A series of experiments across multiple research groups has shown that deliberately implanting atomic-scale defects into ordinary silicon can turn the world’s most common semiconductor into a source of single photons suitable for quantum networking. The work spans peer-reviewed demonstrations of cavity-enhanced light emission, electrically triggered spin initialization, and isotope engineering that extends emitter lifetimes by more than fivefold. Taken together, these results suggest that the infrastructure for a future quantum internet could ride on the same silicon platform that already powers classical computing.
A Single Defect Turns Silicon Into a Light Source
Silicon is an excellent electronic material but a poor light emitter, which is why lasers and LEDs have historically relied on compound semiconductors like gallium arsenide. The core advance reported in a Nature Communications study is that embedding a single atomic emissive center, known as a G center, inside a silicon nanophotonic cavity overcomes that limitation. The cavity confines light so tightly around the defect that the researchers measured greater than 30 times luminescence enhancement and near-unity atom-cavity coupling, meaning almost every photon the defect produces is captured and directed. That level of efficiency had previously required exotic materials incompatible with standard chip fabrication.
Producing these defects reliably is just as important as making them bright. A separate peer-reviewed effort published in Nature Communications detailed controlled formation and engineering of quantum emitters in silicon, establishing repeatable defect creation processes with tunability over emitter properties. Without that repeatability, a quantum light source would remain a laboratory curiosity rather than a scalable technology. The combination of high brightness and programmable fabrication is what gives silicon a credible path toward mass-produced quantum photonic devices.
Electrical Triggering and Spin Control on a Chip
Brightness alone does not make a useful quantum device. A quantum network node must emit single photons on demand and tie each photon to a stored quantum state, typically an electron spin, that can serve as a memory. A preprint on electrically excited emission from a cavity-coupled silicon T centre demonstrated exactly that. The device achieved a second-order correlation value of g^(2)(0) = 0.05(2), a metric that drops toward zero as a source approaches true single-photon purity. The same electrically triggered emission was used to herald and initialize an electron spin state with 99% fidelity, linking photon generation directly to quantum memory preparation.
The electrical triggering aspect matters for practical reasons. Optical pumping, the alternative, requires bulky external lasers and precise alignment. Electrical excitation can be routed through on-chip wiring, which is how every transistor in a modern processor already works. A separate preprint described an electronic-photonic platform fabricated in a high-volume 45 nm CMOS process. That chip included on-chip feedback stabilization for microring photon-pair sources operating in the quantum regime, with high coincidence-to-accidental ratio and low noise. The demonstration shows that quantum photonic components can share a production line with conventional electronics, removing a major bottleneck between laboratory results and deployed hardware.
Heavier Hydrogen Extends Emitter Lifetimes Fivefold
Even with bright, electrically driven emitters, silicon T centres suffer from rapid non-radiative decay: the defect loses its energy as heat before it can emit a photon. A preprint reporting ensemble spectroscopy and first-principles calculations showed that replacing the hydrogen atom in the T centre with its heavier isotope, deuterium, yields greater than five times longer excited-state lifetime than the protium version. The researchers attributed the effect to suppression of non-radiative decay via a lowered local vibrational energy. Because deuterium vibrates at a lower frequency than ordinary hydrogen, it is less efficient at draining the defect’s electronic energy into lattice vibrations.
That lifetime extension has direct consequences for quantum networking. Longer-lived excited states produce narrower spectral lines, which means photons from separate emitters are more likely to be indistinguishable, a prerequisite for entanglement-based protocols. The experiments used isotopically pure silicon cooled below 4 K, as reported by Phys.org, with detailed spectroscopy confirming that the isotope substitution improves optical cyclicity at telecom-band wavelengths compatible with existing fiber infrastructure. Whether these cryogenic conditions can scale to deployed network nodes remains an open engineering question, but the physics now clearly favors deuterium-based T centres over their protium counterparts.
A Hydrogen-Free Alternative and the Road Ahead
Hydrogen itself may be the weakest link in T centre technology. Hydrogen atoms are mobile inside silicon at processing temperatures and can drift away from the defect site, degrading or destroying the emitter. A first-principles prediction paper proposed a hydrogen-free carbon-nitrogen defect complex in silicon as an isoelectronic alternative, with a computed telecom transition at 828 meV in the S-band and a radiative lifetime of 1.2 ns, which would make it significantly faster than current T centres. The CN complex also showed calculated stability against decomposition, addressing the fragility concern head-on and suggesting that quantum-grade emitters could survive the high-temperature steps of mainstream semiconductor processing.
These silicon-based efforts sit alongside broader quantum networking research that explores different host materials and defect species. Work at Harvard has used silicon-vacancy centers in diamond to distribute entanglement over metropolitan distances, with the university’s Gazette coverage framing the experiments as an early glimpse of a next-generation internet based on single photons. Institutions such as Cornell University are also investing in integrated photonics and quantum information science, providing the interdisciplinary expertise needed to merge solid-state physics, device engineering, and large-scale systems design.
Together, these lines of research point toward a plausible architecture in which quantum repeaters, memory nodes, and processors are built directly on silicon, using defect-based emitters that operate at telecom wavelengths and can be fabricated in existing CMOS foundries. G centers and T centres provide near-deterministic single photons and spin-photon interfaces; isotope engineering with deuterium extends coherence and improves indistinguishability; and hydrogen-free complexes promise greater robustness and faster operation. The remaining challenges (scaling to millions of identical emitters, integrating cryogenics, and managing error correction overhead) are substantial but now appear as engineering problems on a familiar semiconductor platform rather than as fundamental roadblocks. If those hurdles can be cleared, the same material that underpins today’s digital world may also carry tomorrow’s entangled networks.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
