The concept at the heart of the experiment is Floquet engineering, named after the nineteenth-century French mathematician Gaston Floquet, whose theory describes systems subjected to periodic driving. In a Floquet system, the energy bands of a material are no longer the static bands familiar from textbook solid-state physics. Instead, the periodic drive creates replicas of the original bands, called sidebands, separated by multiples of the photon energy. When these replicas hybridize with one another, the effective band structure that electrons experience can be fundamentally transformed. A trivial material can, in principle, acquire the hallmarks of a topological insulator: protected conducting edge channels and a bulk that remains insulating.

Graphene has served as the canonical theoretical playground for this idea ever since theorists predicted that circularly polarized light could gap out its Dirac cones and endow the otherwise gapless material with a topological character. In graphene, electrons behave as massless Dirac fermions moving through two inequivalent valleys in momentum space. Circularly polarized light breaks time-reversal symmetry and imprints opposite topological masses on the two valleys, producing a Chern-like Floquet band structure. The prediction was elegant, but verifying and controlling it experimentally proved extraordinarily difficult, because the light-induced gaps exist only while the driving field is present and they are easily masked by heating, disorder, and the many-body complexity of real solids.

The new experiment overcomes these obstacles by combining an intense mid-infrared driving field with ultrafast time- and angle-resolved photoemission spectroscopy, known as tr-ARPES. In this technique, a pump pulse dresses the graphene electrons with photons while a delayed probe pulse ejects them, allowing researchers to reconstruct the occupied and unoccupied electronic structure with femtosecond temporal resolution. By tuning the polarization, intensity, and timing of the pump pulse, the team could watch the Floquet sidebands emerge, measure the light-induced gaps at the Dirac point, and, most importantly, steer the electron populations within the engineered bands in real time.

The decisive advance is the demonstration of control rather than passive observation. By adjusting the parameters of the optical drive, the researchers could modify the size of the Floquet gaps and redistribute electrons among the light-dressed states, effectively programming the electronic landscape of graphene on demand. The measurements reveal that the dressed electrons follow the instantaneous symmetry of the driving field, so that switching the handedness of the circular polarization reverses the topological character imprinted on the two valleys. This level of command over a transient quantum phase had remained elusive in earlier studies, which succeeded in detecting Floquet states but could not reliably manipulate them.

Timing proved to be as important as intensity. The team found that the Floquet bands form and dissolve on femtosecond timescales, and that a window exists during which the light-induced gaps are well defined before carrier relaxation and phonon scattering wash them out. By probing within this window, the researchers obtained clean spectroscopic signatures of the topological band structure: gaps opening at the Dirac crossings, sidebands displaced by integer multiples of the pump photon energy, and spectral weights consistent with theoretical Floquet calculations. The agreement between the measured spectra and simulations based on the Floquet formalism provides strong evidence that the observed states are genuine light-dressed bands rather than artifacts of the measurement.

The implications extend well beyond graphene. Floquet topological insulators are a testbed for a broader vision in which materials properties are not fixed at synthesis but become dynamically programmable. A topological insulator is prized for its robust, dissipation-resistant edge transport, which makes it a candidate platform for low-power electronics and topological quantum information processing. If such phases can be switched on and off with light, devices could in principle route currents along reconfigurable edge channels at terahertz rates, far faster than conventional transistor switching, and without the need to chemically alter or permanently structure the material. The present work demonstrates the elementary operations, gap control and population steering, that such a scheme would require.

The achievement also sharpens a long-standing debate in the Floquet community about how driven quantum systems reach equilibrium, or whether they avoid it altogether. In principle, continuous driving should continuously pump energy into the electrons and heat the system until the band structure loses meaning. Yet the experiments show that on the ultrafast timescales probed here, a coherent, well-defined Floquet band structure exists long enough to be useful. Understanding the interplay between coherent driving, electron-electron scattering, and phonon-mediated relaxation in this regime is a central question for the field, and the new data provide quantitative benchmarks for theories of nonequilibrium many-body dynamics in driven solids.

Challenges remain before optical control of topological electrons can leave the laboratory. The light-induced phase persists only while the drive is on, so practical devices would need continuous or high-repetition-rate illumination, raising questions about efficiency and heat management. The drive intensities required are substantial, although the use of mid-infrared photons resonant with the material’s interband transitions helps maximize the gap size for a given fluence. Extending the approach to other two-dimensional materials, including topological semimetals and engineered moiré superlattices, could lower the required power and broaden the accessible phases, from Chern insulators to Floquet Weyl semimetals and anomalous Floquet phases with no static counterpart.

