A SPectroscopic Survey of Biased Halos in the Reionization Era (ASPIRE): A First Look at the Rest-frame Optical Spectra of z > 6.5 Quasars Using JWST

We perform spectral fitting for each WFSS spectrum using a model consisting of continuum and emission lines. The redshift derived from the [C ii] 158 μm line (Venemans et al. 2020; Yang et al. 2021; Wang et al. 2023) is used as the initial redshift. We use spectral data between observed wavelengths 31600 and 39500 Å, excluding edge regions where the calibration is less certain. This covers the rest-frame wavelength range of ∼4100–5200 Å for these quasars. We use a pseudocontinuum model including a power-law continuum and an empirical optical Fe ii template from Boroson & Green (1992). To fit this pseudocontinuum, we choose continuum windows that are free of strong emission lines in quasar composite, [4150, 4230], [4435, 4700], and [5100, 5200] Å in the rest frame. For the two highest-redshift quasars, J0109–3047 and J0218+0007, we adjust the windows due to their limited spectral coverage on the red side. For the former, we only use the first two windows. For the latter, we extend the blue-side cutoff of the last window to 5060 Å.

Hβ and [O iii ] lines. We adopt the rest-frame vacuum wavelengths of 4862.68 Å for the Hβ line and 4960.30 Å and 5008.24 Å for the [O iii] doublet (Vanden Berk et al. 2001). We employ a line model consisting of one narrow Gaussian component and three broad Gaussian components for the Hβ line (e.g., Shen 2016). For each [O iii] line, we use one Gaussian for the narrow component (the "core" component) plus a second Gaussian profile for the potentially broadened and blueshifted component. For the object with strong blueshifted [O iii] wings (i.e., J0224–4711), we include one more Gaussian for its broad [O iii] component. We set a lower limit to the line FWHM of each component to be 200 km s⁻¹. This is based on the instrument resolution (R ∼ 1600 at 4 μm). We also set an upper limit of 1200 km s⁻¹ for all narrow components (e.g., Shen et al. 2011).

We tie the velocity shift and dispersion of the narrow Hβ to that of the [O iii] "core" component. We assume a fixed line ratio of 3.0 between [O iii] λ5008 Å and [O iii] λ4960 Å for both "core" and broad components (e.g., Osterbrock & Ferland 2006). The velocity shifts of the broad [O iii] components of each transition are also tied. We do not fix the line ratio between narrow Hβ component and narrow [O iii] lines but limit the ratio of the "core" [O iii] 5008 to the narrow Hβ to ≥1. If for an object, there is no narrow [O iii] 5008 line, we consider that there should also be no narrow Hβ. For objects with narrow [O iii] components, we derive [O iii]-based redshifts using their narrow components and find that the [O iii]-based redshifts show good agreement with their redshifts derived from [C ii] 158 μm lines.

He ii line. We test the possibility of including a He ii line with a vacuum wavelength of 4687.02 Å by running alternate spectral decomposition. We include one Gaussian profile without constraining the line width and use a continuum window of [4435, 4635] Å instead of [4435, 4700] Å to avoid potential contamination from He ii emission. We run the same spectral fitting procedure and fit the He ii line together with the Hβ and [O iii] lines. We find that only one quasar (J2232+2930) has an obvious He ii line. Another quasar (J0218+0007) might have a narrow He ii line, but it only has a significance of ∼2σ. For these two, including He ii reduces their BH masses by 4% and 2%, respectively. For other objects, their best-fit models do not show any need for a He ii line, while the shorter continuum windows affect fits of their continuum, especially for z > 6.7 quasars. Thus, we finally only apply the He ii fit for quasar J2232+2930, and for other quasars we keep the continuum window cutoff of 4700 Å. Even if there could be some weak He ii emission in these quasars, their BH masses would not be affected significantly.

