†All three authors contributed equally to this work pg. 2 Recent advances in nonlinear optics have revolutionized the area of integrated photonics, providing on-chip solutions to a wide range of new applications. Currently, the state of the art integrated nonlinear photonic devices are mainly based on dielectric material platforms, such as Si3N4 and SiO2. While semiconductor materials hold much higher nonlinear coefficients and convenience in active integration, they suffered in the past from high waveguide losses that prevented the realization of highly efficient nonlinear processes on-chip. Here we challenge this status quo and demonstrate an ultra-low loss AlGaAs-on-insulator (AlGaAsOI) platform with anomalous dispersion and quality (Q) factors beyond 1.5 × 10 6 . Such a high quality factor, combined with the high nonlinear coefficient and the small mode volume, enabled us to demonstrate a record low Kerr frequency comb generation threshold of ~36 µW for a resonator with a 1 THz free spectral range (FSR), ~100 times lower compared to that in previous semiconductor platform. Combs with >250 nm broad span have been generated under a pump power lower than the threshold power of state of the art dielectric micro combs. A soliton-step transition has also been observed for the first time from an AlGaAs resonator. This work is an important step towards ultra-efficient semiconductor-based nonlinear photonics and will lead to fully integrated nonlinear photonic integrated circuits (PICs) in near future. pg. 3 The extensive research on integrated nonlinear photonics in the last few years, driven by the breakthrough of the microcomb and other on-chip nonlinear devices, has opened up many new opportunities for on-chip integrated photonics, ranging from spectroscopy to atomic clock applications [1-3]. The demand to construct efficient nonlinear devices has motivated the development of different material platforms in nonlinear photonics. A common endeavor of those efforts is the reduction of the waveguide propagation loss, which is essential to enable high Q cavities so as to enhance the build-up power in the resonators and therefore increase the efficiency of the nonlinear optical processes [4]. In this regard, silica on silicon resonators [5-7] have long been dominant offering Q factors as high as 1 billion [6]. These devices can access a wide range of nonlinear effects including microwave rate soliton microcombs [8].However, over the last 5 years, there has been remarkable progress to significantly improve the Q factors of resonators in many other nonlinear integrated optical material platforms. One example is the Si3N4 platform, which delivers high performance in Kerr comb generation on chip [9][10][11]. The Si3N4 micro-resonators have enabled the generation of efficient frequency combs with repetition rates from microwave to THz frequencies [12] and improved Q factor of beyond 30 million [13,14]. Another material, which recently attracted attention, is LiNbO3. It offers additional opportunities for integrated nonlinear...

Microcombs have sparked a surge of applications over the past decade, ranging from optical communications to metrology1–4. Despite their diverse deployment, most microcomb-based systems rely on a large amount of bulky elements and equipment to fulfil their desired functions, which is complicated, expensive and power consuming. By contrast, foundry-based silicon photonics (SiPh) has had remarkable success in providing versatile functionality in a scalable and low-cost manner5–7, but its available chip-based light sources lack the capacity for parallelization, which limits the scope of SiPh applications. Here we combine these two technologies by using a power-efficient and operationally simple aluminium-gallium-arsenide-on-insulator microcomb source to drive complementary metal–oxide–semiconductor SiPh engines. We present two important chip-scale photonic systems for optical data transmission and microwave photonics, respectively. A microcomb-based integrated photonic data link is demonstrated, based on a pulse-amplitude four-level modulation scheme with a two-terabit-per-second aggregate rate, and a highly reconfigurable microwave photonic filter with a high level of integration is constructed using a time-stretch approach. Such synergy of a microcomb and SiPh integrated components is an essential step towards the next generation of fully integrated photonic systems.

The emergence of parallel convolution-operation technology has substantially powered the complexity and functionality of optical neural networks (ONN) by harnessing the dimension of optical wavelength. However, this advanced architecture faces remarkable challenges in high-level integration and on-chip operation. In this work, convolution based on time-wavelength plane stretching approach is implemented on a microcomb-driven chip-based photonic processing unit (PPU). To support the operation of this processing unit, we develop a dedicated control and operation protocol, leading to a record high weight precision of 9 bits. Moreover, the compact architecture and high data loading speed enable a preeminent photonic-core compute density of over 1 trillion of operations per second per square millimeter (TOPS mm−2). Two proof-of-concept experiments are demonstrated, including image edge detection and handwritten digit recognition, showing comparable processing capability compared to that of a digital computer. Due to the advanced performance and the great scalability, this parallel photonic processing unit can potentially revolutionize sophisticated artificial intelligence tasks including autonomous driving, video action recognition and image reconstruction.

We theoretically and experimentally demonstrate a significantly large modulation efficiency of a compact graphene modulator based on a silicon waveguide using the electro refractive effect of graphene. The modulation modes of electro-absorption and electro-refractive can be switched with different applied voltages. A high extinction ratio of 25 dB is achieved in the electro-absorption modulation mode with a driving voltage range of 0 V to 1 V. For electro-refractive modulation, the driving voltage ranges from 1 V to 3 V with a 185-pm spectrum shift. The modulation efficiency of 1.29 V · mm with a 40-μm interaction length is two orders of magnitude higher than that of the first reported graphene phase modulator. The realisation of phase and intensity modulation with graphene based on a silicon waveguide heralds its potential application in optical communication and optical interconnection systems.