An exciton is one of the most seductive objects in solid-state physics. It is a bound state of an electron and a hole, held together by Coulomb attraction, created when a semiconductor absorbs a photon and annihilated when it recombines and emits one. For a brief lifetime it behaves like a neutral, mobile, hydrogen-like particle — an artificial atom that can be created with light, moved with fields and strain, and made to interact with other excitons. It is easy to see why someone would ask whether it could be a qubit.
The honest answer is that an exciton is not a qubit in the way a trapped ion or a superconducting transmon is a qubit. It is a quasiparticle with a finite lifetime, a center-of-mass motion, internal structure, and strong coupling to its environment. Some of its quantum properties — superposition, coherence, entanglement — have been demonstrated in carefully controlled experiments. Others, notably the long-lived, high-fidelity, individually addressable two-level system that fault-tolerant computing demands, remain aspirational. Understanding the gap between those two statements is the key to understanding the whole excitonic quantum program.
What a qubit actually requires
A qubit is not simply “something quantum.” It is a controllable two-level system that satisfies a demanding list of requirements, often summarized in the DiVincenzo criteria: well-characterized qubits, initialization to a known state, long coherence relative to gate times, a universal set of gates, and measurement of individual qubits. Fault tolerance adds the further demand that gate errors fall below a threshold — roughly on the order of one error per thousand to ten thousand operations, depending on the code and architecture — so that error correction can suppress logical errors rather than amplify them.
Most physical qubits meet these criteria imperfectly and compensate with engineering. Superconducting qubits have fast gates but short coherence and require millikelvin cryogenics. Trapped ions have long coherence but slow gates and scaling challenges. Photonic qubits avoid decoherence by flying but struggle with deterministic two-qubit gates. Excitons enter this landscape with a distinctive profile: they are fast, optically addressable, and naturally interfaced to photons, but they are short-lived and strongly coupled to phonons, disorder, and charge noise.
The question “can an exciton be a qubit?” therefore has to be refined. Which exciton, in which material, encoding information in which degree of freedom, and for what task — computation, simulation, sensing, or communication? Different answers apply to different encodings, and conflating them is the most common source of hype in this field.
The exciton as an artificial atom
In a direct-gap semiconductor such as gallium arsenide, an exciton resembles a hydrogen atom with a much larger Bohr radius and a much smaller binding energy, because the crystal screens the Coulomb interaction and the electron and hole have small effective masses. The binding energy is typically on the order of a few to roughly ten millielectronvolts in bulk III-V materials, and substantially larger — hundreds of millielectronvolts — in monolayer transition metal dichalcogenides (TMDs) such as WSe2 and MoSe2, where reduced screening strengthens the attraction. That difference matters enormously: weakly bound excitons ionize at modest temperatures, while tightly bound TMD excitons can survive at room temperature, at least as spectroscopic features.
The internal structure gives excitons several potential degrees of freedom for encoding information. The spin of the electron and hole, the valley index in TMDs (the K versus K-prime valley in momentum space), the orbital state, and the presence or absence of the exciton itself (a “presence” or occupation encoding) have all been proposed. Valley excitons in monolayer TMDs are particularly attractive because the valley index is coupled to the helicity of emitted light through optical selection rules, giving a direct optical handle on a binary degree of freedom.
But an exciton is also a composite object moving through a crystal. Its center-of-mass momentum, its scattering off phonons and defects, and its eventual recombination all compete with any attempt to treat it as a stationary two-level system. The radiative lifetime — often on the order of picoseconds to nanoseconds depending on material and structure — sets a hard ceiling on how long an exciton qubit can exist before it disappears. Any gate sequence must complete well within that window, and any error-correction scheme must reckon with qubit loss, not just qubit error.
Superposition and coherence: what has been shown
The most basic quantum property, superposition, has been demonstrated for excitons in multiple settings. Perhaps the cleanest example is the exciton-polariton, the hybrid light-matter quasiparticle formed when an exciton couples strongly to a cavity photon. Polaritons inherit coherence from their photonic component and interact via their excitonic component, and their condensates exhibit macroscopic phase coherence. That is a genuine quantum phenomenon, but it is coherence of a many-particle condensate, not coherence of an individually addressable qubit.
At the single-exciton level, coherent control has been shown in semiconductor quantum dots — nanoscale islands that confine a single exciton in all three dimensions. Quantum dots are sometimes called artificial atoms precisely because confinement discretizes their levels and suppresses the center-of-mass motion that plagues extended excitons. Rabi oscillations, Ramsey interference, and spin echo experiments have been performed on single quantum-dot excitons and on resident charges in dots, demonstrating coherent manipulation on picosecond to nanosecond timescales. These are real, reproducible results, reported across many groups over two decades.
The catch is coherence time. Excitonic coherence in extended systems is typically limited by phonon scattering and disorder to picosecond timescales at elevated temperatures, improving substantially at cryogenic temperatures and in high-quality samples. Spin coherence of a resident electron or hole in a quantum dot can be much longer — microseconds or more under dynamical decoupling in favorable cases — but that is coherence of a spin qubit hosted in a dot, not coherence of the exciton itself. The distinction matters: once the exciton recombines, the “exciton qubit” is gone. Spin qubits that use optical excitation for initialization and readout are a hybrid story, and a promising one, but they should not be described as excitons serving as long-lived qubits.
Gates: fast, but what kind?
