Multipurpose silicon photonics signal processor core

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.

Complex-valued matrix-vector multiplication using a scalable coherent photonic processor

Matrix-vector multiplication is a fundamental operation in modern signal processing and artificial intelligence. Developing a chip-scale photonic matrix-vector multiplication processor (MVMP) offers the potential for notably enhanced computing speed and energy efficiency beyond microelectronics. Here, we propose and demonstrate a 16-channel programmable on-chip coherent photonic processor capable of performing complex-valued matrix-vector multiplication at a computing speed of 1.28 tera-operations per second (TOPS). Low phase error Mach-Zehnder interferometers mesh and ultralow-loss broadened photonic waveguide delay lines are firstly combined for optical computing, enabling the encoding of amplitude and phase information, along with high-speed coherent detection. The proposed MVMP demonstrates high flexibility in implementing various functions, including arbitrary matrix transformation, parallel image processing, and handwritten digital recognition. Our work demonstrates the MVMP's advantages in scalability and function flexibility, enabled by the low-loss and low phase error designs, making a substantial advancement in high-speed and large-scale photonic computing technologies.

Linear optical wave energy redistribution methods for photonic signal processing

Manipulating the phase of an optical wave over time and frequency gives full control to the user to implement a wide variety of energy preserving transformations directly in the analogue optical domain. These can be achieved using widely available linear mechanisms, such as temporal phase modulation and spectral phase filtering. The techniques based on these linear optical wave energy redistribution (OWER) methods are inherently energy efficient and have significant speed and bandwidth advantages over digital signal processing. We describe several recent OWER methods for optical signal processing, including denoising passive amplification, real-time spectrogram analysis, passive logic computing, and more. These functionalities are relevant whenever the signal is found on a classical or quantum optical wave, or could be upconverted from radio frequencies or microwaves, and they are of interest for a wide range of applications in telecommunications, sensing, metrology, biomedical imaging, and astronomy. The energy preservation of these methods makes them particularly interesting for quantum optics applications. Furthermore, many of the individual components have been demonstrated on-chip, enabling miniaturization for applications where size and weight are a main constraint.

Embedding matrices in programmable photonic networks with flexible depth and width

We show that programmable photonic circuit architectures composed of alternating mixing layers and active layers offer a high degree of flexibility. This alternating configuration enables the systematic tailoring of both the network's depth (number of layers) and width (size of each layer) without compromising computational capabilities. From a mathematical perspective, our approach can be viewed as embedding an arbitrary target matrix into a higher-dimensional matrix, which can then be represented with fewer layers and a larger number of active elements. We derive a general relation for the width and depth of a network that guarantees representing all N × N complex-valued matrix operations. Remarkably, we show that just two such active layers-interleaved with passive mixing layers-are sufficient to universally implement arbitrary matrix transformations. This result promises a more adaptable and scalable route to photonic matrix processors.

Inverse Design of Unitary Transmission Matrices in Silicon Photonic Coupled Waveguide Arrays Using a Neural Adjoint Model

The development of low-loss reconfigurable integrated optical devices enables further research into technologies including photonic signal processing, analogue quantum computing, and optical neural networks. Here, we introduce digital patterning of coupled waveguide arrays as a platform capable of implementing unitary matrix operations. Determining the required device geometry for a specific optical output is computationally challenging and requires a robust and versatile inverse design protocol. In this work we present an approach using high speed neural network surrogate-based gradient optimization, capable of predicting patterns of refractive index perturbations based on switching of the ultralow loss chalcogenide phase change material, antimony triselinide (Sb2Se3). Results for a 3 × 3 silicon waveguide array are presented, demonstrating control of both amplitude and phase for each transmission matrix element. Network performance is studied using neural network optimization tools such as data set augmentation and supplementation with random noise, resulting in an average fidelity of 0.94 for unitary matrix targets. Our results show that coupled waveguide arrays with perturbation patterns offer new routes for achieving programmable unitary operators, or Hamiltonians for quantum simulators, with a reduced footprint compared to conventional interferometer-mesh technology.

Analog parallel processor for broadband multifunctional integrated system based on silicon photonic platform

Sharing the hardware platform between diverse information systems to establish full cooperation among different functionalities has attracted substantial attention. However, broadband multifunctional integrated systems with large operating frequency ranges are challenging due to the bandwidth and computing speed restrictions of electronic circuitry. Here, we report an analog parallel processor (APP) based on the silicon photonic platform that directly discretizes and parallelizes the broadband signal in the analog domain. The APP first discretizes the signal with the optical frequency comb and then adopts optical dynamic phase interference to reassign the analog signal into 2N parallel sequences. Via photonic analog parallelism, data rate and data volume in each sequence are simultaneously compressed, which mitigates the requirement on each parallel computing core. Moreover, the fusion of the outputs from each computing core is equivalent to directly processing broadband signals. In the proof-of-concept experiment, two-channel analog parallel processing of broadband radar signals and high-speed communication signals is implemented on the single photonic integrated circuit. The bandwidth of broadband radar signal is 6 GHz and the range resolution of 2.69 cm is achieved. The wireless communication rate of 8 Gbit/s is also validated. Breaking the bandwidth and speed limitations of the single-computing core along with further exploring the multichannel potential of this architecture, we anticipate that the proposed APP will accelerate the development of powerful opto-electronic processors as critical support for applications such as satellite networks and intelligent driving.

Photonics-assisted analog windowing and matched filtering

In this paper, we propose an integrated method for windowing and matched filtering in the analog domain based on microwave photonic technology, which utilizes dispersion regulation of optical waveguide to achieve the windowing processing of broadband signals in the optical domain and the surface acoustic wave filter (SAWF) to achieve the following matched filtering processing in the radio frequency (RF) domain, thus realizing their integration processing in the analog domain. The proposed method is validated by simulation and experiment, in which the integrated processing of matched filtering and windowing in the analog domain for a linear frequency modulation (LFM) signal with a bandwidth of 1 GHz is carried out and the peak to sidelobe ratio (PSLR) of the output signal is -19.55 dB and the mainlobe width (MLW) broadens to 0.16 ns with the maximum achievable dispersion value. The proposed method can significantly reduce the pressure of signal processing at the back-end of the radar system and improve the response speed of the broadband radar system in detecting the targets.

