Entwicklung von modellbasierten Methoden zur Bestimmung von Materialeigenschaften im THz-Bereich

The determination of the specific material parameters for a sample-substance is the foundation of many scientific, industrial and engineering techniques used for material analysis and recognition. Based on these characteristic features, the comparison and classification of substances can be performed, as well as the derivation of the underlying physical mechanisms and further material properties. The extraction of these material parameters from spectroscopic investigations in the terahertz (THz)-range provides several advantages. Many substances relevant to material analysis display unique absorption characteristics in this frequency-range, which lies between the sections of infrared and microwave radiation in the electromagnetic spectrum. Furthermore, THz-radiation is weakly absorbed by typical packaging materials like textiles, plastics or paper, which are mostly opaque in the optical range of the spectrum. This enables the analysis of substances encapsulated in those materials and of respective composite systems, without relying on measurement techniques that employ ionizing radiation (e.g. x-rays). THz time-domain spectroscopy is one of the most common measurement methods that provides the spectral data from which the material parameters, such as the complex refractive index or complex conductivity, can be extracted. Those systems provide amplitude and phase information, unlike techniques based on power detection that only offer intensities, and at the same time enable broadband measurements.
With the continuous development of novel materials and composites, improved fabrication techniques and the steady evolution of THz-systems, new fields of application for the investigation of material parameters in the THz-range frequently arise. At the same time, it becomes apparent that many classical state of the art methods used for handling and extracting material parameters from the spectral measurement are reaching their limitations. They operate increasingly erroneous and can therefore not be applied unconditionally anymore.
Four of these scenarios are discussed in this work. The existing methods are evaluated in the context of the particular application and the reasons for their restrictions are determined. Based on these findings new and enhanced methods for the determination and processing of material parameters are developed, validated and assessed.
For the first application case it is shown, that the designed recursive propagation model allows the description and analysis of multilayered sample systems of arbitrary thicknesses. In contrast to this, the classical approaches fail in instances of solely thin layers or combinations of thin and thick layers, since they cannot correctly represent the multiple internal reflection inside the layered structures.
In the second case it is demonstrated, that the modelling of material parameters is not limited to predict the transmission properties of different material samples. Using the introduced technique, they can also be utilized to determine the system requirements that are necessary to analyze such samples.
The evaluation of the conductivity of thin layers, e.g. of graphene, is the third application scenario. Here, the utilization of the developed expanded thin-film approximation avoids the systematic errors that are introduced into the calculated sheet conductivities by the typical form of the approximation. Consequently, the material parameters derivable from this sheet conductivity, like the relaxation time and the chemical potential, can be determined without the effects of these errors as well.
In the fourth case, the determination of material parameters from samples with strong absorption characteristics by model-based methods is demonstrated. It is shown that material parameters with a three orders of magnitude lower mean square error, compared to the respective frequency-domain approach and the classical concept without a material-model, can be achieved by the utilization of the developed material-model-based time-domain method. Furthermore, it becomes apparent that this robustness of the time-domain method could be maintained, even if wide segments of the absorption features lie bellow the noise-floor.
The developed methods thereby allow overcoming the various existing limitations of the standard approaches used for the determination and evaluation of material parameters in the THz-range. Hence, the investigation of sample scenarios that were previously not accessible becomes feasible. Moreover, future fields of application of the THz-material analysis can thereby also draw on, and be built upon a more versatile set of methods and procedures.