Optical Parametric Amplification (OPA) Research Papers
Intense optical parametric amplification in dispersion-engineered nanophotonic lithium niobate waveguides
Authors: Luis Ledezma, Ryoto Sekine, Qiushi Guo, Rajveer Nehra, Saman Jahani, and Alireza Marandi
Publisher: Optica Publishing Group
Executive Summary
Optical amplification is a fundamental capability in modern photonics, enabling applications across communications, sensing, quantum technologies, and optical computing. While integrated photonic platforms have made significant progress in recent years, achieving high optical gain on a chip has remained a major challenge. Existing approaches often face tradeoffs between gain, bandwidth, efficiency, and noise performance, limiting their ability to support next-generation photonic systems.
This research demonstrates a breakthrough in integrated optical amplification using a dispersion-engineered lithium niobate nanophotonic waveguide. By combining quasi-phase matching with advanced waveguide engineering, the device achieves intense optical parametric amplification through second-order nonlinear interactions. The resulting amplifier delivers phase-sensitive gains exceeding 50 dB/cm over a broad spectral range, with gains surpassing 100 dB/cm across more than 600 nm of bandwidth around the 2 μm wavelength region. These performance levels represent a substantial improvement over previously demonstrated integrated nonlinear amplification techniques.
A key innovation of this work is the careful engineering of the waveguide to maximize the interaction between ultrashort optical pulses while minimizing dispersion and group velocity mismatch. This enables highly efficient energy transfer from the pump light to the amplified signal without relying on resonant cavities, allowing the device to simultaneously achieve high gain and exceptionally broad bandwidth. The architecture also supports both phase-sensitive and phase-insensitive amplification, making it suitable for a wide variety of classical and quantum photonic applications.
The researchers further demonstrated that the amplifier is capable of amplifying quantum vacuum fluctuations into measurable optical signals, confirming operation in an ultra-high-gain regime. This capability highlights the device's potential not only as an optical amplifier but also as a powerful source for nonlinear optical generation and quantum photonic systems.
By combining record optical gain, broad bandwidth, and compact chip-scale integration, this work establishes a new benchmark for integrated optical amplifiers. The demonstrated platform significantly expands the capabilities of thin-film lithium niobate and provides a practical foundation for next-generation photonic technologies that require efficient, low-noise optical amplification.
This breakthrough opens new opportunities for integrated quantum photonics, mid-infrared light sources, ultrafast frequency combs, laser ranging, optical communications, and photonic computing. As demand for faster and more energy-efficient optical systems continues to grow, high-performance integrated optical amplifiers such as this represent a critical building block for the future of scalable photonic infrastructure.
All-optical ultrafast ReLU function for energy-efficient nanophotonic deep learning
Authors: Gordon H.Y. Li , Ryoto Sekine , Rajveer Nehra , Robert M. Gray , Luis Ledezma , Qiushi Guo and Alireza Marandi
Publisher: De Gruyter Brill
Executive Summary
As artificial intelligence models continue to grow in size and complexity, the performance of traditional electronic hardware is increasingly constrained by power consumption, memory movement, and computational latency. Optical neural networks (ONNs) have emerged as a promising alternative, leveraging light to perform the massive linear computations required for deep learning with exceptional speed and energy efficiency. However, one critical component has remained missing: an all-optical nonlinear activation function capable of operating at the speed and efficiency required for practical AI systems.
This research demonstrates the first all-optical implementation of one of deep learning's most widely used activation functions—the Rectified Linear Unit (ReLU)—using a thin-film lithium niobate nanophotonic waveguide. By exploiting ultrafast second-order nonlinear optical processes, the device performs nonlinear activations entirely in the optical domain, eliminating the need for optical-to-electrical-to-optical conversions that have traditionally limited the performance of photonic neural networks.
The demonstrated device faithfully reproduces the behavior of the ReLU function while operating with signal energies in the femtojoule range and response times of approximately 75 femtoseconds. Together, these characteristics produce a state-of-the-art energy-time product that surpasses previous optical implementations and significantly improves upon the performance of today's digital electronic hardware. The same architecture can also implement other widely adopted activation functions, including ELU and GELU, simply by adjusting the optical bias conditions, making the platform compatible with many of today's most advanced deep learning models, including Transformer-based architectures.
Beyond demonstrating an individual optical component, this work addresses one of the final fundamental building blocks required for fully optical neural networks. Because the device accurately reproduces standard activation functions used throughout modern AI, existing pretrained neural network models can be adapted with minimal retraining rather than requiring entirely new architectures. This significantly lowers the barrier to deploying photonic AI hardware within existing machine learning workflows.
By combining ultrafast operation, exceptional energy efficiency, and compatibility with scalable thin-film lithium niobate manufacturing, this research establishes a practical path toward fully integrated optical AI accelerators. The demonstrated technology provides a foundation for future photonic processors capable of executing deep learning workloads at speeds and efficiencies that extend beyond the limits of conventional electronic computing, enabling new possibilities for large-scale AI inference, scientific computing, communications, and real-time intelligent systems.
