Single-photon switch opens door to terahertz photonic computing

Precise manipulation of photons enables much of today’s digital infrastructure, and a single-photon switch breakthrough by Professor Vladimir Shalaev’s group at Purdue University (West Lafayette, IN) opens a long-sought path to terahertz photonic computing.

Shalaev’s laboratory develops methods to generate ultrafast single photons using solid-state quantum emitters coupled to plasmonic cavities. These systems are extremely bright and can generate single photons at terahertz repetition rates, which correspond to the picosecond timescale (10^-3 nanoseconds).

“It’s about 1,000x faster compared to most existing methods to generate single photons,” says Demid Sychev, a postdoctoral researcher working with Shalaev, a professor of electrical and computer engineering, as well as the scientific director for nanophotonics at Birck Nanotechnology Center at Purdue. “But no existing single-photon detectors can operate at such speeds, which makes terahertz-rate single-photon sources difficult to use in practice.”

Nonlinear optics techniques can detect even femtosecond pulses at terahertz and even higher rates, “but they work only for high-power laser beams and are ineffective at the single-photon level,” Sychev says. “This gap begs the question: Can a single photon be used instead to modulate a macroscopic optical beam, so the modulation can be measured with standard ultrafast techniques to effectively enable ultrafast single-photon detection?”

After exploring possible mechanisms, “we found only one physical process can realistically achieve it: the electron-avalanche multiplication effect used for single-photon detectors,” Sychev says. “Unlike most physical processes, which have a response linearly proportional to the number of input photons or according to a power function, the avalanche multiplication naturally produces a measurable macroscopic response from one photon. This realization became the starting point for our experiment.”

Concepts behind the group’s single-photon switch

Its main principle of operation is similar to a standard photodiode used for measurements of the light’s intensity: Incoming photons generate free electrons within a semiconductor, which can freely move within the material (silicon in this case). Increased concentration of electrons increases its conductivity and makes it act more like metal. If connected to an electrical circuit, it causes a change of current flowing through it and allows researchers to detect an increased light flaw.

“This increase in ‘metallicity’ affects not only its electrical but also optical properties,” Sychev says. “Since metallic materials reflect light better, illuminating the diode also increases its reflectivity. In principle, this effect can be used for purely all-optical modulation—one beam controls transmission of another beam, which was demonstrated in earlier works.”

Importantly, the degree of induced “metallicity” depends on the number of absorbed photons, which means it works well directly only for high-intensity beams and fails at the single-photon level. In this case, adding a single free electron to a conduction band isn’t significant compared to a thermal background of free electrons naturally existing within the material. It makes the direct approach impractical for single photons and quantum applications.

“To overcome this limitation and extend our method to the single-photon regime, we use an avalanche-multiplication process, which is well known within the field of single-photon detectors. For this approach, we apply a high voltage to the diode,” says Sychev. “When a single photon generates a free electron, the strong electric field accelerates it and it gains even more energy. This electron goes on to create more energetic electrons, which are in turn accelerated and produce even more electrons. It initiates an avalanche process that produces a large population of free electrons from the initial single one. This process rapidly amplifies the density of free electrons, which significantly enhances the material’s effective ‘metallicity’ even though only one photon was absorbed.”

Now when the group sends a second beam onto the device, “it experiences the reflectivity change triggered by a single photon in the control beam,” Sychev says. “This behavior effectively mimics a photonic transistor, in which a single-photon-level signal modulates a macroscopic optical beam.”

How does the group’s method work?

For their work, the group uses a commercially available device that operates based on an avalanche mechanism. The device is made of silicon, which is widely used for photodetection applications.

“In general, generating avalanches within intrinsic semiconductors requires extremely high voltages—often several kilovolts,” says Sychev. “But this requirement can be reduced to merely tens of volts via an interface between n- and p-doped semiconductor regions, which forms a p-n junction. This interface creates an intrinsic electric field that helps accelerate electrons to the threshold energy required to initiate impact ionization. But we used a standard off-the-shelf device without any modifications.”

