Fast inflows as the adjacent fuel of supermassive black hole accretion disks in quasars


Quasars, which are exceptionally bright objects at the centres (or nuclei) of galaxies, are thought to be produced through the accretion of gas into disks surrounding supermassive black holes1,2,3. There is observational evidence at galactic and circumnuclear scales4 that gas flows inwards towards accretion disks around black holes, and such an inflow has been measured at the scale of the dusty torus that surrounds the central accretion disk5. At even smaller scales, inflows close to an accretion disk have been suggested to explain the results of recent modelling of the response of gaseous broad emission lines to continuum variations6,7. However, unambiguous observations of inflows that actually reach accretion disks have been elusive. Here we report the detection of redshifted broad absorption lines of hydrogen and helium atoms in a sample of quasars. The lines show broad ranges of Doppler velocities that extend continuously from zero to redshifts as high as about 5,000 kilometres per second. We interpret this as the inward motion of gases at velocities comparable to freefall speeds close to the black hole, constraining the fastest infalling gas to within 10,000 gravitational radii of the black hole (the gravitational radius being the gravitational constant multiplied by the object mass, divided by the speed of light squared). Extensive photoionization modelling yields a characteristic radial distance of the inflow of approximately 1,000 gravitational radii, possibly overlapping with the outer accretion disk.

Access options

Rent or Buy article

Get time limited or full article access on ReadCube.


All prices are NET prices.

Fig. 1: Hydrogen and helium absorption-line spectrum of the quasar JI035 + 1422.
Fig. 2: Close-up of the absorption spectrum of J1035 + 1422 in selective hydrogen, helium and metal lines.
Fig. 3: Probability density distribution in the space of the total hydrogen density and the characteristic radial distance of the inflow in JI035 + 1422.
Fig. 4: Schematic view of the central engine of quasars, with an inflow reflecting our data on redshifted hydrogen and helium broad absorption lines.

Data availability

All observational data that support the findings of this study are available from the corresponding author on request.

Code availability

The custom computer code used to analyse and model the observational data here is available from the corresponding author on request, for the purpose of repeating the published results only.


  1. 1.

    Peterson, B. M. An Introduction to Active Galactic Nuclei (Cambridge Univ. Press, 1997).

  2. 2.

    Krolik, J. H. Active Galactic Nuclei: From the Central black hole to the Galactic Environment (Princeton Univ. Press, 1999).

  3. 3.

    Netzer, H. The Physics and Evolution of Active Galactic Nuclei (Cambridge Univ. Press, 2013).

  4. 4.

    Storchi-Bergmann, T. & Schnorr-Müller, A. Observational constraints on the feeding of super- massive black holes. Nature Astron. 3, 48–61 (2019).

    ADS  Article  Google Scholar 

  5. 5.

    Shi, X.-H. et al. The redshifted hydrogen Balmer and metastable He i absorption line system in mini-FeLoBAL quasar SDSS J112526.12 + 002901.3: a parsec-scale accretion inflow? Astrophys. J. 829, 96–106 (2016).

    ADS  Article  Google Scholar 

  6. 6.

    Pancoast, A. et al. Modelling reverberation mapping data—II. Dynamical modelling of the Lick AGN monitoring project 2008 data set. Mon. Not. R. Astron. Soc. 445, 3073–3091 (2014).

    ADS  CAS  Article  Google Scholar 

  7. 7.

    Grier, C. J. et al. The structure of the broad-line region in active galactic nuclei. II. Dynamical modeling of data from the AGN10 reverberation mapping campaign. Astrophys. J. 849, 146–163 (2017).

    ADS  Article  Google Scholar 

  8. 8.

    Weymann, R. J., Morris, S. L., Foltz, C. B. & Hewett, P. C. Comparisons of the emission-line and continuum properties of broad absorption line and normal quasi-stellar objects. Astrophys. J. 373, 23–53 (1991).

    ADS  CAS  Article  Google Scholar 

  9. 9.

    Rubin, K. H. R. in Gas Accretion onto Galaxies Vol. 430 (eds Fox, A. & Davé, R.) 95−116 (Springer, 2017).

