Emerging Photonic Technologies
Emerging photonic technologies represent the research frontier of light-based science and engineering, encompassing innovations that change how systems generate, manipulate, transmit, and detect light. These technologies move beyond conventional optoelectronics to exploit new physical phenomena, novel materials, and unconventional architectures, with the goal of advancing computing, communications, sensing, energy harvesting, and environmental monitoring.
The convergence of nanotechnology, materials science, and photonics has enabled fine control over light across scales ranging from individual emitters to macroscopic systems. Modern fabrication produces photonic structures with features smaller than the wavelength of light, allowing electromagnetic waves to be shaped in ways that bulk optics cannot achieve. At the same time, growing environmental awareness is driving sustainable approaches that reduce the ecological footprint of photonic devices while applying photonics to environmental benefit.
This category surveys developments that are likely to define the next generation of photonic systems, from brain-inspired computing architectures and reconfigurable photonic processors to topologically protected light propagation and environmentally responsible design.
Articles in This Category
Fundamental Concepts
Nanophotonics and Metamaterials
Nanophotonics operates at length scales where near-field interactions and confinement effects dominate, enabling control of light propagation, emission, and absorption far below the wavelength of light. Metamaterials and metasurfaces extend this control through subwavelength structures with engineered electromagnetic responses not found in natural materials, including negative effective refractive index, tailored dispersion, and near-perfect absorption. These engineered media support devices with capabilities that conventional optics cannot reach, from flat lenses that replace bulk optical assemblies to narrowband absorbers for sensing and energy harvesting.
Metasurface optics have begun to move from laboratory demonstrations into commercial components, particularly beam shapers, polarization optics, and compact lenses for structured-light and depth-sensing modules. Broadband imaging remains harder: a single flat lens imposes strong chromatic dispersion, and correcting it across the full visible band without sacrificing aperture size or efficiency is still an active research problem.
Integrated Photonic Circuits
Integrating optical components onto chip-scale platforms follows the trajectory of electronic integrated circuits, promising similar gains in density, performance, and cost at volume. Silicon photonics leverages mature semiconductor manufacturing to build optical systems on platforms compatible with CMOS processes. Complementary material platforms serve different needs. Silicon nitride offers very low propagation loss for filters, delay lines, and passive routing: research-grade single-mode waveguides have reached losses below one decibel per meter, roughly two orders of magnitude better than the decibel-per-centimeter figures typical of silicon strip waveguides. Thin-film lithium niobate provides high-speed electro-optic modulation, with reported Mach-Zehnder modulators exceeding one hundred gigahertz of bandwidth at drive voltages compatible with CMOS electronics. Indium phosphide supplies on-chip optical gain and integrated lasers through its direct bandgap.
No single material does everything well, so heterogeneous integration has become the dominant strategy: wafer bonding, micro-transfer printing, and flip-chip assembly place III-V gain regions, lithium niobate modulators, and silicon or silicon nitride waveguides on a common substrate. Combining electronics and photonics on shared or bonded substrates enables system architectures that draw on the strengths of both domains, with the electronic layer supplying drivers, transimpedance amplifiers, and control logic close to the optical devices.
Nonlinear and Ultrafast Photonics
Nonlinear optical effects enable wavelength conversion, all-optical switching, frequency-comb generation, and signal processing functions central to advanced photonic systems. Ultrafast phenomena on femtosecond timescales support both fundamental study and practical applications, from precision frequency metrology to laser materials processing. Confining light within micro- and nanoscale structures enhances these effects by concentrating optical fields, bringing capabilities that once required tabletop systems onto a chip.
Microresonator frequency combs illustrate the shift. A single continuous-wave pump laser coupled into a high-quality-factor ring resonator can generate a circulating soliton pulse train whose spectrum is a comb of evenly spaced optical carriers, with the spacing set by the resonator free spectral range, typically tens of gigahertz to about a terahertz. One such device replaces a rack of separate lasers in wavelength-division-multiplexed transmission, and the same combs serve optical clocks, dual-comb spectroscopy, and frequency-modulated ranging. Dispersion engineering of the resonator geometry, rather than bulk material choice, determines the comb bandwidth.
