🔬 mesoscale physics

Beyond Antibunching: Photon Correlation Analysis Reveals Blinking Origin

This paper establishes a consistent framework for analyzing the second-order auto-correlation function ( $g^{(2)}(0)$ ) of blinking quantum emitters by demonstrating that fitting long-time correlation envelopes with correct blinking models yields the most accurate multi-photon contribution estimates, while normalizing to the Poisson level produces the worst results, thereby enabling reliable benchmarking of single-photon sources.

Original authors: Patricia Kallert, Santiago Bermúdez-Feijóo, Giorgio De Pascalis, Nicolas Claro-Rodriguez, Ioannis Caltzidis, Normen Auler, Eva Berger, Rebecca Aschwanden, Tobias Krieger, Saimon. F. Covre da Silva, Sa

Published 2026-09-24

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Patricia Kallert, Santiago Bermúdez-Feijóo, Giorgio De Pascalis, Nicolas Claro-Rodriguez, Ioannis Caltzidis, Normen Auler, Eva Berger, Rebecca Aschwanden, Tobias Krieger, Saimon. F. Covre da Silva, Sandra Stroj, Quirin Buchinger, Michele B. Rota, Thomas Hummel, Sven Höfling, Armando Rastelli, Rinaldo Trotta, Dirk Reuter, Sonja Barkhofen, Klaus D. Jöns, Tobias Huber-Loyola

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

In the quest to build the next generation of quantum computers and unhackable communication networks, scientists rely on tiny, reliable flashes of light. These flashes, known as single photons, are the fundamental units of information in these future technologies. To work correctly, a light source must emit exactly one particle at a time, never two, and never none. Physicists measure the quality of these sources by looking at how the photons arrive. If they arrive in a perfectly spaced, rhythmic pattern, the source is excellent. If they clump together or arrive randomly, the source is flawed. This measurement is the standard way researchers compare different light emitters, from atoms to tiny semiconductor crystals called quantum dots.

However, a common problem plagues these measurements: the light sources often flicker. Much like a faulty lightbulb that dims and brightens unpredictably, quantum emitters can switch between a bright state where they emit light and a dark state where they are silent. This flickering, known as blinking, distorts the measurement of the light's quality. When a source blinks, the standard way of calculating its performance gives a misleading number, making a good source look bad or a bad source look good. This inconsistency has made it difficult for scientists to compare different technologies fairly or to trust the data when building complex quantum systems.

A team of researchers has now solved this puzzle by developing a new way to analyze the data that accounts for the flickering. They started with a perfect, non-flickering quantum dot and artificially introduced blinking to see how different calculation methods reacted. They created two types of artificial flickering: one by digitally masking the recorded light data after the experiment, and another by physically blocking the laser pulses that trigger the light emission. By comparing these controlled scenarios against the known, perfect baseline, they tested five different mathematical methods that scientists currently use to interpret the data.

The study revealed that most of the commonly used methods fail when blinking is present. The most frequent approach, which simply averages all the light peaks over a long period, significantly overestimates the errors caused by the blinking. Another popular method, which tries to normalize the data against a theoretical ideal, produces results that are even worse, sometimes suggesting the light is clumping together when it is actually behaving perfectly. The researchers found that the only way to get the true, intrinsic quality of the light source is to use a specific model that fits the shape of the flickering pattern itself. By mathematically tracing the curve of the blinking envelope back to the very moment the light is emitted, they can strip away the distortion caused by the dark periods and reveal the true nature of the light.

To prove this method works in the real world, the team applied it to a quantum dot that naturally blinks. This specific emitter, embedded in a circular mirror structure, was known to flicker, but the cause of the flickering was unknown. The researchers analyzed the light using their new technique, testing whether the flickering followed a simple on-off pattern or a more complex pattern involving hidden, dark energy states. The data showed that the complex model, involving these dark states, provided a much better fit to the long-term behavior of the light than the simple on-off model. This allowed them to pinpoint the exact physical mechanism causing the flicker and, more importantly, to extract the true quality of the single-photon emission with high precision.

The findings offer a clear path forward for the field. The researchers demonstrated that while blinking is a nuisance, it does not have to ruin the measurement of a light source's quality. By choosing the right analysis method—one that models the flickering rather than ignoring it—scientists can now compare different quantum emitters fairly, even if they are unstable. This ensures that the benchmarks used to judge the best single-photon sources are accurate, paving the way for more reliable quantum technologies. The work confirms that the best way to see the truth in a flickering light is not to average out the darkness, but to understand the rhythm of the flicker itself.

Technical Summary: Beyond Antibunching: Photon Correlation Analysis Reveals Blinking Origin

Problem Statement
The evaluation of single-photon sources (SPS), particularly semiconductor quantum dots (QDs), relies heavily on the second-order auto-correlation function, $g^{(2)}(0)$ . Values below 0.5 are the standard operational criterion for single-photon emission. However, solid-state emitters frequently exhibit "blinking"—stochastic switching between bright (ON) and dark (OFF) states. This phenomenon modulates the intensity correlations on timescales much longer than the radiative lifetime, causing the side peaks in a pulsed correlation histogram to decay rather than remain flat.