  10. 10.

    Leighly, K. M., Dietrich, M. & Barber, S. The discovery of the first He i λ10830 broad absorption line quasar. Astrophys. J. 728, 94–118 (2011).

    ADS  Article  Google Scholar 

  11. 11.

    Hall, P. B. et al. Broad absorption line quasars with redshifted troughs: high-velocity infall or rotationally dominated outflows? Mon. Not. R. Astron. Soc. 434, 222–256 (2013).

    ADS  CAS  Article  Google Scholar 

  12. 12.

    Hall, P. B. A quasar with broad absorption in the Balmer lines. Astron. J. 133, 1271–1274 (2007).

    ADS  CAS  Article  Google Scholar 

  13. 13.

    Ji, T. et al. Unshifted metastable He i* mini-broad absorption line system in the narrow-line type 1 quasar SDSS J080248.18 + 551328.9. Astrophys. J. 800, 56–72 (2015).

    ADS  Article  Google Scholar 

  14. 14.

    Zhang, S. et al. Discovery of extremely broad Balmer absorption lines in SDSS J152350.42 + 391405.2. Astrophys. J. 815, 113–123 (2015).

    ADS  Article  Google Scholar 

  15. 15.

    Liu, W.-J. et al. A comprehensive study of broad absorption line quasars. I. Prevalence of He i* absorption line multiplets in low-ionization objects. Astrophys. J. Suppl. Ser. 217, 11–46 (2015).

    ADS  Article  Google Scholar 

  16. 16.

    Sun, L. et al. Photoionization-driven absorption-line variability in Balmer absorption line quasar LBQS 1206 + 1052. Astrophys. J. 838, 88–109 (2017).

    ADS  Article  Google Scholar 

  17. 17.

    Shi, X.-H. et al. Discovery of variable hydrogen Balmer absorption lines with inverse decrement in PG 1411 + 442. Astrophys. J. 843, L14–L17 (2017).

    ADS  Article  Google Scholar 

  18. 18.

    Pâris, I. et al. The Sloan Digital Sky Survey quasar catalog: fourteenth data release. Astron. Astrophys. 613, A51–A67 (2018).

    ADS  Article  Google Scholar 

  19. 19.

    Cicone, C. et al. Massive molecular outflows and evidence for AGN feedback from CO observations. Astron. Astrophys. 562, A21–A45 (2014).

    Article  Google Scholar 

  20. 20.

    Rudy, R. J. Echelle spectroscopy of the Seyfert 1 galaxy Markarian 231. Astrophys. J. 288, 531–534 (1985).

    ADS  CAS  Article  Google Scholar 

  21. 21.

    Kaspi, S. et al. The relationship between luminosity and broad-line region size in active galactic nuclei. Astrophys. J. 629, 61–71 (2005).

    ADS  CAS  Article  Google Scholar 

  22. 22.

    Greene, J. E. & Ho, L. C. Estimating black hole masses in active galaxies using the Hα emission line. Astrophys. J. 630, 122–129 (2005).

    ADS  CAS  Article  Google Scholar 

  23. 23.

    Netzer, H. Revisiting the unified model of active galactic nuclei. Annu. Rev. Astron. Astrophys. 53, 365–408 (2015).

    ADS  CAS  Article  Google Scholar 

  24. 24.

    Collin, S. & Zahn, J.-P. Star formation and evolution in accretion disks around massive black holes. Astron. Astrophys. 344, 433–449 (1999).

    ADS  Google Scholar 

  25. 25.

    Wang, J.-M. et al. Tidally disrupted dusty clumps as the origin of broad emission lines in active galactic nuclei. Nature Astron. 1, 775–783 (2017).

    ADS  Article  Google Scholar 

  26. 26.

    Aalto, S. et al. The hidden heart of the luminous infrared galaxy IC 860. I. A molecular inflow feeding opaque, extreme nuclear activity. Astron. Astrophys. 627, A147 (2019).

    Article  Google Scholar 

  27. 27.

