THE MASS OF THE CENTRAL BLACK HOLE IN THE NEARBY SEYFERT GALAXY NGC 5273

Over the last ∼25 yr, observational and computational studies have led to the widely accepted belief that supermassive black holes (black holes with masses M_BH = 10⁶–10¹⁰ M_☉) play a significant role in galaxy evolution and cosmology. To truly understand the physical processes at play in an active galactic nucleus (AGN) and the effects on the host galaxy and its surroundings, it is necessary to know the mass of the central black hole that is involved.

Currently, it is only feasible to directly measure black hole masses in relatively nearby galaxies through dynamical modeling (which is limited by spatial resolution) or reverberation mapping (which is limited by time). For quiescent galaxies, M_BH is derived from modeling the dynamics of stars and/or gas on parsec scales in a galaxy nucleus (e.g., Gültekin et al. 2009; McConnell & Ma 2013). For AGNs, the response of milliparsec-scale gas in the broad-line region (BLR) to the accretion disk flux variability probes the gravitational force of the black hole in a technique known as reverberation mapping (e.g., Peterson et al. 2004). Dynamical studies rely on spatial resolution and are therefore limited to the most nearby galaxies (≲ 100–150 Mpc). Reverberation mapping relies on temporal resolution and is not inherently distance limited, but timing arguments and resource availability have generally limited reverberation studies to z < 0.1 (∼400 Mpc). The M_BH measurements that result from directly probing the gravity of the supermassive black hole through these techniques provide the foundation for all black hole scaling relations used to investigate M_BH at cosmological distances. As such, the M_BH estimates provided by the scaling relations are directly affected by the quality of the observations and measurements used to calibrate the relationships.

Both the dynamical and reverberation masses currently suffer from several inherent uncertainties and potential systematic biases (for recent discussions, see Graham et al. 2011 for the dynamical studies and Peterson 2010 for the reverberation studies). One of the most pernicious uncertainties affecting reverberation masses is the exact translation from reverberation measurements to calibrated black hole mass. The so-called "f factor" is an order unity scaling factor that depends on the detailed geometrical and dynamical properties of the BLR gas that is used to probe the gravitational influence of the black hole. Recent progress has shown promise in constraining the "f factor" for individual reverberation targets with high-quality data sets (Pancoast et al. 2014).

What is currently lacking, however, is a set of black holes with both dynamical and reverberation masses that can be directly compared to ensure that all black hole masses are on the same mass scale. Only three AGNs in the current reverberation sample are close enough and have massive enough black holes that they can be investigated through stellar dynamical modeling—NGC 3227 (Davies et al. 2006), NGC 4151 (Onken et al. 2004, 2014), and NGC 6814 (Bentz et al. 2009b for the reverberation mass; observations of the dynamics in the galaxy nucleus are in progress at this time). Comparison of black hole masses derived from independent techniques in the same galaxies is an important cross check, because these black hole masses are the foundation upon which all scaling relationships are built. Stellar dynamical modeling, in particular, is often described as the "gold standard" of supermassive black hole masses. However, all stellar dynamical masses result from axisymmetric modeling codes that are unable to account for the unique dynamics that arise in the presence of a bar. NGC 3227, NGC 4151, and NGC 6814 are all late-type spiral galaxies with bars. Recently, Onken et al. (2014) showed that even the weak bar in NGC 4151 can induce a strong bias in the resultant black hole mass from stellar dynamical modeling, causing the black hole mass to be overestimated. This bias would not be expected to affect the reverberation masses, where the tracer particles or "test masses" are located much deeper in the gravitational potential well of the black hole, at distances of only a few light days (∼0.01 pc for nearby Seyferts), where they would be unaffected by the dynamics of a galaxy-scale bar.

Motivated by the small number of AGNs with reverberation masses that are also suitable targets for dynamical studies, we have begun a program to identify and monitor all AGNs that may be suitable for both reverberation and dynamical studies. The rarity of broad-lined AGNs in the local universe is the fundamental limitation in this endeavor, as this rarity generally translates to a large distance and therefore an inability to spatially resolve the galaxy nucleus on the scales necessary for dynamical modeling. Our goal is to build the largest sample possible, although the sample is not likely to be larger than ∼10 AGNs given the current technological constraints. In this manuscript, we present a reverberation-based mass for the central black hole in NGC 5273, the first reverberation target in an early-type unbarred galaxy at a distance suitable for ground-based dynamical studies. In Section 2, we describe the observations, and in Section 3, the light curve analysis. Section 4 describes the line-width measurements and the black hole mass determination is described in Section 5. Finally, we discuss the results of our analysis and summarize our findings in Sections 6 and 7.