Hγ line. We also apply a two-Gaussian fit to the Hγ line at 4341.68 Å. The two Gaussian profiles are set for one narrow and one broad component. The same limits of narrow line width as used for Hβ are applied. The results are only used for plotting the best-fit but not any science analysis in this work.

We utilize a Monte Carlo (MC) approach to estimate the uncertainties of all spectral measurements, following Yang et al. (2021; also Shen et al. 2019; Wang et al. 2020). For each spectrum, we generate 50 mock spectra by randomly adding Gaussian noise at each pixel with a standard deviation equal to the spectral error at that pixel. We then perform the same spectral fitting procedure to each mock spectrum. The uncertainty of each spectral measurement is estimated by averaging the 16% and 84% percentile deviations from the median value. The best-fit results of all quasars are shown in Figure 1.

The single-epoch virial method is widely used for estimating the BH mass of high-redshift quasars, assuming virial motion of the line-emitting gas in the quasar broad-line region (BLR) and applying the empirical correlation between the BLR size and quasar continuum luminosity (i.e., the R–L relation). With this method, we can estimate the quasar central BH mass utilizing quasar broad emission lines and continuum luminosity. Quasar UV and optical emission lines, C iv λ1549 Å, Mg ii λ2800 Å, Hα λ6563 Å, and Hβ have all been used as virial BH mass tracers (e.g., McLure & Jarvis 2002; Vestergaard 2002; McLure & Dunlop 2004; Onken et al. 2004; Greene & Ho 2005; Vestergaard & Peterson 2006; Vestergaard & Osmer 2009; Shen et al. 2011). Calibration coefficients used in these BH mass estimators are determined using low-redshift quasar samples that have BH mass measurements based on RM. Among these lines, Hβ is often thought to be the most reliable tracer from both RM-based measurements and comparison of single-epoch mass estimators (e.g., Shen & Liu 2012).

We measure the single-epoch virial BH masses of these quasars based on their continuum luminosity at rest-frame 5100 Å and the Hβ line width, obtained from the spectral fitting. We adopt the BH mass estimator in Vestergaard & Peterson (2006),

$$\begin{eqnarray}&&\displaystyle \frac{{M}_{\mathrm{BH},{\rm{H}}\beta }}{{M}_{\odot }}={10}^{6.91}{\left(\displaystyle \frac{\mathrm{FWHM}({\rm{H}}\beta )}{1000\,\mathrm{km}\,{{\rm{s}}}^{-1}}\right)}^{2}{\left(\displaystyle \frac{\lambda {L}_{\lambda }(5100\,\mathring{\rm A} )}{{10}^{44}\,\mathrm{erg}\,{{\rm{s}}}^{-1}}\right)}^{0.5}.\end{eqnarray} \tag{ 1 }$$

The 5100 Å continuum luminosity is derived from the best-fit power-law continuum directly. The systematic scatter of this BH mass estimator relative to RM mass is ∼0.4–0.5 dex, which is not included when we report the errors of BH mass measurements. The uncertainties of all measurements are derived from the MC approach, as described above. We then compute the Eddington ratio (λ_Edd = L_bol/L_Edd) for each quasar. The bolometric luminosity is estimated by multiplying the 5100 Å luminosity by 9.26 (Richards et al. 2006). Table 1 summarizes the redshift, Hβ line width, continuum luminosity, Hβ BH mass, Eddington ratio, and Mg ii BH mass of each of the eight quasars.

