New wavelength domains have become accessible for photonic integrated circuits (PICs) with the development of silicon nitride PICs. In particular, the visible and near-infrared wavelength range is of interest for a range of sensing and communication applications. The integration of energy-efficient III-V lasers, such as vertical-cavity surface-emitting lasers (VCSELs), is important for expanding the application portfolio of such PICs. However, most of the demonstrated integration approaches are not easily scalable towards low-cost and large-volume production. In this work, we demonstrate the micro-transfer-printing of bottom-emitting VCSELs on silicon nitride PICs as a path to achieve this. The demonstrated 850 nm lasers show waveguide-coupled powers exceeding 100 µW, with sub-mA lasing thresholds and mW-level power consumption. A single-mode laser with a side-mode suppression ratio over 45 dB and a tuning range of 5 nm is demonstrated. Combining micro-transfer-printing integration with the extended-cavity VCSEL design developed in this work provides the silicon nitride PIC industry with a great tool to integrate energy-efficient VCSELs onto silicon nitride PICs.
Optical interconnects (OIs) based on vertical-cavity surface-emitting lasers (VCSELs) are the main workhorse within data centers, supercomputers, and even vehicles, providing low-cost, high-rate connectivity. VCSELs must operate under extremely harsh and time-varying conditions, thus requiring adaptive and flexible designs of the communication chain. Such designs can be built based on mathematical models (model-based design) or learned from data (machine learning (ML) based design). Various ML techniques have recently come to the forefront, replacing individual components in the transmitters and receivers with deep neural networks. Beyond such component-wise learning, end-to-end (E2E) autoencoder approaches can reach the ultimate performance through co-optimizing entire parameterized transmitters and receivers. This tutorial paper aims to provide an overview of ML for VCSEL-based OIs, with a focus on E2E approaches, dealing specifically with the unique challenges facing VCSELs, such as the wide temperature variations and complex models.
Flat metaoptics components are looking to replace classical optics elements and could lead to extremely compact biophotonics devices if integrated with on-chip light sources and detectors. However, using metasurfaces to shape light into wide angular range wavefronts with high efficiency, as is typically required in high-contrast microscopy applications, remains a challenge. Here we demonstrate curved GaAs metagratings integrated on vertical-cavity surface-emitting lasers (VCSELs) that enable on-chip illumination in total internal reflection and dark field microscopy. Based on an unconventional design that circumvents the aspect ratio dependent etching problems in monolithic integration, we demonstrate off-axis emission centred at 60° in air and 63° in glass with > 90% and > 70% relative deflection efficiency, respectively. The resulting laser beam is collimated out-of-plane but maintains Gaussian divergence in-plane, resulting in a long and narrow illumination area. We show that metagrating-integrated VCSELs of different kinds can be combined to enable rapid switching between dark-field and total internal reflection illumination. Our approach provides a versatile illumination solution for high-contrast imaging that is compatible with conventional microscopy setups and can be integrated with biophotonics devices, such as portable microscopy, NIR-II range bioimaging, and lab-on-a-chip devices.
Increasing the baud rate in optical interconnects (OIs) will require the use of more sophisticated driver and receiver electronics. This will help overcome the stagnated bandwidth of the Vertical-Cavity Surface-Emitting Laser (VCSEL) and the pin-photodetector. Next generation OIs operating at single lane rates of 50+ Gbaud will therefore require careful co-optimization of the electronics and the optoelectronics. To facilitate this work there is a need of an accurate equivalent circuit for the optoelectronic components, functioning over a broad drive current and ambient temperature range. The VCSEL is the most important and complex to model due to its non-linear behavior and strongly varying characteristics with drive current and ambient temperature. For this purpose, a large-signal equivalent circuit dedicated for high-speed datacom VCSELs has been developed and is presented here. The distributed electrical parasitics in the device layout are carefully considered, the intrinsic speed limitation from carrier transport effects in the Separate-Confinement-Heterostructure (SCH) and the carrier-photon interaction in the Quantum Wells (QWs) are included, and self-heating effects in the device are monitored. The circuit is purposely based on physical instead of empirical models so that it can provide usable feedback to VCSEL designers. For circuit demonstration, it is implemented in Keysight's Advanced Design System (ADS) software and thereafter applied to replicate the performance of a state-of-the-art 28-GHz-bandwidth VCSEL at different temperatures and drive currents. Comparison is made between simulated and measured steady-state characteristics, small-signal behavior, and large-signal response under 28 Gbaud On-Off-Keying (OOK) and Pulse-Amplitude-Modulation 4 (PAM4) modulation, showing good agreement.
Optical interconnects (OIs) continue to require more and more sophisticated driver and receiver electronics as higher baud rates are pushed in datacom, mainly due to the bandwidth stagnation of the optoelectronic components such as the vertical-cavity surface-emitting laser (VCSEL) in the transmitter as well as the photodetector (PD) in the receiver. Another important focus is maintaining high energy efficiency for the OIs. For both cases, a reliable equivalent circuit model for high-speed VCSELs is required for link optimization. This work is an extension of our previously presented large-signal VCSEL equivalent circuit model, where noise is added to the physical processes that the VCSEL model is based on. Thus, a new step is taken in the strive of developing an even more accurate physics-based large-signal VCSEL equivalent circuit model for datacom applications. Following a detailed description of the VCSEL noise modelling, a presentation is given on simulated results of VCSEL relative intensity noise (RIN) spectrum and eye diagrams under 28 Gbaud on-off keying (OOK) and pulse-amplitude 4 (PAM4) modulation, and that are compared with corresponding measurement results. Good agreement is found over a wide range of VCSEL driving conditions and temperatures. A 28-GHz-bandwidth VCSEL is applied for the demonstration, and the presented noise extended VCSEL equivalent circuit model was implemented in Verilog-A, and the simulations were performed using Cadence Spectre.
6 November 2023 in IEEE Transactions on Very Large Scale Integration (VLSI) Systems
Vertical-cavity surface-emitting lasers (VCSELs) are an attractive choice for short-range optical interconnects (OIs) in datacom applications. As the limited bandwidth of VCSEL is the major obstacle to achieve higher data rates in these optical systems, significant effort has been made to develop efficient VCSEL drivers to increase data rate while consuming less power. We review the current state of the art in VCSEL driver design, including choices of semiconductor technology, modulation format, VCSEL configuration, and pre-emphasis techniques, with an emphasis on the challenges associated with VCSEL bandwidth limitations, nonlinearities, and thermal behavior. We illustrate the choices with simulations using a physics-based VCSEL model.