X-ray microfocusing by polycapillary optics


In this work the results on X-ray micro-imaging by means of novel polycapillary optical elements will be presented. To simulate various radiation propagation processes in both single capillary and polycapillary systems, a PolyCAD code was developed.

Many experiments have recently revealed the advantage of confocal optical configuration for the fluorescence studies. Moreover, our experimental prototype layout (developed in the INFN – LNF laboratories, Frascati) enables the possibility to obtain µXRF mapping in simultaneous with X-Imaging. The recorded image of an extended sample is limited to 6 µm by the CCD pixel dimensions; the use of a second polycapillary optics in the confocal scheme followed by a SDD detector provides an additional option for elemental studies. A prototype of compact XRF spectrometer with a spatial resolution less than 100 µm has been designed.


Radiation propagation in capillary optical elements is important due to potential applications in X-ray optics. Regardless of the capillary optics geometry, special feature of these structures is a hollow inner cavity, which could act as a channel for selective radiation penetration. This produces strong radiation redistribution behind capillary systems, revealing essential structural behaviour due to the spatial system geometry.

Research in X-ray propagation in capillary structures shows that diminishing the capillary internal radius down to microns and submicrons results in the ability of handling the radiations of higher quantum energies as well as of focusing in much smaller spots. For instance, while the first polycapillary systems were able to concentrate X-rays in spots of hundreds or tens microns, the last generation lenses (presently, the 5th generation of polycapillary optics has been announced, enable to shape X-ray beams with micron resolution and mrad divergence. The latter becomes very attractive for various applications based on both focusing and imaging efficiencies of the optics used.

A rapidly developed during the last few years for non-destructive analysis by micro-X-ray fluorescence spectrometry (uXRF) is a promising multi-elemental technique. Except for the Synchrotron Radiation, which is a suitable probe for micro spots, it is rather routine to perform a tabel-top μXRF analysis because of the difficulty of producing an original small-size X-ray beam as well as its focusing. Recently developed for X-ray beam focusing, polycapillary optics offers alternative laboratory X-ray micro probes7-12. The combination of polycapillary lens and fine focused micro X-ray tube (with a source spot size less than 50 μm in diameter) can provide the high intensity radiation flux on a sample that is necessary in order to perform the elemental analysis.

In comparison to a pinhole, an optimized “X-ray source-optics” system can result in radiation density gain (the ratio of X-ray intensities with and without the optics at the focal spot) of more than 3 orders by the value. Due to rather high intensity irradiation of a limited space, it becomes possible to realize express non-destructive XRF studies. The most advanced way to get that result is to use the confocal configuration based on two X-ray lenses, one of which is responsible for the fluorescence excitation on a sample and another – for the detection of secondary emission from a sample studied, i.e. X-ray fluorescence. In case of X-ray capillary microfocusing a μXRF instrument designed in the confocal scheme allows us to obtain a 3D elemental mapping.

In this work we will show preliminary results obtained with our prototype, that is a portable X-ray microscope useful, by definition, for X-ray imaging but in the same time providing on-site X-ray fluorescence analysis; the basic motivation of this kind of spectroscopy is in its noon-destructive base. In order to focalize the X-ray beam we are presently using polycapillary optics, while in the future, for X-ray imaging the use of the combination of polycapillary lens, as a concentrator, and compound refractive lens (CRL), as a magnifier, has been planned.


X-ray transmission through polycapillary lenses is based on the phenomenon of Total External Reflection (TER)of radiation by a surface. Any X-ray quantum, which strikes the channel’s wall (typically for the polycapillary optics technology, the wall substance is characterized by a glass composition) at the incidence glancing angle less than the Fresnel angle θc (the critical angle of TER), propagates along a channel through multiple reflections. The number of reflections can vary from one to a few tens and even hundreds, that’s why this optics is called multiple reflection optics.

