With the advent of the ESRF, the quality of the X-ray beam has dramatically increased in the last few years (brilliance for example has roughly gained four orders of magnitude).

Such progress has seen the appearance of new experimental techniques taking advantage of the remarkable properties of synchrotron light, be it coherence, time structure, polarisation or tuneable energy. Some examples of new techniques developed at the ESRF are given below.

At the same time, power densities on windows, absorbers, optical elements and ultimately on the sample have also increased by a large amount, pushing for better adapted optics and sample environment to be worked out. On the other hand, detectors have to accommodate higher count rates and the data stream needs to be handled by an appropriate computing and data storage system.

Improvement of instruments on the beamlines is a continuous process, thanks to all ESRF support groups and beamline scientists whose objective is to offer ESRF users the best experimental conditions. We can show here only a few examples of this intense activity, in two important topics: optics and detectors.


Protein crystallography at ultra-short wavelengths: using anomalous dispersion

The potential benefits of using short wavelength X-rays for protein crystallography have been suggested by several scientists [Arndt, 1984, Helliwell and Fourme, 1983]. These benefits are smaller systematic errors due to absorption in glass capillaries, the mother liquor and the proteins themselves suggesting more accurate intensity data and possibly lower radiation damage. Traditionally, protein crystallography at synchrotron radiation sources have been performed at wavelengths around 0.9-1.5 Å matching the maximum flux and detection methods at the previous generation of synchrotrons. In general the shorter wavelengths have proven to be advantageous. Experiments at even shorter wavelengths have been prevented by the lack of photon flux. The intensity from diffraction varies approximately as 2 (after correction for geometrical factors) whereas the photoelectric absorption varies as 3 away from the absorption edges. Thus a variation of 1/ in the diffracted photon over absorbed photons ratio may be expected and, as a consequence, absorption and radiation damage should be reduced at shorter wavelengths. The Materials Science beamline ID11 produces copious amounts of photons in the high-energy region, and the present study was done to test the above hypotheses as well as to test the feasibility of solving the macromolecular phase problem in crystallography by using anomalous diffraction techniques around the xenon K absorption edge (0.358 Å). In several cases, pressurising macromolecules by noble gas molecules produces specific binding sites for the gas molecules in hydrophobic pockets of the macromolecules without producing structural changes. In the present case, two crystals of porcine pancreatic elastase, a 26 kD protein, were used for two data collections at an energy 50 eV above the xenon absorption edge using the ESRF imaging intensifier CCD system [Moy, 1994] at room temperature (~ 20 °C). In one case, the protein in the capillary was pressurised by 16 bars of Xe gas. For the native data set, 186157 reflections out to a resolution of 1.4 Å were collected; the xenon pressurised crystal yielded 84666 reflections to a resolution of 1.8 Å. The R-merge factors down to a resolution of 2.5 Å were 0.028 and 0.039 respectively. Although the expected anomalous variation in the Bijvoet differences on intensities for the two data collections is only 3.35% assuming one single fully occupied xenon site, isomorphous and anomalous difference Patterson maps readily yielded a single xenon site and phasing of the structure factors could easily be performed. An example of the excellent electron density maps produced is given in Figure 131. In the case of the native data collection, the resolution dependent scale factor varied from 0.0 (first frame) to 0.45 Å2 (last frame) indicative of minor radiation damage. The present study demonstrates the possibility of using ultra-short wavelengths around the Xenon absorption edge for scattering experiments and furthermore shows that excellent very high-resolution data may be obtained even at these very short wavelengths. The minimisation of the radiation damage and the isomorphous structure of the native and Xe pressurised derivative indicate that this method of solving macromolecular structures will be potentially very powerful for future structure determinations.


M. Schiltz (a), Å. Kvick (b), O. Svensson (b), W. Shepard (a), E. de la Fortelle (d), T. Prange (a), R. Kahn (c), G. Bricogne (a,d) and R. Fourme (a), J. Syn. Rad. 4, 287-297 (1997).