Even with those caveats, the study marks a conceptual turning point. For two decades, topological materials have been discovered, characterized, and catalogued as static objects. This work shows that the topology itself can be an operating parameter, adjusted with the twist of a polarization dial and read out within a single optical cycle. The electrons in a Floquet topological insulator are no longer merely passengers on a light-induced band structure; they can now be directed through it. As ultrafast light sources and probe techniques continue to advance, the prospect of circuits whose conducting pathways are written, erased, and rewritten by beams of light moves from theoretical speculation toward experimental engineering.

Subject of Research: Optical control of electrons in a Floquet topological insulator created in light-dressed graphene.

Article Title: Optical control of electrons in a Floquet topological insulator

Article References: Lesko, D. M. B., Weitz, T., Wittigschlager, S., Li, W., Heide, C., Neufeld, O., & Hommelhoff, P. (2026). Optical control of electrons in a Floquet topological insulator. Nature Physics. https://doi.org/10.1038/s41567-026-03429-7

Image Credits: AI Generated

DOI: 10.1038/s41567-026-03429-7

Keywords: Floquet topological insulator, graphene, optical control, topological phases, ultrafast laser, Floquet engineering, light-matter interaction, edge states, condensed matter physics, photoinduced band gaps, Optical, control

The circulator is one of the oldest and most useful pieces of non-reciprocal optics. In its classic form, a circulator is a three-port device: light entering port one exits at port two, light entering port two exits at port three, and light entering port three exits at port one. In fibre-optic systems, this unidirectional routing is what allows a single fibre to carry traffic in both directions, or allows a transmitter and a receiver to share the same fibre facet, without the powerful outgoing signal leaking into, and blinding, the sensitive detector on the receiving path. Conventional circulators achieve this asymmetry using magneto-optic materials such as yttrium iron garnet, whose interaction with light changes depending on the direction of an applied magnetic field. The physics works beautifully, but the hardware does not scale: magnets, garnet crystals and precision-aligned optical assemblies are large, expensive, and impossible to integrate onto a silicon photonic chip.

The new device sidesteps magneto-optics entirely by exploiting the electro-optic effect in thin-film lithium niobate, a material platform in which a nanometre-scale crystalline film of lithium niobate is bonded to an insulating substrate and patterned into waveguides. Lithium niobate possesses one of the strongest Pockels coefficients of any practical optical material, meaning that an applied electric field changes the refractive index seen by light travelling through it essentially instantaneously. By engineering the phase and magnitude of radio-frequency modulation signals applied to travelling-wave electrodes alongside the optical waveguides, the researchers create an interference condition that is asymmetric in time: light propagating in one direction accumulates a modulation-induced phase shift that routes it toward one output, while light propagating in the opposite direction experiences the modulation differently and is routed, or blocked, accordingly. Because the Pockels effect is inherently non-reciprocal in this dynamically driven configuration, the device achieves genuine non-reciprocal routing without any magnetic material.

The performance numbers reported in the study place the device squarely in the territory of practical deployment rather than laboratory curiosity. Peak isolation of 37 decibels means that fewer than one part in roughly five thousand of the reverse-propagating light survives the journey — more than enough to protect a coherent receiver from the far stronger signal launched in the opposite direction along the same fibre. Equally important is the breadth of the transmission window. Optical networks operate across finely divided wavelength channels, and a circulator that works only at a single narrow wavelength would need to be replicated or retuned for each channel. The broad bandwidth demonstrated here means the same chip can serve the densely wavelength-multiplexed traffic that defines modern long-haul and data-centre interconnects.

The system-level demonstration is arguably the headline result. Rather than characterising the chip in isolation, the researchers connected it into a fibre transmission link and pushed bidirectional traffic through a single standard single-mode fibre at 800 gigabits per second of standard throughput. This is the regime in which commercial optical transceivers live, and the fact that the integrated circulator survived the test — supporting full-duplex communication without degrading the signals in either direction — suggests a clear path from the laboratory bench toward the network equipment rack. A circulator of this kind could allow a single fibre pair to do the work of two, or allow transceivers to pack transmit and receive functions onto shared optical infrastructure with less spare capacity held in reserve.

