Circumnuclear Structures in Megamaser Host Galaxies

TL;DR

This study uses HST imaging to analyze circumnuclear structures in megamaser host galaxies, revealing the near-ubiquity of disk-like features at small scales and their varied orientations relative to larger galactic structures.

Contribution

It doubles the sample size of similar studies and provides new insights into the orientation and prevalence of nuclear structures in megamaser host galaxies.

Findings

Disk-like nuclear structures are nearly universal at <200 pc scales.

Nuclear structures show increasing misalignment with host galaxy disks at smaller scales.

The orientations of nuclear structures and megamaser disks are consistent with randomness.

Abstract

Using HST, we identify circumnuclear ($100$-$500$ pc scale) structures in nine new H$_2$O megamaser host galaxies to understand the flow of matter from kpc-scale galactic structures down to the supermassive black holes (SMBHs) at galactic centers. We double the sample analyzed in a similar way by Greene et al. (2013) and consider the properties of the combined sample of 18 sources. We find that disk-like structure is virtually ubiquitous when we can resolve $<200$ pc scales, in support of the notion that non-axisymmetries on these scales are a necessary condition for SMBH fueling. We perform an analysis of the orientation of our identified nuclear regions and compare it with the orientation of megamaser disks and the kpc-scale disks of the hosts. We find marginal evidence that the disk-like nuclear structures show increasing misalignment from the kpc-scale host galaxy disk as the scale of the structure decreases. In turn, we find that the orientation of both the $\sim100$ pc scale nuclear structures and their host galaxy large-scale disks is consistent with random with respect to the orientation of their respective megamaser disks.

Tables

Table 1. Table 1 Orientation of the masing disk, galaxy and nuclear region for each of the galaxies.

	Object	$D_A$	Maser			Galaxy			Nuclear region
		[Mpc]	$i$	PA	ref.	$e$	$i$	PA	Class	$r$ [pc]	$e$	$i$	PA
(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)	(13)	(14)
This work	ESO558$-$G009	109	90	350	G16p$†$	0.97	$104$	${277}^d$	Ch+N	210	0.87	61	99
	J0437+2456	70	90	115	G16p$†$	0.82	$125$	${126}^d$	Bu+N	120	0.77	51	131
	Mrk1029	117	90	300	G16p$†$	0.77	$130$	${176}^d$	D+N	220	0.83	57	126
	Mrk1210	58	101	333	Z17p	0.33	$19$	${36}^r$	D/R+Stw	170	0.45	26	119
	NGC5495	93	90	270	G16p$†$	0.63	$141$	$119,299$	D/R+Sgd	170	0.49	29	57
	NGC5728	41	90	329	K17p$†$	0.91	$114$	${302}^r$	R+Stw	460	0.36	21	121
	NGC5765b	126	95	237	G16	0.67	$42$	${337}^r$	R+Stw	450	0.45	27	67
	UGC3193	60	90	60	W17p$†$	0.95	$72$	${80}^d$	D?+N	220	0.90	65	79
	UGC6093	147	94	70	Z17p	0.47	$152$	$0^r$	Bu?+N	150	0.26	15	92

Equations

$$S=\left[\frac{I}{I\otimes P}\right]\otimes P^T\text{,}$$

$$CD{F}_{3D}(\psi )=\{\begin{array}{ll}0& ,\psi <{\psi }_1;\\ \frac{\cos {\psi }_1-\cos \psi }{\cos {\psi }_1-\cos {\psi }_2}& ,{\psi }_1<\psi <{\psi }_2;\\ 1& ,{\psi }_2<\psi .\end{array}$$

$$CD{F}_{\Delta \text{PA}}(\text{PA})=\{\begin{array}{ll}0& ,\text{PA}<{\text{PA}}_1;\\ \frac{\text{PA}-{\text{PA}}_1}{{\text{PA}}_2-{\text{PA}}_1}& ,{\text{PA}}_1<\text{PA}<{\text{PA}}_2;\\ 1& ,{\text{PA}}_2<\text{PA}.\end{array}$$

Full text

1. Introduction

It is now commonly accepted that supermassive black holes with masses of $10^{5}-10^{10}M_{\odot}$ (SMBHs) ubiquitously reside in galactic centers (Rees 1984, Kormendy & Ho 2013). The evidence of their existence extends to redshifts of $\sim 7$ (Mortlock et al., 2011; Venemans et al., 2013; De Rosa et al., 2014). How such massive black holes can form only $\sim 0.8$ Gyr after the Big Bang is still a mystery.

