Kicking time back in black-hole mergers:
Ancestral masses, spins, birth recoils and hierarchical-formation viability of GW190521

Abstract

Pair-instability supernova (PISN) is thought to prevent the formation of black holes from stellar collapse within the approximate mass range $M\in[65,130]M_{\odot}$. However, such black holes may form through hierarchical formation channels, as the result of merging “ancestor” black holes, whose properties determine those of the “child” black hole, namely its mass, spin, and recoil velocity (or kick). Crucially, the child black hole will be expelled from its host environment if its “birth kick” exceeds the corresponding escape velocity, preventing it from undergoing further mergers. In this work, we exploit tight relations between the final kick and final spin of quasi-circular black-hole mergers to obtain posterior probability distributions for the hypothetical ancestral masses, spins and birth kicks of the component black holes of GW190521, assuming both quasi-circular (with generic spins) and eccentric (aligned-spin) scenarios for the latter. With this, we evaluate the probability $p_{2g}$ that the GW190521 components formed from the merger of stellar-origin black holes, being retained by the host environment. For the primary component BH1, which falls squarely in the PISN mass gap, we find that such scenario is strongly suppressed if GW190521 happened in a Globular Cluster, unless GW190521 was quasi-circular and BH1 had aligned-spin ancestors, characteristic of isolated black-hole binary formation scenarios. Similarly, for denser host environments (excluding AGNs), we only obtain probabilities $p_{2g}>0.5$ if GW190521 was quasi-circular and BH1 had aligned-spin ancestors. If GW190521 was eccentric, we find maximal values $p_{2g}$ of only $p_{2g}\simeq 0.1$ which require ancestral mass ratios $q\geq 2$, untypical of isolated formation scenarios. Our results seem consistent with an AGN origin for GW190521, especially if this was eccentric.

1 Introduction

The way black holes (BHs) form and grow is an open question in astrophysics. While BHs with masses in the approximate range $M\in[5,65]M_{\odot}$ are well explained through stellar collapse, the formation of more massive ones, like intermediate-mass BHs (with $M\in[10^{2},10^{5}]M_{\odot}$) or the supermassive BHs populating the center of most galaxies, is still unclear. In this context, and after more than 90 detections of compact mergers Abbott et al. (2021a) by the Advanced LIGO-Virgo network Aasi et al. (2015); Acernese et al. (2015), now joined by the KAGRA detector, Akutsu et al. (2020), gravitational-wave (GW) astronomy represents a powerful tool to shed light on our understanding of BH formation and population Abbott et al. (2021b).

While the vast majority of current BBH observations display component BHs consistent with a stellar-collapse origin, some of them deviate from this paradigm. In particular, the event GW190521, originally reported Abbott et al. (2020a, b) as a quasi-circular BBH merger with slight evidence for orbital spin-induced orbital precession (hereafter referred to as “precession”), involves a component BH with a mass ($m_{1}=85^{+21}_{-14}M_{\odot}$), squarely fitting within the the so-called pair-instability supernova (PISN) gap, which ranges approximately in $M\in[65,130]M_{\odot}$. According to our current understanding of stellar evolution, BHs in such mass range cannot form from stellar collapse, as the corresponding star would be completely disrupted Heger et al. (2003); Woosley & Heger (2021) . Notably, while orbital precession can be mistaken by eccentricity for GW190521-like events Bustillo et al. (2021a), alternative studies including orbital eccentricity report a consistent primary mass Romero-Shaw et al. (2020); Gayathri et al. (2022); Gamba et al. (2022). The latter is also true for other studies including orbital eccentricity but that find no evidence for it Ramos-Buades et al. (2023). This indicates that while the analysis of this GW190521 may still be subject to significant unknowns including waveform systematics, conclusions on its primary mass seem to be robust, except if very informative prior distributions on the masses are considered Fishbach & Holz (2020). We note that alternative interpretations for GW190521 exist in the literature, including the merger of boson stars Bustillo et al. (2021b); Calderón Bustillo et al. (2023) or that of dwarf galaxies Palmese & Conselice (2021). Additionally, it has been proposed that BHs populating the PISN gap can be populated by processes such as star-mergers Costa et al. (2022), convective overshooting Tanikawa et al. (2021) or the collapse of blue supergiants Winch et al. (2024). Despite these options, most hypotheses point that the primary component BH of GW190521 should be a second-generation BH, i.e., formed through a previous merger of BHs formed through stellar collapse Abbott et al. (2020b); Kimball et al. (2021a); Anagnostou et al. (2022); Mahapatra et al. (2022) .

The above situation has triggered attempts to estimate the masses of the putative “ancestors” of GW190521 Barrera & Bartos (2022, 2023). These found that, indeed, its primary BH is consistent with the merger of two BHs with masses below the PISN gap, therefore consistent with a stellar origin. Such studies and the corresponding astrophysical conclusions, however, are limited by three main fundamental aspects. First, they impose that the posterior for the remnant BH, later observed by GW detectors as a component of a BBH, is either a delta function at $a_{f}=0.7$ (typical of remnants of non-spinning BBHs) or follows a flat distribution within $a_{f}\in[0,1]$. This does not exploit the rich information encoded in the actual probability distribution measured through the observation of GW190521 which, as we will show, is crucial for our conclusions. Second, when considering the ancestral spins, they restrict these to be aligned with the angular momentum, therefore omitting orbital precession Barrera & Bartos (2023) which is highly likely to occur in dynamical environments like globular clusters Talbot & Thrane (2017); Rodriguez et al. (2016). Finally, both works omit one of the key aspects of this work, namely the impact of the gravitational recoil or “birth kick” inherited by BHs born from BBHs. Such information os vital, as the kick determines whether the remnant BH can remain within its environment and undergo the subsequent merger eventually observed by our detectors.

