Compound refractive lens for high-energy X-rays


From the very beginning after Röntgen's discovery, scientists were very tempted to adapt to X-rays methods developed in visible light optics, particularly focusing X-rays by means of refractive lenses. This attempt has been unsuccessful until now because the refractive index decrement for hard X-rays is very small ( ~ 10-5 - 10-6). The manufacture of simple refractive lenses with reasonable focal distance is therefore considered as very difficult or even impossible. It is easy to show that a collecting lens for X-rays has a concave shape and can produce a linear (point) focus at a distance f1 = r/2, where r is the lens radius. For example, = 2.8 x 10-6 for Al at E = 14 keV. This means that the simple concave lens of 300 µm radius will focus the parallel X-ray beam at a distance of 54 m. Despite the small aperture of this lens, the focal distance is relatively large. To shorten the focal distance, it was proposed that the number of holes in the path of the X-ray beam be multiplied, in other words to make a compound refractive lens (CRL). A compound lens with n holes has a focal length fn = r/2n. For the conditions mentioned above, a CRL with n = 30 brings the focal length into a range acceptable for many microfocus experiments (F30 = 1.8 m) (Figure 93). 

Theoretical considerations reveal compelling evidence that compound refractives lenses made from low-Z materials, such as boron, carbon, aluminium and water are very promising for high-resolution X-ray diagnostics in the energy range from 5 to 40 keV. The number of holes is limited by the reasonable length of the array and the possibility to make the spacing between the holes as small as possible. Acceptable hole radii are in the range from 250 to 600 µm.

This corresponds well to the beam size of third generation synchrotron radiation sources such as the ESRF. A first compound lens was implemented by drilling a linear array of 30 closely spaced holes of 0.3 mm radius in an aluminium block (Figure 94). At 14 keV, an X-ray beam was focused to a spot size of 8 microns at a distance of 1.8 m (Figure 95). Two-dimensional focusing by means of CRL can be achieved in two ways. The simplest way is to use two arrays of cylindrical holes in crossed geometry. The second proposition of point focusing uses hollow plastic spheres with a thin wall aligned in tube and space between the spheres is filled with a liquid.



A. Snigirev (a), V. Kohn (b), I. Snigireva (a) and B. Lengeler (c), accepted by Nature (7 November 1996)

(a) ESRF

(b) Kurchatov Institute (Russia)

(c) 2. Physikalisches Institut, RWTH Aachen (Germany)



New generation of Bragg-Fresnel optics


Further progress has been made over the past year in Bragg-Fresnel optics, with many successful applications. A Bragg-Fresnel lens uses phase delay formation on a surface relief to provide focusing of well-monochromatised X-rays in the high -energy domain.

Bragg-Fresnel optics allow the range of the spatial resolution of X-ray techniques to the sub-micrometer level over a wide range of X-ray energies (6-60 keV). At present two types of Bragg-Fresnel lenses are mainly in use: linear ones on a flat crystal substrate in sagittal geometry and circular ones under backscattering geometry. Bragg-Fresnel optics match perfectly with third generation synchrotron radiation sources such as the ESRF. Although it has the best characteristics experimentally observed in the high-energy region of X-rays, some parameters can still be improved, namely the resolution and the total efficiency (light gathering). Therefore new types have been designed and tested in collaboration with the Institute of Microelectronics Technology in Chernogolovka (Russia).

Smaller foci and higher fluxes were obtained with a bent linear Bragg-Fresnel focusing in the meridional and sagittal directions. The results obtained at the bending magnet beamline BM5 and the forecast to an insertion device beamline are summarised in Figure 96.



[1] Ya. Hartman (a), A. K. Freund (b), I. Snigireva (b), A. Souvorov (b), A. Snigirev (b), submitted to Nuclear Instr. & Methods

[2] A. Snigirev (b), I. Snigireva (b), A. Souvorov (b), V. Kohn (c), V. Yunkin (a), to be published

(a) Institute of Microelectronics Technology, Chernogolovka (Russia)

(b) ESRF

(c) Kurchatov Institute, Moscow (Russia)



Enhancement of the Bragg-Fresnel lens light gathering power by ultrasonic modulation


The crystal substrate on which the Bragg-Fresnel lens is based produces a monochromatisation as high as 10-5 -10-7. The Fresnel zone plate itself requires a bandwidth better than / 1/N, where N is the number of zones. In the case of Bragg-Fresnel optics, N = 102 - 103, which leads to a bandwidth of / = 10-2 - 10-3. Thus only a small fraction of the total intensity of a white source is gathered into the focus. To increase the absolute intensity in the focal spot it is necessary to increase the bandwidth of reflection, i.e. to modify the crystal lattice. One way to do this is to introduce stress into the crystal lattice by means of ultrasonic modulation.

