Eppur non si trovano Vol. 2: No Planetary-mass Primordial Black Holes toward the Andromeda Galaxy

This paper refutes a recent claim that twelve short-timescale microlensing candidates in the Andromeda Galaxy indicate a population of planetary-mass primordial black holes, demonstrating instead that all candidates are variable stars and providing no evidence for such black holes as dark matter.

The Gist

The Big Idea: A Cosmic "False Alarm"

Imagine astronomers are playing a high-stakes game of "Where's Waldo?" across the universe. They are looking for something invisible and tiny: Primordial Black Holes (PBHs). These are hypothetical black holes formed right after the Big Bang, and some scientists thought they might be the "missing mass" (Dark Matter) that holds galaxies together.

Recently, a team led by Sugiyama looked at the Andromeda Galaxy (our neighbor, M31) using a super-powerful camera on the Subaru telescope. They claimed to find 12 "blips" in the light of stars. They said these blips were caused by tiny, planet-sized black holes zooming past stars and bending their light (a phenomenon called gravitational microlensing).

If they were right, it would mean the universe is filled with these invisible black holes, solving the mystery of Dark Matter.

However, Przemek Mróz and Andrzej Udalski (the authors of this paper) decided to double-check the work. They looked at the exact same photos with a different, more rigorous set of tools. Their conclusion? It was a false alarm.

The Investigation: Sorting the Noise from the Signal

To understand what happened, let's use a few analogies:

1. The "Sawtooth" vs. The "Bell Curve"

When a black hole passes in front of a star, the star's light should brighten and then fade in a perfectly symmetrical way, like a smooth bell curve (a hill that goes up and down at the same speed).

• What Sugiyama saw: They saw 12 events that looked like hills.
• What Mróz and Udalski found: When they looked closer, they realized the "hills" were actually sawteeth. The light shot up very fast and then trickled down slowly. This is the signature of a pulsating star (specifically an RR Lyrae star), not a black hole.
  • Analogy: It's like mistaking a heartbeat (fast spike, slow recovery) for a smooth wave. The shape didn't match the physics of a black hole.

2. The "Ghost in the Machine"

The original team tried to filter out "variable stars" (stars that naturally change brightness) by checking if the stars flickered on different nights. They thought, "If it's a black hole, it happens once and is gone. If it's a variable star, it will flicker again later."

• The Flaw: The original team was too quick to dismiss the "flickering" on other nights. They assumed the flickering was just a glitch in their camera software (like a bad photo subtraction).
• The Reality: Mróz and Udalski showed that the stars were flickering on multiple nights, and it wasn't a glitch. It was the stars themselves breathing and pulsing.
  • Analogy: Imagine you hear a noise in your house. You think it's a burglar (the black hole). You check the next day, and the noise is gone. You decide it was just the wind. But then you check again, and the noise is back. You realize it wasn't a burglar; it was your cat knocking things over (the variable star). The original team ignored the cat's second visit.

3. The "Wrong Neighborhood"

There was another clue that something was off.

• Where Black Holes should be: If these black holes were dark matter, they should be everywhere, but mostly concentrated in the dense "disk" of the galaxy, where there are lots of stars to act as background lights.
• Where the "Blips" were: All 12 candidates were found in the sparse "halo" (the outer edges) of the galaxy, far away from the main disk.
• The Explanation: This makes perfect sense if they are just old, faint stars (RR Lyrae) that happen to live in the halo. It makes no sense if they are black holes.
  • Analogy: If you are looking for a specific type of fish that only lives in a coral reef, but you find 12 of them in the open ocean, you probably aren't looking for that fish. You're looking for something else entirely.

The Verdict

Mróz and Udalski re-analyzed the data and found:

• 10 of the "black holes" were actually RR Lyrae stars (pulsating stars).
• 1 was an eclipsing binary (two stars orbiting each other and blocking light).
• 1 was an unclassified variable star.

None of them were black holes.

Why This Matters

1. Dark Matter is Still a Mystery: The idea that the universe is filled with tiny, planet-sized black holes is not supported by this data. The "missing mass" of the universe remains unsolved.
2. The "Needle in a Haystack" Problem: This paper highlights how hard it is to find these rare events. When you use giant telescopes to look at millions of stars, you see so much activity (stars pulsing, stars orbiting) that it's easy to get confused.
3. A Lesson for Future Science: The authors warn that just taking pictures very quickly (high cadence) isn't enough. You need to watch for a long time to tell the difference between a one-time event (a black hole) and a repeating pattern (a variable star).

The Bottom Line

The Sugiyama team thought they found a treasure chest of dark matter. Mróz and Udalski opened the chest and found it was full of ordinary, albeit interesting, stars. While it's a disappointment for the black hole hunters, it's a victory for scientific rigor: we didn't let a false alarm fool us.

Technical Summary

1. Problem Statement

The paper addresses a recent claim by Sugiyama et al. (2026), who reported the discovery of twelve short-timescale (less than one day) gravitational microlensing event candidates in the Andromeda Galaxy (M31) using Subaru Hyper Suprime-Cam (HSC) data. Sugiyama et al. attributed these events to a large population of planetary-mass Primordial Black Holes (PBHs) ($10^{-7}$ to $10^{-6} M_{\odot}$), suggesting they could constitute 25–80% of the dark matter in the Milky Way and M31 halos.