The asymmetric emission of GWs by BBHs produces a net emission of linear momentum Thorne (1980). This imparts a gravitational recoil, or “kick” Gonzalez et al. (2007), to the remnant BH which can reach thousands of km/s in the most extreme cases Brügmann et al. (2008); Sperhake et al. (2011) . If the kick exceeds the escape velocity $v_{\text{esc}}$ of the host environment, the BH will abandon it, preventing its participation in subsequent mergers. Therefore, retaining a second-generation BH within its host environment requires the latter to have an escape velocity that exceeds the BH kick. To date, kick estimates in existing GW events have focused on that inherited by the final BH Calderón Bustillo et al. (2018); Varma et al. (2020, 2022); Mahapatra et al. (2021); Bustillo et al. (2022). In contrast, in this work, we perform for the first time estimates of the (putative) birth kicks of the component BHs observed in BBH observations, with the goal of assessing their viability as products of previous mergers of stellar-origin BHs as a function of the properties of the host environment.

Making use of the full parameter estimates for the components of GW190521, we infer the masses and spins of their putative ancestor BHs, together with the corresponding birth kicks. We do this considering both quasi-circular Abbott et al. (2020a) and eccentric Romero-Shaw et al. (2020) scenarios for GW190521, together with three different priors on the ancestor BBH parameters, which correspond to different BBH formation channels: dynamical and isolated. Importantly, to estimate birth kick, we exploit the tight relations between the final kick and final spin of quasi-circular BBH mergers (LABEL:fig:1), which we will later discuss. Finally, we assess the viability of the GW190521 component BHs as the result of the merger of stellar-origin BHs. To this, we estimate the probability that both ancestors have masses $(m_{i1},m_{i2})<65M_{\odot}$ at the same time that the birth kick $v_{f}$ is lower than the escape velocity $v_{esc}$ of given host environments.

Figure 1: Relation between remnant spin and kick in quasi-circular black-hole mergers. The left panel assumes an isotropic spin prior allowing for orbital precession and mass ratios $q\leq 6$ while the central and right ones are restricted to aligned-spins, parallel to the orbital angular momentum of the binary, and mass ratios $q\leq 2$ and $q\leq 6$ respectively. The color code denotes the probability density.

2 Method

2.1 GW190521 parameter estimates

Our starting point is the posterior probability distributions for the masses $(m_{1},m_{2})$ and spin-magnitudes $(a_{1},a_{2})$ of the component BHs of GW190521. We use posteriors obtained both by the LIGO-Virgo-KAGRA (LVK) collaboration Abbott et al. (2020a) (for the case of a quasi-circular merger scenario) and by Romero-Shaw et. al. Romero-Shaw et al. (2020) (for the case of an eccentric merger). Next, we construct prior probability distributions for the masses and spins of the ancestral BHs $p(\theta_{anc}\equiv\{m_{i1},m_{i2},\vec{a_{i1}},\vec{a_{i2}}\})$, which we will later discuss. From these, we obtain the induced prior distributions $p(m_{f},a_{f},k_{f})$ for the final BH mass $M_{f}$ and final spin $a_{f}$ (which we will use to connect the GW190521 components with its ancestors), together with the remnant recoil velocity $v_{f}$. We obtain these using the surrogate model for the final BH properties NRSur7dq4Remnant Varma et al. (2019a, b).

2.2 Estimating the ancestor parameters

The posterior probability $p(\theta_{anc}|d)$ for the ancestor parameters given GW data $d$ is given by

$$p(\theta_{anc}|d)\propto p(\theta_{anc})p(d|\theta_{anc}).$$ (1)

Above, $p(d|\theta_{anc})$ denotes the likelihood for $\theta_{anc}$ given the GW data. This can be expressed as a function of the marginal likelihood for the parameters of the component BHs $\theta_{child}=\{m_{i},a_{i}\}$ of the BBH observed by GW detectors, on which waveform templates depend, allowing us to express

$$p(\theta_{anc}|d)\propto$$ (2)

$$p(\theta_{anc})p(d|\theta_{anc})\propto$$

$$p(\theta_{anc})\int p(\theta_{child}|\theta_{anc})p(d|\theta_{child})d\theta_{child}\propto$$

$$p(\theta_{anc})\int p(\theta_{anc}|\theta_{child})p(\theta_{child})p(d|\theta_{child})d\theta_{child}\propto$$

$$p(\theta_{anc})\int p(\theta_{anc}|\theta_{child})p(\theta_{child}|d)d\theta_{child}.$$

Above, the term $p(\theta_{child}|d)$ denotes the posterior distribution for the parameters $\theta_{child}$ of the putative “child” BHs observed through GWs, which we obtain directly from existing analyses of GW190521.

2.3 Practical implementation

In practice, we implement the following procedure to obtain samples from the posterior distribution of the ancestral BH parameters $p(\theta_{anc}|d)$:

1. First, we generate $2\times 10^{6}$ random samples for the ancestral prior $p(\theta_{anc})$, computing the corresponding $\{m_{f},a_{f},k_{f}\}(\theta_{anc})$.

2. For each sample $\{m_{i},a_{i}\}$ of the posterior distribution for GW190521 $p(\theta_{child}|d)$, we draw random samples from $p(\theta_{anc})$ satisfying $\{m_{f},a_{f}\}(\theta_{anc})=\{m_{i}\pm\delta m_{i},a_{i}\pm\delta a_{i}\}$.

The usage of non-zero tolerances $\delta m_{i}$, $\delta a_{i}$ is motivated by the fact that the discrete nature of our prior samples for $p(\theta_{anc})$ would naturally prevent us from encountering samples exactly matching the condition imposed in Step 2 above. In our case, we set $(\delta m_{i},\delta a_{i})=(1,0.05)$, checking that this produces stable results.