The influence of an ultrasonic modulation of the Bragg-Fresnel lens as a tunable instrument to vary the radiation flux at the focal spot was studied (Figure 97). The reflectivity and integral intensity of the crystal substrate depend on the frequency and amplitude of the excited ultrasonic wave. The intensity in the focal plane of the Bragg-Fresnel lens was measured by means of 2D mapping with a 10 mm pinhole paired with a scintillation detector or a Si PIN diode (Figure 98). The dependence of the focal intensity on the amplitude and frequency of ultrasonic wave was registered under the condition that s >> , where s is the wavelength of the ultrasonic wave, while A is the extinction length of the crystal. A gain by a factor of up to 3 in the focal flux was obtained.



[1] A. Souvorov (a), A. Snigirev (a), I. Snigireva (a), E. Aristova (b), Rev. Sci. Instrum., 67(5), 1733-1736, 1996

[2] A. Souvorov (a), I. Snigireva (a), A. Snigirev (a), E. Aristova (b), Ya. Hartman (b), submitted to Journal of Applied Physics Letters

(a) ESRF

(b) Inst. of Microelectronics Technology, Chernogolovka (Russia)



Composite Bragg-Fresnel lens for high resolution diffraction techniques


The main requirements on advanced X-ray optics are high spatial resolution and high photon flux. To increase the absolute intensity in the focal spot of a Bragg-Fresnel lens it is necessary to increase the aperture or to change the crystal reflectivity, i.e. to modify the crystal lattice. At present the aperture is limited due to lithographic techniques by the width of the outmost zone of 0.15 - 0.3 µm.

A new type of linear Bragg-Fresnel lens allows spatial resolution to be improved to far beyond 1 mm in the hard X-ray region and to increase the intensity in the focal spot. In addition to the Bragg-Fresnel structure on a Si (111) substrate producing a diffraction pattern in first order, the structure was continued beyond the outmost zones with a coarser structure focusing in third order to the same spot. This structure in principle allows the resolution to be improved and moreover the flux to be increased. Such composed zone plates can be further enlarged by adding structures diffracting in fifth and higher orders (Figure 99 and 100).




I. Snigireva (a), A. Souvorov (a), F. Legrand (a), C. Raven (a), A. Snigirev (a), V. Yunking (b), to be published

(a) ESRF

(b) Inst. of Microelectronics Technology, Chernogolovka (Russia)




Moiré diffraction topography as a tool to investigate thin semiconductor layers


X-ray diffraction topography has proved to be an efficient technique to image defects and deformations in rather perfect crystals. In the past it was mainly applied to characterise bulk crystals. With the growing importance of thin layers in science and technology, a large effort is directed to apply these techniques also in thin layer analysis, because they are complementary to the other structural methods (X-ray diffractometry and reflectometry, electron microscopy, chemical etching methods, ...).

Moiré diffraction topography is one of the most sensitive methods when trying to measure very weak strains (inclinations and relative changes of distances of reflecting lattice planes) in the range of 10-6 to 10-9. On ID19 this technique has been applied to thin layers.

The basic condition for the Moiré pattern to occur is that the investigated sample consists of two crystalline parts separated by a gap, and with crystalline lattices very close to each other in mutual orientation and lattice parameters. In that case wave fields created in the first crystal part interfere with those created in the second one, giving rise to interference patterns - the so-called Moiré fringes - whose form and contrast depend very sensitively on the strains and defects present in the layer (Figure 101).

The objects of interest were perfect silicon substrates with a thin layer (thickness from 0.2 µm to 15 µm) on it, which contained lattice defects like threading dislocations. These two parts were separated by a thin silicon oxide layer (thickness about 0.4 µm) with plane interfaces.

The possibilities of the ESRF synchrotron radiation source allows one to collect, from one sample, in a short time, several tens of white beam topographs for different reflections and wavelengths, and with a resolution superior to those taken at other sources. On this basis it was possible to determine the dilatation and the shear components of the relative strain tensor. They were found to be in the order of 10-7, depending strongly on the chosen technological process of the sample production. It appeared that the upper layer symmetry had reduced to triclinic. This result could not be obtained by any other diffraction method.

Another important result of Moiré diffraction topography, not obtained before, is the detection and characterisation of threading dislocations within the thin layers for low dislocation densities (Figure 102).

They ranged between 101 (single dislocations) and 105 cm-2. The interaction of Moiré fringes and defects is the crucial point: it allows to image and characterise these defects, practically impossible to detect with electron microscopy. Two interesting results immediately visible from the topographs are that even a rather high dislocation density does not disturb the interference effect (Figure 99), and that the threading dislocations appear (in our investigated samples) exclusively in pairs with anti-parallel Burgers vectors.

