the crystal ("Darwin width", ) and by both the amplitude and the frequency of the vibration. This can be seen in the insert on Figure 137 where the (angular) amplitude is plotted against time. The amplitude is easily changed by varying the voltage of the actuators. The vertical "set-point" of the Darwin width can be varied by changing the angle between the two crystals which permits to vary the time resolution Dt without changing the chopping frequency that should be an entire fraction of the synchrotron frequency.

The chopper function of the device was tested on beamline BM5 using the 333-reflections from two silicon crystals at 10 keV. The first crystal served as monochromator, the second was mounted with both ends on two piezo-electric motors working in push-pull mode to obtain a pure rotation of the crystal. A rocking curve of 0.9 arcsec width was measured proving that the mounting strain was very small which was a prerequisit for the technique. The results of the dynamic tests at a frequency of 18.5 kHz are shown in Figure 137 where time scans are displayed for different settings of the crystal angular position. As predicted, the width of these scans changed with the angle and the smallest width was about 200 ns. The scatter of the data is due to the limited performance of the electronics. These first results are very encouraging to pursue developing this technique and to apply it to time-resolved experiments and as one element in more complex schemes of X-ray optics for time structure control.

[1] L. Zhang (a), R. Hustache (a), O. Hignette (a), E. Ziegler (a) and A.K. Freund (a), J. Synchrotron Rad. 5, 804-807 (1998).
[2] A.K. Freund (a), F. Comin (a), J.-L. Hazemann (b), R. Hustache (a), B. Jenninger (a), K. Lieb (a) and M. Pierre (a), Proceedings SPIE 3448, 144-155 (1998).
[3] C. Schulze (c), G. Heidenreich (c), H. Auderset (c), D. Vermeulen (c) and A.K. Freund (a), Proceedings SPIE 3448, 156-165 (1998).
[4] A.K. Freund (a), J.P.F. Sellschop (d), K. Lieb (a), S. Rony (a), C. Schulze (c), L. Schroeder (a) and J. Teyssier (a), Proceedings SPIE 3448, 53-63 (1998).
[5] O. Hignette (a), A.K. Freund (a) and E. Chinchio (a), Proceedings SPIE 3152, 188-199 (1997).
[6] U. Lienert (a), C. Schulze (c), V. Honkimäki (a), Th. Tschentscher (a), S. Garbe (e), O. Hignette (a), A. Horsewell (e), M. Lingham (a), H.F. Poulsen (e), N.B. Thomsen (e) and E. Ziegler (a), J. Synchrotron Rad. 5, 226-231 (1998).
[7] Ch. Morawe (a), P. Pecci (a), J.Ch. Peffen (a) and E. Ziegler (a), to be published.
[8] J.F. Eloy (a,f), B. Brullot (g), J. Doublier (g), A.K. Freund (a), R. Marmoret (g), O. Mathon (a), R. Tucoulou (a), B. Villette (g) and R. Wrobel (g), Proceedings SPIE 3451, 96-107 (1998).
[9] J.F. Eloy (a,f) and A.K. Freund (a), Proceedings SPIE 3451, 82-95 (1998).
[10] A.K. Freund (a), O. Mathon (a), R. Tucoulou (a) and R. Le Letty (h). Proceedings SPIE 3451, 130-138 (1998).

(a) ESRF
(b) CNRS, Laboratoire de Cristallographie, Grenoble (France)
(c) Swiss Light source, Villingen (Switzerland)
(d) University of the Witwatersrand, Johannesburg (South Africa)
(e) Riso National Laboratory (Denmark)
(f) CEA/DAM, Le Barp (France)
(g) CE Bruyères le Chatel (France)
(h) CEDRAT Recherche, Meylan (France)