Table 1. Spectral Fitting and Quasar Properties

Name	z[C II]	z[O III] a	MBH,Hβ	FWHMHβ	L5100	λEdd	MBH,Mg II	Ref_Mg ii b
			(109 M⊙)	(km s−1)	(1045erg s−1)		(109 M⊙)
J010953.13–304726.30	6.7904 ± 0.0003	⋯	0.60 ± 0.05	3033.0 ± 117.0	6.40 ± 0.08	0.76 ± 0.06	${1.11}_{-0.36}^{+0.4}$	Farina+2022
J021847.04+000715.20	6.7700 ± 0.0013	6.7668 ± 0.0006	0.61 ± 0.05	3030.0 ± 117.0	6.71 ± 0.16	0.78 ± 0.07	0.61 ± 0.07	Yang+2021
J022426.54–471129.40	6.5222 ± 0.0001	6.5186 ± 0.0004	2.15 ± 0.09	3936.0 ± 78.0	29.19 ± 0.09	0.97 ± 0.04	1.30 ± 0.18	Wang+2021
J022601.87+030259.28	6.5405 ± 0.0001	⋯	1.05 ± 0.09	3131.0 ± 129.0	17.25 ± 0.09	1.17 ± 0.09	2.20 ± 0.39	Wang+2021
J024401.02–500853.70	6.7306 ± 0.0002	6.720 ± 0.003	0.88 ± 0.10	2969.0 ± 158.0	15.08 ± 0.13	1.22 ± 0.12	1.15 ± 0.39	Reed+2019
J030516.92–315056.00	6.6139 ± 0.0002	6.615 ± 0.001	0.62 ± 0.14	3020.0 ± 314.0	6.95 ± 0.06	0.80 ± 0.14	0.70 ± 0.38	Wang+2021
J200241.59–301321.69	6.6876 ± 0.0004	⋯	0.84 ± 0.04	2981.0 ± 79.0	13.65 ± 0.10	1.15 ± 0.06	1.62 ± 0.27	Yang+2021
J223255.15+293032.04	6.666 ± 0.004	6.669 ± 0.0001	1.93 ± 0.13	6044.0 ± 201.0	4.24 ± 0.10	0.16 ± 0.01	3.06 ± 0.36	Yang+2021

Notes.

^a The [O iii] redshift is estimated based on the line center of the "core" component of [O iii] 5008 line. Quasars J0109–3047, J0226+0302, and J2002–3013 do not have narrow [O iii] components, and thus we do not estimate their [O iii] redshift. ^b The references of the Mg ii-based BH masses adopted in this work.

Download table as:  ASCIITypeset image

The Hβ-based measurements show that these quasars have BH masses in the range of (0.6–2.1) × 10⁹ M_⊙. These quasars all have been published with measurements of BH mass based on the Mg ii emission lines from their near-infrared (NIR) spectroscopic data (e.g., Mazzucchelli et al. 2017; Reed et al. 2019; Schindler et al. 2020; Wang et al. 2021; Yang et al. 2021; Farina et al. 2022). We adopt the most recent measurements of these eight quasars from Reed et al. (2019), Wang et al. (2021), Yang et al. (2021), and Farina et al. (2022). All of these Mg ii-based BH masses are computed using the BH mass estimator from Vestergaard & Osmer (2009). Figure 2 shows the Hβ-based BH masses compared to their Mg ii-based BH masses. The BH masses derived from the Hβ lines and 5100 Å luminosity are consistent with the Mg ii-based measurements within their systematic scatters (i.e., ∼0.4–0.5 dex). Recently, a new measurement of BH mass of quasar J0100+2802 at z = 6.3 using the Hβ line from JWST observations also shows agreement with its Mg ii-based BH mass within systematic uncertainties (Eilers et al. 2022).

Zoom In Zoom Out Reset image size

In our quasar sample, the differences between these two BH mass estimates span from −0.32 to 0.22 dex, comparable to the scatter shown in the SDSS low-redshift quasar sample (a standard deviation of ∼0.38 dex; Shen et al. 2011). Among these eight quasars, there is a trend that the Hβ-based BH masses are slightly smaller, with an average M_BH,Hβ –M_{BH,Mg II} of −0.13 dex. However, the current sample is too small to conclude if there is a systematic offset. The Hβ-based BH masses and the comparison between Hβ and Mg ii measurements confirm the existence of a billion solar-mass BHs in the reionization epoch. These SMBHs require ≳10^2–3 M_⊙ seed BHs at z = 30 assuming Eddington accretion across the entire time since BH seeding and a radiative efficiency of 0.1 (e.g., Yang et al. 2021; Farina et al. 2022). This thus raises the questions of how to maintain highly accreting BHs for the long term and how to form massive seed BHs in the early universe (e.g., Volonteri 2012; Davies et al. 2019; Inayoshi et al. 2020).