Fig. 1 A schematic draft of a single channel; the two darker gray paths are meridional trajectories, while the lighter one is spiral

In Fig. 1 it is shown the behavior of a ray propagation inside a channel. The expected trajectory is the meridional one (two darker gray lines), which is an ideal path. Indeed, majority of the rays has a propagation direction different from the meridional case (lighter line). As known, in ideal case the divergence △f obtained behind a capillary (with a single capillary of the diameter d) is defined as

This definition of the divergence, valid just for preliminary estimations of the optics, will provide rather overestimated results on an halo effect. Moreover, having considered a set of capillaries and not only one, we have to underline that, in case of the smallest spot size obtained, the integrated divergence is not the sum of divergences from each channel (see equation in above). It is due to the following important effects. First, the real halo is determined by the divergence from the most of photons converged to the focal spot, that are following not meridional trajectories; and, thanks to the large number of photons, their contribution to the final distribution becomes resulting. Second, the real spot dimension should not be determined by geometrical size, which follows from above equation, but through the full width at half maximum (FWHM) of the radiation density distribution.

Our interest in the development and application of new X-ray systems based on polycapillary optics created a flexible and complete ray-tracing program, named PolyCAD to simulate and visualize the processes of radiation propagation through any kind of polycapillary optical systems and the radiation distributions at the optics output. In the last twenty years, many software packages based on various approximations-simplifications taking into consideration only specific aspects of X-ray propagation, have been developed.

The ray tracing code PolyCAD is a software that evaluates the solution analytically, without any kind on path approximation; the solutions have been obtained considering a system of the equations for photon path together with the channel surface equation. In order to show how PolyCAD can simulate different optical configurations, the behavior of a polycapillary lens in the case of a source having a three-dimensional structure was estimated. It was simulated for a source formed by two spheres, in particular, when a little spherical source is “hidden” by a second larger source located on the optical axis, Fig. 2.

Fig. 2 Intensity distributions for: (2a) two sources (S1 and S2), (b) only the source S1; (c) the distribution 2b subtracted by 2a; (d) evaluation of the intensity distribution S2. From comparison of the two last images, it becomes evident tha2t both the calculated and evaluated focal positions are equal, with the same intensity distributions

For simulations, we have used the following parameters: two X-ray spherical sources with radii of S1=0.3 cm and S2=0.1 cm, respectively, emitting isotropically1 keV photons. The distance of the bigger sphere from the polycapillary inlet plane is 30 cm, while the distance for the smaller one equals to 40 cm. The cylindrical polycapillary has the length of 10 cm, cross secticon radius of 1 cm, single channel radius of 0.9 μm.

In Fig. 2a the radiation intensity distribution extracted from the contour map along the axis x=0 is given. The source S2 is completely hidden by S1: in this case only a little bump due to S2 is present in the upper right side of a contour map (Fig. 2a), but “a priori” it is not possible to evaluate the S2 conjugate position. Now, in order to operate a deconvolution procedure by means of the PolyCAD simulations, for the same optical configuration, we evaluated the situation with the source S1 alone (Fig. 2b). By subtracting the intensity distribution 2b from the distribution 2a in the resulting profile (Fig. 2c), the presence of the little source S2 becomes evident and its position could be evaluated at about 40 cm from the lens. In order to confirm this result we have reconstructed the intensity distribution only due to the radiation coming from the source S2 and passing through the lens (Fig. 2d). Comparing the two last images proves that the second source is located exactly at the 40 cm distance from the lens entrance.


At Laboratori Nazionali di Frascati (INFN – LNF) we have an experimental set-up designed specially for characterization and utilization of X-ray optical systems, Fig. 3. In order both to respect the safety rules and to perform precise spatial and angular measurements, we have designed a cabinet with Pb windows realized by “Officine Elettrotecniche di Tenno”, together with an anti-vibrational optical table (Newport-RS2000).

