Co-packaged optics (CPO): status, challenges, and solutions

Progress in silicon-based reconfigurable and programmable all-optical signal processing chips

Taking the advantage of ultrafast optical linear and nonlinear effects, all-optical signal processing (AOSP) enables manipulation, regeneration, and computing of information directly in optical domain without resorting to electronics. As a promising photonic integration platform, silicon-on-insulator (SOI) has the advantage of complementary metal oxide semiconductor (CMOS) compatibility, low-loss, compact size as well as large optical nonlinearities. In this paper, we review the recent progress in the project granted to develop silicon-based reconfigurable AOSP chips, which aims to combine the merits of AOSP and silicon photonics to solve the unsustainable cost and energy challenges in future communication and big data applications. Three key challenges are identified in this project: (1) how to finely manipulate and reconfigure optical fields, (2) how to achieve ultra-low loss integrated silicon waveguides and significant enhancement of nonlinear effects, (3) how to mitigate crosstalk between optical, electrical and thermal components. By focusing on these key issues, the following major achievements are realized during the project. First, ultra-low loss silicon-based waveguides as well as ultra-high quality microresonators are developed by advancing key fabrication technologies as well as device structures. Integrated photonic filters with bandwidth and free spectral range reconfigurable in a wide range were realized to finely manipulate and select input light fields with a high degree of freedom. Second, several mechanisms and new designs that aim at nonlinear enhancement have been proposed, including optical ridge waveguides with reverse biased PIN junction, slot waveguides, multimode waveguides and parity-time symmetry coupled microresonators. Advanced AOSP operations are verified with these novel designs. Logical computations at 100 Gbit/s were demonstrated with self-developed, monolithic integrated programmable optical logic array. High-dimensional multi-value logic operations based on the four-wave mixing effect are realized. Multi-channel all-optical amplitude and phase regeneration technology is developed, and a multi-channel, multi-format, reconfigurable all-optical regeneration chip is realized. Expanding regeneration capacity via spatial dimension is also verified. Third, the crosstalk from optical as well as thermal coupling due to high-density integration are mitigated by developing novel optical designs and advanced packaging technologies, enabling high-density, small size, multi-channel and multi-functional operation with low power consumption. Finally, four programmable AOSP chips are developed, i.e., programmable photonic filter chip, programmable photonic logic operation chip, multi-dimensional all-optical regeneration chip, and multi-channel and multi-functional AOSP chip with packaging. The major achievements developed in this project pave the way toward ultra-low loss, high-speed, high-efficient, high-density information processing in future classical and non-classical communication and computing applications.

High-density MDM silicon photonic transmitter assisted with MRMs and a two-dimensional grating coupler for single-fiber 4 × 80-Gbps signaling

In this paper, a four-channel silicon photonic mode division multiplexing (MDM) transmitter chiplet is proposed over a 1.85-mm2 footprint, utilizing micro-ring modulators and a two-dimensional grating coupler (2D GC) with side-distributed Bragg reflectors. Through genetic optimization approach, the optimized 2D GC can simultaneously couple the two orthogonal polarizations of the on-chip TE0 and TE1 modes to the four LP modes in the FMF, all exhibiting high coupling efficiency beyond -3.8 dB, so as to support the MDM optical fiber interface. High data rate up to 4 × 80-Gbps signaling is experimentally demonstrated with a bit error rate below the 7% hard-decision forward-error-correction threshold. The proposed transmitter chiplet features single fiber and single-wavelength applications of quad small form-factor pluggable (QSFP) transceivers, pave the way toward future high-density and large-capacity optical interconnects scenarios.

Electronic Chip Package and Co-Packaged Optics (CPO) Technology for Modern AI Era: A Review

With the growing demand for high-performance computing (HPC), artificial intelligence (AI), and data communication and storage, new chip technologies have emerged, following Moore's Law, over the past few decades. As we enter the post-Moore era, transistor dimensions are approaching their physical limits. Advanced packaging technologies, such as 3D chiplets hetero-integration and co-packaged optics (CPO), have become crucial for further improving system performance. Currently, most solutions rely on silicon-based technologies, which alleviate some challenges but still face issues such as warpage, bumps' reliability, through-silicon vias' (TSVs) and redistribution layers' (RDLs) reliability, and thermal dissipation, etc. Glass, with its superior mechanical, thermal, electrical, and optical properties, is emerging as a promising material to address these challenges, particularly with the development of femtosecond laser technology. This paper discusses the evolution of both conventional and advanced packaging technologies and outlines future directions for design, fabrication, and packaging using glass substrates and femtosecond laser processing.

Sub-2W tunable laser based on silicon photonics power amplifier

High-power tunable lasers are intensely pursued due to their vast application potential such as in telecom, ranging, and molecular sensing. Integrated photonics, however, is usually considered not suitable for high-power applications mainly due to its small size which limits the energy storage capacity and, therefore, the output power. In the late 90s, to improve the beam quality and increase the stored energy, large-mode-area (LMA) fibers were introduced in which the optical mode area is substantially large. Such LMA fibers have transformed the high-power capability of fiber systems ever since. Introducing such an LMA technology at the chip-scale can play an equally disruptive role with high power signal generation from an integrated photonics system. To this end, in this work we demonstrate such a technology, and show a very high-power tunable laser with the help of a silicon photonics based LMA power amplifier. We show output power reaching 1.8 W over a tunability range of 60 nm, spanning from 1.83 µm to 1.89 µm, limited only by the seed laser. Such an integrated LMA device can be used to substantially increase the power of the existing integrated tunable lasers currently limited to a few tens of milliwatts. The power levels demonstrated here reach and surpass that of many benchtop systems which truly makes the silicon photonics based integrated LMA device poised towards mass deployment for high power applications without relying on benchtop systems.