NGC 5273 is a broad-lined Seyfert located at R.A. = 13^h42^m083, decl. = +35°39'15'', and z = 0.00362. The host-galaxy morphology is S0 and it exhibits a faint spiral structure. The distance to NGC 5273 has been determined from surface brightness fluctuations of the host galaxy (Tonry et al. 2001, recalibrated by Tully et al. 2008) and is found to be 16.5 ± 1.6 Mpc.

2.1. Spectroscopy

Spectroscopic monitoring of NGC 5273 was carried out at the Apache Point Observatory (APO) 3.5 m telescope from 2014 May 14–July 1 (UT dates here and throughout). The monitoring program was scheduled for the first hour of almost every night during this time period, resulting in observations being taken during evening twilight. The Dual Imaging Spectrograph (DIS) was employed with the low-resolution (B400 and R300) gratings and the 5'' slit rotated to a position angle of 0° (oriented N–S). With central wavelengths of 4398 Å and 7493 Å, the spectra span the entire optical bandpass from the atmospheric cutoff to 1 micron with a nominal dispersion of 1.8 Å pixel⁻¹ and 2.3 Å pixel⁻¹, respectively. A single spectrum with a typical exposure time of 420 s was obtained each night the weather permitted, except for a span of four nights in June when the target was above the 85° altitude limit of the telescope during the time constraints of our program. The spectrophotometric standard star, Feige 66, was also observed throughout this campaign, as were two additional AGNs (we will describe their results elsewhere).

Spectra were reduced and flux calibrated following standard procedures. An extraction width of 12 pixels was used, corresponding to an angular width of 5'' and 48 for the blue and red sides, respectively. The regions for background subtraction determination were set on either side of the spectrum between 75–95 pixels (∼30''–40'') from the nucleus.

Spectroscopic monitoring campaigns generally do not enjoy the luxury of photometric conditions every night, so the rough flux calibration obtained with a spectrophotometric standard star is not accurate enough to measure the few-percent variations we expect during a reverberation campaign. A final, internal calibration of the spectra is required. We follow the typical procedure, which is to intercalibrate the spectra with the scaling method of van Groningen & Wanders (1992) by assuming a constant flux in the narrow emission lines. A reference spectrum is built from the user-determined best spectra, and the differences between each individual spectrum and the reference are minimized in the λλ4959, 5007 region. This method accounts for small differences in wavelength calibration, spectral resolution, and flux calibration from night to night and has been shown to result in relative spectrophotometry that is accurate to 2% (Peterson et al. 1998a). While variability of AGN narrow emission lines has been detected (Peterson et al. 2013), the timescale of variation was much longer (a few years) than the timescales probed by typical reverberation-mapping campaigns such as the one we present here. For the red spectra, we instead used the [S ii] doublet for the internal calibration. The weakness of these emission lines relative to Hα resulted in a less accurate calibration for the spectra and noisier light curves, but we were still able to measure a mean time delay, as we discuss in Section 3.

Figure 1 shows the final mean and rms of the 30 calibrated spectra collected over the seven week period of our monitoring program. The Balmer lines Hα, Hβ, and Hγ, as well as He ii λ4686 are clearly visible in the rms spectrum, evidencing their large variability during the monitoring campaign. Based on the nights where our campaign enjoyed the best weather and observing conditions, we measured a mean integrated flux of f = 93.2 ± 0.6 × 10⁻¹⁵ erg s⁻¹ cm⁻² for the [O iii] λ5007 emission line, which agrees well with the value of 89.9 × 10⁻¹⁵ erg s⁻¹ cm⁻² determined by the Sloan Digital Sky Survey (Ahn et al. 2014) and sets the absolute flux scale of our spectra.

Zoom In Zoom Out Reset image size

2.2. Photometry

Emission-line light curves were determined from the final, scaled spectra by fitting a local, linear continuum under each emission line and integrating the line flux above this continuum. We do not follow the procedure of fitting the entire spectrum with model components because the results are very sensitive to the exact parameters that are included and how they are modeled (Denney et al. 2009). The light curve for the continuum flux density at 5100 Å × (1 + z) was also measured. Table 1 gives our tabulated spectroscopic and photometric light curves.

