A quantum dot that emits single photons on electrical command, with no pump laser anywhere in the setup, is already a useful device. A team led by researchers at the University at Buffalo SUNY reports that such a device does something stranger: the spin of the electrons injected into the dots appears to govern not only the handedness of the emitted light but also the fine-grained timing statistics of two-photon events within a single electrical pulse. When the team erased the magnetization of the spin-injection layer, the fluctuations in two-photon statistics disappeared. The result is a preprint, not yet peer-reviewed, and the microscopic mechanism the authors propose is explicitly phenomenological — two caveats that shape how far the claim should be carried. The paper is “Probing fluctuations in photon emission statistics of spin injected quantum dots with time dependent electronic excitation,” by Chiran Wijesundara, Vincent Iglesias-Cardinale, Yuan Lu, Herbert Fotso, Bo Xu, and Tim Thomay, submitted to arXiv on October 29, 2026, as arXiv:2610.40593.
The headline result is easy to state and hard to interpret, and the two should be kept together. Electrically driven ensembles of InGaAs/GaAs quantum dots, injected with spin-polarized electrons through a ferromagnetic layer, emit circularly polarized single photons at zero applied magnetic field. That part is an extension of established spin-LED physics. The new part is in the higher-order statistics: while single-photon events track the electrical excitation pulse smoothly, two-photon events show pronounced fluctuations at several points within the pulse, and those fluctuations vanish when the spin-injection layer is demagnetized. The authors read this as evidence that spin injection controls higher-order temporal correlations of the emission, not merely its polarization. Whether that reading survives peer review will depend on whether the demagnetized control holds up as a clean comparison, and on whether the cooperative-emission model the authors offer is more than a fitting exercise.
The device: a spin-LED, driven electrically
The device is a spin light-emitting diode built from two parts: a semiconductor LED and a metallic spin injector. The LED is a p-i-n structure grown by molecular beam epitaxy, containing a single layer of In0.3Ga0.7As quantum dots embedded in the active region. The dots have an areal density of 1.6 × 10^10 per square centimeter, an average lateral diameter of about 30 nanometers, and a height of about 9 nanometers, and the structure is delta-doped so that each dot holds approximately one hole — the condition that makes the emitting species a positively charged trion, denoted X+.
The spin injector, deposited after the LED surface was passivated and transferred under vacuum, then annealed at 300 degrees Celsius for three minutes to establish perpendicular magnetic anisotropy, is a stack of 2.5 nanometers of MgO, 1.1 nanometers of Co0.4Fe0.4B0.2, and 5 nanometers of tantalum. The MgO serves as a tunnel barrier; the CoFeB layer, magnetized perpendicular to the plane, supplies spin-polarized electrons. Circular mesas 300 micrometers in diameter were defined by photolithography. Because the CoFeB layer’s perpendicular magnetic anisotropy persists at zero external field, spin-polarized injection — and therefore circularly polarized emission — occurs with no applied magnet, removing a piece of laboratory apparatus that spin-photon experiments have historically required.
The physics linking spin to photon handedness runs through the X+ trion. The trion’s ground state forms a spin singlet, so the only contribution to the average spin comes from the injected electron; combined with the optical selection rules, each two-level emitter then produces a circularly polarized photon whose helicity follows the injected spin. Polarization mapping across the sample shows a spatially uniform circular polarization degree, averaging about 21 percent, and reversing sign when the magnetization is inverted — evidence, the authors argue, that spin injection is not confined to a small subset of dots. Twenty-one percent is modest as a polarization figure, and it matters for the interpretation below: only a fraction of the ensemble’s emitters are spin-aligned at any moment, which is why the authors estimate that fewer than ten emitters are likely to participate in any cooperative effect at a given time.
The device operates in a cryostat at sub-100-kelvin temperatures — a limitation worth stating plainly. This is not a room-temperature source. Electrical excitation comes from an arbitrary waveform generator producing tunable voltage pulses of about 20 nanoseconds’ duration, with rise and fall times the team varied across measurements (5 and 1, 1 and 1, and 1 and 5 nanoseconds). That pulse-shape control is precisely what optical pumping cannot offer, and it is central to the experiment: the team is asking how the photon statistics respond as the electrical drive is reshaped.
