The Diverse Morphology, Stellar Population, and Black Hole Scaling Relations of the Host Galaxies of Nearby Quasars

Our parent sample consists of the 87 low-redshift ($z\lt 0.5$) optically/UV-selected quasars from the Palomar–Green (PG) survey (Schmidt & Green 1983; Boroson & Green 1992). Although the PG survey is incomplete due to large photometric error (Goldschmidt et al. 1992), this representative sample of nearby, bright quasars has been extensively studied for decades. The availability of line widths for the broad Hβ emission line from single-epoch spectroscopy (Boroson & Green 1992; Ho & Kim 2009) and spectrophotometry for the optical continuum (Neugebauer et al. 1987) enables BH masses to be estimated from conventional empirical calibrations (e.g., Vestergaard & Peterson 2006; Ho & Kim 2015). Robust bolometric luminosities, and hence Eddington ratios, can be derived from comprehensive broadband spectral energy distributions (Ho 2008). Figure 1 shows the distribution of V-band absolute magnitude versus redshift for the sample.

Zoom In Zoom Out Reset image size

We undertook a snapshot program (proposal 12903; PI: L.C. Ho) using the HST Wide Field Camera 3 (WFC3; Dressel 2019) to obtain high-resolution rest-frame optical images for a significant subset of the PG quasars. Snapshot programs cannot guarantee that any specific target will be observed, instead drawing from a predesignated list of candidates to fill in gaps in the HST schedule to obtain relatively short observations of typical duration 45 minutes or less (including all overheads). Our program achieved a success rate of ~40%, securing useful data for 35 sources randomly selected from the original sample of 87 quasars. The final sample forms a representative subset of the original sample (Figure 1). Details of the observed targets and the observations are summarized in Table 1.

Table 1. Object Information

Name	Alternative Name	Observation	Filter	Exposure Time	z	DL	AV	MV	$\mathrm{log}\,R$	$\mathrm{log}\,{L}_{5100}$	${\mathrm{FWHM}}_{{\rm{H}}\beta }$	$\mathrm{log}\,{M}_{\mathrm{BH}}$
		Date	(B/I)	(s)		(Mpc)	(mag)	(mag)		($\mathrm{erg}\,{{\rm{s}}}^{-1}$)	($\mathrm{km}\,{{\rm{s}}}^{-1}$)	(${M}_{\odot }$)
(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)	(13)
PG 0050+124 a	I Zw 1	2013-11-02	F438W/F105W	300/147	0.061	282.1	0.171	−23.77	−0.48	44.23	1171	7.29
PG 0052+251		2013-07-31	F475W/F105W	360/147	0.155	762.5	0.125	−24.64	−0.62	44.83	5187	8.90
PG 0804+761		2013-03-10	F438W/F105W	450/147	0.100	475.0	0.095	−24.44	−0.22	44.96	3045	8.51
PG 0923+129	Mrk 705	2012-12-16	F438W/F814W	300/180	0.029	131.1	0.109	−21.59	0.32	43.38	1957	7.28
PG 0934+013	Mrk 707	2013-01-29	F438W/F814W	450/300	0.050	229.4	0.184	−21.43	−0.42	43.51	1254	6.96
PG 0953+414		2013-02-03	F475W/F105W	300/147	0.239	1234.6	0.032	−25.65	−0.36	45.63	3111	8.88
PG 1011−040		2013-03-29	F438W/F814W	300/210	0.058	267.7	0.097	−22.70	−1.00	44.09	1381	7.36
PG 1012+008		2013-05-03	F475W/F105W	360/147	0.185	926.7	0.093	−24.79	−0.30	44.87	2615	8.33
PG 1022+519	Mrk 142	2013-03-04	F438W/F814W	300/210	0.045	205.8	0.044	−21.40	−0.64	43.56	1566	7.19
PG 1049−005		2014-03-05	F555W/F125W	450/147	0.357	1958.1	0.155	−25.93	−0.60	45.59	5351	9.33
PG 1100+772	3C 249.1	2013-05-02	F555W/F110W	390/147	0.313	1680.6	0.096	−25.86	2.51	45.72	6151	9.52
PG 1103−006	PKS 1103−0006	2013-04-12	F555W/F125W	540/147	0.425	2403.6	0.116	−25.96	2.43	45.73	6183	9.53
PG 1114+445		2014-03-25	F475W/F105W	540/147	0.144	703.6	0.042	−24.01	−0.89	44.68	4554	8.71
PG 1119+120	Mrk 734	2013-05-22	F438W/F814W	300/180	0.049	224.7	0.089	−22.49	−0.82	43.95	1773	7.50
PG 1149−110		2014-02-12	F438W/F814W	300/210	0.049	224.7	0.114	−21.90	−0.06	42.72	3032	7.31
PG 1202+281 a	GQ Comae	2013-10-28	F475W/F105W	360/147	0.165	816.7	0.056	−23.75	−0.72	44.29	5036	8.59
PG 1216+069		2013-01-14	F555W/F110W	390/147	0.334	1811.9	0.059	−26.33	0.22	45.81	5180	9.42
PG 1226+023 a	3C 273	2013-03-17	F547M/F105W	360/147	0.158	778.7	0.055	−27.15	3.06	46.20	3500	9.29
PG 1244+026 a		2014-02-16	F438W/F814W	360/270	0.048	219.9	0.068	−21.77	−0.28	43.42	721	6.43
PG 1259+593		2012-11-05	F606W/F125W	360/147	0.472	2722.7	0.022	−26.82	−1.00	46.16	3377	9.24
PG 1302−102	PKS 1302−102	2013-04-23	F555W/F110W	240/147	0.286	1514.8	0.117	−26.60	2.27	45.76	3383	9.03
PG 1322+659		2013-01-05	F475W/F105W	540/147	0.168	833.1	0.051	−24.23	−0.92	44.81	2765	8.34
PG 1351+640		2013-01-06	F438W/F105W	300/147	0.087	409.6	0.055	−24.08	0.64	44.45	5646	8.77
PG 1354+213		2013-03-09	F555W/F110W	480/147	0.300	1600.3	0.067	−24.34	−1.10	44.87	4127	8.73
PG 1444+407		2013-04-28	F555W/F110W	540/147	0.267	1400.2	0.037	−25.18	−1.10	45.49	2457	8.61
PG 1448+273		2014-07-19	F438W/F105W	540/147	0.065	301.4	0.079	−23.30	−0.60	44.14	815	6.93
PG 1534+580	Mrk 290	2013-09-20	F438W/F814W	225/300	0.030	135.7	0.038	−21.44	−0.15	43.65	5324	8.29
PG 1535+547	Mrk 486	2013-04-22	F438W/F814W	180/270	0.038	172.9	0.038	−22.15	−0.85	43.72	1420	7.18
PG 1613+658	Mrk 876	2013-03-25	F475W/F105W	450/270	0.129	624.4	0.071	−24.22	0.00	44.96	8441	9.40
PG 1626+554		2013-07-09	F475W/F105W	450/270	0.133	645.4	0.015	−23.54	−0.96	44.60	4474	8.65
PG 1700+518		2012-12-15	F555W/F110W	240/147	0.282	1490.5	0.089	−26.44	0.37	45.80	2185	8.67
PG 1704+608	3C 351	2012-12-03	F555W/F125W	240/147	0.371	2048.2	0.060	−26.38	2.81	45.90	6552	9.67
PG 2112+059		2013-05-18	F606W/F125W	300/147	0.466	2681.5	0.239	−27.26	−0.49	46.11	3176	9.16
PG 2214+139	Mrk 304	2012-11-12	F438W/F105W	390/147	0.067	311.2	0.194	−23.39	−1.30	44.37	4532	8.54
PG 2251+113	PKS 2251+11	2013-06-10	F555W/F110W	480/147	0.323	1742.9	0.229	−26.24	2.56	45.60	4147	9.12

Notes. Column (1): object name. Column (2): alternative name of the object. Column (3): date of observations. Column (4): filters for rest-frame B and I bands. Column (5): total long exposure time for the rest-frame B-band and I-band images. Column (6): redshift from Boroson & Green (1992). Column (7): luminosity distance. Column (8): galactic extinction in the V band, estimated from Schlafly & Finkbeiner (2011). Column (9): V-band absolute magnitude from Boroson & Green (1992). Column (10): ratio of radio-to-optical luminosity from Kellermann et al. (1989). Column (11): AGN 5100 Å continuum luminosity converted from B-band nuclear emission derived from our GALFIT decomposition. Column (12): FWHM of broad Hβ line from Boroson & Green (1992). Column (13) BH mass calculated in Section 4.3.

