The need for alternatives to conventional memory is becoming more urgent as data-intensive computing expands. Electronic memory has benefited from decades of engineering, but shrinking components further creates difficult trade-offs involving power consumption, switching speed, heat dissipation and physical scaling. Conventional computer architectures also separate memory from processing, a design that forces data to travel back and forth between storage and logic. This so-called von Neumann bottleneck can dominate the energy and time required for machine-learning workloads. Optical computing offers a different route: photons can carry information at high bandwidth and with low propagation delay, while multiple signals may be processed in parallel. Yet optical systems still require memory-like elements capable of retaining information, changing their response in controlled increments and being read without destroying the stored state. The new plasmonic proposal addresses these requirements by using a nanoscale material transition to encode optical weights.

At the heart of the device is GST, a phase-change material whose atomic arrangement can be reversibly altered by short optical pulses. In its amorphous state, the atoms lack the long-range order found in a crystal. A suitable heating pulse can induce crystallization, changing the material’s electrical and optical properties. A stronger, shorter pulse can then melt and rapidly quench the material, returning it to an amorphous configuration. These transformations are non-volatile: after the optical stimulus disappears, the material remains in its new state until another programming pulse is applied. In the proposed memory, crystallization is associated with a higher refractive index and lower electrical resistance, while amorphization produces the opposite trend. Because the refractive index determines how light interacts with the nanostructure, the physical phase of GST can be translated into a measurable transmission level.

The programming process depends on the different thermal requirements of the two transitions. The study estimates that crystallizing GST requires about 7.5 picojoules, delivered by a 150-milliwatt pulse lasting 150 nanoseconds. Amorphization requires approximately 2.2 picojoules from a more powerful 110-milliwatt pulse lasting 20 nanoseconds. These figures describe the energy and pulse conditions used for the proposed operating scheme, rather than proving that a complete commercial device has already been fabricated. The distinction matters because phase-change memories face a familiar engineering compromise: higher pulse energies can accelerate switching but increase thermal stress, while repeated cycling can gradually degrade the material or surrounding structure. Precise control is particularly important when a memory is expected to hold many intermediate states instead of simply switching between binary zero and one.

The optical confinement comes from a metal–insulator–metal, or MIM, plasmonic waveguide. In this geometry, light interacts with conducting metal layers separated by a dielectric region, allowing electromagnetic fields to be compressed far below the scale possible in ordinary dielectric waveguides. The proposed design adds two coupled rings containing GST and uses silver waveguides to shape the resonant response. When light at the device’s operating wavelength—1814 nanometers—enters the structure, the local electromagnetic field is strongly influenced by the phase and refractive index of the GST. Small changes in the material can therefore produce comparatively large changes in transmission. This is the central advantage of plasmonics for memory: it can concentrate light into extremely small volumes, enabling compact devices and strong light–matter interaction. The cost is that metals introduce optical absorption, fabrication becomes demanding and heat must be carefully managed.

Rather than limiting the cell to two states, the researchers map GST conditions onto quantized transmission levels. An 8-bit memory can, in principle, distinguish 256 states, allowing one physical cell to represent a finely adjustable synaptic weight. In an optical neural network, such a weight determines how strongly one signal contributes to another, much as the strength of a biological synapse influences the transmission of information between neurons. A multi-level photonic element could therefore perform more computation in place, reducing the number of separate components needed for multiplication and accumulation operations. The device is not described as a biological neuron, nor does it reproduce the full complexity of learning in the brain. Instead, it supplies a programmable optical transfer function that can be assigned a numerical weight. The non-volatile nature of GST would allow those weights to remain available when the programming light is removed.

The reported simulated performance is unusually strong compared with many earlier plasmonic-memory concepts. The optimized structure produces an optical contrast of 87 percent between relevant states and an extinction ratio of 44.04 decibels. Extinction ratio measures how effectively a device distinguishes high- and low-transmission conditions; a larger value generally indicates cleaner separation during readout. The reported insertion loss is 0.60 decibels for logic state one and 45.60 decibels for logic state zero, although the latter value reflects the strongly attenuated state rather than a low-loss transmission path. The design also predicts a readout time of 62 femtoseconds. Such a response is associated with the optical resonance and propagation dynamics of the modeled structure, not necessarily with the slower thermal process used to rewrite GST. Writing and reading are therefore distinct operations: the material may require nanosecond-scale energy pulses to change phase, while a stored state can be interrogated optically on a much shorter timescale.

