Comparison of TRL, 8-, 10-, and 16-term on-wafer calibration methods for accurate SiGe HBT characterization up to 330 GHz

Abstract This paper discusses the challenges involved in characterizing high-frequency SiGe heterojunction bipolar transistors. Process engineers continually strive to enhance device performance by shrinking dimensions and minimizing parasitic capacitances; however, these capacitances – on the order of only a few femtofarads – are particularly susceptible to measurement errors. At millimeter-wave and sub-THz frequencies, the finite size and electromagnetic-field distribution of RF probes may lead to residual substrate coupling and probe-to-probe crosstalk, especially when the device under test and pad dimensions are strongly scaled. To investigate these issues, we apply on-wafer 16-term calibration techniques using accurately designed calibration standards. Starting from the full 16-term error-correction model and progressively simplifying it in a step-by-step manner, we compare the performance of 16-, 10-, and 8-term calibrations and, finally, traditional TRL calibration up to 330 GHz $330\,\text{GHz}$ 330 GHz . We verify the consistency of these approaches, analyze their influence on the extraction of passive elements, and evaluate how each calibration scheme affects key transistor performance metrics, such as the maximum oscillation frequency ( fmax $f_{\max}$ f max ).