^a Observed in GYRO mode.

A machine-readable version of the table is available.

Download table as:  DataTypeset image

Each target was observed with two filters, which roughly correspond to rest-frame B band and I band, with the intent of providing a sufficiently wide color leverage to constrain the M/L of the stellar continuum in order to estimate the stellar mass of the host galaxy. Each pair of filters was customized carefully according to the redshift of the source to avoid, to the extent possible, contamination from strong emission lines. In general each object was observed with both the ultraviolet-visible (UVIS) and infrared (IR) channels of the WFC3, except for the nine sources with $z\lesssim 0.05$, for which both filters fall in the UVIS channel. The UVIS detector has a pixel scale of 0.0400'' × 0.0398'' and a PSF with an FWHM of ∼0.08''. The IR channel has a pixel scale of 0.135'' × 0.121'' and a PSF with an FWHM ≈ 0.2''. To reduce the readout overheads, we employ the UVIS2-M1K1C-SUB subarray for the UVIS detector and the IRSUB512 subarray for the IR detector, which yield, respectively, a field of view of $40^{\prime\prime} \times 40^{\prime\prime}$ and $69^{\prime\prime} \times 62^{\prime\prime}$, adequate to cover the target and sufficient nearby field for background subtraction. Exposure times varied. Every target was observed with a long exposure with a total integration time of ∼200–500 s in rest-frame B and ∼150–300 s in rest-frame I. We use a three-point and four-point dither pattern for the UVIS and IR observations, respectively, to facilitate rejection of cosmic rays and bad pixels, as well as to better sample the PSF. Given the brightness of the quasar at short wavelengths, saturation in the center is unavoidable in the long UVIS exposure. We acquired an additional short (∼6–100 s) UVIS exposure for the purposes of replacing the saturated nuclear region of the long exposure. The IR images, taken in MULTIACCUM mode with STEP25 timing sequence ($\mathrm{NSAMP}=6$), are not saturated.

Most of the observations were executed in the standard FINE LOCK mode, but four targets (PG 0050+124, 1202+281, 1226+023, and 1244+026) were observed in GYRO mode after a failure in guide star acquisition. This resulted in significant degradation of the pointing precision during the observations, prompting us to apply corrective procedures described in Appendix A.

We reprocessed the calibrated, flat-fielded science files using the auxiliary data sets acquired from the Hubble Data Archive ⁸ on 2016 January 25. We used AstroDrizzle (Gonzaga et al. 2012), largely with default parameters, to process the single-frame short exposure and to combine the dithered frames of the long exposure to remove cosmic rays and correct for geometric distortion. The dithered sub-exposures are weighted by their maps of effective exposure time during the drizzle combination. The pixel scale of the combined image is set to 0.06'' for the IR channel and 0.03'' for the UVIS channel to achieve Nyquist sampling. The final images have a size of $54^{\prime\prime} \times 54^{\prime\prime}$ for the IR channel and $36^{\prime\prime} \times 36^{\prime\prime}$ for the UVIS channel. The saturated nucleus of the long UVIS exposure was replaced by the corresponding pixels from the unsaturated short exposure, which was aligned to the former using AstroDrizzle. Depending on the level of saturation of the cores, we usually replace a circular region with a radius of 0.3''.

Although AstroDrizzle performs background removal, we take additional steps to verify the accuracy of the background determination and quantify its uncertainty. While the pixel-to-pixel variation of the background emission is dominated by Poisson noise, often the main uncertainty of the background comes from larger scale fluctuations. We first use the deeper I-band image to create a source mask from the segmentation image produced by SExtractor (Bertin & Arnouts 1996). The I-band mask is then applied to the B-band image. As in Li et al. (2011), the mask is expanded according to the flux of the source. Special care is paid to mask the region of the quasar and its host galaxy, whose total radial extent is determined from inspection of its one-dimensional (1D), azimuthally averaged surface brightness profile. If the quasar is sufficiently strong, the mask for its diffraction spikes needs to be adjusted manually. In order to identify the spatial scale and amplitude of the background variation, we randomly sample the unmasked pixels 10⁴ times with boxes of increasing size. For each box size, we calculate the rms of the sampled median background values. The scale of the background fluctuation is approximated by the box size above which the rms converges, and the uncertainty of the background is given by the rms value (Li et al. 2011; Huang et al. 2013). A constant background, the median of the sampled values, is subtracted from the image.

Accurate knowledge of the PSF is critical to decomposing the bright nucleus of an active galaxy, a prerequisite to deriving reliable measurements of the host. While the PSF of HST is much more stable than that of ground-based telescopes, it still varies with time due to telescope breathing and changes in instrument focus, and with position across the field due to optical distortion and the properties of the detector. Ideally, an empirical PSF model can be obtained by observing a bright star with a spectrum similar to that of the science target, observed close in time to and using the same observing strategy as the science target (Biretta 2012; Mechtley 2014). This option was not possible for our observations, in view of the severe scheduling limitations of the snapshot program. Moreover, our decision to use subarrays to minimize readout time further significantly reduced the field of view of the observations, to such an extent that hardly any of the images contained useful foreground stars that could serve as a PSF model. In the absence of a dedicatedly observed PSF, many studies have utilized synthetic PSFs (TinyTim; Krist et al. 2011) to analyze HST images of AGN host galaxies (e.g., Greene et al. 2008; Kim et al. 2008a, 2017; Jiang et al. 2011). Here, we explore another strategy, by constructing an empirical PSF using stars extracted from archival images observed non-simultaneously in other HST programs, but otherwise employing the same detector-filter combination and dithering strategy. We stack a number of stars to boost the signal-to-noise ratio. For each detector-filter combination used in our program, we searched the HST archive for useful images that contain suitably isolated, unsaturated stars that are at least one-third as bright as the corresponding quasar. We do not include stars with close contamination (extended sources or bright point sources) within ∼3'', where the diffraction spike is prominent. We use AstroDrizzle to process the images of the stars with exactly the same procedure and pixel scale as the science images. Each image, typically ∼30'' on a side, is inspected carefully for possible faint contaminating sources, which, when present, are replaced manually by random Gaussian noise using the standard deviation measured from the nearby background. The final PSF for each filter, a median-stacked combination of typically ∼10–65 stars, is only slightly (≲10%) broader than the profile of individual stars.

To test the effectiveness of our strategy, we choose a single, bright star having roughly similar brightness as a typical quasar in our sample to serve as the mock image of a "pure" quasar nucleus (one lacking any underlying host galaxy), and we fit it with different PSF models to evaluate how well each can match the point source. Figure 2 compares the performance of (1) a standard synthetic PSF generated from TinyTim, (2) an empirical PSF derived from a single star in the same field of view as the mock quasar with similar flux, and (3) an empirical PSF produced from stacking a large number of stars. ⁹ The image of the mock quasar and the PSF models are convolved with a Gaussian kernel to achieve Nyquist sampling, as discussed in Section 3.2. Using GALFIT to fit the mock quasar, we find that the stacked-star PSF performs slightly better than the single-star PSF, and both empirical PSFs are always superior to the TinyTim PSF, as judged by the reduced ${\chi }^{2}$ reported by GALFIT as well as the rms of the residuals of the 1D profile. This conclusion holds for tests with different sets of stars on images of different bands (e.g., F438W, F814W, F105W, and F110W). We also explored the influence of the number of stars used in the stacking, finding, as a rule of thumb, that combining ≳10 stars produces the most optimal empirical PSF.

Zoom In Zoom Out Reset image size

PSF mismatch in the central pixels of bright nuclei presents a critical challenge in the analysis of the host galaxies of AGNs, especially those bright enough to be considered quasars. The problem is exacerbated when the PSF is undersampled, as is the case for HST. While employing dithering strategies can alleviate the problem, in practice this still proved insufficient for our purposes, especially in the absence of dedicated PSF observations (Section 3.1). Under these circumstances, one can mitigate the PSF mismatch by artificially broadening the PSF image and science image by a modest amount, by convolving them with a Gaussian kernel so that the PSF is Nyquist sampled by the pixel scale of the detector (Kim et al. 2008a). We adopt the same strategy for this study. GALFIT requires a "σ image" to properly calculate ${\chi }^{2}$. Following Casertano et al. (2000), the σ image takes into account noise correlation due to the drizzling process of combining images and the different exposure times if the core of the nucleus is replaced. We nevertheless confirm that the decomposition results, after accounting for systematic uncertainties, are not significantly different if we adopt the σ image generated automatically by GALFIT.