Chiral exceptional point enhanced active tuning and nonreciprocity in micro-resonators

Exceptional points (EPs) have been extensively explored in mechanical, acoustic, plasmonic, and photonic systems. However, little is known about the role of EPs in tailoring the dynamic tunability of optical devices. A specific type of EPs known as chiral EPs has recently attracted much attention for controlling the flow of light and for building sensors with better responsivity. A recently demonstrated route to chiral EPs via lithographically defined symmetric Mie scatterers on the rim of resonators has not only provided the much-needed mechanical stability for studying chiral EPs, but also helped reduce losses originating from nanofabrication imperfections, facilitating the in-situ study of chiral EPs and their contribution to the dynamics and tunability of resonators. Here, we use asymmetric Mie scatterers to break the rotational symmetry of a microresonator, to demonstrate deterministic thermal tuning across a chiral EP, and to demonstrate EP-mediated chiral optical nonlinear response and efficient electro-optic tuning. Our results indicate asymmetric electro-optic modulation with up to 17 dB contrast at GHz and CMOS-compatible voltage levels. Such wafer-scale nano-manufacturing of chiral electro-optic modulators and the chiral EP-tailored tunning may facilitate new micro-resonator functionalities in quantum information processing, electromagnetic wave control, and optical interconnects.

Versatile parallel signal processing with a scalable silicon photonic chip

Silicon photonic signal processors promise a new generation of signal processing hardware with significant advancements in processing bandwidth, low power consumption, and minimal latency. Programmable silicon photonic signal processors, facilitated by tuning elements, can reduce hardware development cycles and costs. However, traditional programmable photonic signal processors based on optical switches face scalability and performance challenges due to control complexity and transmission losses. Here, we propose a scalable parallel signal processor on silicon for versatile applications by interleaving wavelength and temporal optical dimensions. Additionally, it incorporates ultra-low-loss waveguides and low-phase-error optical switch techniques, achieving an overall insertion loss of 10 dB. This design offers low loss, high scalability, and simplified control, enabling advanced functionalities such as accurate microwave reception, narrowband microwave photonic filtering, wide-bandwidth arbitrary waveform generation, and high-speed parallel optical computing without the need for tuning elements calibration. Our programmable parallel signal processor demonstrates advantages in both scale and performance, marking a significant advancement in large-scale, high-performance, multifunctional photonic systems.

Spatial Optical Simulator for Classical Statistical Models

Optical simulators for the Ising model have demonstrated great promise for solving challenging problems in physics and beyond. Here, we develop a spatial optical simulator for a variety of classical statistical systems, including the clock, XY, Potts, and Heisenberg models, utilizing a digital micromirror device composed of a large number of tiny mirrors. Spins, with desired amplitudes or phases of the statistical models, are precisely encoded by a patch of mirrors with a superpixel approach. Then, by modulating the light field in a sequence of designed patterns, the spin-spin interaction is realized in such a way that the Hamiltonian symmetries are preserved. We successfully simulate statistical systems on a fully connected network, with ferromagnetic or Mattis-type random interactions, and observe the corresponding phase transitions between the paramagnetic and the ferromagnetic or spin-glass phases. Our results largely extend the research scope of spatial optical simulators and their versatile applications.

Defect Regulation Strategy of Porous Persistent Phosphors for Multiple and Dynamic Information Encryption

Optical encryption technologies based on persistent luminescence material have currently drawn increasing attention due to the distinctive and long-lived optical properties, which enable multi-dimensional and dynamic optical information encryption to improve the security level. However, the controlled synthesis of persistent phosphors remains largely unexplored and it is still a great challenge to regulate the structure for optical properties optimization, which inevitably sets significant limitations on the practical application of persistent luminescent materials. Herein, a controlled synthesis method is proposed based on defect structure regulation and a series of porous persistent phosphors is obtained with different luminous intensities, lifetime, and wavelengths. By simply using diverse templates during the sol-gel process, the oxygen vacancy defects structures are successfully regulated to improve the optical properties. Additionally, the obtained series of porous Al2O3 are utilized for multi-color and dynamic optical information encryption to increase the security level. Overall, the proposed defect regulation strategy in this work is expected to provide a general and facile method for optimizing the optical properties of persistent luminescent materials, paving new ways for broadening their applications in multi-dimensional and dynamic information encryption.

A programmable topological photonic chip

Controlling topological phases of light allows the observation of abundant topological phenomena and the development of robust photonic devices. The prospect of more sophisticated control with topological photonic devices for practical implementations requires high-level programmability. Here we demonstrate a fully programmable topological photonic chip with large-scale integration of silicon photonic nanocircuits and microresonators. Photonic artificial atoms and their interactions in our compound system can be individually addressed and controlled, allowing the arbitrary adjustment of structural parameters and geometrical configurations for the observation of dynamic topological phase transitions and diverse photonic topological insulators. Individual programming of artificial atoms on the generic chip enables the comprehensive statistical characterization of topological robustness against relatively weak disorders, and counterintuitive topological Anderson phase transitions induced by strong disorders. This generic topological photonic chip can be rapidly reprogrammed to implement multifunctionalities, providing a flexible and versatile platform for applications across fundamental science and topological technologies.

Towards large-scale programmable silicon photonic chip for signal processing

Optical signal processing has been playing a crucial part as powerful engine for various information systems in the practical applications. In particular, achieving large-scale programmable chips for signal processing are highly desirable for high flexibility, low cost and powerful processing. Silicon photonics, which has been developed successfully in the past decade, provides a promising option due to its unique advantages. Here, recent progress of large-scale programmable silicon photonic chip for signal processing in microwave photonics, optical communications, optical computing, quantum photonics as well as dispersion controlling are reviewed. Particularly, we give a discussion about the realization of high-performance building-blocks, including ultra-low-loss silicon photonic waveguides, 2 × 2 Mach-Zehnder switches and microring resonator switches. The methods for configuring large-scale programmable silicon photonic chips are also discussed. The representative examples are summarized for the applications of beam steering, optical switching, optical computing, quantum photonic processing as well as optical dispersion controlling. Finally, we give an outlook for the challenges of further developing large-scale programmable silicon photonic chips.

