Project C05 - Efficient On-Chip Antennas for THz Systems

Principal Investigators: Dr. Andreas Rennings, Prof. Dr. Daniel Erni, UDE ATE

Within the initiated CRC several RF frontends are to be realized either as SiGe BiCMOS chips or by hetero-integrated InP devices embedded in a silicon host. The operation frequencies span a wide range from 150 GHz up to 1.5 THz in the first phase of the CRC. The highly necessary research in this electromagnetic spectrum concerns the transmitter with the limited radiated power that can be generated and also the receiver side with the limited sensitivity that can be achieved. In both cases these limitations are attributable to the inefficient and/or narrow-band antennas. Another challenging issue, especially for real THz transceiver chips, are substrate waves, which usually disturb the controlled, broadsidedirected radiation of the chip.

Thus, in MARIE C05 we will carry out research on innovative antenna solutions that are tailored to the different RF frontends (stack-ups in the chips) of our partners working on C02, C03, C04, C06, C07, C08 and C09. The necessary properties of the on-chip antennas depend on the project, but generally a high-gain broadside beam, basically related to high radiation efficiency, and a broadband operation of 20% and more are desired. In C04, antennas with adjustable polarization, including circular polarization, and in C03 good MIMO performance is required. Planar designs, including the corresponding lenses, are preferable. Due to the project-specific antenna properties, we need to investigate several novel on-chip antenna approaches. One of them is based on Electromagnetic Band Gap (EBG) structures. These periodic structures are utilized to design meta-surfaces with a High Impedance Surface (HIS) and tailored stop-band properties, yielding improved radiation efficiency and the suppression of substrate waves. Alternatively, the two mentioned functionalities can also be obtained by another periodic structure: an artificial dielectric with high effective permittivity that is additionally an-isotropic. Such a meta-material can be obtained by non-resonant metal inclusions on several layers of the Back End of Line (BEoL) of the chip. The high permittivity “attracts” the EM wave into the broadside direction, whereas the anisotropy prevents the unwanted substrate waves based on the lower effective permittivity for off-broadside propagation directions.

Another approach will be based on the Localized Back side Etching (LBE) technique that is offered by the chip manufacturers. We will utilize LBE to open up this layer for the reactive near-field of the antenna yielding an improved bandwidth. We try to find tailored LBE pattern, i.e., the lossy Si is only removed where the corresponding antenna excites strong electric near-fields, again yielding improved efficiencies.

An additional low-dielectric superstrate (BCB or glass) on top of the thin BEoL enables efficient and broadband antennas, again due to more space for the reactive near-field. In this case the antenna on top of the superstrate can be electromagnetically coupled via a coupling structure located in the multilayered BEoL. Alternatively to the layered superstrate we will investigate Dielectric Resonator Antennas (DRAs) of different shape, an approach that might be best suited for THz frequencies due to the missing ohmic losses. The superstrate and DRA concept have a quite freely oscillating resonator in common. Thus, the Characteristic Mode Analyses (CMA) might be applied to find modes that radiate in broadside direction in an efficient manner. The combination of several of these modes might be utilized for the MIMO radar approach of C03, but it is also interesting for the “multi-color” transceiver of C04, in the sense that different colors (frequencies) may correspond to different characteristic modes.

Finally, we cooperate with two MARIE projects in which our antennas are directly connected to active devices (resonant tunnelling diode in C02; and photo diode in C07). For these two cases we will choose antenna structures that organically fulfill the conjugate matching condition for optimal power transfer. For these two cases, the “tailoring” of the antenna will be extreme, due to the required, special input impedance of the antenna far away from the usual 50 Ohms.

The combination of high radiation efficiency and broadband operation for an on-chip environment is a very challenging task from the antenna engineering point of view, because the involved high dielectric, lossy, and sometimes electrically thin materials (for the sub-mm wave case) are hardly well suited for radiating structures. Thus, non-standard, innovative and tailored antenna solutions must be explored. Metamaterial-based concepts, with their extended design possibilities, can offer innovative solutions to some of the issues. The remaining ones will be solved using the other techniques we put forward here in this proposal.

Our part within the MARIE consortium is the research on project-specific, cutting-edge antenna solutions for our MARIE partners, such that their sub-mm wave and THz hardware performs as best possible.