We first focus on the I-band data, which have a higher signal-to-noise ratio than the B-band data. The I band also offers a higher contrast between the host galaxy and the AGN. Once a final, viable model is constructed for the I band, we adopt the best-fit parameters and apply them to the B band, adjusting only the flux amplitude of each component. This approach, similar to that of Li et al. (2021), ensures a consistent bulge-disk decomposition in both bands and mitigates severe PSF mismatch in the B band, even if the B-band model may not be optimized (see Section 3.3). We use the same source mask to estimate the background. We start with the simplest possible model—a point-like nucleus plus a single-component host galaxy—and slowly include more components (e.g., a disk or a bar) to account for additional substructure of the host galaxy if the data require. The centers of all of the components are tied together. Close companions, when sufficiently nearby, need to be fit simultaneously. Whenever possible, we allow the Sérsic index of the spheroidal component (elliptical galaxies and the bulges of disk galaxies) to be fit freely. However, as the solutions are not always meaningful because of parameter degeneracies or strong systematic effects arising from PSF mismatch, we often need to fix the Sérsic index to discrete values between n = 1 and n = 4 and choose that which produces the lowest residuals, with the range of acceptable solutions serving to bracket the uncertainties. Given the many systematic uncertainties in model selection, it is difficult to rely on formal statistical criteria (e.g., reduced ${\chi }^{2}$) to determine the goodness of fit. Instead, we utilize the 1D and 2D residuals to inform us of the relative merits of different models and to decide on whether and in which manner more complex models need to be considered. For example, spiral structure betrays the presence of a disk, and bisymmetric residuals often indicate that a bar is present. We invoke coordinate rotation to capture obvious spiral arms, even if they do not significantly affect the photometric parameters of the bulges of inactive galaxies (Gao & Ho 2017). Bending modes can mimic tidal features effectively, and Fourier modes of order up to 3 can accommodate a variety of non-asymmetric structures in the outermost extended regions (e.g., Kim et al. 2008b). In particular, the absolute amplitude of the m = 1 Fourier mode (A_L ) can effectively gauge the global asymmetry (Peng et al. 2010). Quasar hosts with ${A}_{L}\gt 0.2$ tend to be highly disturbed systems that are ongoing or late-stage mergers (Kim et al. 2017).

Figures 3 and 4 show the fits of the I-band and B-band images, respectively, and Appendix B provides notes on the individual objects. The best-fit parameters are reported in Table 2.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

View figure set (35 panels)

Tarball of figure set images

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

View figure set (35 panels)