The community currently employs inconsistent methods to normalize these blinking data to extract $g^{(2)}(0)$ . Different normalization choices (e.g., averaging all side peaks, using only nearest neighbors, or normalizing to a far-delay baseline) yield significantly different $g^{(2)}(0)$ values for the same physical source. This leads to incomparable benchmarks of source quality and potential misinterpretation of the intrinsic multi-photon emission probability during the emitter's bright state. Furthermore, in experiments involving remote, blinking QDs, incorrect extraction of $g^{(2)}(0)$ can lead to errors in correcting interference visibility.

Methodology
To systematically address these inconsistencies, the authors employed a controlled experimental approach using an inherently non-blinking InAs quantum dot embedded in a p-i-n diode structure. They generated well-defined, artificial blinking patterns using two distinct mechanisms to simulate real-world blinking conditions:

Post-processing Masking: Applying a temporal mask to raw time-tagged photon streams from the non-blinking source to simulate stochastic ON/OFF switching.
Experimental Gating: Using an arbitrary waveform generator (AWG) to gate the laser excitation pulses via a pulse picker, physically creating dark periods.

Three blinking mechanisms were simulated: Random Telegraph Noise (RTN), Dark States (DS), and Powerlaw (PL) processes. The authors then applied five distinct estimators for $g^{(2)}(0)$ , commonly found in literature, to the resulting correlation histograms:

$g^{(2)}_{all}(0)$ : Normalization using the average of all selected side peaks.
$g^{(2)}_{nn}(0)$ : Normalization using only the two nearest neighboring side peaks.
$g^{(2)}_{far}(0)$ : Normalization using the mean of long-delayed side peaks (Poisson level).
$g^{(2)}_{env}(0)$ : Normalization using a fitted envelope extrapolated to zero delay.
$g^{(2)}_{base}(0)$ : Normalization using the fitted envelope at infinite delay (baseline).

These estimators were compared against the known reference $g^{(2)}(0)$ of the non-blinking source. Finally, the framework was applied to an inherently blinking GaAs QD in a circular Bragg grating cavity to demonstrate practical utility.

Key Results

Estimator Performance: The study reveals that the choice of estimator drastically affects the extracted $g^{(2)}(0)$ $g^{(2)} (0)$ as blinking strength (intermittency ratio $r = P_{off}/P_{on}$ $r = P_{o f f} / P_{o n}$ ) increases.
- $g^{(2)}_{base}(0)$ and $g^{(2)}_{all}(0)$ : These methods, which normalize to the Poisson level or average all peaks, systematically overestimate $g^{(2)}(0)$ . For high blinking ratios ( $r \ge 20$ ), $g^{(2)}_{base}(0)$ can exceed 1, falsely implying bunched light emission despite the source being a single-photon emitter during its ON state.
- $g^{(2)}_{env}(0)$ : This model-based estimator, which normalizes the central peak to the fitted envelope extrapolated to zero delay ( $S_{env}(0)$ ), remains closest to the intrinsic non-blinking reference value across all blinking strengths and mechanisms. It effectively corrects for the vacuum contribution introduced by blinking.
- $g^{(2)}_{nn}(0)$ : The nearest-neighbor estimator performs well as a model-independent approximation, closely tracking $g^{(2)}_{env}(0)$ for RTN and PL processes. However, it shows systematic deviations for very rapid blinking (high $r$ ) or Dark State models, where the nearest peaks are already affected by the blinking envelope.
Mechanism Identification: When applied to the inherently blinking GaAs QD, the study demonstrated that the choice of blinking model (RTN vs. DS) significantly impacts the fit quality. By analyzing the reduced $\chi^2$ of the envelope fits over varying correlation windows, the authors identified that a Dark State model provided a superior description of the long-time decay compared to RTN, whereas the nearest-neighbor estimator provided a robust, model-free reference value ( $g^{(2)}_{nn}(0) = 0.0164$ ).

Significance and Claims
The paper claims to provide a consistent, rigorous framework for analyzing blinking single-photon sources. The primary contribution is the demonstration that normalizing to the Poisson level (the "far" or "all" peak average) is fundamentally flawed for blinking sources, as it conflates the intrinsic multi-photon probability with the statistics of the blinking process itself.

The authors assert that:

Envelope Fitting is Optimal: Using a model-based envelope fit ( $g^{(2)}_{env}(0)$ ) to normalize to the zero-delay reference level is the most accurate method to recover the intrinsic $g^{(2)}(0)$ , as it mathematically corresponds to the vacuum-corrected correlation function defined by Grünwald.
Practical Alternative: When the underlying blinking mechanism is unknown or difficult to model, the nearest-neighbor estimator ( $g^{(2)}_{nn}(0)$ ) serves as the best practical, model-free alternative, provided the blinking is not extremely fast.
Standardization: The work highlights the necessity of reporting the specific normalization method and blinking model used when benchmarking quantum emitters to ensure comparability across different research groups and devices.

The study concludes that while blinking is a common feature of solid-state emitters, its impact on $g^{(2)}(0)$ extraction is often underestimated, and the proposed "recipe" allows for the determination of the most realistic single-photon purity values.

Technical Summary: Beyond Antibunching: Photon Correlation Analysis Reveals Blinking Origin

More like this