Imaging X-ray fluorescence microscope with a Wolter-type grazing-incidence mirror

A Wolter-type grazing-incidence mirror was used as an objective for an imaging X-ray fluorescence microscope. The microscope was constructed at the beamline 6C2 of the Photon Factory. The shortest wavelength used was ∼0. 1 nm, which was limited by the grazing-incidence angle of the mirror. To demonstrate the possibility of recording X-ray fluorescence images, several fine grids were used as test specimens. Characteristic X-rays emitted from each specimen could be clearly imaged. Spatial resolution was estimated to be better than 10 µm.

1. Introduction

X-ray fluorescence has been widely used for trace-element analysis for many years. Recent developments of high-brightness electron storage rings have stimulated the construction of various types of X-ray microprobes for X-ray fluorescence microanalysis. Relatively high-energy X-rays are used to produce X-ray fluorescence because the X-ray fluorescence yield becomes larger for heavy elements. Typical X-ray focusing devices are a Kirkpatrick–Baez grazing-incidence mirror (Suzuki & Uchida, 1992), a thick Fresnel (Kamijo et al., 1997) or Bragg–Fresnel zone plate (Erko et al., 1994), and a capillary tube (Bilderback et al., 1994). A new compound refractive lens (Snigirev et al., 1996) has also been reported. All these devices, however, have been used only for producing small spots in the hard X-ray region. As a result, all X-ray fluorescence images were obtained with scanning X-ray microscopes. Owing to the lack of an efficient objective lens for recording amplitudes and phases of X-ray fluorescence, an imaging X-ray fluorescence microscope has not been demonstrated.

In this paper, we propose an imaging X-ray fluorescence microscope with a Wolter-type grazing-incidence mirror and present some preliminary results.

2. Optical systems

Among many optical devices, a Wolter-type grazing-incidence mirror (Wolter, 1952) is one of the most suitable objectives for hard X-ray imaging. Because it has small coma and no chromatic aberration, it can be used for polychromatic X-rays like X-ray fluorescence. In addition, its numerical aperture is not as small as a zone plate, which means that the resolving power is relatively large. Owing to technological difficulties, these factors have not been taken advantage of. Only large Wolter-type mirrors have been successfully used for X-ray astronomy. Recent developments of nanometre fabrication technology, however, enable us to produce small Wolter-type mirrors for hard X-rays.

Fig. 1 shows the parameters of the Wolter-type grazing-incidence mirror we designed and fabricated for hard X-rays. The mirror consists of an axisymmetric hyperboloidal and ellipsoidal mirror. It is known that X-rays can be totally reflected at a grazing incidence angle smaller than the critical angle (θ_c) which is given by

where λ is the wavelength in nanometres and ρ is the density of the reflection surface in g cm⁻³. With a platinum-coated mirror, 0.1 nm X-rays can be reflected at a grazing angle of 7 mrad. The mirror optics satisfy two basic imaging conditions: one is Rayleigh's quarter-wavelength rule and the other is Abbe's sine condition. Ray-tracing calculations show that the field of view is about 0.6 mm or 0.2 mm in diameter, assuming that the spatial resolution is better than 10 µm or 1 µm, respectively.

Figure 1 Geometrical parameters of the Wolter-type mirror. Units: mm.

We fabricated a small Wolter-type mirror by using the glass replica technique which had been developed to produce the same kind of mirrors for soft X-ray microscopes (Aoki et al., 1992). First we polished the master mandrel (tungsten carbide) and then made a glass replica (Pyrex glass). The axial surface deviation from the design parameter was better than 1 µm and the surface roughness was about 1 nm r.m.s. The inner surface of the glass replica was coated with tens of nanometres of platinum film. Details of the fabrication processes have been described in our previous paper (Onuki et al., 1992).

We built an imaging X-ray fluorescence microscope by using this mirror at the Photon Factory. A schematic diagram of the experimental arrangement at beamline 6C2 is shown in Fig. 2. No monochromator was used because the mirror can reflect broad-band X-rays. Two 0.2 mm-thick and two 0.3 mm-thick beryllium plates were put in the beamline as the vacuum windows. These plates worked as high-pass filters so that X-rays above 4 keV could be incident on the samples. The surface of the sample was inclined at 50° to the incident beam and vertical to the optical axis of the objective mirror. The area detector was an X-ray sensitive CCD camera (TI, TC-215). The number of pixels was 1000 × 1018 and the area of one pixel was 12 µm × 12 µm. The detector was cooled by a Peltier cooler to below 253 K to reduce the dark current. The X-ray path between the mirror and the detector was evacuated.

Figure 2 Schematic diagram of the experimental arrangement.

3. Results

Several test samples were imaged to reveal the performance of the system. Fig. 3 shows the X-ray fluorescence spectrum of a copper grid (#100) obtained at the detector plane with a Ge SSD. High-energy X-rays of about 15 keV were observed. Fig. 4(a) shows the X-ray fluorescence image of a stainless-steel netted grid (#200). The exposure time was 50 s. The overlapped parts of the grid produced the shadow of the wire because of the inclined irradiation of the incident X-rays. Fig. 4(b) shows the X-ray fluorescence image of the stainless-steel grid sheet (#200). The exposure time was 90 s. There was no shadow of the grid. Compared with those in Fig. 4(a), the corners of the grid in Fig. 4(b) are round, which are not attributed to the blur of the image but their original patterns. Fig. 4(c) shows the X-ray fluorescence image of 50 nm-thick chromium square patterns (#200) coated on a silicon wafer. The exposure time was 30 min.

Figure 3 X-ray fluorescence spectrum from a copper grid (#100).

Figure 4 X-ray fluorescence images. (a) Stainless-steel netted grid (#200), (b) stainless-steel grid sheet (#200), (c) chromium 50 nm-thick pattern (#200).

4. Conclusions

An imaging X-ray fluorescence microscope has been demonstrated with a Wolter-type grazing-incidence mirror for the first time. Although the experiment could not show the distribution of a specific element, the use of an energy-tunable monochromator or an energy-resolvable area detector can solve this problem. The microscope could image very thin film patterns, though the spatial resolution was not so good. For a further improvement of the spatial resolution, a new mirror-fabrication technique must be developed. The use of a brighter X-ray source is also recommended.