From Flat to Narrow: How a Lattice Remembers and Forgets
09 May 2026, Yanjiang
A Lieb lattice’s flat band traps a quantum emitter’s photon in a bound state, but symmetry breaking warps the band, enabling a switch from memory to forgetfulness.
In the world of quantum photonics, control is everything. To build a quantum network, you need to generate single photons on demand, route them without loss, and detect them with certainty. But perhaps the subtlest challenge is memory—getting an emitter to hold onto a photon just long enough, then release it cleanly. A new theoretical work (arXiv:2605.06296) from a team led by Ying Hu at Shanxi University suggests that the key lies in the shape of an energy band. Zhiyong Liu, Yue Sun, and Ying Hu have shown how a special lattice, by flattening and then slightly warping its bands, can tune an emitter from perfect memory to complete forgetfulness.
Their chosen platform is the one-dimensional Lieb lattice, a structure named after the mathematician Sophus Lie. Picture a chain of repeating unit cells, but instead of a simple row of atoms, each cell contains three sites—labeled A, B, and C—arranged in a line with a gap between A and C where B sits in the middle. B is not directly connected to A or C; the couplings are indirect, hopping through the lattice. This peculiar connectivity produces a band structure with three energy bands: two that curve upwards and downwards like normal dispersive bands, and one that is utterly flat. In a flat band, the kinetic energy of a particle is frozen. Think of it as a perfectly level plane: no matter what force you apply, a marble will not roll. Yet quantum mechanics grants these motionless states a collective power—their wavefunctions overlap in ways that can synchronize behavior across the entire lattice. Unlike classical marbles, which simply sit where placed, quantum particles in flat bands can still interact over long distances through these overlapping tails.
This flatness is not just a mathematical curiosity. In condensed matter, flat bands were first appreciated in the context of ferromagnetism: when electrons cannot move, their spins align to lower their interaction energy. In photonics, flat bands mean light can be slowed to a crawl, enhancing light-matter interactions by a thousandfold. But the team, who work at the State Key Laboratory of Quantum Optics and Quantum Optics Devices in Taiyuan, saw an opportunity for a different kind of engineering—one focused on time and memory.
The central idea was deceptively simple: attach a quantum emitter—a tiny artificial atom—to one of the sublattices and watch it radiate. The emitter’s fate depends on which sublattice it touches, because each hosts a different fraction of the flat-band wavefunction. When the emitter was coupled to the B sublattice, the results were dramatic. The excited-state population did not simply decay away; instead, after an initial drop, it leveled off at about 87% of its original value, remaining there forever. This is the signature of non-Markovian dynamics, where the system retains a memory of its past. The photon is not freely escaping; it is entangled in a bound state with the emitter, and a fraction of the energy forever oscillates between them. Non-Markovian memory is like a conversation in a cathedral: echoes repeat and build, so every new note carries the weight of all that came before. This memory is not a matter of will, but a consequence of how quantum mechanics binds light to matter when the density of states is structured just so.
The 87% trapping efficiency is remarkably high. It tells us that even though the emitter is surrounded by a continuum of modes into which it could radiate, the lattice’s flat band acts as a trap door that snaps shut. The photon is never quite born; it remains in a hybrid state, a bound state in the continuum—a paradox resolved by the destructive interference of other modes.
Coupling the emitter to the C sublattice gave the team a tuning knob: the detuning, or energy mismatch, between the emitter and the flat band. At zero detuning, the trapping was again strong, much like the B-sublattice case. But at a detuning of one times the lattice’s intrinsic coupling J, the population decayed almost completely. The emitter had crossed into the Markovian regime, where the system forgets instantly. Each emission event is independent; the past leaves no trace. Markovian decay is like shouting into a canyon that gives no echo: the sound is gone the moment it leaves your lips. Here, the spectral function—a kind of energy fingerprint of the emitter—tells the story: sharp isolated peaks indicate bound states and memory; broad, featureless continua signal oblivion. The ability to switch from non-Markovian to Markovian with a simple detuning knob is practically useful. In quantum memory applications, you want non-Markovian coherence to store information. In quantum cryptography, you might seek Markovian emission to send photons without correlated noise. The Lieb lattice offers both, side by side.
