Final parsec problem
How do pairs of supermassive black holes get close enough to merge?
Summary
Why it is open
In 1980 Begelman, Blandford and Rees divided the life of a pair of supermassive black holes into three phases [2]. The black holes sink to the centre of the merged galaxy by dynamical friction, the drag from the stars they pass, and form a binary; the binary shrinks by flinging out stars that pass close to it; and once the pair is close enough, gravitational waves carry off the remaining energy and the black holes coalesce [2]. The bottleneck is the middle step, because the binary quickly ejects every star on an orbit that reaches it, the set of orbits called the loss cone, and so cuts off its own supply [1, 2].
Miloš Milosavljević and David Merritt estimated that two black holes of 2 × 10⁷ solar masses each become a tight binary at about one parsec, and that to coalesce within a Hubble time, the usual benchmark for the age of the universe, the pair must then eject 10 to 20 times its reduced mass in stars (for equal black holes, half the mass of one) [1]. In a nearly spherical galaxy the loss cone holds at most about one reduced mass, so the binary may stall while still at least ten times too wide; they called the difficulty of shrinking it by a factor of about 100 "the final parsec problem" [1]. Dong Lai and Diego Muñoz place the gap between about one parsec and about 0.01 parsec, where gravitational radiation takes over [14].
Binaries this close "will remain all but impossible" to image directly "for the foreseeable future", except perhaps with very long baseline radio interferometry of nearby galaxies [14]. No body sets a bar for a resolution. The sources point to two kinds of evidence: a mechanism that keeps binaries shrinking in realistic galaxy models [4, 5], and gravitational-wave measurements of how real binaries evolve [11, 14].
What is known
Indirect evidence and candidates. Luminous early-type galaxies often have depleted cores whose missing stellar mass is about one to two times the mass of the central black hole, as expected if a binary ejected those stars [2]. Merritt and Milosavljević also listed circumstantial evidence that coalescence is efficient: X-shaped radio sources, probably galaxies whose black holes recently merged and flipped their jet direction, appear about as often as bright ellipticals are expected to merge, and off-centre black holes, which repeatedly stalled binaries would produce, seem rare [2]. In 2005 they knew of no "completely convincing" bound binary [2]. By 2023 the most compact pair imaged directly, in the radio galaxy 0402+379, had a projected separation of 7.3 parsecs, and searches for periodic brightness changes had produced hundreds of candidates [14].
Simulations of non-spherical galaxies (2011–2017). In 2011 Fazeel Khan, Andreas Just and Merritt simulated galaxy mergers and found binaries hardening much faster than in spherical models, at rates essentially independent of the number of simulated particles [3]. In 2015 Eugene Vasiliev, Fabio Antonini and Merritt pointed out that collisional relaxation, the slow scattering of stars by each other, is much stronger in such simulations than in real galaxies; with a Monte Carlo method, they found that without relaxation only triaxial galaxies, those with three unequal axes, overcome the problem, with coalescence in about a billion years or less [4]. In 2017 Alessia Gualandris and colleagues found the loss cone efficiently refilled even in very mildly triaxial models, with axis ratios 1:0.9:0.8, and concluded "that there is no 'final parsec problem'" [5].
Pulsar timing arrays (2023). A pulsar timing array times many millisecond pulsars for years; a gravitational-wave background shifts their arrival times in a way common to all of them, with a correlation between pulsar pairs that depends on their angular separation, the Hellings–Downs pattern [7, 8, 9]. At the end of June 2023 four collaborations reported evidence of different strength [7, 8, 9, 10]. NANOGrav, timing 67 pulsars over 15 years, found Hellings–Downs correlations at about 3σ to 4σ depending on the test, and a strain amplitude of 2.4 (+0.7/−0.6) × 10⁻¹⁵ at a frequency of one cycle per year [7]. The European array found evidence at a false-alarm probability of about 0.1% in its newer 10.3-year subset but only marginal evidence in its full 24.7-year data set [8]. The Parkes array, with 30 pulsars over 18 years, measured correlations consistent with a background at about 2σ and found the apparent signal strength to vary with time [9]. The Chinese array, using the FAST telescope, reported Hellings–Downs evidence at 4.6σ near 14 nanohertz [10].
What the signal implies and what it does not. NANOGrav found its signal consistent with a population of supermassive black hole binaries, "although more exotic cosmological and astrophysical sources cannot be excluded" [7]. Its population study reproduced the spectrum with astrophysical models, but only with several parameters at the edges of expected values or a few notably different from standard ones; the consistency "offers a tantalizing prospect for confirming that SMBH binaries are able to form, reach subparsec separations, and eventually coalesce" [11]. The binaries that dominate the signal, above about 10⁹ solar masses, must reach about 0.1 parsec through their environment before gravitational waves take over [11]. Models with and without environmental hardening both fit the data [11]; a July 2026 erratum fixing two software bugs found slightly less evidence for environmental interactions, with conclusions unchanged [11]. A companion paper found that several cosmological sources, among them cosmic inflation, phase transitions and domain walls, can also reproduce the signal, and that their better fits "should not be regarded as evidence for new physics" [12]. A search of the same data for individual binaries found no compelling signal [13].
