1. Introduction

When a monochromatic light wave is incident on a periodic grating, the wave-field at the exit of the grating is replicated at the multiples of the 'Talbot distance', Z_T, downstream of the grating. The Talbot effect, this self-imaging phenomenon of a periodic grating, was first discovered by Talbot [1] in 1836, and demonstrated by Lord Rayleigh a few decades later [2]. In addition, the period in the propagating wave-field is reduced to half at 1/4 Z_T and 3/4 Z_T, and the phase of the grating structure reverses at 1/2 Z_T. The Talbot distance Z_T is given by Z_T = 2a ²/λ, where a is the grating period and λ is the wavelength of the incident light. The Talbot effect is schematically illustrated in Fig. 1 .

 

Fig. 1 Schematic view of the Talbot effect. The grating image is replicated at the Talbot distance, Z_T while its phase is reversed at Z_T/2. The period of the wave-field intensity reduces to half the grating period at Z_T/4 and 3Z_T/4.

Download Full Size | PDF

The Talbot effect has applications in various fields such as atomic optics [3], Bose-Einstein condensation [4], and interferometry of fullerene molecules [5]. Recently, the Talbot effect in x-ray regime has also attracted a great attention in regard to applications in x-ray phase contrast imaging [6,7] where the modification of the Talbot image by the phase variation in the incident wave was used to retrieve the image of an object. The Talbot effect has also great implications in optical lithography based on short wave length light such as x-rays, since patterns with nano-meter scale periods can be recorded by placing a photoresist (PR) at the Talbot distance. Talbot distance reaches up to a few millimeters when a hard X-ray beam is incident on a grating with a period of a few hundred nanometers. This distance is easily accessible for positioning and manipulating PRs.

Case et al. demonstrated the Talbot effect using a green diode laser and a grating with a 200 μm period [8], where the Talbot distance was approximately 15.4 cm. Guerineau et al. demonstrated the Talbot effect using polychromatic light in the visible regime [9]. In these experiments, the grating periods were on the order of hundred micrometers. There are, however, only a few direct observations of the Talbot self-imaging under hard x-ray illumination. Cloetens et al. reported an observation of fractional Talbot images on a polymer film by using monochromatic hard x-ray [10]. The grating period used in their experiment was about 12.5 μm. To apply the Talbot effect to nano-scale lithography, reducing the wavelength of light sources and the dimension of grating period are necessary.

In this study, we report an observation of the Talbot effect using a broadband hard x-ray beam, which is more efficient for lithography than a monochromatic beam since the exposure time on a PR could be reduced significantly using a high flux broadband x-ray beam. We used a specially designed x-ray grating with a sub-micrometer scale period to illustrate the Talbot effect. A simulation based on the Fresnel diffraction theory predicted the experimental observations quite well.

2. Experimental setup

The experiment was performed with synchrotron radiation at beamline 5C1 at Pohang Light Source (PLS) in Korea. The electron beam energy was 2.5 GeV and the nominal bending magnetic field was 1.32 Tesla. No monochromator or focusing element was involved in the beamline. To illustrate the Talbot effect, we used a specially prepared sectioned WSi₂ /Si multilayer grating. WSi₂ and Si layers were alternately grown by sputter deposition on to a silicon substrate to prepare the grating. It was then sectioned vertically into small slices and polished until the section width reaches to about 20 μm. Detailed process of the grating preparation was reported previously [11,12]. Shown in Fig. 2(a) is a schematic illustration of the experimental set up.

 

Fig. 2 (a) Schematic illustration of the experimental configuration. (b) SEM image of the sectioned Si/WSi₂ multilayer grating. The bright and dark regions represent WSi₂ and Si respectively. (c) Transmittance through the transparent Si and the opaque WSi₂ regions. (d) Broken red line represents the spectral density of the bending magnet and the solid line represents the spectral density of the polychromatic beam incident on the multilayer mask.

Download Full Size | PDF

Figure 2(b) shows a scanning electron micrograph (SEM) of the grating. The period of the sectioned multilayer grating corresponding to the thickness of a single WSi₂ /Si pair was 676 nm. The fraction of Si and WSi₂ was incidentally 57.1% and 42.9% respectively. The ratio of section depth to grating period was about 30, which is essential for sufficient x-ray contrast. The transmittance of the white x-rays through the relatively transparent Si regions and the opaque WSi₂ regions of the grating was calculated using the tabulated x-ray attenuation factor [13], and illustrated in Fig. 2(c).