The proposed cell also includes features intended to make it more practical for integrated photonics. The researchers outline a five-stage back-end-of-line CMOS-compatible fabrication route with a maximum process temperature of 200 degrees Celsius. Keeping the thermal budget low is important because photonic memory elements may eventually need to be fabricated alongside electronic circuits and existing interconnects. The analysis further indicates that dimensional deviations of up to plus or minus 5 nanometers cause only minimal changes in performance. That tolerance could be valuable because nanoscale fabrication inevitably introduces variations in ring dimensions, gaps, layer thicknesses and alignment. Still, tolerance in a numerical design does not eliminate the challenges of real manufacturing. Silver can be chemically and thermally vulnerable, nanoscale GST layers must be deposited uniformly, and the optical response of coupled resonators can be sensitive to roughness and defects. Experimental fabrication and cycling tests will be needed to determine whether the predicted characteristics survive outside the simulation environment.

The study places its design within a rapidly developing field of non-volatile optical memories. Earlier concepts have used GST nanoantennas, ring resonators, plasmonic chains, photonic-crystal waveguides and even photochromic molecules. Reported devices have demonstrated different combinations of optical contrast, switching energy, footprint and state density, but no single architecture has solved every problem. Some offer fast switching but suffer from loss or demanding fabrication; others provide strong contrast but occupy larger areas or require complex thermal control. The double-ring MIM design attempts to combine several desirable properties in one cell: small size, multi-bit storage, non-destructive optical readout, external optical programmability and compatibility with neuromorphic weighting. Its stated density of 4.56 bits per square micrometer is a particularly eye-catching feature, but practical system performance will also depend on how cells are connected, how heat spreads through dense arrays, how often states can be rewritten and how reliably adjacent transmission levels can be distinguished in the presence of noise.

The immediate significance of the work is therefore less a finished optical computer than a blueprint for a compact photonic memory element. If experimental devices confirm the predicted contrast, speed and fabrication tolerance, arrays of these cells could act as programmable weight banks for optical neural networks, allowing computation and storage to occur in the same physical platform. Such systems might eventually process high-bandwidth signals for machine learning, communications or sensing without converting every operation into the electronic domain. Major obstacles remain, including fabrication at scale, thermal crosstalk, material fatigue, calibration of 256 analog-like states and the integration of efficient optical sources and detectors. Even so, the proposal highlights why phase-change plasmonics has become a prominent candidate for next-generation neuromorphic hardware: it links a persistent nanoscale material state to a controllable optical response, potentially turning memory from a passive data store into an active computational component.

Optical computing utilizes light waves instead of electrical signals to process information, which opens the door to extraordinary speed and efficiency. Unlike electronic systems constrained by the physical limits of semiconductors, optical computing offers the possibility of parallel processing and lower latency. However, the integration of optical components that maintain stable and coherent light beams presents considerable technical challenges. Researchers have recognized optical diffraction operators, which act akin to computational plates that manipulate light, as a particularly promising avenue for enhancing the feature extraction process.

A groundbreaking advancement in this field comes from a team at Tsinghua University, led by Professor Hongwei Chen. They have introduced an innovative optical feature extraction engine known as OFE² designed to tackle the obstacles faced in optical computations. This engine is reported to perform feature extractions at unprecedented speeds, making it a viable candidate for applications across various practical domains. Published in the academic journal Advanced Photonics Nexus, their research outlines the specific capabilities and underlying mechanics that make OFE² a notable breakthrough in the realm of optical computing.

The OFE² engine incorporates an exceptional data preparation module that plays a critical role in generating high-speed optical signals. This is particularly vital for behind-the-scenes optical cores functioning in a coherent light environment. The conventional reliance on fiber optic components for power splitting tends to introduce phase perturbations, which challenges consistent output. The research team has ingeniously developed an integrated on-chip system featuring tunable power splitters that alleviate these issues by enabling precise delays and stream management.

The magic unfolds as optical waves transition through the specialized diffraction operator, triggering a mathematical modeling akin to matrix-vector multiplication that facilitates feature extraction. By manipulating the phase of the input lights through an adjustable integrated phase array, the OFE² engine successfully directs diffracted light into specific output paths, consequently allowing it to track variations in input signals over time. This cutting-edge mechanism enables the system to discern crucial features from the continuous flow of input data.

