Abstract

Fresnel zone plates with minimum zone widths below 70 nm have been fabricated for the joint high resolution x-ray microscopy effort with Stony Brook University at Brookhaven National Laboratory. The x-ray scanning microscope requires zone plates as objective lenses with the central zones blocked by thick absorber material (apodized) and also the outermost zone surrounded by a wide ring of thick absorber. This is achieved by a double development/double plating technique.¹ A 120-nm thick Si₃N₄ membrane substrate with 30 nm of gold evaporated as a plating base and coated with ≃200 nm of PMMA resist is exposed in a high resolution vector scan electron beam lithography system at a dose of 230 μC/cm² for the zones and 500 μC/cm² for the thick absorber areas at a beam voltage of 25 kV. After partial development the thick absorber areas can be electroplated with 0.4–0.8 μm of gold. After final development and oxygen-plasma descumming the zones can be plated with 140 nm of gold. Monitoring the development status of the fine features of the zone plates (200 nm to <70 nm) is an important issue during the fabrication procedure. Although the optical microscope cannot resolve these features, using polarized light and crossed analyzers, a distinct change of the reflected intensity can be observed when the developed lines approach the substrate. This can be explained in terms of form birefringence caused by the submicron discontinuities on the surface, which rotates the vector of polarization and therefore increases the light intensity passing the analyzer. For the performance of the zone plates (resolution, efficiency), correct spacing, circularity, and width of the zones is of importance. Absorber thickness, zone width, and local pattern integrity was determined by scanning electron microscopy. Double exposure of scanning electron micrographs with translation or rotation of the plates between exposures, resulted in moire patterns, which allow estimates of the zone placement accuracy and circularity. The plates showed ellipticities ranging from ≃1 % to values not measurable with this method (better than 0.3%). Their performance is reported in an accompanying paper.²

© 1986 Optical Society of America