The spectrum of the x-ray beam incident on the grating shown in Fig. 2(d) was estimated by considering preferential absorption of low energy x-rays by the Be-windows together with the spectral density calculated using the bending magnet parameters [14]. Two Be-windows of 254 μm thick were placed upstream of the grating, separating the beamline from the front-end and the experimental station. The spacing between the last Be window and the grating was 48.5 cm. The wavelength at the maximum flux in the spectrum occurred at 1.2Ǻ, and the spectral width (full width at the half maximum: FWHM) was about 1.2Ǻ. Long wavelength components were attenuated by the Be-windows while short wavelength components were limited by the critical energy set by the bending magnet parameters.

The wave-field intensity of the x-ray beam through the grating was recorded on a PR (ZEP520A-7, Zeon corporation) spin-coated on Si (001) substrates. The thickness of the PR was set to 250 nm. The grating-to-PR gap varied from 200 μm to 6 mm, and the exposure was adjusted at each gap distance to obtain images with the clearest contrast. The PR exposed by x-rays was developed in ZED-N50 for 40 sec, and rinsed in isopropyl alcohol for 60 sec.

3. Simulation of the Talbot effect under broadband x-ray illumination

In order to understand the propagation of the broadband x-ray beam through a periodic grating structure, we carried out a simulation of the wave-field propagation using the experimental parameters given in the previous section. The wave-field at a position (X, Y) in the downstream of a delta-function like monochromatic source located at origin is given by the impulse response function of the Fresnel diffraction,

The grating function f(x,y;λ) represents the amplitude of the wave-field at the exit surface of the grating. It can be evaluated by considering the grating structure and the transmission coefficient through the structure. In the case when a broadband beam is incident on the grating, the total wave-field through the grating can be obtained by summing the contribution of each wavelength component given in Eq. (2).

The simulated wave-field intensity profile downstream of the grating calculated from Eq. (4) using the experimental grating structure and the spectral density ${\left|B(\lambda )\right|}^2$shown in Fig. 2(d), is illustrated in Fig. 3(a) . We defined the effective Talbot distance$Z_T^{eff}$ by identifying the distance where the period becomes half in the simulation as $Z_T^{eff}/4$. $Z_T^{eff}$was 5.76 mm. For comparison, a simulated intensity profile under a monochromatic beam at an averaged wavelength ${\lambda }_{av}$defined by,

 

Fig. 3 Simulated wave-field intensity downstream of a periodic grating under broadband illumination (a) and monochromatic illumination (b). We used the grating structure and the spectral density used in this experiment. For the monochromatic case, the averaged wavelength obtained from the spectral density was used.

Download Full Size | PDF

The major features of the Talbot effect were still present even in the broadband illumination shown in Fig. 3(a), although the detailed interference structure of the Talbot carpet shown in Fig. 3(b) was smeared out. As the quarter of the effective Talbot distance $Z_T^{eff}$was approached, the intensity in the transparent regions became weaker while the opaque regions became brighter both in Figs. 3(a) and 3(b). The fine structure presented in Fig. 3(b), however, were not present in Fig. 3(a). The line profile at $Z_T^{eff}/4$ showed that the period in the wave-field intensity was reduced to half the grating period. Downstream of this distance, the intensity in the opaque regions kept increasing, and the phase of the wave-field intensity was reversed at the half Talbot distance $Z_T^{eff}/2$. The grating image was replicated at this distance, but the bright and dark regions are reversed.

In the case of monochromatic illumination shown in Fig. 3(b), the wave-field intensity from $Z_T^{eff}/2$ to $Z_T^{eff}$ was almost exactly opposite of the intensity from 0 to $Z_T^{eff}/2$. The period in the wave-field intensity became half at $3Z_T^{eff}/4$ and the grating structure was completely replicated at $Z_T^{eff}$. On the other hand, in the case of broadband illumination, the intensity in the opaque regions decreased, but did not disappear completely even at the Talbot distance. Therefore, the replication of the grating structure at $Z_T^{eff}$ was incomplete as shown in Fig. 3(a).

The presence of the key features of the Talbot effect using a broadband polychromatic x-ray beam in the simulation is rather surprising but understandable. A broadband x-ray beam generated from a typical white x-ray beamline with Δλ/λ about 100% can still exhibit every feature of the Talbot effects, since the spectral distribution is narrow enough.

