Performance Analysis of Materials for Plasmonic Computing

Plasmonic metal-insulator-metal (MIM) slot waveguide (WG), a promising building block of the plasmonic integrated circuits, has been heavily utilized to explore and demonstrate devices like logic gates, modulators, power splitters, frequency filters, and multiplexers. For the MIM slot WG, the choice of material is critical since it primarily controls the degree of signal attenuation and the extent of confinement of the supported modes. A trade-off always exists between the signal attenuation and mode confinement in plasmonic WGs that define the energy consumption and integration density of the components, respectively. Therefore, a systematic analysis of the possible material combinations is essential for the MIM WG to optimize the performance of the plasmonic devices. This work presents a comprehensive numerical analysis of the possible material combinations of the plasmonic MIM WG for computation purpose. To demonstrate the material effect on computing related context, the analysis incorporates plasmonic MIM interconnect and MIM majority logic gate as use cases. The potential material combinations are initially selected from material quality factor and CMOS compatibility. For each selected material combinations, the confinement factor, propagation length, and coupling length are calculated for the WG. Also, using the coupling length and power transfer data between two nearby WGs, the WG pitch is optimized considering the trade-off between crosstalk and footprint/integration density of the devices. The manipulated plasmon signal in the MIM WG needs to be converted into current/voltage using a detector for readout purposes. Therefore, the effect of material selection on plasmon detection efficiency is next studied by introducing a Ge-based plasmonic metal-semiconductor-metal (MSM) WG as a plasmon detector. For each material case, the MIM WG width is optimized to match the characteristics impedance of the two WG sections resulting in maximum power transmission. Based on various trade-offs, a system-level holistic performance metric, Emin/bit or minimum required energy to detect a single bit sent from the MIM interconnect/logic gate, is introduced, and compared for all the material combinations. This holistic metric encompasses the effect of signal attenuation, mode confinement, WG coupling efficiency, plasmon detection efficiency, and detection bandwidth. Moreover, signal to noise ratio (SNR) and footprint of the devices are calculated. Lastly, considering the trade-off between Emin/bit, SNR and footprint, the optimized material combination is identified. The study shows that Al and Cu are promising metals for MIM WGs for computing. Regarding dielectrics, SiO2 offers a low Emin/bit value and large device area, whereas a high-k dielectric material like Si offers a more compact layout at the cost of a higher energy per bit. It is found that Cu-TiO2-Cu and Al-Al2O3-Al are the two optimum material combinations for MIM WG for plasmonic computing circuits.