CPM Seminar

A two-dimensionally programmable photonic chip for classical and quantum information processing

Tatsuhiro Onodera

Cornell University

Integrated photonics plays a pivotal role across several domains, including optical networking, light generation, and photonics-based quantum computing.  However, its utility has been limited by lengthy design and fabrication cycles, which result in chips that are generally single-purpose and prone to fabrication errors.  In this talk, I will introduce our recent work on developing a programmable integrated photonics platform [1] that is well-suited to tackle these challenges.  Our approach utilizes the electro-optic effect in a Lithium Niobate planar waveguide and a photoconductive film to develop a 2D-programmable planar waveguide.  By shaping the light incident on the top of the chip, we can make in-situ, real-time modifications to its spatial refractive index distribution, n(x,z).  The chip has a large number of parameters (~10,000); to optimize/train these parameters for a given task, we leveraged a recently developed physics-aware machine learning algorithm [2].  By directly training them, we reconfigurably realize different photonic devices such as lenses and waveguides, and leverage the complex multimode wave propagation of light to perform machine learning computations.  Finally, I will present preliminary results on extending the platform to spatially program the nonlinear optical susceptibility, χ²(x,z). This extension opens up possibilities for programmable nonlinear and quantum photonic applications.

[1] T. Onodera, M.M. Stein, et al., arXiv:2402.17750 (2024)
[2] L.G. Wright, T. Onodera, et al., Nature 601, 549 (2022)