These measurements are derived from the spectral fits of the NIR spectra and JWST spectra separately. Thus, for both data sets, the limited continuum windows might lead to additional uncertainties of the spectral decomposition and BH mass measurements. In the next step, we will utilize the entire ASPIRE sample to perform a detailed comparison between the Hβ and Mg ii BH masses when observations of the full sample are complete. With the final sample, we will combine the NIR and JWST spectra as well as the JWST F115W, F200W, and F356W photometries and will apply joint spectral analysis. The combined data set will provide a wider continuum window and allow a better separation of the Fe ii emission from the featureless continuum.

The most important spectral quantities covered by the WFSS spectra are the Hβ, [O iii], and Fe ii emission. In general, all eight quasars have strong and broad Hβ lines, with FWHM ranging from 3000 to 6000 km s⁻¹. Optical Fe ii emission is clearly seen in all eight quasars. The observed [O iii] doublets in these quasars have relatively diverse properties. Three of these quasars, J0109–3047, J0226+0302, and J2002–3013, show similar profiles in their [O iii] line regions, with very weak broad-only [O iii] doublets but no narrow [O iii] lines (i.e., the "core" component, <1200 km s⁻¹). The other five quasars all have both narrow and broad [O iii] components. Quasar J2232+2930 has the strongest narrow [O iii] lines among these quasars. It is the only object observed with stronger narrow [O iii] lines relative to its broad components.

Using the WFSS data, we are able to visit the eigenvector 1 (EV1) relations (Boroson & Green 1992) in the most distant quasars. The EV1 relations describe the relationships between emission-line quantities and the strength of the optical Fe ii emission and has been widely investigated in the past three decades with quasar samples at z < 4 (e.g., Boroson & Green 1992; Marziani et al. 2001; Boroson 2002; Shen & Ho 2014). The anticorrelation between the strength of [O iii] and Fe ii emission is one of the most important EV1 relations observed. The trend of Fe ii strength, as the Hβ FWHM decreases, is also a probe of the physical drivers of the EV1 relations. We study the correlation of the [O iii] 5008 EW and FWHM_Hβ with the R_{Fe II} (Shen 2016) and compare them with the observations of the luminous intermediate-redshift (1.5 < z < 3.5) sample in Shen (2016) and the low-redshift (z < 1) quasars from the SDSS DR7 (Shen et al. 2011). The EW_{[O III]5008} used here is the EW of the entire [O iii] λ5008 line in order to compare our results with the measurements from Shen (2016). FWHM_Hβ is the line width of the broad Hβ. R_{Fe II} is the ratio of EW_{Fe II} and EW_Hβ , and the Fe ii EW is measured between 4434 and 4684 Å.

As shown in Figure 3, overall, the optical spectral quantities in these high-redshift quasars follow the relations defined in lower-redshift quasars. Our high-redshift quasars show slightly lower [O iii] EW than the mean values of SDSS low-redshift quasars and are more similar to the luminous 1.5 < z < 3.5 quasar sample, which is due to their high luminosity and the Baldwin effect (Baldwin 1977). There is no significant outlier among our high-redshift quasars. The three lowest [O iii] EWs are from the quasars with weak and broad-only [O iii] emission. In the FWHM_Hβ versus R_{Fe II} plane, our objects distribute in a narrow FWHM_Hβ range with lower FWHM_Hβ than the intermediate-redshift sample. This is consistent with the high Eddington ratios of most (7 of 8) of our high-z quasars. The general consistency between our sample and low-z distributions suggests that the EV1 relations might exist in quasars as early as the reionization epoch, and our full ASPIRE sample will enable more detailed comparisons with low-z samples.

