GOALS-JWST: Pulling Back the Curtain on the AGN and Star Formation in VV 114

The MIRI MRS observations include three grating settings (SHORT, MEDIUM, and LONG) in order to cover the entire wavelength range accessible to the four IFU channels (4.9–27.9 μm). A four-point dither pattern was employed to recover the extended emission and avoid saturation, with separate off-source observations for background subtraction. The field of view (FOV) and position angle (PA) vary by channel, but the FOV with full wavelength coverage is defined by channel 1, ∼5'' × 4'' at PA = 255° as shown in Figure 1.

Zoom In Zoom Out Reset image size

Our procedure for reducing the MRS data of the ERS 1328 targets is described in detail by U et al. (2022), but a brief summary is given here: uncalibrated data are processed using the most recently available developmental release of the JWST Science Calibration Pipeline (Bushouse et al. 2022), version 1.8.3, using calibration reference data system (CRDS) context file jwst_0963.pmap. The resulting data products generated by the pipeline are 12 background-subtracted, fringe-corrected, wavelength- and flux-calibrated data cubes–one for every combination of channel and grating setting combining the four dither pointings. These cubes are used to generate 1D spectra of regions of interest and 2D maps of the properties of particular emission line features.

We extract spectra from 10 regions of interest within the channel 1 FOV to investigate IR-bright sources resolved in the MIRI imaging as well as diffuse emission surrounding those sources and in the eastern nucleus (Figure 1). We chose five locations coincident with the two brightest IR sources, the NE and SW cores ((a), (c)), and several bright clumps identified in the MIRI imaging data and in both the submillimeter and radio ((d), (e), and (f)) (Saito et al. 2018; Evans et al. 2022; Song et al. 2022). We also chose five regions intended to capture diffuse emission between (b) and around ((g), (h), (i), and (j)) the bright clumps coincident with some tidal and shock features previously identified using ALMA (Saito et al. 2017, 2018).

Spectra are extracted from each of the 12 data cubes using apertures with a radius of 04 (160 pc). This results in 12 1D spectra for each extracted region, with some overlap in adjacent wavelength regions. The 12 individual bands for each spectra have slight offsets in flux (a few percent) which are multiplicatively scaled, trimmed, and stitched in order to create continuously smooth 1D spectra over the full MIRI wavelength range. This process begins by using the overlapping wavelength region to scale the Channel 4 MEDIUM spectrum to the longest wavelength Channel 4 LONG spectrum (∼23–25 μm), trimming the overlapping values from the noisier spectrum of the two, and continuing the process to channels at shorter and shorter wavelengths. Finally, a wavelength-dependent aperture correction is applied to each spectrum (U et al. 2022).

NIRSpec IFU observations were taken with three grating and filter combinations: G140H/F100LP, G235H/F170LP, and G395H/F290LP to cover the spectral range from 0.97 to 5.3 μm. Calculations in this paper were made using the wavelength region covered by the G235H/F170LP and G395H/F290LP settings, ∼1.7–5.3 μm. This wavelength range allows us to measure the AGN-sensitive 3.3 μm PAH feature. Again a four-point dither pattern was employed to sample the PSF completely and to avoid saturation, with an additional "Leakcal" image taken for each grating setting. The FOV of the combined dither pattern is ∼36 × 38 centered on the SW Core at a PA of ∼32° (Figure 1).

We reduce the NIRSpec data in a similar fashion to the MRS data, using CRDS context file jwst_1009.pmap. Uncalibrated four-point dither pattern science images and Leakcal images, one for each of the two NIRSpec chips, are downloaded using MAST resulting in 10 image files. These are first put through Stage 1 processing with the Detector1 pipeline, which applies detector-level calibrations and produces count rate files calculated from the nondestructive "ramp" readouts. These rate files are then processed using the Spec2 pipeline, which applies physical corrections and flux and wavelength calibrations. At this step in the overall pipeline, the Leakcal files are also used to correct for any stray light that may fall on the detector due to failed open MSA shutters. Finally we run the Spec3 pipeline step, which produces a final combined data cube sampled with 01 spaxels.

For our analysis in this paper, we extract from the final data cube spectra from the two brightest IR sources, the NE and SW cores ((a) and (c), respectively). We matched our apertures to the MIRI MRS extraction radius of 04 centered at the same two locations. For the G395H/F290LP setting this produces flux- and wavelength-calibrated 1D spectra covering 2.87–5.27 μm, with a gap in coverage from 4.06 to 4.18 μm in the middle of the spectrum due to the gap between the two NIRSpec chips. For G235H/F170LP the range is 1.66–3.17 μm with a gap from 2.40 to 2.45 μm. We use the overlapping wavelength range between the NIRSpec and MIRI data to scale the G395H/F290LP spectrum to match the MIRI spectrum, and the overlapping region between the two NIRSpec settings to scale and stitch the shorter wavelength spectra (G235H/F170LP) to the longer wavelength NIRSpec spectra.

We perform fits to features in each 1D MIRI spectrum using the lmfit package (Newville et al. 2014) with the resulting values given in Table 1. Atomic and H₂ emission lines are fit with a single Gaussian component combined with a polynomial fit to the local continuum over a range of 0.1 μm. The resulting Gaussian parameters are used to derive the observed flux of each line. The width of the Gaussian fit is used to determine the intrinsic FWHM of each emission line by subtracting in quadrature the instrumental resolution of MRS at the observed wavelength (Labiano et al. 2021).

Several fine structure lines and H₂ lines are well detected and resolved, as well as the hydrogen recombination lines Pfund α and Humphreys α and β. The emission line ratios show variation between apertures indicative of both widespread star formation and shock excitation. We do not detect any of the high-ionization coronal lines typically found in the mid-IR spectra of AGN-dominated galaxies (e.g., [Ne v]; Genzel et al. 1998; Lutz et al. 2000; Sturm et al. 2002; Weedman et al. 2005; Armus et al. 2007) in contrast to JWST observations of NGC 7469 that show nine well-detected coronal lines excited by high-energy photons from the central AGN (Armus et al. 2023).

Emission line ratios sensitive to AGN activity (e.g., [S iv]/[Ne ii] and [O iv]/[Ne ii]) show values indicative of a composite of AGN and starburst activity (Inami et al. 2013). Apertures (g), (h), and (j) show elevated values of [Fe ii]/Pfα (>5) and (g) and (j) show [Fe ii] FWHM values that are higher than the other apertures (200–300 km s⁻¹).

Apertures (a), (b), (c), (d), and (e) have fine structure lines ([Ar ii], [Ar iii], [Ne ii], [Cl ii], [Ne iii], and [S iv]) with FWHM values ranging between 150 and 200 km s⁻¹ and show no trend with the emission line ionization potential. In the NE core (a) the H₂ FWHM is ∼50 km s⁻¹ narrower than the fine structure lines, while in the SW core (c) the H₂ and fine structure line widths are similar. This trend is reversed in aperture (f) where the H₂ FWHM values are ∼200 km s⁻¹ versus ∼100 km s⁻¹ for the fine structure lines. The broadest emission line widths in our data are observed in the spectrum of aperture (i), which samples the diffuse H₂ bright gas near the western edge of our FOV close to the "overlap" region defined by Saito et al. (2017).

In order to assess the morphology and kinematics of the region observed with MIRI/MRS, we have also carried out spaxel-by-spaxel fits of the [Ne ii] 12.8 μm fine structure emission line and H₂ S(2) 12.3 μm molecular emission line. These two emission lines are generally quite luminous and are well detected across nearly every spaxel. The lines are covered by Channel 3B (13.29–15.52 μm) which provides a wider FOV (7'' × 6'') while still maintaining relatively high spectral resolution (R ∼ 2800–3000) with somewhat larger spaxels (∼02). Fits to the two emission lines are carried out on a spaxel-by-spaxel basis using the Channel 3B sub-band data cube, with a single Gaussian component fit to each line independently, in the same manner as the aperture-extracted spectra. The resulting maps are shown in Figure 2.

Zoom In Zoom Out Reset image size

The variations in the flux and FWHM of both the [Ne ii] and H₂ lines agree with the values measured in our extracted apertures. The broadest FWHMs in both lines lay outside the Channel 1 FOV, and our closest aperture extractions are (i) and (j). The increase in H₂ FWHM at the NW corner of our map corresponds to a portion of the "overlap" region observed by Saito et al. (2017) with a similar increase in FWHM seen at submillimeter wavelengths.

We measure the EW of the 3.3 μm and 6.2 μm PAH emission features (EW_3.3μm and EW_6.2μm, respectively) in the NIRSpec and MIRI spectra by applying the same method to both. First, portions of the spectrum adjacent to each PAH feature are used to perform a linear interpolation of the continuum. We then integrate a spline fit from 3.20 to 3.28 and 5.95 to 6.55 μm (rest frame, EW_3.3μm and EW_6.2μm, respectively) to calculate the PAH feature flux. The integrated flux is divided by the continuum flux density at the wavelength of the peak of each PAH feature.

The EW_6.2μm values range from 0.11–0.72 μm, bracketing the published Spitzer IRS value of 0.3 μm (Stierwalt et al. 2013). The NE core has an EW_6.2μm of 0.264 μm and a relatively high EW_3.3μm of 0.121 μm, while the SW core has a low EW_6.2μm of 0.106 μm and a very low values of EW_3.3μm of 0.016 μm, indicative of an AGN (e.g., Imanishi et al. 2010; Petric et al. 2011; see the discussion in Section 4).

The largest EW_6.2μm of 0.72 μm is measured for source (f), a very young, highly enshrouded star cluster revealed by JWST NIRCam (Linden et al. 2022). The remaining apertures have a range of values from ∼0.20 to 0.65 μm tracing the presence of extended star formation and the influence of the AGN in the SW core.

To compare with previously published values we also calculate the apparent 9.7 μm silicate absorption feature strength in a manner consistent with Spoon et al. (2007)'s measurements of PAH-dominated sources. We assume an extrapolated power-law fit for the continuum, constrained by portions of the spectrum at 5.5 μm and 14 μm and take the natural logarithm of the ratio of the observed and interpolated continuum flux density at 9.7 μm (s_9.7μm).

The majority of the s_9.7μm values measured range from −0.73 to −1.44, bracketing the published Spitzer IRS measurement of −0.98, with the exception of the NE core (a). Visual inspection of the NE core's spectrum indicates a much deeper silicate absorption feature, measured to be s_9.7μm = −2.45. For absorption-dominated sources Spoon et al. (2007) used a spline fit to determine a continuum. If we assume the NE core is absorption-dominated and apply the same process, the measured s_9.7μm would be stronger (−2.87). Several other features unique to the spectrum of the NE source are shown in Figure 3 including aliphatic hydrocarbon absorption and crystalline silicate absorption, both observed in ULIRGs with deep 9.7 μm silicate absorption (Spoon et al. 2007). We note that these strong absorption features are not seen in either the SW core (c) or the embedded cluster (f).

Zoom In Zoom Out Reset image size

The MIRI spectra allow high-resolution vibration−rotation spectroscopy of gas-phase molecules toward dust-enshrouded regions in the VV 114 system. We report the detection in absorption of the ν₂ 14.04 μm bending mode of hydrogen cyanide (HCN) and tentatively also the ν₅ 13.7 μm bending mode of acetylene (C₂H₂) toward the NE core (a). C₂H₂ is a key ingredient in the gas-phase formation of large molecules such as HC₃N, and HCN is one of the most abundant nitrogen bearing molecules in dense (n > 1 × 10⁴ cm⁻³) molecular clouds.

These lines have been previously detected by Spitzer toward dust-enshrouded young stellar objects (YSOs) (e.g., Lahuis & van Dishoeck 2000) and toward LIRGs (Lahuis et al. 2007). Lahuis et al. (2007) find HCN column densities ranging between N(HCN) = 1−12 × 10¹⁶ cm⁻² and warm gas with excitation temperatures T_ex = 230–700 K. Owing to the higher spatial and spectral resolution, the JWST MIRI spectrum of the NE core is more complex than those found by Lahuis et al. (2007) using Spitzer.

We perform preliminary fits using the methodology of Lahuis & van Dishoeck (2000) which assumes LTE. Our preliminary results are consistent with an excitation temperature of 300–500 K and an N(HCN) of 1−5 × 10¹⁶ cm⁻². For an HCN abundance (with respect to H₂) of 10⁻⁸–10⁻⁷ (e.g., Schilke et al. 1992; Lahuis & van Dishoeck 2000; Harada et al. 2013), this would be consistent with a high obscuration with N(H₂) 10²³–10²⁴ cm⁻². The HCN and C₂H₂ spectra require further analysis, including multiple temperature component modeling and inclusion of non-LTE effects (V. Buiten et al. 2023, in preparation).

It is also possible that the nuclear obscuration is even higher with column densities in excess of N(H₂) 10²⁵ cm⁻²—the so called compact obscured nuclei (CONs; e.g., Aalto et al. 2015). Such objects are characterized by high millimeter and submillimeter continuum surface brightnesses and luminous emission from millimeter-wave rotational transitions of HCN within the vibrationally excited ladder (HCN-vib) (e.g., Sakamoto et al. 2010; Aalto et al. 2015, 2019; Falstad et al. 2021; Sakamoto et al. 2021). For such deeply enshrouded objects, the HCN 14 μm line, or silicate absorption, may not trace the full N(H₂), but only its surface. The millimeter/submillimeter HCN-vib line (and the millimeter/submillimeter continuum) would probe deeper, revealing the full obscuring column. Falstad et al. (2021) propose that CONs have surface brightnesses of HCN-vib of Σ(HCN-vib) > 1 L_⊙ pc⁻². A recent ALMA study by Saito et al. (2018) does not detect HCN-vib emission toward the NE Core of VV 114 with a limit of Σ(HCN-vib) < 0.12 L_⊙ pc⁻². The relatively faint millimeter continuum found by Saito et al. (2017) is consistent with the HCN-vib non-detection. This suggests that the NE core either does not fulfill the CON criteria of Falstad et al. (2021) or that the radius of the CON region is smaller than 12 pc.

Previous multiwavelength observations of VV 114 hinted at an AGN contribution in the X-ray, mid-IR, and submillimeter (e.g., Le Floc'h et al. 2002; Grimes et al. 2006; Iono et al. 2013). As the brightest compact sources from the mid-IR to radio, the NE and SW cores are the most likely to harbor an AGN, with Iono et al. (2013) identifying the NE core as a potential AGN. We do not detect coronal lines in either the NE or SW cores, but this does not rule out an AGN; ULIRG Mrk 231 is a well-known optically classified Seyfert 1 with no observed coronal lines in the mid-IR. Instead, very low PAH EW and silicate absorption indicate the presence of the AGN in Mrk 231 (Armus et al. 2007; Imanishi et al. 2010; Inami et al. 2013; Stierwalt et al. 2014).

To compare with other AGN and starburst-dominated LIRGs we plot our measured values on several PAH diagnostic diagrams (Figure 4). We first plot s_9.7 versus EW_6.2μm (Figure 4(d)): the apertures extracted from diffuse emission (with the exception of (d)) as well as the star-forming clump in aperture (f) have values consistent with starburst galaxies and lie near the literature values for M82 (Spoon et al. 2007). The SW core shows contribution from an AGN (e.g., Spoon et al. 2007; Marshall et al. 2018) and lies near Mrk 273, a ULIRG with a Seyfert 2 nucleus (Armus et al. 2007; Stierwalt et al. 2013; U et al. 2013). Aperture (d) is directly adjacent to the SW core, in fact partially overlapping, and is likely showing some contribution from the AGN that is more clearly seen in the SW core. The NE core has a correspondingly higher obscuration and lies near the values observed for ULIRG Arp 220, a deeply embedded starburst that also has C₂H₂ and HCN absorption crystalline silicate features in its mid-IR spectrum (Spoon et al. 2006; Lahuis et al. 2007).

Zoom In Zoom Out Reset image size

The spectrum of the SW core at shorter IR wavelengths is consistent with the heating and processing of dust by an AGN, which reduces the 3.3 μm PAH EW (Figures 4(a)–(c)) as well as the 6.2 μm EW (EW_3.3μm < 0.04 μm and EW_6.2μm < 0.20 μm; Imanishi et al. 2010; Petric et al. 2011). The 3.3 μm PAH feature is directly adjacent to a broad absorption feature caused by H₂O ice that may impact our continuum measurement and in turn the measured EW_3.3μm (Imanishi et al. 2008; Inami et al. 2018). Moreover, the attenuation of the continuum and the 3.3 μm feature may be different depending on the geometry of the dust and PAH emisison (Lai et al. 2020). We make a simple estimate of the impact by performing a power-law fit to the 1.7–5.0 μm spectrum and dividing our integrated flux by the continuum flux density at 3.3 μm. This decreases the EW_3.3μm slightly from 0.12 μm and 0.017 μm to 0.11 μm and 0.015 μm for the NE and SW core, respectively, but does not affect our conclusions regarding the nature the NE and SW cores.

Inami et al. (2018) used AKARI to propose revised AGN diagnostics including the 3.3 μm PAH feature as well as the F_ν (4.3)/F_ν (2.8) flux density ratio. When placed on a plot of EW_6.2μm versus EW_3.3μm (Figure 4(e); following Inami et al. 2018), the SW core lies in a region populated by known AGNs while the high EW_3.3μm of the NE core is more consistent with ULIRGs dominated by star formation. We also place the NE and SW cores on the EW_3.3μm versus flux density ratio diagnostic proposed by Inami et al. (2018) in Figure 4(f). The SW core again clearly lies in the AGN-dominated portion of the diagram.

To estimate the contribution of the AGN in the SW core to the total luminosity of VV 114 we take the estimation of L_IR ∼ 5 ± 0.5 × 10¹⁰ L_⊙ by Evans et al. (2022), about 12% of the total L_IR, as an upper limit. The presence of PAH features in the SW core indicates a blended contribution to the IR of star formation and AGN. Sources with similar EW_3.3μm and colors have a bolometric AGN contribution of ∼30%–50% (Díaz-Santos et al. 2017; Inami et al. 2018), which means the contribution of the AGN to the total luminosity of the entire VV 114 system is ∼5%.

Although Iono et al. (2013) identified the NE core as potentially harboring an AGN due to the enhanced HCN/HCO+ ratio, an analysis of X-ray and millimeter data by Privon et al. (2020) showed that in fact the HCN/HCO+ is not a robust indicator of total AGN luminosity or its fractional contribution to IR luminosity. Our findings regarding the nature of the two bright cores in VV 114E are not in agreement with the analysis of JWST MIRI data by Donnan et al. (2023) who propose the NE core as a potential AGN host based on its compact nature.

Interestingly, the NE core does have a somewhat low EW_6.2μm despite a strong EW_3.3μm, placing it in a region of Figure 4(e) with few other galaxies. The galaxies with the most similar values to the NE core are II Zw 96 and IRAS F19297-0406, which also have similar s_9.7μm when comparing the values measured using Spitzer for all three galaxies (Stierwalt et al. 2013). II Zw 96 is an unusually compact and powerful starburst (Inami et al. 2010, 2022) and IRAS F19297-0406 has a powerful starburst-driven outflow (Soto et al. 2012; Veilleux et al. 2013), physical conditions which may warm dust in a way that lowers EW_6.2μm. The curious nature of both the NE and SW cores warrants a more thorough follow-up analysis decomposing the relative contributions of the stellar, PAH, dust, and AGN components of the SED.

Evidence for extended shocks in VV 114 has previously been suggested at visible wavelengths via enhanced [O i] and [S ii] emission and broadened line profiles across VV 114 (Rich et al. 2011, 2015) and in the submillimeter from the presence of methanol (CH₃OH) in the "overlap" region between VV 114W and E. Apertures (f), (g), (h), and (i) lie closest to the overlap region and both emission line ratios and molecular and atomic line widths show evidence of shocked gas.

The atomic lines in aperture (f) all have narrow line widths (∼100–150 km s⁻¹) and ratios typical of buried, young star-forming regions as this source was revealed to be in (Linden et al. 2022). Aperture (i) encompasses a fainter star-forming knot seen in both MIRI and NIRCAM imaging, and several emission line profiles show a double peaked profile. The broader H₂ lines in apertures (f) and (i) (∼200–300 km s⁻¹) are similar to the values extracted just to the north in apertures (g) and (h), which show elevated [Fe ii]/Pfα values indicative of shocks, also seen in MRS observations of NGC 7469 (U et al. 2022).

These regions are consistent with ALMA observations of VV 114 where the elevated FWHM of 150–300 km s⁻¹ in the molecular gas (e.g., CO(1–0), CH₃OH) is found in the "overlap" region and is suggested to be the product of both shocks, and overlapping kinematic components due to the merger (Saito et al. 2017, 2018) and may be similar in nature to the shocked bridge seen in Stephan's Quintet (Guillard et al. 2012b; Appleton et al. 2017). However, the clusters in this region of VV 114 are 1–2 orders of magnitude more massive and ∼2–3 times dustier than those seen in the bridge of Stephan's Quintet (Fedotov et al. 2011; Linden et al. 2022).

Resolved emission line profiles in aperture (j) show a strong indication of both blue and redshifted wings, potentially due to projection effects of the tidal arm that extends from VV 114W across the IR-bright cores southward beyond the FOV of our aperture extractions (Saito et al. 2015). These values are supported by examining the [Ne ii] emission line profiles which fall in Channel 3 and has a wider FOV than Channel 1. Looking several arcseconds southeast of the SW core, the [Ne ii] line profiles show significant broadening and wings in several spaxels (Figure 2). We also see enhanced H₂/[Ne ii] in this region and at the furthest SE region of our spaxel-by-spaxel maps. The value of [Fe ii]/Pf α ∼ 9 in aperture (j) also indicates the presence of shocked gas which likely extends beyond the Channel 1 FOV and follows the elevated emission line ratios caused by shocks seen in visible light IFU observations (Rich et al. 2011).

Aperture (b) appears to be dominated by diffuse emission when examining the MIRI and NIRCAM images, but does not display the same characteristics of shock excitation as the other apertures that trace diffuse gas. Previous observations of shock excitation in VV 114 have suggested both galactic winds and tidally driven gas flows as sources of shock excitation (Rich et al. 2011; Saito et al. 2017). The JWST data show some shocked gas coincident with tidal features previously observed in the submillimeter and radio as well as kinematic profiles potentially associated with galactic winds. Follow-up work mapping the 2D kinematics of the molecular and atomic gas, as well as the temperature and distribution of H₂ gas using these data will provide a more complete picture of the shock excitation in VV 114, especially when combined with wide field optical IFU observations from MUSE.

If we assume that the atomic and fine structure line emission is dominated by star formation in the apertures with bright star-forming clumps ((a), (e), and (f)), we can calculate star formation rates (SFRs) using a hydrogen recombination line flux or neon emission. If we scale either Pfund or Humphreys α to Hα assuming Case B recombination (Hummer & Storey 1987) and use Equation (2) of Murphy et al. (2011) the estimated SFR for each spectrum is ∼1 M_⊙ yr⁻¹. Using the [Ne ii] and [Ne iii] fluxes and Equation (3) of Ho & Keto (2007) yields an SFR from ∼1.5−2.5 M_⊙ yr⁻¹. These values are consistent with those found by Song et al. (2022) using the radio emission of bright knots measured with the VLA and amount to ∼2%–3% of the total SFR per aperture.