Thin-film lithium niobate has been on a remarkable run over the past several years. The platform combines the ultra-low optical loss and strong electro-optic response of bulk lithium niobate with the compactness and fabrication scalability of chip-based photonics. Researchers have used it to build modulators with bandwidths exceeding one hundred gigahertz, frequency comb sources, quantum photonic circuits, and high-performance filters. What has often been missing from the toolbox, however, is non-reciprocity. Passive integrated photonics built on silicon or silicon nitride is fundamentally reciprocal: light travels through the same component identically in either direction, which is a direct consequence of the linearity and time-independence of the underlying physics. Lasers on optical chips therefore remain vulnerable to back-reflections, and full-duplex links have required external, discrete circulators — exactly the kind of bulky component that integrated photonics was invented to eliminate.

The electro-optic approach demonstrated here changes that calculus by making non-reciprocity a matter of circuit design rather than materials sourcing. Because the circulator is built with the same lithographic processes and electrode structures used for lithium niobate modulators, it can in principle be co-fabricated on the same chip as the high-speed modulators, switches and filters that already exist on the platform. A complete transceiver front end — laser-coupled modulator on the transmit path, circulator sharing the fibre, and coherent receiver on the return path — could then live on a single lithographic die. The authors’ demonstration of standard 800G throughput through single-mode fibre speaks directly to the engineering requirements of that vision, since it shows the device operating with the modulation formats, channel counts and power levels that real systems actually use.

The implications extend beyond telecommunications. Data-centre interconnects are consuming fibre and transceiver capacity at a pace that strains both supply chains and power budgets, and any component that lets one fibre carry two directions of traffic efficiently translates directly into infrastructure savings. In fibre-to-the-home networks, circulators are already standard equipment for sharing fibre between downstream and upstream signals; an integrated, magnet-free version could shrink the optical line terminals at the heart of those networks. Coherent transceivers, sensing systems such as fibre-optic distributed acoustic sensors, and free-space laser communication terminals all rely on separating transmitted and received light, and all would benefit from a compact, low-loss, broadband circulator that integrates with the rest of the photonic circuit.

There remain, as with any first demonstration, engineering questions on the road to volume manufacturing. Insertion loss, the fraction of signal power sacrificed in passing through the device, must be minimised so that network link budgets can absorb the circulator without shortening reach. The radio-frequency drive electronics consume power and add complexity, and the modulation scheme must be stabilised against temperature drift and fabrication variation across large wafers. Scalability of the fabrication process — yields, wafer-scale uniformity, and packaging of the optical fibre interfaces — will determine whether the device makes the leap from the laboratory to the production line. Yet the platform’s rapid commercial maturation, with thin-film lithium niobate foundry services now offered by multiple suppliers, gives the field reason for optimism that these are problems of engineering refinement rather than fundamental physics.

The broader significance of the work lies in what it says about the direction of photonics as a discipline. For half a century, the non-reciprocal components at the heart of optical networks have been the last holdouts against integration, stubbornly magnetic, bulky and discrete while everything around them shrank onto chips. By demonstrating a high-isolation, broadband circulator with genuine system-level throughput on thin-film lithium niobate, the researchers have shown that even this last holdout can be brought into the integrated fold. If the technology follows the trajectory of the platform’s modulators — from laboratory record to commercial product in a handful of years — the optical circulator may soon be as unremarkable a fixture inside a transceiver as the amplifier and the modulator, quietly enabling the two-way flow of data over the single strands of glass that carry the world’s information.

Subject of Research: Integrated electro-optic circulator on thin-film lithium niobate for bidirectional optical fibre transmission

Article Title: Integrated electro-optic circulator on thin-film lithium niobate for bidirectional optical fibre transmission

Article References: St-Arnault, C., Laperle, C., Kita, D. M., Reimer, C., & Plant, D. V. (2026). Integrated electro-optic circulator on thin-film lithium niobate for bidirectional optical fibre transmission. Nature Photonics. https://doi.org/10.1038/s41566-026-02008-9

Image Credits: AI Generated

DOI: 10.1038/s41566-026-02008-9

Keywords: Integrated, electro-optic, circulator, thin-film, lithium, niobate, bidirectional, optical, fibre, transmission, scientific research