Terahertzmetrologie voor on-wafer S-parametermetingen

Metrology plays a significant role in the development of electronic circuits, not only by verifying their good behavior after fabrication, but by intervening much earlier in the development phase. In fact, the accuracy of models for individual components, such as transistors, diodes, or even just simple transmission lines, has a direct consequence on the number of iterations needed to achieve the required performance. This is where on-wafer measurements play an essential role by enabling the definition of scattering parameters directly in printed transmission lines, significantly improving accuracy over measurements defined at connector level. In on-wafer measurements, signals are guided along a small probe that is put directly in contact with the fabricated chip. A separate calibration substrate is usually sufficient to get accurate measurements at frequencies where the wavelength is well below the dimensions of the probes, and the calibration plane can easily be located at the probe tips level. At higher signal frequencies, however, when the wavelength approaches the dimensions of the probe tips, the impact of a change in substrate material or the geometry of the landing pads on the probe behavior becomes too significant, introducing a substantial error in the determined S-parameters. This explains the need for custom calibration unique to each RF integrated circuit IC technology. The dissertation provides new insights into high-frequency on-wafer S-parameters measurements from four distinct perspectives: measurement methodology, calibration kit development, calibration algorithm, and uncertainty quantification. To support this investigation, two sub-THz calibration substrates were designed, fabricated, and analyzed, each implemented using a different technology. The first, manufactured by a university, was based on simple coplanar waveguide transmission lines and was measured from DC to 500 GHz. This study revealed the impact of undesirable effects like probe contact repeatability and slot-line mode propagation. The second calibration kit was implemented in a grounded multi-layered substrate using grounded coplanar waveguides. Several strategies were employed to mitigate the impact of parasitic modes on the measurements: the transition was modified by adding backside vias to isolate the transition from surface modes, ground straps were added to the transmission lines to avoid higher-order slot-line modes, and surface-wave absorbers were implemented under the probes to reduce the impact of neighboring structures. The research also validates a systematic design approach for on-wafer calibration kits that is becoming an industry standard. Primarily, the methodology relies on verifying the calibration kit's performance by applying calibration algorithms to results obtained from full-wave electromagnetic simulations using realistic probe models. This work also extends the capabilities of the multiline Thru-Reflect-Line method by including an additional systematic effect within the calibration: the probe position on the contacting pads. This not only permitted us to analyze the impact of probe misplacements during on-wafer measurements, but we were even able to correct for such misplacements, opening the path for compensated calibration algorithms. Finally, an uncertainty propagation framework was built around this new calibration algorithm to better analyze the impact of different uncertainty sources on on-wafer S-parameters measurements.