⚛️ general relativity

Evading the CMB $μ$ -distortion bound on Supermassive Primordial Black Hole seeds with Non-Gaussian tails

This paper demonstrates that the CMB $\mu$ -distortion constraints preventing the formation of supermassive primordial black hole seeds can be evaded if the primordial curvature perturbations exhibit non-Gaussian tails, specifically algebraic or heavy log-normal distributions, rather than the standard Gaussian or light exponential tails.

Original authors: Sanket Dave, Sheng-Feng Yan, Amara Ilyas, Yi-Fu Cai

Published 2026-07-07✓ Author reviewed ⓘ

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Sanket Dave, Sheng-Feng Yan, Amara Ilyas, Yi-Fu Cai

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 by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Big Mystery: Giant Black Holes Too Young to Exist

Imagine you are looking at a photo of a massive, ancient oak tree that was planted only yesterday. It's impossible, right? Trees take centuries to grow.

In our universe, astronomers have found Supermassive Black Holes (SMBHs) that are as heavy as billions of suns. The problem? They existed when the universe was less than a billion years old. If these black holes started as tiny "seeds" (like the remnants of dead stars), they would need to eat matter at an impossible speed to grow that big that fast. It's like trying to fill a swimming pool with a teaspoon in an hour.

To solve this, scientists propose that the seeds started out huge to begin with—Primordial Black Holes (PBHs) formed right after the Big Bang, weighing millions of suns. But there's a catch.

The "Gaussian Barrier": The Universe's Speed Limit

The universe has a strict rulebook called the CMB µ-distortion bound. Think of this as a cosmic speed limit or a "noise meter" for the early universe.

The Rule: If you try to create too many giant black holes, you have to shake the fabric of space-time violently. This shaking creates "static noise" (distortions) in the cosmic background radiation.
The Limit: The COBE/FIRAS satellite measured this noise and said, "It's too quiet." The shaking couldn't have been that strong.
The Barrier: For a long time, scientists assumed the universe's fluctuations were Gaussian (like a perfect bell curve). In a bell curve, if you limit the "width" (variance) to satisfy the noise limit, the "tails" (the rare, huge fluctuations needed to make giant black holes) become so thin that they effectively disappear.
- Analogy: Imagine a bell curve is a mountain. The "noise limit" says the mountain can't be very wide. If you make the mountain narrow, the peak is high, but the sides drop off so steeply that you can never find a path to the top. The "Gaussian Barrier" says: "You can't have the noise limit and the giant black holes at the same time."

The Escape Route: Non-Gaussian Tails

The authors of this paper ask: What if the mountain isn't a perfect bell curve?

They propose that the universe's fluctuations might be Non-Gaussian. Instead of a smooth bell curve, the "tails" of the distribution (the rare, extreme events) could be shaped differently.

The Analogy: Imagine the bell curve is a steep cliff. The "Gaussian Barrier" says you can't climb it. But what if the cliff suddenly turns into a gentle, long ramp (a "heavy tail")? Even if the base of the mountain is narrow (satisfying the noise limit), the ramp allows you to reach the very top (creating giant black holes) without breaking the noise rule.

The Four "Tail" Shapes They Tested

The researchers analyzed four different shapes for these "ramps" to see which ones could let giant black holes form without breaking the cosmic noise limit. They used a mathematical tool called the $\delta N$ formalism (think of it as a map that translates how the universe expands into how black holes form).

The "Stretched Exponential" (The Standard Ramp):
- Simple inflation models only generate Gaussian tails. To get these non-Gaussian (stretched exponential) tails, the universe requires 'non-attractor' behavior, such as 'ultra-slow-roll'-like scenarios with non-trivial fractional potential dynamics or multiplicative dynamics.
- Result: It's still too steep. It's like a ramp that is slightly less steep than a cliff, but still too high to climb. It fails.
The "Power-Law" (The Gentle Slope):
- This comes from specific, complex physics where the energy of the universe changes in a "fractional" way.
- Result: This is a very long, gentle slope. It allows enough giant black holes to form while keeping the noise low enough to pass the test. It works.
The "Log-Normal" (The Super Ramp):
- This comes from "multiplicative" physics, where effects stack on top of each other (like compound interest).
- Result: This is the flattest, longest ramp of all. It easily allows giant black holes to form. It works (though the physics behind it is a bit more speculative).
The "Generalized Normal" (The Symmetric Ramp):
- This is a mathematically symmetric shape.
- Result: It works mathematically, but it's physically unlikely because it would create giant "voids" (empty spaces) just as easily as black holes, which we don't see.

