Nikolay G. Pavlov

Artificial neural networks set the pace in machine vision, natural language processing, and scientific discovery, but their performance depends on fast and efficient tensor computations. Analog photonic systems are a promising alternative to digital electronics because they enable ultra-fast, low-latency computing while avoiding capacitive charging losses and electrical crosstalk. Here we present a photonic tensor processor for deep neural network inference, integrated into a standard 19-inch rack unit with a high-speed electronic interface to PyTorch for seamless hardware deployment. The processor implements an all-optical crossbar with nine inputs and three outputs for parallel intensity-based accumulation of weighted signals. Fabricated in imec’s iSiPP50G silicon photonics platform, the chip integrates electro-absorption modulators and photodiodes for scalability and compatibility with high-volume manufacturing. An integrated self-injection-locked microcomb provides stable multi-wavelength carriers. We demonstrate inference on MNIST and CIFAR-10 with 98.1% and 72.0% accuracy, highlighting a compact, reprogrammable platform toward scalable high-speed optical AI accelerators.

Artificial neural networks set the pace in machine vision, natural language processing, and scientific discovery. Yet their success comes with a rising need for fast and efficient tensor computations, the key operation underpinning neural networks. Analog photonic systems offer a promising solution to perform tensor operations more efficiently than digital electronics because of ultra-fast signal propagation for low-latency computing as well as the elimination of charging, discharging capacitances and electrical crosstalk. Here we present an all-optical photonic tensor processor capable of deep neural network inference. Integrated in a standard 19-inch rack unit with a complete high-speed electronic interface to the PyTorch framework, the processor enables seamless deployment of neural networks on photonic hardware. Our photonic processor realizes an all-optical crossbar with nine input and three output channels that performs parallel intensity-based accumulation of weighted signals. The chip is fabricated using imec’s iSiPP50G silicon photonics platform, integrating electro-absorption modulators and photodiodes to ensure scalability and compatibility with high-volume manufacturing. An integrated, self-injection-locked microcomb provides a stable multi-wavelength light source for simultaneous optical carriers. We demonstrate inference on MNIST and CIFAR-10, achieving 98.1% and 72.0% classification accuracy, respectively. Together, these advances demonstrate a compact, reprogrammable photonic computing platform compatible with industrial silicon processes as a key step toward scalable, high-speed optical accelerators for artificial intelligence.

Extremely high-Q microresonators provide an attractive platform for a plethora of photonic applications including optical frequency combs, high-precision metrology, telecommunication, microwave generation, narrow linewidth lasers, and stable frequency references. Moreover, the desire for compactness and a low power threshold for nonlinear phenomena have spurred investigation into integrated and scalable solutions. Historically, crystalline microresonators with Q~10^9 were one of the first material platforms providing unprecedented optical performance in a small form factor. A key challenge, though, with these devices is in finding alternatives to fragile, bulky, and free-space couplers, such as tapered fibers, prisms, and cleaved fibers. Here, we present for the first time, the coupling of a photonic wire bond (PWB) to a MgF2-based microresonator to generate solitons and a pure, low-noise microwave signal based on Kerr-microcombs. These results open a path towards scalable heterogeneous integration of crystalline microresonators with integrated photonics. Moreover, because PWBs possess advantages over traditional coupling elements in terms of ease of fabrication, size, and flexibility, they constitute a more advanced optical interface for linear and nonlinear photonics.

Laser-driven microresonators have enabled chip-integrated light sources with unique properties, including the self-organized formation of ultrashort soliton pulses and frequency combs (microcombs). While poised to impact major photonic applications such as spectroscopy, sensing and optical data processing, microcombs still necessitate complex scientific equipment to achieve and maintain suitable single-pulse operation. Here to address this challenge, we demonstrate microresonators with programmable synthetic reflection providing tailored injection feedback to the driving laser. Synthetic reflection achieves independence from random sample scattering properties and, when designed appropriately, enables deterministic access to self-injection-locked microcombs operating exclusively in the single-soliton regime. These results provide a route to easily operable microcombs at scale for portable sensors, autonomous navigation or extreme-bandwidth data processing. The novel concept of synthetic reflection may also be generalized to other integrated photonic systems.

The original version of this Article contained an error in the first sentence of the Acknowledgements, which incorrectly read ‘This publication was supported by Contract HR0011-15-C-0055 (DODOS) from the Defense Advanced Research Projects Agency (DARPA), Defense Sciences Office (DSO).’ The correct version states ‘Microsystems Technology Office (MTO)’ in place of ‘Defense Sciences Office (DSO)’. This has been corrected in both the PDF and HTML versions of the Article.

Microcombs provide a path to broad-bandwidth integrated frequency combs with low power consumption, which are compatible with wafer-scale fabrication. Yet, electrically-driven, photonic chip-based microcombs are inhibited by the required high threshold power and the frequency agility of the laser for soliton initiation. Here we demonstrate an electrically-driven soliton microcomb by coupling a III–V-material-based (indium phosphide) multiple-longitudinal-mode laser diode chip to a high-Q silicon nitride microresonator fabricated using the photonic Damascene process. The laser diode is self-injection locked to the microresonator, which is accompanied by the narrowing of the laser linewidth, and the simultaneous formation of dissipative Kerr solitons. By tuning the laser diode current, we observe transitions from modulation instability, breather solitons, to single-soliton states. The system operating at an electronically-detectable sub-100-GHz mode spacing requires less than 1 Watt of electrical power, can fit in a volume of ca. 1 cm³, and does not require on-chip filters and heaters, thus simplifying the integrated microcomb.

We simultaneously create stable solitons in up to three distinct mode families of a single crystalline MgF 2 microresonator. The resulting Kerr combs are mutually coherent and have distinct repetition rates and are thus well suited for dual-comb applications.

We present the analysis of the self-injection locking of a single-frequency laser diode to a high-Q whispering gallery mode (WGM) microresonator with Rayleigh backscattering.Simple analytical formulas for the width of the locking band and resulting laser linewidth are derived.