Periodic Reporting for period 1 - RaPPIC (Rapid Programmable Photonic Integrated Circuits)
Période du rapport: 2023-07-01 au 2024-12-31
Information processing is currently undergoing yet another revolution, with the rise of large-language models enabled by massive computing power. However, this remarkable progress is facing fundamental limits to keep up with the required demand. Therefore, alternative forms of computing are highly necessary for future growth and development. It has long been hypothesized that photonic integrated circuits, operating with light instead of electric charge, could pave the way to computational performance beyond the limitations of transistor miniaturization. In particular, they could operate much faster, while being much more energy efficient at the same time. One of the key bottlenecks in the development of photonic computing is the current lack of an efficient photonic counterpart of electronic transistors as universal building blocks of integrated circuits. To date, complementary approaches in photonics have been utilized for controlling the phase of light and thus its interference effects. However, they all fall short in providing an efficient and universal platform for programmable photonic circuits. In the RaPPIC project, we proposed to develop novel electro-optical gates based on atomically thin two-dimensional semiconductors as a new paradigm for the realization of universal building block of rapid and programmable photonic integrated circuitry. Using these gates, a photonic circuit can be reprogrammed at ultra-high speeds with highest phase manipulation efficiency, allowing small footprints and enabling scalability that is indispensable for leveraging the fundamental advantages of photonic computing.
The primary work area of the project was the development of a prototype photonic device for efficient phase modulation utilizing the novel material platform of atomically thin semiconducting transition metal dichalcogenides. The device parameters and geometry were optimized to study phase modulation of 2D-gates in a foundry-produced silicon photonic integrated circuits. Additionally, conventional phase modulators based on thermo-optic phase shifters were also designed and benchmarked in terms of key performance metrics, such as energy consumption, modulation efficiency and optical losses.
Second activity involved directly utilizing foundry-manufactured chip to train machine-learning algorithm for computer vision application. As part of this, custom software and electronics were developed. This includes custom electronic driving circuit for more than 40 modulators with a custom printed circuit board. A python-based software code was developed to interface digital controller (Raspberry Pi) and implement computation on the photonic chip.
The main outcomes of the action are twofold. First, the potential of transition metal dichalcogenide for integrated photonic application was demonstrated. Second, various tools, including nanofabrication techniques, electronics, optical setups and software, were developed to assemble and test the photonic integrated circuits.
Second activity involved directly utilizing foundry-manufactured chip to train machine-learning algorithm for computer vision application. As part of this, custom software and electronics were developed. This includes custom electronic driving circuit for more than 40 modulators with a custom printed circuit board. A python-based software code was developed to interface digital controller (Raspberry Pi) and implement computation on the photonic chip.
The main outcomes of the action are twofold. First, the potential of transition metal dichalcogenide for integrated photonic application was demonstrated. Second, various tools, including nanofabrication techniques, electronics, optical setups and software, were developed to assemble and test the photonic integrated circuits.
The project results have validated the potential of low-loss programmable circuits for information processing tasks. Further steps for technology transfer and commercialization have been recognized by receiving further financing for the main members involved in the RaPPIC project, including support from EXIST grant worth 1.3 M€ from German Ministry of Economics and Climate. The innovative modulator components require further research and development to optimize the design and study the feasibility of scalable implementation, further activities here are supported by recently funded project from German Ministry of Education and Research in a joint collaboration with two research groups in Taiwan which will bring expertise in semiconductor manufacturing.