Modifying single-crystal silicon and trimming silicon microring devices by femtosecond laser irradiation

Single-crystal silicon (c-Si) is a vital component of photonic devices and has obvious advantages. Moreover, femtosecond-pulsed laser interactions with matter have been widely applied in micro/nanoscale processing. In this paper, we report the modification mechanisms of c-Si induced by a femtosecond laser (350 fs, 520 nm) at different pulse fluences, along with the mechanism of this technique to trim the phase error of c-Si-based devices. In this study, several distinct types of final micro/nanostructures, such as amorphization and ablation, were analyzed. The near-surface morphology was characterized using optical microscopy, scanning electron microscopy, and atomic force microscopy. The main physical modification processes were further analyzed using a two-temperature model. By employing Raman spectroscopy, we demonstrated that a higher laser fluence significantly contributes to the formation of more amorphous silicon components. The thickness of the amorphous layer was almost uniform (approximately 30 nm) at different induced fluences, as determined using transmission electron microscopy. From the ellipsometry measurements, we demonstrated that the refractive index increases for amorphization while the ablation decreases. In addition, we investigated the ability of the femtosecond laser to modify the effective index of c-Si microring waveguides by either amorphization or ablation. Both blue and red shifts of resonance spectra were achieved in the microring devices, resulting in double-direction trimming. Our results provide further insight into the femtosecond laser modification mechanism of c-Si and may be a practical method for dealing with the fabrication errors of c-Si-based photonic devices.

Monolithic back-end-of-line integration of phase change materials into foundry-manufactured silicon photonics

Monolithic integration of novel materials without modifying the existing photonic component library is crucial to advancing heterogeneous silicon photonic integrated circuits. Here we show the introduction of a silicon nitride etch stop layer at select areas, coupled with low-loss oxide trench, enabling incorporation of functional materials without compromising foundry-verified device reliability. As an illustration, two distinct chalcogenide phase change materials (PCMs) with remarkable nonvolatile modulation capabilities, namely Sb2Se3 and Ge2Sb2Se4Te1, were monolithic back-end-of-line integrated, offering compact phase and intensity tuning units with zero-static power consumption. By employing these building blocks, the phase error of a push-pull Mach-Zehnder interferometer optical switch could be reduced with a 48% peak power consumption reduction. Mirco-ring filters with >5-bit wavelength selective intensity modulation and waveguide-based >7-bit intensity-modulation broadband attenuators could also be achieved. This foundry-compatible platform could open up the possibility of integrating other excellent optoelectronic materials into future silicon photonic process design kits.

Optimized wideband and compact multifunctional photonic device based on Sb2S3 phase change material

In this paper, a 1 × 2 photonic switch is designed based on a silicon-on-insulator (SOI) platform combined with the phase change material (PCM), Sb2S3, assisted by the direct binary search (DBS) algorithm. The designed photonic switch exhibits an impressive operating bandwidth ranging from 1450 to 1650 nm. The device has an insertion loss (IL) from 0.44 dB to 0.70 dB (of less than 0.7 dB) and cross talk (CT) from -26 dB to -20 dB (of less than -20 dB) over an operating bandwidth of 200 nm, especially an IL of 0.52 dB and CT of -24 dB at 1550 nm. Notably, the device is highly compact, with footprints of merely 3 × 4 µm2. Furthermore, we have extended the device's functionality for multifunctional operation in the C-band that can serve as both a 1 × 2 photonic switch and a 3 dB photonic power splitter. In the photonic switch mode, the device demonstrates an IL of 0.7 dB and a CT of -13.5 dB. In addition, when operating as a 3 dB photonic power splitter, the IL is less than 0.5 dB.

Integrated lithium niobate microwave photonic processing engine

Integrated microwave photonics (MWP) is an intriguing technology for the generation, transmission and manipulation of microwave signals in chip-scale optical systems1,2. In particular, ultrafast processing of analogue signals in the optical domain with high fidelity and low latency could enable a variety of applications such as MWP filters3-5, microwave signal processing6-9 and image recognition10,11. An ideal integrated MWP processing platform should have both an efficient and high-speed electro-optic modulation block to faithfully perform microwave-optic conversion at low power and also a low-loss functional photonic network to implement various signal-processing tasks. Moreover, large-scale, low-cost manufacturability is required to monolithically integrate the two building blocks on the same chip. Here we demonstrate such an integrated MWP processing engine based on a 4 inch wafer-scale thin-film lithium niobate platform. It can perform multipurpose tasks with processing bandwidths of up to 67 GHz at complementary metal-oxide-semiconductor (CMOS)-compatible voltages. We achieve ultrafast analogue computation, namely temporal integration and differentiation, at sampling rates of up to 256 giga samples per second, and deploy these functions to showcase three proof-of-concept applications: solving ordinary differential equations, generating ultra-wideband signals and detecting edges in images. We further leverage the image edge detector to realize a photonic-assisted image segmentation model that can effectively outline the boundaries of melanoma lesion in medical diagnostic images. Our ultrafast lithium niobate MWP engine could provide compact, low-latency and cost-effective solutions for future wireless communications, high-resolution radar and photonic artificial intelligence.

General-purpose programmable photonic processor for advanced radiofrequency applications

A general-purpose photonic processor can be built integrating a silicon photonic programmable core in a technology stack comprising an electronic monitoring and controlling layer and a software layer for resource control and programming. This processor can leverage the unique properties of photonics in terms of ultra-high bandwidth, high-speed operation, and low power consumption while operating in a complementary and synergistic way with electronic processors. These features are key in applications such as next-generation 5/6 G wireless systems where reconfigurable filtering, frequency conversion, arbitrary waveform generation, and beamforming are currently provided by microwave photonic subsystems that cannot be scaled down. Here we report the first general-purpose programmable processor with the remarkable capability to implement all the required basic functionalities of a microwave photonic system by suitable programming of its resources. The processor is fabricated in silicon photonics and incorporates the full photonic/electronic and software stack.

Non-volatile 2 × 2 optical switch using multimode interference in an Sb2Se3-loaded waveguide

We propose a non-volatile 2 × 2 photonic switch based on multimode interference in an Sb2Se3-loaded waveguide. The different modal symmetries of the TE0 and TE1 modes supported in the multimode region change their propagation constants distinctly upon the Sb2Se3 phase transition. Through careful optical design and FDTD optimization of the multimode waveguide dimensions, efficient switching is achieved despite the modest index contrast of Sb2Se3 relative to Ge2Sb2Te5. The fabricated optical switch demonstrates favorable characteristics, including low insertion loss of ∼1 dB, a compact length of ∼27 µm, and small cross talk below -15 dB across a 35 nm bandwidth. Such non-volatile and broadband components will be critical for future high-density programmable photonic-integrated circuits for optical communications and signal processing.