In the future, the device will need to be optimized for all-optical modulation, “which will require dedicated research,” Sychev adds.

Avalanche process simulations

To simulate the avalanche process, “we used standard finite-element simulations based on numerically solving differential equations that describe charge evolution,” says Sychev. “This model predicts how the charge distribution develops over time. But a more refined simulation should also account for the stochastic behavior of individual electrons within the semiconductor, which we plan to implement in future work.”

Interestingly, they discovered that within certain regions the charge distribution evolves faster than for the photodiode as a whole. “Because our system uses an optical beam—and allows optical readout of the avalanche—it can, in principle, capture much faster dynamics than a purely electronic device, which was confirmed by our simulations. In the future, this capability could enable the development of components operating at terahertz rates,” Sychev says.

Light-light interaction

“The concept of light-light interaction is special because it produces strong interactions between two optical beams (independent of power or wavelength),” says Sychev. “While many single-photon-level approaches, such as quantum emitters coupled to optical cavities, can mediate interactions between two weak beams, most other methods enable all-optical modulation only at macroscopic power levels. But, effectively, there are no approaches in which a single-photon-level beam can reliably control or modulate a high-power macroscopic beam.”

The group’s approach also operates at room temperature, doesn’t require an optical cavity (which significantly increases operating speed), and is fully compatible with complementary metal-oxide semiconductor (CMOS) fabrication. “These features make it ideally suited for building ultrafast, large-scale all-optical photonic circuits,” Sychev adds.

Challenges along the way

This work essentially explored “uncharted territory.” During the experiment, they tested several configurations of their setup because they were unsure what to expect for modulation amplitude or what timescale would be needed to get meaningful time-dependence data.

“Conventional optical pump-probe measurements rely on ultrashort delays—typically up to ~100 picoseconds—and were achieved by introducing optical delay lines,” explains Sychev. “But for nanosecond scale, long delay lines of a few meters are required. To avoid scanning such long delays, we modified our setup so the delay between the control and probe pulses was set electronically using a pulse generator.”

Another challenge was “reducing electrical noise in the system,” says Sychev. “The circuitry connected to the avalanche photodiode produces significant electromagnetic noise, which can distort the measurements. We spent considerable time mitigating these effects to ensure reliable data.”

Photonic computing and data center applications ahead

The group’s single-photon switch may find use for a variety of information-processing tasks, including computing and communications. “Its main principle of operating can lead to a new class of devices for photonic computation and communications, operating at terahertz frequencies without significant power consumption, which could potentially become the next platform for computers after electronics. But more research is needed to better understand its potential,” says Sychev.

It may also be used for other applications—think bioimaging, lasing, and related photonic applications—that demand strong optical nonlinearities. “Importantly, this platform also offers significant advantages for quantum photonic technologies,” Sychev says. “Although our method doesn’t preserve coherence between interacting beams, it can still support achieving all-optical quantum circuits operating at extremely high clock rates. With appropriate protocols, it may enable certain photonic quantum gates.”

Timeline

At this stage, the group has demonstrated a proof-of-principle result showing that modulation is indeed possible. “For our initial demonstration, we used a commercial single-photon detector not optimized for this purpose in any way,” says Sychev. “We believe the underlying idea has enormous potential, but creating a practical single-photon switch will require substantial further development that might take several years.”

Progress is still needed on both the fundamental and engineering sides. “Fundamentally, a deeper understanding of how the avalanche process evolves in time and space is essential—a direction we plan to focus on during our next steps,” Sychev adds. “It will allow us to figure out an optimal design for all-optical switches. But we still need to make advances in device geometry, diode design, coupling to photonic structures, explore different electrical regimes, and evaluate new materials.”

The group envisions their concept may open an entirely new research direction and ultimately enable fully optical photonic circuits for both quantum and classical applications.

FURTHER READING

D. V. Sychev et al., Nat. Nanotechnol. (2025); https://doi.org/10.1038/s41565-025-02056-2.