Main approaches
Stars in non-spherical galaxies. Merged galaxies are not spherical, and in triaxial ones stars on orbits through the centre keep refilling the loss cone [3, 4]. Strongest evidence: merger simulations with hardening rates independent of particle number [3], and efficient refilling even in mildly triaxial models [5]. Main difficulty: published results are contradictory, and in isolated non-rotating triaxial models the hardening rate falls as the particle number rises, a sign of spurious relaxation [5]; without relaxation, axisymmetric galaxies are not enough, and the hardening rate stays well below the maximum and declines with time [4].
Gas. A circumbinary disc, a gas disc around both black holes, could drive the binary inward with the accretion flow, as a giant planet migrates in a protoplanetary disc; most cosmological studies assume this [14]. Strongest evidence: simulations find inward migration when the disc's thickness is below about 0.04 of its radius, a transition that "appears to be robust", and discs around supermassive black holes are thought to be about 0.01 or thinner [14]. Main difficulty: in thicker discs, equal-mass and moderately unequal binaries expand, so circumbinary accretion "does not always lead to binary orbital decay" [14]; whether gas pushes binaries inward or outward "is still unclear" [11].
A third black hole. If a binary stalls for longer than the time between galaxy mergers, a third black hole can arrive, speed coalescence and possibly eject the lightest one [11]. Strongest evidence: in simulations of black hole triplets in spherical galaxies, Matteo Bonetti and colleagues found that 20–30% of binaries that would otherwise stall coalesce within a Hubble time, most often after about 300 million years [6]. Main difficulty: in the same simulations most stalled binaries are not rescued this way [6].
Self-interacting dark matter. In 2024 Gonzalo Alonso-Álvarez, James Cline and Caitlyn Dewar argued that friction from the dense spike of dark matter around the black holes bridges the final parsec if dark-matter particles scatter off each other with a cross-section of order one square centimetre per gram; with collisionless cold dark matter the friction disrupts the spike [15]. Strongest evidence: the same effect softens the gravitational-wave spectrum at low frequencies, as the pulsar timing data suggest, with a velocity dependence like one proposed for the small-scale problems of cold dark matter [15]; see Dark matter. Main difficulty: it works only if dark matter self-interacts at that strength [15], and at current precision binary models without it also reproduce the spectrum [11].
Open sub-questions
- Triaxial nuclei. Do the centres of real merger remnants have the mild triaxiality, near axis ratios of 1:0.9:0.8, that collisionless loss-cone refilling needs [4, 5], and which stellar-kinematic data would measure it?
- Low-frequency turnover. Will longer timing data show the flattening at the lowest frequencies that environmental hardening predicts [11], and can its shape separate stellar scattering from dark-matter friction [15]?
- Thin discs. In simulations that resolve the gas around each black hole, does inward migration persist for discs 0.01 of their radius thick or thinner [14]?
- Triples in the background. What share of the background could come from binaries merged by triple interactions, and would their small eccentricities near plunge, up to about 0.1, be measurable [6]?
- Masses behind the amplitude. The background is dominated by the most massive binaries and needs population parameters at the edge of expectations [11]; do black hole mass functions that fit how the first black holes grew allow it? See Early supermassive black holes.
- Individual binaries. Which periodic candidates from time-domain surveys [14] fall inside the volumes that pulsar-timing searches already exclude [13]?
Work on this in Papex
If you have a testable answer to one of the sub-questions above, you can draft it as a private hypothesis in Papex, or prepare the draft from Claude or ChatGPT with the Papex connector. You review the exact version, sign it and decide whether to publish it; a public record that names this problem is listed on this page. A Papex record shows who signed which version and when. It is not peer review and does not mean the hypothesis is correct.