One of the main mechanisms proposed to fuel SMBHs is inflow of cold gas from the large-scale galaxy (Heckman et al., 1978). The final accretion itself is obviously facilitated on extremely small scales, where the infalling material radiates as an active galactic nucleus (AGN). However, for the gas to travel from the kpc-scale galaxy to the outer accretion structures at $\lesssim 1$ pc, more than four orders of magnitude in angular momentum must be lost. The details of this angular momentum extraction process are not yet fully understood.

On galaxy-wide scales, there are several potential mechanisms responsible for dissipation of angular momentum. Secular interactions between a (collisionless) stellar component and (collisional) gas in non-axisymmetric kpc-scale galactic structures (such as bars or spiral arms) can drive gas inwards (e.g. Hopkins & Quataert, 2010), and these interactions may be triggered in a variety of ways. During gas-rich mergers, tidal forces destabilize galactic disks to form non-axisymmetries, as seen by the numerical experiments of Hernquist (1989), Barnes & Hernquist (1996), Hernquist & Mihos (1995) and others. Observational results confirm that mergers accompany nuclear activity in some sources (Canalizo & Stockton, 2001; Ellison et al., 2011), but there is growing evidence that mergers are not the main kpc-scale driving mechanism of AGN activity, at least at moderate luminosity (Cisternas et al., 2011; Ellison et al., 2011; Kocevski et al., 2012; Villforth et al., 2014). The driver of kpc-scale galactic gas inflow likely depends on AGN luminosity and redshift (Hopkins & Hernquist, 2009; Treister et al., 2012; Comerford & Greene, 2014). Transient interactions between galaxies, or non-axisymmetric gravitational instabilities in isolated disk galaxies (Cavaliere & Vittorini, 2000; Hopkins & Quataert, 2010; Gatti et al., 2015), can also drive gas inflows. At high redshift, torques may be supplied by massive star-forming regions in gas-rich galaxies; such “clumps” are most frequently seen at redshift of $\sim 2$ (Shibuya et al., 2016) and models suggest that they sink in the gravitational potential of their host, driving gas inflow (Noguchi, 1998; Bournaud et al., 2007; Genzel et al., 2008).

However, torques induced by large-scale structures cannot efficiently extract angular momentum at distances $<1$ kpc from the galactic center (Goodman, 2003). Viscosity-related effects are not efficient enough beyond the last parsec from the black hole (Shlosman & Begelman, 1989; Goodman, 2003). Gravitational torques may again be a viable solution. The “bars within bars” model, originally proposed by Shlosman et al. (1989), assumes the presence of a series of embedded bars, stretching all the way from kpc-scales to the central SMBH accretion structures. These structures are expected to gradually remove angular momentum from the gas, letting it reach the galactic center sufficiently fast to explain AGN activity. The “bars within bars” mechanism, later revised to the “non-axisymmetric features all the way down” model (also referred to as the “stuff within stuff” model) to account for non-axisymmetries other than bars (Hopkins & Quataert, 2010, 2011), has been found to arise in close to self-consistent nested zoom-in simulations of Hopkins & Quataert (2010, 2011) and tested in several high-resolution numerical studies (Escala, 2007; Anglés-Alcázar et al., 2017). Moreover, central $\sim 100$ pc scale dust structures with various morphologies have been observed in a number of active and inactive galaxies (Jungwiert et al., 1997; Regan & Mulchaey, 1999; Márquez et al., 2000; Martini et al., 2003) and their morphologies confirm the viability of gravitational instabilities as the main gas inflow mechanism at $\sim 100$ pc scales (Maiolino et al., 2000; Davies et al., 2009; Haan et al., 2009; Combes et al., 2014; Davies et al., 2014).

If the “non-axisymmetric features all the way down” model is true, how do its features manifest in observations? One general feature is theoretically predicted by Hopkins et al. (2012). They report that the non-axisymmetric structures are expected to progressively misalign from the disk of their host galaxies as they reach further into the galaxy. Here, we search directly for these structures on $\sim 100$ pc scales using HST data, for a special sample of AGN where we know the orientation of the accretion disk precisely.

