© ESO, 2014

1. Introduction

The focus on merger studies often lies on major-major (equal mass spirals) mergers and their evolution, although minor (unequal-mass) mergers constitute the major phase of interactions in the local Universe and at higher redshifts. Understanding how the gas is feeding starburst and active galactic nucleus (AGN) activities in these objects is therefore paramount in understanding the overall evolution of the Universe. In a numerical simulation study of minor- or intermediate mergers, Bournaud et al. (2005) found that the gas brought in by the disturbing companion galaxy is generally found at large radii in the merger remnant. The gas returns to the system from tidal tails and often forms rings – polar, inclined or equatorial – that will appear as dust lanes when seen edge-on (e.g., Combes 1988; Shlosman et al. 1989).

Evidence has been presented for an increased fraction of star formation happening in clusters within starbursts that are compared to that in quiescent galaxies, which potentially are linked to the interstellar medium (ISM) structure changes, while the starburst process takes place (Kennicutt & Evans 2012). Massive, young star clusters (super star clusters (SSCs), 10⁵ − 10⁸M_⊙) are observed to form in on-going starbursts, such as those associated with interactions and (minor and major) mergers and those in smaller numbers in other types of galaxies with elevated star formation rates (SFR; e.g., Larsen 2002; de Grijs et al. 2003; Mora et al. 2009).

A nearby example of a surprisingly efficient starburst is the inner 2 kpc of the Medusa merger (NGC 4194). With a luminosity of L_FIR = 8.5 × 10¹⁰L_⊙ (at D = 39 Mpc, 1′′ = 189 pc), this E+S minor merger is an order of magnitude fainter than well known ultraluminous galaxies (ULIRGs), such as Arp 220 (Aalto & Hüttemeister 2000; Manthey et al. 2008). Despite the moderate FIR luminosity, NGC 4194 has a L_FIR/L_CO ratio similar to those typical for ULIRGs, which suggests that its star formation efficiency (SFE) rivals that of the compact ULIRGs. Despite this, its CO/HCN 1−0 luminosity ratio is ~25, indicating that the fraction of dense gas is significantly lower despite the similar SFE to ULIRGs. The picture becomes even more interesting when one considers that most of the ongoing star formation in NGC 4194 is not traced by the FIR or radio emission: The Hα SFR is ~40 M_⊙ yr^-1 (Hancock et al. 2006), while the FIR estimated SFR is 6−7 M_⊙ yr^-1. The spatial correlation between the molecular gas distribution and the 1.4 GHz continuum is also poor (unlike the case for most nearby galaxies, Aalto & Hüttemeister 2000). In NGC 4194, a large fraction of the star formation is going on in SSCs with a kpc-scale distribution, which is separated from the molecular gas distribution. No obvious correlation between these young SSCs (5−15 Myr) and the CO can be found. The CO emission also traces the two prominent dust lanes that cross the central region and extends into the northern tidal tail. The majority of the CO (~70%, ¹²CO, Aalto et al. 2001; Lindroos 2011) is found in the central 2 kpc of the galaxy with 15% of this gas (¹³CO, Aalto et al. 2010), which resides in a compact region 1.5′′ south of the radio nucleus. The knots where the optically traced star formation is going on only occasionally correlate with the radio continuum or the molecular distribution.

Throughout the paper, we are concerned with pure rotational transitions in CO between upper state J′ = 2 and lower state J′′ = 1 that are labeled 2−1.

In this paper, we present a study of the molecular gas close to the AGN in the Medusa merger. In Sect. 2, we describe the observations and data reduction; Sect. 3 reports on the results of the observations. The discussion follows in Sects. 4 and 5, and conclusions are drawn in Sect. 6.

2. Observations

NGC 4194 was observed with the SubMillimeter Array (SMA) on February 21, 2010 in the very-extended configuration and on April 8 in the compact configuration, which provides the highest angular resolution CO observations of NGC 4194 to date. For analysis purposes, we shifted the phase center to the position of the 1.4 GHz radio continuum peak at α = 12:14:09.660 and δ = + 54:31:35.85 (J2000, Beswick et al. 2005). The heterodyne receivers were tuned to the redshifted frequency of the ¹²CO 2−1 transition at 228.63 GHz in the upper sideband, while the ¹³CO 2−1 transition was observed in the lower sideband. With baseline lengths between 38 m and 509 m, these SMA imaging data are sensitive to scales smaller than 16.5′′. The correlator was configured to provide a spectral resolution of 0.8125 MHz (corresponding to a velocity resolution of ~1.1 km s^-1). The bright quasars J0854+201 and 3C 454.3 were used as bandpass calibrators; Vesta and Ganymede were observed as primary flux calibrators, and we regularly observed the close-by quasars J1153+72, J0927+390, and J0721+713 for complex gain calibration.

After calibration within the dedicated MIR/IDL SMA reduction package, both visibility sets were converted into FITS format and imported in the GILDAS/MAPPING¹ and AIPS packages for further imaging.