Tarball of figure set images

Table 2. Fitted Parameters

	Nucleus		Bulge						Total
Name	mI	mB	mI	mB	Re	n	${\mu }_{e,I}$	${\mu }_{e,B}$	Parametric mI	Nonparametric mI	Parametric mB	Nonparametric mB	B/T	AL
	(mag)	(mag)	(mag)	(mag)	('')	(mag arcsec−2)	(mag arcsec−2)	(mag)	(mag)	(mag)	(mag)
(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(17)	(18)	(19)	(20)	(21)	(22)
PG 0050+124	13.60 ± 0.05	15.17 ± 0.05	14.08 ± 0.33	17.24 ± 0.32	1.33 ± 0.05	1.69 ± 0.05	17.64 ± 0.05	20.80 ± 0.05	13.36 ± 0.32	13.36 ± 0.05	16.23 ± 0.35	16.22 ± 0.09	0.52 ± 0.05	0.15 ± 0.05
PG 0052+251	14.80 ± 0.06	15.92 ± 0.06	17.31 ± 0.47	19.33 ± 0.51	1.92 ± 0.06	1	21.42 ± 0.13	23.37 ± 0.26	15.75 ± 0.44	15.70 ± 0.06	17.66 ± 0.48	17.22 ± 0.10	0.24 ± 0.05	0.03 ± 0.05
PG 0804+761	13.56 ± 0.06	14.55 ± 0.06	17.20 ± 0.36	19.15 ± 0.35	0.48 ± 0.05	0.58 ± 0.07	18.06 ± 0.13	20.01 ± 0.15	15.33 ± 0.35	15.29 ± 0.11	17.77 ± 0.35	⋯	0.18 ± 0.05	0.02 ± 0.05
PG 0923+129	15.20 ± 0.05	15.74 ± 0.05	14.53 ± 0.32	17.04 ± 0.32	1.01 ± 0.05	1.04 ± 0.05	17.26 ± 0.05	19.77 ± 0.05	13.44 ± 0.35	13.43 ± 0.05	15.60 ± 0.34	15.54 ± 0.05	0.37 ± 0.05	0.02 ± 0.05
PG 0934+013	15.63 ± 0.05	16.51 ± 0.05	18.50 ± 0.50	21.25 ± 0.44	0.57 ± 0.09	0.47 ± 0.08	19.65 ± 0.11	22.40 ± 0.05	15.19 ± 0.55	15.20 ± 0.05	17.10 ± 0.45	16.98 ± 0.05	0.05 ± 0.05	0.00 ± 0.05
PG 0953+414	14.42 ± 0.06	15.04 ± 0.05	18.12 ± 0.59	19.68 ± 0.51	3.34 ± 0.83	1	23.43 ± 0.06	24.99 ± 0.33	18.12 ± 0.59	17.77 ± 0.43	19.68 ± 0.51	⋯	1	0.17 ± 0.05
PG 1011-040	14.95 ± 0.05	15.50 ± 0.05	16.68 ± 0.45	19.44 ± 0.44	0.84 ± 0.07	2.84 ± 0.66	19.51 ± 0.11	22.27 ± 0.05	14.52 ± 0.52	14.51 ± 0.05	16.71 ± 0.46	16.55 ± 0.05	0.14 ± 0.05	0.10 ± 0.05
PG 1012+008	15.77 ± 0.06	16.26 ± 0.06	16.23 ± 0.50	18.73 ± 0.44	1.32 ± 0.74	7.00 ± 2.82	20.52 ± 0.96	23.02 ± 0.23	15.11 ± 0.45	15.12 ± 0.05	17.71 ± 0.45	17.56 ± 0.12	⋯	0.51 ± 0.05
PG 1022+519	15.68 ± 0.05	16.33 ± 0.05	16.63 ± 0.46	18.51 ± 0.44	0.43 ± 0.05	1.06 ± 0.15	17.52 ± 0.09	19.40 ± 0.05	14.99 ± 0.46	14.98 ± 0.05	17.03 ± 0.46	16.93 ± 0.05	0.22 ± 0.05	0.05 ± 0.05
PG 1049-005	14.88 ± 0.05	16.02 ± 0.05	18.13 ± 0.36	20.21 ± 0.30	0.87 ± 0.06	0.34 ± 0.07	20.08 ± 0.17	22.16 ± 0.05	16.89 ± 0.62	16.66 ± 0.05	18.79 ± 0.41	⋯	⋯	0.54 ± 0.31
PG 1100+772	14.60 ± 0.05	15.44 ± 0.05	16.92 ± 0.32	18.48 ± 0.31	2.09 ± 0.17	1.08 ± 0.19	21.25 ± 0.15	22.81 ± 0.06	16.92 ± 0.32	16.68 ± 0.05	18.48 ± 0.31	⋯	1	0.03 ± 0.05
PG 1103-006	15.43 ± 0.05	16.12 ± 0.05	17.06 ± 0.34	19.54 ± 0.30	2.03 ± 0.38	1.81 ± 0.45	21.58 ± 0.33	24.06 ± 0.05	17.06 ± 0.34	16.87 ± 0.05	19.54 ± 0.30	⋯	1	0.04 ± 0.05
PG 1114+445	15.26 ± 0.06	16.23 ± 0.06	16.28 ± 0.38	18.88 ± 0.32	0.97 ± 0.25	2.40 ± 0.88	19.34 ± 0.47	21.94 ± 0.05	15.87 ± 0.33	15.82 ± 0.05	18.34 ± 0.42	⋯	0.69 ± 0.08	0.03 ± 0.05
PG 1119+120	14.76 ± 0.05	15.48 ± 0.05	15.79 ± 0.36	18.05 ± 0.32	0.60 ± 0.07	1	17.37 ± 0.10	19.63 ± 0.05	14.62 ± 0.38	14.49 ± 0.05	16.82 ± 0.33	16.53 ± 0.05	0.34 ± 0.05	0.06 ± 0.05
PG 1149-110	17.60 ± 0.05	18.54 ± 0.05	14.40 ± 0.33	17.25 ± 0.32	2.25 ± 0.16	2.28 ± 0.10	19.26 ± 0.11	22.11 ± 0.05	13.87 ± 0.39	13.93 ± 0.05	16.27 ± 0.35	16.33 ± 0.05	0.61 ± 0.08	0.12 ± 0.05
PG 1202+281	16.37 ± 0.06	17.48 ± 0.06	16.28 ± 0.32	18.55 ± 0.30	1.18 ± 0.17	4.50 ± 0.79	20.09 ± 0.29	22.36 ± 0.05	16.28 ± 0.32	16.29 ± 0.05	18.55 ± 0.30	18.52 ± 0.08	1	0.04 ± 0.05
PG 1216+069	14.52 ± 0.05	15.40 ± 0.05	17.70 ± 0.62	18.65 ± 0.35	5.35 ± 1.59	1.17 ± 0.62	24.11 ± 0.62	25.06 ± 0.18	17.70 ± 0.62	17.52 ± 0.10	18.65 ± 0.35	⋯	1	0.02 ± 0.05
PG 1226+023	12.12 ± 0.06	12.71 ± 0.08	14.87 ± 0.68	16.43 ± 0.40	11.61 ± 4.13	4	23.58 ± 0.15	25.14 ± 0.05	14.87 ± 0.68	14.56 ± 0.27	16.43 ± 0.40	16.27 ± 0.90	1	0.09 ± 0.07
PG 1244+026	16.11 ± 0.05	16.81 ± 0.05	18.12 ± 0.44	18.72 ± 0.44	0.10 ± 0.05	1	15.81 ± 0.05	16.41 ± 0.05	15.54 ± 0.46	15.54 ± 0.05	17.35 ± 0.44	17.23 ± 0.05	0.09 ± 0.05	0.04 ± 0.05
PG 1259+593	14.68 ± 0.05	15.47 ± 0.05	16.72 ± 0.54	18.63 ± 0.60	11.07 ± 3.33	4	25.33 ± 0.33	27.24 ± 0.45	16.72 ± 0.54	17.55 ± 0.05	18.63 ± 0.60	18.45 ± 0.09	1	0.09 ± 0.05
PG 1302-102	14.45 ± 0.05	15.10 ± 0.06	17.12 ± 0.44	18.92 ± 0.53	1.60 ± 0.05	0.87 ± 0.05	20.77 ± 0.05	22.55 ± 0.30	16.12 ± 0.44	15.92 ± 0.25	18.07 ± 0.54	⋯	⋯	0.18 ± 0.05
PG 1322+659	15.56 ± 0.05	16.24 ± 0.06	18.08 ± 0.33	20.80 ± 0.55	0.58 ± 0.05	1	19.59 ± 0.13	22.31 ± 0.45	16.79 ± 0.32	16.85 ± 0.07	18.91 ± 0.38	⋯	0.30 ± 0.05	0.03 ± 0.05
PG 1351+640	14.71 ± 0.06	15.57 ± 0.05	14.67 ± 0.31	17.23 ± 0.30	1.53 ± 0.24	4.98 ± 0.79	19.10 ± 0.32	21.66 ± 0.05	14.67 ± 0.31	14.71 ± 0.05	17.23 ± 0.30	16.96 ± 0.05	1	0.02 ± 0.05
PG 1354+213	16.34 ± 0.05	17.49 ± 0.05	18.92 ± 0.36	20.82 ± 0.32	0.55 ± 0.05	0.19 ± 0.10	19.71 ± 0.25	21.61 ± 0.05	17.10 ± 0.33	17.11 ± 0.05	19.51 ± 0.33	19.21 ± 0.05	0.19 ± 0.05	0.03 ± 0.05
PG 1444+407	14.93 ± 0.05	15.69 ± 0.05	17.99 ± 0.35	19.82 ± 0.32	0.95 ± 0.05	0.23 ± 0.07	20.02 ± 0.16	21.85 ± 0.05	16.88 ± 0.33	16.82 ± 0.05	18.66 ± 0.32	⋯	0.36 ± 0.05	0.00 ± 0.05
PG 1448+273	14.57 ± 0.06	15.66 ± 0.05	14.94 ± 0.44	17.21 ± 0.44	1.54 ± 0.05	2.43 ± 0.15	19.01 ± 0.05	21.28 ± 0.05	14.53 ± 0.44	14.51 ± 0.05	16.76 ± 0.45	16.60 ± 0.05	⋯	0.62 ± 0.05
PG 1534+580	14.58 ± 0.05	15.23 ± 0.05	15.57 ± 0.40	17.87 ± 0.30	0.73 ± 0.18	1.61 ± 0.48	17.81 ± 0.35	20.11 ± 0.05	14.30 ± 0.34	14.28 ± 0.05	16.37 ± 0.31	16.23 ± 0.05	0.31 ± 0.05	0.03 ± 0.05
PG 1535+547	14.41 ± 0.05	15.58 ± 0.05	16.89 ± 0.53	19.60 ± 0.35	0.99 ± 0.28	1	19.56 ± 0.11	22.27 ± 0.15	15.47 ± 0.35	15.28 ± 0.05	17.74 ± 0.33	17.50 ± 0.05	0.27 ± 0.08	0.02 ± 0.05
PG 1613+658	14.46 ± 0.05	15.22 ± 0.05	14.05 ± 0.36	16.64 ± 0.33	5.99 ± 1.91	6.11 ± 0.99	21.55 ± 0.56	24.14 ± 0.06	13.71 ± 0.34	13.82 ± 0.05	16.27 ± 0.34	16.29 ± 0.06	⋯	0.30 ± 0.05
PG 1626+554	15.37 ± 0.06	16.26 ± 0.06	18.16 ± 0.34	20.70 ± 0.59	0.71 ± 0.05	1	20.11 ± 0.11	22.65 ± 0.49	15.93 ± 0.32	15.91 ± 0.05	18.46 ± 0.44	18.08 ± 0.18	0.13 ± 0.05	0.01 ± 0.05
PG 1700+518	13.99 ± 0.05	14.99 ± 0.05	17.46 ± 0.36	18.63 ± 0.30	2.09 ± 0.88	4	22.45 ± 0.93	23.62 ± 0.05	16.94 ± 0.32	16.73 ± 0.05	18.45 ± 0.30	⋯	⋯	0.19 ± 0.05
PG 1704+608	14.17 ± 0.05	15.44 ± 0.05	16.42 ± 0.37	17.93 ± 0.34	6.38 ± 1.81	4	23.83 ± 0.38	25.34 ± 0.15	16.42 ± 0.37	16.57 ± 0.05	17.93 ± 0.34	⋯	1	0.12 ± 0.05
PG 2112+059	14.31 ± 0.05	15.34 ± 0.05	18.12 ± 0.53	19.71 ± 0.98	5.35 ± 2.05	4	25.15 ± 0.80	26.74 ± 0.89	18.12 ± 0.53	17.96 ± 0.05	19.71 ± 0.98	⋯	1	0.01 ± 0.18
PG 2214+139	13.75 ± 0.06	14.99 ± 0.05	13.77 ± 0.31	16.88 ± 0.31	3.44 ± 0.32	2.42 ± 0.13	19.58 ± 0.17	22.69 ± 0.06	13.77 ± 0.31	13.77 ± 0.05	16.88 ± 0.31	⋯	1	0.01 ± 0.05
PG 2251+113	14.53 ± 0.05	15.66 ± 0.05	16.53 ± 0.34	18.26 ± 0.32	4.96 ± 1.93	4	23.40 ± 0.63	25.13 ± 0.10	16.53 ± 0.34	16.82 ± 0.05	18.26 ± 0.32	⋯	1	0.07 ± 0.05

Note. Column (1): object name. Column (2): apparent nuclear magnitude in the rest-frame I band. Column (3): apparent nuclear magnitude in the rest-frame B band. Column (4): apparent bulge magnitude in the rest-frame I band. Column (5): apparent bulge magnitude in the rest-frame B band. Column (6): effective radius of bulge in the rest-frame I band. Column (7): Sérsic index of the bulge component. Column (8): surface brightness of bulge at the effective radius in the rest-frame I band. Column (9): surface brightness of bulge at the effective radius in the rest-frame B band. Column (10): flags on disk (d), bar (b), merger companions (c), or tidal features (f). Column (11): apparent magnitude of disk, bar, merger companions, or tidal features for the rest-frame I-band image. Column (12): apparent magnitude of disk, bar, merger companions, or tidal features for the rest-frame B-band image. Column (13): effective radius of disk, bar, merger companions, or tidal features in the rest-frame I band. Column (14): Sérsic index of disk, bar, merger companions, or tidal features. Column (15): surface brightness of disk, bar, merger companions, or tidal features at the effective radius in the rest-frame I band. Column (16): surface brightness of disk, bar, merger companions, or tidal features at the effective radius in the rest-frame B band. Column (17): parametric host magnitude of the rest-frame I -band image from combining the luminosity of the bulge and disk, bar, merger companions, or tidal features. Column (18): host magnitude measured nonparametrically after subtracting the best-fit nucleus in the rest-frame I band. Column (19): parametric host magnitude of the rest-frame B-band image from combining the luminosity of the bulge and disk, bar, merger companions, or tidal features. Column (20): host magnitude measured nonparametrically after subtracting the best-fit nucleus in the rest-frame B band. Column (21): bulge-to-total ratio (B/T). Column (22): absolute amplitude of the first Fourier mode, which measures the lopsidedness of the galaxy. Columns (10)–(16) only appear in the machine-readable version of the table.

