Biexciton–Exciton Cascade in Semiconductor Nanostructures

• Biexciton–exciton cascade is a sequential radiative decay in semiconductor nanostructures emitting two correlated photons via a biexciton to exciton to ground state transition.
• It features a three-level (or four-level with fine structure) energy ladder where exciton binding energies and fine structure splitting govern spectral and temporal photon characteristics.
• The cascade enables advanced quantum optics experiments by allowing spectral tuning through electric fields, pressure, and cavity QED, with applications in secure communications and quantum computing.

The biexciton–exciton cascade is a sequential radiative decay in which a doubly excited semiconductor nanostructure relaxes from a biexciton state to a single-exciton state and then to the ground state, emitting two photons in the process. In its minimal form the ladder is written as $|XX\rangle \to |X\rangle \to |0\rangle$, or, when the bright exciton doublet is resolved, as $$|XX\rangle \to |X_H\rangle,|X_V\rangle \to |0\rangle$$. The cascade is central to solid-state quantum optics because the two decay paths can generate photon pairs with strong quantum correlations, including polarization entanglement, and because the same mechanism now appears across self-assembled epitaxial dots, colloidal nanocrystals, localized emitters in monolayer semiconductors, chiral waveguides, and more general radiative quantum cascades in interacting bosonic ladders (Kaniber et al., 2010, Lubin et al., 2021, He et al., 2017, Scarpelli et al., 2022).

1. Level structure, transition energies, and binding-energy conventions

The cascade is built from three neutral configurations: the ground state $|0\rangle$, the single exciton $|X\rangle$, and the biexciton $|XX\rangle$. In self-assembled InGaAs quantum dots, the neutral exciton $1X^0$ is a bound electron–hole pair, while the biexciton $2X^0$ contains two electrons and two holes, with electron–electron and hole–hole repulsions in addition to electron–hole attraction. The radiative sequence is

$$|XX\rangle \to |X\rangle + \gamma_1,\qquad |X\rangle \to |0\rangle + \gamma_2,$$

with photon energies $\hbar\omega_{XX\to X}$ and $\hbar\omega_{X\to 0}$ set by the two transition energies (Kaniber et al., 2010).

A persistent technical point is that “biexciton binding energy” is used with more than one convention. In the colloidal-dot spectroscopy work, the biexciton photon is denoted $$|XX\rangle \to |X_H\rangle,|X_V\rangle \to |0\rangle$$0, the exciton photon $$|XX\rangle \to |X_H\rangle,|X_V\rangle \to |0\rangle$$1, and the binding energy is defined as

$$|XX\rangle \to |X_H\rangle,|X_V\rangle \to |0\rangle$$2

so $$|XX\rangle \to |X_H\rangle,|X_V\rangle \to |0\rangle$$3 means the biexciton transition is red-shifted relative to the exciton transition (Lubin et al., 2021). In the hydrostatic-pressure study of self-assembled dots, the experimentally convenient definition is

$$|XX\rangle \to |X_H\rangle,|X_V\rangle \to |0\rangle$$4

again in terms of observed transition energies; $$|XX\rangle \to |X_H\rangle,|X_V\rangle \to |0\rangle$$5 is called binding and $$|XX\rangle \to |X_H\rangle,|X_V\rangle \to |0\rangle$$6 antibinding (Wu et al., 2013). The same physical ladder is therefore often discussed either through total-state energies or through directly measured photon energies.

For entanglement applications the minimal three-level picture is usually refined to a four-level scheme with two bright excitons, $$|XX\rangle \to |X_H\rangle,|X_V\rangle \to |0\rangle$$7 and $$|XX\rangle \to |X_H\rangle,|X_V\rangle \to |0\rangle$$8, split by the fine-structure splitting (FSS). In the linear basis the exciton Hamiltonian is commonly written as a splitting term between the two bright states, for example

$$|XX\rangle \to |X_H\rangle,|X_V\rangle \to |0\rangle$$9

under continuous-wave pumping (Tur et al., 5 Mar 2025). In chiral-waveguide formulations one may instead use circular excitons $|0\rangle$0, while in symmetric colloidal nanocrystals a triplet bright exciton $|0\rangle$1 can replace the conventional heavy-hole doublet entirely (González-Ruiz et al., 2023, Mantsevich et al., 7 Nov 2025).