For the ¹²CO 2−1 data, sets of visibilities from the compact and very extended configuration observations were combined and deconvolved using the Clark method (Clark 1980) with robust weighting. This results in a synthesized beam size of 0.65′′ × 0.52′′ with a position angle (PA) of 67°. We smoothed the data to a velocity resolution of ~25 km s^-1, which yields a 1σ rms noise level per channel of ~6.8 mJy beam^-1. To look at the most compact component in the CO distribution, we used robust weighting, putting additional weight on the longest baselines, resulting in a 0.43′′ × 0.38′′ beam with a position angle of 52°. This provides a very high resolution map in which we look for giant molecular associations (GMAs).

To compare our results with high resolution images of the radio continuum structure of NGC 4194, we obtained 1.4 GHz radio continuum data at the VLA (Jütte et al., in prep.) and combined them with published data from Beswick et al. (2005). These combined data have an angular resolution of 0.50′′ × 0.52′′ corresponding to a linear scale of ~95 pc × 98 pc.

3. Results

3.1. ¹²CO 2–1

3.1.1. Integrated intensity

The integrated intensity map (Fig. 1a) shows an asymmetric CO 2−1 distribution in the east-west direction. The eastern part contains the peak of the CO emission at about 1−1.5′′ southeast of the radio continuum peak, which is located in the northern part of a ring-like structure or possibly tightly wound spiral arms, with a maximum extent of ~4′′ that has no CO emission peak at its center (Fig. 1a).

The total CO 2−1 flux within the 2σ contours (see Fig. 1a) amounts to 411.0 ± 4.1 Jy km s^-1. A comparison of the total CO 2−1 flux within the inner 13′′ of NGC 4194 obtained from single-dish observations Casoli et al. (1992) shows that we recover approximately 50% (~642 Jy km s^-1) of the flux with our SMA observations. Our high-resolution data are sensitive to structures smaller than 16.5′′ (see Sect. 2), hence 50% of the flux seems to be associated with structures larger than this size. Approximately one third of the total flux is located in the east of the ring-like structure (128.3 ± 4.5 Jy km s^-1) and one fourth (S_ν = 103.7 ± 2.7 Jy km s^-1) in the elongated western part of the CO 2−1 distribution.

Casoli et al. (1992) show the integrated intensity of CO 2−1 to be 80% of that of CO 1−0. Thus, we can scale our CO 2−1 flux by this amount to obtain an estimated CO 1−0 flux for our observations. Assuming a CO-to-H₂ conversion factor of X_CO = 2.0 × 10²⁰ cm^-2 (K km s^-1)^-1 (Narayanan et al. 2012; Sandstrom et al. 2013)², this then translates to a mass of ~1.6 × 10⁹M_⊙ for the overall CO 2−1 molecular gas distribution. The molecular gas in the ring sums up to a mass of ~4.1 × 10⁸M_⊙.

3.1.2. Kinematics of the ¹²CO 2–1 gas

Figures 1c, d, and 2 show the velocity field and the velocity dispersion of the CO 2−1 emission in NGC 4194. The velocity fields (Figs. 1c, d) show a rather regular pattern for the main body of the molecular emission, which is typical for solid body rotation, with a position angle of −20° in which the velocities range from 2400 km s^-1 to 2700 km s^-1. The contours of the velocity field in the western component of the molecular gas emission, which we call the western arm, deviate from the typical solid-body rotation pattern. These deviations could be indicators for streaming motions, as found in other galaxies, such as in M 51 and NGC 5248 (Aalto et al. 1999; van der Laan et al. 2013). Aalto & Hüttemeister (2000) have shown that molecular gas is associated with the dust lanes that cross NGC 4194 in front of the main body. The velocities of this gas emission are almost constant across large distances along the way to the center. This behavior changes just before the gaseous material turns into the plane of the merger (see, e.g., the most eastern CO associated with the eastern dust lane, as seen in Fig. 1c). This seems to be true as well for the dust lane located further west and closer to the center of NGC 4194.

The velocity dispersion distribution in the Medusa (Fig. 2a) shows the most prominent peak at the position of the radio continuum peak (Beswick et al. 2005) with a line dispersion roughly twice the size than in the rest of the CO distribution. The line dispersion at this peak rises to about 70 km s^-1 compared to the surrounding material where the dispersion lies between 15 km s^-1 and 35−40 km s^-1.

The position-velocity diagrams along the major (north to south, left panel) and the minor (east to west, right panel) axes are shown in Fig. 3. Both pv diagrams were obtained by averaging over slits with a width of 1′′ with one positioned along the axis of solid-body rotation and the other perpendicular to that direction. Both slits do include emission from parts of the molecular ring-like structure. In the major axis pv diagram (Fig. 3a), the pattern of the rather smooth solid-body rotation, especially when considering the merger history, already seen in the velocity field (Figs. 1c, d) has been reproduced. The two most distinctive peaks in this distribution represent the structure of the ring-like gas component at the center of NGC 4194. The minor axis pv diagram (Fig. 3b) shows two distinct components in the distribution: one molecular emission peak with a steep velocity gradient across a few arcseconds in the west and two distribution peaks close to the nucleus. The feature with the steep velocity gradient originates from gas in the same location that already showed an exceptional behavior in the velocity field (Figs. 1c, d) in the western arm. This might hint toward the presence of streaming motions in the western arm. The second component close to the nucleus represents the gas in the ring-like structure by showing a difference in velocity for the different locations in this structure.