Finally, we compute the probability that the component BHs of GW190521 are the product of the merger of stellar-origin BHs. To do this, we compute the probability that both ancestors have masses below $65M_{\odot}$, at the same time that $k_{f}<v_{esc}$, with $v_{esc}$ denoting the escape velocity of the host environment. This is, we compute

$$p(k_{f}<v_{esc},m_{1,2}<65M_{\odot}).$$ (3)

2.4 Priors for the ancestor parameters

The priors for the ancestral parameters $\theta_{anc}$ can have a strong impact on the results, specially in cases where the spins of the component BHs are not well constrained. First, the priors for the ancestral masses will clearly impact the chances that both of them are below $65M_{\odot}$. Second, regarding the ancestral spins, it is well known that while precessing BBHs can produce recoils exceeding $3000$ km/s, those for aligned-spin BBHs are capped at around $500$ km/s Healy et al. (2014). Therefore, different assumptions for the hypothetical ancestral masses and spins can strongly impact the hierarchical-origin nature of the observed BH.

In light of the above, we consider three priors for the ancestral BBHs:

a) A “precessing prior” (PP) where we consider generically spinning binaries with spins isotropically distributed on the sphere with spin magnitudes uniformly distributed in $a_{i}\in[0,0.99]$. We pair this with priors on the masses $m_{i}\in[5,170]$ $M_{\odot}$, large enough to encompass the ancestral masses. We constrain the mass ratio to $q\in[1,6]$ which corresponds to the limits of NRSur7dq4Remnant.

b) A “restricted” aligned-spin prior (APq2) where we consider binaries with masses uniformly distributed in $m_{i}\in[5,170]M_{\odot}$, restricted to mass ratios $q\in[1,2]$, and aligned spins with direction parallel to the orbital angular momentum of the binary and spin magnitudes uniformly distributed in $a_{i}\in[0,0.99]$.

c) An aligned-spin prior (APq6), identical to the APq2 in b), but expanded to mass ratios $q\leq 6$.

The three above priors closely relate to the possible formation channels of the ancestral binaries and their host environments Mapelli et al. (2020); Farr et al. (2017). While isotropic spin distributions are expected in dynamical formation scenarios characteristic of dense environments Tutukov & YungelSon (1993); Belczynski et al. (2016), aligned spins are characteristic of isolated binary formation Sigurdsson & Hernquist (1993); Portegies Zwart & McMillan (2000); Rodriguez et al. (2016). The restriction of the aligned-spin prior to mass ratios $q\leq 2$ is motivated by results indicating that binaries originated from isolated binaries tend to display such small mass ratios Belczynski et al. (2016); Giacobbo & Mapelli (2018); Mandel & de Mink (2016). As we will see, each case leads to significantly different final spin distributions, as well as different kick distributions for a given final spin. As a consequence, on the one hand, this means these scenarios will have different abilities to produce GW190521-like component BHs. On the other hand, even if both scenarios could form such BHs, the inherited kicks will widely differ, leading to very different probabilities for the BHs to retain their environments and undergo subsequent mergers. We note, however, that we have found that the restriction to $q\leq 2$ in our APq2 prior fails to reproduce most of the posterior distribution for the spin magnitudes of GW190521 $a_{1,2}$. For this reason, we also explore our c) prior, expanded to $q\leq 6$. Finally, we note that several works Fishbach & Holz (2020); Nitz & Capano (2021); Chandra et al. (2023) have shown the important role of mass-prior choices in analysing high-mass events. For this reason, we have also considered the above three priors, but using mass-priors uniform in total mass $M$ and mass ratio $q\in[1,6]$. We have found, however, that our results are barely sensitive to this change.

3 Results

Figure 2: Spin magnitudes for the GW190521 component black holes. The left (right) panel corresponds to the primary (secondary) black hole. Blue distributions correspond to the quasi-circular precessing scenario for GW190521, originally reported by the LVK Abbott et al. (2020a), while the orange ones correspond to the eccentric non-precessing scenario described in Romero-Shaw et. al. Romero-Shaw et al. (2020).

Figure 3: Posterior distributions for the ancestral masses and birth kicks of the primary component black hole of GW190521. We show the posterior two-dimensional $68\%$ and $90\%$ credible regions, together with the corresponding one-dimensional posterior distributions. The left panel corresponds to a prior allowing for precessing ancestors while the central and right panels correspond to aligned-spin ancestors, respectively restricted to mass ratios $q\leq 2$ and $q\leq 6$. Results obtained under the original LVK analysis of GW190521 are shown in blue while those obtained under the eccentric merger analysis by Romero-Shaw et. al. are shown in orange. Prior distributions are shown in gray. We note that the prior corresponding to our central panel fails to reproduce around $90\%$ of the posterior distribution for the spin magnitude of the GW190521 component BHs.

Figure 4: Posterior distributions for the ancestral masses and birth kicks of the secondary component black hole of GW190521. Same as Figure 3 but for the secondary component of GW190521.

3.1 Final spin vs. final kick relation

Figure 1 shows the final spin and kick distributions obtained under each of our three priors for the ancestral BBHs. The left panel corresponds to our “precessing prior”, which includes orbital precession. The central and right panels corresponds to our two “aligned-spin” priors. There are several aspects that must be highlighted. First, a tight relation between the final spin and kick can be observed. This clearly shows that accurate estimates of the “child” BH spin magnitude can place strong constraints on its putative birth kick. Second, for our precessing prior, low spins impose significantly tighter (lower and upper) bounds on the birth kick than large spins. In particular, note that while spins smaller than $\simeq 0.2$ lead to kicks below $500$ km/s (which would be retained in Milky Way like galaxies and denser) these must also surpass $\simeq 100$ km/s, enough to expel the final BH from any globular cluster. A qualitatively similar situation is observed for our “aligned spin” priors, with kicks capped at $500$ km/s due to the absence of precession.