Fig. 3 A picture of the X-ray testing facility, at LNF – INFN. In the image are showed the X-ray source, the detectors and the optical mechanics

As a radiation source, for our analysis we have utilized an Oxford Apogee 5000 tube (CuKa), with µm and power of 50 kV and 1 mA. As the detectors, we have three kind of detectors: a scintillator one (Saint Gobain), with an active area of 2.56 mm, and two CCD (FDI 1.61:1 and FDI 1:1.61, by Photonic Science) with a complete software for “X-ray rough/fine imaging”. The first camera has a sensitive area of 4×3 mm with a 3.5×3.5 pm resolution, while the second one has an active area of 14.4×10.8 mm with a pixel resolution of 10.4×10.4 μm. The third detector is aSDD detector with 5 mm² effective area. Optical mounts and remote controller actuators designed by Newport enable controlling in remote by means of a LabView software. Polycapillary optics(cylindrical, semilens and full lens) are manufactured by Unisantis S.A. and IRO (Institute for Roentgen Optics, Moscow), while the CFRL samples are provided by Dr. Dudchik (Institute of Applied Physical Problems,Minsk).

One of the main activity of our laboratory is to test and characterize polycapillary optics: in these years we perform a procedure of characterization, that is able to deterimine all the phishycal and geometrical properties of each kind of polycapillary optics. As a first step, we characterized the behaviour of the transmission coefficient versus the channel alignment, by the relationship between the source position and the capillary axis and by a set of the 2D angular mappings obtained rotating a capillary by m

Fig. 4 Geometry of the experiment for angular measurements, and the acceptance angle. The axis source-Gimbal mount center-detector is fixed. The scanning is obtained moving the polycapillary optics in ϕ, θ space

eans a Gimbal mount, Fig. 4.

Secondly, we focused on the image characterization, as obtained by the aligned capillary on a plane (CCD detector) at various distances. Finally, we compare our experimental results with our simulated by PolyCAD evaluations.

In order to align the lens and to take pictures at high resolutions, the first prototype of a transmission imaging microscope was realized at LNF INFNF; it is composed by an X-ray source, a semi-lens and two detectors, one of which is a scintillator and another one is a CCD.For this unit, our choice for a semi-lens was postulated by the possibility to get a very small blurring effect due to rather small radiation divergence behind the optics (without taking into account diffraction effects on sample edges, i.e. far away from a wave zone, as well as multiple scattering radiation in matter). As known, for any kind of the lenses, there is a residual divergence responsible for the resolution graduation; if we suppose that the sample dimension △X0 is reproduced at the detector as △X, then the effect can be estimated by

Hence, in order to reach the higher resolution, keeping the sample-detector distance I unchanged, we have to diminish a residual divergence △θ.

Introducing a second polycapillary lens, where the angles source-sample and sample-detector are equal, it is possible to develop an X-ray microscope prototype for elemental studies as an additional option. This kind of set-up is well known as a “confocal scheme”, with a nominally focal spot size less than~100×100 μm². Many works were presented in the last years, showing advantages of this optical configuration: its peculiar property allows a micro-fluorescence mapping simultaneously with imaging to be performed. In Fig. 5 the last prototype scheme version is shown.

Fig. 5 Experimental layout for µXRF and imaging: 1 – the X-ray source; 2 and 4 – the 1st and 2nd polycapillary optical elements, respectively; 3 – the sample; 5 – the laser sensor used for fine lens positioning; 6 – the detector


The first microscope prototype designed at LNF-INFN last year was based on transmission imaging and composed by an X-ray source, a CCD detector and a semi-lens. Our choice for a semi-lens was postulated by the possibility to get a very small blurring effect due to rather small radiation divergence behind the optics (without taking into account diffraction effects on sample edges, i.e. far away from a wave zone, as well as multiple scattering radiation in matter). In order to evaluate the highest resolution available, we used a sample of standard mesh: Au 1000 with a hole width of 19 μm, and bar width-of 6 μm)(Fig.6).