2. Entanglement generation, fine structure, and path distinguishability

In the ideal symmetric limit, the biexciton can decay through two indistinguishable intermediate excitons, so the two-photon output is Bell-like: $|0\rangle$2 This is the standard polarization-entanglement mechanism of the cascade: which-path information is absent except for polarization, and the pair is emitted in a coherent superposition of the two branches (Kaniber et al., 2010).

Finite FSS modifies, but does not trivially destroy, this picture. In the time-resolved tomographic study of a resonantly prepared biexciton, the emitted two-photon state at a fixed delay $|0\rangle$3 between the biexciton and exciton photons is

$|0\rangle$4

where $|0\rangle$5 is the exciton precession period induced by the FSS. For any fixed $|0\rangle$6, the state remains maximally entangled; the measured negativity is reduced only by finite temporal resolution, with

$|0\rangle$7

for a square time window $|0\rangle$8 (Winik et al., 2017). This makes the observed entanglement a joint property of the emitter and the measurement bandwidth.

Two broad strategies then recur across the literature. One is to suppress the FSS so that the two exciton branches become degenerate. The other is to engineer the spectral–temporal structure of the cascade so that the path information becomes irrelevant even with finite FSS. The lateral-field and hydrostatic-pressure works fall in the second category: both tune the exciton and biexciton transition energies into special resonant conditions needed for time-reordering schemes, rather than canceling the FSS itself (Kaniber et al., 2010, Wu et al., 2013).

The same logic reappears in integrated nanophotonics, but with path instead of polarization as the encoded qubit. In a chiral waveguide, ideal spin–momentum locking maps the two circular cascade branches into opposite propagation directions and yields the path-entangled state

$|0\rangle$9

For imperfect chirality and finite FSS, the concurrence becomes

$|X\rangle$0

showing explicitly how chirality, exciton precession, and timing jitter trade against one another (González-Ruiz et al., 2023).

Recent theory also isolates a deeper limit: hyperfine coupling to host nuclei. In symmetric colloidal nanocrystals with a triplet bright exciton, the concurrence can be written as

$|X\rangle$1

with $|X\rangle$2 the Overhauser-field dispersion, $|X\rangle$3 the exciton lifetime, and $|X\rangle$4 an effective hyperfine coupling set by anisotropy and exchange. At $|X\rangle$5, one obtains $|X\rangle$6, which suppresses the dominant electron-hyperfine contribution; with representative CdSe/CdTe parameters the predicted upper limit is $|X\rangle$7 (Mantsevich et al., 7 Nov 2025).

3. Spectral and temporal engineering of the cascade

The biexciton–exciton splitting is a controllable system parameter rather than a fixed spectroscopic detail. In a lateral electric-field device containing single self-assembled In$|X\rangle$8Ga$|X\rangle$9As dots, the neutral exciton $|XX\rangle$0 redshifts while the biexciton transition $|XX\rangle$1 blueshifts under the in-plane quantum confined Stark effect. For the illustrated dot, $|XX\rangle$2 meV and $|XX\rangle$3 meV at zero field, so $|XX\rangle$4 meV, but the opposite Stark shifts drive the lines into resonance at $|XX\rangle$5 kV/cm (Kaniber et al., 2010). This electrically realizes the core spectral condition of a time-reordered cascade.

Hydrostatic pressure provides a second route with a much larger tuning range. In single (In,Ga)As/GaAs dots, exciton transition energies blue-shift linearly with slopes between 81 and 93 meV/GPa, and the maximum shift reaches 330–380 meV near 4.2–4.4 GPa. Because the biexciton line shifts slightly differently, the biexciton binding changes nearly linearly with pressure and can cross from antibinding to binding. In one dot the $|XX\rangle$6-polarized XX and X lines coincide at 1.62 GPa, the $|XX\rangle$7-polarized pair at 2.07 GPa, and an across-generation coincidence $|XX\rangle$8, $|XX\rangle$9 appears at 1.97 GPa (Wu et al., 2013). Pressure therefore offers access to color-indistinguishable and across-generation resonances, although it also increases the FSS by tens of $1X^0$0eV/GPa.

A third approach is environmental rather than energetic. In a weak-coupling quantum-dot–metal-nanoparticle hybrid, the structured plasmonic local density of states broadens the exciton and biexciton lines. For suitable nanoparticle radius $1X^0$1 and separation $1X^0$2, the broadened $1X^0$3- and $1X^0$4-polarized spectra overlap strongly enough that the FSS becomes irrelevant at the level of detection, even though the exciton energies themselves remain split (Moradi et al., 2017). In this formulation, entanglement recovery is driven by linewidth engineering rather than level tuning.