The V-band filter (λ_c = 5483 Å and Δλ = 827 Å) does not include a significant contribution from any of the broad emission lines in the spectrum of NGC 5273, so we combined the V-band photometry with the 5100 Å continuum flux densities. We identified pairs of points in the two light curves that were contemporaneous within 0.75 days. A linear relationship was then fit to the pairs of points to determine the multiplicative and additive factors necessary to bring the V-band fluxes into agreement with the 5100 Å fluxes, taking into account the differences in galaxy background light and bandpass.

We then compared the merged continuum light curve with the Gunn–Thuan g- and r-band photometry from APO. We detected no significant time delay between the continuum and the g band, so we merged those light curves together in the same way. However, we did detect a small time delay between the continuum light curve and the r-band photometry (see Figure 2 and Table 3, so we did not merge the r band with the continuum light curve. The small delay determined for the r band is likely due to the strong contribution to the bandpass from the broad Hα emission line (∼90% of the r-band flux, see Table 2). A similar effect would be expected from Hβ for the g band, but the smaller equivalent width of Hβ relative to the total bandpass (∼40% of the g-band flux) is likely the reason we did not detect any time delay for g.

Zoom In Zoom Out Reset image size

Table 3. Time Lags, Line Widths, and Virial Products

	Time Lags				Mean Line Width			rms Line Width			$\frac{c \tau _{\rm cent} \sigma _{\rm line,rms}^2}{G}$
Line	τcent	τpeak	τJAV		σline	FWHM		σline	FWHM
	(days)	(days)	(days)		(km s−1)	(km s−1)		(km s−1)	(km s−1)		(106 M☉)
Hα	2.06$^{+1.42}_{-1.31}$	2.75$^{+1.00}_{-2.00}$	$2.44^{+0.15}_{-0.94}$		1781 ± 36	3032 ± 54		1783 ± 66	3579 ± 145		$1.28^{+0.89}_{-0.82}$
Hβ	2.22$^{+1.19}_{-1.61}$	2.50$^{+1.25}_{-2.00}$	$1.45^{+1.09}_{-0.15}$		1821 ± 53	5688 ± 163		1544 ± 98	4615 ± 330		$1.03^{+0.57}_{-0.76}$
Hγ	2.14$^{+1.09}_{-1.08}$	2.00$^{+1.75}_{-1.00}$	$1.32^{+0.17}_{-0.67}$		1378 ± 55	4015 ± 101		1600 ± 54	4231 ± 604		$1.06^{+0.55}_{-0.54}$
He ii	0.35$^{+2.17}_{-1.66}$	0.00$^{+3.00}_{-1.75}$	$-0.24^{+0.26}_{-0.28}$		⋅⋅⋅	⋅⋅⋅		2201 ± 102	4204 ± 882		$0.41^{+2.50}_{-1.92}$
r	0.35$^{+0.47}_{-0.59}$	0.00$^{+0.50}_{-0.25}$	$0.12^{+0.13}_{-0.13}$		⋅⋅⋅	⋅⋅⋅		⋅⋅⋅	⋅⋅⋅		⋅⋅⋅

Note. Time lags are presented in the observed frame, but are corrected for cosmological dilation when calculating the virial product—a tiny effect that is much smaller than the uncertainties in the case of NGC 5273 with z = 0.00362.

Download table as:  ASCIITypeset image

Figure 2 shows the light curves for the continuum (including the V-band and g-band photometry), the r-band photometry, and for the optical recombination lines Hα, Hβ, Hγ, and He ii. NGC 5273 exhibited fairly strong variability over the course of the monitoring campaign. Table 2 gives the variability statistics for the final light curves displayed in Figure 2. Column 1 lists the spectral feature and Column 2 gives the number of measurements in the light curve. Columns 3 and 4 list the average and median time separation between measurements, respectively. Column 5 gives the mean flux and standard deviation of the light curve, and Column 6 lists the mean fractional error (based on the comparison of observations that are closely spaced in time). Column 7 lists the excess variance, computed as

$$\begin{equation} F_{\rm var} = \frac{\sqrt{\sigma ^2 - \delta ^2}}{\langle F \rangle }, \end{equation} \tag{ 1 }$$

where σ² is the variance of the fluxes, δ² is their mean-square uncertainty, and 〈F〉 is the mean flux. Column 8 is the ratio of the maximum to the minimum flux in the light curve. In general, the true level of variability was somewhat higher than these measurements describe if taken at face value. The continuum light curve includes the host-galaxy starlight contribution, as does the r-band light curve. The broad emission-line light curves include the contribution from the narrow emission lines. These non-variable components serve to dampen the variations we observe in the light curves.