The rotation curve: we fitted a Brandt curve (Eq. (1), Brandt 1960) to the velocity field of the molecular emission using a fixed inclination value of 40° (from the optical isophotes, Aalto & Hüttemeister 2000). The result is shown in Fig. 3a with the pv diagrams for the major and minor axes. The fitted rotation curve starts to flatten (fall) at a radius R_max of 2.3′′, which indicates a deprojected rotational velocity v_rot of 148 km s^-1. The dynamical mass inferred from these values is M_dyn ~ 2.2 × 10⁹M_⊙. A plot of the residuals after subtracting the fit from the velocity field is presented in Fig. 2b. There, the residuals in the western arm again illustrate the presence of typical features of streaming motions in this location. We cannot draw clear conclusions whether or not some type of streaming motion is also going on in the eastern part of the molecular emission as well due to the likely overlap of different structures and processes taking place there. ${\mathit{v}}_{\mathrm{rot}}\mathrm{(}{\mathit{R}}^{\mathrm{)}}\mathrm{=}\frac{{\mathit{v}}_{\max }\mathrm{(}\mathit{R}\mathit{/}{\mathit{R}}_{\max }\mathrm{)}}{{\mathrm{[}\frac{\mathrm{1}}{\mathrm{3}}\mathrm{+}\frac{\mathrm{2}}{\mathrm{3}}{\mathrm{(}\mathit{R}\mathit{/}{\mathit{R}}_{\max }\mathrm{)}}^{\mathrm{n}}\mathrm{]}}^{\frac{\mathrm{3}}{\mathrm{2}\mathrm{n}}}}\mathrm{·}$(1)

3.2. The Eye of the Medusa

Figures 1a and b show a top-heavy ring-like structure, where the majority of the molecular mass is located in the northern part of the structure. This structure dominates the CO 2−1 emission in the eastern part of NGC 4194, which is close to its center. We determined the ring-like structure to have a molecular mass of ~4.1 × 10⁸M_⊙ (corresponding to one third of the total CO 2−1 mass in these data), enclosed in a radius of ~1.7′′ (equivalent to ~320 pc) with a width of 1.5′′ (~285 pc) in the northern part of the ring and a width of 0.9′′ (~170 pc) in the south. The nucleus of NGC 4194 is located at the northern tip of the ring, while the bulk emission of the ring is located south of the kinematical center in a molecular shell or bubble. In contrast to 1.4 GHz radio continuum observations (Fig. 6, Beswick et al. 2005), where a secondary peak is located at the center of the ring, the CO 2−1 does not show a clear emission peak at its center. This structure was named the Eye of the Medusa (Lindroos 2011).

The CO 2−1 emission distribution of the Eye, shows hot spots of enhanced ¹²CO 2−1 intensity (Fig. 5). The mass estimates of some of those clumps place them in the mass range for GMAs (Table 1, Vogel et al. 1988). We identified GMA candidates as peaks (peak values larger than three times the noise level) in the integrated CO 2−1 map and with a clear detection (≳5σ) in the spectrum. The spectra were obtained by averaging over an area of 5 × 5 pixels around the peak. This resulted in the identification of 11 GMA candidates. How many of these candidates are real is an issue that can only be solved by higher resolution, high-sensitivity observations. The molecular properties of the Eye are discussed in Sect. 4.

4. Molecular structure of NGC 4194

Two major dust lanes to the east of the nucleus, which are visible in the optical (e.g., HST, Fig. 1b), cross the main optical body of NGC 4194. The one crossing directly below the galaxy’s dynamical center is associated with large parts of the molecular gas. The second dust lane further to the east crosses further south of the nucleus. The molecular gas is clearly associated with both dust lanes tracing filamentary-like features (see the lower resolution velocity field in Fig. 1d). The CO emission in the northeast stretching out toward these dust lanes appears filamentary. The western part of the CO distribution with a north-south orientation agrees well with peaks c and d found by Aalto & Hüttemeister (2000) in their CO 1−0 data (see Fig. 7). There seems to be a gap in the CO distribution between the molecular gas in the east and the west, apart from a small connection between the two parts with a width of ~1′′. Within the asymmetric CO 2−1 emission distribution, the ring-like structure, the Eye of the Medusa, shows hot spots of enhanced ¹²CO 2−1 intensity (Fig. 5). A more detailed discussion of these features can be found in Sect. 4.2.

Connected to the CO in the Eye seems to be a gas component that shows the highest velocity dispersion at the 1.4 GHz radio continuum peak found by Beswick et al. (2005). A weak AGN component might be present at this position in the center of NGC 4194. This is discussed in more detail in Sect. 4.1.

Observations of different star formation tracers have been reported covering the larger scale structures associated with the CO 1−0 observations and the dust lanes down to the region surrounding the nucleus of NGC 4194. Weistrop et al. (2004) and Hancock et al. (2006) found a number of stellar clusters in the UV and visible light (VIS, see, e.g., Figs. 8, 9). The majority of these clusters does not seem to be closely associated with the molecular gas. Indeed, most of these clusters are positioned in a void of molecular gas between the eastern and western parts in the CO 2−1 emission. For further details, see Sect. 4.3.

In Sect. 5.1, we compare the properties of NGC 4194 to other minor mergers and, in Sect. 5.2, we compare our results for NGC 4194 with galaxies that harbor molecular shells and blowouts.