3.2 Spin magnitudes of the GW190521 component black holes

We now discuss the estimates for the spin magnitudes of the component BHs of GW190521, which are key to infer the putative birth kicks. The left and right panels of Figure 2 show respectively the posterior distributions for the spin magnitudes of the primary and secondary component BHs of GW190521. Blue posteriors correspond to the original analysis performed by the LVK Abbott et al. (2020a), using the state-of-the-art model for quasi-circular precessing BBHs NRSur7dq4 Varma et al. (2019a). Orange posteriors correspond to the analysis by Romero-Shaw et. al. Romero-Shaw et al. (2020) using a non-precessing model for BBHs with eccentricities in the range $e_{10}=[0,0.2]$ Cao & Han (2017); Liu et al. (2019), estimated at an orbital frequency $f_{\rm ref}=10$ Hz.

From the above posteriors, it is rather direct to expect that birth-kick estimates will not be particularly informative under the quasi-circular scenario for GW190521. The reason is that while the posteriors for both component BHs clearly rail against the Kerr limit, these are somewhat flat, offering good support for all spin magnitudes. In contrast, under the eccentric hypothesis, both spins are constrained to values below $\sim 0.4$, clearly peaking at $0$, which will lead to strong constraints on the possible birth kicks.

Table 1: 90% credible intervals for the masses and spins of the ancestors and birth kick of the component black holes of GW190521 for our precessing ancestor prior (PP) and aligned-spin ones (APq2, APq6). We display also the corresponding effective-spin and effective precessing-spin. The asterisk accompanying the APq2 prior for the eccentric scenario for GW190521, denotes that such prior fails to reproduce around $90\%$ of the posterior distribution for the spin magnitude of the component BHs.

	GW190521 Scenario	Ancestors Prior	$m_{1}$ ($M_{\odot}$)	$m_{2}$ ($M_{\odot}$)	$v_{f}$ (km/s)	$a_{1}$	$a_{2}$	$\chi_{eff}$	$\chi_{p}$
		PP	$64^{+20}_{-18}$	$22^{+21}_{-10}$	$404^{+1423}_{-271}$	$0.75^{+0.24}_{-0.64}$	$0.52^{+0.43}_{-0.47}$	$0.10^{+0.64}_{-0.74}$	$0.39^{+0.45}_{-0.29}$
	Circular	APq2	$53^{+15}_{-11}$	$36^{+13}_{-9}$	$104^{+240}_{-93}$	$0.72^{+0.26}_{-0.64}$	$0.59^{+0.38}_{-0.53}$	$0.21^{+0.70}_{-1.06}$	$N/A$
		APq6	$61^{+20}_{-17}$	$26^{+19}_{-13}$	$117^{+208}_{-100}$	$0.66^{+0.32}_{-0.58}$	$0.52^{+0.44}_{-0.46}$	$0.15^{+0.71}_{-0.89}$	$N/A$
Primary BH
		PP	$78^{+27}_{-16}$	$17^{+9}_{-5}$	$237^{+147}_{-81}$	$0.70^{+0.27}_{-0.44}$	$0.53^{+0.43}_{-0.47}$	$-0.55^{+0.42}_{-0.23}$	$0.16^{+0.24}_{-0.10}$
	Eccentric	APq2∗	$64^{+22}_{-15}$	$35^{+23}_{-8}$	$264^{+119}_{-178}$	$0.89^{+0.10}_{-0.39}$	$0.69^{+0.29}_{-0.60}$	$-0.77^{+0.42}_{-0.20}$	$N/A$
		APq6	$76^{+22}_{-16}$	$18^{+14}_{-5}$	$225^{+98}_{-74}$	$0.75^{+0.23}_{-0.41}$	$0.50^{+0.45}_{-0.46}$	$-0.60^{+0.35}_{-0.28}$	$N/A$
		PP	$49^{+17}_{-16}$	$17^{+17}_{-8}$	$437^{+1415}_{-307}$	$0.77^{+0.22}_{-0.66}$	$0.51^{+0.44}_{-0.45}$	$0.16^{+0.59}_{-0.77}$	$0.40^{+0.44}_{-0.30}$
	Circular	APq2	$41^{+12}_{-11}$	$28^{+10}_{-8}$	$96^{+244}_{-87}$	$0.73^{+0.25}_{-0.65}$	$0.60^{+0.37}_{-0.54}$	$0.27^{+0.65}_{-1.10}$	$N/A$
		APq6	$47^{+17}_{-15}$	$20^{+15}_{-10}$	$104^{+218}_{-88}$	$0.67^{+0.31}_{-0.59}$	$0.52^{+0.44}_{-0.47}$	$0.23^{+0.64}_{-0.95}$	$N/A$
Secondary BH
		PP	$55^{+15}_{-16}$	$12^{+6}_{-4}$	$243^{+175}_{-89}$	$0.70^{+0.27}_{-0.47}$	$0.49^{+0.45}_{-0.43}$	$-0.53^{+0.42}_{-0.26}$	$0.17^{+0.27}_{-0.10}$
	Eccentric	APq2∗	$42^{+13}_{-13}$	$24^{+7}_{-7}$	$267^{+115}_{-179}$	$0.92^{+0.08}_{-0.39}$	$0.71^{+0.27}_{-0.61}$	$-0.81^{+0.45}_{-0.16}$	$N/A$
		APq6	$53^{+15}_{-17}$	$13^{+9}_{-4}$	$228^{+98}_{-77}$	$0.75^{+0.23}_{-0.44}$	$0.50^{+0.45}_{-0.45}$	$-0.59^{+0.37}_{-0.28}$	$N/A$

Table 2: Probability $p_{2g}$ for ancestor masses falling below the PISN mass gap at the same time that birth kick is lower that the typical escape velocity of a Globular Cluster (${\rm GC}$), the Milky Way (${\rm MW}$), a Nuclear Star Cluster (${\rm NSC}$) or an Elliptical Galaxy (${EG}$). We show values obtained for two orbital models under two prior settings. The asterisk accompanying the APq2 prior for the eccentric scenario for GW190521, denotes that such prior fails to reproduce around 90% of the posterior distribution for the spin magnitude of the component BHs.