Only a portion of this table is shown here to demonstrate its form and content. A machine-readable version of the full table is available.

Download table as:  DataTypeset image

The formal uncertainties reported by GALFIT are usually small. In practice, the error budget for most of the derived parameters are dominated by several sources of systematic uncertainty. First, as described in Section 3.2, many of the fitting results, especially those that pertain to the bulge, are extremely sensitive to the PSF. While we have made every effort to secure a reliable empirical PSF based on stacking a number of observed stars (Section 3.1), of course, the final PSF depends on the exact number and choice of stars. To account for this factor, we repeat the fits with different realizations of the empirical PSF constructed from various combinations of observed stars. Next, due to the limited field of view of our images, we opt to fix the background to its predetermined value (Section 2.3) in order to facilitate the convergence of the fit (Kim et al. 2008a), but we explore the impact of large-scale fluctuations of the background by repeating the fits with the background value perturbed around its rms value. Finally, as documented in Appendix B, a major source of uncertainty occasionally arises from the ambiguity of model selection. Because of the complexity of AGN host galaxies, the robustness of the fits must be scrutinized individually for each source, and often the "best fit" is based less on any rigorous statistical criteria than on a reasonable choice bracketed by a range of plausible choices. The adopted range of acceptable solutions then defines the true uncertainty range. The final uncertainties reported in Table 2 reflect the quadrature sum of the statistical uncertainties from GALFIT and the systematic uncertainties due to the (1) the PSF, (2) background subtraction, and (3) model selection.

Following Kim et al. (2008a), we cross-check the decomposition results by comparing the parametrically derived total magnitudes with those measured nonparametrically (Figure 5). After removing the nuclear emission and close contaminating sources based on the overall GALFIT model, we sum all the remaining flux to estimate the total magnitude of the host galaxy. The uncertainty of the nonparametric magnitude is estimated by adopting different PSF models and background subtraction. The nonparametric magnitudes usually agree well with the sum of the parametric components, with the median difference between the parametric and nonparametric magnitudes and their standard deviation being ${\rm{\Delta }}m=0.01\pm 0.19$ mag in the I band. PSF mismatch poses a more serious problem in the B band and strongly affects the nonparametric magnitudes if the host galaxy is faint. For the 20 sources for which a meaningful comparison can be made, the B-band parametric magnitudes are fainter than the nonparametric magnitudes by ${\rm{\Delta }}m=0.16\pm 0.13$ mag. The parametric measurements, constrained by the structural parameters based on the I-band decomposition, may miss more extended emission in the B band. Nevertheless, the difference is much smaller than the uncertainty of the integrated magnitudes, and in terms of stellar mass it amounts to only 0.07 dex, which is much smaller than the mass uncertainty.

Zoom In Zoom Out Reset image size

On account of the strong contamination by the AGN core, coupled with the large distances of the sources and the relative shallowness of the HST images, the morphology of the host galaxies is challenging to ascertain. We base our morphological classification in part on our quantitative decomposition and in part on visual inspection of the original images and the residual images after subtracting the best-fit parametric model (Figures 3 and 4). Except for a single case (Appendix B), the 11 galaxies (31% of the sample) that can be fit adequately with a single Sérsic component are classified as ellipticals, even if in deep observations nearby ellipticals often exhibit more than one major photometric component (Huang et al. 2013; Zhu et al. 2021, hereafter ZHG21. Although the visual classification is usually challenging when a source is at $z\gtrsim 0.2$, particularly when the nuclear emission overwhelms the starlight, the host galaxies, robustly detected, are sufficiently luminous and massive to be regarded securely as ellipticals. Disk galaxies are recognizable as two-component systems with a Sérsic bulge plus an exponential disk, which is often accompanied by an additional non-axisymmetric feature that can be readily identified with a bar. Spiral features in the residual maps offer further evidence in support of the presence of a disk. Nearly half of the sample (49%, 17/35) resides in disks, among them ∼35% (6/17) that are barred. Six objects (17%) exhibit irregular substructure indicative of unambiguous ongoing or recent merger activity. Throughout the paper, we mainly focus on galaxy major mergers with stellar mass ratio typically ≲0.3. PG 1202+281 has a close projected neighbor, but in the absence of detectable tidal features or redshift information for the nearby source, we do not consider it a merger system. The single-component hosts are systematically more luminous (median ${M}_{I}=-23.71$ mag) than those decomposed into two (bulge and disk) or three (bulge, bar, and disk) components (median ${M}_{I}=-22.66$ mag), consistent with the known luminosity difference between nearby ellipticals and spirals (e.g., Burstein et al. 2005). This gives us some confidence that the ellipticals have been properly classified. The mergers are the most luminous, with median ${M}_{I}=-24.21$ mag, not too dissimilar from the ellipticals.

We estimate the total stellar mass of the host galaxy, and separately for the bulge alone if present, from its rest-frame, extinction-corrected I-band magnitude and B − I color, using the color-dependent M/L prescribed by Bell & de Jong (2001). To convert the measured HST-based magnitudes to rest-frame magnitudes in the Johnson B and I bands, we adopt Bruzual & Charlot (2003) stellar population synthesis models covering a grid of stellar ages provided by HST calibration reference data. ¹⁰ The models assume an exponentially decaying star formation history with an e-folding time of 0.6 Gyr, solar metallicity, and a Chabrier (2003) stellar initial mass function (IMF). The model spectra are redshifted to the observed frame of each target, corrected for foreground extinction using the extinction values of Schlafly & Finkbeiner (2011) and the Galactic extinction curve of Cardelli et al. (1989), and then convolved with the respective HST filter response function using synphot. ¹¹ The model spectrum that provides the closest match to the observed color and its associated uncertainty is used to calculate the K correction. Typical ages for the bulge component range between 1 and 12 Gyr, overlapping with but systematically larger than the ages of 1–5 Gyr that characterize the entire galaxy. The uncertainty of the K correction is typically small (≲0.05 mag; Kim et al. 2017). The derived quantities are not sensitive to the star formation history of the adopted stellar synthesis model. Whether we use the simple stellar population model or the exponential decaying model with an e-folding time of 1.0 Gyr, our derived bulge and host galaxy masses varies by typically ≲0.1 dex.

The calculations of Bell & de Jong (2001) are based on the IMF of Salpeter (1955). Scaling the stellar masses by a factor of 0.61 (Madau & Dickinson 2014) to match the IMF of Chabrier (2003),

$$\begin{eqnarray}&&\mathrm{log}\left(\displaystyle \frac{{M}_{* }}{{M}_{\odot }}\right)=-0.4({M}_{I}-{M}_{I,\odot })+0.439(B-I)-0.609,\end{eqnarray} \tag{ 2 }$$

where ${M}_{I,\odot }=4.07$ mag is the I-band absolute magnitude of the Sun (Blanton & Roweis 2007). The final uncertainties of the stellar masses are significant (median ∼0.5 dex; Table 3), accounting not only for the ∼0.3 dex uncertainty of M/L (Conroy 2013), but also for the significant uncertainty associated with the photometric decomposition. Our subsequent analysis will use the results based on the parametric magnitudes, except for the hosts of objects classified as mergers, for which the nonparametric magnitudes should be more reliable when the nuclei are properly removed. The concept of a "bulge," of course, is not meaningful in an ongoing merger. As all the mergers in our sample appear to be quite substantial, for simplicity we assume that the final product of the merger will be an elliptical, or at the very least a highly bulge-dominated galaxy, even if a disk can reform (e.g., Peschken et al. 2020).