To determine the average time lag of the emission lines relative to the continuum, we first cross-correlated the light curves using the interpolation cross-correlation method (Gaskell & Sparke 1986; Gaskell & Peterson 1987) with the modifications of White & Peterson (1994). The method determines the cross-correlation function (CCF) twice: first, by interpolating the continuum light curve, and then by interpolating the emission-line light curve. The final CCF is the average of the two. We characterize the results of the CCF by recording the maximum value of the CCF (r_max), the time delay of the CCF maximum (τ_peak) and the centroid of the points about the peak (τ_cent) above a threshold value of 0.8r_max. CCFs for each light curve relative to the continuum are displayed in Figure 2.

To quantify the uncertainties on the time lag measurements, we employ the Monte Carlo "flux randomization/random subset sampling" method of Peterson et al. (1998b, 2004). From the N available data points, a selection of N points is chosen without regard to whether a datum has been previously selected. The uncertainty on a point that is sampled 1 ⩽ n ⩽ N times is scaled by a factor of n^1/2 and the typical number of points that is not sampled in any specific realization is ∼1/e. This is the "random subset sampling," and it quantifies the uncertainty that arises from the inclusion of any specific data point in the light curve. The flux values in the selected subset are then modified by a Gaussian deviation of the flux uncertainty. This is the "flux randomization" and it accounts for the measurement uncertainties. The final sampled and modified light curves are then cross-correlated and the CCF measurements are recorded. The process is repeated many times (N = 1000) and a distribution of CCF measurements are built up (see the right-hand panels in Figure 2). We take the means of the cross-correlation centroid distribution and the cross-correlation peak distribution as τ_cent and τ_peak, respectively. The uncertainties on τ_cent and τ_peak are determined so that 15.87% of the realizations fall above and 15.87% fall below the range of uncertainties, corresponding to ±1σ for a Gaussian distribution.

Table 3 lists the measured time lags and their uncertainties for the emission lines and for the r band relative to the continuum light curve. We also investigated the time lags with the software package JAVELIN, previously known as SPEAR (Zu et al. 2011). JAVELIN uses a damped-random walk to fit the continuum light curve, and then determines the best model for the reprocessed (shifted and smoothed) emission-line light curves by maximizing the likelihood of the model. Uncertainties in the time delay are determined through a Bayesian Markov Chain Monte Carlo method.

Unfortunately, we were unable to obtain reasonable constraints on the smoothing and delay parameters when simultaneously modeling multiple emission-line light curves with JAVELIN. Instead, we ran multiple models and each time we focused on modeling the continuum and a single emission line. Figure 3 shows the modeling results obtained for the Hβ emission-line light curve. The time lags we obtained with JAVELIN were generally consistent with those determined through cross-correlation methods (see Table 3), but the difficulties we encountered leads us to focus on the time lags obtained through cross-correlation methods throughout the remainder of this manuscript.

Zoom In Zoom Out Reset image size

Finally, we can compare the time delay measured for Hβ with that expected from the relationship between emission-line time delay and AGN luminosity (the R_BLR–L relationship) determined for the full reverberation sample of AGNs. The 5100 Å luminosity of NGC 5273 as determined from the mean spectrum contains a large contribution from host-galaxy starlight. Fortuitously, the Hubble Space Telescope archive contains WFPC2 imaging of NGC 5273 through the F547M filter, which allowed us to determine the starlight correction and deduce the AGN luminosity. Following the methods of Bentz et al. (2013), we modeled the surface brightness distribution of the drizzled and combined WFPC2 image in two dimensions with Galfit (Peng et al. 2002, 2010). The galaxy was well fit with an exponential disk and a Sérsic bulge with index n = 3.3 (see Figure 4). The modeling allowed us to accurately subtract the AGN contribution and create a "PSF-free" image of NGC 5273, from which we determined the starlight flux through the monitoring aperture of (4.80 ± 0.48) × 10⁻¹⁵ erg s⁻¹ cm⁻¹ Å at 5100 × (1 + z) Å. The mean flux at 5100 Å (as shown in Table 2) was corrected for the starlight contribution, giving an AGN luminosity of log L_AGN/erg s⁻¹ = 41.534 ± 0.144. Figure 5 shows our Hβ time delay measurement and starlight corrected luminosity for NGC 5273, which agrees well with the expectations from the R_BLR–L relationship.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The width of a broad emission line is a measure of the line-of-sight velocity of the gas in the BLR. Narrow emission lines are emitted from gas that does not participate in the same bulk motion as the BLR gas, so we report the width of only the broad components of the emission lines.

