Silicon nitride programmable photonic processor with folded heaters

Single-chip silicon photonic engine for analog optical and microwave signals processing

We present a photonic engine that processes both optical and microwave signals, and can convert signals between the two domains. Our photonic chip, fabricated in IMEC's iSiPP50G silicon photonics process, is capable of both generation and detection of analog electrical and optical signals, and can program user-defined filter responses in both domains. This single chip integrates all essential photonic integrated components like modulators, optical filters, and photodetectors, as well as tunable lasers enabled by transfer-printed indium phosphide optical amplifiers. This makes it possible to operate the chip as a black-box microwave photonics processor, where the user can process high-frequency microwave signals without being exposed to inner optical operation of the chip. The system's configuration is locally programmed through thermo-optic phase shifters and monitored by photodetectors, and can select any combination of optical or microwave inputs and outputs. We construct multiple systems with this engine to demonstrate its capabilities for different RF and optical signal processing functions, including optical and RF signal generation and filtering. This represents a key step towards compact and affordable microwave photonic systems that can enable higher-speed wireless communication networks and low-cost microwave sensing applications.

Low-voltage-tunable electromechanical photonic directional coupler in silicon nitride for telecom wavelengths

Silicon nitride is a low-loss photonic integrated circuit (PIC) platform. However, silicon nitride also shows small nonlinear optical properties and is dielectric, which makes the implementation of programmability challenging. Typically, the thermo-optic effect is used for this, but modulators based on this effect are often slow and cross talk-limited. Here, we present a different approach to programmability in silicon nitride photonics. Micro-electromechanical elements are added to a photonic directional coupler, forming two H-shaped structures. The coupling can be changed by applying a voltage to electrodes placed onto the H-structure, which are then attracted by an electrostatic force. These suspended directional couplers show an insertion loss of 0.67 dB and demonstrate switching with 1.1±0.1 µs rise times, representing a valuable addition to the thermal photonic modulators in silicon nitride technology that offer higher modulation speeds while keeping a comparable insertion loss.

Low-loss fiber-to-chip edge coupler for silicon nitride integrated circuits

Silicon nitride (SiN) integrated optical waveguides have found a wide range of applications due to their low loss, broad wavelength transmission band and high nonlinearity. However, the large mode mismatch between the single-mode fiber and the SiN waveguide creates a challenge of fiber coupling to these waveguides. Here, we propose a coupling approach between fiber and SiN waveguides by utilizing the high-index doped silica glass (HDSG) waveguide as the intermediary to smooth out the mode transition. We achieved fiber-to-SiN waveguide coupling efficiency of lower than 0.8 dB/facet across the full C and L bands with high fabrication and alignment tolerances.

Ultraflat bandpass, high extinction, and tunable silicon photonic filters

We report on the design and the demonstration of silicon photonic ultraflat bandpass filters with low insertion loss and high out-of-band rejection for an operation near the 1550 nm wavelength band. These filters are based on cascading low (2^nd) order Ring-Assisted Mach-Zehnder Interferometer (RAMZI) filter stages. The cascade design enables high out-of-band rejection while keeping the unit cells of each stage low order to be more tolerant to fabrication imperfections. The characterization of filters shows an insertion loss of ∼1 dB, an in-band ripple of <0.1 dB, an out-of-band rejection of >50 dB for a filter 3-dB bandwidth of ∼1.1 nm, and tunable up to ∼6 nm. We also investigate the filter's spur-free dynamic range at high input optical powers, which is important for RF photonics applications, and quantify a dynamic range of >60 dB for a laser power as high as ∼11.6 mW sent to the filter. Such integrated filters are promising for applications in pump wavelength rejection in four-wave mixing photon pair generation, and in RF antenna remoting where multiple RF signals are carried on different coarse wavelengths to be separated.

Two-photon polymerization simulation and fabrication of 3D microprinted suspended waveguides for on-chip optical interconnects

Quantum and neuromorphic computational platforms in integrated photonic circuits require next-generation optical functionalities. Often, increasingly complex on-chip light-routing that allow superpositions not attainable by planar technologies are paramount e.g. for artificial neural networks. Versatile 3D waveguides are achievable via two-photon polymerization (TPP)-based microprinting. Here, a 3D morphology prediction tool which considers experimental TPP parameters, is presented, enabling on-chip 3D waveguide performance simulations. The simulations allow reducing the cost-intensive systematic experimental optimization process. Fabricated 3D waveguides show optical transmission properties in agreement with simulations, demonstrating that the developed morphology prediction methodology is beneficial for the development of versatile on-chip and potentially inter-chip photonic interconnect technology.

Photonic Integrated Reconfigurable Linear Processors as Neural Network Accelerators

Reconfigurable linear optical processors can be used to perform linear transformations and are instrumental in effectively computing matrix–vector multiplications required in each neural network layer. In this paper, we characterize and compare two thermally tuned photonic integrated processors realized in silicon-on-insulator and silicon nitride platforms suited for extracting feature maps in convolutional neural networks. The reduction in bit resolution when crossing the processor is mainly due to optical losses, in the range 2.3–3.3 for the silicon-on-insulator chip and in the range 1.3–2.4 for the silicon nitride chip. However, the lower extinction ratio of Mach–Zehnder elements in the latter platform limits their expressivity (i.e., the capacity to implement any transformation) to 75%, compared to 97% of the former. Finally, the silicon-on-insulator processor outperforms the silicon nitride one in terms of footprint and energy efficiency.