A team of engineers has packed light generation, routing, and detection onto a single chip by encoding data in a quantum property called the valley state, demonstrating the ability to process two distinct images simultaneously at room temperature. Lead author Chi Li and senior author Haoran Ren, working across Monash University and Macau University of Science and Technology, described a device that produces chiral photons from a tungsten disulfide monolayer, steers them through a selective waveguide, and reads the valley information electrically. The result, reported in Nature Photonics, represents the first time these three functions have been integrated on one programmable nanocircuit.
Valley-encoded photonics and the data-movement bottleneck
Moving data between processors burns energy and generates heat, two problems that worsen as AI workloads scale. Conventional electronic interconnects face physical limits on how densely they can carry signals without interference. Photonic chips, which use light instead of electrons, promise higher bandwidth and lower power draw, but most designs still rely on external laser sources and bulky optical components that resist miniaturization.
The valley state offers a different encoding axis. In certain two-dimensional materials, electrons occupy energy minima, or “valleys,” that carry distinct quantum signatures. Light emitted from these valleys inherits a left- or right-handed circular polarization, a property called chirality. By treating each chirality as a separate data channel, engineers can encode and transmit independent streams of information through the same physical waveguide without crosstalk. The Monash-led team exploited this principle to build a circuit that handles generation, routing, and readout in a footprint small enough to sit on a chip.
How the WS2 nanocircuit generates and steers chiral light
According to the Nature report, the circuit starts with an encapsulated WS2 monolayer that produces valley-dependent chiral photons through second-harmonic generation. When a pump laser hits the monolayer, it emits photons at twice the input frequency, with the chirality determined by the valley from which the photon originates. A chirality-selective meta-waveguide then routes left-handed and right-handed photons in opposite directions, creating two independent optical channels on the same chip.
The demonstration encoded and processed two images simultaneously at room temperature. Each image was mapped onto a different valley channel, transmitted through the waveguide, and reconstructed at the detector end. By showing that image information survives the full generate–steer–read cycle, the experiment moves beyond proof-of-principle light routing and into functional signal processing. The authors argue that this architecture could, in principle, support parallel data streams for neuromorphic, quantum-inspired, or AI accelerators without duplicating hardware for each channel.
The device’s selectivity hinges on the meta-waveguide’s ability to distinguish between left- and right-handed circularly polarized light. Its nanoscale patterning enforces directionality: photons of one chirality couple efficiently into modes that travel to the left, while the opposite chirality couples into modes that travel to the right. Because the valley state in WS2 is tied to chirality, the waveguide effectively separates valley information spatially. At the far ends of the guide, integrated photodetectors convert the optical signals into electrical readouts, completing an on-chip valleytronic loop.
Separate peer-reviewed work on valley photonic crystals has shown that valley kink states can route light through sharp bends and intersections at telecom wavelengths with minimal loss. That study focused on topological protection, where the crystal’s structure prevents backscattering even at tight corners. Together, these two bodies of research suggest that valley-based routing can handle both the generation-to-detection pipeline and the physical routing challenges that on-chip optical networks demand.
Missing metrics that would prove commercial viability
The dual-image demonstration is striking, but the published record lacks several numbers that engineers would need before designing valley-encoded chips into real products. Neither the Nature Photonics paper nor institutional releases report bit-error rates for the full generate–steer–read cycle. Power consumption per bit, a standard benchmark for photonic interconnects, is also absent from the available data. Without these figures, claims about energy savings relative to electronic interconnects remain qualitative rather than quantitative.
A practical hypothesis-that valley-encoded waveguides could sustain aggregate throughput above 100 Gbps at less than 1 picojoule per bit across centimeter-scale on-chip paths without active cooling-cannot be tested against the current dataset. The reported demonstration used second-harmonic generation from a single monolayer, and the papers do not specify insertion loss, propagation distance, or thermal stability under sustained operation. The absence of a detailed link budget leaves open questions about how the architecture would scale beyond a laboratory-scale testbed.
Foundry-compatible fabrication details are another blind spot. The reports do not address whether the WS2 encapsulation process can survive standard CMOS back-end-of-line temperatures or how the meta-waveguide patterning would integrate with existing lithography flows. Without a clear path to wafer-scale manufacturing, the technology remains closer to a bespoke research platform than a near-term commercial component.
Independent replication is also missing. The collaboration between Monash University and Macau University of Science and Technology produced the core result, and a linked notice confirms the publication and its framing as room-temperature on-chip valley information processing. But no data from outside labs has yet appeared to corroborate the performance claims. Until a second group reproduces the generate–steer–read cycle on a comparable device, the result stands as a single-team achievement, impressive but not yet broadly validated.
What to watch as valley photonics matures
Two distinct research threads are converging on the same goal. The Monash-led nanocircuit proves that valley-dependent chiral photons can be born, directed, and read on one chip. The separate work on electrically tunable valley photonic crystals shows that valley-protected modes can survive complex routing geometries that resemble real on-chip networks. Together, they outline a roadmap toward valley-based optical fabrics that might one day interconnect compute tiles the way copper traces do today.
Several milestones will determine whether that roadmap leads to practical hardware. First, researchers will need to quantify link-level metrics: bit-error rates across realistic distances, energy per bit including all control circuitry, and robustness to temperature fluctuations typical of high-density packages. Demonstrations that move beyond static images to high-speed modulation patterns would help translate valley encoding into familiar communications benchmarks.
Second, integration with electronics must progress from conceptual compatibility to process-level detail. That includes proving that two-dimensional materials like WS2 can be deposited, encapsulated, and patterned on full wafers without degrading their valley-selective properties, and that meta-waveguides can be aligned with acceptable tolerances alongside conventional metal interconnects. Hybrid packages that combine a valley-photonic die with a standard CMOS logic die could serve as an intermediate step.
Third, the community will be watching for demonstrations that combine valley generation, topologically protected routing, and detection in a single platform. A chip that uses valley kink states to carry chiral photons around multiple bends, through intersections, and into integrated detectors would begin to resemble a genuine network-on-chip rather than a linear optical path. If such a device can operate at room temperature with repeatable performance, it would mark a significant advance beyond today’s prototypes.
For now, the WS2 nanocircuit highlights what is possible when quantum degrees of freedom are engineered directly into photonic hardware. By co-locating generation, routing, and readout of valley-encoded light on a single chip, the work points toward interconnects that use the same physical channel to carry multiple independent data streams. The missing metrics and manufacturing questions are nontrivial, but they define a clear agenda for the next wave of experiments. As valley photonics moves from isolated demonstrations to reproducible platforms, its promise-to ease the data-movement bottleneck without paying a steep energy price-will be tested in the unforgiving environment of practical systems design.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