Cavity QED can be used more selectively still. A recent open-microcavity study analyzes the regime where the cavity is resonant only with the $1X^0$5 transition. Resonant two-photon excitation prepares $1X^0$6, the H-polarized upper leg is Purcell enhanced, and the lower $1X^0$7 leg remains far detuned by approximately the biexciton binding energy. Experimentally this yields $1X^0$8 and HOM visibility $1X^0$9 for the emitted XX photon, with theory indicating that selective Purcell enhancement overcomes the usual lifetime-ratio limitation of the cascade as a single-photon source (Heinisch et al., 20 Feb 2026).

Strong coherent driving opens still another regime. Under two-photon-resonant excitation of a dressed biexciton, the bare $2X^0$0 cascade is replaced by a ladder of dressed states, and Purcell-enhanced two-photon “leapfrog” transitions can dominate over the ordinary sequential cascade. The same biexcitonic level structure then supports antibunched or bunched photon pairs and, with a polarization-symmetric cavity, entangled two-photon emission (Muñoz et al., 2015).

4. Material platforms and physical realizations

Self-assembled III–V quantum dots remain the canonical implementation. They provide the narrow cryogenic lines, bright-state doublets, and Coulomb-controlled biexciton shifts on which the standard cascade picture was built, and they support multiple external controls including lateral fields, hydrostatic pressure, cavities, chiral waveguides, and coherent biexciton preparation (Kaniber et al., 2010, Wu et al., 2013, Winik et al., 2017, González-Ruiz et al., 2023). Site-controlled pyramidal quantum dots extend this toward deterministic placement; stacked double-pyramidal structures have been reported with biexciton binding energies close to zero and a sequence of two photons with nearly the same energy from the biexciton–exciton–ground-state cascade (Moroni et al., 2018).

Colloidal CdSe/CdS/ZnS quantum dots bring the same ladder to room temperature, but with thermal broadening, spectral diffusion, and strong Auger processes. Heralded spectroscopy isolates rare radiative cascades by selecting two photons from the same excitation pulse, assigning the earlier photon to $2X^0$1 and the later to $2X^0$2. In the representative dot the extracted biexciton binding energy is $2X^0$3 meV despite linewidths of order 35–50 meV, and across 30 dots the mean binding is about 6 meV; the biexciton quantum yield is low, about 9%, but the cascade remains observable and microscopically informative (Lubin et al., 2021).

Localized emitters in monolayer WSe$2X^0$4 realize a two-dimensional variant of the cascade. In a localized center on GaInP, the exciton line P2 appears at $2X^0$5 eV and the biexciton-related line P1 at $2X^0$6 eV, giving a separation of about $2X^0$7 meV. The power-law exponents are $2X^0$8 for P2 and $2X^0$9 for P1, the lifetimes are $$|XX\rangle \to |X\rangle + \gamma_1,\qquad |X\rangle \to |0\rangle + \gamma_2,$$0 ns and $$|XX\rangle \to |X\rangle + \gamma_1,\qquad |X\rangle \to |0\rangle + \gamma_2,$$1 ns, and the cross-correlation shows bunching at positive delay and antibunching at negative delay, establishing a localized XX–X cascade in a monolayer semiconductor (He et al., 2017).

A complementary WSe$$|XX\rangle \to |X\rangle + \gamma_1,\qquad |X\rangle \to |0\rangle + \gamma_2,$$2 platform is the BN-encapsulated monolayer. There the neutral biexciton exists only in the charge-neutral regime, with a binding energy of about 16–17 meV, while a trion–exciton complex $$|XX\rangle \to |X\rangle + \gamma_1,\qquad |X\rangle \to |0\rangle + \gamma_2,$$3 appears in lightly electron-doped material (1806.03775). This work does not measure photon cascades directly, but it fixes the spectral and charge-state conditions under which a neutral XX–X ladder can exist in that material.

5. Experimental observables and cascade-specific spectroscopies

The simplest diagnostics of a biexciton–exciton cascade are still power dependence, lifetime hierarchy, and second-order correlations. In localized WSe$$|XX\rangle \to |X\rangle + \gamma_1,\qquad |X\rangle \to |0\rangle + \gamma_2,$$4, the XX line is superlinear, the X line sublinear, both lines are antibunched individually, and their cross-correlation is asymmetric: under continuous-wave excitation $$|XX\rangle \to |X\rangle + \gamma_1,\qquad |X\rangle \to |0\rangle + \gamma_2,$$5 for negative delay and $$|XX\rangle \to |X\rangle + \gamma_1,\qquad |X\rangle \to |0\rangle + \gamma_2,$$6 for positive delay, which is the direct signature that XX precedes X (He et al., 2017).