Operating at a remarkable frequency of 12.5 GHz, the OFE² engine boasts a latency of less than 250.5 picoseconds, establishing a new benchmark for optical computing systems. This performance eclipses existing implementations and opens the door to enhanced real-time decision-making capabilities. The implications are profound, with potential impacts spanning various sectors including healthcare, finance, and image processing where rapid data analysis is paramount.

The practical demonstrations conducted by the research team confirm the versatility of OFE² across multiple tasks. For instance, in image processing applications, OFE² displayed a remarkable ability to extract edge features and generate distinctive feature maps that highlight ‘relief and engraving.’ This advancement in image classification could revolutionize sectors such as medical imaging, enabling more accurate diagnostics through tools that utilize optical computing technologies.

Similarly, in a digital trading environment, the OFE² engine was tested on time-series market data. Traders input real-time price signals into the system, which after appropriate training, outputs actionable trading signals. This capability allows for immediate buy or sell decisions based on optimized strategies, facilitating a process that could ultimately yield consistent profitability while maintaining a significant edge in speed thanks to optical processing.

The transition from traditional electronic computation towards photonic computing marks a shift in how we approach data-intensive tasks. The energy efficiency and speed advantages offered by optical systems like OFE² suggest exciting possibilities for the upcoming generation of real-time AI applications. Professor Chen emphasizes that this development promotes the necessary acceleration and efficiency required for advanced applications in digital finance, healthcare, and beyond.

Analyzing the overarching transformation, the research underscores a growing trend where computational barriers are rapidly dissipating thanks to innovative methods in integrated optical systems. By engaging with industrial stakeholders, the team at Tsinghua University aims to elevate the practical deployment of these technologies in sectors reliant on high-demand computational tasks.

In conclusion, the advancements encapsulated in the OFE² optical feature extraction engine signify a substantial leap toward achieving enhanced capabilities in AI processing tasks. The profound advantages of light over electronic components not only pave the way for faster computations but also herald a new era marked by lower energy demands. Exciting future collaborations promise to produce viable solutions to the computational challenges faced in data-rich environments across numerous applications while ensuring sustainable practices.

As researchers continue to hone their techniques and bridge the gap between complex data processing and the limitations of current technologies, optical computing stands at the forefront of the next wave of computational innovation. This momentum demonstrates how the synergy between AI and photonics could redefine our capabilities for real-time, responsive systems that drive the future of technology forward.

The ongoing exploration into and application of optical computing brings with it promising avenues for research and development. The path ahead not only raises the potential for groundbreaking advancements across various fields but also reorients our approach to understanding and utilizing the physics of light in practical computing applications, thus forever altering the landscape of technology.

Subject of Research: Optical Computing and Feature Extraction
Article Title: High-speed and low-latency optical feature extraction engine based on diffraction operators
News Publication Date: 8-Oct-2025
Web References: Link to the full text
References: Advanced Photonics Nexus, DOI 10.1117/1.APN.4.5.056012
Image Credits: Credit: H. Chen, Tsinghua University

Keywords

Optical computing, Optoelectronics, Data analysis.

The core innovation lies in the device’s unique dual-mode operational capability. Traditionally, photodetectors rely on external power sources and have limited angular sensitivity, constraining their functionality in practical scenarios. This newly reported device overcomes these limitations by delivering wide-angle, self-driven photodetection, while simultaneously enabling efficient neuromorphic computing operations at ultralow power consumption levels. Such an integration is unprecedented, elegantly combining light detection with brain-inspired computation on a transparent substrate that allows for seamless embedding in various environments without visual interference.

At the heart of this technological breakthrough is an intricate design that employs transparent materials engineered to achieve both photodetection and neuromorphic functionalities simultaneously. By leveraging carefully tuned semiconductor components layered within an optically clear matrix, the researchers succeeded in fabricating a device that can respond to light stimuli from virtually any direction—accomplishing what they term 360° quasi-omnidirectional photodetection. This capability dramatically expands the spatial coverage of light sensing beyond conventional planar devices, ensuring consistent performance regardless of illumination angle.

Moreover, the device operates in a truly self-driven mode. In other words, it harnesses the incident light not only as the stimulus to detect but also as the sole energy source driving its operational processes. This attribute eliminates the reliance on battery power or external electrical sources, making the photodetector highly suitable for sustainable and autonomous applications. The energy harvested from ambient light is efficiently converted into electrical signals that subsequently feed into the neuromorphic computing elements embedded within the device.

