Katsura Otaki

Development of the point diffraction interferometer for extreme ultraviolet lithography: Design, fabrication, and evaluation

A point diffraction interferometer (PDI) for extreme ultraviolet lithography (EUVL) aspheric mirror measurement has been developed. In order to realize an accuracy of 0.1 nm rms, various optical error factors have been numerically analyzed and the maximum tolerable error has been determined. From the error estimation results, the optimal pinhole diameter has been determined as 0.5 μm. In a PDI, air turbulence reduces the precision and accuracy because of the long optical path. In order to avoid this problem, the apparatus is filled with helium gas, which has a smaller refractive index than that of air. By using this apparatus, precision of 0.03–0.04 nm rms and a system error of 0.10 (0.16) nm rms have been obtained for a spheric mirror with numerical aperture (NA) 0.08 (0.15). In aspheric mirror measurement, an accuracy of 0.74 (1.18) nm rms for NA 0.08 (0.15) has been obtained. The accuracy becomes 0.34 (0.97) nm rms for NA 0.08 (0.15) with 36-term Zernike polynomial fitting.

Impact of the EUV mask phase response on the asymmetry of Bossung curves as predicted by rigorous EUV mask simulations

EUV masks generally mandate rigorous scattering models as the thickness to wavelength ratio is typically on the order of 30:1. In addition, boundary conditions at the absorber/air interface lead to non-zero electric fields even in the absorber region while the phase itself generally experiences a complex cross mask behavior that is pattern type as well as size dependent. Using two different types of rigorous simulators that are based on the differential method (LithoLand) and the time-domain finite-element method (EMFLEX) we explore and quantify the subtleties associated with the EUV mask when compared to a thin mask obeying the Kirchoff approximation. Both rigorous simulators predict that 30 nm isolated pattern, especially for NA >0.20 experience a focus shift of about 50 nm relative to best focus for the thin mask case. This effect occurs even when the illumination is normally incident to the mask. Furthermore, the Bossung curve for isolated patterns show an asymmetry through focus that is absent for the thin mask case and is sensitive to the partial coherence condition of the illuminator. Nested features seem to be far more immune to this unexpected anomaly. The origin of the predicated focus shift and asymmetry is explained through the complex phase behavior for the EUV mask which is fundamentally different from the thin mask case.

Accuracy evaluation of the point diffraction interferometer for extreme ultraviolet lithography aspheric mirror

We evaluated the accuracy of the point diffraction interferometer, which has been completed at the Atsugi Research Center’s Association of Super-Advanced Electronics Technologies for the precise measurement of extreme ultraviolet lithography optics. To evaluate the absolute accuracy, more precision is required. A pinhole with 0.5 μm diameter was used to generate a near completely spherical reference wave front, and helium gas filled the inside of the chamber to suppress air turbulence. With this apparatus, precision of 0.04 nm root-mean-square (rms) was achieved. Absolute accuracy was evaluated from the measurement of mirror rotation and displacement and an absolute accuracy of 0.17 nm rms was obtained for a spherical mirror with numerical aperture of 0.145. Absolute accuracy was improved to 0.11 nm rms by limiting the numerical aperture to 0.08.

Angular spectrum calculations for arbitrary focal length with a scaled convolution

Nyquist sampling theorem in an image calculation with angular spectrum method restricts a propagation distance and a focal length of a lens. In order to avoid these restrictions, we studied suitable expressions for the image computations depending on their conditions. Additionally, a lateral scale in an observation plane can be magnified freely by using a scaled convolution in each expression.

Simulation of Multilayer Defects in Extreme Ultraviolet Masks

We have employed a scalar simulation based on Fresnel formulas to predict approximately how strongly a multilayer defect in a reflective mask affects the aerial image in extreme ultraviolet lithography. This method enables us to obtain the field reflected from a defective mask, using considerably fewer computational resources than those required for a rigorous electromagnetic simulation. This method was applied to two-dimensional masks with multilayer defects modeled by a simple structure in which the coverage profile was identical throughout the layers. For such defects, we confirmed the validity of the scalar simulation by comparing the normalized peak intensity to that obtained from an electromagnetic simulation. The dependence of the printability of a defect on its size and position is also discussed.

EUV wavefront measurement of six-mirror optics using EWMS

The wavefront measurements have been performed with the EUV Wavefront Metrology System (EWMS) for the first time using a prototype projection optic as a test optic. The wavefronts of the test optic was measured at the five positions in the exposure field with the Digital Talbot Interferometer (DTI). The RMS magnitude of the wavefront errors ranged from 0.71 λ (9.58 nm) to 1.67 λ (22.75 nm). The results obtained with the DTI were compared to those with the Cross Grating Lateral Shearing Interferometer (CGLSI). As a result of a repeatability assessment, it was found that the EWMS can stably measure the wavefronts of the test optic. Additionally, unwrapping of the phase map was found to be related to the precision of the measurement.

Rigorous wavefront analysis of the visible-light point diffraction interferometer for EUVL

In visible-light point diffraction interferometer (PDI), in order to achieve measurement error <0.1 - 0.2 nm rms, wavefront irregularity from the pinhole must be supressed as 0.05 - 0.1 nm rms in designing. It is so difficult to execute such high accurate (10-4λ) simulation because the numerical electromagnetic simulation shows slow convergence in the visible-region. We discussed this problem by using 2D-model and found simulation conditions to obtain significant results. By using the simulator, several kind of systematic erros have been analyzed and optimized.

Advanced point diffraction interferometer for EUV aspherical mirrors

An advanced point diffraction interferometer for measuring EUV aspherical mirrors with high accuracy has been developed. It is designed for measuring various EUV mirrors with high accuracy and high precision. It can measure the surface figure of all mirrors that will be used in high numerical aperture systems. Using this interferometer, 0.1nm rms precision and 0.2nm rms accuracy are expected.

Recent Progress of EUV Wavefront Metrology in EUVA

The recent experimental results of EUV wavefront metrology in EUVA are reported. EUV Experimental Interferometer (EEI) was built at the NewSUBARU synchrotron facility of University of Hyogo to develop the most suitable wavefront measuring method for EUV projection optics. The result is to be reflected on EWMS (EUV Wavefront Metrology System) that measures wavefront aberrations of a six-aspherical mirror projection optics of NA0.25, of a mass-production EUV lithography tool. The experimental results of Point Diffraction Interferometer (PDI) and Lateral Shearing Interferometer (LSI) are shown and the error factors and the sensitivity of astigmatism measurements of these methods are discussed. Furthermore, for reducing these kinds of errors, another type of shearing interferometer called DTI (Digital Talbot interferometer) is newly introduced.

Polarization Effect on Signal from Optical ROM Using Solid Immersion Lens

In a high-numerical-aperture optical head using a solid immersion lens (SIL), it is anticipated that the polarization of the diffracted beam is disturbed, this affects the signals. To study this problem, a rigorous simulator based on the three-dimensional vector diffraction theory has been developed. Polarization disturbance for various NAs has been estimated and it has been clarified that the polarization is disturbed not only in the SIL head but also in the conventional high-NA head whose disk substrate is not covered with protective glass. The influence of polarization disturbance on the readout signals has been determined and discussed for the various polarizations.