Quasar Mass Function Overview

• Quasar mass function is a measure of the space density of actively accreting SMBHs per logarithmic mass bin at various redshifts, defining black hole growth history.
• Modern observational methods use virial mass estimators and forward modeling to address flux-limited surveys and correct selection biases.
• Empirical findings show a steep high-mass slope with a redshift-dependent break mass, highlighting challenges in completeness and uncertainty at high redshift.

A quasar mass function quantifies the space density of actively accreting supermassive black holes (SMBHs) as a function of black hole mass and redshift. It is fundamental for understanding the growth history of black holes, the fueling of quasars, and their link to galaxy and halo assembly across cosmic time. Modern measurements of the quasar mass function are built on the synthesis of large redshift surveys, forward modeling frameworks, and empirical or semi-analytic accretion prescriptions. This article reviews the observational methodology, analytic parameterizations, physical interpretation, and current constraints on the quasar mass function, with particular emphasis on results from flux-limited optical surveys and their extensions to high redshift.

1. Formal Definition and Analytic Forms

The differential quasar mass function, denoted $\Phi(M_{\mathrm{BH}}, z)$, is the comoving number density of actively accreting SMBHs per unit logarithmic mass at a given redshift $z$: $$\Phi(M_{\mathrm{BH}},z) \equiv \frac{d^2N}{dV\,d\log M_{\mathrm{BH}}}$$ with units of $\mathrm{Mpc}^{-3}\,\mathrm{dex}^{-1}$ (Vestergaard, 2010). Analytically, it is most commonly parameterized as a broken (double) power law: $$\Phi(M_{\mathrm{BH}},z) = \frac{\Phi^*(z)}{(M_{\mathrm{BH}}/M^*(z))^{\alpha(z)} + (M_{\mathrm{BH}}/M^*(z))^{\beta(z)}}$$ where $\Phi^*$ is the normalization, $M^*$ the "break" mass, $\alpha$ the faint-end slope and $\beta$ the bright-end slope (Vestergaard, 2010, Wu et al., 2022, Lai et al., 2024, Zhang et al., 2023).

Empirical fits find that, at $z \lesssim 3.5$, the high-mass slope is steep ($\beta \approx -3.3$), flattening to $\beta \approx -1.8$ at $z \sim 4$ (Vestergaard, 2010). Many Bayesian inference studies, notably those using SDSS broad-line quasars, represent the mass function as a mixture of lognormals or Gaussians in $\log M_{\mathrm{BH}}$ for greater flexibility in modeling measurement scatter, selection incompleteness, or redshift evolution (Kelly et al., 2012, Kelly et al., 2010, Shen et al., 2011).

2. Observational Estimation: Methods and Selection Effects

Observational determination of the quasar mass function combines high-completeness optical/IR spectroscopic samples with virial black hole mass estimators and rigorous forward modeling frameworks.

Virial Mass Estimates

Black hole masses are inferred from single-epoch spectra via scaling relations of the form

$$M_{\rm BH} = 10^{A} \left(\frac{L_{\text{cont}}}{10^{44}~\text{erg}~\text{s}^{-1}}\right)^B \left(\frac{\mathrm{FWHM}}{1000~\mathrm{km}~\mathrm{s}^{-1}}\right)^2~M_\odot$$

using broad emission lines (H$\beta$, Mg II, or C IV), with empirical calibrations anchored by reverberation mapping (e.g., Mg II: $A=6.86$, $B=0.50$ (Willott et al., 2010, Vestergaard, 2010, Lai et al., 2024)). Systematic uncertainties in these mass estimators are typically 0.3–0.5 dex.

Forward Modeling, Completeness, and Statistical Treatment

Flux-limited surveys miss low-luminosity (and hence low-mass or low-Eddington ratio) quasars, resulting in strong and mass-dependent incompleteness. Modern studies correct for this using forward-modeling and Bayesian mixture techniques, explicitly modeling the joint intrinsic distribution in $\{\log M_{\mathrm{BH}}, \log L, z\}$, survey selection functions, measurement uncertainties, and possible biases in the mass estimators:

• Intrinsic models are convolved with observational errors and weighted by completeness in redshift-magnitude space before comparison with the observed distribution (Kelly et al., 2012, Kelly et al., 2010, Wu et al., 2022).
• Maximum likelihood or MCMC approaches yield the best-fit mass function and its uncertainties, as well as marginalized constraints on systematic effects such as luminosity-dependent mass biases, scatter in virial estimates, and Eddington-ratio distributions (Shen et al., 2011, Wu et al., 2022).