Key Research Directions
Neuromorphic and Optical Computing
Neuromorphic photonic systems implement neural-network operations with optical hardware, aiming for high throughput and energy efficiency on the matrix multiplications and nonlinear activations that dominate machine-learning inference. Light signals propagate in parallel through meshes of interferometers and wavelength channels, exploiting the bandwidth and parallelism of optics. More broadly, optical computing targets specific functions where light excels, such as Fourier transforms and linear-algebra operations, while hybrid optical-electronic architectures use optics for data movement and parallel computation and electronics for control, memory, and nonlinear logic.
Terahertz and Mid-Infrared Technologies
The terahertz band, conventionally taken as roughly 0.1 to 10 THz, or wavelengths from about three millimeters down to thirty micrometers, sits between the reach of conventional electronics and that of conventional optics. Together with the adjacent mid-infrared it was long described as the terahertz gap, because neither transistor-based sources nor ordinary semiconductor lasers performed well there. Quantum cascade lasers, photoconductive antennas driven by femtosecond pulses, and nonlinear difference-frequency generation have progressively closed that gap. Terahertz quantum cascade lasers in particular have improved sharply: pulsed devices emitting near 4 THz have operated at heat-sink temperatures around 250 K, which brings compact thermoelectric coolers within reach and removes the need for cryogenic hardware in some instruments.
These wavelengths pass through many materials that are opaque at visible wavelengths, including plastics, ceramics, paper, and clothing, while carrying distinctive rotational and vibrational signatures of molecules. Terahertz time-domain spectroscopy exploits both properties, recovering amplitude and phase in a single measurement and therefore the complex refractive index without a separate reference. Practical systems support security screening, non-destructive testing of composites and coatings, pharmaceutical tablet inspection, and materials characterization. Photon energies in this band are far too low to ionize matter, which distinguishes terahertz inspection from X-ray methods.
Communication is a second driver. Terahertz links promise very wide channels: IEEE 802.15.3d-2017 defines a switched point-to-point physical layer spanning approximately 253 to 322 GHz, with channel bandwidths that scale in multiples of 2.16 GHz up to 69.12 GHz and target data rates as high as 100 Gb/s. Atmospheric absorption, high free-space path loss, and the need for tightly directional beams confine such links to short ranges and largely line-of-sight geometries, so the near-term applications are wireless backhaul, kiosk-style bulk data transfer, and intra-facility connections rather than wide-area coverage.
Sustainable and Green Photonics
Environmental sustainability is becoming a central concern in photonic technology, motivating research into energy-efficient devices, recyclable and biodegradable materials, and circular-economy manufacturing. Photonics also provides tools for environmental protection, from remote sensing of greenhouse gases to monitoring of water quality and pollutants. This dual role, as both a technology to be made sustainable and a means of addressing environmental challenges, makes green photonics a distinctive area of emerging research.
Applications
Artificial Intelligence and Machine Learning
The computational demands of artificial intelligence motivate photonic accelerators that perform the large matrix multiplications underlying neural networks at high speed and low energy per operation. Several device families compete: meshes of Mach-Zehnder interferometers that implement a matrix directly in the phase settings of the mesh, microring weight banks that assign one wavelength channel per input and sum photocurrents at a shared detector, and diffractive networks in which passive layers perform the transformation as light propagates through them. Optical reservoir computers take a different route, leaving a fixed nonlinear optical system untrained and training only a simple electronic readout.
The nearer-term commercial impact of photonics on artificial-intelligence infrastructure lies less in optical arithmetic than in optical data movement. Co-packaged optics and optical chip-to-chip interconnects address the bandwidth and energy cost of moving activations and model weights between accelerators, which is now the binding constraint in large training clusters. Optical computation and optical interconnect draw on the same integration platforms, so progress in one advances the other.
Next-Generation Communications
Emerging photonic technologies address growing demands on communication networks through higher data rates, greater capacity, and new transmission media. Space-division multiplexing carries independent data streams on separate cores or spatial modes of one fiber, lifting capacity past the nonlinear Shannon limit of a single-mode core. Laboratory transmission experiments have exceeded one petabit per second in a multicore fiber built to the standard 0.125-millimeter cladding diameter, which keeps the fiber compatible with existing cabling and connectorization, and have reached roughly 23 petabits per second when multicore and multimode transmission are combined with wideband multiplexing across the S, C, and L bands. These are research demonstrations; deployed systems remain far below such figures, and the practical obstacles are mode-dependent loss, crosstalk between cores, and the cost of the multiple-input multiple-output digital signal processing needed to unscramble coupled modes.
