An optical delay line is a device that makes a light signal arrive later than it otherwise would, by routing it through extra path length or through a material that slows it down. The concept sounds simple, but the engineering behind it touches radar, medical imaging, telecommunications, and quantum computing. What makes optical delay lines interesting is not just the delay itself but the challenge of producing it precisely, tuning it on demand, and keeping losses low enough that the signal is still useful when it comes out the other end.
How a Light Signal Gets Delayed
Light in a vacuum travels at roughly 300,000 kilometers per second. An optical delay line exploits two basic levers to slow that journey. The first is path length: force the light to travel a longer physical route, and it takes longer to arrive. A simple spool of optical fiber does this, and so does a set of mirrors that bounce a beam back and forth across a gap. The second lever is the refractive index of the material the light passes through. A higher refractive index means a slower effective speed for the light pulse. Some designs combine both approaches, using switchable waveguide segments of different lengths alongside materials whose refractive properties can be tuned electronically or thermally.
In practice, engineers tune the delay by changing the group refractive index of a waveguide, adjusting the dispersion characteristics of that index, or switching between waveguide paths of different physical lengths.1Chinese Optics Letters. Integrated optical delay lines: a review and perspective These three knobs give designers a surprisingly rich set of trade-offs to play with, depending on whether they need a large delay range, fine resolution, fast switching, or some combination of all three.
Why Radar Systems Care About Optical Delays
Phased-array radar antennas steer their beams electronically by feeding each antenna element a slightly time-shifted version of the same signal. Traditionally, those shifts are created with electronic phase shifters. The trouble is that a phase shift is frequency-dependent: set it for one frequency and the beam points in the right direction at that frequency, but at neighboring frequencies the beam drifts off target. Engineers call this “beam squint,” and it gets worse the wider the bandwidth of the radar signal.
Optical true time delay lines solve beam squint by replacing a frequency-dependent phase shift with a genuine time delay that is the same across all frequencies in the signal’s bandwidth. The radar’s microwave signal is first converted to an optical signal using a modulator, routed through optical paths of carefully chosen lengths, and then converted back to microwaves at each antenna element. Because the delay is a true time offset rather than a phase offset, the beam points in the same direction whether the radar operates at 27, 28, or 29 GHz.2J. Eur. Opt. Society-Rapid Publ. A review of research on optical true time delay technology This frequency-independent steering is the main reason military and civilian radar developers have invested heavily in photonic delay line technology over the past three decades.3Defence Science Journal. Evolutionary Trends in True Time Delay Line Technologies for Timed Array Radars
Looking Inside the Body With Delay Lines
Optical coherence tomography, commonly known as OCT, is the technology behind the detailed cross-sectional images an eye doctor takes of your retina. It works by splitting a light beam into two paths. One path bounces off tissue inside the body, and the other travels through a reference arm that contains an optical delay line. By scanning the delay in the reference arm, the system measures how deep different reflecting layers sit inside the tissue, building up a depth profile one point at a time.
Early OCT systems used a simple moving mirror as the delay line, but speed and stability limited how fast images could be acquired. Newer designs use more sophisticated delay mechanisms. One approach employs an all-reflective design with flat and curved mirrors plus a galvanometer-driven oscillating mirror, achieving a scanning depth of 1.5 mm with a duty ratio of about 89% and repetition rates above 4 kHz.4Optics Letters. Rapid scanning all-reflective optical delay line for real-time optical coherence tomography Another uses a polygonal scanner in a Fourier-domain configuration, which allows in vivo imaging at speeds that would be impractical with a simple translating mirror.5Applied Optics. Fast-Fourier-domain delay line for in vivo optical coherence tomography with a polygonal scanner Researchers have also explored thermo-optical delay lines built on silicon, aiming for cycling speeds above 1 kHz and scanning ranges of more than 1 mm to enable video-rate acquisition of tissue volumes.6Proceedings of SPIE. Thermo-optical delay line for optical coherence tomography
Handling Data Collisions in Optical Networks
In an optical packet-switching network, data arrives as bursts of light. When two packets try to use the same output port at the same time, one of them has to wait. Electronic networks solve this with memory buffers, but light cannot be parked in a register. Instead, optical networks use delay-line buffers: loops or lengths of fiber that force a packet to circulate for a controlled time until its intended output port is free. These buffers are the closest thing the optical domain has to RAM.