4.1. The AGN

Several indicators speak in favor of the presence of an AGN in NGC 4194, even if it might be weak. X-ray and [Ne V] 14.3 μm line observations have shown the possibility for the presence of a small AGN contribution to the energy output of NGC 4194. Kaaret & Alonso-Herrero (2008) and Lehmer et al. (2010) suggest the presence of an AGN in the nucleus based on their 2−10 and 2−8 keV point source identifications and nuclear count rates. Spitzer observations of the [Ne V] 14.3 μm line yielded detections for Bernard-Salas et al. (2009) and Lehmer et al. (2010) which leads them to conclude the presence of a weak AGN being located in NGC 4194.

Spectra taken of the Eye from the high resolution data (0.43′′ × 0.38′′) show a double-peaked velocity distribution (Fig. 4) with velocity components centered at ~2600 km s^-1 and at ~2450 km s^-1. The latter component is only present in the region closest to the AGN (GMAs 3, 5 and 6). The gas mass determined from this velocity component is ~4.4 × 10⁷M_⊙. Isolating this velocity component and making an integrated intensity map confirms that this gas is only present in the northern part of the central high density gas complex (Fig. 5b). The total gas mass surrounding the AGN (both CO velocity components) results to ~1.1 × 10⁸M_⊙. The dynamical mass enclosed in this region is ~5.0 × 10⁸M_⊙. With their HI absorption observations, Beswick et al. (2005) were able to put a limit on the dynamical mass of the central nuclear region of ~2 × 10⁹M_⊙, which is in good agreement with the results from this work. This CO 2−1 complex is located at the unresolved center of the larger scale molecular gas reservoir in NGC 4194 (CO 1−0, see Fig. 7, Aalto & Hüttemeister 2000).

Speaking against the presence of an AGN, however, are the findings from Beck et al. (2014). Their radio continuum and [Ne ii] observations indicate the compact radio emission sources in NGC 4194 to be dense stellar clusters.

4.2. The structure of the Eye

The highest surface brightness CO 2−1 emission emerges in a ring-like structure south of the nucleus (Fig. 5b) – the Eye of the Medusa. The molecular gas in this complex seems to be connected to the 1.4 GHz radio continuum peak, which possibly marks the position of a weak AGN. Most of the gas, however, is associated with the eastern dust lane that comes in closest to the galactic center (Fig. 1). The center of the Eye is not located at the dynamical center of NGC 4194, as determined from the rotation curve (see Sect. 3.1.2). The location of the possible AGN in the northwestern part of this structure agrees with the center coordinates. This implies that this structure might not have been formed from purely dynamical processes (e.g., orbit crowding, resonances).

In Figs. 5 and 6 we show that there is a hole in the CO emission. Located right at the center of that hole is a secondary 1.4 GHz radio continuum peak (Fig. 6, see also Beswick et al. 2005). The radio continuum very nicely traces the peak of the gas in the northern part of the Eye but it seems to avoid gas in the southern part: the radio continuum peaks at the center of the structure where there is no CO, but most of the southern part of the CO in this region is not traced by the 1.4 GHz emission as is GMA 7 in the north. The radio continuum associated with GMA 8 most likely spills over from the peak at the AGN position.

The fact that the CO depression in the south of the ring-like structure corresponds to a secondary peak in the 1.4 GHz emission and the maximum in HI absorption against the continuum (Beswick et al. 2005), suggests that the hole in the CO 2−1 distribution might be a shell or bubble in the molecular gas caused by a massive explosion of supernovae (SNe). The funneling of the gas to the very center of NGC 4194 leeds to a burst of star formation in the central molecular gas. Subsequently, supernovae explode and the energy freed by the SNe going off drove the molecular gas outwards, which is away from the center of the explosion, thereby shaping the central molecular gas reservoir to the observed morphology: the expansion of the shell/bubble causes a spherical gas pile-up at the shock fronts, where new star formation is triggered and pinpoints the interaction between the accelerated shell/bubble material and the surrounding ISM. The result is the Eye of the Medusa, the molecular ring-like structure with a shell or bubble that we observe in the CO 2−1 emission. ${\mathit{E}}_{\mathrm{0}}\mathrm{=}\mathrm{5.3}\mathrm{\times }{\mathrm{10}}^{-7}{\mathit{n}}_{\mathrm{0}}^{\mathrm{1.12}}{\mathit{v}}_{\mathrm{sh}}^{\mathrm{1.4}}{\mathit{R}}^{\mathrm{3.12}}\mathit{.}$(2)To derive the energy output necessary for the formation of the shell/bubble with the properties we observe in our CO 2−1 data, we applied Chevalier’s equation (Eq. (2), Chevalier 1974). Using a hydrogen number density of the surrounding pre-bubble medium of ~59 cm^-3 (value determined from the density and volume of the surrounding gas not yet influenced by the expansion of the shell), a radius of 180 pc (the inner radial expansion of the shell material), and an expansion velocity of 55 km s^-1 (from the pv diagram), we derive an energy output of 1.4 × 10⁵⁵ ergs, which is the equivalent of ~10 000 SNe of type II. We derived the kinematical age, using t_kin ≈ αR/v where R is the radius of the shell or bubble, v is the expansion velocity, and α is a parameter to account for nonlinear expansion. We assume α to be ~0.5, a value in the middle of the range of possible values (Sakamoto et al. 2006). Due to the uncertainty in the assumption of α, we expect an uncertainty in the time estimate of a factor of 2. Hence, the kinematical age for the shell in NGC 4194 is ~1.6 × 10⁶ yr, and the supernova rate thus amounts to ~6 × 10^-3 SN yr^-1. A comparison to other galaxies hosting shells/bubbles is discussed in Sect. 5.2.

