Doubling NIRSpec/IFS capability to calibrate the single epoch black hole mass relation at high redshift

TL;DR

The study addresses potential biases in single-epoch black hole mass estimators at high redshift by anchoring SE masses to reverberation-mapping results. It develops and releases a data-reduction method for JWST/NIRSpec IFU that doubles usable wavelength coverage, enabling detection of the Hα line at z~2 and direct SE-vs-RM comparisons, with an extended spectral range achieving up to about R ~ 2500. An empirical, RM/GRAVITY-anchored high-z SE calibration is derived for Hα and Hβ, showing Hβ-based estimators typically agree with RM within ~0.5 dex, while Hα-based estimators show larger scatter and a notable outlier at low mass and high Eddington ratio. The results yield a practical high-z SE calibration applicable to newly discovered JWST AGN populations, while cautions about extrapolation to lower masses and the need for larger samples. Overall, the paper providing the extended data-reduction pipeline and the first high-z SE calibration advances BH demographics studies in the early universe by enabling more reliable mass estimates from rest-frame optical lines.

Abstract

The recent discovery of a large population of overmassive black holes (BHs) in the early Universe challenges the validity of the BH-host galaxy coevolution framework. However, the reliability of the estimated BH masses (M$_{BH}$) is being questioned, as these are typically derived using single-epoch (SE) relations calibrated locally. Calibrating SE relations at high redshift would therefore enable more accurate M$_{BH}$ estimates and help identify potential biases. In this work, we release a data-reduction technique for JWST/NIRSpec IFU observations that doubles the effective wavelength coverage, enabling detection of otherwise inaccessible emission. Whenever adjacent dispersers are required, observers should carefully evaluate the tradeoff between integrating longer in the bluer configuration alone versus distributing the exposure time across two dispersers. We apply this pipeline to a sample of 5 quasars at z~2 with M$_{BH}$ independently measured through reverberation mapping (RM). This enables a joint analysis of both H$β$ and H$α$; the latter lying beyond the nominal wavelength range. We assess the reliability of the most widely adopted SE calibrations, finding that H$β$ yields the closest agreement with RM-based M$_{BH}$ estimates, whereas H$α$-based estimators exhibit a larger scatter. For the least massive BH in our sample ($M_{BH,RM}$~$10^{7.5}M_\odot$), which is accreting at a rate close to the Eddington limit ($λ_{Edd}=0.8$), all SE calibrators overestimate M$_{\rm BH, RM}$ by one order of magnitude. This may indicate a systematic overestimation of M$_{BH}$ for highly accreting BHs at high redshift. Finally, we provide the first high-redshift SE calibration based on H$α$ and H$β$. Although a larger sample is needed to reduce the uncertainties, our calibration can already be applied to the newly discovered BH population in the early Universe.

Figures (22)

• Figure 1: Count-rate images for the G140M/F100LP observations of RM332, a QSO at $z\sim 2.6$ downloaded from MAST of the NRS1 and NRS2 detectors, on left and right, respectively. The vertical red line shows the maximum wavelength range of the nominal coverage of the observations. The inset panels show a zoomed-in view of the [Oiii]-H$\beta$ complex on the left, and H$\alpha$ emission on the right.
• Figure 2: Observed flux of the extended G140M/F100LP filter for the star P330-E observed as part of the programs 1538 and 6645. The different color curves show the same flux for the star observed in cycle 1 (green), in cycle 3 at the center of the FOV (black), and in the bottom and top parts of the FOV (blue and purple, respectively). The same for the G235M/F170LP is shown in Appendix \ref{['appendix:fluxcalibration235m']}.
• Figure 3: Values of $\tilde{\alpha(\lambda)}$ (orange), $\tilde{\beta(\lambda)}$ (blue), $k(\lambda)$ (violet), as function of wavelength for the G140M/F100LP and the G235M/F170LP on the left and right panels, respectively.
• Figure 4: Ratio between the observed flux in the nominal wavelength range of the filters and the flux observed with the extended G140M/F100LP filter. The same figure for the G235M/F170LP is shown in Figure \ref{['fig:second_order_correction_g235m']}.
• Figure 5: Examples showing the spectrum in the nominal range (black), and the extended spectra before and after the calibration reported in Eq. \ref{['eq:spectrum_final']} (green and red, respectively). In the top panel, we show the spectrum extracted from a circular aperture of radius 0.5$^{\prime\prime}$ centered on the brightest pixel of the AGN SDSSJ0841 (PID: 2654), the nominal G140M/F100LP+G235M/F170LP spectrum, and the extensions of the G140M/F100LP. In the bottom panel, we show the spectrum extracted from a circular aperture of radius 0.5$^{\prime\prime}$ centered on the brightest pixel of the ULRIG UGC-5101 (PID: 2186) from the nominal G235M/F170LP and G395M/F290LP (black) and the extended G235M/F170LP spectrum. The nominal wavelength range encompassed by each filter is shown as the pink, purple, and blue lines for the G140M/F100LP, G235M/F170LP, and G395M/F290LP, respectively