This last result may be inferred from the fact that no global extra Moiré fringes due to the dislocations appear in the topographs (Figure 101 and Figure 103).



[1] M. Ohler (a,b), E. Prieur (a), J. Härtwig (a), J. Appl. Cryst, in print

[2] E. Prieur (a), C. Guilhalmenc (c), J. Härtwig (a), M. Ohler (a,b). A. Garcia (c), B. Aspar (c), J. Appl. Phys. 80 (1996).

[3] E. Prieur (a), M. Ohler (a,b), J. Härtwig (a), submitted to Phys. Stat. Sol.

(a) ESRF

(b) Max Planck Arbeitsgruppe «Röntgenbeugung an Schichtsystemen», Berlin (Germany)

(c) LETI, CEA, Grenoble (France)



Better resolution of ultra small-angle scattering


The relation between the scattering angle 2, the wavelength and a typical correlation length d is given by Bragg's law 1/d = 2sin/. For l = 0.1 nm and 2 = 15 µrad, a correlation length as large as 7 µm can be measured, which overlaps conveniently the regime of visible light scattering and ordinary microscopy. Such a resolution was obtained with crystal collimators in a modified Bonse-Hart camera arrangement. Due to the good collimation of synchrotron radiation from the undulator at ID2, a photon flux at the sample position as high as 5 x 1010 ph/s was obtained. The background next to the direct beam is so well rejected that at 2 = 1 mrad its ratio to the direct beam is 10-8. The instrument is described in Figure 104 and an example of a measurement is shown in Figure 105.



[1] O. Diat (a), P. Bösecke (a), C. Ferrero (a), A.K. Freund (a), J. Lambard (b) and R. Heintzmann (c), NIM A 356 (1995)

[2] O. Diat (a), P. Bösecke (a) and J. Lambard (b), to be published

(a) ESRF

(b) CNRS Saclay (France)

(c) Univ. of Osnabrück (Germany)



A perfect crystal X-ray analyser with 1.5 meV energy resolution


A new method to construct a spherical crystal X-ray energy analyser has been developed for high energy resolution inelastic X-ray scattering. The energy analysis is based on high-order Bragg reflections from a silicon perfect crystal at angles very close to 90°. In order to preserve the perfect crystal properties in a focusing optics, necessary for meV energy resolution and large angular acceptance, we developed a procedure to mount ~ 12,000 independent small crystals, obtained from the same silicon wafer, on a spherical substrate. The method is based on computer controlled gluing and cycles of etching for each crystal. We obtained analysers with an energy resolution of 1.5 ± 0.2 meV for 21.75 keV X-rays, using the Si (11, 11, 11) reflection, and with 100 mrad2 angular acceptance. This is shown in Figure 106, where we report the elastic X-ray scattering spectrum form a plastic scatterer at a momentum transfer Q corresponding to the maximum of the static structure factor. here, as a consequence of the de Gennes narrowing, the scattering is dominated by the elastic component.



C. Masciovecchio (a), U. Bergmann (a), M. Krisch (a), G. Ruocco (a), F. Sette (a) and R. Verbeni (a), Nucl. Instr. and Meth. 1B-111, 181 (1996) and Nucl. Instr. and Meth. B (1996)

(a) ESRF



Subnanosecond time resolved X-ray excited optical luminescence spectroscopy


The subnanosecond pulse structure of the ESRF source offers a unique possibility to investigate and to exploit fast radiative de-excitation processes which develop after the creation of a core-hole by innershell photoionisation. In a number of systems, the emission of optical photons occurs in the final stale of the relaxation mechanism: this emission is commonly referred to as X-ray Excited Optical Luminescence (XEOL). This emission can also be used to record time-resolved XANES or EXAFS spectra with site selectivity. We have exploited the subnanosecond temporal structure of the ESRF source in multibunch mode to record time-resolved XEOL spectra. In our experiment, the key component was a dissector tube operated in a stroboscopic mode developed originally by E.I. Zinin in 1978. Its measured time resolution is 12 ps and it can be operated at rather high repetition rates (up to several hundreds of Mhz).

We have evaluated the performance of the instrument by monitoring the fast component of the decay of a tetraphenyl-porphyrinato zinc complex (TPP: Zn) at the Zn K-edge. The luminescence decay of TPP: Zn is controlled by intramolecular transitions which are responsible for very fast decay times ( < 1 ns). We have reproduced in Figure 107 the XEOL decay curve of TPP: Zn measured at 100 K with an excitation energy of 9.67 keV (i.e. slightly above the Zn K-edge). This decay curve is clearly non-exponential and could be fitted as a sum of two exponentials with different decay times: 1 < 15 ps and 2 650 ps. This result confirms that the XEOL emission is strongly quenched by non-radiative energy transfer. We have also reproduced in Figure 108 the Zn K-edge XANES spectrum recorded by monitoring only the fast component. This is the first XANES spectrum recorded using XEOL emission and subnanosecond time delay relative to the excitation pulse.