Figure 1 shows the mean and rms spectra for NGC 5273 (dotted lines) as well as the narrow-line subtracted mean and rms spectra (solid lines). For the narrow-line subtraction, we used the [O iii] λ5007 emission line as a template for the λ4959 and Hβ narrow lines. The ratio of [O iii] λ4959/[O iii] λ5007 was set at 0.34 (Storey & Zeippen 2000), and we derived a ratio of Hβ/[O iii] λ5007 = 0.10. Due to the lack of a suitable narrow emission line in the red spectra for use as a template, we did not attempt any narrow-line subtraction with Hα.

The widths of the broad emission lines were measured in the narrow-line subtracted mean and rms spectra and are reported as two separate measures: the FWHM flux, and the line dispersion, σ_line, which is the second moment of the emission-line profile (Peterson et al. 2004). The uncertainties in the line widths were set using a Monte Carlo random subset sampling method. In this case, from a set of N spectra, a random subset of N spectra were selected without regard to whether a spectrum had previously been chosen, and mean and rms spectra were created from the chosen subset. The FWHM and the σ_line were measured and recorded for each realization, and distributions of line-width measurements were built up over 1000 realizations. The mean and standard deviation of each distribution are taken to be the line width and uncertainty, respectively. This method quantifies the weight that any individual spectrum has on the final line-width measurements. Additionally we also quantify the uncertainty from the exact placement of the continuum region on either side. For each line, we define a maximum continuum window (typically 30–50 Å wide) on either side of an emission line. For each realization, a subset of each continuum window of at least 7 pixels (∼14 Å) is randomly selected, from which the local linear continuum is fit. This additional step has little effect on the rms line widths, because their uncertainties are dominated by the noise in the spectra, but has a small effect that increases the uncertainties we measure for the line widths derived from the mean spectra (Bentz et al. 2009b).

Emission-line widths were corrected for the dispersion of the spectrograph following Peterson et al. (2004). Specifically, the observed line width Δλ_obs can be described as a combination of the intrinsic line width, Δλ_true, and the spectrograph dispersion, Δλ_disp, such that

$$\begin{equation} \Delta \lambda _{\rm obs}^2 \approx \Delta \lambda _{\rm true}^2 + \Delta \lambda _{\rm disp}^2. \end{equation} \tag{ 2 }$$

We determined Δλ_true by taking our measurement of the FWHM of [O iii] λ5007 as Δλ_obs = 14.26 Å and the FWHM measured by Whittle (1992) through a small slit and with a high resolution as Δλ_true = 130 km s⁻¹. We deduced a final spectral resolution of Δλ_disp = 14.1 Å, which we used to correct our line-width measurements. The final rest-frame, resolution-corrected line-width measurements determined from the mean and the rms spectra are listed in Table 3.

The black hole mass is generally determined from reverberation-mapping measurements as

$$\begin{equation} M_{\rm BH} = f \frac{c \tau V^2}{G}, \end{equation} \tag{ 3 }$$

where τ is the time delay for a specific emission line relative to continuum variations, and V is the line-of-sight velocity width of the emission line, with c and G being the speed of light and gravitational constants, respectively. The emission-line time delay is therefore a measure of the responsivity-weighted average radius of the BLR for the emission of a particular species (e.g., Hβ). Only gas that is optically thick to ionizing radiation will be included in this measure, because optically thin gas will not respond to changes in the ionizing flux and will not reverberate.

