PhD - Leuven | Just now
As VLSI device downscaling nears atomic dimensions, traditional charge-based devices face huge challenges. As a result, the last years have seen the advent of a surge in so-called beyond-CMOS device concepts which are mostly non-charge based. One of these promising directions is the optical-based
plasmonic wave computing paradigm. In that case, the light waves traveling through a metal-insulator-metal waveguide can be exploited for highly parallel wave computing as long as the coherency is maintained. Furthermore, the inherent speed of these waveguides is unparallelled. A successful implementation of this paradigm however also requires an energy-efficient and fast (near-THz range) electronic interface circuit to bridge multiple of such plasmonic waveguides.
Several initial results at KU Leuven ESAT-MICAS clearly indicate to feasibility of using CMOS technology to generate and detect near-THz waves. By exploiting the non-linear behavior of transistors, operation above fmax becomes possible and several examples up to 600GHz have already been demonstrated in 28nm CMOS. This clearly indicates the potential of building such very high frequency circuits at reasonable power consumption. However, these concepts have never been transferred to the domain of plasmonic wave computing.
The aim of this PhD thesis will be threefold:
i) investigate innovative near-THz circuits for realizing the functionality of an ALU processor core
which is complementary to the carry chain function implemented with the plasmonic waveguides,
ii) explore and propose novel circuit extensions which allow to find the sweet spot between above 200 GHz operation speed, which maintaining the
lowest total energy consumption. In this part the PDK for the IMEC A10 Si-CMOS node will be used as basis but also recommendations for DTCO co-optimizing
will be provided.
iii) explore the wave pipelining concept to mitigate the potential difficulties with above 200 GHz clocking and synchronisation
Required background: Electrical Engineering
Type of work: 90% modeling/simulation, 10% literature
Supervisor: Patrick Reynaert
Co-supervisor: Dennis Lin
Daily advisor: Patrick Reynaert
The reference code for this position is 2026-055. Mention this reference code on your application form.