Science1 publisher2 min readPublished
Programmable photonic chip reads topology in momentum space while mimicking energy loss
Researchers programmed a photonic chip to read a lossy system's topology in momentum space, finding a Zak phase of 1.02 pi where theory predicts pi. The chip simulates each momentum point from a model, so it can map many phases on one device but has not yet measured a real material.
The Scientist · Science desk
Drafted by a language model from the sources cited here and checked against its claim ledger before publication. How we use AISend a correction
What happened
- Instead of fabricating a lattice, the team programs the chip to mimic the system at one point in momentum space, then reprograms it for the next until the whole space is scanned.
- Light on the chip travels without loss, so the team added helper channels that absorb what would be lost, making the signal channels behave like the leaky system.
- Mixing the output with reference light at four phase shifts lets the team recover both the strength and the phase of the signal from intensity measurements.
- Looping around an exceptional point, where a lossy system's modes merge, the measured phase wound once, giving +1 in one direction and -1 in the other.
- For a two-dimensional Rice-Mele pump, the team found a Chern number of 1 when running the topological cycle and 0 when running the trivial cycle.
Compiled by The ScientistSomething wrong?How this is made
Why it matters
- capability One reprogrammable chip can replace the many fabricated samples otherwise needed to survey a family of lossy topological phases, including strongly lossy ones.
- constraint Labs measuring real lossy materials still depend on edge states and other indirect signatures, because the chip reproduces only the models it is programmed with.
- constraint Studies of the highest-loss regimes will hit the weak-signal limit first, since the topology has to be recovered from a phase that becomes hard to pin down as the signal fades.
Topological properties are labelled by whole numbers that do not change under small imperfections [1]. Those numbers live in momentum space, which describes how a wave travels, not where it is [2]. Most experiments never look there directly. They infer the topology from its consequences, usually special states at the edges of a fabricated sample [3]. That works, but "it is a bit like working out whether a rope is knotted by looking only at its ends," one of the researchers wrote in an account published by phys.org [4]. The study itself is in Physical Review Letters [11].
Loss makes the indirect route harder. Non-Hermitian systems leak energy to their surroundings, every real device does, and strong loss shrinks the signal that carries the topological information [5]. The group's chip is a network of waveguides whose light paths are set electrically [6]. With that design there is no sample to fabricate. A trivial chain and a topological one are two settings on the same hardware [7].
The topology is carried by the signal's phase. "The phase is where the topology hides," the researcher wrote [10]. On the non-Hermitian Su-Schrieffer-Heeger chain, the measured Zak phase sits 0.02 pi from the prediction, a gap of about 2%, and the ordinary phase came out close to zero [12] [1]. The account does not give error bars for those figures, or the raw values behind the reported integer windings and Chern numbers [13] [14]. On the exceptional point, the researcher wrote: "I find it striking that such an exotic feature leaves such a clean fingerprint in a few intensity measurements" [15].
The thing this doesn't tell you is how the method would fare on a system whose topology nobody knows in advance. "The chip simulates these systems; it isn't a real material," the researcher wrote [16]. Each momentum point is programmed in from the model under test [7]. The close match to theory shows that the chip reproduces that model's predictions under simulated loss [12].
Loss is also where the readout strains. "When the signal becomes very weak, the phase becomes hard to determine," the researcher wrote [17]. The group still reports working in the strongly lossy regime, which the researcher described as the hardest to access [18].
I think the fair description is a programmable test bench for lossy topological models, and a carefully built one. The four-phase interferometry that recovers the phase from intensities is a simple measurement [9]. Larger, more complex models will need more elaborate chips, according to the account [19].
What to watch
- A published threshold for how much simulated loss the four-phase interferometric readout tolerates before the phase becomes indeterminate.
- Larger chips running non-Hermitian models more complex than the Su-Schrieffer-Heeger chain and the Rice-Mele pump.
- Any attempt to apply direct momentum-space phase readout to a fabricated lossy photonic material instead of a programmed simulation.