	GW190521 Scenario	Ancestors Prior	$p^{\rm GC}_{2g}$	$p^{\rm MW}_{2g}$	$p^{\rm NSC}_{2g}$	$p^{\rm EG}_{2g}$
		PP	0.002	0.293	0.461	0.518
	Circular	APq2	0.259	0.903	0.903	0.903
		APq6	0.135	0.605	0.605	0.605
Primary BH
		PP	0.000	0.114	0.116	0.116
	Eccentric	APq2*	0.011	0.572	0.572	0.572
		APq6	0.000	0.143	0.143	0.143
		PP	0.004	0.563	0.834	0.927
	Circular	APq2	0.326	0.997	0.997	0.997
		APq6	0.255	0.951	0.951	0.951
Secondary BH
		PP	0.000	0.839	0.849	0.849
	Eccentric	APq2*	0.026	0.998	0.998	0.998
		APq6	0.002	0.878	0.878	0.878

3.3 Ancestral BBHs of GW190521

Figure 3 shows the posterior two-dimensional $90\%$ credible regions, together with the corresponding one-dimensional posterior distributions, for the ancestral masses and birth kick of the primary BH of GW190521. Figure 4 shows the same results for the secondary BH. The left panel shows results obtained under our “precessing” prior for the ancestors while the central and right panels show those corresponding to our “aligned-spin” priors, respectively restricted to mass ratios $q\leq 2$ and $q\leq 6$. The results in blue and orange respectively correspond to the hypothesis that GW190521 was a quasi-circular and an eccentric BBH . Prior distributions are represented in grey. We note that we have restricted the plot ranges to values encompassing our posteriors, to improve the visibility of the results. Next, Fig. 5 shows the posterior distributions for the individual spin magnitudes of the ancestral BHs. The left (right) panels correspond to our precessing (aligned-spin) ancestral priors. Finally, Fig. 6 shows the posterior distributions for the effective-spin parameter $\chi_{\rm eff}$ Santamaria et al. (2010) and the effective precessing-spin parameter $\chi_{p}$ Schmidt et al. (2012, 2015). All results are reported numerically in 1, in terms of median values accompanied by symmetric $90\%$ credible intervals. As expected, our results show a strong dependence on both the prior choice for the ancestors and the scenario considered for GW190521, as we will discuss in detail. While we report results for the putative ancestors of both components of GW190521, in the following we will mostly focus our discussion on those of the primary BH, as this has an extremely low probability of residing below the PISN gap.

Figure 5: Ancestral spin magnitudes. Posterior distributions for the spin magnitudes of the ancestral BBHs of GW190521. Left panels correspond to our precessing prior for the ancestors while right ones correspond to our aligned-spin prior. The blue and orange distributions correspond respectively to a quasi-circular and eccentric scenario for GW190521. Prior distributions are represented in grey.

Figure 6: Ancestral effective spin parameters. Same as Figure 5 for the effective spin $\chi_{\rm eff}$ and effective spin-precession $\chi_{p}$ parameters of the ancestral black-hole binaries of GW190521.

3.3.1 LVK quasi-circular precessing scenario

Under our precessing prior, we find ancestral masses for the primary BH $(m_{11},m_{12})=(64^{+20}_{-18},22^{+21}_{-10})M_{\odot}$. Given our reference range for the PISN gap, we obtain a $\sim 50\%$ chance that both ancestors reside below it. For the primary ancestor, we obtain an informative posterior distribution for the spin magnitude $a_{11}$ that rails against the Kerr limit, similar to that of $a_{1}$ itself. In contrast, we obtain a totally uninformative posterior for $a_{12}$, owing to its smaller impact on the resulting final spin. As per the spin orientations, we obtain a quite unconstrained but informative posterior for $\chi_{\rm eff}$. This deviates from the prior distribution through peaks at highly positive and negative $\chi_{\rm eff}$, while showing less support than the prior for nearly null $\chi_{\rm eff}$. This result reflects the fact that the production of nearly null (extremal) remnant spins, both well supported by the $a_{1}$ posterior, requires highly negative (positive) ancestral $\chi_{\rm eff}$. In contrast, the prior (which peaks at $\chi_{\rm eff}=0$) highly suppresses such remnant spin values, showing a preference for $a_{f}\in[0.4,0.6]$ (see Figure 1). Finally, we obtain a totally uninformative posterior for $\chi_{p}$. The latter, together with the uninformative posterior for $a_{1}$, leads to a highly unconstrained posterior for the the putative birth recoil $v_{f}=404^{+1423}_{-271}$km/s. This yields almost zero chanceS for this BH to be retained in a Globular Cluster, which are typically considered to have maximal escape velocities of $50$ km/s. While we defer detailed explanations to the next section, the latter already shows the dramatic role of the inclusion of the recoil velocity in this type of study.

Our aligned-spin ancestral prior restricted to $q\leq 2$ (APq2) leads to slightly smaller(larger) ancestral primary(secondary) masses $(m_{11},m_{12})=(53^{+15}_{-11},36^{+13}_{-9})M_{\odot}$, notably raising the probability that both ancestral masses lay below the PISN gap. This is simply due to the limitation imposed in the ancestral mass ratio. We note, however, that this prior fails to produce BHs with spins $a_{f}\leq 0.2$, therefore failing to capture around $10\%$ of the posterior distribution for the primary BH of GW190521. Expanding the allowed ancestral mass ratio to $q\leq 6$, we obtain masses totally consistent with those obtained under our precessing prior. The kick estimation, however, is radically different, owing to the lower kicks allowed by aligned-spin mergers. In particular, we obtain $v_{f}=104^{+240}_{-93}$ km/s and $v_{f}=117^{+208}_{-100}$ km/s under each of our aligned-spin priors. While, once again, this essentially ensures the ejection of the final BH from a Globular Cluster, the probability of retention in denser environments is significantly increased, as we will later discuss.

