The wafer of the 800G optical transceiver chip

on display at the Light of Internet Expo in 2021

The 800G optical transceiver chip and optical engine technology, one of the 14 World Leading Internet Scientific and Technology Achievements released in 2021, has updates.

In May, Optics Express, an authoritative optical journal, published the project update "High linearity silicon DC Kerr modulator enhanced by slow light for 112 Gbit/s PAM4 over 2 km single mode fiber transmission" about 800G optical transceiver chip and optical engine technology. The online article proposes and demonstrates an innovative design idea, and was included in the "selected articles". The corresponding author of the article is YU Hui, an expert researcher of ZJ Lab's Research Institute of Intelligent Networks.

The rapid development of big data, AI, 5G, and more emerging technologies and industries depends on the swift transmission and computation of mega data. According to statistics, the traffic of the world's data centers grew 32% every year, putting forward an urgent need for elevating data transmission and processing capacities of data centers. "Large data centers use light to transmit high-speed data and use electricity to compute them. Optical transceiver chip and optical module are vital to optical interconnection. This is a transcriber bridge between light and electricity," said YU Hui, "the modulator is key to the transcription."

In practice, the speed, power dissipation and accuracy of transcription are subject to various factors such as long transmission, insertion loss and bandwidth. How to improve speed, reduce error and minimize loss are the three major challenges for electro-optic transmission and transcription. "The popular modulating method is to use "carrier plasma dispersion effect", namely to use an external electric field to modulate light transmission," explained ZHANG Qiang, member of the project team and senior researcher of ZJ Lab, "this is a quite established method, but the results are certain insertion loss and low modulation efficiency. It cannot meet the demands of data transmission for high efficiency, high speed and high linearity degree (note: an index for data accuracy during electro-optic transcription)."

Improving electro-optic modulator performance is   long stress of global academic and industrial circles as well as the target   of YU Hui and his teammates ZHANG Qiang and XIA Penghui.

The key to technical bottlenecks lies in   fundamental research. Upon the launch of the project, the team tried to find   breakthrough points from fundamental physical effects, signal modulation   theories and opponent structural design. The finding is that in the context   of current technology, improving electro-optic modulation is mainly through   the "slow light effect", i.e. altering optical waveguide structure   to slow down light and amplify the light-matter reaction. Along with recent   material science progress, a new method, the "DC Kerr effect",   namely using an external electric field to break up silicon lattice symmetry   to alter optical waveguide refractivity, is gradually applied to optical   modulator design.   

"'Carrier plasma dispersion effect', 'slow   light effect' and 'DC Kerr effect' have their strengths and weaknesses.   Individually, none of them could help achieve our target. So, we carefully   studied the three effects and found they well supplement each other in terms   of modulation. We are the first to propose taking their respective strengths   to design a new type of modulator that outperforms all existing   products," said YU Hui. 

Once the design idea was defined, the new modulator is just a matter of time. The first edition opponent was soon taped out in late 2020. Unfortunately, the test result was not good. "The 'slow light effect' was not significant, thus the performance improvement was not good. There was a long way to the theoretical value," said XIA Penghui. Again and again, the project team deliberated the blueprint and reconsidered the three effects, without missing any possible improvement. "We redid modeling simulation to define the improvement scope and then set different parameters for tests, in order to work out the best design plan." Hard work paid back. In late 2021 when they had the latest tape-out test result, the project team knew they'd found the "optimal solution" for integrating the three effects. The test proved this new optical modulator, compared with previous products, boasts 100% improvement in bandwidth, 30% reduction in terms of insertion loss, and great improvements in terms of modulation speed and linearity degree, suggesting world-leading comprehensive performance.

YU Hui said recognition by the academic circle is one of their "fast outputs". For the next step, they'll probe into more possible performance improvements. That will be a sub-project of 800G optical transceiver chip and light engine technology - 1.6T optical transceiver chip and light engine technology. "The new type optical modulator will be first applied to ZJ Lab's Center for Computing and Data. Wish it could further improve the Lab's data transmission and computing strength, help the Lab's study on intelligent computing and support state strategies," said YU Hui.