Beyond fiber, free-space optical links provide high-bandwidth connections for applications ranging from data-center interconnects to inter-satellite crosslinks in large constellations, where narrow beams offer both bandwidth and resistance to interception, at the cost of pointing complexity and, for ground links, sensitivity to cloud and turbulence. Visible light communication reuses lighting infrastructure for indoor wireless networking, trading range for spatial reuse and freedom from radio-spectrum congestion.
Advanced Sensing and Imaging
Photonic sensing is reaching new levels of sensitivity, resolution, and functionality through advances in sources, detectors, and computational methods. Single-photon detectors enable imaging in extremely low light and time-resolved measurement with picosecond precision; arrays of single-photon avalanche diodes fabricated in standard CMOS have made photon counting a commodity capability in consumer depth cameras and in lidar and active optical sensing. Computational imaging combines optical hardware with reconstruction algorithms to relax traditional trade-offs among resolution, field of view, and depth of field, allowing lensless and metasurface cameras that would be unusable without their reconstruction step. Spectroscopic and hyperspectral imaging add chemical and material information beyond conventional intensity images, and chip-scale spectrometers built from tunable filters or reconstructive algorithms are shrinking instruments that once occupied a laboratory bench.
Energy and Environment
Photonics contributes to sustainable energy through improved solar-cell efficiency, efficient solid-state lighting, and optical monitoring of energy infrastructure. Environmental applications include atmospheric sensing for climate research, water-quality monitoring, and detection of pollutants and hazardous materials. Developing the underlying photonic technologies with environmentally friendly materials and processes helps ensure that these solutions do not create new environmental burdens.
Engineering Challenges
The gap between a laboratory demonstration and a manufacturable product is wide in photonics, and several obstacles recur across the subtopics in this category.
Conversion Overhead and the Absence of Optical Memory
Optics moves and transforms data efficiently, but converting between the electrical and optical domains costs energy and latency. In an optical accelerator, digital-to-analog converters, modulators, photodetectors, and analog-to-digital converters can consume more power than the optical computation itself, so the advantage appears only when each conversion feeds a large amount of optical work. There is also no practical optical equivalent of static random-access memory: light must be stored electronically or circulated in delay lines, which constrains architectures to feed-forward or partially electronic designs.
Precision, Noise, and Calibration
Analog optical computation is limited by shot noise, thermal noise, detector noise, and component tolerances rather than by clock rate, so effective precision is typically far below the thirty-two-bit floating point of digital hardware. Machine-learning inference tolerates reduced precision better than most workloads, which is why it is the leading target, but training and scientific computation are less forgiving. Large interferometer meshes compound the problem: fabrication variation offsets every phase shifter, and thermal tuning both drifts and consumes power, so programmable photonic processors require calibration routines and often closed-loop monitoring to hold a configuration.
Packaging, Test, and Ecosystem Maturity
Fiber attachment, hermetic packaging, and thermal management frequently dominate the cost of a photonic chip, inverting the economics familiar from electronics, where packaging is a small fraction of unit cost. Optical testing is slower than electrical probing because each device must be aligned to fiber or free-space optics. Process design kits, standardized building blocks, and multi-project wafer services have matured considerably and now make foundry access routine, yet component libraries and modeling tools remain less complete than their electronic counterparts. Some enabling devices carry further constraints of their own, as superconducting nanowire single-photon detectors deliver excellent efficiency and timing resolution but require cryogenic cooling.
Future Directions
The subtopics in this category sit at different points on the path from discovery to deployment, and it is worth separating them. Integrated photonics, microresonator frequency combs, and space-division multiplexing are engineering problems now: the physics is settled, and the remaining work is yield, packaging, cost, and standardization. Programmable photonic meshes and photonic accelerators are at the prototype stage, where the decisive question is whether the system-level gain survives conversion overhead and calibration burden. Topological photonics and much of neuromorphic photonics remain closest to fundamental research, valuable first for what they reveal about robust light transport and unconventional computation, and only later for the devices they may yield.
Two forces cut across all of them. Heterogeneous integration is steadily removing the constraint that a single material must serve every function, and sustainability is becoming a design input rather than an afterthought, shaping material selection, energy budgets, and end-of-life handling. Read together, these topics show where photonics is heading: not toward replacing electronics, but toward a division of labor in which light handles the movement and parallel transformation of information and electronics retains control, memory, and decision logic.