    Gaskell, C. M. & Goosmann, R. W. Line shifts, broad-line region inflow, and the feeding of active galactic nuclei. Astrophys. J. 769, 30–39 (2013).

    ADS  Article  Google Scholar 

  28. 28.

    Elvis, M. Quasar rain: the broad emission line region as condensations in the warm accretion disk wind. Astrophys. J. 847, 56–67 (2017).

    ADS  Article  Google Scholar 

  29. 29.

    Ferland, G. J. et al. The 2017 Release of Cloudy. Rev. Mex. Astron. Astrofis. 53, 385–438 (2017).

    ADS  CAS  Google Scholar 

  30. 30.

    Cushing, M. C., Vacca, W. D. & Rayner, J. T. Spextool: a spectral extraction package for SpeX, a 0.8-5.5 micron cross-dispersed spectrograph. Publ. Astron. Soc. Pacif. 116, 362–376 (2004).

    ADS  Article  Google Scholar 

  31. 31.

    Vanden Berk, D. E. et al. Composite quasar spectra from the Sloan Digital Sky Survey. Astron. J. 122, 549–564 (2001).

    ADS  Article  Google Scholar 

  32. 32.

    Gordon, K. D., Clayton, G. C., Misselt, K. A., Landolt, A. U. & Wolff, M. J. A quantitative comparison of the small magellanic cloud, large magellanic cloud, and Milky Way ultraviolet to near-infrared extinction curves. Astrophys. J. 594, 279–293 (2003).

    ADS  CAS  Article  Google Scholar 

  33. 33.

    Eracleous, M. & Halpern, J. P. Completion of a survey and detailed study of double-peaked emission lines in radio-loud active galactic nuclei. Astrophys. J. 599, 886–908 (2003).

    ADS  CAS  Article  Google Scholar 

  34. 34.

    Wu, X.-B. & Liu, F. K. Black hole mass and accretion rate of active galactic nuclei with double-peaked broad emission lines. Astrophys. J. 614, 91–100 (2004).

    ADS  Article  Google Scholar 

  35. 35.

    Lewis, K. T. & Eracleous, M. Black hole masses of active galaxies with double-peaked Balmer emission lines. Astrophys. J. 642, 711–719 (2006).

    ADS  CAS  Article  Google Scholar 

  36. 36.

    Runnoe, J. C., Brotherton, M. S. & Shang, Z. Updating quasar bolometric luminosity corrections. Mon. Not. R. Astron. Soc. 422, 478–493 (2012).

    ADS  Article  Google Scholar 

  37. 37.

    Zhang, S. et al. Outflow and hot dust emission in broad absorption line quasars. Astrophys. J. 786, 42 (2014).

    ADS  Article  Google Scholar 

  38. 38.

    Barvainis, R. Hot dust and the near-infrared bump in the continuum spectra of quasars and active galactic nuclei. Astrophys. J. 320, 537–544 (1987).

    ADS  Article  Google Scholar 

  39. 39.

    Brandt, W. N., Laor, A. & Wills, B. J. On the nature of soft X-ray weak quasi-stellar objects. Astrophys. J. 528, 637–649 (2000).

    ADS  Article  Google Scholar 

  40. 40.

    Green, P. J., Aldcroft, T. L., Mathur, S., Wilkes, B. J. & Elvis, M. A Chandra survey of broad absorption line quasars. Astrophys. J. 558, 109–118 (2001).

    ADS  CAS  Article  Google Scholar 

  41. 41.

    Gibson, R. R. et al. A catalog of broad absorption line quasars in Sloan Digital Sky Survey data release 5. Astrophys. J. 692, 758–777 (2009).

    ADS  CAS  Article  Google Scholar 

  42. 42.

    Glikman, E., Helfand, D. J. & White, R. L. A near-infrared spectral template for quasars. Astrophys. J. 640, 579–591 (2006).

    ADS  CAS  Article  Google Scholar 

  43. 43.

    Eracleous, M. et al. The ultraviolet spectra of active galaxies with double-peaked emission lines. Proc. Int. Astron. Union 222, 29–32 (2004).