At high redshift ($z\gtrsim4$), the difficulty of assembling large, unbiased samples with robust black hole masses renders the mass function highly uncertain at $M_{\mathrm{BH}} \lesssim 10^{9}\,M_\odot$ due to severe flux incompleteness (Wu et al., 2022, Lai et al., 2024).

3. Parametric Results: Shape, Redshift Evolution, and Mass Scales

Comprehensive surveys suggest the following empirical trends in the quasar mass function:

• High-mass slope: Steep ($\beta\sim-3$) for $z\lesssim3.5$; flattens at higher $z$ (Vestergaard, 2010, Wu et al., 2022, Lai et al., 2024).
• Break mass: The "knee" or characteristic mass $M^*$ increases with redshift, reaching $M^*\sim10^{9}\,M_\odot$ at $z\sim6$ (Wu et al., 2022, Lai et al., 2024).
• Space density: At fixed mass ($10^9\,M_\odot$), space density $\Phi$ rises from $$\sim 3\times10^{-8}\,\mathrm{Mpc}^{-3}\mathrm{dex}^{-1}$$ at $z\sim0.5$ to $\sim10^{-7}$ at $z\sim2$, then declines to $\sim10^{-8}$ at $z\sim4$, and further at higher $z$ (Vestergaard, 2010, Wu et al., 2022).
• Redshift evolution: The number density of massive black holes ($M_{\rm BH}\gtrsim10^9\,M_\odot$) peaks at earlier epochs ($z\sim2$–$3$). The redshift of the peak shifts to lower $z$ for lower-mass SMBHs, consistent with AGN "downsizing" (Kelly et al., 2012, Kelly et al., 2010).

The characteristic double power-law (or "broken power-law") form is seen in both high-redshift ($z\simeq5$–6) and low-redshift ($z\lesssim1$) measurements, with the normalization at $z=6$ orders of magnitude below the local relic black hole mass function (Willott et al., 2010, Wu et al., 2022).

Representative double power-law fits for high-$z$ ($z\simeq6$; Wu et al. 2021, (Wu et al., 2022)): $$\Phi^* = 10^{-6.47}\,\mathrm{Mpc}^{-3}\,\mathrm{dex}^{-1}, \quad M^* = 10^{9.04}\,M_\odot, \quad \alpha = -0.93, \quad \beta = -4.47$$ or (z~5, Onken et al. 2024, (Lai et al., 2024)): $$\Phi^* = 4.6 \times 10^{-7}\,\mathrm{Mpc}^{-3}\,\mathrm{dex}^{-1}, \quad M^* = 10^{8.64}\,M_\odot, \quad \alpha = -1.26, \quad \beta = -3.12$$

4. Eddington Ratio Distributions, Duty Cycles, and Accretion Physics

Because the quasar mass function reflects only the actively accreting SMBH population, its normalization is sensitive to both the Eddington ratio distribution (ERDF) and the active fraction (duty cycle).

• The intrinsic $P(\log \lambda)$ is typically lognormal or Schechter-like with a peak at $\log \lambda \sim -0.9$ (i.e., $\lambda\sim0.12$), scatter $\sim0.3$–$0.4$ dex, and very weak mass dependence (Kelly et al., 2012, Wu et al., 2022, Lai et al., 2024).
• At $z\sim6$, quasars are found to accrete at or near the Eddington limit (Willott et al., 2010, Wu et al., 2022), while at $z<2$, typical Eddington ratios drop and the duty cycle for luminous broad-line quasars can fall to $\sim10^{-2}$ for $M_{\mathrm{BH}}\sim10^9\,M_\odot$ (Kelly et al., 2012, Kelly et al., 2010).
• At fixed mass bin, the typical quasar growth time $$t_{\mathrm{grow}} = (\lambda U [1-\epsilon]/\epsilon)^{-1} t_{\mathrm{Salp}}$$ is comparable to or exceeds the Hubble time for the most massive systems at low $z$, requiring more rapid, likely obscured, growth in an earlier phase (Lai et al., 2024, Kelly et al., 2010).

