GA-NIFS: A massive black hole in a low-metallicity AGN at $z\sim5.55$ revealed by JWST/NIRSpec IFS

Abstract

We present JWST/NIRSpec Integral Field Spectrograph rest-frame optical data of the compact $z=5.55$ galaxy GS_3073. Its prominent broad components in several hydrogen and helium lines (while absent in the forbidden lines), and the detection of a large equivalent width of He II$\lambda4686$, EW(He II) ~ 20A, unambiguously identify it as an active galactic nucleus (AGN). We measure a gas-phase metallicity of $Z_{\rm gas}/Z_\odot\sim0.21^{+0.08}_{-0.04}$, lower than what has been inferred for both more luminous AGN at similar redshift and lower redshift AGN. We empirically show that classical emission line ratio diagnostic diagrams cannot be used to distinguish between the primary ionisation source (AGN or star formation) for such low-metallicity systems, whereas different diagnostic diagrams involving He II$\lambda4686$ prove very useful, independent of metallicity. We measure the central black hole mass to be $\log(M_{\rm BH}/M_\odot)\sim8.2\pm0.4$ based on the luminosity and width of the broad line region of the H$α$ emission. While this places GS_3073 at the lower end of known high-redshift black hole masses, it still appears to be over-massive compared to its host galaxy properties. We detect an outflow with projected velocity $\gtrsim700$ km/s and infer an ionised gas mass outflow rate of about $100\ M_\odot/$yr, suggesting that GS_3073 is able to enrich the intergalactic medium with metals one billion years after the Big Bang.

Figures (10)

• Figure 1: Integrated spectrum extracted from the central three by three spaxels in the wavelength range $2.86\mu m < \lambda < 4.85\mu m$ with flux in linear scale (top) and log scale (bottom). Several emission lines are present and indicated by vertical lines at the top. We detect seven He i lines, He ii$\lambda4686$, H$\beta$, [O iii]$\lambda\lambda4959,5007$, H$\alpha$, [N ii]$\lambda\lambda6548,6583$, and [S ii]$\lambda\lambda6716,6731$. We also report the detection of [Ar iv]$\lambda4711$, [Ar iv]$\lambda4740$ and [Ar iii]$\lambda7136$ (note that [Ar iv]$\lambda4711$ is blended with He i$\lambda4713$). The redshifted auroral line [O iii]$\lambda4363$ is only partly covered by the spectral band of our observation with the G395H grating. [O i]$\lambda6003$ falls into the detector gap masked here in the region $4.06\mu m < \lambda < 4.14\mu m$. BLR components are present in H$\beta$, H$\alpha$, He ii, and the He i lines. In addition, an outflow component is present, best visible in the broadened, asymmetric line base of the [O iii] doublet. We indicate the positions of possible coronal lines [Fe xvi], [Ca v], [Fe xiii], [Fe v], of the auroral line [N ii]$\lambda5755$, and of another line at $\lambda\sim7167.5$Å the position of which is consistent with Si i, as grey dotted vertical lines in the bottom panel.
• Figure 2: Zoom-in on the integrated spectrum extracted from the central three by three spaxels including our best fit (blue) with individual components for the emission lines (narrow: green, outflow: orange, BLR: purple) for the wavelength range including He ii$\lambda4686$, [Ar iv]$\lambda4711$, He i$\lambda4713$, [Ar iv]$\lambda4740$, H$\beta$, He i$\lambda4922$, [O iii]$\lambda4959$, and [O iii]$\lambda5007$ (left) and the wavelength range including [N ii]$\lambda6548$, H$\alpha$, [N ii]$\lambda6584$, He i$\lambda6678$, [S ii]$\lambda6716$, and [S ii]$\lambda6731$ (right). The bottom panels show the residuals, res. = data -- best fit. Note that for some of the weaker lines no BLR or outflow component is preferred by the fit. We further note a faint flux excess red-wards of [O iii]$\lambda5007$ that is undetected in individual spaxels, potentially suggesting some higher velocity (nuclear) outflow components not captured by our fiducial fit.
• Figure 3: Top and middle row: projected maps of flux (left), velocity (middle), and velocity dispersion corrected for instrumental resolution (right) as measured from the narrow [Oiii] component (top) and the narrow H$\alpha$ component (middle) tracing the host galaxy kinematics. Bottom: projected maps of flux (left), V10, the velocity at the 10$^{th}$ percentile of the emission-line profile (middle), and V90, the velocity at the 90$^{th}$ percentile (right), from the outflow [Oiii] component. North is up and East is to the left. The spaxel size is $0.1\arcsec$. A bar in the first column indicates $1\arcsec$, corresponding to roughly 6 kpc at the source redshift. The highest projected velocities are found in the East-South-East region in both the narrow H$\alpha$ and [O iii] velocity maps, and are possibly related to an ongoing interaction with a faint companion (see Appendix \ref{['a:r100_comp']}). Apart from this region, a velocity gradient of $\Delta v\sim70$ km/s along the North-North-West to South-South-East direction is visible in both the narrow [Oiii] and H$\alpha$, possibly indicative of rotation. There is a small offset between the integrated [Oiii] and H$\alpha$ line centroids of about 8 km/s. We find elevated velocity dispersions in the galaxy centre, as expected for observations of a rotating disc affected by beam-smearing. The outflow is visible in the nuclear region, with positive and negative velocities respectively the systemic velocity of |600-700| km/s.
• Figure 4: Central black hole mass $M_{{\rm BH}}$ as a function of host galaxy stellar mass $M_{\rm star}$ (left) and host galaxy dynamical mass $M_{\rm dyn}$ (right) for GS_3073 (filled blue star) and literature compilations. In the left panel, we compare our galaxy to local AGN by Reines15 for which the black hole mass has been determined from a broad line region or from reverberation mapping (red circles, with representative error bar in grey). The black line with shaded region shows the best fit with uncertainties to this data by Reines15, with the dashed lines indicating intrinsic scatter plus measurement uncertainties. As teal diamonds we show two data points by Kocevski23, for which we calculate the black hole mass based on Eq. \ref{['eq:mbh']} for consistency with our measurement and the $z=0$ data. In the right panel, we compare our galaxy to $z\geq6$ QSOs by Izumi21 (green diamonds). For our dynamical mass estimate, we adopt Sérsic index $n_S=4$ and effective radius $R_e=0.18$ kpc, and indicate uncertainties by varying $R_e$ between 0.11 and 0.9 kpc. The black line shows the local $M_{\rm BH}-M_{\rm bulge}$ relation with uncertainties and intrinsic scatter by Kormendy13. The black hole of GS_3073 is at the lower end of the $M_{\rm BH}-M_{\rm dyn}$ distribution when compared to the measurements by Izumi21, but still appears over-massive for its host galaxy dynamical mass compared to the local relation by Kormendy13. Considering the $M_{\rm BH}-M_{\rm star}$ relation, the black hole of GS_3073 is much more massive compared to local broad line AGN with a similar host galaxy stellar mass.
• Figure 5: GS_3073 in the mass-metallicity plane (blue star). The $z\sim0.08$ relation based on SDSS data by Curti20 is shown as grey shading with the best fit in black. The red and yellow dash-dotted lines show the best-fit relations at $z\sim2.3$ and $z\sim3.3$, respectively, by Sanders21 based on data from the MOSDEF survey Kriek15. The dashed green line shows the best-fit relation obtained by Nakajima23 based on a compilation of early JWST data at redshifts $4<z<10$. GS_3073 at $z=5.55$, although being slightly more massive, is compatible with the extrapolation of this relation.