For the ancestors of the secondary BH of GW190521, we obtain similar results of which we highlight two aspects. First, while the mass estimates are naturally smaller, we still obtain a non-zero chance for the primary ancestor to be within the PISN gap. Second, we also obtain non-zero chances for the recoil to be below the $50$ km/s limit characteristic of globular clusters Baumgardt & Hilker (2018).

3.3.2 Eccentric non-precessing scenario

The assumption of this scenario for GW190521 has dramatic consequences due to the significantly different spin estimates it leads to. First, we note that our APq2 aligned-spin prior fails to reproduce almost $90\%$ of the posterior spin distribution for both GW190521 components, therefore not posing a viable scenario for the ancestral masses. Under our precessing prior, we obtain ancestral masses for the primary component of GW190521 $(m_{11},m_{12})=(78^{+27}_{-16},17^{+9}_{-5})M_{\odot}$. Importantly, this leaves only a $\simeq 10\%$ chance for both ancestral masses to reside below the PISN gap, much lower than the $\simeq 50\%$ chance obtained under the LVK quasi-circular scenario. Equivalent results are obtained for the masses under the aligned-spin APq6 prior. For the ancestral spins, as expected from the $a_{1}$ posterior, we obtain results that are very different from those of the quasi-circular case. First, we obtain an informative estimate $a_{11}=0.70^{+0.27}_{-0.44}$, clearly peaking around $a_{11}=0.7$, together with an uninformative distribution for $a_{12}$. Second, we obtain informative posteriors for $\chi_{\rm eff}$ and $\chi_{\rm p}$. These respectively yielding $\chi_{\rm eff}=-0.55^{+0.42}_{-0.23}$, essentially ruling out positive values, and $\chi_{p}=0.16^{+0.24}_{-0.10}$, rather discarding strong precession. As discussed earlier, these informative posteriors arise from the informative posterior for $a_{1}$, which peaks at $a_{1}=0$ and is negligible beyond $a_{1}=0.3$. On the one hand, the production of such remnant BHs, requires highly negative effective spins. On the other hand, the latter also requires the spins to form rather angles of nearly 180 degrees with the orbital angular momentum (i.e., the spins must be essentially anti-aligned with the orbital angular momentum), which prevents the system from having large in-plane components and, therefore, strong precession. These properties lead to a highly informative kick estimate $v_{f}=237^{+147}_{-81}$km/s, dramatically different from that inferred under the LVK scenario. This means that the upper bound for the birth kick is significantly decreased with respect to the LVK case, therefore increasing the chances of retention in dense environments, even under the precessing ancestral prior. On the contrary, we find that assuming an aligned-spin ancestral prior, the lower bound for the birth kick $225^{+98}_{-74}$ km/s is much larger under LVK quasi-circular scenario for GW190521 (owing to the $\chi_{\rm eff}$) posterior, making it impossible to retain the final BH in a Globular Cluster.

Regarding the putative ancestors of the secondary BH, we find again a non-zero chance for the primary ancestor to reside in the PISN gap. However, just as for the primary BH, we estimate a birth kick of $v_{f}={243^{+175}_{-89}}$ km/s that cannot be retained in a Globular Cluster.

3.4 Viability of the GW190521 components as second-generation black holes

We now evaluate the probability that the component BHs of GW190521 can be the remnant of a previous merger of stellar-origin BHs. To do this, for each BH, we compute the probability $p_{2g}$ that both ancestors have masses $(m_{i1},m_{i2})<65M_{\odot}$ at the same time that the corresponding kick satisfies $v_{f}<v_{esc}$, where $v_{esc}$ denotes the escape velocity of the host environment.

Figure 7 shows $p_{2g}(v_{esc})$ as a function of the escape velocity $v_{esc}$. We highlight in vertical bars the maximal escape velocities for some particular types of environments, namely Globular Clusters, Nuclear Star Clusters, Elliptical Galaxies and the Milky Way. The left and right panels respectively show results for the primary and secondary BH. The corresponding numerical values of $p_{2g}$ are listed in Table 2. As expected from the previous section, results vary significantly depending on the chosen combination of scenario for GW190521 and prior for the ancestor BHs. Also, in order to highlight the dramatic impact of the inclusion of the recoil in the assessment of the viability of the GW190521 components as second-generation BHs, we note that rightmost value in the panels (i.e., for $v_{\rm esc}$ equal to the maximum kick velocity that NRSur1dq4 can predict) corresponds to the case where only the masses are considered. Once again, we will mostly focus our discussion on the primary BH.

3.4.1 “Ignoring” the birth kick: the infinite escape velocity limit

For high enough escape velocities, the remnant BH is always retained in its environment, so that $p_{2g}$ is simply given by $p(m_{i1},m_{i,2})<65M_{\odot}$. Even in this case, the choice of scenario and ancestor prior dramatically impacts $p_{2g}$. For the quasi-circular scenario, we that $p_{2g}$ is respectively bounded by approximately 0.5, 0.6 and 0.9 under our precessing, APq6 and APq2 priors. The latter prior yields a larger $p_{2g}$ due to the lower ancestral masses it leads to. These probabilities are respectively reduced to 0.1, 0.1 and 0.6 if we assume an eccentric scenario for GW190521. We note again, however, that the aligned-spin APq2 prior fails to reproduce most of the spin posterior for the primary BH of GW190521. Therefore, we conclude that since mass ratios $q\geq 2$ are needed to produce the component BHs of GW190521 if this was indeed eccentric, then the probability that the primary BH is a second-generation one is bounded by $p_{2g}\leq 0.1$.