This claim creates a significant tension with previous high-cadence surveys (specifically the OGLE survey toward the Magellanic Clouds), which found no such events and placed strict upper limits on the abundance of planetary-mass PBHs (at most 1% of dark matter). Furthermore, Sugiyama et al.'s results contradict their own earlier work (Niikura et al., 2019b) and exhibit anomalous temporal and spatial distributions inconsistent with theoretical microlensing expectations.

2. Methodology

Mróz and Udalski performed an independent reanalysis of the exact same Subaru HSC dataset used by Sugiyama et al. to verify the nature of the twelve candidates.

• Data Source: Subaru HSC observations of M31 from ten nights across 2014, 2017, and 2020. The authors utilized raw data and calibration frames from the SMOKA archive.
• Photometric Pipeline: Instead of the pipeline used by Sugiyama et al., the authors employed the OGLE Difference Image Analysis (DIA) pipeline (based on Woźniak 2000), which is well-tested for dense stellar fields.
  • Preprocessing: Standard bias, dark, and flat-field corrections were applied.
  • Reference Image Construction: Created from 15 high-quality images with the best seeing (notably from 2014-11-24).
  • Coordinate Alignment: Cross-matched with Gaia DR3 to derive accurate quadratic polynomial transformations, correcting for approximate WCS in raw FITS files.
  • Photometry: Used PSF photometry (DOPHOT) with a 3-pixel fitting radius. Fluxes were converted to magnitudes and calibrated against Pan-STARRS1 DR2.
• Quality Control: Images with seeing $> 1.3''$ or those taken through thick clouds were discarded, reducing the sample to 1,090 images.
• Analysis: The authors generated full light curves for all twelve candidates across all available nights, rather than relying on the partial light curves (single-night) presented by Sugiyama et al.

3. Key Contributions

• Independent Verification: Provided a rigorous, independent photometric reduction of the Subaru M31 data using a proven pipeline (OGLE DIA) to test the validity of the PBH claim.
• Full Light Curve Reconstruction: Revealed the multi-night variability of the candidates, which was obscured or ignored in the original study.
• Classification of Candidates: Systematically identified the true astrophysical nature of all twelve "microlensing" candidates.
• Statistical and Spatial Critique: Highlighted inconsistencies in the temporal clustering and spatial distribution of the candidates that contradict microlensing theory.

4. Results

The reanalysis yielded the following definitive results:

• No Microlensing Events: None of the twelve candidates exhibited the symmetric light curve profile characteristic of single-lens gravitational microlensing.
• Reclassification:
  • 10 objects were identified as RR Lyrae stars (pulsating variable stars). Their light curves showed the characteristic asymmetric "saw-tooth" shape (rapid rise, slow decline) and periodicity consistent with RR Lyrae periods (0.2–1 day).
  • 1 object (Candidate #8) was identified as an eclipsing binary.
  • 1 object (Candidate #12) was classified as an unclassified variable star (possibly a Cepheid).
• Temporal Anomalies: The original study claimed events occurred on only two nights (2014-11-24 and 2017-09-20) despite 25.8 hours of observation in 2020. The authors calculated that if the event rate were real, ~23 events should have been detected in 2020. The probability of detecting zero events is $p \approx 1.1 \times 10^{-10}$.
• Spatial Anomalies: All twelve candidates were located in the M31 halo, whereas genuine microlensing events toward M31 should be concentrated in the disk due to higher optical depth and source star density. The halo location is consistent with the distribution of RR Lyrae stars.
• Asymmetry: The light curves were highly asymmetric, a feature Sugiyama et al.'s pipeline failed to reject as a non-microlensing signal.

5. Significance

• Refutation of PBH Dark Matter: The study provides no evidence for a significant population of planetary-mass PBHs as dark matter components. The Sugiyama et al. results are reinterpreted as upper limits on PBH abundance, which are consistent with the stringent limits set by the OGLE survey ($<0.2\%$ for $10^{-7}$ to $10^{-6} M_{\odot}$).
• Methodological Warning: The paper serves as a critical cautionary tale for high-cadence microlensing surveys (e.g., DECam, Rubin, Roman). It demonstrates that high cadence alone is insufficient to distinguish microlensing from variable stars.
  • Variable Star Rejection: Robust rejection of variable stars (especially RR Lyrae) requires long-baseline observations to detect periodicity and multi-night monitoring to identify non-symmetric variability.
  • Pipeline Rigor: Subjective manual cuts and insufficient asymmetry checks can lead to false positives that mimic rare physical phenomena.
• Future Outlook: The authors plan to reanalyze the entire Subaru M31 dataset using their pipeline to establish robust, contamination-free limits on the abundance of compact dark matter objects.

In conclusion, the "discovery" of planetary-mass PBHs by Sugiyama et al. is an artifact of misidentifying variable stars (RR Lyrae) as microlensing events, driven by insufficient data coverage and inadequate variable-star rejection criteria.