An especially precise measurement of the orientation of the central accretion flow in galactic nuclei is possible with observations of H2O megamasers (see Greene et al. 2013 and a review by Lo 2005). The maser emission in these systems originates in a ring of material illuminated by the AGN at a distance of $\sim 0.1-0.5$ pc from the black hole. The ring is located in the viscous-torque-dominated region of gas inflow and is thus expected to align with the accretion disk around the black hole. These megamaser disks are only detected when they are close to edge-on, since in that orientation the optical depth for maser action is maximized. Mapping with VLBI at sub-pc resolution (e.g., Greenhill et al. 1990; Kuo et al. 2011, see also Lo 2005) provides a very precise three-dimensional orientation for the accretion disk, a great advantage over other methods. Combined with HST/WFC3 images of the host galaxy, these properties make them excellent targets for investigation of how the orientation of nuclear non-axisymmetries correlates with the orientation of SMBH’s accretion disk. Such a comparison, along with morphological characterization of identified nuclear structures, is the main goal of this work.

The paper is organized as follows. In Sect. 2 we describe the instruments and technical details of the observations, as well as initial data reduction leading to the results presented in the following sections. In Sect. 3, the methods of identification and classification of the nuclear regions are presented. Calculations related to the orientation of galactic structures based on optical images are described in Sect. 3.5. Our results concerning the morphologies of nuclear regions are given in Sect. 4 and the relative orientation of various components within our galaxies is presented in Sect. 5. We discuss and summarize our findings in Sect. 6. In Appendix A we present detailed information on our analysis of each of the 9 new megamaser galaxies. In Appendix B we consider how the morphological classification of the nuclear regions in our sample depends on galaxy distance, scale of the region, and available resolution.

2. Observations and data reduction

There are 34 known megamaser disk galaxies (Pesce et al., 2015). We focus here on a subset of 18 megamaser disk galaxies with reliable BH mass measurements from Keplerian fitting to the maser dynamics (Kuo et al. 2011; Gao et al. 2016, 2017 and W. Zhao et al. 2017, in prep.). The list of our sources is given in Table 4. In general, the galaxies are early-type spiral galaxies (e.g., Greene et al. 2010) and we have studied the detailed morphological structure of roughly half of the galaxies in Läsker et al. (2016) using the Hubble Space Telescope (HST).

Each target was observed in two orbits with HST between Dec 1st 2014 and Aug 29th 2015. We obtained F336W, F438W, F814W, F110W, and F160W (roughly UBIJH) images of each galaxy with integration times of 1320, 430, 2140, 150, and 420 s, respectively. In the optical, we use a three-point dither pattern for cosmic-ray removal, and in the NIR we use the 4-point dither pattern. We use the default output of the MultiDrizzle pipeline, which performs cosmic-ray rejection and optimally combines the images.

3. Data analysis

The goal of this paper is to characterize the innermost structures in galaxies with maser disks. Here we describe the classification of these structures and the process of deriving their orientation from HST data.

3.1. Ellipse fitting

In order to support the identification of nuclear structures in our sample of galaxies, we used the algorithm of Jedrzejewski (1987) to fit ellipses to the galaxy isophotes, implemented as the IRAF script ellipse111ellipse is included in STSDAS (version 3.17) available as a package for IRAF (version 2.16.1 used)..

The initial analysis was performed using IRAF. First, the foreground stars and background galaxies were manually masked (in DS9222SAOImage DS9, http://ds9.si.edu.). The centers of the galaxies in NIR- and UVIS-band filters were found using the task imexamine and fed to an ellipse parameter file as initial ellipse center positions. The ellipse centers were then further refined by ellipse. In some cases (all filters for Mrk 1029, F336W for ESO 558 and J0437+2345, F438W for UGC 3193, as well as F336W and F438W for NGC 5765b), the “object locator’s k-sigma threshold” was also lowered from the default value of $1.0$ to $0.5$ in order for the algorithm to correctly identify the galactic center.

For each of our targets, the ellipse run conducted as above resulted in a list of elliptical fits to image isophotes for each of the filters – we use the parameters of those fits in further analysis.

3.2. Structure maps

As a second method to characterize nuclear morphology, we use structure maps to remove large-scale smooth galaxy components and highlight the small-scale features (such as dust lanes). The concept of structure maps was introduced by Pogge & Martini (2002). The technique is designed to remove low-frequency (smooth) features of the map, and highlight high-frequency features around the scale of the PSF. The method is closely related to Richardson-Lucy deconvolution (Richardson, 1972; Lucy, 1974). It is also similar in spirit to unsharp masking, but with structure maps the convolution is done with the PSF itself rather than a boxcar. Mathematically, a structure image is given by eq. (1) of Pogge & Martini (2002):

[TABLE]

where $S$ is the structure image pixel matrix, $I$ is the original image, $P$ is the point-spread function (PSF), $P^{T}$ is the transposed PSF and $\otimes$ denotes convolution. Structure maps emphasize high frequency features that are nearly unresolved in the original image.