The Conclusion

The paper concludes that the "Gaussian Barrier" is not a dead end; it's just a dead end for bell curves.

If the early universe had specific, complex dynamics (like fractional potentials or multiplicative growth) that created heavy tails in the distribution of fluctuations, we can explain the existence of those ancient, giant black holes without breaking the cosmic noise limits set by the CMB.

In short: The universe didn't need to break the rules to make giant black holes; it just needed to use a different shape of "rule" (a heavy tail) that we hadn't fully considered before. This keeps the idea that giant black holes are "primordial" (born at the start of time) alive and well.

Technical Summary: Evading the CMB $\mu$ -distortion bound on Supermassive Primordial Black Hole seeds with Non-Gaussian tails

1. The Problem: The "Gaussian Barrier"

The existence of luminous quasars at redshifts $z \gtrsim 6$ implies the presence of supermassive black holes (SMBHs) with masses $\sim 10^8 - 10^{10} M_\odot$ within the first billion years of the universe. Standard accretion models starting from stellar-mass remnants ( $\sim 100 M_\odot$ ) struggle to explain this rapid growth without invoking super-Eddington accretion or fine-tuned environments. This motivates the hypothesis that SMBHs formed from "heavy seeds"—Primordial Black Holes (PBHs) with masses in the range $10^5 - 10^7 M_\odot$ .

However, generating such a population requires a localized enhancement of primordial curvature perturbations on small scales ( $k \sim 10^2 - 10^4 \text{ Mpc}^{-1}$ ). These scales are constrained by the Cosmic Microwave Background (CMB) spectral distortion limits, specifically the COBE/FIRAS bound on the chemical potential distortion $\mu < 9 \times 10^{-5}$ .

The Conflict: The $\mu$ -distortion is sensitive to the variance ( $\sigma_\zeta^2$ ) of the curvature perturbations. For Gaussian statistics, the variance uniquely determines the tail of the probability distribution function (PDF). The FIRAS bound limits $\sigma_\zeta^2 \lesssim 10^{-4}$ on seed-relevant scales. Under Gaussian statistics, this small variance suppresses the probability of fluctuations exceeding the collapse threshold ( $\zeta_c \approx 0.67$ ) to negligible levels ( $\beta \sim 10^{-1600}$ ), far below the abundance required for SMBH seeds ( $\beta \sim 10^{-19} - 10^{-13}$ ).
The "Gaussian Barrier": The authors define this obstruction as the "Gaussian barrier," where the variance required to satisfy the distortion bound simultaneously kills the tail probability needed for PBH formation.

2. Methodology

The paper proposes that this barrier can be evaded if the statistics of the curvature perturbations are non-Gaussian, thereby decoupling the variance (controlled by the distortion bound) from the deep tail (controlled by the collapse probability).

Non-Perturbative Approach: The authors argue that perturbative expansions in $f_{NL}$ and higher cumulants are insufficient because the collapse threshold lies $50-100$ standard deviations from the mean. Instead, they utilize the non-perturbative $\delta N$ formalism, where the curvature perturbation $\zeta$ is a map of the initial Gaussian field fluctuations: $\zeta = \delta N(\delta \phi)$ . The asymptotic shape of the PDF tail is determined by the global shape of this $\delta N$ map.
Standardization Procedure:
1. Variance Cap: The FIRAS bound is applied as a hard cap on the total variance: $\sigma_\zeta^2 \leq \sigma_{\zeta, \text{max}}^2(M)$ .
2. Standardization: The PDFs are standardized to unit variance ( $\sigma_u = 1$ ) and zero mean. This isolates the shape of the tail from the amplitude.
3. Abundance Calculation: The PBH mass fraction at formation, $\beta$ , is calculated by integrating the standardized PDF from the standardized threshold $u_c = \zeta_c / \sigma_{\zeta, \text{max}}$ to infinity.
Four Tail Families: The authors analyze four families of Gaussian-cored distributions with varying asymptotic tails:
1. Generalized Normal (Family A): Symmetric tails with index $p$ .
2. Stretched-Exponential (Family B): Asymmetric tails arising from single-field non-attractor dynamics (e.g., ultra-slow-roll).
3. Power-Law (Family C): Algebraic tails arising from fractional-potential dynamics.
4. Log-Normal (Family D): Heavy tails arising from multiplicative dynamics (treated phenomenologically).