Few-cycle vacuum squeezing in nanophotonics
Authors: Rajveer Nehra, Ryoto Sekine, Luis Ledezma, Qiushi Guo, Robert M. Gray, Arkadev Roy, and Alireza Marandi
Publisher: Science
Executive Summary
Squeezed states of light are one of the fundamental resources of quantum information science, enabling applications such as quantum computing, secure communications, precision sensing, and quantum-enhanced metrology. Although these quantum states have traditionally been generated using large laboratory optical systems, creating and measuring high-quality squeezed light on an integrated photonic chip has remained a major challenge. Existing approaches often suffer from limited bandwidth, significant optical losses, and measurement techniques that restrict scalability.
This research demonstrates a fully integrated lithium niobate nanophotonic platform capable of both generating and measuring squeezed quantum states on the same chip. By combining two dispersion-engineered optical parametric amplifiers (OPAs), the device first generates a squeezed vacuum state and then performs an all-optical measurement through high-gain, phase-sensitive amplification. This approach eliminates many of the limitations associated with conventional photodetection, allowing quantum states to be measured with significantly greater tolerance to optical losses while preserving their quantum properties.
The platform achieved up to 4.9 dB of directly measured squeezing across more than 25 terahertz of optical bandwidth, supporting ultrashort quantum states lasting only a few optical cycles. After accounting for system losses, the inferred squeezing exceeds 10 dB—a performance level considered sufficient for many fault-tolerant continuous-variable quantum computing architectures. Achieving this level of performance on a compact integrated chip represents an important milestone toward practical, scalable quantum photonic systems.
Beyond demonstrating high-quality squeezed light, the research introduces a new method for all-optical quantum measurement that overcomes one of the longstanding bottlenecks in integrated quantum photonics. Traditional balanced homodyne detection is highly sensitive to optical losses and limited in bandwidth, whereas the demonstrated phase-sensitive amplification technique enables broadband, loss-tolerant measurements that can fully leverage the ultrafast nature of integrated photonic devices.
By combining broadband quantum light generation, integrated measurement, and the scalability of thin-film lithium niobate, this work establishes a practical foundation for next-generation quantum photonic processors. The demonstrated architecture enables compact, wafer-scale quantum systems capable of supporting continuous-variable quantum computing, quantum communications, precision sensing, and ultrafast quantum information processing. As integrated quantum technologies continue to mature, this research represents a significant step toward scalable, high-performance quantum photonic platforms that can move beyond laboratory demonstrations into real-world applications.
Femtojoule femtosecond all-optical switching in lithium niobate nanophotonics
Authors: Qiushi Guo, Ryoto Sekine, Luis Ledezma, Rajveer Nehra, Devin J. Dean, Arkadev Roy, Robert M. Gray, Saman Jahani & Alireza Marandi
Publisher: Nature Photonics
Executive Summary
All-optical switching is a fundamental capability for the future of photonic computing, communications, and signal processing, enabling information to be processed entirely with light rather than through slower and more energy-intensive electronic components. However, realizing practical all-optical switches on integrated photonic chips has remained a significant challenge. Most existing platforms rely on inherently weak optical nonlinearities, requiring large amounts of energy, resonant cavities, or complex material integration that limit performance, scalability, and speed.
This research demonstrates a breakthrough in integrated photonics by achieving cavity-free all-optical switching using thin-film lithium niobate nanowaveguides. By combining dispersion engineering with quasi-phase matching, the researchers harnessed the material's strong second-order nonlinear optical response to create an ultrafast nonlinear optical splitter capable of switching light with exceptionally low energy and without the need for resonant structures. This architecture simplifies device design while overcoming many of the limitations that have constrained previous integrated optical switching technologies.
The demonstrated device achieves switching energies as low as 80 femtojoules with an ultrafast switching time of approximately 46 femtoseconds, representing one of the fastest and most energy-efficient all-optical switching demonstrations reported in integrated photonics. Together, these results produce a record-low energy-time product, highlighting the platform's ability to perform optical switching at unprecedented speed while minimizing power consumption.
Unlike conventional approaches that require optical cavities to enhance weak nonlinear interactions, this cavity-free design supports broadband operation, avoids resonance-related constraints, and provides a more scalable foundation for large photonic circuits. The use of thin-film lithium niobate further enables compatibility with advanced integrated photonic manufacturing while leveraging one of the strongest nonlinear materials available for chip-scale devices.
By combining ultrafast response, exceptional energy efficiency, and compact integrated implementation, this work establishes a new benchmark for nonlinear optical switching in nanophotonics. The demonstrated technology provides a practical foundation for next-generation all-optical systems that can process information at the speed of light while dramatically reducing latency and energy consumption.
This breakthrough opens new opportunities for optical computing, photonic neural networks, high-speed communications, ultrafast signal processing, and integrated light sources. As demand for faster and more energy-efficient computing infrastructure continues to grow, cavity-free all-optical switching represents a critical building block for future photonic processors and scalable optical information processing systems.