In our analysis we used structure maps to highlight dust features in the galaxies of our sample. Two filters in our data set, F438W and F814W, are suitably sensitive to dust to derive the structure maps from them. F438W provides higher spatial resolution and is more sensitive to dust. However, in our data the F814W filter has significantly better signal-to-noise ratio than F438W. In our analysis we have therefore used the structure maps based on the F814W images.

The structure map derivation proceeded as follows. We extracted a sky-subtracted and re-centered PSF from each of the images. A lower cutoff $n_{0}$ for the count rate per pixel of $0.00001$ was set and all the pixels on both the original F814W image and the PSF with count rates $n<n_{0}$ were assigned count rates $n_{0}$ in order to avoid division by zero and negative values. Each of the F814W images was then convolved with its respective PSF obtained in previous steps333For Mrk 1210 and UGC6093 the PSFs obtained using point-sources in their F814W images were strongly asymmetric, resulting in dipole-like artifacts visible on the structure maps. Therefore we decided to use a more stable PSF from NGC 5495 in calculation of the structure maps in these two cases.. Finally, a copy of the original image was divided by the result of the previous step and the resulting image was convolved with a transposed PSF template. The structure maps for all the new sources can be found in Figs. 1 and A1 – A8.

The re-analysis of the sources of Greene et al. (2013) was performed analogously.

3.3. Identifying nuclear structures

We consider a range of scales in this work: the $0.1$–$0.5$ pc scale of the megamaser disk, the nuclear scales of $100$–$500$ pc where we seek kinematically cold or flattened structures (“nuclear regions”) along with spiral features (“nuclear spirals”) and the galaxy-wide disk on kpc scales.

Many galaxies are known to have nuclear disks on $10$-$100$ pc scales (e.g., Combes et al. 2014; García-Burillo et al. 2014) and we resolve the $100$ pc scales in $10$ of our $18$ galaxies. With the theoretical resolution limits of the F110W ($\sim 1153.4$ nm), F160W ($\sim 1536.9$ nm), and F814W ($\sim 802.4$nm) filters being $0.12$, $0.16$ and $0.08$ arcsec, respectively, we identify nuclear structures at least $0.3$ arcsec in size (which limits the visibility of $100$ pc structures to $\sim 70$ Mpc; see Table 4 for distances to the galaxies in our sample). The outer radii of our nuclear regions range between $0.3$ and $2.3$ arcsec, with a median of $0.6$ arcsec. In most cases we operate at the very limits of what can be robustly resolved and identified. However, we sometimes select larger structures that have a clearer interpretation to be able to analyse morphology and / or be better equipped to extract the orientation of the nuclear regions.

To ensure that the lower limit of the nuclear region’s angular size of $0.3$ arcsec is sufficient, we have re-derived all the results presented in Sections 4 and 5 adopting a more restrictive limit on the angular size of a nuclear region of $0.5$ arcsec (i.e., excluding 8 galaxies that host nuclear regions with radii $<0.5$ arcsec from the analysis). This corresponded to removing most of the galaxies beyond $100$ Mpc from the sample. Our conclusions (see Sections 4 and 5) remained mostly unchanged – for specific results of this trial and their discussion see Appendix B.2.

With ellipse fits and structure maps in hand, we proceed to identify nuclear structures in our sample galaxies. While we took into account all the available filters, we concentrated our efforts on F110W (or F160W) and F814W (the deepest image in the UVIS band); with F110W tracking starlight and F814W$-$F110W interstellar dust. For each of our galaxies we have chosen a set of ellipses that we associate with the large-scale galaxy, assuming that the isophotes on large scales are fit by an axisymmetric disk. These define the “kpc-scale” galaxy to which we refer in the following sections. We also identify a set of ellipses associated with the outer edge of a nuclear structure of size $\sim 50-600$ pc. We carefully pick those isophotes to correspond to changes in PA and ellipticity profiles, so that they correspond to a physical feature in the galactic nucleus (for details on how the structures are identified see Sect. 3.4).