Figure 7: Viability of the components of GW190521 as the result of a merger of stellar-origin black holes as a function of the escape velocity of the host environment. The left (right) panel corresponds to the primary (secondary) black hole. Solid (dashed) curves correspond to a quasi-circular (eccentric) interpretation for GW190521. Blue (red) curves correspond to our precessing (aligned-spin) prior for the ancestor black holes. We highlight typical maximal values for four different types of host environments, namely Globular Clusters, the Milky Way, a Nuclear Star Clusters and Elliptical Galaxies.

3.4.2 Finite escape velocities: the impact of the birth kick

The inclusion of the birth kick strongly suppresses the viability of certain environments as hosts of GW190521 (assuming that the primary BH is of second generation) as these decrease their escape velocity. For the case of GW190521 being quasi-circular, the results are mostly driven by the choice of ancestor priors, as this determines almost completely birth-kick posterior. This is reflected, for instance, by the fact that the solid-blue curve slowly decreases for $v_{esc}<3000$ km/s, i.e., the maximal kick we can obtain for precessing mergers, while the dashed-blue starts dropping at $v_{esc}<400$ km/s, around the maximal allowed kick for aligned-spins. In contrast, the results under the eccentric scenario are strongly driven by the actual data, namely the strong constrain that the GW190521 spin magnitudes impose on the putative birth kicks, which get restricted to $v_{f}\in\sim[150,350]$ km/s even if the ancestors are allowed to precess.

For a Globular Cluster, we obtain $p_{2g}\sim 0$ in all cases except if GW190521 was quasi-circular and its ancestors had aligned spins, for which we obtain nevertheless small probabilities of $p_{2g}=0.26(0.14)$ when the ancestral mass ratio is respectively restricted to $q\leq 2(6)$. Therefore, if GW190521 happened in a Globular Cluster, its primary BH can only be of second generation if its ancestors had aligned spins. For the remaining environments, we obtain the same $p_{2g}$ as when we assume an infinite $v_{esc}$, except for the case of a quasi-circular GW190521 with precessing ancestors. For the latter, we respectively obtain $p_{2g}=\{0.29,0.46,0.52\}$ for the case of the Milky Way, a Nuclear Star Cluster and an Elliptic Galaxy.

As a summary, for the primary component of GW190521 we obtain three main conclusions:

1. If GW190521 was eccentric and we allow for generic ancestral spins, the chances that its primary BH is a second-generation BH are highly suppressed, with $p_{2g}\leq 0.1$, regardless the host environment. Moreover, mass ratios $q\geq 2$, untypical of isolated formation, are needed to produce the GW190521 component BHs.

2. If GW190521 happened in a Globular Cluster, its primary BH cannot be of second generation, unless GW190521 was quasi-circular and its ancestors had aligned spins. In such a case, we get $p_{2g}=0.26(0.14)$ if we restrict the ancestral mass ratio to $q\leq 2(6)$.

3. We only obtain $p_{2g}>0.5$ if GW190521 was hosted in a Nuclear Star Cluster, an Elliptical Galaxy or a Milky-Way-like Galaxy, provided that its parents had aligned spins. For precessing parents, we obtain $p_{2g}$ near 0.5 for Nuclear Star Clusters and Elliptic Galaxies.

3.4.3 Secondary component of GW190521

For the secondary component of GW190521, we obtain results with a similar qualitative behaviour. The main quantitative difference resides in that $p_{2g}$ is much larger in the infinite escape velocity limit, due to the lower mass estimates for its ancestors. In particular, we always obtain $p_{2g}>0.5$ for all hosts except for a Globular Cluster. For the latter, we obtain again $p_{2g}\simeq 0$ for in all cases, except when we consider aligned-spin ancestors together with a quasi-circular scenario for GW190521, for which we obtain $p_{2g}\simeq 0.3$.

4 Conclusions

We have described a framework to obtain the masses and spins of the putative ancestors of the component merging black observed in gravitational-wave events, together with their corresponding “birth kicks”. Such measurements are of particular relevance in the context of BBH observations involving component masses populating the PISN gap which. In particular, these allow us to assess the viability of such BHs as products of previous mergers as a function of the properties of the host environment. We have applied this framework to the mysterious GW190521 event, whose primary black hole populates the PISN gap. We have done this both under the hypotheses that this was a quasi-circular precessing merger and a non-precessing eccentric merger; and for three different priors for the properties of the ancestral BHs. Next, we have estimated the probabilities that the component black holes of GW190521 are the result of a merger of stellar-origin black holes, as a function of the host environment.

Our estimates of the ancestor properties and the corresponding birth kicks can strongly depend on the combination of the scenario for GW190521 (eccentric or quasi-circular) and on the priors for the properties of the ancestors of its components. Far from denoting lack of robustness, this underlines the potential of our this method to discard some of such combinations, specially if the host environment can be identified. Therefore, our method can drive highly relevant astrophysical conclusions, both about the properties of host environments and on the viability and existence of hierarchical BH formation channels. For instance, our results show that if GW190521 was eccentric, its primary component has less than $10\%$ chance of being born from the merger of stellar-born black-holes; and null chances if the host was a Globular Cluster. Moreover, for such an environment the only viable scenario is that where GW190521 is a quasi-circular merger, with the ancestors of its primary BH having aligned spins. Our results are broadly consistent with those by Kimball et al. (2021b), which concluded that the primary BH of GW190521 has a strong chance to be a second-generation BH if its host escape velocity exceeds $100$ km/s. In addition, our results support that GW190521 originated in an Active Galactic Nuclei (AGN) Ford & McKernan (2022), especially if it was eccentric. On the one hand, the escape velocity of AGNs largely exceed the recoil BBH remnants. On the other hand, orbital eccentricity in GW190521, together with the large mass ratio and small spins needed to produce its components are also characteristic of BBHs in AGNs Ford & McKernan (2022).