But flat bands are ideals. In any real material or photonic chip, imperfections, disorder, or intentional design always impart some small dispersion. Liu, Sun, and Hu therefore introduced a controlled symmetry-breaking term, a small perturbation V₀ that altered the hopping strengths between sites. The result: the perfectly flat band morphed into a narrow dispersive band, acquiring a bandwidth W that grew linearly with V₀. The formerly frozen marbles could now drift on a gently sloped table. This simple deformation gave rise to a rich competition. When the bandwidth W was much smaller than the emitter’s coupling strength g, the non-Markovian oscillations persisted—the bound states found enough shelter from the continuum. When W grew large enough to dominate g, the system transitioned to exponential Markovian decay. The researchers derived explicit scaling laws governing this crossover. For example, the total bound-state residue, a measure of the emitter’s memory, scaled as a function of g divided by W. By tuning a single knob—the symmetry-breaking V₀—one could dial up any degree of memory, from perfectly coherent to completely forgetful.
The team’s insight was that the bandwidth is not just a perturbation; it is a new control parameter. By plotting the crossover as a function of g divided by V₀, they found a universal curve—a master equation for how memory escapes. This scaling relation is the kind of thing experimentalists dream of: a single formula that tells you what will happen before you build the device.
The implications ripple outward. Flat-band photonics has been a white-hot field, with experiments on moiré photonic crystals and laser-written waveguide arrays demonstrating slow light and enhanced nonlinearities. But most of those systems operate with narrow bands, not exactly flat ones. The Shanxi team’s framework connects these worlds, offering a continuous theoretical bridge from idealized flatness to practical narrowness. Their scaling laws provide a quantitative blueprint: knowing your lattice’s bandwidth and your emitter’s coupling lets you predict exactly how much memory to expect. The team also examined the emitter on the A sublattice, where the coupling is to both flat and dispersive bands simultaneously. There, the dynamics were richer still—a hybridization of trapping and decay—but the same theoretical machinery handled it elegantly. Throughout, they validated their analytics against numerical simulations, the two tracing each other with impressive fidelity.
What makes this work resonate beyond the specific geometry of the Lieb lattice is its general approach. The idea of engineering band flatness to control emission dynamics is portable. One can imagine similar strategies applied to twisted bilayer photonic crystals, where moiré potentials naturally create narrow bands, or to ultracold atoms in optical lattices, where symmetry can be tuned with lasers. The scaling laws distill the essential physics into simple power laws between bandwidth and coupling, making them easy to test and use.
Perhaps the deepest thread here concerns the nature of time and memory in quantum mechanics. Non-Markovian dynamics, which depend on the entire history of the system, stand in tension with the usual Markovian assumption that the future depends only on the present. The Lieb lattice, in the hands of Liu, Sun, and Hu, becomes a laboratory where this tension is turned into a quantitative dial. By warping a flat band, we can watch a quantum system lose its memory in real time, transitioning from a world of echoes to a world of silence.
The road ahead is clear, even if the timeline remains uncertain. Photonic Lieb lattices have already been fabricated in waveguide arrays, and the kind of symmetry-breaking the team describes is achievable with standard laser-writing techniques. The next step is to bring an actual quantum emitter—a semiconductor quantum dot or a defect center in diamond—to one of these sites and watch the memory spectrum unfold. If the theory holds, we will have a new, predictable handle for shaping light at the single-photon level.
There is a poetic resonance here. The Lieb lattice, with its flat band of frozen motion, resembles a kind of quantum stasis. By breaking symmetry, the team releases that stasis into a gentle flow. In a world where we measure time by the ticks of atomic clocks, perhaps we will one day measure quantum memory by the bandwidth of a lattice. The work from Taiyuan is not just a theoretical advance; it is an invitation—to experimentalists, to engineers, to anyone who has ever wondered how a quantum system learns to forget.
Yanjiang is an online editor of LoomSci
References
- Zhiyong Liu et al., From flat to narrow bands: Engineering quantum emission in a one-dimensional Lieb lattice, arXiv:2605.06296