Systematic uncertainties in the mass function normalization at high $z$ arise primarily from the correction for obscured AGN (which can boost the intrinsic space density by factors $\sim2$–3) and the assumed duty cycle ($DC\sim0.5$–1.0) (Willott et al., 2010, Wu et al., 2022), as well as selection effects that bias observed Eddington ratio distributions toward higher $\lambda$ (Li et al., 2022, Wu et al., 2022).

5. Physical Interpretation and Theoretical Implications

The observed shape and evolution of the quasar mass function encode important aspects of SMBH and galaxy co-evolution.

• Cosmic downsizing: The shift in the space density peak toward lower $z$ for lower masses ("anti-hierarchical" growth) matches predictions from merger-triggered, feedback-limited growth models (Kelly et al., 2012, Kelly et al., 2010, Li et al., 2022, Zhang et al., 2023).
• High-$z$ quasar formation: Double power-law behavior and steep high-mass slope at $z\sim6$ are consistent with rapid, episodic, and sometimes super-Eddington growth, as modeled in continuity- or burst-growth frameworks with seed black holes in the $10^3$–$10^5\,M_\odot$ range. The normalization of $\Phi(M)$ at $z=6$ is $\sim10^4$ times below $z=0$, a much sharper drop than in the global stellar mass function ($\sim10^2$ times), requiring constraints on early black hole seeding and radiative growth efficiency (Willott et al., 2010, Li et al., 2022, Li et al., 2023).
• Direct collapse and analytic models: In direct-collapse scenarios, the mass function at the end of the super-Eddington phase is a "tapered power-law," featuring a shallow slope at intermediate masses and a sharp high-mass cutoff, naturally reproducing the observed knee and rapid decline seen in $z\sim6$–7 QLFs (Basu et al., 2019).
• Empirical models: Simple linear mappings from the halo–stellar–BH mass hierarchy with log-normal ERDFs reproduce the observed QLF and clustering, requiring only a mass-independent duty cycle and $M_{\mathrm{BH}}$–$M_{\star}$ relation at each epoch (Conroy et al., 2012, Zhang et al., 2023).

6. The Quasar Halo Mass Function and Environment

Recent clustering and lensing studies allow direct inference of the host halo mass function for quasars:

• At $z\sim1.7$, the typical halo mass occupied by quasars is $M_{\text{halo}} \sim 10^{12.2}$–$10^{12.5}\,h^{-1}M_\odot$ with a narrow dispersion: two-thirds occupy halos within a factor of $\sim3$ of the mean (Eltvedt et al., 2024). This result holds across multiple estimation techniques (angular auto-correlation, CMB lensing cross-correlation, HOD modeling).
• The abundance as a function of $M_\text{halo}$, $$n_q(M) \propto (M/10^{12.4})^{-1.8} \exp(-M/10^{13.3})$$, closely tracks expectations from hierarchical structure formation and is consistent with independently inferred AGN–stellar–halo mass mappings.
• The quasar halo mass function is nearly invariant with redshift below $z\sim2$, suggesting that the observed luminosity evolution of the QLF reflects changes in fuel supply rather than in the abundance of suitable host halos (Eltvedt et al., 2024).

7. Prospects, Systematics, and Future Directions

The leading systematic uncertainties in quasar mass functions include:

• Population of heavily obscured or Compton-thick AGN, especially at $z>4$ (Willott et al., 2010, Wu et al., 2022);
• Uncertainties and biases in virial mass estimators, particularly for Mg II and C IV at high $z$ (Shen et al., 2011, Wu et al., 2022);
• Completeness at low masses ($M<10^9\,M_\odot$) and low Eddington ratios due to survey flux limits (Kelly et al., 2012, Kelly et al., 2010).

Forthcoming wide/deep IR and X-ray surveys (JWST, Euclid, Roman, eROSITA, Athena) will probe lower-luminosity populations and fainter mass regimes, enabling direct measurement of the low-mass slope, as well as robust determination of the high-mass cutoff and true duty cycles at $z>6$ (Li et al., 2023, Li et al., 2022, Zhang et al., 2023). Continuity-equation and population synthesis approaches, incorporating luminosity-dependent obscuration and evolving Eddington ratio distributions, will further refine constraints on SMBH fueling and quasar phase lifetimes.

Theoretically, further ties between the mass functions of quasars, galaxies, and their halos through empirical and semi-empirical models, and their coherent connection to hydrodynamic simulations, are essential for quantitatively linking the observed black hole growth to cosmological structure formation and galaxy feedback mechanisms (Zhang et al., 2023, Conroy et al., 2012).