A. Rogalev (a), J. Goulon (a), M.-E. Couprie (b), E.I. Zinin (c)

(a) ESRF

(b) LURE, Orsay (France)

(c) Institute of Nuclear Physics, Novosibirsk (Russia)



Development of efficient area detector systems


During the last year the image intensifier CCD detector developed at the ESRF [1] has been converted from a laboratory prototype to a routinely used detector system for diffraction experiments that produces useful data of high quality.

The image intensifier CCD system covers the energy range 6 - 30 keV with a detective quantum efficiency above 80 %, providing artefact free images of spatial resolution < 200 µm. The useful input field in the corrected flat field image has a diameter of about 185 mm. The optical image produced by the image intensifier is relayed in the standard set-up onto commercially available slow scan CCD cameras of typically 1000 x 1000 pixel resolution, which provide a signal dynamic range > 50,000 for an image read-out time £ 10 seconds.

These cameras are almost always controlled by PCs and their associated software, whereas the beamline control software package (SPEC) is operating on UNIX workstations. In order to operate the CCD cameras from SPEC, the controller card in the PC bus has been replaced with one fitting directly into the UNIX workstation and the necessary software to perform the basic operations of the camera has been written by the Programming Group. At present there are three systems used in this way at the ESRF: Princeton, Photometrics and the "Frelon" camera, a high speed read-out CCD based camera developed at the ESRF by the Detector Group [2]. Table 4 shows the current state of the controlling software for each of the cameras (The references to SDV and BIT3 cards concern the hardware cards plugged into the host workstation).

A crucial aspect of providing high quality data has been the development of appropriate corrections for both spatial distortions and non-uniform response, which is now implemented in the program FIT2D.

A very sensitive test of the quality of the data produced by a new detector for diffraction is to measure the anomalous differences of protein crystals containing a few selenium atoms. Such differences amount to an average value of 1 - 2 % and the signal disappears rapidly if there are errors or noise in the detector system. Before releasing this detector for general use on beamline BM14 a test was therefore made on an enzyme, hydroxymethylbilane synthase, which is involved in the biosyntheses of heme [3]. The structure of a selenomethionine form of the enzyme had previously been determined by a MAD analysis based on data measured at room temperature at the Daresbury Synchrotron Radiation Source and data were collected at low temperature on BM14 to establish the performance of the image intensifier CCD detector for MAD. On BM14 a total of four separate wavelengths could be measured from one frozen crystal in 6 shifts of beam time comprising a total of 630 degrees of diffraction data. The exposure time of 90 seconds per degree per 1 degree image versus a read-out time of 15 seconds per image demonstrates the highly effective duty cycle of the CCD device. The quality of the measured data with the detector can be assessed via the quality of the anomalous Patterson map as shown in Figure 109. By comparing the calculated map and the observed map using data from the image intensifier CCD detector, it is obvious that all the significant features are present, hence the weak signal is present in the data and the detector is useful for protein crystallography, including MAD. Subsequent use of the detector for solving new structures has confirmed the high quality of the data obtained (see page 54).

In contrast to the Photometrics and Princeton cameras which are slow-scan and require several seconds to acquire their images, the Frelon camera is fast-scan. The development of this camera has now been completed and the prototype system tested at beamlines both with image intensifier and scintillant screens as X-ray to visible light input stages. The new Frelon camera is read out directly over a fibre-optic link into the beamline workstation, allowing 1024 x 1024 pixel resolution image acquisition and display at rates of up to 5 images/second, with true signal dynamic range >10,000. Four cameras are now in the final stages of assembly and earmarked for September '96 delivery for small-angle scattering (ID2), diffraction (ID9), and topography (ID19) beamline applications. The Frelon camera opens up possibilities for time-resolved measurements but also poses new problems on data storage and processing: a preliminary microtomography experiment on a single bone sample generated 20 Gbytes of image data from 1125 2D projections (see page 8).



[1] J.P. Moy (a), A. Thomson (a, b) et al., J. of Syn. Rad. (1996)

[2] J.C. Labiche (a), J. Segura-Puchades (a), D. van Brussel (a) and J.P. Moy (a), ESRF Newsletter n° 25 (1996)

[3] A. Haedener (c), J.R. Helliwell (d), S. Harrop (d), A. Cassetta (d), A.P. Hammersley (a), S.O. Svensson (a), A. Thompson (a, b), to be published

(a) ESRF

(b) EMBL, Grenoble (France)

(c) Univ. of Basel (Switzerland)

(d) Univ. of Manchester (UK)