At higher technical sophistication, the cascade can be reconstructed in both energy and polarization. Room-temperature colloidal-dot heralded spectroscopy builds a two-dimensional histogram of the first-photon energy versus the second-photon energy, directly revealing $$|XX\rangle \to |X\rangle + \gamma_1,\qquad |X\rangle \to |0\rangle + \gamma_2,$$7 on a pair-by-pair basis even when the two spectra completely overlap in conventional photoluminescence (Lubin et al., 2021). Under continuous-wave excitation in a nanowire QD, a Lindblad model fitted to 36 time-resolved polarization correlations $$|XX\rangle \to |X\rangle + \gamma_1,\qquad |X\rangle \to |0\rangle + \gamma_2,$$8 reproduces the full set of measurements and links the observed oscillations and asymmetries to FSS, radiative decay, and pumping (Tur et al., 5 Mar 2025). Under pulsed resonant excitation, full time-resolved tomography shows that the experimentally observed entanglement of a biexciton cascade is set by temporal resolution rather than by intrinsic loss of coherence (Winik et al., 2017).

The same cascade logic extends beyond two-photon pair sources. In quantum-cascade correlation spectroscopy with exciton polaritons, the emitted field is treated as a radiative cascade down an anharmonic many-body ladder whose individual transitions are not spectrally resolved. Narrowband filtering and $$|XX\rangle \to |X\rangle + \gamma_1,\qquad |X\rangle \to |0\rangle + \gamma_2,$$9 measurements then reveal two- and three-body Feshbach resonances through the energy dependence of the photon correlations (Scarpelli et al., 2022). This does not reproduce a biexciton–exciton ladder literally, but it generalizes the idea that cascade correlations are a sensitive probe of the underlying bound-state structure.

6. Extensions, limitations, and current frontiers

The biexciton–exciton cascade is no longer confined to the original four-level entangled-pair picture. One extension is upward in excitation number: deterministic triexciton preparation in an InGaAs dot produces a three-photon radiative cascade through triexciton, biexciton, and exciton manifolds, and third-order intensity correlations recover the expected lower-order cascade statistics by temporal averaging (Schmidgall et al., 2014). Another extension is coherent state preparation: in a GaAs dot in a low-$\hbar\omega_{XX\to X}$0 micropillar, coherent swing-up excitation can prepare either the exciton directly or the biexciton via a distinct SUPER resonance. The XX-mediated case is identified by biexciton emission, unpolarized exciton emission, and an extended exciton lifetime due to the XX–X cascade (Piccinini et al., 24 Oct 2025).

Several practical constraints recur across otherwise disparate platforms. In lateral-field devices, tuning X and XX into resonance reduces photoluminescence intensity at high field because carrier capture is suppressed and tunneling escape increases (Kaniber et al., 2010). Under hydrostatic pressure, the same large tuning range that enables XX–X coincidences also amplifies the FSS, so pressure is a tool for spectral engineering rather than for standard FSS cancellation (Wu et al., 2013). In colloidal dots, the biexciton quantum yield is only about 9% and valid pair rates scale quadratically with detector efficiency, while inter-pixel crosstalk must be corrected at the $\hbar\omega_{XX\to X}$1 level per detection (Lubin et al., 2021). In chiral-waveguide implementations, imperfect chirality, asymmetric decay rates, and timing jitter reduce concurrence even when the underlying path-entangled state is deterministic (González-Ruiz et al., 2023). Under high-power SUPER excitation, microsecond-scale bunching indicates charge-noise dynamics that can degrade source performance even when $\hbar\omega_{XX\to X}$2 remains low (Piccinini et al., 24 Oct 2025).

What emerges from these developments is not a single protocol but a family of related quantum-optical ladders. The biexciton–exciton cascade remains the fundamental two-step process, yet current research treats it simultaneously as an entangled-pair source, a spectroscopic probe of exciton–exciton correlations, a tunable testbed for cavity and plasmonic electrodynamics, a route toward higher-order cascades, and a platform whose ultimate performance may be set by hyperfine physics rather than by fine-structure splitting alone (Mantsevich et al., 7 Nov 2025).