This paper presents a systematic comparison of the relationship between transmission line characteristic impedance and Q-factor of CPW, slow-wave CPW, microstrip, and slow-wave microstrip in the same CMOS back-end-of-line process. It is found that the characteristic impedance for optimal Q-factor depends on the ground-to-ground spacing of the slow-wave transmission line. Although the media are shown to be similar from a mode of propagation point of view, the 60-GHz optimal Q-factor for slow-wave transmission lines is achieved when the characteristic impedance is ≈23 Ω for slow-wave CPWs and ≈43 Ω for slow-wave microstrip lines, with Q-factor increasing for wider ground plane gaps. Moreover, it is shown that slow-wave CPW is found to have a 12% higher optimal Q-factor than slow-wave microstrip for a similar chip area. The data presented here may be used in selecting Z0 values for S-MS and S-CPW passives in CMOS that maximize transmission line Q-factors.

In this paper, a slow-wave transmission line implemented in coplanar waveguide technology, based on simultaneous inductive and capacitive loading, is presented for the first time. The shunt capacitors are achieved by periodically etching transverse strips in the back substrate side, connected to the central strip through metallic vias. The series inductors are implemented by etching rectangular slots in the ground plane. The effect of these reactive elements is an enhancement of the effective shunt capacitance and series inductance of the line, leading to a significant reduction of the phase velocity (slow-wave effect). Consequently, the guided wavelength is also reduced, and these lines can be applied to the miniaturization of microwave components. Moreover, due to periodicity, these artificial lines exhibit stop bands (Bragg effect) useful for spurious or harmonic suppression. A compact harmonic suppressed power splitter, based on a slow wave 35.35 Ω impedance inverter, has been designed and fabricated in order to demonstrate the potential of the proposed approach. The length of the inverter is 48% the length of the conventional counterpart, and measured power splitting at the first (3f0) and second (5f0) harmonic frequencies is rejected more than 49 and 23 dB, respectively.