Our data mapped at the highest resolution (beam: 0.43′′ × 0.38′′, PA = 52°, 1′′ = 189 pc) revealed a clumpy structure in the Eye component of the CO 2−1 distribution, as in NGC 1097, NGC 1365, or NGC 1614, for example. We identified 11 possible GMAs (see Fig. 5); the majority (GMA 1 to GMA 8) are located in the northern part of the structure. Masses M(H₂) between ~1.0 × 10⁷M_⊙ and ~3.6 × 10⁷M_⊙ place them in the mass range typically found for GMAs (Vogel et al. 1988). Their dispersion values and sizes (spherical radii) range from ~40 km s^-1 to ~67 km s^-1 and 35 pc to 56 pc, respectively. Due to their large velocity dispersion, the GMA candidates do not fall on the Larson size-linewidth law for molecular clouds (e.g., Larson 1981; Solomon et al. 1987) but instead lie above this relation. A comparison of the CO spectra of the GMA candidates reveals a single spectral peak in all GMAs not associated with gas surrounding the AGN position. The spectra of some of the GMAs in the northern part of the shell show a double-peaked emission line (GMA 5, 8) and broad line widths probably due to blending of the two velocity components (GMA 6). These three complexes are closest to the nucleus of NGC 4194. The GMAs 7 and 8 are located in the northern part of the shell; GMA 9 to 11 are located in the southern part. The majority of these hot spots of enhanced ¹²CO 2−1 intensity (GMA 1, 4, 7, 9−11) is actually associated with the eastern dust lane that crosses the molecular gas shell. This can also be seen in NGC 1614, for example, where half of the GMAs identified in CO 2−1 observations are associated with dust lanes.

The average mass of the GMA candidates in NGC 4194 is ~2.2 × 10⁷M_⊙, the line widths average at a value of ~120 km s^-1, and the fitted average FWHM size (spherical radius) of the GMAs is ~109 pc. The GMAs in NGC 1614 seem to lie on a different scale. They are on average more massive (6.0 × 10⁷M_⊙), show the same line widths (120 km s^-1), and are significantly larger in size (270 pc) than the GMAs in NGC 4194 (König et al. 2013). The size estimate is difficult, though. Therefore, this comparison criterion should not be overestimated. Another point to keep in mind is that the ring-structure in NGC 4194 is only half of the size of the ring in NGC 1614. Another galaxy bearing GMAs that are comparable to the ones in NGC 1614 is NGC 1097 (average size ~250 pc, mass 9.2 × 10⁷ M_⊙, line width 84 km s^-1, Hsieh et al. 2011), whereas the GMAs in NGC 1365 (line widths between 60 and 90 km s^-1, masses of ~10⁶ M_⊙, Sakamoto et al. 2007) seem to be even less massive and have smaller line widths than the ones in NGC 4194.

4.3. Circumnuclear molecular gas and star formation

A number of different gas tracers has been observed toward NGC 4194 with different spatial resolutions. The CO 1−0, radio continuum (at 3.5 cm, 6 cm and 20 cm), and HST observations (e.g., at wavelengths of 150 nm and 600 nm), for example, show brightness distribution peaks at the location of the AGN (e.g., Aalto & Hüttemeister 2000; Condon et al. 1990; Beswick et al. 2005; Armus et al. 1990; Weistrop et al. 2004). Spectroscopic Hα observations find the majority of the star formation going on in the central 8′′ (Weistrop et al. 2012). Several star forming clusters have been identified in the UV (160 nm and 200 nm, Weistrop et al. 2004) and the VIS (430 nm and 780 nm, Hancock et al. 2006); some of them are associated with the AGN at the nucleus of NGC 4194 (Fig. 8).

Most of the brightness peaks identified in the high resolution CO 1−0 data (Aalto & Hüttemeister 2000) are situated in the main dust lane (peaks a, b, d, and e). Peaks a and b represent positions in the southern and northern part of the central high surface brightness density complex in the CO 2−1 maps; peaks c and d are located in the northern and the southern part of the western arm. Two of the CO 1−0 peaks (b and c) seem to be counterparts to brightness maxima found in Hα observations (see Fig. 7, Armus et al. 1990). Peak b is furthermore associated with the peak position in the high-resolution 1.4 GHz continuum maps (see, e.g., Fig. 6, Beswick et al. 2005).

Our CO 2−1 observations together with all these tracers indicate that a big part of the overall star forming activity seems to be on-going in the central few arcseconds around the AGN. However, to gain more insight into the locations and the properties of the star forming regions in the center of NGC 4194, further observations are needed. For example, high resolution Brγ narrow-band images would give valuable insight into the positions of regions of massive star formation; HCN/HCO⁺ observations would help to distinguish high density regions (n ≳ 10⁴ cm^-3).

