Eccentricity constraints disfavor single-single capture in nuclear star clusters as the origin of all LIGO-Virgo-KAGRA binary black holes

By analyzing 85 binary black hole mergers from the O4a observing run, this study finds no statistically significant evidence for orbital eccentricity, thereby ruling out single-single gravitational wave captures in nuclear star clusters as the dominant formation channel for all observed LIGO-Virgo-KAGRA events.

The Gist

The Big Picture: A Cosmic Mystery

Imagine the universe is a giant, crowded dance floor. Occasionally, two massive partners (black holes) grab each other and spin wildly until they crash together, sending out ripples in space-time called gravitational waves.

Scientists have detected hundreds of these crashes. But there is a big mystery: How did these partners meet?

There are two main theories:

1. The "Slow Dance" (Isolated Evolution): Two stars are born next to each other, grow old together, and slowly spiral inward over billions of years. They are very polite and move in perfect circles.
2. The "Mosh Pit" (Dynamical Capture): In crowded, chaotic places (like the centers of galaxies or dense star clusters), black holes bump into strangers. They get grabbed by gravity, swing wildly around each other in crazy, oval-shaped paths, and then crash.

The key difference is eccentricity.

• Circular orbit: Like a car driving on a perfect round track.
• Eccentric orbit: Like a car swerving wildly, taking a sharp turn, and then straightening out before turning again.

The Investigation: Checking the Dashcams

The authors of this paper acted like cosmic detectives. They looked at 85 recent black hole crashes detected by the LIGO-Virgo-KAGRA network (specifically from the "O4a" observing run).

They used a super-advanced computer model (a "waveform model") to check the "dashcam footage" of these crashes. They were looking for the tell-tale signs of that wild, swerving motion (eccentricity) that would prove the "Mosh Pit" theory.

The Findings:

• No Wild Swerving: Even though a few events looked slightly like they might be swerving, none were convincing enough to say, "Yes, this was a wild capture!"
• The Glitch Factor: Some of the data looked messy, like a camera lens getting smudged or a static noise (called a "glitch"). The researchers realized that when they cleaned up this static noise properly, the "wild swerving" often disappeared. It turned out the noise was tricking the computer into thinking the orbit was weird when it was actually just a glitch.
• The Verdict: None of the 85 crashes showed strong evidence of being formed by a chaotic, single-black-hole capture. They all looked like they were moving in nice, calm circles.

The Big Test: The "Speed Limit" of the Dance Floor

Since they didn't find any wild swerving, the authors asked a "What If" question:

"What if ALL of these black holes actually came from the chaotic Mosh Pit (Nuclear Star Clusters), but we just haven't seen the swerving yet?"

To test this, they used a clever trick. They imagined a giant, crowded room (a star cluster) where people (black holes) are running around.

• In a slow room (like a Globular Cluster), people move slowly. If two people bump, they might grab each other gently and start a slow, circular dance.
• In a fast room (like a Nuclear Star Cluster), people are sprinting. If two sprinters bump, they grab each other and spin wildly in a crazy, oval shape before crashing.

The researchers calculated: "If all these black holes came from the Fast Room, how fast would the room have to be for us to NOT see any wild spinning?"

The Result:
They found that for the data to look this calm, the "room" would have to be moving at a speed of less than 24 km/s.

However, we know that Nuclear Star Clusters (the Fast Rooms) usually have speeds between 20 and 200 km/s.

• The Conclusion: The "Fast Room" is too fast! If all these black holes came from there, we would have seen wild spinning. Since we didn't, it is highly unlikely that all of them came from Nuclear Star Clusters via single captures.