4. Experimental results and discussion

Figure 4 shows SEM images of the x-ray wave-field intensity at several distances d downstream of the grating recorded on the PR. The bright regions, where the PR material survived, represent the opaque low wave-field intensity regions, and the dark regions correspond to the high intensity regions. At z = 200 μm, the image recorded on the PR shown in Fig. 4(a) replicated the grating structure shown in Fig. 1 well. The period of the wave-field intensity in the image was 676 nm, exactly matching the grating period. This distance was much shorter than the effective Rayleigh distance $a^2/{\lambda }_{av}\sim 997\mu \text{m}$ estimated using the averaged wavelength, 0.14 nm with the opening size a corresponding to the Si layer thickness of 386 nm. The PR was located well in the near field region and the recorded image represented the shadow of the grating closely.

 

Fig. 4 Wave-field intensity recorded on PR at 0.2 mm(a), 1.4 mm(b), 2.8 mm(c), 4.2 mm (d), 5.6 mm (e) downstream of the grating. On the right is the simulated result for comparison.

Download Full Size | PDF

The grating structure was replicated quite clearly as z approached 2.8 mm for the first time as shown in Fig. 4(c), although a broadband x-ray beam was used. Experimentally, this distance corresponds to half the Talbot distance, since the grating structure was replicated first at half the Talbot distance as shown in the simulation results. A series of measurements were performed around twice this distance to find out the Talbot distance experimentally, which turned out to be 5.6 mm. This is very close to the effective Talbot distance estimated in the simulation, 5.71 mm. Figure 4(e) shows an SEM image of the wave-field intensity recorded on the PR at 5.6 mm. The grating structure was mostly replicated. At this position, however, a dark line signifying high wave-field intensity showed up in the middle of the bright stripe. This was consistent with the simulation where the intensity in the opaque regions did not disappear completely at the Talbot distance.

It is interesting to note that the wave-field intensity at a distance corresponding to a quarter and three quarters of this distance, 1.4 mm and 4.2 mm respectively, was quite different from the grating structure. As shown in Figs. 4(b) and 4(d), the period of the recorded intensity was reduced to 338 nm, half the original grating period. We think that this phenomenon was due to the relatively narrow spectral distribution of the x-ray beam used in this experiment as noted in the description of the simulation.

In fact, the grating structure replicated at 2.8 mm, which we consider to be effectively half the Talbot distance, shown in Fig. 4(c), is not exactly the same as the near field image shown in Fig. 4(a). Upon close inspection, we noticed that the number of lines in the image was smaller by one in Fig. 4(c) than in Fig. 4(a). This is in fact due to the phase reversal expected at this distance, which is illustrated clearly in Figs. 5(a) , 5(b) where we placed the two intensity profiles next to each other. The bright and dark regions are reversed in the two images. The phase reversal is also well illustrated in the line profiles of the simulated result shown in Figs. 5(c) and 5(d). The phase reversal at the half the Talbot distance is due to the interference of the light wave emanating from each transparent Si region that becomes destructive in the opening positions and constructive in the closed positions in the downstream region. The edge diffraction effect appeared near the edge of the grating in the simulation, was not recorded well in Fig. 5(b). We think that the contrast was not enough to record the detailed features related to the edge diffraction.

 

Fig. 5 Recorded wavefield intensity in the near field region (a) and at half the Talbot distance (b). The number of line patterns is reduced by one in (b) due to phase shift of electric field. The observed phase shift is consistent with the line profile of the simulated intensity profile in the near field (c) and at half the Talbot distance (d). Dashed line indicates the relative phase shift between the two regions.

Download Full Size | PDF

It is interesting that the interference effect was observed even under broadband white x-ray illumination. We note that the key features of the Talbot effect, i.e. the periodic replication of the grating structure at half the Talbot and the Talbot distances were exhibited even under the broadband illumination.

5. Conclusion

We have observed the Talbot effects using a broadband x-ray beam generated from a synchrotron beamline. The grating structure is well replicated at the Talbot distance. The half period image and the phase shifted images were recorded at fractional Talbot distances. A simulation based on the Fresnel diffraction theory predicts the experimentally observed wave-field propagation well. We believe that high flux broadband x-rays generated from a bending magnet synchrotron beamline can be utilized in hard x-ray lithography utilizing the observed Talbot effect. Our simulation suggests that hard x-ray lithography with a resolution close to 10 nm is possible with the multilayer grating technology. As feature size decreases to 10 nm, it is impractical to place a PR in the shadow near field regime. Therefore utilizing the Talbot effect observed in this experiment is critical. The broad band polychromatic x-ray beam at synchrotron is much more intense than monochromatic beam.