Optical buffers avoid the need to convert a photonic signal to an electronic one and back again, which saves energy and latency.7Chinese Optics Letters. Integrated optical delay lines: a review and perspective Designing them well, however, is surprisingly tricky. The length of each delay element needs to be optimized for the traffic load, and the optimal granularity turns out to depend on how heavily the network is loaded rather than on the buffer size itself. Adaptive buffer structures that reconfigure their delay granularity based on traffic conditions can minimize packet loss across a range of network scenarios.8Applied Optics. Optimized design of delay-line buffers with an input-feedback mechanism for asynchronous optical packet switching networks
Quantum Applications and Entanglement Preservation
Quantum networks and quantum computers face a timing problem similar to the one in optical packet switching: quantum operations do not all finish at the same moment, so photonic qubits sometimes need to be stored briefly while other parts of a computation catch up. Unlike classical signals, quantum states cannot be copied or amplified, which makes any form of storage far more delicate. A delay line used for quantum information has to preserve not only the intensity and timing of a photon but also its quantum state, including entanglement with a partner photon that may be somewhere else entirely.
Recent work has demonstrated a free-space optical delay line using a nested multipass cell architecture that achieves a single-transit delay of up to about 687 nanoseconds while maintaining a photon retrieval efficiency above 95%. Quantum state tomography on polarization-entangled photon pairs sent through the device confirmed an entanglement fidelity of 99.6%, meaning the quantum correlations survive the trip almost perfectly.9arXiv. Highly Efficient and Broadband Optical Delay Line towards a Quantum Memory That combination of low loss and high fidelity is what makes the delay line a viable stand-in for a quantum memory in near-term experiments.
The Bandwidth-Delay Trade-Off
Every optical delay line designer bumps into the same fundamental tension: you can have a large delay or a wide bandwidth, but getting both at once is hard. This constraint shows up as a limited “bandwidth-delay product,” and it is especially punishing for continuously tunable delay lines.10Optics Letters. Variable optical true-time delay line breaking bandwidth-delay constraints
The physics behind this trade-off is easiest to see in slow-light approaches, where a medium is engineered to have a very low group velocity. In principle, you can make light creep through a material almost arbitrarily slowly. But very slow group velocity always comes at the cost of very low bandwidth or throughput, and the delay-bandwidth product remains stubbornly limited.11Journal of Lightwave Technology. Slow-Light Optical Buffers: Capabilities and Fundamental Limitations In other words, the narrower the frequency window you can tolerate, the longer the delay you can achieve, and vice versa. Designers working on radar or wideband telecom applications, where bandwidth is non-negotiable, have to find workarounds, often by cascading multiple delay stages or by using wavelength-conversion tricks that shift the signal to a spectral region where a dispersive element provides the right delay.12Optics Communications. All-optical continuously tunable delay with a high linear-chirp-rate fiber Bragg grating based on four-wave mixing in a highly-nonlinear photonic crystal fiber
Slow Light as a Delay Mechanism
One way to delay a pulse without routing it through kilometers of fiber is to slow the pulse itself. Stimulated Brillouin scattering in an ordinary single-mode optical fiber can do this at room temperature and at standard telecom wavelengths. A pump laser creates an acoustic wave inside the fiber, and that acoustic wave interacts with the signal pulse to reduce its group velocity. By tuning the pump laser’s wavelength, you choose which signal wavelength gets delayed. By adjusting pump intensity, you control how much delay the pulse experiences, with continuously tunable delays demonstrated up to about 25 nanoseconds for pulses as short as 15 nanoseconds.13Physical Review Letters. Tunable all-optical delays via Brillouin slow light in an optical fiber
The appeal of slow light is that it is all-optical: no electronic switching, no moving parts. But the fundamental bandwidth-delay limitation described above means slow-light delays tend to be modest compared to what switched-path architectures can provide. Their niche is in applications where a small, finely tunable delay is more important than a large total delay range.
Shrinking Delay Lines Onto Chips
Bulk fiber spools and free-space mirror assemblies work, but they are large, sensitive to vibration, and hard to mass-produce. The push toward integrated photonics aims to put optical delay lines on chips, using the same lithographic manufacturing that produces electronic circuits. A variety of chip-scale architectures have been demonstrated, and their performance spans a wide range.