Counting photons by number, inside one pulse
The measurement apparatus is an extended Hanbury Brown–Twiss interferometer: three fiber beam splitters feeding four avalanche photodiodes, with time-tagged readout that records every photon’s arrival time relative to the electrical trigger. Four detectors are the minimum needed to resolve up to three-photon states, and the team uses them to resolve one- and two-photon Fock states within a single excitation cycle. The detectors are ID Quantique ID100 units with 40-picosecond timing jitter; the PicoQuant MultiHarp 150 time tagger adds 20 picoseconds; coincidence histograms are binned at 10 picoseconds with a 500-picosecond coincidence window defining a two-photon event. Each dataset represents eight hours of integration — a reminder of how rare two-photon events are in a source designed to suppress them. The polarized sample yielded 3.8 × 10^8 single-photon events and 7 × 10^4 two-photon events; the demagnetized sample yielded 5.1 × 10^8 and 3 × 10^5.
The standard summary statistic, g(2)(0), averages over the entire pulse, and the authors argue it is the wrong tool here — particularly when the excitation pulse outlasts the emitter lifetime of 800 picoseconds. An ensemble driven by a 20-nanosecond pulse behaves, in the averaged statistic, much like a coherent source, even though individual emitters antibunch. The time-resolved, photon-number-resolved measurement is the paper’s methodological contribution: instead of one number per pulse, it produces curves of one-photon and two-photon event counts as a function of position within the pulse. The single-photon curve follows the pulse shape, as expected. The two-photon curve does not: it fluctuates, with pronounced deviations from the smooth single-photon profile at multiple temporal positions within the pulse, reproducible across the three pulse shapes tested.
This is where the demagnetization control does its work, and it is the experiment’s strongest card. To demagnetize the sample, the team rotated it at high speed in a strong magnetic field while ramping the field down, breaking the CoFeB film into many small domains with randomly oriented magnetization. Net magnetization averages to zero, net spin injection averages to zero, and the polarization bias of the emitted photons disappears — while everything else about the device and the excitation stays the same. In the demagnetized data, the two-photon fluctuations are gone. Secondary coverage has quantified the reduction at over 60 percent; the paper itself presents the comparison graphically rather than as a single percentage, so that figure should be treated as coverage’s summary, not the authors’ stated number. Either way, the qualitative point is the control’s point: the same device, the same pulses, the same detectors — and the fluctuations track the presence of spin injection, not the electrical drive alone.
A phenomenological model, honestly labeled
The authors do not leave the fluctuations unexplained, but the explanation they offer is labeled phenomenological in the abstract and again in the conclusion, and readers should hold them to that label. They model the ensemble with a Monte Carlo wavefunction simulation: three spin-aligned emitters, a natural lifetime of 800 picoseconds, a time-dependent excitation operator fitted so the simulated single-photon curve matches the experiment, and — the novel ingredient — a set of joint collapse operators that annihilate pairs of emitters simultaneously, producing a two-photon event. A tunable rate parameter, kappa = 0.0016, is chosen to match the experimental ratio of one- to two-photon events. Over 100,000 simulation trials, this scheme reproduces the fluctuating two-photon curves across the tested pulse shapes.
The authors are explicit about the status of these joint operators: they “do not arise from a standard treatment of the Weisskopf-Wigner Hamiltonian nor are they typical of any decay channels known to the authors,” and “the microscopic mechanism that couples the emitters is not identified here.” The joint operators capture the effect phenomenologically. That honesty is to the paper’s credit, and it defines the boundary of the claim. What the experiment establishes, if the control holds, is a correlation: spin injection is necessary for the two-photon fluctuations. What it does not establish is a mechanism: no identified interaction couples the emitters, and no measurement distinguishes cooperative emission from some other spin-conditioned dynamics of the injection process itself. The authors’ suggestion — that spin alignment supplies the shared polarization axis cooperative emission requires — is plausible, and the simulation’s agreement is suggestive, but a fitted phenomenological operator is not a microscopic theory, and the paper does not pretend otherwise.
What this is not
Three overreadings should be avoided. First, this paper does not report photon indistinguishability. There is no Hong–Ou–Mandel measurement in it, and no indistinguishability figure to quote; the paper’s concern is photon-number statistics within a pulse, not the interference of photons from different emissions. Any summary claiming this work advances indistinguishability is inventing a result. For photonic quantum computing, where interference between photons is the resource, indistinguishability remains a separate measurement this device has not yet passed.
Second, this is an ensemble measurement, not a single-emitter demonstration. The dots are many — about 11 per square emission wavelength at 904 nanometers — and the two-photon events are ensemble events. The authors’ three-emitter toy model is a deliberate simplification, justified by the polarization fraction, not a claim that three dots were isolated. Scaling this toward a deterministic two-photon Fock-state source would require isolating and controlling a small, known number of emitters, which is a different device.
