Research: Thin-Film Lithium Niobate Photonic ICs are Scalable, Cost-Effective

Estimated reading time: 2 minutes

Scientists from HyperLight, a leader in the commercialization of lithium niobate (LN) integrated optical circuits, have teamed with Harvard University researchers to achieve a significant technical milestone for photonic integrated circuits (PICs). For the first time, ultrahigh performance LN PICs are demonstrated to be compatible with mass production technologies. LN PICs produced at large scale with high performance and low cost using HyperLight technology, will alleviate bottlenecks in data center and telecommunications networks, while minimizing energy consumption.

In their manuscript entitled “Wafer-scale low-loss lithium niobate photonic integrated circuits,” published in Optics Express on August 17, 2020, the team consisting of Kevin Luke, Prashanta Kharel, Christian Reimer, Lingyan He, and Mian Zhang from HyperLight and Marko Loncar from Harvard, demonstrated monolithic LN PICs fabricated on 4- and 6-inch wafers. Using deep ultraviolet lithography and smooth and uniform etching, the team achieved 0.27 dB/cm optical propagation loss on wafer-scale allowing efficient light guiding on this material.

Unlike traditional circuits, PICs relay information signals through light, not electricity. In integrated electronics, silicon is the dominant and perhaps the best material. In integrated photonics, silicon is also widely used but the performance of silicon photonic devices are limited by this material’s non-ideal properties. On the contrary, LN is commonly used for realization of high performance optical devices due to its superior material properties that allow efficient and high speed conversion of signals from electrical to optical domains. Until recently, all previous LN work was done using slow serial lithography techniques. For LN to reach its true potential, large scale, silicon-like fabrication technology is needed.

“By making high performance LN PICs at scale we demonstrated that these devices are not only limited to high-end applications but allow for mass production at competitive price points. Our disruptive technology will have a very broad impact on performance and cost sensitive technologies such as data center networking, 5G radio communication systems and automotive LIDAR.” said Kevin Luke, lead author, co-founder and Head of Manufacturing of HyperLight. “LN PICs in recent years showed drastically better electro-optic performance than their silicon counterparts, despite decades of optimization of silicon photonics. But LN PICs’ manufacturability fell far behind silicon photonics, which is leveraging all of the existing infrastructure of the CMOS transistor industry. In this work, we show that you really can have the best of both worlds - performance and scaling using only standard CMOS fab tools, making LN a serious candidate for future PICs,” added Luke.

“Large scale integration of ultra-low loss and very fast optical components in LN, demonstrated by HyperLight, will enable novel applications such as reconfigurable and programmable photonic neural networks. Such PICs will allow for unprecedented control of temporal and spectral properties of photons, which is essential for realization of photonic quantum computers,” said Marko Loncar, co-founder and Chief Sciences Advisor of HyperLight and Tiantsai Lin Professor of Electrical Engineering and Applied Physics at John A. Paulson School of Engineering and Applied Sciences, Harvard University.

Previous performance milestones and technical achievements related to LN PICs frequency were achieved with research findings published in Optica (2017), Nature (2018, 2019), Nature Photonics (2019) and Nature Communications.

HyperLight was formed in 2018 and received seed investment from The Engine, MIT’s venture capital firm. HyperLight is pioneering the commercialization of LN PICs and has served many industrial and academic customers with applications spanning datacom, telecom, quantum, sensing and LIDAR in the US and around the world with ultrahigh performance LN PICs since its inception.