3.4. Classification of nuclear structure

To discuss the morphology of nuclear structures, we classify them in two ways. First, we assign a class to the region itself, according to the key:

• •

D – disk,

• •

R – ring,

• •

Bu – bulge,

• •

B – bar,

• •

Ch – no discernible morphology, chaotic dust structure.

An additional “?” sign marks class assignment as unsure.

Nuclear spirals are almost ubiquitously found in late-type spiral galaxies (e.g., Pogge & Martini 2002; Martini et al. 2003). We therefore add a classification of potential spiral dust structure surrounding our set of nuclear isophotes, which we append to the nuclear region classification after a “$+$” sign:

• •

N – no surrounding spiral dust structure;

• •

Sx – spiral structure visible, where x denotes its type (“gd” – two-arm grand-design, “tw” – tightly wound or flocculent).

As an example, class D/R+Stw is assigned to a galaxy with a central disk or ring with a tightly-wound spiral structure – as in the case of Mrk 1210, see Fig. A3 and Table 4.

The outputs from ellipse aid our identification of nuclear structures in the following manner (cf. Greene et al. 2013):

• •

bars are characterized by a region of constant position angle (PA), ellipticity ($\epsilon$) decreasing inwards, and rapid $\epsilon$ and PA changes at their outer edge (Maciejewski et al., 2002; Erwin & Sparke, 2003) – see, e.g., the $\sim 4^{\prime\prime}$-scale bar in UGC 6093, Fig. A8;

• •

spiral structure is identified by smoothly rotating position angle with $\epsilon$ constant or changing (Martini et al., 2003) – see the spiral structure outside $5^{\prime\prime}$ in UGC 6093, Fig. A8;

• •

disks can be recognized by relatively constant PA and significant ellipticity – see the nuclear disk at $\sim 0.6^{\prime\prime}$ in Mrk 1210, Fig. A3;

• •

rings exhibit features similar to disks, but are distinguished by discontinuities in PA at their edges, as well as “bumps” in the surface brightness profiles (Buta, 1986) – see $\sim 2^{\prime\prime}$ nuclear ring of NGC 5728, Fig. A5;

• •

while the ellipse results should show bulges as round (i.e., with low ellipticity), with constant PA, and a surface brightness profile that is steadily rising towards the center (see, e.g., the $\sim 3^{\prime\prime}$-scale bulge in UGC 3193, Fig. A7), confident classification of bulges requires detailed 2D modelling of a galaxy (see, e.g., Läsker et al. 2016); we do not attempt such decomposition here and, therefore, nuclear regions classified as bulges in this work should be treated as tentative.

These considerations are only one part of our analysis. We also consider color and structure maps. For example, flocculent or tightly wound nuclear spirals are best detectable with structure maps, which improve the visibility of any PSF-scale structure, regardless of its symmetry, while they would fall below the spatial resolution of ellipse profiles. In turn, if narrow line regions (NLRs) are present in a galaxy image, they are best distinguished from dust structures by color maps, on which they appear very blue. The NLR emission may hide any dust components in the structure map, but the contamination due to the NLR can be visible in a color image (see the conical structure in NGC 5728, Fig. A5, which we interpret following Schommer et al. 1988 and Wilson et al. 1993 as an AGN ionization cone). Finally, the F110W data traces stellar light, and allows us to distinguish features corresponding to the stellar component of galactic nuclei, enabling us to verify whether a circular feature visible in the galaxy image may correspond to a bulge.

In some parts of our analysis it is beneficial to divide the nuclear regions in our sample into smaller and larger regions. The radius separating these two groups, $r_{b}=200$ pc, provides an equal number of objects in each size bin: there are 9 small nuclear regions ($r<200$ pc) and 9 large nuclear regions ($r\geq 200$ pc); see Table 4. We admit that the $200$ pc boundary is somewhat arbitrary, but small changes in this value that roughly keep sample sizes similar do not yield different results of our analysis. As noted in Sect. 3.3, all of our nuclear structures were identified at sizes of at least $0.3$ arcsec. While re-deriving all the results presented in Sections 4 and 5 with a more restrictive limit on angular size of a nuclear region of $0.5$ arcsec (see Sect. 3.3 and Appendix B.2), we also made sure that the determination of orientation for small ($r_{b}<200$ pc) nuclear regions is not affected by resolution effects. In the restricted sample the differences between large and small nuclear regions are still apparent and the sample is still evenly divided at $200$ pc – for specific results see Appendix B.2.