(a) LURE, University of Paris-Sud, Orsay (France)
(b) ESRF
(c) IBS, Grenoble (France)
(d) MRC Laboratory of Molecular Biology, Cambridge (UK)




Atomic scattering factors measured by high-energy X-ray interferometry

The aim of the experiment was the measurement of relativistic corrections to the atomic scattering factors. In a simple picture, the scattering power of the electrons bound in an atom is reduced by the relativistic mass increase due to their kinetic energy [Smith]. Recent calculations propose that this reduction is independent of the energy of the scattered X-rays but scales with the total binding energy of the atom and therefore roughly increases with Z2, Z being the number of electrons in the atom [Kissel et al.]. The relativistic reduction is predicted to be about 1% for Z = 70. It becomes more important at lower energies and large scattering angles when only a few of the Z electrons contribute. So far, no experimental data exist to which the various calculations could be compared.

Although the relativistic reduction is most important at low energies, it should be measured at energies which are large compared to the K-shell binding energy in order to separate the relativistic contribution from dispersion corrections. X-ray interferometry at high energies provides a direct measurement and the required accuracy.

X-ray interferometry measures the phase shift in a sample due to the interference of the forward scattered amplitude with the incident beam. This phase shift is proportional to the atomic scattering factor. In the triple crystal Laue interferometer two separated, coherent beams are produced by diffraction on a first Si wafer and recombined after a second reflection (Figure 132). The resulting interference pattern is analysed by a third wafer. The phase shift due to the sample is directly related to a shift of the interference pattern.

The required precision of the phase shift determination is achieved by the amazing capability of monolithic interferometers to displace the analyser wafer with an accuracy of 0.01 Å. These displacements are achieved by a weak link and a force produced by a solenoid acting on a permanent magnet glued to the interferometer. The experiment was performed at the High Energy beamline ID15 utilising very high photon energies provided by the Superconducting Wavelength Shifter (SCWS) [1]. The interferometer was tuned with its first harmonic (220) to 135 keV and the interference pattern could be obtained up to the 4th harmonics at 540 keV. The distance between the interferometer wafers and the small scattering angles determine the separation of the beams and demand a very precise positioning of the samples, but samples up to 20 mm length can be measured. Investigated materials were Si, CaF2, Ge, Ag and Sn. When the analyser wafer scans the interference pattern, all four harmonics are recorded simultaneously allowing for a consistency check of the evaluated relativistic reductions. Figure 133 shows a typical result for an Ag sample. Without sample the zero phases for all harmonics coincide. With sample, the fundamental is almost absorbed but higher harmonics can be used to evaluate the phase shift. In order to determine the relativistic reduction, the energy, sample thickness and density must be known to a relative accuracy of 10-4.

In a preceding study, the relativistic reduction was measured using characteristic WK radiation from an X-ray tube (60 keV) [Lienert]. The results show that the commonly used dipole approximation indeed overestimates the relativistic reduction by a factor of approximately two, as predicted by recent calculations. Ge was the heaviest element that could be measured. The dispersion contribution, which must be taken from theory, is almost twice as large as the relativistic reduction. This situation was dramatically improved by the use of high-energy synchrotron radiation. The relativistic reduction for Ag is about 2.5 times larger than for Ge and the dispersion contribution at higher energies amounts to only 10 to 20% of the relativistic reduction. The results confirm the importance of higher order terms in the calculation of the relativistc reduction.


[1] U. Lienert (a, b), M. Hart (c), D. Laundy (d), K.-D. Liss (a), to be published.

(a) ESRF
(b) Risø National Laboratory (Denmark)
(c) Brookhaven National Laboratory (USA)
(d) University of Warwick (UK)




Quasi-elastic X-ray scattering by time domain interferometry

Scattering experiments resolving small (< meV) energy transfers at modest momentum transfers (~ Å-1) are of interest for probing slow motions on atomic length scales, including, for example, structural relaxations in glasses. Here we describe a new technique to do this using synchrotron radiation and nuclear resonant scattering. This method is a type of interferometry. However, where most interferometry measures spatial interference patterns to learn about static structure, here a temporal interference pattern is measured to learn about sample dynamics.

