A condensate that cannot be seen is a poor witness to its own existence. That is the problem a new theoretical study from the University of Hong Kong sets out to solve for twisted bilayer MoTe₂ — and the solution it proposes is a fingerprint written in light. In a preprint posted to arXiv on November 1, 2026, Haochen Wang, Shengxian Li, Hongyu Lu, Mingxing Chen, and Wang Yao predict that condensates of hybrid moiré excitons in twisted bilayer MoTe₂ radiate in two sharply distinct ways: the biexciton condensate — the molecular type — decays by superradiant emission of entangled photon pairs, while the exciton condensate — the atomic type — is dark at rest and becomes superradiant only when it carries a supercurrent.

Every verb in that summary needs its qualifier, and this desk will keep applying it. This is a calculation, not a measurement: no condensate has been observed emitting these biphotons, and no supercurrent has been switched on in a moiré exciton condensate in the laboratory. What the paper offers is a prediction with specific, testable conditions attached.

The emitter: one trap, a structured photon

The starting point is a single moiré trap in twisted bilayer MoTe₂, a material this desk has covered before as the host of fractional Chern insulators and unconventional superconductivity. The moiré superlattice confines excitons into an array of traps with a periodicity up to two orders of magnitude smaller than the optical wavelength. That deep-subwavelength spacing is what makes cooperative radiation possible: emitters packed much closer than a wavelength can radiate together, coherently, rather than independently.

The exciton in each trap is a hybrid: an interlayer exciton — electron and hole in opposite layers, tightly confined into compact S- and P-type orbitals — mixed with an intralayer exciton, whose orbitals are far more extended. The mixing is set by the local band alignment and is electrically tunable through a displacement field. In the paper’s worked example, a twist angle of 3.15 degrees and zero displacement field, the hybridization weight W is 0.12, fixed by a binding-energy difference between the intra- and interlayer excitons of 72 meV.

The layer-hybridization selection rule gives the ground-state exciton a property no ordinary quantum-dot exciton has: it emits a photon carrying orbital angular momentum — an azimuthally polarized vector-vortex mode with one unit of OAM, its angular intensity proportional to sin²θ. But the same selection rule that makes the emission structured also makes it feeble: the single-site radiative rate is of order ten per millisecond, a lifetime of order a hundred microseconds. A lone ground-state exciton in a moiré trap is, for all optical purposes, dark.

Why the ground state stays dark — until it doesn’t

Projecting the dipole-dipole radiative kernel onto the Bloch states of the exciton array yields a surprise: the exciton Bloch state at zero in-plane momentum — the state a condensate at rest would occupy — is completely dark. Its radiative rate vanishes at k = 0 and grows with momentum, reaching roughly ten thousand times the single-site rate at k of order one inverse optical wavelength, a lifetime of order nanoseconds. The P-orbital character of the intralayer component is responsible: the same wavepacket structure that gives the single exciton its orbital-angular-momentum photon makes the zero-momentum collective state optically silent.

A condensate sitting still in its lattice would therefore not announce itself in photoluminescence at all — the darkness is itself the signature of the hybrid exciton’s orbital structure, and it sets up the contrast with the biexciton channel, which behaves in exactly the opposite way.

Two condensates, one lattice

The many-body side treats the hybrid moiré excitons as bosons on the moiré lattice, governed by an extended Bose-Hubbard model with parameters computed from first-principles Wannier orbitals. For the 3.15-degree twist at W = 0.12: hopping t = 0.63 meV, onsite repulsion V₀ = 62 meV, nearest-neighbor attraction V₁ = −6 meV, next-nearest-neighbor repulsion V₂ = 1.8 meV, and third-nearest-neighbor attraction V₃ = −1.4 meV. The sign pattern is the physics: the dipole-dipole interaction is strongly attractive between excitons on nearest-neighbor sites of opposite sublattices — the MX and XM stacking regions, whose moiré electric polarizations point in opposite directions — and repulsive within each sublattice. The attraction binds pairs of excitons on neighboring MX–XM bonds into quadrupolar biexcitons. The system therefore supports two kinds of bosonic matter at once: atomic excitons and molecular biexcitons, coexisting rather than competing.