The search for star clusters has yielded detections in the visible and in the UV wavelength regimes (see Figs. 8 and 9). Weistrop et al. (2004) used HST imaging to look for star clusters in the UV. They identified 39 young stellar clusters between the ages of 5 and 15 Myr distributed over the optical body of NGC 4194. Hancock et al. (2006) extended the properties of star clusters in this galaxy by identifying 14 clusters in a slit positioned along a position angle of 147° using visible and UV light. Some of the clusters from the two samples seem to coincide within the placement of the slit, but there are also clusters at slit positions in the visible light that seem to have no UV counterparts. The majority of the UV and VIS identified star clusters does not seem to be closely associated with the molecular gas. Indeed, most of these clusters are positioned in a void of molecular gas between the eastern and western parts in the CO 2−1 emission. Most of the UV-identified clusters are located in the north and west of the CO 2−1 distribution; none are associated with molecular gas in the east and north-east of the nucleus (Figs. 8, 9).

A comparison to the high resolution CO 2−1 shows that the GMAs closest to the nucleus (GMA 3, 5 and 6) seem to be associated with some of the Weistrop et al. (2004) UV clusters (Fig. 9). One of the clusters identified by Hancock et al. (2006) is also located in that region. The southern part of the CO shell, on the other hand, harbors no star clusters, although we identified three GMAs in that region. One of the VIS-identified star clusters, however, is located right at the center of the shell, where there is no CO, but the radio continuum has a secondary peak in this location.

Using the CO 2−1 data, we are now able to better distinguish between the higher density gas close to the nucleus and the lower density gas in the surrounding environment described by the CO 1−0 observations of Aalto & Hüttemeister (2000). Indeed, we find that the gas in the Eye closest to the nucleus is of higher surface brightness density than the larger scale CO emission. This reservoir of high-surface-brightness gas is located inside the solid body part of the rotation (Fig. 3). The gas surface density close to the nucleus amounts to ~4.1 × 10³M_⊙ pc^-2, whereas we find a value of ~1.4 × 10³M_⊙ pc^-2 for the surrounding medium; hence, we find a difference of a factor of ~3. This value would increase even more when taking the surface density of the CO 1−0 (500−1000 M_⊙ pc^-2, Aalto & Hüttemeister 2000) into account instead. The difference between the obtained values from our CO 2−1 observations and the CO 1−0 data from Aalto & Hüttemeister (2000) is that we do suffer from resolving out large parts of the lower brightness gas due to the high resolution of these observations.

The bulk of the star formation ongoing in the SSCs that are distributed on extended scales is not associated with the regions of highest surface brightness density (Figs. 8, 9). The clusters identified with the highest Hα SFR, however, are located in this high surface density region (Hancock et al. 2006). Using the CO 2−1 surface brightness density, the Kennicutt-Schmidt law predicts an SFR of ~4.9 M_⊙ yr^-1, which within the error bars is in good agreement with the 6−7 M_⊙ yr^-1 value that Aalto & Hüttemeister (2000) estimated from the FIR but is much below the value of ~46 M_⊙ yr^-1 that Hancock et al. (2006) derive from their Hα observations. The SFE we obtained from the highest surface brightness density CO 2−1 emission is ~1.5 × 10^-8 yr^-1. Although NGC 4194 has a brightness surface density of one magnitude lower than typical ULIRGs, its SFEs rival that of these extraordinary objects (e.g., SFE(Arp 220) ~2 × 10^-8, Franceschini 2003). Compared to values in typical mergers, the star formation in NGC 4194 is two orders of magnitude more efficient (Rownd & Young 1999).

Although the gas surface density at the center of NGC 4194 has now increased to ~4.1 × 10³M_⊙ pc^-2, these values show that the surface density even in the densest regions in the center is still lower by one order of magnitude compared to the gas in the centers of ULIRGs like Arp 220 or Mrk 231 that have surface densities of up to ~3.3 × 10⁴M_⊙ pc^-2 (Bryant & Scoville 1999). The central high surface brightness gas density values in NGC 4194 are comparable to numbers for other similar sources, such as NGC 1614 (~3.1 × 10³ M_⊙ pc^-2, König et al. 2013) and NGC 5218 (~1.8 × 10³ M_⊙ pc^-2, Cullen et al. 2007).

4.4. Infalling gas

1.4 GHz radio continuum observations show two compact radio components at the center of NGC 4194 (Beswick et al. 2005). They concluded that there is the possibility of a weak, buried AGN that contributes to the powering of the star formation. The position that is inferred from the radio continuum brightness peak coincides with the velocity dispersion peak, which we find in our CO 2−1 data (Fig. 2a). The rotation curve fitted to the CO 2−1 data cube (Fig. 2a, Sect. 3.1.2) further indicated the location of the dynamical center just ~0.4′′ toward the south of the radio continuum and the velocity dispersion peaks. This is in positional agreement within one beam size.