Nuclear scattering from two foils containing 57Fe, placed upstream and downstream of the (non-resonant) sample (see Figure 134) provides the object and reference waves, respectively. A pulse of synchrotron radiation excites the first foil, scatters from the sample and then excites the second foil. Due to the long, 140 ns, lifetime of the nuclear resonance, the foils continue to radiate after the exciting pulse has passed. However, the wave reaching the detector from the second foil reflects the state of the sample when the exciting pulse was scattered, while the wave from the first foil scatters from the sample at later times. The two waves interfere in the detector and the intensity measured as a function of time provides information about sample dynamics. The measured intensity may be related to the intermediate scattering function, S(q,t), of the sample.

The method was tested using a well-known glass former, glycerol. Figure 135 shows the temporal interference pattern measured at different temperatures in the neighbourhood of the first structure factor maximum at 1.5 Å-1. Constant velocity motion of the first foil was used to introduce a Doppler shift between the nuclear response frequencies of the two foils. This leads to a 13 ns period quantum beat pattern (Figure 135a, b) in the time response if the sample is static (pure elastic scattering). At higher temperatures, however, structural relaxations in the sample lead to a progressive dephasing of the object wave from the reference wave, reducing the contrast in the quantum beats at later times, relative to that at earlier times. This is particularly clear in Figure 135d. One also notes that inelastic scattering from fast motions of the sample causes an instantaneous (on the ns time scale) dephasing of the two waves and leads to a time-independent damping of the quantum beats. Solid lines in the figures are fits, giving results that agree with previous measurements, both in the magnitude and the time scales of the structural relaxation.

The impact of this new technique remains to be seen, but one can compare it to older methods. In particular, while there is some overlap with neutron spin-echo studies, this method more easily reaches, simultaneously, smaller energy transfers and large momentum transfers. This technique is somewhat similar to Rayleigh Scattering of Mössbauer Radiation (RSMR) measurements. However, with a synchrotron radiation source (instead of the radioactive sources of RSMR) smaller samples may be studied with better angular resolution. In addition, this technique seems to be sensitive to longer time scales (smaller energy transfers) than RSMR.


A.Q.R. Baron (a), H. Franz (b), A. Meyer (a, b), R. Rüffer (a), A.I. Chumakov (a), E. Burkel (c) and W. Petry (b), Phys. Rev. Lett., 79, 2823 (1997)

(a) ESRF
(b) Physik-Department, Technische Universität München, Garching (Germany)
(c) FB Physik, Universität Rostock (Germany)




Fresnel diffraction by slits

The ESRF X-ray source has been significantly improved over the last year, achieving a size of about 25 - 30 microns in the vertical direction and getting closer to the diffraction limit. This has led to further progress in coherent imaging and diffraction techniques. A quantitative study and development of the simple and reliable techniques related to spatial coherence is therefore of great interest. The most important point is that the technique should not influence the beam coherence. A holographic technique using a calibrated object such as a round boron fibre has been proposed and successfully used and reported recently. Nevertheless this technique requires an additional fibre installation and detector system with a resolution of about 1 micron.

It is well-known in classic optics that coherent illumination of the slit generates an interference pattern called "Fresnel fringes". The contrast (visibility) of the fringes strongly depends on the source size, i.e. on the spatial coherence of the beam. It turned out that Fresnel fringes can be easily observed for a slit size of about 50 ­ 200 microns at energies of 6 ­ 60 keV. This interference pattern can be used to probe the beam coherence by means of Fresnel diffraction. Comparing the experimentally measured visibility V of the central fringe with the visibility of the point (fully coherent) source V0, one can easily calculate the source size s (fvhm).

On the ID22 beamline, the Fresnel diffraction pattern was studied for different primary slits opening (Figure 136a). The images were recorded on a high-resolution X-ray camera. The source-to-slit distance was z0 = 31 m and the slit-to-detector distance was z1 = 10 m. Figure 136b shows the relative intensity distribution of the Fresnel fringe pattern for a slit size s = 100 µm: the red curve corresponds to the measured intensity distribution and the black curve was calculated for the point source. Thus the source size is about 30 µm, which is in good agreement with the source characteristics.