Neuromorphic computing, inspired by the human brain’s neural architecture, represents a paradigm shift in information processing by mimicking synaptic functionalities at a hardware level. The device integrates synaptic transistors that emulate neuronal behavior, allowing it to process and interpret optical signals in situ—reducing latency and power consumption while improving computational efficiency. This synergy between sensing and processing within a single transparent entity eliminates the need for separate components and complex wiring, simplifying device architecture and enhancing scalability.

The ultralow-power nature of this neuromorphic unit is particularly impressive. By utilizing novel materials with low threshold voltages and energy-efficient switching dynamics, the researchers achieved computation at power consumption levels orders of magnitude below traditional processors. This feature is crucial for deploying electronics in portable or remote scenarios where power budgets are severely constrained or where perpetual operation on harvested energy is paramount.

Crucially, the transparent quality of the device does not compromise its performance. Conventional electronic devices often introduce opacity and bulky form factors, limiting their integration into applications requiring aesthetic discretion or unhindered light transmission, such as augmented reality glasses or smart windows. This transparent device maintains high optical clarity, ensuring it can be layered onto or embedded within surfaces and displays without detracting from their appearance or function.

The fabrication process adopted in this research combines advanced materials synthesis with precision layering techniques. The semiconductor layers responsible for light absorption and photogeneration are carefully deposited to maximize responsivity while maintaining transparency. The neuromorphic components, composed of emerging two-dimensional materials and oxide semiconductors, are integrated using state-of-the-art lithographic methods that preserve the delicate balance between optical and electrical functionality.

An exhaustive characterization of the device reveals its robust performance over a wide spectral range and diverse angles of incidence. The photodetection capability remains stable and sensitive even under varying environmental lighting conditions, a testament to the device’s adaptability and reliability. Furthermore, the synaptic behavior exhibits long-term plasticity and rapid response times, essential traits for practical neuromorphic applications requiring learning and adaptation.

Potential applications for this dual-mode device span a vast technological landscape. In the realm of wearable health monitors, the device could enable continuous, self-powered sensing of environmental light factors coupled with on-site processing for real-time feedback. In robotics and autonomous systems, it could underpin intelligent vision systems that adaptively filter and interpret optical signals with minimal energy overhead. Moreover, integration into building materials like transparent facades could allow smart windows to dynamically respond to light stimuli and perform local data processing, contributing to energy-efficient architectures.

This innovation stands at the confluence of multiple research frontiers—optoelectronics, neuromorphic engineering, and materials science—showcasing what interdisciplinary collaboration can achieve. Its dual-mode operation, self-sufficiency, and transparency collectively push the boundaries of what is currently possible in integrated photodetection and computation systems. The work lays a solid foundation for future devices that could seamlessly blend into everyday objects, smart environments, and intelligent interfaces with minimal energy and visual cost.

To harness the full commercial and societal impact of this technology, further developments are anticipated. Scaling the device to larger areas, enhancing durability under diverse environmental stresses, and incorporating complex neuromorphic learning algorithms will be pivotal. Additionally, exploring new transparent materials with even greater carrier mobilities and synaptic efficiencies could amplify the device’s capabilities, paving the way toward fully autonomous, intelligent, and visually unobtrusive sensors.

In conclusion, the advent of this dual-mode transparent photodetector and neuromorphic computing device represents a bold stride forward. It unites wide-angle light sensing and brain-like computation within an ultralow-power, self-supporting, and visually transparent architecture, setting the stage for revolutionary applications across multiple domains. As the research community builds upon these findings, the dream of ambiently powered, intelligent, and invisible electronics edges tantalizingly closer to reality.

Subject of Research: Dual-mode transparent device combining 360° quasi-omnidirectional self-driven photodetection and ultralow-power neuromorphic computing

Article Title: A dual-mode transparent device for 360° quasi-omnidirectional self-driven photodetection and efficient ultralow-power neuromorphic computing

Article References:
Jiang, M., Zhao, Y., Liu, T. et al. A dual-mode transparent device for 360° quasi-omnidirectional self-driven photodetection and efficient ultralow-power neuromorphic computing. Light Sci Appl 14, 273 (2025). https://doi.org/10.1038/s41377-025-01991-y

Image Credits: AI Generated

DOI: https://doi.org/10.1038/s41377-025-01991-y