    ADS  Article  Google Scholar 

  44. 44.

    Dai, X., Shankar, F. & Sivakoff, G. R. The intrinsic fractions and radio properties of low-ionization broad absorption line quasars. Astrophys. J. 757, 180–189 (2012).

    ADS  Article  Google Scholar 

  45. 45.

    Halpern, J. P. Line emission from another relativistic accretion disk: 3C 332. Astrophys. J. Lett. 365, 51–54 (1990).

    ADS  CAS  Article  Google Scholar 

  46. 46.

    Eracleous, M. & Halpern, J. P. Double-peaked emission lines in active galactic nuclei. Astrophys. J. Suppl. Ser. 90, 1–30 (1994).

  47. 47.

    Wang, H., Zhou, H., Yuan, W. & Wang, T. Metallicity and quasar outflows. Astrophys. J. Lett. 751, 23–27 (2012).

Download references


We acknowledge the use of the Hale 200-inch Telescope at Palomar Observatory through the Telescope Access Program (TAP), which made these observations possible. Funding for the third Sloan Digital Sky Survey (SDSS-III; has been provided by the Alfred P. Sloan Foundation, the participating institutions, the US National Science Foundation and the US Department of Energy Office of Science. H.Z., L.S. and X.P. acknowledge support from the National Natural Science Foundation of China (NSFC11473025) and the SOC programme (CHINARE2017-02-03). X.C. and G. Li are grateful for support from the Key Research Program of the Chinese Academy of Sciences (XDPB09-02).

Author information




H.Z. conceived the project. X.P. led the data acquisition and reduction, with J.G., T.J., P.J. and Z.Z. X.S. analysed the data and performed the photoionization simulations, with W.L., H.L. and G. Li helping with data analysis. W.Y., X.S., L.H., X.C. and G. Liu contributed to writing the manuscript. B.L., J.S., X.S., L.S., Q.T., H.W., T.W., S.W., C.Y. and S.Z. provided critical feedback and helped to shape the manuscript.

Corresponding author

Correspondence to Hongyan Zhou.

Ethics declarations

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Peer review information Nature thanks Thaisa Storchi-Bergmann and Francesco Tombesi for their contribution to the peer review of this work.

Extended data figures and tables

Extended Data Fig. 1 Sensitive range of broad absorption lines in H i* Balmer series and metastable He i*multiplets in gas near a black hole.

Assuming a black hole of mass M = 109Mʘ, accreting at an Eddington ratio of 0.1, the column densities of H0 at the n = 2 level \({N}_{{\rm{col}}}\left({{\rm{H}}}_{n=2}^{0}\right)\) and He0 at the metastable level 23S (Ncol(He023S)) can be evaluated through photoionization simulations for gas of various densities (total hydrogen density; nH), total column density (NH) and distance (in units of Rg). Assuming a Gaussian velocity dispersion (FWHM = 3,000 km s−1), a specified line is considered to be sensitive in measuring the ionic column density as long as the optical depth is in the range 0.05–3 at the line centre. The coloured areas show the sensitive ranges for only three lines in each series for clarity: af, Hα, Hδ and Hκ at λrest = 6,564, 4,102 and 3,750 Å; and gl, \({{\rm{H}}{\rm{e}}\,{\rm{I}}}_{6,4,3}^{\ast }\)≡ He i* λλ2,829, 3,889, 10,830.

Extended Data Fig. 2 Optical spectra for our sample of quasars with redshifted H i* and He i*broad absorption lines.

We plot the SDSS-observed spectra of seven quasars with pure redshifted broad absorption lines (or mini-broad ones) in the H i* Balmer and metastable He i* series in their rest frames. The systemic redshifts are determined from narrow emission lines including [O ii] (grey dotted-dashed vertical lines). As in Fig. 1, the blue dash-dotted lines mark the rest wavelengths of the H i* and He i* transitions. The wavelength ranges of the absorption lines are shaded in orange.

Extended Data Fig. 3 Close-ups of selective hydrogen lines from the absorption lines.