Density-matrix renormalization group calculations on cylinders thirty sites long and four, six, and eight sites wide show both the exciton and biexciton momentum-space populations developing a macroscopic peak centered at zero momentum, both extensive — growing linearly with cylinder width and diverging in the thermodynamic limit, the hallmark of condensation. Crucially, the two condensates have comparable weights — a coupled atomic-molecular condensate, the excitonic analogue of those familiar from ultracold gases, except here the “molecules” are bound by dipole attraction on a lattice and the whole system is, in principle, optically readable.

A finite-temperature cluster mean-field calculation puts the transition temperature for both condensates at roughly 2 to 14 K, at W around 0.12 to 0.13 — a range the authors note has already been reached in existing twisted MoTe₂ experiments. But it is a mean-field estimate, and mean-field estimates of transition temperatures in two-dimensional systems are exactly the kind of number experiments have a habit of revising downward. Read it as “not obviously out of reach,” not as a guarantee.

The biphoton channel: the molecule announces itself

The headline result concerns how the biexciton condensate decays. A biexciton on a bond decays by emitting two photons in a cascade, passing through an intermediate recoil-exciton state — the first photon carrying the biexciton-to-exciton energy difference, the second the remaining exciton energy. Because the biexciton Bloch state is superradiant at zero momentum — the opposite of the exciton’s darkness there — the condensate decays through superradiant emission of a momentum-entangled photon pair, at a rate scaled by the condensate occupation itself.

The two-photon wavefunction is entangled in the photons’ momenta, with emission angles fixed by in-plane momentum conservation and a time ordering enforced by the cascade. The superradiant enhancement for any single biexciton state is modest, but the macroscopic occupation of the condensate multiplies it by the number of condensed pairs — which is what makes the emission a fingerprint of condensation rather than of biexcitons in general: uncondensed biexcitons would emit pairs too, but without the enhancement and momentum-space concentration that mark the condensed state.

Here the language needs care. The paper’s title describes the pairs as “path-entangled biphotons”; the derivation describes a momentum-entangled photon pair. These are two descriptions of the same predicted object, not two different claims — but a reader looking for a Bell-pair source should not overread either phrase: the paper computes no Bell parameter, concurrence, or fidelity to a target entangled state. What is predicted is a source of entangled light, not a characterized one.

The supercurrent switch: the atom announces itself, on demand

If the biexciton condensate is the talkative partner, the exciton condensate is the silent one — until it is set in motion. The second signature is a way to switch the atomic condensate’s radiation on, electrically, without touching the molecular condensate at all.

The mechanism is a gradient in the displacement field, applied in the plane of the sample. Because excitons on the MX and XM sublattices carry opposite electric dipoles, the gradient imprints a phase gradient on the condensate that is opposite on the two sublattices — the excitonic analogue of a London-type acceleration term. It drives a counterflow of exciton supercurrent, pushing the exciton condensate’s momentum peak away from zero and splitting it into two peaks at finite momenta, while the biexciton peak stays put: the quadrupolar biexciton, with no net dipole, does not couple to the field gradient. At finite momentum the exciton Bloch state is superradiant, so the displaced condensate emits highly directional single photons, with an intensity set by the supercurrent density and a direction encoding the phase gradient.

The numbers make this a switch, not a dial. A small phase gradient of about 0.02 times the optical wavevector is enough to flip the dominant radiation channel from the biexciton’s photon pairs to the exciton’s single photons; push the gradient past one optical wavevector, and the exciton superradiance is quenched again and the biexciton pairs return as the sole channel. The field gradient is therefore predicted to toggle the sample’s light between two qualitatively different outputs — entangled pairs, then directional single photons, then pairs again. That specificity is its strength: an experiment that sees the predicted sequence as the gradient is swept would be hard to explain any other way. It is also, to be blunt, a demanding experiment — patterned gates with a designed field profile, at kelvin temperatures — and the proposed detection scheme, oblique emission detection combined with displacement-field-gradient control, is chosen precisely because the signatures live at finite momentum and finite angle, where normal-incidence photoluminescence would miss them.

What would count as confirmation — and what would not

A prediction this specific invites a checklist, because the most likely misreading would be to treat any two-photon coincidence from a twisted MoTe₂ device as confirmation. First, the pairs. Coincidence counts between photons at the biexciton and exciton energies would be consistent with biexciton decay — but biexcitons emit pairs without any condensate. Confirmation requires the superradiant signature: a pair rate that grows with the condensed population, concentrated at the predicted momentum and angle, and appearing only below the condensation temperature. A pair signal present at all temperatures would be biexciton luminescence, not a condensate fingerprint.