Zoom In Zoom Out Reset image size

Broad, skewed, and blue/redshifted [O iii] profiles have been discovered in quasars at z ≲ 3.5 and used to probe kiloparsec-scale outflows in quasar host galaxies (e.g., Cano-Díaz et al. 2012; Carniani et al. 2015; Zakamska et al. 2016; Bischetti et al. 2017). At z > 6, outflows powered by the central active galactic nucleus are a common prediction of a broad range of cosmological simulations (e.g., Dubois et al. 2013; Costa et al. 2014, 2022; Lupi et al. 2022). However, observational evidence of large-scale outflows in quasar hosts remains anecdotal at this redshift. The JWST data set provides us with the first chance to search for [O iii] outflows in reionization-era quasars. Two quasars, J0218+0007 and J0224–4711, have prominently strong broad and blueshifted [O iii] components, with a velocity shift of −630 km s⁻¹ and −1690 km s⁻¹ relative to the narrow [O iii], respectively. Figure 4 shows their zoomed-in [O iii] λ5008 lines, with other components subtracted out according to the best-fit models. We compute the median velocity, v₅₀, the velocity at the 50th percentile of the line flux; v₁₀ and v₉₀, the velocities at the 10th and 90th percentiles; and w₈₀ (=v₉₀ − v₁₀), containing 80% of the total line power, which is close to the commonly used FWHM (w₈₀ = 1.088 FWHM) for a Gaussian profile. The velocities are calculated relative to their [C ii] redshifts. J0218+0007 has a median velocity (v₅₀) of ∼−610 km s⁻¹, with a v₉₀ of −1510 km s⁻¹ and a w₈₀ of 1590 km s⁻¹. J0224–4711 has a median velocity (v₅₀) of ∼−1430 km s⁻¹, a v₉₀ of −4010 km s⁻¹, and a w₈₀ of ∼4140 km s⁻¹.

Zoom In Zoom Out Reset image size

Their broad [O iii] widths and high-velocity shifts make them stand out among the quasar population. Particularly, J0224–4711 has one of the most extreme broad and blueshifted [O iii] lines observed to date, even compared to the observations of lower-redshift quasars. For example, the 1.5 < z < 3.5 luminous quasars in Shen (2016) have a median [O iii] FWHM of 900 km s⁻¹, with the broadest one to 2100 km s⁻¹. Zakamska et al. (2016) observed extreme outflows in red luminous quasars at z ∼ 2–3 and discovered broad [O iii] with w₈₀ ranging between 3600 and 5500 km s⁻¹, strongly blueshifted up to 1500 km s⁻¹. The more recent study of hyperluminous quasars at z ∼ 2–4 revealed a population of [O iii] lines with FWHM of 1200–2200 km s⁻¹ (Bischetti et al. 2017). Large velocity widths and high velocities of blueshifted [O iii] suggest powerful ionized outflows, while the nature of high-velocity large-scale gas in quasars is still debated. The most blueshifted [O iii] lines observed in the extremely red z = 2–3 quasars are thought to be caught during a "blow-out" phase of quasar evolution and associated with fast ionized outflows (e.g., Zakamska et al. 2016; Perrotta et al. 2019; Vayner et al. 2021). So far, there is no evidence of J0218+0007 and J0224–4711 being extremely red quasars. A detailed multiwavelength analysis will be conducted in the future to better understand the physical properties of their fast outflows. In addition, a joint analysis of the 1D spectra and 2D images will be performed in subsequent works to examine the host galaxies and outflows. In the next step, JWST NIRSpec/IFU follow-up observations will be carried out to reveal the nature of their outflows. The entire ASPIRE sample will allow us to systematically study [O iii] outflows in the reionization-era quasars, in synergy with multiwavelength imaging and spectroscopic data.