Dielectric optical waveguide fabricated on a transparent substrate

Transparent glass substrates are routinely used in the fabrication of metasurfaces, augmented reality (AR), virtual reality (VR), and holographic devices. While readily compatible with photolithographic patterning methods, when electron beam (E-Beam) techniques are used, field distortion and stitching errors can result due to the buildup of charge. A common approach to overcome this issue is to deposit a thin conductive polymer layer (E-Spacer). However, if high-voltage E-Beam is used to achieve nano-features, the polymer conductivity is not sufficient. We have shown that by using chromium (Cr) as an overcoating conductive layer on the resist, we can achieve accurate and seamless patterning in multiple writing fields and used the method to fabricate on-chip Si3N4 waveguides on SiO2. This technique has the potential to enable the fabrication of large-scale integrated photonic systems on transparent or dielectric substrates.

Integrated Bragg grating filters based on silicon-Sb2Se3 with non-volatile bandgap engineering capability

Integrated optical filters show outstanding capability in integrated reconfigurable photonic applications, including wavelength division multiplexing (WDM), programmable photonic processors, and on-chip quantum photonic networks. Present schemes for reconfigurable filters either have a large footprint or suffer from high static power consumption, hindering the development of reconfigurable photonic integrated systems. Here, a reconfigurable hybrid Bragg grating filter is elaborately designed through a precise, modified coupling mode theory. It is also experimentally presented by integrating non-volatile phase change material (PCM) Sb2Se3 on silicon to realize compact, low-loss, and broadband engineering operations. The fabricated filter holds a compact footprint of 0.5 µm × 43.5 µm and maintains a low insertion loss of < 0.5 dB after multiple levels of engineering to achieve crystallization. The filter is able to switch from a low-loss transmission state to the Bragg reflection state, making it a favorable solution for large-scale reconfigurable photonic circuits. With a switching extinction ratio over 30 dB at 1504.85 nm, this hybrid filter breaks the tradeoff between insertion loss and tuning range. These results reveal its potential as a new candidate for a basic element in large-scale non-volatile reconfigurable systems.

Flexible dispersion engineering using polymer patterning in nanophotonic waveguides

We demonstrate the engineering of waveguide dispersion by lithographically patterning the polymer cladding on silicon nitride waveguide resonators. Both normal and anomalous dispersion, ranging from - 462 to 409 ps/nm/km, can be achieved for the same waveguide dimension within an integrated photonic chip. In the meantime, this simple process shows no impact on the waveguide loss and the quality factor of the waveguide resonators, offering flexibility in tailoring designable dispersion for a universal photonic platform. In addition, by adjusting the coverage ratio of cladding, relatively low dispersion (≈ - 130 ps/nm/km) is also demonstrated in the same waveguide resonator, yielding the potentials for zero-dispersive waveguide resonators by a proper coverage ratio of the polymer cladding.

Dual-layer optical encryption fluorescent polymer waveguide chip based on optical pulse-code modulation technique

Information encryption technique has broad applications in individual privacy, military confidentiality, and national security, but traditional electronic encryption approaches are increasingly unable to satisfy the demands of strong safety and large bandwidth of high-speed data transmission over network. Optical encryption technology could be more flexible and effective in parallel programming and multiple degree-of-freedom data transmitting application. Here, we show a dual-layer optical encryption fluorescent polymer waveguide chip based on optical pulse-code modulation technique. Fluorescent oligomers were doped into epoxy cross-linking SU-8 polymer as a gain medium. Through modifying both the external pumping wavelength and operating frequency of the pulse-code modulation, the sender could ensure the transmission of vital information is secure. If the plaintext transmission is eavesdropped, the external pumping light will be switched, and the receiver will get warning commands of ciphertext information in the standby network. This technique is suitable for high-integration and high-scalability optical information encryption communications.

Ultrabroadband and compact 2 × 2 3-dB coupler based on trapezoidal subwavelength gratings

We propose and experimentally demonstrate an ultrabroadband and compact 2 × 2 3-dB coupler based on the trapezoidal subwavelength gratings (SWGs). The adiabatic coupling is achieved between a trapezoidal SWG waveguide and a reversely tapered strip waveguide, which contributes to the ultrabroad operation bandwidth and the compact footprint of the coupler. Numerical results prove that our device has a power splitting imbalance of < ± 0.5 dB and an excess loss of < 0.2 dB in the ultrabroad bandwidth of 300 nm from 1400 nm to 1700nm, with a coupling length of 4.4 µm and a total length of 24.4 µm. The fabricated device is characterized in a 270-nm bandwidth from 1400 nm to 1670 nm, showing a measured power splitting imbalance of < ± 0.7 dB and an excess loss of < 0.5 dB.

On-chip non-uniformly spaced multi-channel microwave photonic signal processor based on an ultrahigh-Q multimode micro-disk resonator

Multi-channel microwave photonic (MWP) signal processing can simultaneously perform different task operations on multiple signals carried by multiple wavelengths, which holds great potential for ultrafast signal processing and characterization in a wavelength-division-multiplexed (WDM) network. As emerging telecommunication services create more data, an elastic optical network, which has a flexible and non-uniform spectrum channel spacing, is an alternative architecture to meet the ever-increasing data transfer need. Here, for the multi-channel ultra-fast signal processing in the elastic optical network, we propose and demonstrate an on-chip non-uniformly spaced multi-channel microwave photonic signal processor based on an ultrahigh-Q multimode micro-disk resonator (MDR). In the proposed signal processor, an MDR supporting multiple different order whispering-gallery modes (WGMs) with an ultrahigh Q-factor is specifically designed. Benefiting from the large and different free spectral ranges (FSRs) provided by the different order WGMs, a non-uniformly spaced multi-channel microwave photonic signal processor is realized, and various processing functions are experimentally demonstrated including bandpass filtering with a narrow passband of 103 MHz, a rejection ratio of 22.3 dB and a frequency tuning range from 1 to 30 GHz, multiple frequency measurement with a frequency measurement range from 1 to 30 GHz, a frequency resolution better than 200 MHz and a measurement accuracy of 91.3 MHz, and phase shifting with a phase tuning range from -170°∼160°, an operational bandwidth of 26 GHz from 6 GHz to 32 GHz and a small power variation of 0.43 dB. Thanks to the coexistence of different order WGMs supported by the MDR, non-uniformly spaced multi-channel signal processing is enabled with the key advantages including a broad operation bandwidth, an ultra-narrow frequency selectivity, and a large phase tuning range with a small power variation. The proposed signal processor is promising to be widely used in future elastic optical networks with flexible spectrum grids.