Table 3. Physical Parameters

		Bulge				Total
Name	Morphology	MB	MI	$(B-I)$	$\mathrm{log}{M}_{\mathrm{bulge}}$	MB	MI	$(B-I)$	$\mathrm{log}{M}_{\mathrm{total}}$
		(mag)	(mag)	(mag)	(${M}_{\odot }$)	(mag)	(mag)	(mag)	(${M}_{\odot }$)
(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)
PG 0050+124	D	−20.72 ± 0.36	−22.69 ± 0.34	1.97 ± 0.50	10.96 ± 0.49	−21.72 ± 0.38	−23.42 ± 0.33	1.70 ± 0.50	11.13 ± 0.49
PG 0052+251	B	−20.44 ± 0.51	−21.77 ± 0.48	1.33 ± 0.70	10.31 ± 0.63	−22.11 ± 0.49	−23.33 ± 0.45	1.22 ± 0.67	10.89 ± 0.60
PG 0804+761	B	−19.64 ± 0.41	−20.90 ± 0.37	1.26 ± 0.55	9.93 ± 0.53	−21.23 ± 0.41	−22.67 ± 0.36	1.44 ± 0.55	10.72 ± 0.52
PG 0923+129	D	−19.00 ± 0.32	−21.16 ± 0.32	2.16 ± 0.45	10.43 ± 0.47	−20.39 ± 0.35	−22.25 ± 0.35	1.86 ± 0.49	10.74 ± 0.50
PG 0934+013	B	−16.29 ± 0.49	−18.49 ± 0.50	2.20 ± 0.70	9.38 ± 0.64	−20.21 ± 0.49	−21.77 ± 0.55	1.56 ± 0.74	10.41 ± 0.67
PG 0953+414	U	−21.14 ± 0.55	−22.09 ± 0.59	0.95 ± 0.81	10.27 ± 0.72	−21.14 ± 0.55	−22.09 ± 0.59	0.95 ± 0.81	10.27 ± 0.72
PG 1011-040	B	−18.39 ± 0.45	−20.61 ± 0.45	2.22 ± 0.64	10.24 ± 0.59	−21.01 ± 0.48	−22.76 ± 0.52	1.75 ± 0.71	10.89 ± 0.65
PG 1012+008	M	−22.69 ± 0.15	−24.40 ± 0.07	1.71 ± 0.17	11.53 ± 0.32	−22.69 ± 0.15	−24.40 ± 0.07	1.71 ± 0.17	11.53 ± 0.32
PG 1022+519	B	−18.36 ± 0.47	−20.01 ± 0.46	1.65 ± 0.66	9.75 ± 0.60	−19.97 ± 0.48	−21.67 ± 0.46	1.70 ± 0.66	10.43 ± 0.61
PG 1049-005	M	−23.06 ± 0.44	−24.06 ± 0.63	1.00 ± 0.77	11.08 ± 0.71	−23.06 ± 0.44	−24.06 ± 0.63	1.00 ± 0.77	11.08 ± 0.71
PG 1100+772	E	−22.89 ± 0.33	−23.86 ± 0.32	0.97 ± 0.46	10.99 ± 0.47	−22.89 ± 0.33	−23.86 ± 0.32	0.97 ± 0.46	10.99 ± 0.47
PG 1103-006	E	−23.07 ± 0.38	−24.29 ± 0.34	1.22 ± 0.51	11.27 ± 0.50	−23.07 ± 0.38	−24.29 ± 0.34	1.22 ± 0.51	11.27 ± 0.50
PG 1114+445	D	−20.62 ± 0.32	−22.53 ± 0.39	1.91 ± 0.50	10.87 ± 0.51	−21.16 ± 0.42	−22.94 ± 0.34	1.78 ± 0.54	10.98 ± 0.51
PG 1119+120	D	−19.22 ± 0.35	−21.09 ± 0.36	1.87 ± 0.50	10.28 ± 0.50	−20.45 ± 0.35	−22.26 ± 0.38	1.81 ± 0.52	10.72 ± 0.51
PG 1149-110	D	−20.15 ± 0.33	−22.50 ± 0.33	2.35 ± 0.47	11.05 ± 0.48	−21.05 ± 0.38	−23.02 ± 0.39	1.97 ± 0.54	11.09 ± 0.53
PG 1202+281	E	−21.32 ± 0.31	−22.93 ± 0.32	1.61 ± 0.45	10.90 ± 0.47	−21.32 ± 0.31	−22.93 ± 0.32	1.61 ± 0.45	10.90 ± 0.47
PG 1216+069	E	−22.88 ± 0.35	−23.23 ± 0.62	0.35 ± 0.71	10.47 ± 0.68	−22.88 ± 0.35	−23.23 ± 0.62	0.35 ± 0.71	10.47 ± 0.68
PG 1226+023	E	−22.73 ± 0.44	−24.24 ± 0.69	1.51 ± 0.82	11.38 ± 0.76	−22.73 ± 0.44	−24.24 ± 0.69	1.51 ± 0.82	11.38 ± 0.76
PG 1244+026	B	−18.20 ± 0.44	−18.65 ± 0.44	0.45 ± 0.62	8.68 ± 0.58	−19.71 ± 0.47	−21.26 ± 0.46	1.55 ± 0.66	10.20 ± 0.60
PG 1259+593	E	−23.73 ± 0.61	−24.90 ± 0.54	1.17 ± 0.81	11.49 ± 0.71	−23.73 ± 0.61	−24.90 ± 0.54	1.17 ± 0.81	11.49 ± 0.71
PG 1302-102	M	−23.11 ± 0.54	−24.36 ± 0.44	1.25 ± 0.70	11.31 ± 0.62	−23.11 ± 0.54	−24.36 ± 0.44	1.25 ± 0.70	11.31 ± 0.62
PG 1322+659	D	−19.19 ± 0.56	−21.12 ± 0.33	1.93 ± 0.65	10.32 ± 0.56	−21.01 ± 0.39	−22.47 ± 0.33	1.46 ± 0.51	10.65 ± 0.50
PG 1351+640	E	−21.33 ± 0.35	−22.98 ± 0.32	1.65 ± 0.47	10.94 ± 0.48	−21.33 ± 0.35	−22.98 ± 0.32	1.65 ± 0.47	10.94 ± 0.48
PG 1354+213	D	−20.47 ± 0.33	−21.67 ± 0.36	1.20 ± 0.49	10.21 ± 0.50	−21.78 ± 0.34	−23.49 ± 0.33	1.71 ± 0.47	11.17 ± 0.48
PG 1444+407	D	−21.05 ± 0.32	−22.28 ± 0.36	1.23 ± 0.48	10.47 ± 0.49	−22.21 ± 0.32	−23.39 ± 0.34	1.18 ± 0.47	10.89 ± 0.48
PG 1448+273	M	−21.24 ± 0.05	−22.48 ± 0.05	1.24 ± 0.07	10.56 ± 0.31	−21.24 ± 0.05	−22.48 ± 0.05	1.24 ± 0.07	10.56 ± 0.31
PG 1534+580	D	−18.12 ± 0.32	−20.15 ± 0.40	2.03 ± 0.51	9.97 ± 0.52	−19.60 ± 0.32	−21.42 ± 0.34	1.82 ± 0.47	10.39 ± 0.48
PG 1535+547	D	−17.03 ± 0.36	−19.38 ± 0.53	2.35 ± 0.64	9.80 ± 0.62	−18.83 ± 0.35	−20.79 ± 0.35	1.96 ± 0.49	10.20 ± 0.50
PG 1613+658	M	−22.92 ± 0.06	−24.72 ± 0.05	1.80 ± 0.08	11.70 ± 0.31	−22.92 ± 0.06	−24.72 ± 0.05	1.80 ± 0.08	11.70 ± 0.31
PG 1626+554	D	−18.53 ± 0.59	−20.43 ± 0.35	1.90 ± 0.69	10.03 ± 0.59	−20.77 ± 0.44	−22.66 ± 0.33	1.89 ± 0.55	10.91 ± 0.52
PG 1700+518	M	−22.59 ± 0.31	−23.56 ± 0.33	0.97 ± 0.45	10.87 ± 0.47	−22.59 ± 0.31	−23.56 ± 0.33	0.97 ± 0.45	10.87 ± 0.47
PG 1704+608	E	−23.95 ± 0.34	−24.60 ± 0.37	0.65 ± 0.50	11.14 ± 0.51	−23.95 ± 0.34	−24.60 ± 0.37	0.65 ± 0.50	11.14 ± 0.51
PG 2112+059	E	−22.70 ± 0.99	−23.57 ± 0.53	0.87 ± 1.12	10.83 ± 0.87	−22.70 ± 0.99	−23.57 ± 0.53	0.87 ± 1.12	10.83 ± 0.87
PG 2214+139	E	−21.36 ± 0.35	−23.24 ± 0.32	1.88 ± 0.47	11.14 ± 0.48	−21.36 ± 0.35	−23.24 ± 0.32	1.88 ± 0.47	11.14 ± 0.48
PG 2251+113	E	−23.35 ± 0.34	−24.37 ± 0.34	1.02 ± 0.48	11.22 ± 0.49	−23.35 ± 0.34	−24.37 ± 0.34	1.02 ± 0.48	11.22 ± 0.49