At one end of the spectrum, silicon-nitride relay-type true-time delay lines can provide delay ranges above 12 nanoseconds with a resolution of 850 picoseconds, but they require chip areas in the thousands of square millimeters. At the other end, MEMS-actuated delay lines fit in a footprint smaller than a thousandth of a square millimeter but offer only about 94 picoseconds of delay range. Between these extremes, coupled-resonator optical waveguides (CROWs) achieve about 800 picoseconds in 5 square millimeters, and recirculating loop designs reach around 1.1 nanoseconds in 60 square millimeters.14Chinese Optics Letters. Integrated optical delay lines: a review and perspective
Signal loss is another axis of comparison. The best integrated designs achieve delay losses below a thousandth of a decibel per picosecond of delay, meaning you can accumulate a large delay before the signal degrades unacceptably. Grating-based delay lines, for instance, report losses around 0.001 dB/ps, while photonic-crystal waveguide designs are considerably lossier at about 0.17 dB/ps.15Chinese Optics Letters. Integrated optical delay lines: a review and perspective
Material Platforms and Tuning Speed
The choice of material for an integrated delay line determines not only the optical loss but also how fast the delay can be adjusted. The two dominant tuning mechanisms on chip are the thermo-optic effect and the free-carrier dispersion effect. Thermo-optic tuning changes the refractive index by heating the waveguide, offering a large tuning range but with response times on the order of microseconds. Free-carrier tuning injects or removes charge carriers to change the index, achieving nanosecond-scale switching but with a smaller tuning range and additional absorption loss from the carriers themselves.16Chinese Optics Letters. Integrated optical delay lines: a review and perspective
Thin-film lithium niobate has emerged as an attractive material for delay lines that need both low loss and fast switching. Its strong electro-optic effect allows switching speeds in the tens of nanoseconds without the absorption penalty of free-carrier tuning. A recent demonstration at the 2-micrometer wavelength band achieved a 9-bit delay line with a tuning range from 0 to 511 picoseconds in 1-picosecond steps, using waveguides with only 0.27 dB/cm of loss, all within a footprint of about 34 square millimeters.17ACS Publications. Tunable Optical Delay Line Based on Thin-Film Lithium Niobate at 2 μm Waveband The 2-micrometer band is relevant for atmospheric sensing and next-generation communications, so having a compact, precisely stepped delay line at that wavelength fills a practical gap.
Keeping Long Delays Stable
The longer the delay you need, the harder it is to keep it stable. A fiber-based delay line producing hundreds of microseconds of delay involves kilometers of fiber, and even small temperature changes cause the fiber to expand and its refractive index to shift. At the femtosecond level, these environmental drifts are large enough to ruin applications like precision metrology, coherent radar, and high-resolution spectroscopy.
One approach to taming this drift is a homodyne phase-locked loop that continuously measures and compensates for environmentally induced jitter. A fiber-based tunable delay line built with this technique has demonstrated a maximum delay of 905 microseconds, with delay jitter held below 20 femtoseconds and an overlapping Allan deviation reaching roughly 2 × 10⁻¹⁷ at an averaging time of 1,000 seconds. Precise delay tuning is achieved by adjusting the reference phase in steps as small as 0.5 picoseconds without breaking the lock.18Optics Letters. Highly stable fiber-based photonic delay line with large and tunable delay Those stability numbers are remarkable: they mean the delay drifts by less than a few parts per quadrillion over a roughly 15-minute window, putting this class of delay line in the same stability league as atomic clocks.
How Different Architectures Compare
Choosing the right delay line architecture depends entirely on what you need from it. Here is a rough landscape of the trade-offs:
- Fiber spools: Offer enormous delays (microseconds to milliseconds) with low loss per unit length, but they are bulky, temperature-sensitive, and not easily tunable without additional components.
- Free-space multipass cells: Achieve low loss and preserve quantum states well, but they require careful alignment and do not lend themselves to mass production.
- Switched-path integrated circuits: Provide digitally selectable delays with fine resolution and a compact footprint, at the cost of cumulative loss from switches and waveguide bends.
- Resonator-based chips: Use coupled ring resonators or similar structures to achieve slow-light-like delays in a small area, but their usable bandwidth is narrow.
- Electro-optic platforms: Offer the fastest tuning (nanoseconds) with moderate delay ranges, especially on lithium niobate, though fabrication maturity still lags behind silicon photonics.
No single architecture dominates across all metrics. A radar system that needs broadband, squint-free beam steering will gravitate toward switched-path true-time delay lines with many bits of resolution. An OCT system that needs fast, continuous scanning favors a galvanometer-driven mirror or a thermo-optic approach. A quantum networking experiment will prioritize loss and fidelity above all else and accept a fixed, non-tunable delay if it means keeping more photons alive.
The Emerging Role of the 2-Micrometer Band
Most optical delay line development has historically targeted the 1.55-micrometer telecom window, where fiber losses are lowest and component ecosystems are most mature. But there is growing interest in the 2-micrometer band for applications in atmospheric sensing, gas spectroscopy, and free-space communications. Water vapor and carbon dioxide have absorption features near 2 micrometers that make this wavelength range useful for remote sensing, and several proposed satellite communication links operate there to exploit a different atmospheric transmission window.
Building delay lines at 2 micrometers has required new waveguide designs and coatings optimized for a wavelength where standard telecom components do not work. The thin-film lithium niobate delay line mentioned earlier is one of the first demonstrations of a fully integrated, multi-bit tunable delay at this wavelength, and the 0.27 dB/cm loss it achieves is competitive with the best results at 1.55 micrometers in the same material.19ACS Publications. Tunable Optical Delay Line Based on Thin-Film Lithium Niobate at 2 μm Waveband As demand for mid-infrared photonic systems grows, the techniques developed here will likely influence delay line design more broadly.