Hard X-ray microscopy with Zernike phase contrast

Authors

  • HWA SHIK YOUN,

    1. Pohang Accelerator Laboratory, Pohang University of Science and Technology, 31 San, Hyoja-dong, Pohang, KyungBuk, Korea, 790-784
      *Nano Mechatronics Research Center, Korea Electronics Technology Institute, 455-6 Masan-ri, Jinwi-myon, Pyungtaek, Kyungki-do, Korea, 451-865
    Search for more papers by this author
  • and * SUK-WON JUNG

    1. Pohang Accelerator Laboratory, Pohang University of Science and Technology, 31 San, Hyoja-dong, Pohang, KyungBuk, Korea, 790-784
      *Nano Mechatronics Research Center, Korea Electronics Technology Institute, 455-6 Masan-ri, Jinwi-myon, Pyungtaek, Kyungki-do, Korea, 451-865
    Search for more papers by this author

Hwa Shik Youn. Tel: 82 54 279 1531; fax: 82 54 279 1599; e-mail; hsyoun@postech.edu

Summary

Zernike phase contrast has been added to a full-field X-ray microscope with Fresnel zone plates that was in operation at 6.95 keV. The spatial resolution has also been improved by increasing the magnification of the microscope objective looking at the CsI(Tl) scintillation crystal. Cu no. 2000 meshes and a zone plate have been imaged to see the contrast as well as the spatial resolution. A Halo effect coming from the Zernike phase contrast was clearly visible on the images of meshes.

Introduction

The important advantage of X-ray microscopes over electron microscopes is the capability of observing the inside structure of samples without destruction. In the soft X-ray region of the electromagnetic spectrum, spatial resolution around 20 nm has already been achieved (Chao et al., 2003). The natural contrast of the so-called water window, together with the spatial resolution, has enabled soft-ray microscopy to be successfully applied to biosamples. However, there are a few shortcomings of this form of microscopy; one is the short depth of focus, and the other is the weak penetration. There have been attempts to overcome the former with the use of harder X-rays. We have developed a new hard X-ray microscope with submicrometre spatial resolution at Pohang Light Source, the detailed description of which can be found elsewhere (Youn et al., 2005). Because there is no privileged spectral region such as a water window in the hard X-ray, one has to implement ways to enhance the contrast for transparent samples in biological applications. In this article, we would like to introduce the upgrades of our microscope in the contrast with Zernike phase contrast (Zernike, 1935) as well as in the spatial resolution.

Optics

As shown in Fig. 1, our microscope is analogous to an optical microscope. We just use highly efficient Fresnel zone plate optics instead of optical lenses. This microscope has been installed at the 1B2/microprobe beam line at Pohang Light Source, which is a third-generation synchrotron radiation facility with an operating energy of 2.5 GeV at Pohang, Korea. The light from a bending magnet of the synchrotron radiation source is focused at the sample position by a condenser zone plate, and the X-ray image of a sample is magnified 27.6 times by an objective (micro) zone plate. The outer and the inner diameter of the condenser zone plate are 4 and 1.5 mm, respectively. This zone plate demagnifies the source at the sample by a factor of 8.85. The objective zone plate has a diameter of 160 µm. Both zone plates are made of Au, and have an outermost zone width of 100 nm and a nominal thickness of 1.6 µm. A Fresnel zone plate with a thickness of 1.6 µm should have an efficiency of 25.6% at 6.95 keV, based upon theoretical calculations. We measured the efficiency of a comparable objective zone, and it turned out to be around 20% (Youn et al., 2005). The depth of focus of this objective zone plate is 226 µm at 7 keV, which defines the upper bound of the allowed sample thickness. The magnified X-ray image is converted into a visible one on the scintillator. This visible image is further enlarged by a microscope objective and captured by a full-frame CCD camera. Thus, we have two lenses to focus in this microscope, the objective zone plate and the visible microscope objective.

Figure 1.

The beam line optics of the microscope. The centre stop was replaced with the annular aperture.

Results: spatial resolution and contrast

Because most biospecimens are transparent to hard X-rays, they would not be visible under bright field imaging (amplitude contrast), so we produced an analogue of Zernike phase contrast (Zernike, 1935) in optical microscopy (Fig. 1). In the soft X-ray region of the electromagnetic spectrum, Zernike phase contrast was found to give considerably better contrast than amplitude contrast (Schmahl et al., 1995). In Zernike phase contrast, the direct beam is isolated from that scattered by a sample, giving a phase shift of either π/2 (positive) or 3π/2 (negative) between them. An annular aperture was installed 3 cm downstream of the condenser zone plate. The inner and the outer diameter of the annular aperture are 2.4 and 3.2 mm, respectively. Thus, the available flux was reduced to 32.5% of that through the condenser. A 2.1-µm-thick Au phase plate was located at the back focal plane of the objective zone plate. The inner and the outer diameter of the phase plate are 85.9 and 114.6 µm, respectively. A 2.1-µm-thick Au would give a 3π/2 phase shift at 6.95 keV, i.e. negative phase contrast, and transmits 29.5% of the direct radiation through the annular aperture. The annular aperture is fixed on the condenser zone plate holder, whereas the phase plate can be in or out of the way of the X-rays, depending upon the phase or the amplitude contrast.

Figures 2 and 3 show a bright field and a negative phase contrast image, respectively, of Cu no. 2000 mesh taken at an energy of 6.95 keV. The insides of the square regions of the latter look dark, whereas the sides of them look brighter, due to the ‘negative’ phase contrast. Inside the first three squares on the second row, there are structures that are hardly visible in the former. There is a noticeable difference in the contrast from those of other recent reports using harder X-rays (Kagoshima et al., 2003; Yokosuka et al., 2003).