Note. Column (1): object name. Column (2): galaxy morphology: D = unbarred disk galaxy; B = barred disk galaxy; E = elliptical galaxy; M = merger; U = uncertain. Column (3): B-band absolute magnitude of the bulge. Column (4): I-band absolute magnitude of the bulge. Column (5): B − I color of the bulge. Column (6): mass of the bulge. Column (7): B-band absolute magnitude of the host galaxy. Column (8): I-band absolute magnitude of the host galaxy. Column (9): B − I color of the host galaxy. Column (10): total mass of the host galaxy.

A machine-readable version of the table is available.

Download table as:  DataTypeset image

Despite the large uncertainties of the stellar masses, it is reassuring that the bulge masses are always smaller than the total masses of the host (Figure 6(a)). The median bulge-to-total stellar mass ratio is ∼0.4. While this may seem an obvious point, it does offer some measure of reassurance that our masses are not unreasonable despite the challenges associated with estimating M/L for the bulge and the galaxy separately, which can occasionally produce bulge masses that formally exceed those of the whole galaxy (Bentz & Manne-Nicholas 2018; Li et al. 2021). Prior to this study, the work of Zhang et al. (2016) has provided the largest collection of (total) stellar masses for the PG quasars, derived from assembling heterogeneous, literature-based HST photometry of 55 sources assuming a fixed M/L for the optical or near-IR. We compare our host galaxy stellar masses with those of Zhang et al. (2016) in Figure 6(b), for the subset of 22 overlapping sources. ¹² The two sets of stellar masses show overall agreement, with most sources lying within an envelope of ±0.3 dex. The most deviant points are PSF-dominated sources and mergers. The most striking example is PG 0953+414, for which our stellar mass of ${M}_{* }=1.9\times {10}^{10}\,{M}_{\odot }$ is ∼10 times lower than that reported by Zhang et al. (2016). The optical photometry used by Zhang et al. (2016) originated from the analysis of an HST WFPC2/F675W image by Hamilton et al. (2002), who seem to have severely overestimated the flux of the host (Kim et al. 2017), likely on account of the extremely dominant nuclear emission of PG 0953+414. Similarly, our stellar mass for PG 1700+518 is a factor of ∼3 lower than that reported by Zhang et al. (2016), who based it on an HST NICMOS/F160W observation. Bentz & Manne-Nicholas (2018) analyzed the same image for this source and found a stellar mass similar to ours.

Zoom In Zoom Out Reset image size

Following Ho & Kim (2015), BH mass is derived as

$$\begin{eqnarray}\begin{array}{rcl}\mathrm{log}\left(\displaystyle \frac{{M}_{\mathrm{BH}}}{{M}_{\odot }}\right) & = & \mathrm{log}\left[{\left(\displaystyle \frac{{\mathrm{FWHM}}_{{\rm{H}}\beta }}{1000\mathrm{km}{{\rm{s}}}^{-1}}\right)}^{2}{\left(\displaystyle \frac{{L}_{5100}}{{10}^{44}\mathrm{erg}{{\rm{s}}}^{-1}}\right)}^{0.533}\right]\\ & & +\,7.03,\end{array}\end{eqnarray} \tag{ 3 }$$

where we assume, for concreteness, the virial factor for classical bulges. ¹³ The uncertainty of BH masses is 0.35 dex. In Equation (3), ${\mathrm{FWHM}}_{{\rm{H}}\beta }$ gives the line width of broad Hβ derived from the single-epoch spectra of Boroson & Green (1992), and ${L}_{5100}\equiv \lambda {L}_{\lambda }(5100\,\mathring{A})$, with ${L}_{\lambda }(5100\,\mathring{A})$ the specific luminosity of the nonstellar continuum at 5100 Å.

We obtain ${L}_{5100}$ from our image analysis, converting the nuclear magnitude from the B-band GALFIT decomposition to rest-frame 5100 Å flux density assuming a power-law spectrum ${f}_{\nu }\propto {\nu }^{-0.44}$ (Vanden Berk et al. 2001) and correcting for Galactic extinction (Schlafly & Finkbeiner 2011). Uncertainties arising from variations in PSF model and background subtraction (Section 3.2) are typically less than 0.05 dex. As shown in Figure 7, our values of ${L}_{5100}$ agree within ∼0.3 dex with those given by Vestergaard & Peterson (2006), who made use of the ground-based spectrophotometry from Neugebauer et al. (1987). The lower luminosity sources with ${L}_{5100}\,\lesssim {10}^{44.5}\,\mathrm{erg}\,{{\rm{s}}}^{-1}$, predominantly narrow-line Seyfert 1s (NLS1s; ${\mathrm{FWHM}}_{{\rm{H}}\beta }\leqslant 2000\,\mathrm{km}\,{{\rm{s}}}^{-1};$ Osterbrock & Pogge 1985), are systematically brighter in Neugebauer et al.'s measurements, no doubt because of the significant host galaxy light included in the large, 15'' diameter aperture used for the ground-based observations. The extreme outlying point for PG 1149-110 requires another explanation. Its nucleus is quite modest in strength compared to the host in the B-band and I-band HST images, both much fainter than reported in earlier observations (e.g., Neugebauer et al. 1987; Jahnke & Wisotzki 2003). Inspection of multi-epoch photometry in the W1 (3.4 μm) and W2 (4.6 μm) bands recorded during 2011–2019 by the Wide-field Infrared Survey Explorer (Wright et al. 2010) reveals that the light curves of PG 1149-110 reached a minimum in 2014, close in time to when our HST observations were conducted, and then quickly rose by more than 0.5 mag. This level of near-IR variability, large even by the standards of "changing-look" AGNs, likely induces a brightnening of ≳1 mag in the optical (Yang et al. 2018). Therefore, we suspect that for PG 1149-110 intrinsic variability is the culprit for the large difference between our measurement of its nuclear brightness and that recorded more than 20 yr earlier. The BH mass for this source was computed using the value of ${L}_{5100}$ from Vestergaard & Peterson (2006).

Zoom In Zoom Out Reset image size

Figure 8(a) shows the relation between BH mass and bulge stellar mass for our sample of PG quasars. For comparison, we overplot the M_BH–M_bulge relation of nearby, inactive elliptical galaxies and classical bulges from KH13, as well as the revised version of this relation that replaces ellipticals by their "core," the component most closely linked to the compact, high-redshift progenitors of present-day ellipticals, which were most directly related to the growth phase of the BH (ZHG21). As with KH13, the zero-point of the M/L used by ZHG21 was calibrated dynamically and does not depend on the choice of IMF. Within the relatively large scatter of the quasar sample, both of the relations for inactive galaxies intersect our data points for the PG quasars, especially for the group BHs more massive than ${M}_{\mathrm{BH}}\gtrsim {10}^{8}\,{M}_{\odot }$. By contrast, the objects with ${M}_{\mathrm{BH}}\lesssim {10}^{7.5}\,{M}_{\odot }$, which are predominantly NLS1s hosted by disk, and especially barred, galaxies, depart notably from the M_BH–M_bulge or M_BH–M_core relations of their more massive counterparts. This is qualitatively consistent with the known behavior of pseudo bulges in inactive galaxies (KH13 ), and suggests that the lower mass BHs reside in pseudo bulges.