Excitons offer one genuine advantage: speed. Optical pulses can manipulate excitonic states on femtosecond to picosecond timescales, orders of magnitude faster than microwave control of superconducting qubits or laser control of trapped ions. In principle, many gate operations could fit inside even a short coherence window if the gates are fast enough. This speed is why excitonic and polaritonic systems are attractive for quantum simulation and for ultrafast optical switching, where the relevant figure of merit is the ratio of interaction strength to loss, not the absolute coherence time.
The difficulty is the two-qubit gate. Useful quantum computation requires entangling operations between individually addressable qubits with high fidelity. Exciton-exciton interactions exist — dipole-dipole repulsion, exchange, and phase-space filling all contribute — and nonlinearities at the few-exciton level have been observed. But demonstrating a high-fidelity, deterministic, entangling gate between two individually addressable exciton qubits, with the kind of characterization (tomography, randomized benchmarking) routine in superconducting and ion-trap platforms, has not been achieved. What has been shown is nonlinear optical response, conditional phase shifts in polariton systems, and entanglement in photon pairs emitted by biexciton cascades — all important, none equivalent to a universal excitonic gate set.
The biexciton cascade deserves a note because it is sometimes overclaimed. When a quantum dot containing two electron-hole pairs (a biexciton) decays, it emits two photons in succession. Under suitable symmetry conditions, those two photons can be polarization-entangled. This is a demonstrated source of entangled photon pairs, and it is a genuine quantum technology. But the entanglement resides in the emitted photons, and the exciton system serves as the emitter, not as a pair of qubits undergoing a gate. Conflating a good entangled-photon source with a quantum computer is a category error the field should resist.
The lifetime problem, stated plainly
Every exciton qubit proposal must confront the same arithmetic: the qubit disappears when the exciton recombines. Radiative lifetimes vary widely — from sub-picosecond for some polaritons to nanoseconds for localized quantum-dot excitons — but they are short compared with the coherence times of mature qubit platforms. Proposals to circumvent this include encoding information in longer-lived degrees of freedom (spin, valley, or dark excitons that couple weakly to light), transferring quantum information from excitons to photons or spins before recombination, and using excitons only transiently as mediators of interaction between longer-lived qubits.
Dark excitons are an instructive example. In TMDs, certain spin-forbidden or momentum-forbidden exciton states cannot recombine optically without assistance, giving them substantially longer lifetimes than bright excitons. They are harder to create and read out precisely because they are dark, but they illustrate the general trade: the properties that make an exciton easy to control optically also make it decay quickly. Long-lived and optically bright are in tension, and every architecture chooses a point on that tradeoff.
None of this means excitonic quantum information is a dead end. It means the most credible near-term roles for excitons in quantum technology are not as standalone qubits in a universal gate-based computer, but as interfaces, transducers, and simulators: converting between photons and matter qubits, mediating interactions, generating entangled photons, and emulating many-body quantum systems that are hard to simulate classically. Those are substantial roles, and they play to excitons’ actual strengths rather than to a borrowed qubit narrative.
Where the field actually stands
A fair summary of the state of play is as follows. Single-exciton coherence and optical control are demonstrated, most cleanly in quantum dots. Entangled photon generation from biexciton cascades is demonstrated. Polariton condensates and their coherence are demonstrated. Valley polarization and valley coherence in TMDs are demonstrated, with coherence times that depend strongly on material quality, temperature, and encapsulation. What has not been demonstrated is a universal, fault-tolerant, exciton-qubit processor, or even a small multi-qubit excitonic processor with benchmarked gate fidelities comparable to leading platforms.
That gap is not a failure; it reflects the youth of the field and the difficulty of the problem. Excitonic quantum research is best understood as exploring a different region of the design space — fast, optically interfaced, potentially room-temperature-adjacent — rather than as a direct competitor to superconducting or ion-trap processors on their own metrics. Its most likely near-term impact is in quantum simulation, photonic quantum technologies, and hybrid systems where excitons connect components that are each good at one thing.
Readers should therefore treat claims of “exciton quantum computers” with the same skepticism they would apply to any claim that a newly demonstrated quantum effect implies a computer. Ask which degree of freedom encodes the qubit, what the measured coherence and gate times are, whether two-qubit gates have been benchmarked, and whether the result is a proposal, a simulation, or a measurement. The excitonic literature contains all three, and the distinction is everything.
Sources and further reading
- Reviews of exciton-polariton physics and polariton condensation in semiconductor microcavities, published in Reviews of Modern Physics and Nature Physics, covering strong coupling, condensation, and coherence.
- Review literature on monolayer transition metal dichalcogenides, valley physics, and valley coherence, in Nature Materials, Nature Reviews Materials, and Annual Review of Condensed Matter Physics.
- Review and experimental literature on semiconductor quantum dots, single-exciton coherent control, and entangled photon generation from biexciton cascades, in Nature, Science, Physical Review Letters, and Nature Photonics.
- The DiVincenzo criteria for quantum computation, as originally formulated in the quantum information literature and summarized in standard quantum computation textbooks.
- Review literature on dark excitons and long-lived exciton states in two-dimensional semiconductors, in Nature Communications and Physical Review B.
- Note on verification: no specific device-level coherence or gate-fidelity figures are quoted above as established records, because such numbers vary by material, structure, temperature, and measurement method. Any article citing a single record number for “exciton coherence” should be read as referring to one particular system and protocol, not to excitons in general.