Ultrafast dynamic machine vision with spatiotemporal photonic computing

Ultrafast dynamic machine vision in the optical domain can provide unprecedented perspectives for high-performance computing. However, owing to the limited degrees of freedom, existing photonic computing approaches rely on the memory's slow read/write operations to implement dynamic processing. Here, we propose a spatiotemporal photonic computing architecture to match the highly parallel spatial computing with high-speed temporal computing and achieve a three-dimensional spatiotemporal plane. A unified training framework is devised to optimize the physical system and the network model. The photonic processing speed of the benchmark video dataset is increased by 40-fold on a space-multiplexed system with 35-fold fewer parameters. A wavelength-multiplexed system realizes all-optical nonlinear computing of dynamic light field with a frame time of 3.57 nanoseconds. The proposed architecture paves the way for ultrafast advanced machine vision free from the limits of memory wall and will find applications in unmanned systems, autonomous driving, ultrafast science, etc.

Experimentally realized in situ backpropagation for deep learning in photonic neural networks

Integrated photonic neural networks provide a promising platform for energy-efficient, high-throughput machine learning with extensive scientific and commercial applications. Photonic neural networks efficiently transform optically encoded inputs using Mach-Zehnder interferometer mesh networks interleaved with nonlinearities. We experimentally trained a three-layer, four-port silicon photonic neural network with programmable phase shifters and optical power monitoring to solve classification tasks using "in situ backpropagation," a photonic analog of the most popular method to train conventional neural networks. We measured backpropagated gradients for phase-shifter voltages by interfering forward- and backward-propagating light and simulated in situ backpropagation for 64-port photonic neural networks trained on MNIST image recognition given errors. All experiments performed comparably to digital simulations ([Formula: see text]94% test accuracy), and energy scaling analysis indicated a route to scalable machine learning.

CMOS-compatible electro-optical SRAM cavity device based on negative differential resistance

The impending collapse of Moore-like growth of computational power has spurred the development of alternative computing architectures, such as optical or electro-optical computing. However, many of the current demonstrations in literature are not compatible with the dominant complementary metal-oxide semiconductor (CMOS) technology used in large-scale manufacturing today. Here, inspired by the famous Esaki diode demonstrating negative differential resistance (NDR), we show a fully CMOS-compatible electro-optical memory device, based on a new type of NDR diode. This new diode is based on a horizontal PN junction in silicon with a unique layout providing the NDR feature, and we show how it can easily be implemented into a photonic micro-ring resonator to enable a bistable device with a fully optical readout in the telecom regime. Our result is an important stepping stone on the way to new nonlinear electro-optic and neuromorphic computing structures based on this new NDR diode.

On-chip Ce:YIG/Si Mach-Zehnder optical isolator with low power consumption

The integrated optical isolator is an essential building block in photonic integrated chips. However, the performance of on-chip isolators based on the magneto-optic (MO) effect has been limited due to the magnetization requirement of permanent magnets or metal microstrips on MO materials. Here, an MZI optical isolator built on a silicon-on-insulator (SOI) without any external magnetic field is proposed. A multi-loop graphene microstrip operating as an integrated electromagnet above the waveguide, instead of the traditional metal microstrip, generates the saturated magnetic fields required for the nonreciprocal effect. Subsequently, the optical transmission can be tuned by varying the intensity of currents applied on the graphene microstrip. Compared with gold microstrip, the power consumption is reduced by 70.8%, and temperature fluctuation is reduced by 69.5% while preserving the isolation ratio of 29.44 dB and the insertion loss of 2.99 dB at1550 nm.

Vacuum-sealed silicon photonic MEMS tunable ring resonator with an independent control over coupling and phase

Ring resonators are a vital element for filters, optical delay lines, or sensors in silicon photonics. However, reconfigurable ring resonators with low-power consumption are not available in foundries today. We demonstrate an add-drop ring resonator with the independent tuning of round-trip phase and coupling using low-power microelectromechanical (MEMS) actuation. At a wavelength of 1540 nm and for a maximum voltage of 40 V, the phase shifters provide a resonance wavelength tuning of 0.15 nm, while the tunable couplers can tune the optical resonance extinction ratio at the through port from 0 to 30 dB. The optical resonance displays a passive quality factor of 29 000, which can be increased to almost 50 000 with actuation. The MEMS rings are individually vacuum-sealed on wafer scale, enabling reliable and long-term protection from the environment. We cycled the mechanical actuators for more than 4 × 10⁹ cycles at 100 kHz, and did not observe degradation in their response curves. On mechanical resonance, we demonstrate a modulation increase of up to 15 dB, with a voltage bias of 4 V and a peak drive amplitude as low as 20 mV.

Asymmetric, non-uniform 3-dB directional coupler with 300-nm bandwidth and a small footprint

Here we demonstrate a new, to the best of our knowledge, type of 3-dB coupler that has an ultra-broadband operational range from 1300 to 1600 nm with low fabrication sensitivity. The overall device size is 800 µm including in/out S-bend waveguides. The coupler is an asymmetric non-uniform directional coupler that consists of two tapered waveguides. One of the coupler arms is shifted by 100 µm in the propagation direction, which results in a more wavelength-insensitive 3-dB response compared to a standard (not shifted) coupler. Moreover, compared to a long adiabatic coupler, we achieved a similar wavelength response at a 16-times-smaller device length. The couplers were fabricated using the silicon nitride platform of Lionix International. We also experimentally demonstrated an optical switch that is made by using two of these couplers in a Mach-Zehnder interferometer configuration. According to experimental results, this optical switch exhibits -10 dB of extinction ratio over the 1500-1600 nm wavelength range. Our results indicate that this new type of coupler holds great promise for various applications, including optical imaging, telecommunications, and reconfigurable photonic processors where compact, fabrication-tolerant, and wavelength-insensitive couplers are essential.

