Photon Design, a global leader in photonic simulation CAD software, enables rigorous 3D simulations of entire, Thin Film Lithium Niobate (TFLN), Mach-Zehnder Modulators (MZM’s), within a single design environment. Designing fast modulators for high-speed data transmission is central to driving coherent communications, AI, and co-located optics-based processers. TFLN’s strong and near instant response to applied voltage, and its wafer-scale fabrication capability, make it the key material for MZMs and next-generation modulators 

Dr. Dominic Gallagher, CEO of Photon Design, said, “MZM device simulation is critical to many, new and emerging, high-data rate, photonics technologies. Photon Design’s Multi-Topology (MT)-FIMMPROP simulator enables thorough, MZM, 3D simulation, with export to layout by GDS-II. It is unique in enabling engineers to use a single tool for designing MMIs, Y-splitters, and waveguide bends; investigating the waveguide’s response under voltage; building a circuit simulator; and producing a layout tool. Additionally, the MT-FIMMPRP’s electro-optic solver simulates the Pockles effect of the isotropic, TFLN material in MZMs, as it changes refractive properties with voltage, to switch phase.

“Accurately simulating a full MZM within a single design environment is only possible using EigenMode Expansion (EME) for processing, found in our MT-FIMMPROP simulator. EME can perform 3D simulations on a standard laptop, with results back within minutes. Whereas, the more commonly available FDTD-based simulators are barely viable, with equivalent results taking hours, not minutes to process, and often requiring expensive, cloud computing services for the calculations. 

“EME’s simulation efficiency comes from only processing the active areas of the MZM, not the entire bounding box, wasting no resources simulating empty space. EME is unaffected by the MZM’s size, which can often be large in photonics terms, measuring 1600um in some cases. Since MZM design is an iterative process, design and simulation of MZMs using FDTD modelling is barely viable.”