Third, the “electrically driven, microelectronics-compatible” framing is real but bounded. Electrical driving removes the pump laser and allows pulse-shape control that optics cannot match, and the spin-LED lineage here is mature — the same groups have published on zero-field spin injection and electrical helicity control, including a 2024 Nature paper on controlling light helicity by electrical magnetization switching. But the device still requires cryogenic cooling below 100 kelvin, the polarization degree is about 21 percent rather than near unity, and two-photon events are rare enough to need eight-hour acquisitions. Compatibility with microelectronics is a direction, not a delivered system.
Why it matters anyway
With those boundaries stated, the result is genuinely interesting for two reasons. The first is diagnostic. Electrically driven quantum-dot sources are attractive precisely because they drop the laser, but electrical injection is stochastic in a way optical excitation is not: a bulk current couples to individual emitters through resistive and capacitive elements whose details are hard to predict, and the paper notes frankly that the precise interaction between the bulk current and the individual emitters is not known. A measurement that resolves photon-number statistics inside the pulse turns the emitted light into a probe of that internal dynamics — each photon corresponds to a recombined carrier, so the light reports what the current did.
The second reason is the prospect the authors name: electrically driven, spin-controlled single- and multi-photon Fock-state sources. Higher-order Fock states matter for quantum metrology and photonic quantum information, where they are usually produced probabilistically and heralded. A source in which an electrical control — the magnetization state of a layer — tunes multi-photon statistics would be a different kind of tool. The present paper is far from that: it shows fluctuations in a two-photon population, not an on-demand two-photon source, and it shows them in an ensemble at cryogenic temperature. But it identifies a control knob — spin injection — that no one had previously connected to multi-photon timing, and it provides the control experiment that makes the connection hard to dismiss as an artifact of the drive electronics.
What to watch next
The obvious next experiment is the one the authors flag as future work: identifying the microscopic coupling — by varying the dot density to test whether the fluctuations scale with emitter number as a cooperative model predicts, or by isolating few-dot devices to see whether the effect survives when the ensemble is reduced to a known, small number of emitters. A second follow-up is peer review itself: the demagnetization control is the load-bearing result, and reviewers will probe whether rotating the sample in a decaying field changes anything besides net magnetization that could independently alter two-photon statistics.
A third question is whether the effect can be pushed toward utility: higher polarization degree, which would strengthen any cooperative signal; shorter, shaped pulses that isolate the fluctuating features; and eventually a Hong–Ou–Mandel measurement, which would connect this photon-statistics work to the indistinguishability metric that photonic applications actually require. Until that measurement exists, the honest summary of this paper is narrower than its most exciting reading — and also more interesting than a routine device report.
Sources and further reading
- Chiran Wijesundara, Vincent Iglesias-Cardinale, Yuan Lu, Herbert Fotso, Bo Xu, and Tim Thomay, “Probing fluctuations in photon emission statistics of spin injected quantum dots with time dependent electronic excitation,” arXiv:2610.40593 [quant-ph], submitted October 29, 2026. Preprint, not peer-reviewed. DOI: 10.48550/arXiv.2610.40593.
- Background on the device lineage: A. E. Giba et al., “Spin Injection and Relaxation in p-Doped InGaAs/GaAs Quantum-Dot Spin Light-Emitting Diodes at Zero Magnetic Field,” Physical Review Applied 14, 034017 (2020); P. A. Dainone et al., “Controlling the helicity of light by electrical magnetization switching,” Nature 627, 783–788 (2024); Y. Lu et al., “Spin light-emitting diodes,” Nature Reviews Electrical Engineering 3, 316–333 (2026).
- Method background: T. Thomay et al., “Simultaneous, Full Characterization of a Single-Photon State,” Physical Review X 7, 041036 (2017), on the extended Hanbury Brown–Twiss photon-number-resolving technique.
- This desk’s earlier article on single-photon sources from excitons, for the purity, indistinguishability, and brightness metrics used to evaluate quantum light sources.
- Note on verification: all device parameters, event counts, pulse shapes, and simulation details above are taken from the arXiv preprint text. The paper reports no indistinguishability measurement and no single “over 60 percent” figure; the latter appears in secondary coverage and is attributed as such. The cooperative-emission mechanism is described by the authors themselves as phenomenological, and this article preserves that characterization.