Ultrahigh dynamic range and low noise figure programmable integrated microwave photonic filter

Microwave photonics has adopted a number of important concepts and technologies over the recent pasts, including photonic integration, versatile programmability, and techniques for enhancing key radio frequency performance metrics such as the noise figure and the dynamic range. However, to date, these aspects have not been achieved simultaneously in a single circuit. Here, we report a multi-functional photonic integrated circuit that enables programmable filtering functions with record-high performance. We demonstrate reconfigurable filter functions with record-low noise figure and a RF notch filter with ultra-high dynamic range. We achieve this unique feature using versatile complex spectrum tailoring enabled by an all integrated modulation transformer and a double injection ring resonator as a multi-function optical filtering component. Our work breaks the conventional and fragmented approach of integration, functionality and performance that currently prevents the adoption of integrated MWP systems in real applications.

Ultra-compact and broadband all-silicon TM-pass power splitter using subwavelength holey-structured metamaterial waveguides

Power splitters with polarization management features are highly desired to construct high-density silicon photonic integrated circuits. However, few attempts have been made to design a single device that can act as both a power splitter and a TE- or TM-pass polarizer. In this paper, for the first time, we experimentally demonstrate an ultra-compact and broadband all-silicon TM-pass power splitter, where a triple-guide directional coupler (TGDC) composed of three parallel subwavelength holey-structured metamaterial waveguides (SHMWs) is located at central coupling region and three regular strip waveguides are connected at the input/output ports. Such a SHMW can enhance the reflection to realize a wide stop-band for the undesired TE polarized light, while achieving the low loss transmission for the TM polarized light. Besides, the TM dispersion can be significantly flattened by the designed SHMWs, leading to a broadband power splitting for TM polarization. Simulated results show that an ultra-compact device of 1.7 × 4 µm² in size is obtained with an insertion loss (IL) of 0.34 dB and an extinction ratio (ER) of 36 dB at 1550 nm, and its working bandwidth can be extended to ∼240 nm by keeping IL < 0.9 dB and ER > 16 dB. The measurements of the fabricated devices show low IL (<1 dB) and high ER (>15 dB) over the measured wavelength range of 1460 to 1580 nm, which is consistent with the simulation results.

Quantum arbitrary waveform generator

Controlling the temporal waveform of light is the key to a versatile light source in classical and quantum electronics. Although pulse shaping of classical light is mature and has been used in various fields, more advanced applications would be realized by a light source that generates arbitrary quantum light with arbitrary temporal waveforms. We call such a device a quantum arbitrary waveform generator (Q-AWG). The Q-AWG must be able to handle various quantum states of light, which are fragile. Thus, the Q-AWG requires a radically different methodology from classical pulse shaping. Here, we invent an architecture of Q-AWGs that can operate semi-deterministically at a repetition rate over gigahertz in principle. We demonstrate its core technology via generating highly nonclassical states with temporal waveforms that have never been realized before. This result would lead to powerful quantum technologies based on Q-AWGs such as practical optical quantum computing.

Thermo-optic tuning of silicon nitride microring resonators with low loss non-volatile [Formula: see text] phase change material

A new family of phase change material based on antimony has recently been explored for applications in near-IR tunable photonics due to its wide bandgap, manifested as broadband transparency from visible to NIR wavelengths. Here, we characterize [Formula: see text] optically and demonstrate the integration of this phase change material in a silicon nitride platform using a microring resonator that can be thermally tuned using the amorphous and crystalline states of the phase change material, achieving extinction ratios of up to 18 dB in the C-band. We extract the thermo-optic coefficient of the amorphous and crystalline states of the [Formula: see text] to be 3.4 x [Formula: see text] and 0.1 x 10[Formula: see text], respectively. Additionally, we detail the first observation of bi-directional shifting for permanent trimming of a non-volatile switch using continuous wave (CW) laser exposure ([Formula: see text] to 5.1 dBm) with a modulation in effective refractive index ranging from +5.23 x [Formula: see text] to [Formula: see text] x 10[Formula: see text]. This work experimentally verifies optical phase modifications and permanent trimming of [Formula: see text], enabling potential applications such as optically controlled memories and weights for neuromorphic architecture and high density switch matrix using a multi-layer PECVD based photonic integrated circuit.

Silicon Photonic Phase Shifters and Their Applications: A Review

With the development of silicon photonics, dense photonic integrated circuits play a significant role in applications such as light detection and ranging systems, photonic computing accelerators, miniaturized spectrometers, and so on. Recently, extensive research work has been carried out on the phase shifter, which acts as the fundamental building block in the photonic integrated circuit. In this review, we overview different types of silicon photonic phase shifters, including micro-electro-mechanical systems (MEMS), thermo-optics, and free-carrier depletion types, highlighting the MEMS-based ones. The major working principles of these phase shifters are introduced and analyzed. Additionally, the related works are summarized and compared. Moreover, some emerging applications utilizing phase shifters are introduced, such as neuromorphic computing systems, photonic accelerators, multi-purpose processing cores, etc. Finally, a discussion on each kind of phase shifter is given based on the figures of merit.

Ultra-broadband on-chip beam focusing enabled by GRIN metalens on silicon-on-insulator platform

Metalens has emerged as an important optical block in free-space optical systems, which shows excellent performance. Even the metalens based on gradient index (GRIN) profiles can be implemented for on-chip beam focusing behavior. However, for most previous schemes, the GRIN metalenses can only achieve on-chip beam focusing behavior in one dimension, which limits their applications in low-loss waveguide interconnecting or fiber-to-chip coupling. In this paper, an on-chip half Maxwell's fisheye lens based on GRIN profiles with subwavelength features, integrated with silicon waveguides, is experimentally demonstrated. Benefitting from the index distribution and beam focusing characteristics of the half Maxwell's fisheye lens, an on-chip beam transforming can be achieved for transverse electric (TE) fundamental mode in two waveguides with different heights and widths. The simulated 1 dB bandwidth can reach 1100 nm, which exhibits great prospects in integrated photonic circuits. The measured insertion loss of an on-chip 5.4 μm-length lens is less than 1 dB to connect a 220 nm-height, 8 μm-wide waveguide, and a 60 nm-height, 0.5 μm-wide waveguide in the wavelength range of 1280-1620 nm.