The Takeaway

1. No "Smoking Gun" yet: We haven't found a black hole crash that definitely proves the "chaotic capture" theory, though we are getting better at looking.
2. The "Mosh Pit" isn't the only source: The idea that every single black hole merger we see comes from a chaotic capture in a dense nuclear cluster is probably wrong. The data suggests that if they did come from there, the clusters would have to be surprisingly calm, which doesn't match what we know about them.
3. Clean Data Matters: The study showed that being careful about "static noise" (glitches) in the data is crucial. If you don't clean the noise, you might think you see a wild dance when it's just a camera glitch.

In short: The universe's black holes seem to be mostly dancing in calm circles, not the wild, chaotic mosh pits we hoped to find in these specific events. This helps scientists narrow down where these cosmic couples are actually meeting.

Technical Summary

1. Problem Statement

The origin of Binary Black Hole (BBH) mergers detected by the LIGO-Virgo-KAGRA (LVK) collaboration remains uncertain. Two primary formation channels are proposed:

1. Isolated Binary Evolution: Stars evolve in isolation, radiating away orbital eccentricity efficiently. By the time they enter the detector band, their eccentricity is negligible ($e \sim 10^{-11}$).
2. Dynamical Assembly: Occurs in dense stellar environments (Globular Clusters or Nuclear Star Clusters). Interactions can produce BBHs with significant residual eccentricity ($e \gtrsim 10^{-4}$) at merger.

While previous LVK studies have used mass and spin distributions to constrain these channels, orbital eccentricity is considered a "smoking gun" for dynamical formation. Specifically, single-single gravitational wave (GW) captures (where two unbound black holes pass close enough to become bound via GW emission) are analytically tractable and produce the highest eccentricities. The paper aims to determine if the observed BBH population in the first part of the fourth observing run (O4a) is consistent with a scenario where all mergers originate from single-single captures in dense clusters, particularly Nuclear Star Clusters (NSCs) which have high velocity dispersions.

2. Methodology

Data and Waveform Models

• Dataset: The authors analyzed 85 confidently detected BBH signals from the O4a observing run.
• Waveform Models: They utilized the SEOBNRv5 family of Effective-One-Body (EOB) models:
  • SEOBNRv5EHM: An eccentric, aligned-spin model used for the primary analysis. It offers improved accuracy over previous generations (SEOBNRv4EHM) and includes corrections up to 3rd Post-Newtonian (PN) order.
  • SEOBNRv5PHM: A quasicircular, precessing-spin model used as a competing hypothesis.
  • SEOBNRv5HM: A quasicircular, aligned-spin model.
• Reference Frequency: Eccentricity ($e_{10\text{Hz}}$) and relativistic anomaly ($\zeta_{10\text{Hz}}$) are defined at an orbit-averaged detector-frame frequency of 10 Hz.

Parameter Inference Techniques

• Neural Posterior Estimation (NPE): The primary tool is DINGO, a machine-learning-based inference code. It uses embedding networks and normalizing flows to generate proposal posteriors in minutes, followed by importance sampling to obtain accurate Bayesian evidences.
• Nested Sampling: For events where DINGO struggled (e.g., due to glitches or low masses), the authors used Bilby with the dynesty sampler.
• Glitch Mitigation: To address non-Gaussian noise ("glitches"), the authors employed glitch marginalization. Instead of subtracting a single glitch realization, they jointly inferred signal and glitch properties (using parameterized scattering or wavelet sums) and marginalized over 100 glitch realizations. This prevents bias in eccentricity estimation caused by overfitting glitches as signal features.

Statistical Framework

• Bayes Factors: They computed the Bayes factor ($B_{EAS/QCAS}$) comparing the Eccentric Aligned-Spin (EAS) model against the Quasicircular Aligned-Spin (QCAS) model.
• Odds Ratios: To account for astrophysical priors, they calculated odds ratios ($O_{EAS/QCAS}$) by multiplying the Bayes factor by the prior odds ($R_{EAS}/R_{QCAS} \approx 0.023$), reflecting the expectation that only ~2% of BBHs are measurably eccentric.
• Hierarchical Inference: To test the "single-single capture" hypothesis, they performed a population-level analysis. They constructed a forward model $p(e|\sigma, M, \mu)$ predicting eccentricity distributions based on the host cluster's velocity dispersion ($\sigma$) and binary masses. They then inferred the posterior distribution of $\sigma$ given the O4a data.

