Conference papers

The majority of nanopositioning and nanomeasuring machines (NPMMs) are based on three independent linear movements in a Cartesian coordinate system with a repeatability in the nanometer range. This in combination with the specific nature of sensors and tools limits the addressable part geometries. This article contributes to the enhancement of multiaxial machine structures by the implementation of rotational movements while keeping the precision untouched. A parameter based dynamic evaluation system with quantifiable technological parameters has been set up and employed to identify general solution concepts and adequate substructures. It further on contains data based on comprehensive design catalogues, uncertainty calculations and CAD-model based footprint analysis for specific setups. First evaluations show high potential for sample scanning mode variants considering linear movements of the object in combination with angular movements of the tool, considering a goniometer setup in specific. Based on this, positioning systems for the tool rotation of a NPMM were selected and the positioning properties of different arrangements were determined in test series using autocollimators. General properties of the influence of the arrangement were derived. The arrangement of the substructures which fulfils the previous given requirements is integrated into the NPMM and investigated for long-term stability using a retroreflector as a tool and various laser interferometers. The influence of the additional positioning systems on the existing structure of NPMMs are investigated and solutions for the optimization of the overall system with regard to reproducibility and long-term stability are developed. For this purpose, comprehensive FEA simulations are carried out and structural adjustments are derived via topology optimizations. After all, the knowledge gained is formed into general rules for the verification and optimization of design solutions for multiaxial nanopositioning machines.

Nanometre accuracy and resolution metrology over large areas is becoming more and more a necessity for the progress of precision and especially for nano manufacturing. In recent years, the TU Ilmenau has succeeded in developing the scientific-technical basics of new ultra-high precision, so called nanopositioning and nanomeasuring machines. In further development of the first 25 mm machine, known as NMM-1 from SIOS Meßtechnik GmbH, we have developed and built new machines having measuring ranges of 200 mm x 200 mm x 25 mm at a resolution of 20 pm and enable measuring reproducibility of up to 80 pm. This means a relative resolution of 10 decades. The enormous accuracy is only made possible by the consistent application of error-minimum measurement principles, highly accurate interferometric measurement technology in combination with highly developed measurement signal processing and comprehensive error correction algorithms. The probing of the measurement objects can optionally be carried out with the aid of precision optical, interference-optical, tactile or atomic force sensors. A complex 3D measurement uncertainty model is used for error analysis. The high performance could be demonstrated as an example in step height measurements with a reproducibility of only 73 pm. The achieved resolution of 10-10 also presents new challenges for the frequency stability of the He-Ne lasers used. Here, the approach of direct coupling of the lasers to a phase-stabilized optical frequency comb synchronized with an atomic clock is pursued. The frequency stability is thus limited by the relative stability of the RFreference to better than 4×10-12 (1s).

In principal, optical measurement methods suffer from physical limits related to finite wavelengths and diffraction. In laser focus scanning, vertical resolutions below 1 nm can be achieved while the lateral accuracy is more or less restricted by the diameter of the focused laser beam, i.e. values of half the wavelength can be reached in the best case. We present a model based approach having the potential to show a way out of this limitation. It is based on the rigorous modeling of the complete measurement device including sophisticated ray tracing in combination with Maxwell based modeling of the sample diffraction and scattering providing a simulated signal for an assumed sample profile. Furthermore, the sample profile is parametrized on the basis of a priori information. The simulated signal is then iteratively compared with the measured signal while updating the floating parameters of the model in order to improve the match between the two signals. Eventually, the improved sample profile obtained in this way is considered to represent the real sample profile as soon as a certain goodness of fit is achieved. On the other hand, the profile model has to be changed in case there is no satisfying fit. In this way, the lateral accuracy can be increased considerably. Edge detection errors below a few tens nanometer or even below 10 nm become possible while measuring with visible light. This is demonstrated by first comparisons of modeled and measured signals and validation by alternative metrology techniques.