Fig. 6 High resolution image of a gold mesh 1000 recorded by the detector placed at about 44 cm from the sample

Introducing a second polycapillary optics, as mentioned below, is possible to obtain the confocal configuration. The two full lenses are chosen so that each focal spot is as smallest as possible, and the transmission is as greater as possible. The first lens has provided a focal spot of~ 90μm at the transmission of about 50%,while the second one was characterized by a spot of~100μm and the transmission of 42%. Fig. 7 shows our first μXRF results for a neolithic human bone; as seen, all main elements were well resolved.

Fig. 7 XRF spectrum of a neolithic human bone obtained in confocal configuration

A real mapping of μXRF Spectroscopy was obtained on standard sample of Ferric Oxide (Fe2O3); the standard monophasic minerals of Fe2O3 was prepared in a 1000 class clean room at DISAT, University of Milano Bicocca. Sample of 2 mg of mineral was weighed by an analytical balance (sensitivity 10-5 g), in an ultra clean cuvette with 20 ml of ultra pure(MilliQ)water and homogenised by a mechanical stirrer; 100 μl of mineral suspensions were diluted in 20 ml of MilliQ water. Mineral suspensions were tested by using Coulter Counter technique to check the concentration and size distribution of the particles.

In order to not overlap two close measures, we made a scan, by a remote system, of a 4×4 mm of region with a step movent of 200×200 μm. In the Fig. 8 it is shown the image result; the red colour represents iron (in intenisty scale of black-bright red), while the green pixels are manganese (probably there is a contamination; for this reason we are desing a glow box in order to make measure in an helium atmosphere).

Fig. 8 µXRF mapping spectrum Ferrum Oxide sample, deposited on a silicon wafer. The gray pixels represents iron (in a scale of gray), while the white bright spots are manganese trace. The measured area is 4×4 mm with a step sized of 200×200 µm

For comparison, in the Fig. 9 is shown the relative sample, obtaned with a visible microscope.

Fig. 9 The same image of Fig. 8 obtained by a visible optical microscope


In our work we tried to utilize some specific characteristics of polycapillary optics in order to get the high resolution images of the extended objects and to obtain micro Fluorescence Spectroscopy.

In a set of experiments, measurements of a test capillalry optics showed a reduction of scatter fraction; images of a contrast detail phantom revealed a corresponding increase in image contrast when compared to anti-scatter grid and no grid methods. For this reason, in order to obtain high contrast imaging a specific polycapillary semilens was designed, by which the divergence of 1.5 mrad were obtained. A small divergent beam was then used to record an image on the CCD with 6 μm resolution. This result proves the fact that a small divergent beam will provide high contrast image transfer. In order to increase the resolution, the next step is to combine a full polycapillary lens with a CRL.

Another important thing to be underlined is the fact that the capillary optical elements provide high X-ray flux into a small spot size, permitting short exposures of small-volume samples. Such a capability is ideal for a number of applications. As above discussed, polycapillary lenses in the confocal scheme are the ideal candidates to overcome some of the main problems of laboratoryX-ray instruments including the portable units, namely, the rather low photon flux delivered on a sample. We have shown first results both for X-Ray Imaging and for XRF Spectroscopy. For the first case, the simply use of a polycapillary semi-lens allows us to obtain transmission X-ray imaging with a spatial resolution of about 6 μm.

In the second case we have shown that a confocal optical scheme used with a low power(50 W) conventional tube and high resolution SDD detector becomes a powerful instrument for mapping samples, with a micro-size spot, through X-ray fluorescence analysis (about l00x100 μm spot. In order to decrease the X-ray spot dimensions, we would like to substitute present polycaloillary lenses with the last generation ones; these lenses enable obtaining focal spots of a few microns. In nearest future, we would like to combine our experimental setups in order to design a new one, compact and portable, which allows performing simultaneously sample elemental analysis and X-imaging.