Compact nonvolatile 2×2 photonic switch based on two-mode interference

On-chip nonvolatile photonic switches enabled by phase change materials (PCMs) are promising building blocks for power-efficient programmable photonic integrated circuits. However, large absorption loss in conventional PCMs (such as Ge₂Sb₂Te₅) interacting with weak evanescent waves in silicon waveguides usually leads to high insertion loss and a large device footprint. In this paper, we propose a 2×2 photonic switch based on two-mode interference in a multimode slot waveguide (MSW) with ultralow loss Sb₂S₃ integrated inside the slot region. The MSW supports two lowest order TE modes, i.e., symmetric TE₀₀ and antisymmetric TE₀₁ modes, and the phase of Sb₂S₃ could actively tune two-mode interference behavior. Owing to the enhanced electric field in the slot, the interaction strength between modal field and Sb₂S₃ could be boosted, and a photonic switch containing a ∼9.4 µm-long Sb₂S₃-MSW hybrid section could effectively alter the light transmission between bar and cross ports upon the phase change of Sb₂S₃ with a cross talk (CT) less than -13.6 dB and an insertion loss (IL) less than 0.26 dB in the telecommunication C-band. Especially at 1550 nm, the CT in the amorphous (crystalline) Sb₂S₃ is -36.1 dB (-31.1 dB) with a corresponding IL of 0.073 dB (0.055 dB). The proposed 2×2 photonic switch is compact in size and compatible with on-chip microheaters, which may find promising applications in reconfigurable photonic devices.

Ultra-low-energy programmable non-volatile silicon photonics based on phase-change materials with graphene heaters

Silicon photonics is evolving from laboratory research to real-world applications with the potential to transform many technologies, including optical neural networks and quantum information processing. A key element for these applications is a reconfigurable switch operating at ultra-low programming energy-a challenging proposition for traditional thermo-optic or free carrier switches. Recent advances in non-volatile programmable silicon photonics based on phase-change materials (PCMs) provide an attractive solution to energy-efficient photonic switches with zero static power, but the programming energy density remains high (hundreds of attojoules per cubic nanometre). Here we demonstrate a non-volatile electrically reconfigurable silicon photonic platform leveraging a monolayer graphene heater with high energy efficiency and endurance. In particular, we show a broadband switch based on the technologically mature PCM Ge₂Sb₂Te₅ and a phase shifter employing the emerging low-loss PCM Sb₂Se₃. The graphene-assisted photonic switches exhibited an endurance of over 1,000 cycles and a programming energy density of 8.7 ± 1.4 aJ nm^-3, that is, within an order of magnitude of the PCM thermodynamic switching energy limit (~1.2 aJ nm^-3) and at least a 20-fold reduction in switching energy compared with the state of the art. Our work shows that graphene is a reliable and energy-efficient heater compatible with dielectric platforms, including Si₃N₄, for technologically relevant non-volatile programmable silicon photonics.

The Design of a Low-Loss, Fast-Response, Metal Thermo-Optic Phase Shifter Based on Coupled-Mode Theory

A thermo-optic phase shifter is of great importance in silicon photonics. However, it is difficult to simultaneously achieve a good performance in insertion loss and in response speed by using traditional thermal tuning through a metallic heater. In this paper, based on coupled-mode theory, we propose a method to place high-loss materials close to the optical waveguide while maintaining the low loss of the optical device, which ensures the low insertion loss (~0.78 dB) of the phase shifter. Additionally, thanks to the very short distance between the rib waveguide and the chromium (Cr) heater, the phase shifter exhibits a high response speed (1.15 μs in rise time and 2.18 μs in decay time) with a measured bandwidth (BW) of 186 kHz. Moreover, we further optimize the structure of phase shifters, leading to the reduction of π-shift power consumption from 25.1 mW to 13.6 mW. Our proposed phase shifters have great potential in large-scale silicon photonic integrated circuits.

A Review of Capabilities and Scope for Hybrid Integration Offered by Silicon-Nitride-Based Photonic Integrated Circuits

In this review we present some of the recent advances in the field of silicon nitride photonic integrated circuits. The review focuses on the material deposition techniques currently available, illustrating the capabilities of each technique. The review then expands on the functionalisation of the platform to achieve nonlinear processing, optical modulation, nonvolatile optical memories and integration with III-V materials to obtain lasing or gain capabilities.

High-performance silicon PIN diode switches in the 2-µm wave band

The 2-µm wave band has attracted significant research interest due to its potential applications for next-generation high-capacity optical communication and sensing. As the key component, fast optical switches are essential for an advanced and reconfigurable optical network. Motivated by this prospect, we propose and demonstrate two typical silicon PIN diode switches at 2 µm. One is based on a coupled microring resonator (CMRR), and the other is based on a Mach-Zehnder interferometer (MZI) with a push-pull-like configuration. The measured insertion loss of the CMRR switch is <2.5 dB, and the cross talk is <-10.8 dB. The insertion loss of the MZI switch is <2 dB, and the cross talk is <-15.6 dB. The switch times of these two structures are both lower than 12.5 ns.

Chip-integrated van der Waals PN heterojunction photodetector with low dark current and high responsivity

Two-dimensional materials are attractive for constructing high-performance photonic chip-integrated photodetectors because of their remarkable electronic and optical properties and dangling-bond-free surfaces. However, the reported chip-integrated two-dimensional material photodetectors were mainly implemented with the configuration of metal-semiconductor-metal, suffering from high dark currents and low responsivities at high operation speed. Here, we report a van der Waals PN heterojunction photodetector, composed of p-type black phosphorous and n-type molybdenum telluride, integrated on a silicon nitride waveguide. The built-in electric field of the PN heterojunction significantly suppresses the dark current and improves the responsivity. Under a bias of 1 V pointing from n-type molybdenum telluride to p-type black phosphorous, the dark current is lower than 7 nA, which is more than two orders of magnitude lower than those reported in other waveguide-integrated black phosphorus photodetectors. An intrinsic responsivity up to 577 mA W^-1 is obtained. Remarkably, the van der Waals PN heterojunction is tunable by the electrostatic doping to further engineer its rectification and improve the photodetection, enabling an increased responsivity of 709 mA W^-1. Besides, the heterojunction photodetector exhibits a response bandwidth of ~1.0 GHz and a uniform photodetection over a wide spectral range, as experimentally measured from 1500 to 1630 nm. The demonstrated chip-integrated van der Waals PN heterojunction photodetector with low dark current, high responsivity and fast response has great potentials to develop high-performance on-chip photodetectors for various photonic integrated circuits based on silicon, lithium niobate, polymer, etc.

