Problem of roughness detection for supersmooth surfaces

The requirements for multilayered x-ray elements for diffraction quality imaging optics (EUV - lithography, x-ray microscopy) achieves 0.2-0.3 nm roughness in spatial frequency range 10(-3) - 10(3) mcm(-1); it's also true for the substrates. Although, there are plenty of publications on studying a surface, when it comes to angstrom-quality substrates there is still a problem. In some cases we observe, standard methods like x-ray diffuse scattering (XRDS), atomic force microscopy (AFM) and optical interferometric microscopy (OIM) give notably different results in surface characterization. The goal of the attestation procedure is choosing the sample for sputtering a multilayer coating with better reflection properties, that's why it's important to understand the physical causes of the difference and get reliable information about the surface. In this work we discuss the limitation for aforesaid standard methods. OIM is seems to be inapplicable for supersmooth surface investigation because of applying references. It's also shown, that examination substrates with damaged layers in the volume (caused, for example, by ion-beam etching) by XRDS can lead to incorrect results. Imaging systems are composed by nonplanar optical elements with radiuses from 10 mm to 1 meter. That makes impossible using hard x-rays and also limited AFM applicability to high frequencies. Therefore, we propose the diffuse scattering of soft x-rays as an alternative approach. We also describe a new reflectometer, based on soft x-ray and visible light diffuse scattering, which can be used for surface investigation in middle and high spatial frequency ranges for both plane or curved substrates.