Second, the switch. The gradient experiment is the sharper test, because it predicts a sequence, not just a signal: pairs dominant at zero gradient, directional single photons dominant at a small gradient, pairs again at a large gradient. Observing the full sequence would be strong evidence for the whole picture; observing only one step would be suggestive, not conclusive.

Third, the condensate itself. All of this presupposes that an exciton-biexciton condensate actually forms in twisted MoTe₂ at kelvin temperatures — which has not been demonstrated — and the fingerprints are proposed as the way to establish that formation, so there is a circularity to be honest about: the fingerprints are the evidence for the condensate, and the condensate is the explanation of the fingerprints. The way out of the circle is the specificity of the predictions, which are collectively hard to mimic without the underlying many-body order. The usual caveats bound how literally the numbers should be taken: the Bose-Hubbard parameters come from one twist angle and one hybridization weight; the DMRG results are on finite cylinders; the transition-temperature estimate is mean-field; and the radiative treatment is perturbative.

Why this matters, stated carefully

It is tempting to read any result about entangled photon pairs as a step toward quantum communication, and the paper does describe its predicted pairs as “a novel source of entangled light.” The honest version is longer: a usable source needs brightness, collection efficiency, a known state fidelity, and a heralding scheme, none of which this paper establishes. What it establishes is that a moiré exciton condensate, if it forms, would emit entangled pairs by the very nature of its decay — the entanglement is the system’s own radiation, not something engineered on top of it. That is a genuinely new idea. But it is an idea about a source, not a source.

The deeper significance, and the one the authors foreground, is diagnostic. Moiré excitons are a Bose-Hubbard quantum simulator, with electrical tunability that cold atoms cannot match — but a simulator whose phases cannot be read out is a black box. Cooperative radiation, the paper argues, is the readout: the light a moiré lattice emits encodes its many-body order, because the emitters are spaced far below the wavelength and radiate as one. The two fingerprints would give optics a way to tell the two condensate species apart in a single device, in real time, without destroying the sample.

The right summary, then: five theorists in Hong Kong and Hunan have calculated that twisted bilayer MoTe₂ should host coexisting exciton and biexciton condensates, that the molecular condensate should decay by superradiant emission of entangled photon pairs while the atomic condensate stays dark at rest, and that a displacement-field gradient should switch the atomic condensate on by driving a supercurrent. Until experiments report, the claims remain what they are: a precise, falsifiable prediction, awaiting its measurement — and the version of the story in which the pairs have already been seen should be resisted.

Sources and further reading

  • Haochen Wang, Shengxian Li, Hongyu Lu, Mingxing Chen, and Wang Yao, “Superradiance fingerprints of moiré exciton-biexciton condensates: path-entangled biphoton vs. supercurrent-enabled one-photon channel,” arXiv:2611.01515 (submitted November 1, 2026; manuscript dated November 3, 2026). All figures quoted above are taken from this preprint.
  • H. Xu, W. Yao, and J. Li, “Exotic cooperative quantum optics of moiré exciton superlattices,” arXiv:2603.06998 (2026): the cooperative-radiation framework this preprint builds on.
  • J. Kumlin, A. Srivastava, and T. Pohl, “Superradiance of strongly interacting dipolar excitons in moiré quantum materials,” Physical Review Letters 134, 126901 (2025): the closest prior treatment of superradiance in this setting.
  • H. Yu and W. Yao, “Luminescence anomaly of dipolar valley excitons in homobilayer semiconductor moiré superlattices,” Physical Review X 11, 021042 (2021): the hybrid moiré exciton physics underlying the preprint’s emitter model.
  • Note on verification: this article reports a theoretical prediction. No moiré exciton-biexciton condensate of the kind described has been experimentally demonstrated as of this writing, and no superradiant biphoton emission from such a condensate has been measured. All figures attributed to the preprint are results of model calculations, not measurements. The preprint was submitted to arXiv on November 1, 2026; its manuscript is dated November 3, 2026. No numerical value for the biexciton binding energy is quoted, because the preprint states the pair energy only as the exciton energy minus the binding energy, without giving a headline number.