We plot selective H i* Balmer absorption lines (orange) from the SDSS spectra of the seven quasars in Extended Data Fig. 2 in their common velocity space. The data are normalized by the continuum after subtracting the best-fit emission-line models (as for J1035 + 1422 in Fig. 2). The error bars denote 1σ flux uncertainties. ad, Four bona fide broad-absorption-line quasars with absorption troughs spanning a wide range of velocities from roughly 0 to 4,000 km s−1 (much wider than the criterion for definition of a broad absorption line, that is, 2,000 km s−1). eg, The absorption troughs of the remaining three quasars have widths of roughly 1,000−2,000 km s−1 and are formally classified as ‘mini-broad absorption lines’.

Extended Data Fig. 4 Velocity structure of the redshifted Hα broad absorption line of J1035 + 1422.

The optical depth (red) and local covering factor (blue) of the line, derived from the continuum-normalized spectrum, are shown as a function of velocity shift with respect to the quasar’s rest frame. The error bars represent 1σ uncertainties.

Extended Data Fig. 5 Probability density distribution in the parameter space of the total hydrogen density (nH) and distance from the central engine (dinflow) for different inflow models.

a, The simplest primordial models are applied, and the redshifted H i* and He i* broad absorption lines are used to evaluate the probability density. However, the highly probable models predict much higher column densities of C3+ ions (\({N}_{{\rm{col}}}\left({{\rm{C}}}_{{\rm{ground}}}^{3+}\right)\)) than that estimated from the redshifted C iv broad-absorption-line trough. b, To resolve this problem, only the region beyond the C3+ region (the ‘post-C3+ region’) in the primordial models is used to describe the inflow gas, and probability density is recalculated by including C iv. The results of these refined model calculations are shown here. The probability density shows two peaks around nH ≈ 107 cm−3 and dinflow ≈ 2,000Rg, and nH ≈ 109.5 cm−3 and dinflow ≈ 100Rg (see also Fig. 3).

Extended Data Fig. 6 Comparison between observation and model calculations for selected metal absorption lines.

a, Recovered spectra of C iv, corrected for redshifted absorption, assuming C3+ ion column densities of \({N}_{{\rm{col}}}\left({{\rm{C}}}_{{\rm{ground}}}^{3+}\right)\) = 1015, 1016 and 1017 cm−2 in the quasar’s rest frame. The error bars on the observed flux denote 1σ uncertainty. Compared with the best-fit SDSS composite spectrum (blue dashed line), the recovered flux is much too weak for the absorption when \({N}_{{\rm{col}}}\left({{\rm{C}}}_{{\rm{ground}}}^{3+}\right)\) = 1015 cm−2, while it is too high (showing two extra deceptive peaks at around 1,550 and 1,570 Å) for the absorption when \({N}_{{\rm{col}}}\left({{\rm{C}}}_{{\rm{ground}}}^{3+}\right)\)  = 1017 cm−2. The absorption model with \({N}_{{\rm{col}}}\left({{\rm{C}}}_{{\rm{ground}}}^{3+}\right)\) ≈ 1016 cm−2 predicts unabsorbed flux that is reasonably consistent with the composite spectrum, and is thus adopted. b, Absorption-corrected UV Fe ii spectra between 2,000 and 2,750 Å for the post-C3+ inflow models with nH = 107  cm−3 (red), 107.5 cm−3 (yellow), 109 cm−3 (green), and 109.5 cm−3 (violet) in the high ‘probability’ zone of Extended Data Fig. 5b. The error bars on observed flux are 1σ uncertainties. Compared with the best-fit composite (blue dashed line), the models with higher densities can be clearly ruled out.