As far as slits are standard beamline components, the proposed technique can be applied at any ESRF beamline. It should be noted that this technique does not require a high-resolution detector and can be performed by the scanning of a 3-5 micrometer pinhole (or a secondary slit) coupled with an X-ray counter.


A. Snigirev (a), V. Kohn (b), C. Raven (a), I. Snigireva (a), to be published.

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




Silicon germanium crystals for fixed-exit double-crystal monochromators

One of the most important devices at synchrotron beamlines is the monochromator. Here double-crystal monochromators are widely used because they can provide a reflected beam parallel to the incoming beam over the whole energy range. The mechanical design of such a fixed-exit monochromator foresees a first crystal with one freedom of movement, which is a rotation, to satisfy the Bragg law for a chosen energy. The movement of the second crystal has to have three degrees of freedom. First it has to be kept parallel to the first crystal to meet the Bragg law as well. Then it has to be translated horizontally and vertically to fit the condition of a fixed exit beam. These three movements have to be carried out precisely, so that the reflection will not be lost. Therefore mechanical stability and reproducibility of the second crystal stage movement is a crucial point. With the use of good monochromator mechanics, a parallelism better then 0.l arcsec can be achieved.

At undulator beamlines, the first crystal, due to the high heat load of the incoming beam, has to be cooled cryogenically. Because then both crystals have different temperature, their lattice parameters are slightly different, namely the first crystal has a smaller lattice parameter. Cooling a silicon crystal from 300 K down to 77 K causes a relative expansion of the lattice of - 2.4 x 10-4. This difference leads to a slightly dispersive set-up and therefore induces a systematic angular deviation of the exit beam, which varies with the energy. For a Si (111) reflection, the angular deviation for 30 keV and 5 keV is 7 arcsec and 43 arcsec, respectively. For a Si (311) reflection, it is 13 arcsec and 115 arcsec. In comparison with the mechanical precision of a monochromator and the Darwin width of a reflection, these are large values.

The corresponding change in the Bragg angle of the second crystal as well as the spatial displacement of the exit beam at a certain distance can be compensated mechanically, but the angular deviation remains.

To overcome this problem, an optical solution can be applied. A silicon crystal with a small amount of germanium has a slightly bigger lattice parameter than a pure silicon crystal, and this difference could compensate the negative expansion from cooling.

To verify this solution an experiment was carried out at the Optics Beamline (BM5).

The lattice parameters of a single Si-crystal and a single SiGe-crystal with 0.7 atomic percent Ge were measured in the temperature range between 77 K and 300 K. The results are shown in Figure 137, demonstrating that the lattice parameter of a Si99.3Ge0.7-crystal at 77 K, nearly matches the lattice parameter of a Si-crystal at 300 K.

With these values for the d-spacing of both crystals at 77 K, one can interpolate that the necessary concentration of germanium to match exactly the right value has to be 0.6 atomic percent.

The application of such SiGe crystals in a double-crystal monochromator is a very promising technique to eliminate the energy dispersion of the reflected beam, caused by the temperature difference between the crystals.


[1] A. Souvorov (a) and A. Snigirev (a), Rev. Sci. Instrum. 68 (1997).
[2] A. Souvorov (a), M. Drakopoulos (a), A. Freund (a), I. Snigireva (a), A. Snigirev (a), A. Erko (b), W. Gudat (b), N. Abrosimov (c), S. Rassolenko (c), W. Schöder (c), to be published.

(a) ESRF
(b) BESSY, Berlin (Germany)
(c) Institute of Crystal Growth, Berlin (Germany)




A multi-purpose, micro-focusing monochromator for high energies

It might be surprising that the unique opportunities which high energy X-rays add to the classic X-ray scattering techniques often arise from their weak interaction with matter. The exploitation of this new field has therefore often been hampered by too low intensities. For the first time, third generation high-energy synchrotrons provide brilliant high-energy sources, but the existing low-energy optics failed to transmit the brilliance onto the sample. Up to now, such important optical tools like a fixed-exit, focusing, energy-tuneable monochromator or efficient micro-focusing optics were not available. Many of the apparent difficulties in the handling of high-energy X-rays may turn into valuable advantages if adapted optical concepts are used. Exploitation of meridional focusing and Laue geometry for instance allows the construction of a focusing, fixed-exit, energy-tuneable monochromator for an energy range of 40-120 keV as shown in Figure 138 [Suortti and Schulze]. The crucial problem of sagittal focusing, anticlastic bending, can simply be neglected. Furthermore, the horizontal and vertical scattering planes are clearly separated, allowing for vertical focusing by a bent multilayer. A WB4C multilayer was deposited in the ESRF multilayer facility and provided a peak reflectivity of 80% as expected from low absorption. At 68.5 keV, 1010 ph/s were focused at BM5 into a line focus of 4 x 400 µm2 (v x h). The vertical focus size was caused by the deliberately large focal length of the multilayer, to provide space for bulky sample environment, and could be reduced by moving the multilayer closer to the focus. The focused beam was utilised to measure strain gradients in structural CuNi multilayers. High energy X-rays are required because of their penetration power, as it is known that sample thinning changes the residual strains. Figure 139 shows that Cu and Ni layers are clearly separated. Analysis of the peak positions indeed reveals the existence of steep strain gradients at the buried interfaces - underlining the important role that focused high-energy X-rays may play in the characterisation of polycrystalline materials filling the gap between neutron diffraction and electron microscopy. New spectroscopic applications open up if the periodic multilayer can be replaced by a wide band pass super-mirror.


U. Lienert (a), V. Honkimäki (a), M. Lingham (a), Ch. Morawe (a), E. Ziegler (a), S. Garbe (b), N. B. Thomsen (c), H. F. Poulsen (b), A. Freund (a), to be published in SRI 97 proceedings

(a) ESRF
(b) Risø National Laboratory, Materials Department, Roskilde (Denmark)
(c) Danfoss A/S, Materials Technology, Nordborg (Denmark)




A new image plate detector system for diffraction experiments at high pressure

High-pressure diffraction experiments require a large-area, on-line detector, with good sensitivity to high-energy radiation. Furthermore, a fast cycle time (expose-read-erase) is a considerable advantage as it could open the field of transition-kinetics studies, and allow measurements on single crystals with monochromatic radiation.

Such a detector, based on the image plate technology, has been developed for the High Pressure beamline, ID30, as none of the devices currently available commercially satisfy these conditions.

The image read-out time is 12 seconds in the fastest mode, which is extremely valuable not only for kinetics and single-crystal work, but generally for a more efficient use of the measuring time, as the high brilliance of the ESRF beams as well as optics improvements are rapidly reducing the exposure time in many experiments. In order to maximise the sensitivity, a laser beam of 150 mW is used to scan the image plate, and hence extract more information for a given X-ray dose. This gives an improvement of about 10 over the commercial system used previously, and hence allows much shorter exposures or higher image statistics (see Figure 140). The X-ray sensitive area is 250 mm by 305 mm, hence larger than that of CCD-based systems with equivalent resolution. This is very important to improve the signal-to-background ratio: the background signal decreases as the inverse of the square of the distance, but the diffraction signal decreases with the inverse of the distance, thus a large sample-to-detector distance is an important advantage. In order to obtain a high spatial resolution, the scanning laser beam is focused by a high performance lens to a focal spot of about 20 µm fwhm on the image plate. The final resolution of the image is mainly due to the broadening of the laser profile by the light scattering in the phosphor layer of the image plate, and as this is proportional to the thickness of the phosphor layer, we can have the additional advantage of resolution tunability: different phosphor thicknesses will give different resolutions. For the tests presented below, the plates used gave a resolution of about 0.2 mm fwhm; these are a good compromise between efficient X-ray absorption (55% at 50 keV) and high resolution. Another feature of the device is its variable scanning pixel size: 40, 60 or 80 microns, which can be tuned for image point-density and intrinsic plate resolution. In powder diffraction experiments, the rings obtained are integrated to give one-dimensional spectra, and hence the spatial distortion of the images caused by the scanning process should be as small as possible. After calibration, the latter is of no more than 16 microns, and hence makes a negligible contribution to the final integrated diffraction profile. Finally, to cope with a large range of X-ray doses, the sensitivity of the detector can be varied over more than five orders of magnitude, which is very useful to adapt its sensitivity to the experimental conditions, and at a specific sensitivity setting. The image data is digitised with a resolution of 14-bits, which is sufficient for diffraction at high pressure.

There are many areas of high pressure research in which this detector is extremely useful, amongst which we find studies of materials in extreme conditions of pressure and temperature where the experimental stability requirements are hard to meet, studies of reaction kinetics and single crystal diffraction, when high-speed data collection is needed, as well as large-volume cell work when separate images of the sample and its surroundings have to be collected in close succession (see Figure 141).


M. Thoms (a), D. Häusermann (b) , S. Bauchau (b), M. Kunz (b), T. Le Bihan (b), M. Mezouar (b) and D. Strawbridge (b), to be published.

(a) Institute of Material Science VI, University of Erlangen- Nürnberg, (Germany)
(b) ESRF




Luminescent screen - CCD camera with 1.6 µm spatial resolution

Micro-imaging techniques with synchrotron radiation demand fast, on-line X-ray detectors with a spatial resolution in the micron- or submicron range. We developed a camera where detection of X-ray images is accomplished with transparent, i.e. non-scattering luminescent screens. High image quality is achieved if depth of focus of the optical system and the X-ray absorption length in the scintillator are matched. The detector developed for the ESRF beamline ID22 consists of a 5 µm thick YAG: Ce screen, microscope optics and a low noise CCD camera, operated at X-ray energies between 10 - 50 keV. A spatial resolution of 1.6 µm fwhm was measured by recording the hologram of a Fresnel zone plate. Applications are in topography, phase-contrast imaging and microtomography (see page 11, "X-ray tomography with micrometer spatial resolution").

Detection systems employing transparent luminescent screens permit thicker screens, i.e. higher absorption of X-rays, compared to powder phosphor screens for the same spatial resolution. The reason is that the lens of the optical system "looks" into the transparent luminescent screen. The resolution, therefore, is not limited by light scattering within a granular powder phosphor layer. Instead, depth of focus, and to a lesser extent spherical aberration of the optics, determine the resolution. Parallax between the incident X-ray beam and the optical axis may reduce resolution further. The difficulties of producing small grain-sized phosphors are replaced by the need to obtain thin homogeneous layers of high-Z scintillators with optical quality surfaces.

Figure 142 shows the set-up. A conventional microscope objective together with a cooled CCD camera is used, which provides a compact and light (~ 2 kg) camera. The detector components are listed below. As an option, an in-house developed fast and low noise CCD camera (Frelon camera, 14 bit dynamic range, 10242 pixels, 20 images/s) may be employed, which is especially advantageous for tomography where Gbytes of image data have to be acquired. The detective quantum efficiency (DQE) of the system is mainly determined by X-ray absorption. A higher Z, more dense scintillator than YAG could improve the performance of the camera considerably, and work is progressing in this direction.

The present system is close to its theoretical performance, in terms of spatial resolution. A possible improvement in resolution by a factor of four is predicted in theory if screens of 1 µm thickness are used. Using scintillators which emit light at a shorter wavelength will further enhance the resolution.



Camera components and performance

X-ray entrance window
Light tight aluminised polyester foil
Foil thickness: 24 µm

5 µm thick YAG: Ce (Y3Al5O12: Ce)
on 1mm thick substrate of undoped YAG

Microscope objective, magnification *40, NA = 0.5

CCD camera
14 bit dynamic range, Peltier/water cooled, 10242 pixels,
(24 µm)2 pixel size, read-out time 5 s

Sensitive area
~ 1 mm2

Spatial resolution
1.6 µm fwhm of PSF, directly measured with a Fresnel zone plate.

Detective Quantum Efficiency (DQE)
~ 2%, calculated (10 - 30 keV), limited by absorption of X-rays in the scintillator.



A. Koch (a), C. Raven (a).

(a) ESRF