Key references
- Milosavljević, M. & Merritt, D. (2003). The final parsec problem. In J. Centrella (ed.), The Astrophysics of Gravitational Wave Sources, AIP Conference Proceedings 686, 201–210. 10.1063/1.1629432 · Checked 15 Oct 2026
- Merritt, D. & Milosavljević, M. (2005). Massive black hole binary evolution. Living Reviews in Relativity 8, 8. 10.12942/lrr-2005-8 · Checked 15 Oct 2026
- Khan, F. M., Just, A. & Merritt, D. (2011). Efficient merger of binary supermassive black holes in merging galaxies. The Astrophysical Journal 732(2), 89. 10.1088/0004-637X/732/2/89 · Checked 15 Oct 2026
- Vasiliev, E., Antonini, F. & Merritt, D. (2015). The final-parsec problem in the collisionless limit. The Astrophysical Journal 810(1), 49. 10.1088/0004-637X/810/1/49 · Checked 15 Oct 2026
- Gualandris, A., Read, J. I., Dehnen, W. & Bortolas, E. (2017). Collisionless loss-cone refilling: there is no final parsec problem. Monthly Notices of the Royal Astronomical Society 464(2), 2301–2310 (published online 2016). 10.1093/mnras/stw2528 · Checked 15 Oct 2026
- Bonetti, M., Haardt, F., Sesana, A. & Barausse, E. (2018). Post-Newtonian evolution of massive black hole triplets in galactic nuclei – II. Survey of the parameter space. Monthly Notices of the Royal Astronomical Society 477(3), 3910–3926. 10.1093/mnras/sty896 · Checked 15 Oct 2026
- Agazie, G., Anumarlapudi, A., Archibald, A. M., Arzoumanian, Z., Baker, P. T., Bécsy, B. et al. (NANOGrav) (2023). The NANOGrav 15 yr data set: evidence for a gravitational-wave background. The Astrophysical Journal Letters 951(1), L8. 10.3847/2041-8213/acdac6 · Checked 15 Oct 2026
- EPTA Collaboration and InPTA Collaboration: Antoniadis, J., Arumugam, P., Arumugam, S., Babak, S., Bagchi, M. et al. (2023). The second data release from the European Pulsar Timing Array. III. Search for gravitational wave signals. Astronomy & Astrophysics 678, A50. 10.1051/0004-6361/202346844 · Checked 15 Oct 2026
- Reardon, D. J., Zic, A., Shannon, R. M., Hobbs, G. B., Bailes, M., Di Marco, V. et al. (2023). Search for an isotropic gravitational-wave background with the Parkes Pulsar Timing Array. The Astrophysical Journal Letters 951(1), L6. 10.3847/2041-8213/acdd02 · Checked 15 Oct 2026
- Xu, H., Chen, S., Guo, Y., Jiang, J., Wang, B., Xu, J. et al. (2023). Searching for the nano-hertz stochastic gravitational wave background with the Chinese Pulsar Timing Array Data Release I. Research in Astronomy and Astrophysics 23(7), 075024. 10.1088/1674-4527/acdfa5 · Checked 15 Oct 2026
- Agazie, G., Anumarlapudi, A., Archibald, A. M., Baker, P. T., Bécsy, B., Blecha, L. et al. (NANOGrav) (2023). The NANOGrav 15 yr data set: constraints on supermassive black hole binaries from the gravitational-wave background. The Astrophysical Journal Letters 952(2), L37; erratum in 1006(2), L67 (2026), DOI 10.3847/2041-8213/ae8764. 10.3847/2041-8213/ace18b · Checked 15 Oct 2026
- Afzal, A., Agazie, G., Anumarlapudi, A., Archibald, A. M., Arzoumanian, Z., Baker, P. T. et al. (NANOGrav) (2023). The NANOGrav 15 yr data set: search for signals from new physics. The Astrophysical Journal Letters 951(1), L11; erratum in 971(1), L27 (2024), DOI 10.3847/2041-8213/ad68fc. 10.3847/2041-8213/acdc91 · Checked 15 Oct 2026
- Agazie, G., Anumarlapudi, A., Archibald, A. M., Arzoumanian, Z., Baker, P. T., Bécsy, B. et al. (NANOGrav) (2023). The NANOGrav 15 yr data set: Bayesian limits on gravitational waves from individual supermassive black hole binaries. The Astrophysical Journal Letters 951(2), L50. 10.3847/2041-8213/ace18a · Checked 15 Oct 2026
- Lai, D. & Muñoz, D. J. (2023). Circumbinary accretion: from binary stars to massive binary black holes. Annual Review of Astronomy and Astrophysics 61, 517–560. 10.1146/annurev-astro-052622-022933 · Checked 15 Oct 2026
- Alonso-Álvarez, G., Cline, J. M. & Dewar, C. (2024). Self-interacting dark matter solves the final parsec problem of supermassive black hole mergers. Physical Review Letters 133(2), 021401. 10.1103/PhysRevLett.133.021401 · Checked 15 Oct 2026
Source
Open, as listed by Wikipedia · Unsolved problems in astronomy. Checked 29 Sep 2026.
The status is the source’s. Papex does not decide whether a problem is solved.
Records on this problem
No public record addresses this problem yet.