Chip Chip

Scientists Create a Three-State Optical Memory Chip Using Less Power Than a Laser Pointer

Researchers have demonstrated optical tristability on a silicon microchip using only 240 microwatts of input power, creating a potential building block for low-energy photonic memory, optical logic and neuromorphic computing.

The device allows light to remain in three stable intensity states under the same external conditions. Conventional digital electronics ordinarily represent information through two states, commonly interpreted as zero and one. Introducing a dependable third state could allow individual optical components to represent more information and support computing architectures that operate differently from traditional binary processors.

The achievement was produced by researchers from Peking University, Harbin Engineering University, ITMO University and Peng Cheng Laboratory. Their peer-reviewed study was published in Nature Nanotechnology under the title “Optical multistability in a compact microcavity enabled by near-exceptional coupling.”

What Optical Tristability Means

A stable optical state is a condition in which a device can maintain a particular light intensity after settling into operation. Optical bistability provides two possible stable outputs under identical external conditions, allowing the component to behave somewhat like a binary memory element.

Tristability extends that behaviour to three stable outputs. The same device can settle into a low, intermediate or high optical-intensity state depending on how it was previously stimulated. That memory of its earlier condition produces hysteresis, meaning the output depends not only on the current input but also on the path used to reach it.

The researchers describe multistability as an important property for multilevel optical memory and photonic computing. Three stable states could provide greater information density than an equivalent two-state element, although practical benefits would also depend on reliable reading, error tolerance and integration with complete computing systems.

Optical tristability has been demonstrated in other physical systems, but producing it inside a compact, silicon-based device has remained difficult. Light normally interacts only weakly with itself, especially when confined to microscopic chip structures. Previous approaches have often required larger devices, special materials, cascaded components or demanding operating conditions.

The Device Is Smaller Than a Human Hair

The team created its tristable system inside a circular silicon photonic-crystal microcavity measuring approximately 20 micrometres across. For comparison, a typical human hair is several times wider than the complete optical structure.

A photonic crystal contains a carefully arranged nanoscale pattern that controls how light propagates and becomes confined. The microcavity is designed to trap selected wavelengths, allowing the optical field to circulate repeatedly rather than immediately escaping.

The device produced resonances with quality factors approaching one million. A high quality factor, usually written as Q, means that the cavity can store optical energy for a comparatively long time relative to each oscillation. Light can therefore build up inside the tiny structure, strengthening nonlinear effects that would otherwise be too weak to use efficiently.

The published study reports efficiently excited resonances with measured quality factors reaching approximately 800,000, while the broader engineered regime approached one million. This combination of strong confinement and efficient external coupling was essential to achieving the three-state response at low power.

Near-Exceptional Coupling Made the Breakthrough Possible

The central innovation involves a concept called near-exceptional coupling.

The researchers began with two optical modes created through the symmetry of the photonic-crystal structure. They then introduced controlled structural perturbations that caused both modes to couple through a shared radiation channel.

This is described as a non-Hermitian system because energy can enter and leave through radiation and other forms of loss. Under carefully controlled conditions, the two modes approach an exceptional point, a special physical state where their properties become strongly connected.

Near that point, the modes develop almost identical wavelengths and similar linewidths while remaining accessible from outside the cavity. The researchers call this operating region near-exceptional coupling, or NEC. It lets the cavity confine light strongly while still accepting and releasing optical power efficiently.

That balance is difficult to achieve. A resonance that traps light exceptionally well can also become difficult to excite from an external source. The NEC design helps preserve a high Q factor without isolating the cavity so completely that it becomes impractical to control.

Heat Helps Create the Three Stable States

The observed tristability is based on thermo-optical nonlinearity.

When light accumulates inside the silicon cavity, part of its energy is absorbed and converted into heat. That small temperature increase changes silicon’s refractive index, shifting the wavelengths at which the cavity resonates.

The shifted resonance changes how much optical power enters and remains inside the cavity. That power then changes the temperature again, creating a feedback process between light intensity, heating and resonance position.

In a conventional single-mode system, this feedback commonly produces bistability. The researchers’ two closely spaced, strongly coupled modes created a more complex response, allowing three stable optical intensity levels to exist within part of the operating range.

The team observed clear hysteresis loops at an incident input power of approximately 240 microwatts. That is 0.24 milliwatts, below the output power of many ordinary laser pointers. The figure refers to optical input power used in the experiment, not the total electricity that a future processor containing lasers, control electronics and cooling equipment would consume.

Researchers Built a Three-State Optical Memory Prototype

The experiment went beyond observing an unusual optical effect. The team used the three stable states to demonstrate a proof-of-concept optical random-access memory element.

By adjusting the input light’s power or wavelength, the researchers could move the system between three distinct intensity levels. The device maintained the selected state until another controlled optical input caused it to switch.

The Peking University summary of the research says the prototype demonstrated fast and reliable switching among the three states. The research paper presents time-domain switching experiments that show how the multistable response can function as a controllable memory mechanism.

This does not mean a complete three-state photonic computer has been built. The prototype represents one memory component operating under laboratory conditions. A practical processor would require large arrays of devices, precise fabrication, integrated light sources, detectors, control circuits and methods for preventing neighbouring components from affecting one another.

Why It Could Matter for AI Hardware

Photonic computing attempts to use light for selected information-processing tasks that are traditionally performed by electronic circuits. Light can move through optical structures at high speed, and several wavelengths can travel through the same channel simultaneously.

Multistable components are particularly relevant to neuromorphic computing, which develops hardware inspired by features of biological neural systems. Neural and synaptic elements often require more than a simple on-or-off response. They may need several persistent levels to represent weights, activation states or previous inputs.

A compact three-state optical memory could potentially serve as a building block for reconfigurable optical neural networks, multivalued logic and all-optical signal processing. The researchers describe their method as a general strategy for producing energy-efficient multistability in nonlinear microcavities.

Silicon is also significant because silicon photonics can use manufacturing techniques related to those already developed for semiconductor fabrication. The experiment therefore offers a more integration-friendly direction than multistable systems relying on large free-space optical arrangements.

Major Engineering Challenges Remain

The research is an important physics and device-engineering demonstration, but several steps separate it from commercial computing hardware.

The thermo-optical effect relies on heating and cooling, which can be slower than purely electronic or ultrafast optical nonlinearities. Large arrays could also create thermal interference if heat from one memory cell changes the behaviour of nearby components.

Manufacturing variation is another concern. The system depends on two resonances having carefully controlled spacing, linewidths and coupling. Small fabrication differences could shift those properties and require individual calibration.

Future designs must also demonstrate long-term stability, repeatable switching over many cycles and compatibility with integrated lasers and detectors. The energy required by the complete system will matter more commercially than the optical input threshold of one isolated cavity.

The study nevertheless resolves a central challenge: it shows that stable three-level optical behaviour can be produced in a tiny silicon structure without requiring large input power.

A Small Chip With a Third Optical State

The breakthrough changes what a microscopic silicon cavity can do. Instead of producing only two stable optical outputs, the device can maintain three and switch controllably among them.

Its 20-micrometre footprint, quality factors approaching one million and 240-microwatt operating threshold make the result particularly relevant to integrated photonics. The accompanying optical-memory demonstration also shows that the effect can be controlled rather than merely observed.

Commercial photonic processors will require much more development, but the experiment provides a credible foundation. By engineering how two optical modes interact near an exceptional point, the researchers strengthened weak nonlinear behaviour enough to create a compact, low-power, three-state memory element.

Leave a Reply

Your email address will not be published. Required fields are marked *