Figure 2.

A bright field image of Cu no. 2000 mesh. The 10× microscope objective was used. The width of a bar and a square in Cu no. 2000 are 5 and 7.5 µm, respectively. The annular aperture is fixed on the condenser zone plate, and the phase plate was moved away for the bright field imaging. The length of the white scale bar is 12.5 µm.

Figure 3.

A Zernike phase contrast image of the Cu no. 2000 mesh in Fig. 2.

In our previous paper (Youn et al., 2005), the visible image was magnified by 10 times. We found that the spatial resolution of 20-µm-thick CsI(Tl) scintillation crystal allowed higher magnification. Therefore, the 10× lens was replaced with a 20× lens. With the new magnification, we show another two images of Cu no. 2000 mesh in the amplitude and the phase contrast in Figs 4 and 5, respectively. As with Fig. 3, we can see bright boundaries of the squares in Fig. 5. This is the so-called halo effect, which has been well known in optical microscopy (see, for example, http://www.microscopyu.com/tutorials/java/phasecontrast/shadeoff/index.html). This effect is pronounced whenever there is a significant difference in the indices of refraction between the neighbouring regions such as the grids and the squares of a mesh. To see the halo effect of the Zernike phase contrast quantitatively, we plotted the intensity against the position along the white dotted lines of Figs 4 and 5 in Fig. 6. The changes in the intensity are much larger and overexpressed in the phase contrast compared to the amplitude counterpart whenever boundaries are crossed. Because we are in negative phase contrast, the change is in the opposite direction. We can see two parallel bright lines in the horizontal sides of the squares, which overlapped in the lower-magnification image of Fig. 3. In Zernike phase contrast, the 3λ/4 phase difference is introduced by the phase plate. The wall of this part of the grid has more angles relative to the incoming X-rays. As one crosses the horizontal grid bar, following the arrow in the figure, the phase difference caused by the mesh changes from 0 to the thickness of the mesh, which is 4.6 µm. This results in a modulation in the intensity, which is shown as double bright lines. Apparently, a Cu mesh is not a phase object, and there is a severe halo effect. Therefore, Zernike phase contrast better serves phase objects like the features in the squares in Figs 2 and 3. In the bright field images of Figs 2 and 4, the grid looks flat, whereas there is a variation in the intensity of the grid in the phase contrast version, reflecting the fact that the thickness of the grid is not uniform. The structure of Cu no. 2000 mesh is not as fine as that of Cu interconnect structures (Neuhäusler et al., 2003), resulting in broader boundaries. Apparently, there is a noticeable gain in spatial resolution with the 20× lens. The side of a square hole of the Cu no. 2000 mesh that we used is 7.5 µm. From Fig. 4, the field of view is estimated as 17 × 25 µm2.

Figure 4.

A bright field image of Cu no. 2000 mesh. The 10× microscope objective was replaced with a 20× one from this point on.

Figure 5.

A Zernike phase contrast image of the Cu no. 2000 mesh observed in Fig. 4.

Figure 6.

Intensity across the vertical grid bars along the white dotted lines in Figs 4 and 5. We integrated two points along the vertical direction and subtracted one from the lower curve representing the amplitude contrast mode to separate it from the upper curve of the negative Zernike phase contrast mode. npc, negative Zernike phase contrast; ac, amplitude contrast.

In Fig. 7 we show a bright field image of the outer part of a zone plate that was used as an objective zone plate lens in our previous publication (Youn et al., 2005). This zone plate also has an outermost zone width of 100 nm. The zones are visible. Figure 8 is the image of the same area that we took after inserting the phase plate at the back focal plane of the objective zone plate. A dust particle is hardly visible at the lower left of Fig. 7, whereas it is clearly visible with a moderate halo effect along the boundary in the phase contrast image of Fig. 8. Also, we can immediately see from Fig. 8 that some of the zones are not evenly distributed. A conservative estimate of the spatial resolution of the image in Figs 7 and 8 is better than 122 nm, which is consistent with Chao et al. (2003) in that ‘with partially coherent imaging, the spatial resolution can be better (smaller) than the outermost zone width of the MZP’.

Figure 7.

A bright field image of a zone plate. This zone plate has a diameter of 160 µm, an outermost zone width of 100 nm, and a nominal thickness of 1.6 µm.

Figure 8.

A Zernike phase contrast image of the zone plate in Fig. 7.

Discussion

We have succeeded in implementing Zernike phase contrast in hard X-ray microscopy, which would be especially useful in studying the inner structure of thick and transparent biosamples as well as material samples. We have confirmed that there is a halo effect in the X-ray part of Zernike phase contrast like the analogue in visible light. Zernike phase contrast is much more useful for phase objects in which there is not too much variation in the indices of refraction. As for the spatial resolution, our microscope can explore samples at a scale that is not accessible with an optical microscope. Therefore, our microscope can be used to investigate a territory that is not accessible with an optical microscope or electron microscopes.

Acknowledgements

We would like to thank Dr Wenbing Yun and Dr David Attwood for their illuminating comments and discussions. The excellent support of the staff at the Pohang Accelerator Laboratory especially, Dr S. Y. Rah and Mr Hyo-Jin Choi, is gratefully acknowledged. Hwa Shik Youn wishes to thank Dr Steve Wilkins for having provided a nice channel-cut crystal. This work was partially supported by the Korean MOST (Ministry of Science and Technology) through the X-ray/Particle-beam Nano-Characterization Program. The research at the beam line 1B2 is supported by MOST and the POSCO (Pohang Iron and Steel Company).

Ancillary