Interferometric testing of optical surfaces at its current limit

Today the production of microlithographic projection lenses drives interferometry to its very limits. Resolutions to a fraction of an Angstrom when measuring surface figure are necessary. Especially EUV lithography needs mirror surface qualities of approximately 0.1 nm RMS, with the consequence that metrology needs accuracies significantly below 1 Angstrom. Repeatability of a measurement is limited by the quality of the interferometric setup itself. Reproducibility is mainly influenced by handling, adjustment and thermal adoption to the testpiece, whereas accuracy is limited by the calibration process and procedure. Carl Zeiss uses a low coherent Fizeau interferometer with multi-fringe evaluation technique (DMI = Direct Measuring Interferometry) to accumulate many phase maps at video rate. Cavities are kept short, air ventilation of the cavity during the measurements is applied. A weak aspheric surface for EUV application (so called ELT2-EUV asphere) was tested with a special interferometric setup using a single lens compensating system. The reproducibility of the measurement came out as 0.07 nm RMS (surface figure). The setup has been calibrated using a spherical calibration surface, whose accuracy was cross checked by two different absolute measurement procedures. We found that both measurements differ by an amount of only 0.15 nm RMS (surface figure). Taking into account the systematic errors of the interferometric setup, the accuracy of the asphere measurement can be estimated to 0.2 nm RMS (surface figure). Further improvements are necessary to drive accuracy to even lower limits and to make test procedures ready for production process.