Mueller matrix spectroscopic ellipsometry becomes increasingly important for determining structural parameters of periodic line gratings. Because of the anisotropic character of gratings, the measured Mueller matrix elements are highly azimuthal angle dependent. Measurement results are interpreted by basic principles of diffraction on gratings. The spectral and azimuthal angle dependent intensity changes are correlated to so-called Rayleigh singularities, i.e., wavelengths where the number of diffraction orders changes. The positions of the Rayleigh singularities are calculated analytically and overlapped with measured spectra of two different types of photomasks with transparent and reflecting substrates. For both types of gratings, the Rayleigh singularities reproduce the contours of the spectra. Increasing grating periods result in a shift of these contours to longer wavelengths. Characteristic differences between the two photomasks are explained by the influence of the transmission orders, which are determined by the substrate transparency.

No abstract

Critical dimension and line edge roughness on photomask arrays are determined with Mueller matrix spectroscopic ellipsometry. Arrays with large sinusoidal perturbations are measured for different azimuth angels and compared with simulations based on rigorous coupled wave analysis. Experiment and simulation show that line edge roughness leads to characteristic changes in the different Mueller matrix elements. The influence of line edge roughness is interpreted as an increase of isotropic character of the sample. The changes in the Mueller matrix elements are very similar when the arrays are statistically perturbed with rms roughness values in the nanometer range suggesting that the results on the sinusoidal test structures are also relevant for "real" mask errors. Critical dimension errors and line edge roughness have similar impact on the SE MM measurement. To distinguish between both deviations, a strategy based on the calculation of sensitivities and correlation coefficients for all Mueller matrix elements is shown. The Mueller matrix elements M 13 /M 31 and M 34 /M 43 are the most suitable elements due to their high sensitivities to critical dimension errors and line edge roughness and, at the same time, to a low correlation coefficient between both influences. From the simulated sensitivities, it is estimated that the measurement accuracy has to be in the order of 0.01 and 0.001 for the detection of 1 nm critical dimension error and 1 nm line edge roughness, respectively.

We investigated the potentials, applicability and advantages of spectroscopic ellipsometry (SE) for the characterization of high-end photomasks. The SE measurements were done in the ultraviolet-near infrared (UV-NIR) wavelength range from 300 nm to 980 nm, at angle of incidences (AOI) between 10 and 70° and with a microspot size of 45 x 10 µm² (AOI=70°). The measured and spectra were modeled using the rigorous coupled wave analysis (RCWA) to determine the structural parameters of a periodic array, i.e. the pitch and critical dimension (CD). Two different types of industrial photomasks consisting of line/space structures were evaluated, the reflecting extreme ultraviolet (EUV) and the transmitting opaque MoSi on glass (OMOG) mask. The and spectra of both masks show characteristic differences, which were related to the Rayleigh singularities and the missing transmission diffraction in the EUV mask. In the second part of the paper, a simulation based sensitivity analysis of the Fourier coefficients and is presented, which is used to define the required measurement precision to detect a CD deviation of 1%. This study was done for both mask types to investigate the influence of the stack transmission. It was found that sensitivities to CD variations are comparable for OMOG and EUV masks. For both masks, the highest sensitivities appear close to the Rayleigh singularities and significantly increase at very low AOI. To detect a 1% CD deviation for pitches below 150 nm a measurement precision in the order of 0.01 is required. This measurement precision can be realized with advanced optical hardware. It is concluded that UV-NIR ellipsometry is qualified to characterize photomasks down to the 13 nm technology node in 2020.