3.5. Angular momentum orientation from ellipse fits

If a nuclear region is disk- or ring-like, the orientation of its angular momentum in space is easily recovered from the ellipse fits. For each source we calculate the positions of the angular momenta of the galaxy as a whole and the nuclear region, utilizing the groups of ellipses described in Sect. 3.3. We take the average position angle (PA) and eccentricity ($e=\sqrt{1-(1-\epsilon)^{2}}$) within the region and calculate $i=\arccos{\left(\sqrt{1-e_{\rm avg}^{2}}\right)}$. Note that the PA we quote is for the angular momentum vector and is thus aligned with the minor axis of the projected disk. For instance, $\textrm{PA}=30\degree$ describes a disk with its projected angular momentum pointing $30\degree$ from North towards East on the sky and the image of the nuclear region in this case appears elongated along the $\textrm{PA}=120\degree$ direction with the blue-shifted edge in the direction of $\textrm{PA}=120\degree$ and the red-shifted edge in the direction of $\textrm{PA}=300\degree$. Four possible three-dimensional angular momentum directions are allowed by $\textrm{PA}_{\rm avg}$ and $i$. The position angle can be $\textrm{PA}_{\rm avg}$ or $\textrm{PA}_{\rm avg}+180\degree$ and an inclination of either $i$ or $180\degree-i$ is allowed (where $i=0$ corresponds to the angular momentum of the disk pointing towards the Earth). We are assuming that the nuclear region is a disk (treated as infinitely thin) to make this assignment. We quantify the accuracy of such an assumption by assigning a class to each nuclear region (see Sect. 3.4). The nuclear bulges are ignored in our analysis of nuclear regions’ orientations as their orientation cannot be established using the method described above. We do, however, consider the 3 nuclear regions with chaotic dust structures in our sample to correspond to flattened structures, and take their orientations into account.

There are important biases associated with our estimation of 3D positions of angular momenta, especially for nuclear regions. If a nuclear region is edge-on, ellipse will still fit a finite-width ellipse to its isophotes due to vertical structure in the disk and finite PSF, giving $i\neq 0$ (the floor appears to be $i_{\rm min}\sim 20\degree$, see Table 4). We also avoid face-on nuclear regions due to random structure in the plane of the disk distorting ellipse fits and potential misclassification of face-on disks as bulges.

In the case of the set of ellipses associated with the kpc-scale disk, additional information can be used to further constrain the orientation of the angular momentum. All of the galaxies in our sample are spirals. If we assume that the spiral arms are trailing, as is observed in most spiral galaxies (see Binney & Tremaine 2008 and references therein), this fixes the inclination and leaves only two possible orientations. In some cases, the rotation curves of the galaxies are also available, leaving only one angular momentum orientation allowed by the data.

For the galaxies without rotation curves available, we have used the relative prominence of dust lanes in the galaxy to constrain the orientation, a method originally suggested by Hubble (1929) and used by, e.g., Sharp & Keel (1985); Väisänen et al. (2008). The dust lanes of the part of the galaxy in front of its nucleus as seen by the observer are expected to be more pronounced than those behind it due to their being back-lit by stronger galactic emission closer to the nucleus. The only galaxy where a rotation curve is not available and dust lanes prominence method does not yield a reliable orientation (due to the galaxy being almost face-on) is NGC 5495, where we have kept both possible position angles of the angular momentum in the analysis. Table 4 gives all the resulting PA and inclination values for each of the galaxies.

In our investigations, we also use results and expand the analysis of Greene et al. (2013) in order to derive the statistics of the total sample of $18$ sources. The data related to these galaxies are included in the summary in Table 4.

4. Results: Morphology of nuclear regions

Figures 1 and A1 – A8 show the ellipses corresponding to the large-scale galaxy and the nuclear region.

When it comes to the nuclear regions, our sample of 18 galaxies contains $12$ disky structures, $4$ of which are rings, $2$ are disk/ring structures, and $6$ do not exhibit any additional morphology. We also identify $3$ bulges and $3$ chaotic dust structures. There are $3$ sources with grand-design nuclear spirals and $5$ with tightly-wound ones. Two of the former belong to galaxies with a large-scale bar (the exception being UGC 3789), while all of the latter belong to non-barred galaxies. Ten galaxies do not show any nuclear spirals associated with the identified regions.