Zoom In Zoom Out Reset image size

While BH mass correlates most closely with bulge stellar velocity dispersion and stellar mass (KH13), its relation with the total galaxy stellar mass (Reines & Volonteri 2015), albeit looser, nonetheless still offers a pragmatic and observationally less demanding empirical tool to investigate BH-galaxy coevolution. Figure 8(b) illustrates that the PG quasars obey the M_BH–M_total relation of inactive galaxies, especially when separately divided into early-type galaxies and late-type galaxies, as recently calibrated by GSH20. Note that GSH20 derived stellar masses using the color-based M/L from Bell et al. (2003), which, on account of the "diet Salpeter" IMF assumed, need to be scaled by 0.06 dex to be consistent with our adopted Chabrier (2003) IMF.

The most luminous quasars in our sample are mainly at $z\gtrsim 0.2$ and are thus not as well resolved. However, so long as they are detected robustly, their massive hosts can be securely classified as elliptical galaxies (Section 4.1). Moreover, these host galaxies usually display faint surface brightnesses and large effective radii that follow the Kormendy (1977) relation of inactive elliptical galaxies (Section 5.3). Therefore, we conclude that most of the luminous (${M}_{B}\lt -23.5$ mag or ${L}_{5100}\gtrsim {10}^{45}\,\mathrm{erg}\,{{\rm{s}}}^{-1}$) quasars in our sample likely reside in elliptical galaxies, consistent with early HST observations (Disney et al. 1995; Bahcall et al. 1997; McLure et al. 1999; Dunlop et al. 2003).

Figure 9 compares the distribution of the B/T of our sample with that of the 235 type 1 AGNs studied by Kim et al. (2017), which, while heterogeneous, is the largest and most representative sample to date of nearby ($z\lesssim 0.35$) AGNs with detailed morphological decomposition based on HST imaging. The Anderson–Darling test (Scholz & Stephens 1987) cannot reject the null hypothesis that the samples are drawn from the same parent distribution (p-value > 0.25). Our sample contains a moderately higher fraction of elliptical ($B/T=1$) galaxies, likely because our sample extends to slightly higher redshift (∼0.4) than that of Kim et al. (2017), and it comprises among the most luminous sources within its redshift range. As in Kim et al. (2017), mergers make up only a minority (17%; 6/35) of the sample, emphasizing that mergers are not a prerequisite for triggering quasar-level activity in galaxies (Villforth et al. 2014; Hewlett et al. 2017).

Zoom In Zoom Out Reset image size

Quasars hosted by disk galaxies exhibit relatively modest bulge fractions ($B/T\lesssim 0.5$) and low bulge Sérsic indices ($n\lesssim 2$). Those that are barred, in particular, have the lowest B/T (Figure 10), consistent with the overall trend observed among nearby inactive galaxies (gray symbols) from the Carnegie-Irvine Galaxy Survey (Ho et al. 2011) with detailed morphological decomposition (Gao et al. 2019). ¹⁴ In contrast, the elliptical galaxies and major mergers have large Sérsic indices. While rigorous distinction between pseudo and classical bulges is still controversial (Gao et al. 2020 and references therein), pseudo bulges are generally characterized by a low B/T and low Sérsic n. Adopting, for simplicity, $n\lt 2$ as a criterion for recognizing pseudo bulges, 82% (14/17) of the PG quasars hosted by disk galaxies contain a pseudo bulge.

Zoom In Zoom Out Reset image size

Our sample contains a total of nine NLS1s (Figure 10). All but one reside in disk galaxies, among which 50% (4/8) are barred. This fraction is consistent with that of Kim et al. (2017) but lower than that of earlier studies of (lower luminosity) Seyfert galaxies (e.g., 65%–85%; Crenshaw et al. 2003; Ohta et al. 2007). The bar fraction of inactive disk galaxies decreases with increasing galaxy stellar mass (Díaz-García et al. 2016; Li et al. 2017; Erwin 2018; Zhao et al. 2020), averaging ∼50% for stellar masses $\sim {10}^{10}\mbox{--}{10}^{11}\,{M}_{\odot }$, similar to our finding. The higher bar fraction reported in previous NLS1 samples likely can be attributed to their lower typical galaxy masses. The lower bar fraction of 22% (2/9) among the targets with ${\mathrm{FWHM}}_{{\rm{H}}\beta }\gt 2000\,\mathrm{km}\,{{\rm{s}}}^{-1}$ ("broad-line" objects), while in agreement with previous studies (Crenshaw et al. 2003; Ohta et al. 2007; Kim et al. 2017), nevertheless may be underestimated by observational bias (Erwin 2018) due to the larger distances of the sample (median $z\approx 0.14$ compared to $z\approx 0.05$ for the NLS1s).

With $R\equiv {L}_{\nu }(6\,\mathrm{cm})/{L}_{\nu }(B)\gt 100$, six of the quasars in our sample (Table 1) can be considered radio loud (Kellermann et al. 1989). Apart from a single exception in a merger, the rest of the radio-loud sources are hosted by massive ellipticals, which are systematically more massive than their radio-quiet counterparts. Dunlop et al. (2003) and Kim et al. (2017) reached similar conclusions based on more extensive statistics.

The effective radius (R_e ) forms a strong inverse correlation with the mean surface brightness within R_e in elliptical galaxies (Kormendy 1977) and classical bulges (Carollo 1999). By contrast, pseudo bulges depart markedly from the Kormendy relation by manifesting lower $\langle {\mu }_{e}\rangle$ at a given R_e (Kormendy & Kennicutt 2004; Gadotti 2009; Fisher & Drory 2010; Gao et al. 2020). Early HST observations revealed that quasar host galaxies, when properly decomposed from the bright nucleus, follow the Kormendy relation of inactive galaxies (Dunlop et al. 2003). However, with the vastly expanded and more diverse sample of type 1 AGNs of Kim et al. (2017), Kim & Ho (2019) found that the bulges of AGN host galaxies tend to be more luminous than those of inactive galaxies of the same bulge size, concluding that their bulge surface brightnesses are likely enhanced by newly formed stars associated with the current episode of BH accretion.

We calculate the mean effective surface brightness by averaging the flux inside the effective radius of the bulge,

$$\begin{eqnarray}&&\langle {\mu }_{e,I}\rangle ={M}_{I}+5\mathrm{log}\left(\displaystyle \frac{{R}_{e}}{10\,\mathrm{pc}}\right)+28.57,\end{eqnarray} \tag{ 4 }$$

where M_I is absolute magnitude of the bulge. As shown in Figure 11, the bulges of PG quasars generally overlap with the distributions defined by the classical bulges of inactive galaxies and AGNs studied by Kim & Ho (2019). At the same time, the quasar hosts designated by us as ellipticals in this study indeed possess large effective radii and correspondingly faint surface brightnesses similar to their inactive counterparts (Kormendy et al. 2009; Huang et al. 2013; Gao et al. 2020), which lends confidence to our morphological assignment. Major mergers tend to have systematically higher surface brightnesses than the rest of the sample, likely a reflection of boosted central brightness from merger-induced starbursts, by analogy with ultraluminous IR galaxies (Rothberg et al. 2013). At a given R_e , pseudo bulges in inactive galaxies display lower surface brightness than classical bulges or elliptical galaxies (Kormendy & Kennicutt 2004; Gadotti 2009; Fisher & Drory 2010; Gao et al. 2020). Judging by a variety of evidence (disk morphologies, presence of a bar, low B/T, low n), a number of the PG quasars in our sample, particularly those classified as NLS1s, should possess pseudo bulges, but, intriguingly, they are located mostly above the Kormendy relation of pseudo bulges. As for the mergers, the simplest explanation for this trend is to posit that the AGNs hosted by pseudo bulges have central surface brightnesses enhanced by recent or ongoing star formation. Zhao et al. (2019) reached the same conclusion studying the HST images of nearby type 2 quasars.

Zoom In Zoom Out Reset image size

The results presented here, in concert with the analysis of Kim & Ho (2019), paint a consistent picture. The host galaxies of quasars, most notably those with accretion rates high enough to be NLS1s, which tend to be relatively late-type, often barred, disk galaxies with pseudo bulges, show evidence of enhanced young stellar populations relative to their inactive counterparts. Mergers do as well—whether they be active or not—when compared to classical bulges and ellipticals. Nearby AGNs, notably those powerful enough to qualify as quasars, lack neither gas nor the ability to form stars (e.g., Husemann et al. 2017; Shangguan et al. 2018, 2020a; Jarvis et al. 2020; Yesuf & Ho 2020; Zhuang & Ho 2020; Xie et al. 2021). BH accretion and host galaxy star formation are tightly linked by the molecular gas (Shangguan et al. 2020b; Zhuang et al. 2021), which may be preferentially more centrally concentrated (Molina et al. 2021).