3. Key Contributions

1. Comprehensive O4a Analysis: This is one of the first large-scale analyses of O4a data specifically targeting eccentricity using state-of-the-art, high-accuracy waveform models (SEOBNRv5EHM).
2. Robust Glitch Handling: The paper demonstrates the critical importance of glitch marginalization over simple subtraction. They show that for events like GW190701 and GW231114 043211, subtracting a glitch can artificially inflate eccentricity support, whereas marginalization correctly reduces the significance.
3. Population-Level Constraints: Rather than just looking for individual eccentric events, the authors derived a constraint on the velocity dispersion ($\sigma$) of the host environments. This allows them to rule out specific astrophysical scenarios (like NSCs) even if no single event is confidently eccentric.
4. Synthetic Validation: They verified their constraints by injecting synthetic eccentric events (motivated by O3 analyses) into the catalog, confirming that their upper bound on $\sigma$ is driven by the lack of eccentricity in the majority of events, not just the absence of high-significance detections.

4. Results

Individual Event Analysis

• No Confident Detections: None of the 85 events reached a high enough statistical significance to claim a confident detection of eccentricity.
• Odds Ratios: While 9 events had positive Bayes factors ($\log_{10} B_{EAS/QCAS} > 0.2$), none had an odds ratio ($\log_{10} O_{EAS/QCAS}$) exceeding 0 after incorporating astrophysical priors.
• Impact of Glitches:
  • GW190701: The Bayes factor dropped from $\log_{10} B = 3.68$ (glitch subtraction) to $0.42$ (glitch marginalization), eliminating the claim of eccentricity.
  • GW231114 043211: The Bayes factor dropped from $0.36$ to $0.06$ after marginalization.
  • This highlights that previous claims of eccentricity in O3 events may have been biased by insufficient glitch modeling.

Population-Level Constraints

• Velocity Dispersion ($\sigma$): Assuming all O4a BBHs originate from single-single captures, the authors inferred the velocity dispersion of the host clusters.
  • Result: $\sigma < 24.3$ km/s (95% credible upper bound), with a median of $9.6$ km/s.
• Implication for Formation Channels:
  • Globular Clusters (GCs): Typical $\sigma \approx 1–20$ km/s. The results are consistent with GCs being a source.
  • Nuclear Star Clusters (NSCs): Typical $\sigma \approx 20–200$ km/s. The results strongly disfavor NSCs as the dominant source of all observed BBH mergers via single-single capture.
• Hyper-model Analysis: When modeling the population as a mixture of GCs and NSCs, the fraction of events from NSC-like environments ($\beta$) was found to be $0.13^{+0.37}_{-0.12}$ (90% CI), consistent with a GC-dominated population.

5. Significance

• Ruling Out Extreme Scenarios: The study provides strong evidence that the single-single capture mechanism in high-velocity-dispersion environments (NSCs) cannot be the sole origin of the observed BBH population.
• Methodological Rigor: The work underscores the necessity of advanced glitch mitigation (marginalization) in eccentricity searches. It suggests that previous detections of eccentricity in O3 may have been artifacts of noise modeling.
• Future Directions: The results set improved upper limits on the number of eccentric BBHs. As detector sensitivity improves, the ability to distinguish between formation channels via eccentricity will become a primary tool for astrophysical inference. The authors note that future work will incorporate mass and spin population models and selection effects to refine these constraints further.

In summary, the paper concludes that while eccentricity remains a powerful probe for dynamical formation, the current O4a data shows no evidence of high eccentricity, effectively ruling out high-velocity-dispersion Nuclear Star Clusters as the exclusive birthplace of all observed binary black hole mergers.