The factor f includes the details of the inclination of the system to our line of sight and the exact geometrical arrangement of the responding gas, as well as its kinematics. These details are generally unknown for any particular object because the gas only extends across angular scales of milliarcseconds or less from our vantage point. Current practice is to apply a population-averaged value, 〈f〉, to reverberation masses to get the overall mass scale correct for the sample as a whole. The value of 〈f〉 is determined from a comparison of the M_BH–σ_⋆ relationship for dynamical black hole masses and the M_BH–σ_⋆ relationship for AGNs with reverberation measurements. The overall multiplicative factor that must be applied to the AGN masses to bring the two relationships into agreement is taken to be 〈f〉. This value has varied in the literature from 5.5 (Onken et al. 2004) to 2.8 (Graham et al. 2011), depending on which objects are included and the specifics of the measurements. We adopt the value determined by Grier et al. (2013) of 〈f〉 = 4.3 ± 1.1 as it uses the most up-to-date set of measurements and includes four high-luminosity AGNs with large M_BH that better anchor the high-mass end of the AGN M_BH–σ_⋆ relationship, although we note that the Kormendy & Ho (2013) recalibration of M − σ for the dynamical sample suggests a slightly larger value for 〈f〉.

While the use of the M_BH–σ_⋆ relationship to set the absolute mass scale for reverberation masses has been fairly standard for the past 10 yr, we note that there are several potential problems with this method. In particular, several studies have uncovered large scatter and morphological biases in the M_BH–σ_⋆ relationship for quiescent galaxies (e.g., Hu 2008; Graham et al. 2011; Kormendy et al. 2011). There are several lines of evidence that demonstrate, however, that adopting a mean scaling factor from comparison of M_BH–σ_⋆ relationships does put reverberation masses in the correct vicinity. Bentz et al. (2009a) compared the M_BH–L_bulge relationship for reverberation-mapped AGNs and quiescent galaxies and found that the two relationships were consistent when an average scaling factor determined from the M_BH–σ_⋆ relationship was adopted. Pancoast et al. (2014) directly determined AGN black hole masses through dynamical modeling of reverberation-mapping spectra and found an average 〈f〉 = 4.8 for the five AGNs they examined. They were also able to constrain the AGN inclinations, one of the largest expected contributions to the value of 〈f〉, and found inclinations of 10°–50° to our line of sight. This agrees well with the inclination of ∼29° implied by a scaling factor of 〈f〉 = 4.3, if other effects are neglected. Additionally, Fischer et al. (2013) analyzed the spatially resolved biconical narrow-line regions of several nearby AGNs with three-dimensional geometric models and found that the Seyfert 1s in their sample had inclinations to our line of sight of 12°–49° with an average inclination of ∼24°. All of these independent studies suggest that adopting 〈f〉 = 4.3 for reverberation masses will result in a sample of unbiased masses, although the mass of any particular AGN is likely only accurate to a factor of 2–3.

Each emission line in our analysis provides an independent measurement of the black hole mass in NGC 5273. We can investigate the reliability of a reverberation-based black hole mass determination by comparing the results for all the emission lines. Figure 6 (left) shows the black hole mass derived from the time delay and line width of each of the optical recombination lines we were able to probe: Hα, Hβ, Hγ, and He ii. The right panel of Figure 6 shows the emission-line width versus the time delay for each line. The dotted line is a power law with slope of −0.5 which is the expected relationship if the gravity of the black hole dominates the dynamics of the gas we are probing. The results are consistent with the expectation, as has been seen for other reverberation studies (e.g., Peterson et al. 2004; Bentz et al. 2009b).

Zoom In Zoom Out Reset image size

Combining our time lags (τ_cent) and line widths (σ_{line, rms}) for each emission line and scaling by 〈f〉, we determine a final weighted mean of the black hole mass in NGC 5273 of M_BH = (4.7 ± 1.6) × 10⁶ M_☉.

The sphere of influence of the black hole is generally defined as

$$\begin{equation} r_{\rm h} = \frac{GM_{\rm BH}}{\sigma _{\star }^2}, \end{equation} \tag{ 4 }$$

and it is a useful metric for comparing with the spatial resolution of stellar dynamical observations, to determine whether or not a reliable black hole mass is likely to be obtained from dynamical modeling.