Extended Data Fig. 7 Photoionization models for inflow/outflow and the transmitted spectral energy distributions.

a, The ionization structure of a primordial model with nH = 107 cm−3 and U = 100.5 that is directly illuminated by the central continuum source of the quasar. If integrated from the illuminated surface, the model with \({N}_{{\rm{c}}{\rm{o}}{\rm{l}}}({{\rm{H}}}_{n=2}^{0})\) and \({N}_{{\rm{col}}}\left({{\rm{He}}}^{0}{2}^{3}{\rm{S}}\right)\) values that are comparable to the measurements predicts \({N}_{{\rm{col}}}\left({{\rm{C}}}_{{\rm{ground}}}^{3+}\right)\) values of more than 1019 cm−2, far from the estimated \({N}_{{\rm{col}}}\left({{\rm{C}}}_{{\rm{ground}}}^{3+}\right)\) in the redshifted broad absorption line. An alternative solution is that the inflow in fact corresponds to the grey area, which is gas behind the C3+ region (the light green area, where C3+ and other high-ionization ions dominate). In such a picture, the outflow is suggested to play an equivalent role to the C3+ region in this panel in eliminating high-energy ionizing photons. b, Plot of the ionization structure for an outflow with nH = 109.5 cm−3 and U = 100.5. The requirement for transmitted radiation (which should have the same spectral energy distribution as the incident radiation on inflow) could constrain the thickness of outflow model. The outer surface of this model (red dashed line) coincides highly with the extension of the C3+ region. However, Ncol(He023S), measured using the blueshifted He i* λ10,830, defines a thinner outflow gas (blue dashed line) if we assume that the local covering factor, Cf, is wavelength independent. c, The transmitted spectral energy distributions through the spectral-energy-distribution-constrained outflow and the Ncol(He023S)-defined outflow are plotted. The former (red) naturally coincides with the incident spectral energy distribution for the inflow model, while the latter (blue) shows considerable excess in soft X-ray, which would result in a much larger \({N}_{{\rm{col}}}\left({{\rm{C}}}_{{\rm{ground}}}^{3+}\right)\) in the inflow than the measurement. d, Transmitted spectral energy distributions through a Ncol(He023S)-defined outflow model with nH = 109.5 cm−3, U = 100.5 and different metallicities. The spectral-energy distribution depends sensitively on the metallicity. As the metallicity increases from 1Zʘ to 10Zʘ, the transmitted spectral energy distribution seems to match the incident spectral energy distribution required by the inflow model, suggesting that a metal-rich outflow model could explain the measurement in both the redshifted and the blueshifted broad-absorption-line systems.

Extended Data Fig. 8 Comparison of the Hα emission line in observations and simple model calculations.

a, Observed (black line) and absorption-corrected (orange dotted line) broad Hα emission line spectra of J1035 + 1422, normalized to the continuum. Overplotted for comparison are the Hα emission lines predicted by photoionization models from the inflow (red dashed line), the outflow (blue dashed line), and both (green dotted line) (assuming a covering factor of 0.5 (Ωi = Ω0= 0.5 in Supplementary Table 1) and a radial velocity 5,000 km s−1 for both inflow and outflow). Clearly, the observed Hα is much stronger than the model prediction. This may be due to the oversimplicity of the models, in which a much broader velocity range is missing. Alternatively, and more likely, the excess Hα flux might be contributed by the accretion disk. b, The residual line profile (cyan dotted line; the zigzag shape is caused by the oversimplified model assumption of a single velocity instead of the actual large velocity gradient), which largely resembles the Hα line observed in the well studied disk-emitting quasar 3C332 (ref. 43; grey solid line). Note that 3C332 shows a substantial excess component with respect to the best-fit disk line model (violet dashed line), which is redshifted with a velocity range of roughly 0−5,000 km s−1. This is reminiscent of the redshifted Hα broad absorption line found in J1035 + 1422 here, suggesting the interesting possibility that this excess Hα emission in 3C332 might originate from inflows.

Supplementary information

Supplementary Table 1

Predicted emission line equivalent widths from the outflow and inflow of J1035+1422 in comparison with the averaged broad-emission-line equivalent widths of quasars.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Zhou, H., Shi, X., Yuan, W. et al. Fast inflows as the adjacent fuel of supermassive black hole accretion disks in quasars. Nature 573, 83–86 (2019).

Download citation

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.


Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing