Extraplanar [C II] and Hα in the Edge-On Galaxy NGC 5775

Abstract

Spiral galaxies are thin and susceptible to being disrupted vertically. The largest star clusters, and nuclear starbursts, generate enough energy from winds and supernovae to send disk material to the halo. Observations of edge-on galaxies allow for the clearest view of vertical disruptions. We present new observations of the nearby, edge-on galaxy NGC 5775 with SOFIA in [C II] $157.7 \mathsf{\mu }$m and archival images from Hubble in H$\mathsf{\alpha }$ to search for extraplanar gas. The extraplanar [C II] extends 2 kpc from the midplane over much of the star-forming disk. The extraplanar [C II] at 2 kpc from the midplane approximately follows the rotation of the disk, with a lag of approximately 40 km ${\mathrm{s}}^{-1}$; this lag is similar to what has been previously reported in H$\mathsf{\alpha }$. Significant vertical extensions (to 3 kpc) are seen on the northeast side of the galaxy, potentially due to super star clusters in the NGC 5775 disk combined with gravitational interaction with the companion galaxy NGC 5774. The H$\mathsf{\alpha }$ narrow-band image reveals a narrow plume that extends 7 kpc from the nucleus and is almost exactly perpendicular to the disk. The plume shape is similar to that seen from the comparable galaxy NGC 3628 and may arise from the nuclear starburst. Alternatively, the H$\mathsf{\alpha }$ plume could be a relic of past activity.

1. Introduction

Galaxies are composed of dark matter, stars, and interstellar gas. In the solar neighborhood of the Milky Way, these constituents comprise 42%, 31%, and 14% of the mass surface density, respectively [1]. The fact that these densities are in the same order of magnitude suggests a significant mutual influence among the constituents. Since two-thirds of the baryonic material in the solar neighborhood is in stars, the net star formation efficiency cannot be higher than this. In fact, the efficiency of star formation is significantly lower (<1%, [2]), requiring a mechanism that regulates star formation and prevents all gas forming stars or being lost. Otherwise, the Milky Way would be like a globular cluster, dominated by old stars and nearly gas free [3]. On scales of kiloparsecs (kpc) and larger, two disruptive mechanisms contribute: gravitational interactions with other galaxies, and combined kinetic energy from localized bursts of star formation. The effects of galaxy interactions are known from observing groups and clusters [4]. Galaxies not in groups or clusters, and with no recent interaction with large nearby galaxies, require an internal feedback source to prevent total star formation.

The circumgalactic medium of a galaxy depends on a balance between infall, due to the pull of dark matter and stars toward the midplane, and outflow, due to the kinetic energy imparted by supernovae and outflows from clusters in the disk (reviewed by [5]). The cycling of material from the disk of a galaxy is a key process for understanding galaxy evolution, but simulations yield divergent views of whether star formation feedback extends beyond halos or actually cycles through a circumgalactic medium [6]. Bursts of localized star formation drive material into the halo through ‘chimneys’ [7]. In balance with infall, the cycling is referred to as a ‘galactic fountain’ [8,9]. However, if the kinetic energy imparted to the outflowing material is sufficiently large, it escapes as a ‘superwind’ [10].

Edge-on galaxies afford the best view of inflow and outflow from the disk. In this paper, we present observations of the edge-on spiral galaxy NGC 5775, which is part of the Virgo Cluster, at a distance of 17.4 Mpc [11]. The neighboring galaxy, NGC 5774, is evidently connected to NGC 5775 by an $\mathrm{H}\mathrm{I}$ bridge [12], indicating that the two galaxies are undergoing a minor merger. The $\mathrm{H}\mathrm{I}$ disk has a larger-than-normal scale height due to the merger [13].

Multiple lines of evidence point to a large-scale outflow from NGC 5775, with vertical excursions from the midplane of 7 kpc in H I [14] and 13 kpc in [N II] [15]. In X-rays, a halo of hot gas is evident, potentially forming a “limb-brightened outflow cone” [16]. Distinguishing between individual supershells or filaments and a galaxy-wide flow is challenging, as some extraplanar features appear to be associated with in-disk star formation [17]. Radio continuum observations tracing synchrotron radiation, which arises from cosmic rays interacting with the large-scale magnetic field of the galaxy, reveal organized vertical magnetic field structures [18] that are distinct from the tightly wound magnetic field in the midplane of a galaxy. NGC 5775 is relatively bright in X-rays, consistent with other high star formation Virgo galaxies and unlike galaxies that appear more faintly in X-rays, possibly due to ram pressure stripping of their halos [19].

This paper completes our SOFIA observational study of edge-on galaxies, adding new observations of NGC 5775 (Figure 1) to complement those obtained for NGC and NGC 5907 [20]. The observations target the far-infrared fine-structure line of ionized carbon ([C II]) at 157.751 $\mathsf{\mu }$m. The upper energy level of this transition is only 91 K above the ground state, making it relatively easily excited, except in the coldest interstellar conditions. Carbon is also easily ionized (singly) under most interstellar conditions, except when deep in cold molecular clouds (where the upper level is not excited anyway) and very close to energy sources. The combination of its low-excitation energy, typical ionization state, and far-infrared emission’s relative immunity to the heavy extinction near the galactic midplane makes the [C II] ground-state fine-structure line an ideal tracer of the diffuse interstellar medium.

2. Observations

The Stratospheric Observatory for Infrared Astronomy (SOFIA; [22]) observed NGC 5775 as part of project 06_0010, on three flights operating from Santiago, Chile.

Table 1. SOFIA/FIFI-LS observation log.

Date	Flight	Altitude (feet)	Elevation (°)	Duration (min)	Chop (″)	Transmission	Exposure (s)
23 March 2022	843	40,500	36	55	120	89	1443
24 March 2022	844	42,400	45	107	120	93	2488
27 March 2022	845	42,900	49	30	160	94	768

summarizes the observing conditions. The atmospheric transmission, averaged within $\pm 250 \mathrm{km}{\mathrm{s}}^{-1}$ of the [C II] $157.741 \mathsf{\mu }$m line redshifted to $1676 \mathrm{km}{\mathrm{s}}^{-1}$, exceeded 89%, which is only possible from an airborne or space telescope due to the high opacity of the Earth’s atmosphere in the far-infrared. The science instrument utilized was the Far Infrared Field-Imaging Line Spectrometer (FIFI-LS; [23]), which generates two images of $5\times 5$ spaxels: one red and one blue. The red-side grating was scanned over the wavelength of the [C II] $2{}^2\mathrm{P}_{3/2}\to {}^2\mathrm{P}_{1/2}$ transition redshifted to $1676 \mathrm{km}{\mathrm{s}}^{-1}$. The blue-side grating was scanned over the wavelength of the [O III] $3{}^3\mathrm{P}_1\to {}^3\mathrm{P}_0$ transition at $88.4 \mathsf{\mu }$m; those data are not discussed further in this paper. The observations had an angular resolution of ${16}^{″}$, corresponding to 1.3 kpc at the distance of NGC 5775. The spectral resolution of the observations was $250 \mathrm{km}{\mathrm{s}}^{-1}$ [24]. The observations utilized a matched chop-nod choreography, with the secondary mirror chopping by a total separation listed in Table 1 as ‘Chop’ at a frequency of 1 Hz, with a position angle ${305}^{\circ }$ E of N. The exposure times in Table 1 comprise 37% of the duration of the respective legs of the observing flights, with the same amount spent on chopping using the secondary mirror for sky subtraction; the remainder of the time was spent on telescope motions and nod spectra.

The FIFI-LS data were processed using the SOFIA data processing pipeline sofia-redux [25] version 2.5.2. Data from all flights were processed together, with the following adjustments to the default parameters. In the Spatial Calibrate step, the option to rotate the images by the detector angle was disabled. Since our observing strategy aligned the galaxy’s midplane with the detector, the resulting images are naturally aligned with the galaxy. In the Resample step, the ‘spatial edge threshold’ was decreased to 0.01 to avoid generating artifacts from outside the boundary of the actual observed image. The exposure time per sample in the final cube (i.e., at a given position and wavelength) was 72 s within 1 kpc of the midplane, 60 s at distances of 1.5 kpc from the midplane, and 44 s at distances of 2–3.5 kpc from the midplane.

A baseline was subtracted from each pixel of the FIFI-LS data cube by fitting a straight line to the brightness measured between −860 and +860 km ${\mathrm{s}}^{-1}$ (relative to the galaxy’s central velocity), while excluding velocities between −400 and 400 km ${\mathrm{s}}^{-1}$. Figure 2 shows the integrated [C II] surface brightness image over the entire galaxy in the $Y,Z$ coordinate system we will use for the along-plane and vertical directions, respectively. The left panel of Figure 3 shows a slice through the [C II] cube along the major axis of the galaxy. In this and all subsequent velocity plots, velocities are shifted such that $v=0$ corresponds to a heliocentric systemic velocity of 1590 km ${\mathrm{s}}^{-1}$. An approximation of the [C II] rotation curve in this reference frame is

$$v=150{tan}^{-1}\left({\displaystyle \frac{Y}{1.8\mathrm{kpc}}}\right){\mathrm{km}\mathrm{s}}^{-1}.$$

(1)

This approximation is illustrated in Figure 3.

3. Distribution of Extraplanar [C II]

3.1. Vertical Profile

To derive the vertical extent of the [C II] emission, we extracted spectra at various altitudes (Z) above the midplane, averaging over elongated rectangular regions within one beam of each Z. Since the galaxy’s rotation speed exceeds the instrument’s velocity resolution, averaging the spectra over the northern and southern halves of the galaxy would artificially dilute the signal. To mitigate this effect, we created three sets of vertical spectra, each corresponding to a different range of projected distances (Y) from the nucleus, measured parallel to the midplane. Figure 4 shows the spectra for the approaching and receding halves of the galaxy, as well as the average over the disk.

The [C II] emission decreases steeply with Z, but its vertical distribution remains well-resolved. The SOFIA full width at half maximum (FWHM) beam is 1.35 kpc at the distance of NGC 5775, corresponding to a spatial extent of $\pm 0.67$ kpc. Emission is clearly detected at both $Z=+2$ and $Z=-2$ kpc from the midplane. The telescope is diffraction-limited at the observing wavelength. For a Gaussian approximation of the beam, and a disk significantly thinner than the beam, the signal of the midplane that an observer would detect at separation Z = 0, 1, and 2 kpc would be 100%, 22%, and <0.3% of its peak signal. This applies to a disk with a scale height much smaller than the beam, $H<100$ pc. For a thicker disk with an exponential scale height, $H=0.5$ kpc, the signal that would be observed at 1 and 2 kpc above the midplane would be 45% and 0.07% of the midplane brightness, respectively. This scale height approximately matches the observed brightness at 1 kpc but is 4 times too low to explain the brightness at 2 kpc.

Thus, the [C II] emission does not arise solely from the thin disk of star formation in the midplane. This same result was obtained for NGC 891 and NGC 5907, where a two-scale-height fit was found to have a thinner disk of 0.4 kpc and a thick disk scale height of 2.8 kpc [20]. Using those two scale heights, the variation of [C II] with Z in NGC 5775 can be reasonably well fit, suggesting the vertical distribution is similar. By fixing the thinner disk scale height to 0.4 kpc and fitting the vertical profile as shown in Figure 3 (right), we derive a thick disk scale height of $H_2=2.3\pm 0.7$ kpc for NGC 5775.

The diffuse ionized gas is significantly extended on the northeast side of the galaxy, beyond what can be explained by thermal, magnetic, turbulent, or cosmic ray pressure [27]. A possible explanation is that the halo gas is aligned with the direction toward the nearby major galaxy NGC 5774, which lies west of NGC 5775 (see Figure 1) and exerts greater gravitational influence on the negative Z side of the galaxy. Irwin [12] identified an $\mathrm{H}\mathrm{I}$ bridge extending from the northern side of NGC 5775 to NGC 5774, connecting the galaxies and originating (based on the observed radial velocity) from the latter galaxy. The $\mathrm{H}\mathrm{I}$ bridge and associated radio continuum emission form an envelope encompassing both galaxies [14].

3.2. Rotation of Extraplanar Gas

The extraplanar [C II] approximately follows the rotation of the disk, but with a systematic slowing away from the midplane. On the northeastern side of the galaxy, the line center shifts to a lower line-of-sight velocity by $40\pm 8$ km ${\mathrm{s}}^{-1}$ at $\left|Z\right|=2$ kpc, corresponding to a gradient of $20\pm 4$ km ${\mathrm{s}}^{-1}$ per kpc. The apparent lag seen in [C II] is similar in sign and magnitude to the velocity lag observed in H$\mathsf{\alpha }$ on the northeastern side of the galaxy. The lag represents a small fraction of the 150 km ${\mathrm{s}}^{-1}$ amplitude of the rotation curve (Equation (1)) and is below the 250 km ${\mathrm{s}}^{-1}$ velocity resolution, so it is only measured statistically with the SOFIA data. Part of this apparent lag could be caused by line-of-sight geometrical effects, as the galaxy is not perfectly edge-on, and the [C II]-emitting material may not be uniformly distributed along the line of sight. For comparison, the H$\mathsf{\alpha }$ emission has a clear velocity gradient along the minor axis of 25 km ${\mathrm{s}}^{-1}$ per kpc on the northeast side and 6 km ${\mathrm{s}}^{-1}$ per kpc on the southwest side of the galaxy [27], while an earlier study found an average gradient of 8 km ${\mathrm{s}}^{-1}$ per kpc [28]. For stars, vertical gradients of azimuthal velocity may be due to the influence of a spheroidal mass distribution [29]. For the gas, models of ejected clouds following ballistic trajectories show some resemblance to the observed trends but cannot fully explain the kinematics, which may be significantly affected by the interaction with NGC 5774 [30].

3.3. Potential Chimneys and Extraplanar Clouds

Local deviations exist in the overall vertical distribution of extraplanar gas. Notably, [C II] emission is observed even at 3 kpc above the midplane on the northeast side of the galaxy (positive Z and positive Y), and there are some spots of enhanced brightness relative to their surroundings. Figure 5 shows [C II] deviation spectra for four individual regions (the locations of which are labeled in Figure 1 and Figure 2). The deviation spectra were calculated for each position by subtracting the average of reference spectra at the same Z but offset by 1.5 beam widths in $\pm Y$.

Position A is on a prominent radio polarization spur, and it shows a broad velocity distribution, with FWHM 350 km ${\mathrm{s}}^{-1}$ and center −150 km ${\mathrm{s}}^{-1}$. The contours of [C II] and radio polarization closely agree. This is the best example of a potential [C II] in NGC 5775. Position B is on a [C II] spur with no corresponding radio polarization feature. Its spectrum has a broad distribution like position A, but with the center shifted to −80 km ${\mathrm{s}}^{-1}$, primarily due to galactic rotation (as this position is closer to the nucleus). The small-scale structure in the spectrum of position B is finer than the instrumental resolution of 250 km ${\mathrm{s}}^{-1}$, but the centroid shift is well-measured and indicates deviations from symmetric rotation at $+Z$ and $-Z$.

Position C is extraplanar peak of [C II] brightness detached from the disk of NGC 5775. It has a somewhat narrower [C II] velocity distribution, with FWHM 300 km ${\mathrm{s}}^{-1}$. The centroid at +190 km ${\mathrm{s}}^{-1}$ clearly shifted from the other spectra due to galactic rotation (as this position is on the opposite side of the galaxy from the others).

Position D is another [C II] spur, without a clear radio counterpart. The [C II] velocity distribution here is narrow, with FWHM 293 km ${\mathrm{s}}^{-1}$ primarily due to the instrumental resolution.

4. Large-Scale Feedback from the Nuclear Region

The far-infrared surface brightness of the nuclear region is significantly elevated, indicating a high star formation rate from the inner portions of NGC 5775’s disk. The concentrated star formation can lead to a nuclear starburst and superwind, as seen in M 82 [10]. Such winds are broad, extending symmetrically on both sides of the galaxy with a ‘U’-shaped opening. The central black hole can also release vast amounts of energy into its surroundings when material is accreted. Accretion disks generally drive jets parallel to their spin axis, as they must shed angular momentum.

While there is no evidence of a nuclear superwind from the far-infrared continuum or [C II] observations, the ionized extraplanar gas far from the disk is traced using optical H$\mathsf{\alpha }$ emission. NGC 5775 was among the galaxies in studies that showed the extent and frequency of galactic winds potentially driven by supernovae [31,32]. H$\mathsf{\alpha }$ images [33] show vertical filaments from the disk and compared to models of ballistic ejection [30]. NGC 5775 was included in a systematic study of extraplanar H$\mathsf{\alpha }$ emission [34,35], where prominent extraplanar features were detected in both H$\mathsf{\alpha }$ and dust.

In this section, we search for nuclear feedback to complement the SOFIA [C II] observations presented earlier in this paper, while not repeating the existing H$\mathsf{\alpha }$ studies on extraplanar gas. We use archival Hubble Space Telescope images of NGC 5775, which were obtained as part of observation project 10416 (PI J. Rossa). A comparison of these Hubble data to Chandra has been made [17]. The Hubble Legacy Archive continuum and H$\mathsf{\alpha }$ narrow-band filter images were reprojected onto a common grid. The ratio of H$\mathsf{\alpha }$ to continuum filter brightness for starlight was used to scale and subtract the continuum from the H$\mathsf{\alpha }$ image.

Figure 6a shows that a large-scale (>7 kpc) nuclear plume extends southward from the galaxy, directly below the nucleus. A counterplume may also be present in the images, though it is fainter. The plume closely resembles, both in size and morphology, the one found in NGC 3628, which was interpreted as being part of a nuclear starburst wind outflow. Deep optical spectra show that the nuclear plume in NGC 3628 has high ratios of [N II]/H$\mathsf{\alpha }$ and [S II]/H$\mathsf{\alpha }$, suggesting shock origin [36]. Alternatively, the observed line ratio could be due to a dilute radiation field at high Z.

The plume is clearly distinct from the larger-scale disk wind in that it is much more collimated. From its morphology alone, the plume resembles a jet, similar to those produced by supermassive black holes in active galactic nuclei. Although there is no known evidence for an active nucleus in NGC 5775, this does not rule out past activity. The H$\mathsf{\alpha }$ plume could be a relic of a prior active jet.

The H$\mathsf{\alpha }$ surface brightness of the plume in the HST image is approximately $5\times {10}^{-7}$ erg ${\mathrm{cm}}^{-2}{\mathrm{s}}^{-1}{\mathrm{sr}}^{-1}$, measured by making a slice through the plume parallel to the midplane and 5 kpc below it. This corresponds to an emission measure EM $=6T_4^{0.92}{\mathrm{cm}}^{-6}$ pc, where $T_4$ is the gas temperature in units of ${10}^4$ K [37]. There is no evidence of the plume in high-resolution radio images taken with the Very Large Array (VLA; [26]). The predicted radio brightness temperature of the plume at 5 GHz is $T_B=0.3T_4^{-0.35}$ mK. The VLA resolution at this frequency is ${12}^{″}$ and dilutes the plume, which has a diameter $\sim 7^{″}$, further reducing the observable brightness temperature. An approximate upper limit from the VLA images is $T_B<3$ mK, which is not enough to detect this faint structure.

The emission measure is defined as the line-of-sight integral EM $\equiv \int n_{\mathrm{H}}^2{f}^{{\displaystyle \frac{1}{3}}}\mathrm{d}L$, assuming the hydrogen is fully ionized, and has a local density $n_{\mathrm{H}}$ and a volume filling factor f. The gas density is then $n_{\mathrm{H}}\sim 0.1T_4^{0.46}{f}^{-1/6}{\mathrm{cm}}^{-3}$, over the line-of-sight path length $L\sim 700$ pc. Approximating the plume as a cylinder, its total mass is $6\times {10}^7\mu n_{\mathrm{H}}{f}^{{\displaystyle \frac{5}{6}}}$$M_{\odot }$, where $\mu \sim 1.4$ is the mean weight per hydrogen, including helium and heavier elements. The total rate of ionizing photons required to keep the plume ionized, using the case B recombination coefficient [38], is $2\times {10}^{50}{f}^{{\displaystyle \frac{5}{6}}}{\mathrm{s}}^{-1}$. This much ionizing radiation could be provided by as few as four extraplanar O5-type stars or 100 of the more common O9-type stars. However, the present ionized state of the plume does not require extraplanar young stars. Instead, the material could have been injected into the halo at a high temperature, already ionized. The recombination time, using the density estimated above, is ${10}^4{f}^{{\displaystyle \frac{1}{2}}}{T}_{\mathrm{init}}$ yr, where $T_{\mathrm{init}}$ is the peak temperature. If the material had been ejected in a high-energy jet, at a temperature of ${10}^8$ K, it would remain ionized for ${10}^8$ yr. Thus, the plume may be a relic of a past active phase of the supermassive black hole at the center of the galaxy, appearing in a jet-like form despite the current absence of AGN activity in NGC 5775.

5. Conclusions

The observations of NGC 5775 presented in this paper suggest two distinct modes of mass injection into the halo, which are schematically shown in Figure 6b:

(1) A galaxy-wide superwind is produced by a widespread burst of star formation, including the nucleus as well as super star clusters throughout the disk. New far-infrared observations with SOFIA detect far-infrared [C II] emission from NGC 5775 rising up to at least 2 kpc from the midplane, similar to what was found in NGC 891 and NGC 5907 [20]. The extraplanar gas distribution is more extended closest to the nearby major galaxy NGC 5774, which exerts significant gravitational influence.

The surface brightness of the extraplanar [C II] at 2 kpc from the midplane of NGC 5775 (24 nW ${\mathrm{m}}^{-2}{\mathrm{sr}}^{-1}$) is comparable to the northern starburst region of NGC 891 (20 nW ${\mathrm{m}}^{-2}{\mathrm{sr}}^{-1}$). If the [C II] arises from gas where the H is ionized, the implied emission measure is $1000{\mathrm{cm}}^{-6}$ pc, and the gas density is of order $1{\mathrm{cm}}^{-3}$. Such a high gas density at 2 kpc above the midplane exceeds values predicted by state-of-the-art simulations of galactic outflows [39]. If the [C II] instead arises from gas where the H is neutral, an even higher volume density of order $40 {\mathrm{cm}}^{-3}$ is required, due to the lower efficiency of excitation via neutral collisions. Transporting such dense gas to high altitudes above the midplane, or forming it in situ, presents a challenge and may have important implications for the study of circumgalactic multi-phase gas.

The [C II] velocity in NGC 5775 is similar to that of other tracers of the interstellar medium in the disk, but it lags galactic rotation at high altitudes. High-sensitivity polarized radio continuum images may trace individual conduits between the disk and halo from individual super star clusters [40]. Soft (0.2–0.5 keV) X-ray emission reveals ‘spurs’ that extend from the midplane to 2 kpc, with possible H$\mathsf{\alpha }$ and [C II] counterparts, originating approximately 6 kpc along the midplane from the nucleus [16]. These spurs may delineate the walls of a galaxy-scale supershell, driven by the combined star formation activity of the entire inner disk.

(2) A more collimated plume of material, originating in the nucleus, is detected in faint H$\mathsf{\alpha }$ emission. The plume may be a remnant of a nuclear jet from a supermassive black hole that was active in the last ${10}^8$ yr but was not active at the time of the HST observation.