Programmable photonic waveguide arrays: opportunities and challenges

TL;DR

The paper addresses the scalability and loss limitations of conventional mesh-based photonic processors by proposing programmable waveguide arrays (PWAs) that realize 'always-on' Hamiltonians through continuous coupling and local tuners. It develops modeling frameworks (including a tri-diagonal Hamiltonian with $H$ and unitary $U = e^{-iHL}$) and analyzes material platforms, layout constraints, and control strategies, while surveying applications in universal multiport interferometry, quantum computing, photonic neural networks, topological photonics, and nonlinear optics. A key result is the universality of cascaded PWAs for arbitrary unitaries, with decomposition bounds and practical section-count reductions demonstrated in recent work (e.g., $n = O(N^3 k)$ with error $O(N^3/k)$ for general cases and a 4-section requirement for $U(2)$). The paper also outlines four critical benchmarks (scalability, per-section control, reporting metrics, and deployment gaps) and highlights model-driven and data-driven strategies to characterize and control PWAs, indicating a path toward compact, high-fidelity, reconfigurable photonic processors. Overall, PWAs promise lower optical loss, smaller footprints, and versatile functionality across quantum simulation, DAQC, and neuromorphic photonics, with clear research directions to realize practical, large-scale devices.

Abstract

The rising complexity of photonic applications, ranging from quantum computing to neuromorphic processing, has driven the demand for highly programmable and scalable photonic integrated circuits. While mesh-based architectures built from Mach-Zehnder interferometers have enabled significant advances, their reliance on beam splitting and light bending introduces optical loss, fabrication challenges, and scalability bottlenecks. Continuously lateral-coupled integrated waveguide arrays (WAs), by contrast, offer compact systems with no direct free-space analogs, but their static nature has limited their utility. Recently, programmable waveguide arrays (PWAs) have emerged as a promising alternative, combining the Hamiltonian richness of WAs with tunable control. This perspective outlines the conceptual foundations, recent progress, and future potential of PWAs across quantum simulation, photonic neural networks, topological photonics, and nonlinear optics. We examine the theoretical and practical challenges of modeling, fabrication, and control, and propose PWAs as a next-generation architecture for compact, reconfigurable photonic processors.

Figures (4)

• Figure 1: Increasing complexity of programmable photonic circuits for advanced applications, quantified by the number of tunable phase shifters per chip. The development trend is indicated by an exponential function fitting represented by dashed lines. Although recent milestones have been achieved primarily with MZI mesh networks, the continued exponential growth in complexity highlights the demand for alternative architectures. PWAs may provide a more scalable and compact solution to continue this trend. Notable milestones in the field of optical quantum computing (OQC) include the first static CNOT gate (a static device with 0 tunable components in 2008) politi2008silica, the first programmable quantum processor (8 tunable components in 2011) shadbolt2012generating, the first universal linear circuits (30 tunable components in 2015) carolan2015universal, the first large-scale circuit (93 tunable components in 2018) wang2018multidimensional, and a recent large-scale circuit (216 tunable components in 2023) bao2023very. Key demonstrations in the field of photonic neural networks (PNN) include the first demonstration of unscrambling light (12 tunable components in 2017) annoni2017unscrambling, the first demonstration of deep learning (24 tunable components in 2017) Shen_2017, the first implementation of complex-valued neural networks (56 tunable components in 2021) zhang2021optical, the first realization of backpropagation training for deep learning (30 tunable components in 2023) doi:10.1126/science.ade8450, the first demonstration of a fully-integrated coherent optical neural network (169 tunable components in 2024) bandyopadhyay2024single.
• Figure 2: Schematics of an integrated PWA architecture. (a) The architecture incorporates photon sources and detectors with potential for integration as well as cascaded $n$ multisection PWAs. The black lines represent optical waveguides, and the orange lines indicate control connections. Each section of PWA has independent tuners and implements a $N$-dimensional unitary transformation matrix $U_k(\vec{v}_k)$, where $\vec{v}_k$ is the control parameter applied to the tuners in the $k$-th section. The entire architecture implements an overall unitary transformation matrix $U$. For a given input state $\ket{\psi_\text{in}}$, the output state is calculated as $\ket{\psi_\text{out}}=U\ket{\psi_\text{in}}$. (Note: This architecture can be used for both classical and quantum light.) (b) Cross-section of an example PWA featuring three waveguides (black areas indicate the optical modes) based on the device recently demonstrated in Ref Yang_2024yang2024programmablesingle. The device is fabricated using annealed proton-exchange waveguides on x-cut lithium niobate lenzini2015anisotropic. Gold microelectrodes functioning as microtuners were patterned on top of the silicon dioxide buffer layer above the waveguides. Electric fields are generated by applying control pulses to the electrodes and are confined by the shielding effect of neighboring electrodes, effectively eliminating crosstalk 10011218. This ensures precise control of individual Hamiltonian terms and their corresponding unitary transformations.
• Figure 3: Evolution of waveguide array systems. (a) The first experimental demonstration of a directional coupler fabricated in GaAs in 1973 10.1063/1.1654468. (b) The first demonstration of a nonlinear waveguide array fabricated on an AlGaAs platform in 1997 millar1997nonlinear. (c) The two-dimensional optically induced waveguide array in a bulk photorefractive crystal in 2002 PhysRevE.66.046602. (d) The two-dimensional helical waveguides fabricated using femtosecond laser writing technology in 2013 rechtsman_2013. (e) The fully electro-optically programmable waveguide array fabricated in bulk lithium niobate used in recent demonstrations youssry2024experimentalYang_2024yang2024programmablesingle (Red light shines through the chip to highlight the device. Credit to photographer: James Thomas).
• Figure 4: Cross-section of a two-dimensional PWA concept design. The optical modes are arranged in a hexagonal pattern rechtsman_2013, as indicated by dashed lines. Four tuners between neighboring optical modes control their coupling coefficients, while six tuners placed around an optical mode adjust their propagation coefficients. The substrate is made of an ideal isotropic material, such as one with a homogeneous high electro-optic coefficient in all directions.