COMPARISON OF INTERFACE QUALITY IN EVAPORATED AND SPUTTERED Mo/Si MULTILAYERS FOR X-UV OPTICS

M. Jergel¹, V. Holý², E. Majková¹, Š. Luby¹, R. Senderák¹, H.J. Stock³, D. Menke³, U. Kleineberg³, U. Heinzmann³

¹ Institute of Physics, Slovak Academy of Sciences, Dúbravská cesta 9, 842 28 Bratislava Slovakia
² Laboratory of Thin Films and Nanostructures, Faculty of Science, Masaryk University, Kotlárská 2, 611 37 Brno, Czech Republic
³ University of Bielefeld, Faculty of Physics, Postfach 1001 31, D-33501 Bielefeld, Germany

Keywords : multilayer, interface, X-UV optics, X-ray reflectivity, diffuse scattering

Multilayers composed of a heavy refractory metal and a light spacer element operate as Bragg reflectors for ultraviolet and soft X-ray (X-UV) wavelength range. The interface quality plays a crucial role to achieve a desired optical performance. The electron-beam evaporation and sputtering are the most common techniques to prepare multilayer X-UV optical elements. This contribution deals with a comparison of the interface quality and vertical replication of the interface profiles in Mo/Si multilayer mirrors prepared by these two techniques.

The Mo/Si MLs were prepared by the UHV electron beam evaporation and dc magnetron sputtering. The evaporation started with Mo, the deposition rate being 0.01 nm/s. To increase the surface mobility of the adatoms, the Si(100) substrate was heated up to 160o C. The nominal ML period d=8.7 nm was deposited 50 times. The sputtered sample consisted of 50 bilayers, too, which were deposited on a float glass substrate in an Ar plasma with a partial pressure of 0.3 Pa, the ML period being 10.2 nm. The distance between target and substrate was 32 mm and the deposition rates were 0.59 nm/s and 0.38 nm/s for Si and Mo, respectively, Si being the first deposited layer.

The internal structure of the Si layers in both MLs was found to be amorphous by X-ray diffraction while the Mo layers were polycrystalline and strongly textured with the (110) planes parallel to the surface. The interfaces were studied by the X-ray reflectivity and diffuse scattering measurements at grazing incidence on a high-resolution diffractometer equipped with a double-crystal GaAs monochromator using CuK₁ radiation. The reflectivity was evaluated by the Fresnel optical computational code in the coherent scattering approximation and provided basic ML parameters - ML period d_Mo+d_Si, individual layer thicknesses, relative thickness error D, and starting estimate of the rms interface roughness s. The Mo₅Si₃-like interlayers had to be considered in both MLs to obtain satisfactory fits. The diffuse scattering distribution in the reciprocal space was mapped by the detector, sample, and offset scans which were evaluated within the distorted-wave Born approximation (DWBA). The substrate was assumed as an undisturbed system and the whole ML including the interface roughness was considered as a disturbance. This calculation strategy of the DWBA does not include the secondary processes and is equivalent to the semikinematical method, well-known from the scattering theory.

The Gaussian type of the self-correlation function with a correlation length x was supposed to be the same for all interfaces. The cross-correlation function of the interface profiles was described by an attenuation of the self-correlation function according to

where the a parameter is related to the degree of the vertical interface profile correlation (along z) and q_x is the lateral wavector transfer on the corresponding frequency component of the interface roughness. This frequency-dependent decay of the vertical interface correlation follows from the kinetic roughening growth model. Numerical parameters of the simulations are given in the following table.

The most distinct difference is that the parameter a is lower by more than one order of magnitude for the sputtered sample which implies a lower degree of the vertical interface conformality. As our sputtering system at the Ar plasma pressure of 0.3 Pa was below the thermalization threshold, the sputtered adatoms impinged on the surface ballistically. A high energy of the sputtered adatoms, which arrived at the surface at large velocities and migrated for long distances before getting incorporated in an aggregate, together with larger deposition rate resulted into a poorer replication of the interface profiles in the sputtered ML. The surface diffusion length of the sputtered adatoms was estimated to be 5 times larger than that of the evaporated ones and is reflected also in somewhat larger lateral correlation length x.

Except for a, other ML parameters are not affected dramatically by the preparation technique and the reflectivity on the 1st ML Bragg maximum reaches 70% in both MLs. Regarding the intensity, there is obviously no generally superior technique to prepare Mo/Si ML mirrors for X-UV optics. However, the X-rays diffusely scattered on vertically conformal interfaces undergo a constructive interference which results into concentration of the diffuse intensity around ML Bragg points in the reciprocal space. Such an effect is observable also in our results. Therefore, a higher conformality of the interfaces in the evaporated ML may have a rather detrimental effect on contrast if it is used X-UV imaging.