It is possible that the nucleus is fed by molecular gas returning via the dust lane. Bournaud et al. (2005) have shown that the gas in minor mergers that is brought in by a disturbing galaxy companion is generally found at large radii in the merger remnant. Gas located in this largest scale gas reservoir is then returned to the system via tidal tails, where it often ends up forming rings, which might be associated with or appear as dust lanes seen edge-on. In NGC 4194, a number of dust lanes has been found. The dust lanes most probably playing a role in the feeding of the AGN are the two large dust lanes to the east of the nucleus and a smaller one just northeast of the AGN. All those dust lanes are associated with the CO emission (Aalto & Hüttemeister 2000, see also Fig. 1b) and might funnel the CO from the larger scale molecular gas reservoir, as traced by the CO 1−0 emission, to the center and ultimately to the AGN. We found evidence for this scenario for example in the western part of the molecular gas distribution in the velocity field (Figs. 1c, d) and the map showing the residuals from the subtraction of the rotation curve (Fig. 2b), which show typical features of streaming motions as, for example, found in M 51 (Aalto et al. 1999) and M 83 (Rand et al. 1999).

Another mode of transport providing gas to the center of NGC 4194 might be streaming motions. We found indications for streaming motions in the western arm of NGC 4194. Gas could be piled up in the center of the galaxy due to orbit crowding processes of inflowing gas streams. In the galactic potential, the gas brought in through the streams encounters shocks and migrates to new orbits accumulating in these locations.

5. Comparisons to other galaxies

Comparisons between the features of NGC 4194 and other interacting galaxies with similar types of structure can help in the interpretation of our data. In this section, we consider how the properties of NGC 4194 relate to those observed in a comparison sample of interacting galaxies and galaxies with shells/bubbles.

5.1. Comparison to other minor axis dust lane minor mergers

One galaxy sharing a number of properties with NGC 4194 is the S+s minor merger NGC 1614. Both galaxies are associated with large reservoirs of molecular gas (e.g., Aalto & Hüttemeister 2000; Olsson et al. 2010; König et al. 2013). Whereas the star formation in NGC 1614 is mostly associated with molecular gas (CO) and other tracers like Paα, radio continuum, etc. (König et al. 2013), most of the star formation in NGC 4194 is taking place in star clusters (Weistrop et al. 2004; Hancock et al. 2006), which are not associated with the bulk of the ¹²CO emission, but instead, mostly with lower surface brightness gas to the west of the mergers nucleus.

Looking into the molecular gas content in these two galaxies in more detail, NGC 4194 and NGC 1614 share even more properties. In both galaxies, a large reservoir of CO 1−0 emission is associated with the minor axis dust lanes crossing the mergers main body. This association between molecular gas and dust lanes led to suggestions that the dust lanes are playing a key role in the transport of the molecular gas into the centers of galaxies of this type (minor mergers). Both galaxy centers harbor ring-like molecular structures that are clearly identified in high-resolution ¹²CO 2−1 observations (König et al. 2013, this work) that might be connected to the larger scale CO 1−0 reservoirs via the respective dust lanes. The ring-structures are, however, different in size and mass. The ring in NGC 1614 is twice as large and twice as massive (König et al. 2013) than the shell complex in NGC 4194 (Fig. 9, Table 2). Both ring-like structures are located off the actual nuclei of their galaxies: the ring in NGC 1614 is located symmetrically off-nucleus with its center located right at the galaxy’s nucleus (König et al. 2013), whereas the high density gas complex in NGC 4194 including the shell is located asymmetrically off nucleus. Only the gas to the north of the shell, containing the AGN, is located at the very nucleus of this galaxy. Both rings are closely associated with radio continuum emission, whereas the emission in NGC 1614 is actually associated with the ring, the radio continuum in NGC 4194 has a secondary peak at the center of the shell but does not coincide with most of the CO emission located in the shell. It does seem as if the radio continuum in NGC 4194 is avoiding the regions filled with CO in the shell. Furthermore, we have identified GMAs (Fig. 9) in both NGC 4194 and NGC 1614–11 in NGC 4194 and ten in NGC 1614. The GMAs in both minor mergers are associated with the dust lanes. In general, the GMAs in NGC 4194 are smaller in size and mass then the ones in NGC 1614. This is most probably due to the AGN-shell-complex in NGC 4194 itself being much smaller than the ring in NGC 1614.

Other nearby minor axis dust lane galaxies at different evolutionary stages in the merger sequence share some of these properties as well (e.g., NGC 5218, NGC 3718, and NGC 4441). The object NGC 5218 is a merger at an earlier evolutionary stage with a molecular ring about 2.5 times larger and twice more massive than the ring in NGC 4194 (Table 2Cullen et al. 2007; Olsson et al. 2007). Two galaxies at a later evolutionary stage in the merger sequence are NGC 3718 and NGC 4441. In their morphology, NGC 3718 and NGC 4441 are very similar to NGC 4194; they have tidal tails and minor axis dust lanes crossing the galaxies. Closely associated with these features are molecular gas reservoirs extended out to large scales (e.g., Pott et al. 2004; Krips et al. 2005; Jütte et al. 2010) as in NGC 4194. Like NGC 4194, NGC 3718 hosts a large number of star clusters (Trinh et al. 2006). Both mergers also harbor molecular disks/ring-like structures at their centers (Schwarz 1985; Pott et al. 2004; Krips et al. 2005; Sparke et al. 2009; Jütte et al. 2010).

In a nutshell, galaxies with minor axis dust lanes involved in interactions/mergers can show extended molecular gas emission over scales of several kpc, and many harbor molecular gas rings at their centers. There is no obvious correlation between the size and mass of the observed molecular rings with the stage of the interaction/merger process the comparison galaxies represent.