NGC 5273 was included in the ATLAS^3D (Cappellari et al. 2011) sample of early-type galaxies. Cappellari et al. (2013) derive a bulge stellar velocity dispersion of σ(R_e/8) = 74.1 ± 3.7 km s⁻¹ from integral-field spectroscopy of the galaxy. Combined with the black hole mass we have derived here, we estimate r_h = 005. This estimate depends rather sensitively on the specific value of 〈f〉 that is adopted. Figure 7 shows where NGC 5273 lies on the most recent M_BH–σ_⋆ relationship for AGNs, as determined by Grier et al. (2013). NGC 5273 sits slightly above the relationship, however it is within the scatter, lending credence to the value of M_BH determined here with 〈f〉 = 4.3.

Zoom In Zoom Out Reset image size

Current ground-based large-aperture telescopes are capable of obtaining spatial resolutions of ∼008 in the near-infrared with adaptive optics and integral-field units. The James Webb Space Telescope (JWST) will be able to achieve a similar resolution, but will have the added advantage of being located above the Earth's atmosphere, allowing for a much lower background and a more stable PSF with a significantly higher Strehl ratio. Gültekin et al. (2009) show that a strict insistence on resolving r_h is not necessary for determination of an accurate stellar dynamical mass, which seems to be evidenced by the quality of the dynamical mass obtained for NGC 3227 (where r_h is not resolved; Davies et al. 2006). The early-type host galaxy of NGC 5273 makes it the ideal candidate for stellar dynamical modeling in order to directly test the masses obtained from reverberation mapping and those obtained from stellar dynamical modeling.

It is interesting that NGC 5273 sits above the relationship, as it is hosted by an early-type galaxy. The AGNs sitting slightly below the relationship at similar black hole masses all live in barred spiral galaxies, and Hu (2008) and Graham et al. (2011) have found that barred or pseudobulge galaxies have black hole masses that lie preferentially below the M_BH–σ_⋆ relationship for early-type galaxies. It is currently unclear, however, whether this is because the black holes are undermassive or because the bar dynamics are artificially broadening the bulge stellar absorption line signatures in the spectra. The majority of the bulge stellar velocity dispersions in the AGN sample with reverberation masses are determined from fitting the stellar absorption lines in a single long slit spectrum, so it is likely that there is contamination from the bar dynamics in many of the stellar velocity dispersion measurements for the AGNs. Grier et al. (2013) attempted to quantify such a bias, but were unable to detect it among the current sample. On the other hand, Onken et al. (2014) found that using an axisymmetric dynamical modeling code to determine a stellar dynamical black hole mass could lead to a biased black hole mass measurement in barred galaxies, although in the case of NGC 4151, the object in their study, the bias led to an overestimate of M_BH, not an underestimate. Clearly, additional study is necessary to determine the exact effects caused by galaxy bars on black hole mass determinations, and NGC 5273 is an important addition to the reverberation sample given its proximity and the unbarred early-type morphology of its host galaxy.

We have carried out a spectroscopic and photometric monitoring campaign of the AGN in the nearby Seyfert galaxy NGC 5273. From the time delays measured between the broad optical recombination lines and the continuum flux, we determine a black hole mass of M_BH = (4.7 ± 1.6) × 10⁶ M_☉. Combined with the bulge stellar velocity dispersion, we estimate that the black hole sphere of influence for NGC 5273 should be just at the limit of the resolution achievable with current ground-based instrumentation and with the integral field capabilities of JWST. NGC 5273 is the newest addition to a very short list of AGNs with reverberation-based masses and black hole spheres of influence capable of being probed with current technology. It is also the only one of these few AGNs in an early-type unbarred galaxy, which makes NGC 5273 an obvious candidate for stellar dynamical modeling for a direct comparison of the black hole masses determined from different techniques.

We thank the anonymous referee for comments and suggestions that improved the presentation of this manuscript. We thank Chris Onken, Kelly Denney, Kate Grier, Gisella de Rosa, and Ying Zu for helpful feedback and discussions. This research is based on observations obtained with the Apache Point Observatory 3.5 m telescope, which is owned and operated by the Astrophysical Research Consortium. We heartily thank the staff at APO—especially telescope operators Alaina, Alysha, Jack, and Russet—for all their help with this program. It was a pleasure to interact with you all every night for two months! M.C.B. gratefully acknowledges support from the NSF through CAREER grant AST-1253702. This research has made use of the NASA/IPAC Extragalactic Database (NED) which is operated by the Jet Propulsion Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space Administration and the SIMBAD database, operated at CDS, Strasbourg, France.