Progress in Research on Co-Packaged Optics

In the 5G era, the demand for high-bandwidth computing, transmission, and storage has led to the development of optoelectronic interconnect technology. This technology has evolved from traditional board-edge optical modules to smaller and more integrated solutions. Co-packaged optics (CPO) has evolved as a solution to meet the growing demand for data. Compared to typical optoelectronic connectivity technology, CPO presents distinct benefits in terms of bandwidth, size, weight, and power consumption. This study presents an overview of CPO, highlighting its fundamental principles, advantages, and distinctive features. Additionally, it examines the current research progress of two distinct approaches utilizing Vertical-Cavity Surface-Emitting Laser (VCSEL) and silicon photonics integration technology. Additionally, it provides a concise overview of the many application situations of CPO. Expanding on this, the analysis focuses on the CPO using 2D, 2.5D, and 3D packaging techniques. Lastly, taking into account the present technological environment, the scientific obstacles encountered by CPO are analyzed, and its future progress is predicted.

Challenges and opportunities in engineering next-generation 3D microelectronic devices: improved performance and higher integration density

In recent years, nanotechnology and materials science have evolved and matured, making it increasingly easier to design and fabricate next-generation 3D microelectronics. The process has changed drastically from traditional 2D microelectronics, resulting in improved performance, higher integration density, and new functionalities. As applications become more complex and power-intensive, this technology can address the demands of high-performance computing, advanced sensors, and cutting-edge communication systems via wearable, flexible devices, etc. To manufacture higher-density microelectronics, recent advances in the fabrication of such 3D devices are discussed. Furthermore, the paper stresses the importance of novel materials and architectures, such as monolithic 3D integration and heterogeneous integration, in overcoming these challenges. We emphasize the importance of addressing complex issues to achieve better performance and higher integration density, which will play an important role in shaping the next generation of microelectronic devices. The multifaceted challenges involved in developing next-generation 3D microelectronic devices are also highlighted.

Multimodal deep learning using on-chip diffractive optics with in situ training capability

Multimodal deep learning plays a pivotal role in supporting the processing and learning of diverse data types within the realm of artificial intelligence generated content (AIGC). However, most photonic neuromorphic processors for deep learning can only handle a single data modality (either vision or audio) due to the lack of abundant parameter training in optical domain. Here, we propose and demonstrate a trainable diffractive optical neural network (TDONN) chip based on on-chip diffractive optics with massive tunable elements to address these constraints. The TDONN chip includes one input layer, five hidden layers, and one output layer, and only one forward propagation is required to obtain the inference results without frequent optical-electrical conversion. The customized stochastic gradient descent algorithm and the drop-out mechanism are developed for photonic neurons to realize in situ training and fast convergence in the optical domain. The TDONN chip achieves a potential throughput of 217.6 tera-operations per second (TOPS) with high computing density (447.7 TOPS/mm2), high system-level energy efficiency (7.28 TOPS/W), and low optical latency (30.2 ps). The TDONN chip has successfully implemented four-class classification in different modalities (vision, audio, and touch) and achieve 85.7% accuracy on multimodal test sets. Our work opens up a new avenue for multimodal deep learning with integrated photonic processors, providing a potential solution for low-power AI large models using photonic technology.

Compact lithium niobate plasmonic modulator

Lithium niobate (LN)-based modulators offer superior modulation performances, including high-speed modulation, linearity, and temperature stability. However, these devices exhibit larger sizes due to the low light-matter interaction despite a significant electro-optic coefficient. In this work, we present a compact LN-based modulator using a plasmonic mode that confines the optical mode in a very narrow gap. By filling the gap with LN, the confinement factor in the LN is significantly enhanced. The proposed modulator provides an extremely small half-wave voltage-length product, VπL of 0.02 V/cm at an optical communication wavelength (λ = 1.55 µm). The proposed modulator scheme can be utilized in a wide range of optical communication devices that demand small footprints and a high-speed operation.

Optimization of the cavity length and pulse characterization based on germanene as a saturable absorber in an Er-doped fiber laser

In this study, germanene-nanosheets (NSs) were synthesized by liquid-phase exfoliation, followed by an experimental investigation into the nonlinear saturable absorption characteristics and morphological structure of germanene. The germanene-NSs were employed as saturable absorbers, exhibiting saturation intensity and modulation depth values of 22.64M W/c m 2 and 4.48%, respectively. This demonstrated the feasibility of utilizing germanene-NSs passively mode-locked in an erbium-doped fiber laser (EDFL). By optimizing the cavity length, improvements in the output of EDFL characteristics were achieved, resulting in 883 fs pulses with a maximum average output power of 19.74 mW. The aforementioned experimental outcomes underscore the significant potential of germanene in the realms of ultrafast photonics and nonlinear optics.