5.2. Shells in other galaxies

A number of shells and bubbles have been discovered in external galaxies using different atomic and molecular gas tracers (e.g., HI, OH, CO, Weiß et al. 1999; Wills et al. 2002; Sakamoto et al. 2006; Olsson et al. 2007; Argo et al. 2010). Nearby starburst galaxies known to host molecular shells/bubbles are M 82, NGC 253, and NGC 5218. These bubble hosts are no mergers, but M 82 and NGC 5218 are involved in interactions. All three of them have a bar transporting molecular gas to the galaxies center (Petitpas & Wilson 2000; Mauersberger et al. 1996; Olsson et al. 2007) like NGC 4194 (Beck et al. 2014; Jütte et al., in prep.).

One process, that causes the formation of shells/bubbles can be the explosion of a large number of supernovae, as suggested for the bubbles in M 82 (e.g., Wills et al. 2002; Argo et al. 2010) and the shell in NGC 5218 (Olsson et al. 2007). It has been determined that the energy output of up to several thousand type II supernovae is necessary to reach the current extents that are found for these structures. An alternative scenario for shell/bubble formation is that they might be caused by stellar winds coming from massive SSCs, as suggested for the bubbles in NGC 253 (Sakamoto et al. 2006). NGC 4194 does have a large population of star clusters, but the majority of these clusters is located away from the bulk of the molecular gas. In particular, we do not find a large conglomerate of star clusters at the center of the molecular shell in NGC 4194. This points toward supernova explosions as the driving mechanism of the shell in this galaxy.

An in-detail comparison between the shells/bubbles in these three galaxies with the shell in NGC 4194 (for more details see Table 2) shows that the shell in NGC 4194 is older, has a larger size, expands with a higher velocity, and is more massive than the shells/bubbles found in the starburst galaxies M 82, NGC 253, and NGC 5218.

This shows that NGC 4194 shares properties with several of the comparison galaxies – for example, being a merger, having dust lanes, harboring shells/bubbles and/or GMAs, having asymmetric molecular ring-like structures, and massive star clusters. However, we always found significant differences as well. Therefore, NGC 4194 cannot be classified as a typical barred spiral galaxy, or a typical merger of any flavor; it remains a very interesting yet puzzling object.

6. Summary

We studied the properties of the molecular gas in the center of NGC 4194 using high angular resolution ¹²CO 2−1 emission line observations.

• 1.

  We found a ring-like structure, the Eye of the Medusa, at the center of NGC 4194. A large part of the gas in this region of high surface brightness is associated with the major dust lane. The structure contains molecular gas associated with the AGN and a molecular shell located southeast of the nucleus.

• 2.

  The event causing the formation of the molecular shell southeast of the dynamical center of NGC 4194 was most probably a number of supernovae explosions, which now pushes the molecular gas outward.

• 3.

  The kinematics of the molecular gas distribution suggest that gas is transported along the dust lanes to the center of NGC 4194, where it replenishes the gas in the Eye. Smaller dust lanes in the very center, together with the pressure provided by the expanding material in the shell, then funnel this gas to the immediate surroundings of the AGN.

• 4.

  We identified individual GMAs in the Eye, which are associated with both the shell structure and the gas surrounding the AGN. The GMAs are not associated with star clusters previously identified in the UV and visible wavelength regimes.

In summary, we found that NGC 4194 is a very interesting lab to study different physical properties of the molecular gas in different states. It is remarkable how well-ordered the velocity field looks when comparing all the different processes going on at different scales of this merger. We believe we found another example of a minor merger, besides NGC 1614, where molecular gas is transported via the dust lanes to the very center of the galaxy. There, we found molecular gas associated with the AGN at the dynamical center and with a shell to the southeast of the very nucleus, which is most likely formed by explosions of supernovae. The GMAs we located in the central molecular structure might be the result of the formation of denser gas resulting from shocks from the supernovae explosions.

In spite of these results, many questions remain to be answered, such as the origin of the gas surrounding the AGN, the feeding of the AGN itself on even smaller scales, and the origin of the GMAs, or how the kinematics of cold molecular gas affect the evolution of star formation and nuclear activity in minor mergers in general. A more detailed study of the molecular gas content of NGC 4194 on different spatial scales is needed to address these questions. New studies with instruments, such as the PdBI and the SMA, will enable us to answer them.

---

Acknowledgments

We thank the referee for useful comments. S.A. thanks the Swedish Research Council (grant 621-2011-414) and the Swedish National Space Board (SNSB, grant 145/11:1-3) for support. J.S.G. thanks the College of Letters & Science, University of Wisconsin-Madison for partial support of this work. The Submillimeter Array is a joint project between the Smithsonian Astrophysical Observatory and the Academia Sinica Institute of Astronomy and Astrophysics and is funded by the Smithsonian Institution and the Academia Sinica. MERLIN/eMERLIN is a National Facility operated by the University of Manchester at Jodrell Bank Observatory on behalf of STFC. AIPS is produced and maintained by the National Radio Astronomy Observatory, a facility of the National Science Foundation operated under cooperative agreement by Associated Universities, Inc. 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.

References

All Tables

All Figures