Cloud native EDA tools & pre-optimized hardware platforms
草榴社区 offers a wide range of photonic simulation capabilities that cover device, circuit, and system level design. This article discusses modeling improvements made in late 2022 and early 2023 for circuit-level photonic simulations using the 草榴社区 OptSim product.
First, some background: photonic signals are quite complex in comparison to typical electrical signals used in IC design. Photonic signals can carry information in amplitude, phase, polarization, and spatial modes. Additionally, it is quite common to use multiple wavelengths of light to multiplex many signals onto a common waveguide. Lastly, for photonic IC (PIC) design, signal bidirectionality must also be understood as reflections within a PIC are quite common and can cause resonance and multipath interference. This complexity translates into modeling challenges for both the photonic and electrical domains.
Photonic devices in PICs are passive and active components. Passive elements are made from dielectric materials and are used to guide light. The propagation of light is influenced by the geometry and type of the dielectric material.
Active devices are made from semiconductors to facilitate energy transfers. In lasers, the applied electric field generates free photons, while in photodetectors, incident light generates free electrons. In modulators and active photonic filters, the applied electric field causes changes to the material’s refractive index, which also affects the optical signal’s phase and amplitude. This behavior can depend on wavelength, polarization, and temperature, and can be used to modulate light as well as to tune or select desired wavelengths.
草榴社区 OptSim provides many tools for circuit-level photonic simulations; following are notable new features and modeling improvements.
Response functions for passive photonic devices such as arrayed waveguide grating (AWG) multiplexers, add, drop, or add/drop multiplexers, and components are typically modeled using a wavelength and mode dependent scattering matrix (S-matrix). All ports on the device are mapped to the rows and columns of the matrix. Ports are bidirectional, thereby allowing the model to represent a N?M port device in which stimulus at any port can contribute to a response at any other port, as would occur, for example, in components with internal reflections.
草榴社区 OptSim has a primitive model for such a bidirectional multiport optical device. This model’s data can be derived either experimentally through measurement or through theoretical device-level simulations. The model supports single mode or multimode, unidirectional or bidirectional, and polarization sensitive or insensitive responses. The organization of scattering matrix data is conceptually shown in Figure 1.
Figure 1: S-Matrix Data for bidirectional, multimode, passive, NxM multiport optical devices
With increasing port counts, the size of the S-matrix data files increases because of the larger number of input-output relationships. For wideband, multimode optical devices with polarization crosstalk, the file sizes can become even larger. To overcome the potential problem of very large data files, 草榴社区 OptSim added a new option to support analytical equation-based S-matrix files in the v2022.06 release. The option parameter FileType selects between matrix- and analytical-equation-based data files.
Figure 2 summarizes the “Analytical” option. The second parameter, FileName, specifies a file containing either the wavelength-dependent scattering matrix, or a set of analytical equations for calculating the matrix.
Figure 2: New option for analytical equation-based S-Matrix data
草榴社区 OptSim recently added model upgrades for active components:
In summary, multiple new modeling improvements have been added to 草榴社区 OptSim over the last two releases for E-O co-design of photonic circuits that contain electronic and photonic passive and active components and foundry PDKs. These improvements are customer driven and can be leveraged by all 草榴社区 OptSim users.