Our work is limited by the following main aspects. First, we have obtained final masses, spins and kicks through the NRSur7dq4Remnant model. While this model provides accurate estimates for BBHs with precessing spins, this also forces us to limit the mass ratio of the ancestor BBHs to $q\leq 6$, which in turn imposes constraints not only on the possible final spins, but also on the probability distribution of the kick for a given final spin. Second, while we have considered the case where GW190521 is an eccentric binary, we have worked under the assumption that the ancestor BHs described quasi-circular ones. Relaxing this assumption shall lead to significantly different probability distributions for the final mass and spins of the corresponding remnants and, possibly, to different astrophysical conclusions. Third, we have not considered a scenario for GW190521 combining both precession and eccentricity, as such study would require of fast-to-evaluate waveform models including both effects, which do not yet exist. We note that Gayathri et al. (2022) performed a study combining both effects, finding a good fit to GW190521, using numerical relativity waveforms. However, this study only reports a single point-like parameter estimate for a best-fit numerical simulation with fixed mass ratio, spins and eccentricity, making it unrealistic to estimate how our results would look under such scenario . Fourth, it has been shown that environmental effects can induce changes in the GW emission around the merger and ringdown stages of black-hole mergers Leong et al. (2023), which can in turn modify the kick with respect to the vacuum case Zhang et al. (2023). Finally, while GW190521 is the most confident BBH detection to date with component masses within the PISN gap, less confident events like GW200220 Abbott et al. (2021a) or GW190426 Collaboration & the Virgo Collaboration (2021) pose also suitable candidates for displaying second-generation BHs. In forthcoming studies, we aim to expand our study to further final-parameter models, suitable ancestral BH priors and a larger suite of BBH detections, as these are provided by current and future observing runs.

Our work sets the ground for full parameter inference of ancestral BBHs, highlighting the crucial role of birth kick estimates and how these shall allow to select between combinations of potential ancestor populations and host environments. This, in turn, can help to extract crucial astrophysical information such as the properties of host environments of BBH mergers and the viability of hierarchical formation channels as those giving origin to the high-mass BHs observed by GW detectors. The rather poor accuracy of current spin-magnitude estimates in BBH events significantly limits that of the putative birth kicks, preventing stronger astrophysical conclusions. This situation shall be much improved as the GW detector network progressively improves its sensitivity, particularly with the arrival of third-generation detectors like Einstein Telescope Punturo et al. (2010); Hild et al. (2010) or Cosmic Explorer Reitze et al. (2019); Abbott et al. (2017).

Finally, we note that while finalizing this paper, we learned that P. Mahapatra et. al. Mahapatra et al. (In Preparation) have also looked into the problem of identifying the parents of GW190521 assuming hierarchical merger scenario.

5 acknowledgments

We thank Isobel Romero-Shaw for kindly sharing the posterior distributions for GW190521 under the eccentric scenario and for many useful comments on the manuscript. We thank Thomas Dent, Michella Mapelli, Parthapratim Mahapatra, Christopher Berry, Imre Bartos and Ester Rfor useful comments. We also thank Kiki Wing Lo for useful discussions during the early stages of this study. JCB is funded by a fellowship from “la Caixa” Foundation (ID100010474) and from the European Union’s Horizon2020 research and innovation programme under the Marie Skodowska-Curie grant agreement No 847648. The fellowship code is LCF/BQ/PI20/11760016. JCB is also supported by the research grant PID2020-118635GB-I00 from the Spain-Ministerio de Ciencia e Innovación and by the European Horizon Europe staff exchange (SE) programme HORIZON-MSCA2021-SE-01 Grant No. NewFunFiCO-101086251. We acknowledge the use of IUCAA LDG cluster Sarathi for the computational/numerical work. The authors acknowledge computational resources provided by the CIT cluster of the LIGO Laboratory and supported by National Science Foundation Grants PHY-0757058 and PHY0823459; and the support of the NSF CIT cluster for the provision of computational resources for our parameter inference runs. This material is based upon work supported by NSF’s LIGO Laboratory which is a major facility fully funded by the National Science Foundation. The analysed LIGO-Virgo data and the corresponding power spectral densities, in their strain versions, are publicly available at the online Gravitational-Wave Open Science Center Abbott et al. (2021c); Collaboration et al. (2023). This research has made use of data or software obtained from the Gravitational Wave Open Science Center (gwosc.org), a service of LIGO Laboratory, the LIGO Scientific Collaboration, the Virgo Collaboration, and KAGRA. LIGO Laboratory and Advanced LIGO are funded by the United States National Science Foundation (NSF) as well as the Science and Technology Facilities Council (STFC) of the United Kingdom, the Max-Planck-Society (MPS), and the State of Niedersachsen/Germany for support of the construction of Advanced LIGO and construction and operation of the GEO600 detector. Additional support for Advanced LIGO was provided by the Australian Research Council. Virgo is funded, through the European Gravitational Observatory (EGO), by the French Centre National de Recherche Scientifique (CNRS), the Italian Istituto Nazionale di Fisica Nucleare (INFN) and the Dutch Nikhef, with contributions by institutions from Belgium, Germany, Greece, Hungary, Ireland, Japan, Monaco, Poland, Portugal, Spain. KAGRA is supported by Ministry of Education, Culture, Sports, Science and Technology (MEXT), Japan Society for the Promotion of Science (JSPS) in Japan; National Research Foundation (NRF) and Ministry of Science and ICT (MSIT) in Korea; Academia Sinica (AS) and National Science and Technology Council (NSTC) in Taiwan. This manuscript has LIGO DCC number P2400073.