3. Key Results

The study quantifies the maximum possible PBH abundance ( $\beta_{\text{max}}$ ) for each family under the FIRAS constraint.

Failure of Standard Exponential Tails: The standard exponential tail ( $p=1$ ), which is the generic output of single-field non-attractor dynamics (e.g., ultra-slow-roll or step potentials), is found to be too light. Even with the variance saturated to the FIRAS limit, the abundance remains $\beta \lesssim 10^{-30}$ , insufficient for seeding.
Success of "Heavy" Tails ( $D_\infty = 0$ ): The seed window reopens only for distributions where the asymptotic logarithmic slope $D_\infty = 0$ $D_{\infty} = 0$ (sub-exponential decay).
- Power-Law Tails (Family C): Derived from fractional potentials ( $V \sim |\phi - \bar{\phi}|^m$ with $2 < m < 2.5$ ), these algebraic tails ( $P \propto \zeta^{-(q+1)}$ ) yield $\beta \sim 10^{-13} - 10^{-15}$ across the seed mass window, satisfying the requirements for SMBH seeds.
- Log-Normal Tails (Family D): Treated as a phenomenological proxy for multiplicative dynamics (e.g., multifield scenarios), these provide the heaviest tails, easily exceeding the seed requirements for shape parameters $s \gtrsim 0.5$ .
Parameter Space: The viable regions are mapped in the $(p, x_\times)$ and $(q, x_\times)$ planes (where $x_\times$ is the transition point from the Gaussian core to the tail). The results show that only specific, non-generic dynamical regimes (fractional potentials or multiplicative accumulation) can satisfy both the distortion bound and the seed abundance requirements.

4. Significance and Claims

The paper claims that the "Gaussian barrier" is not a fundamental prohibition on PBH seeds but a consequence of assuming Gaussian or mildly non-Gaussian statistics.

Decoupling Variance and Tail: The primary theoretical contribution is demonstrating that the $\mu$ -distortion bound (a second-moment observable) and PBH formation (a deep-tail observable) can be decoupled via non-Gaussian statistics.
Dynamical Constraints: The work narrows the class of viable inflationary dynamics capable of producing SMBH seeds. It explicitly rules out the generic single-field non-attractor scenarios (which produce exponential tails) while identifying fractional-potential and multiplicative dynamics as the necessary ingredients.
Methodological Shift: The authors advocate for studying PBH constraints using the full non-perturbative one-point PDF rather than relying solely on the power spectrum or perturbative non-Gaussianity parameters ( $f_{NL}$ ), as the latter fail to capture the physics of the deep tail.

5. Limitations and Caveats

The authors maintain a modest stance regarding their findings:

Phenomenological Nature: The log-normal tail (Family D) is treated as a phenomenological proxy for multiplicative dynamics rather than a derived prediction from a specific single-field model.
Approximations: The analysis relies on the narrow-feature approximation for the $\mu$ -distortion integral and the Press-Schechter formalism with a fixed threshold. The authors note that while these approximations affect absolute abundance values, they do not alter the relative ordering of the tail families (i.e., heavy tails remain superior to light tails).
Model Specificity: The paper does not claim that any specific published model currently achieves these tails at the required amplitude; rather, it identifies the type of dynamics required.

In conclusion, the paper establishes that a primordial origin for high-redshift SMBHs remains viable, provided the early universe dynamics generate sufficiently heavy, sub-exponential tails in the curvature perturbation distribution, thereby evading the CMB $\mu$ -distortion bound that would otherwise exclude Gaussian seeds.