Photonic matrix multiplication lights up photonic accelerator and beyond

Matrix computation, as a fundamental building block of information processing in science and technology, contributes most of the computational overheads in modern signal processing and artificial intelligence algorithms. Photonic accelerators are designed to accelerate specific categories of computing in the optical domain, especially matrix multiplication, to address the growing demand for computing resources and capacity. Photonic matrix multiplication has much potential to expand the domain of telecommunication, and artificial intelligence benefiting from its superior performance. Recent research in photonic matrix multiplication has flourished and may provide opportunities to develop applications that are unachievable at present by conventional electronic processors. In this review, we first introduce the methods of photonic matrix multiplication, mainly including the plane light conversion method, Mach-Zehnder interferometer method and wavelength division multiplexing method. We also summarize the developmental milestones of photonic matrix multiplication and the related applications. Then, we review their detailed advances in applications to optical signal processing and artificial neural networks in recent years. Finally, we comment on the challenges and perspectives of photonic matrix multiplication and photonic acceleration.

Self-controlling photonic-on-chip networks with deep reinforcement learning

We present a novel photonic chip design for high bandwidth four-degree optical switches that support high-dimensional switching mechanisms with low insertion loss and low crosstalk in a low power consumption level and a short switching time. Such four-degree photonic chips can be used to build an integrated full-grid Photonic-on-Chip Network (PCN). With four distinct input/output directions, the proposed photonic chips are superior compared to the current bidirectional photonic switches, where a conventionally sizable PCN can only be constructed as a linear chain of bidirectional chips. Our four-directional photonic chips are more flexible and scalable for the design of modern optical switches, enabling the construction of multi-dimensional photonic chip networks that are widely applied for intra-chip communication networks and photonic data centers. More noticeably, our photonic networks can be self-controlling with our proposed Multi-Sample Discovery model, a deep reinforcement learning model based on Proximal Policy Optimization. On a PCN, we can optimize many criteria such as transmission loss, power consumption, and routing time, while preserving performance and scaling up the network with dynamic changes. Experiments on simulated data demonstrate the effectiveness and scalability of the proposed architectural design and optimization algorithm. Perceivable insights make the constructed architecture become the self-controlling photonic-on-chip networks.

Arbitrary optical wave evolution with Fourier transforms and phase masks

A large number of applications in classical and quantum photonics require the capability of implementing arbitrary linear unitary transformations on a set of optical modes. In a seminal work by Reck et al. [Phys. Rev. Lett.73, 58 (1994)10.1103/PhysRevLett.73.58], it was shown how to build such multiport universal interferometers with a mesh of beam splitters and phase shifters, and this design became the basis for most experimental implementations in the last decades. However, the design of Reck et al. is difficult to scale up to a large number of modes, which would be required for many applications. Here we present a deterministic algorithm that can find an exact and efficient implementation of any unitary transformation, using only Fourier transforms and phase masks. Since Fourier transforms and phase masks are routinely implemented in several optical setups and they do not suffer from some of the scalability issues associated with building extensive meshes of beam splitters, we believe that our design can be useful for many applications in photonics.

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.

Photonic Matrix Computing: From Fundamentals to Applications

In emerging artificial intelligence applications, massive matrix operations require high computing speed and energy efficiency. Optical computing can realize high-speed parallel information processing with ultra-low energy consumption on photonic integrated platforms or in free space, which can well meet these domain-specific demands. In this review, we firstly introduce the principles of photonic matrix computing implemented by three mainstream schemes, and then review the research progress of optical neural networks (ONNs) based on photonic matrix computing. In addition, we discuss the advantages of optical computing architectures over electronic processors as well as current challenges of optical computing and highlight some promising prospects for the future development.

Silicon nitride programmable photonic processor with folded heaters

General-purpose programmable photonic processors rely on the large-scale integration of beamsplitters and reconfigurable phase shifters, distributed within unit cells or photonic gates. With their future evolution threatened by several hardware constrains, including the integration density that can be achieved with current mesh topologies, in this work, we present a unit cell topology design to increase the integration density of waveguide mesh arrangements based on folded Mach-Zehnder Interferometers. We report the design details of a 40-unit cell waveguide mesh integrated in a 11mm x 5.5 mm silicon nitride chip achieving, to the best of our knowledge, the highest integration density reported to date for a general-purpose photonic processor. The chip is electrically interfaced to a PCB and we report examples of reconfigurable optical beamsplitters, basic tunable microwave photonic filters with high peak rejection (40 dB approx.), as well as the dynamic interconnection and routing of 5G digitally modulated signals within the photonic mesh.

Parallel convolutional processing using an integrated photonic tensor core

With the proliferation of ultrahigh-speed mobile networks and internet-connected devices, along with the rise of artificial intelligence (AI)¹, the world is generating exponentially increasing amounts of data that need to be processed in a fast and efficient way. Highly parallelized, fast and scalable hardware is therefore becoming progressively more important². Here we demonstrate a computationally specific integrated photonic hardware accelerator (tensor core) that is capable of operating at speeds of trillions of multiply-accumulate operations per second (10¹² MAC operations per second or tera-MACs per second). The tensor core can be considered as the optical analogue of an application-specific integrated circuit (ASIC). It achieves parallelized photonic in-memory computing using phase-change-material memory arrays and photonic chip-based optical frequency combs (soliton microcombs³). The computation is reduced to measuring the optical transmission of reconfigurable and non-resonant passive components and can operate at a bandwidth exceeding 14 gigahertz, limited only by the speed of the modulators and photodetectors. Given recent advances in hybrid integration of soliton microcombs at microwave line rates^3-5, ultralow-loss silicon nitride waveguides^6,7, and high-speed on-chip detectors and modulators, our approach provides a path towards full complementary metal-oxide-semiconductor (CMOS) wafer-scale integration of the photonic tensor core. Although we focus on convolutional processing, more generally our results indicate the potential of integrated photonics for parallel, fast, and efficient computational hardware in data-heavy AI applications such as autonomous driving, live video processing, and next-generation cloud computing services.