THE BURIED STARBURST IN THE INTERACTING GALAXY II Zw 096 AS REVEALED BY THE SPITZER SPACE TELESCOPE

Spitzer observations of II Zw 096 were taken with the Infrared Array Camera (IRAC; Fazio et al. 2004) at 3.6 μm, 4.5 μm, 5.8 μm, and 8.0 μm on 2004 October 29 and the Multiband Imaging Photometer for Spitzer (MIPS; Rieke et al. 2004) at 24 μm, 70 μm, and 160 μm on 2004 November 30 (PID 3672; PI: J. Mazzarella). All IRAC data were collected with five frames of 30 s high dynamic range (HDR) mode. The 8.0 μm data saturated the 30 s integrations, so only the short 1.2 s frames were used. Total exposure times were 150 s at 3.6 μm, 4.5 μm, and 5.8 μm and 6 s at 8.0 μm. The MIPS images were taken in photometry mode, using super-resolution at 70 μm. Sixteen images of 3 s each were obtained. The total exposure times were 48.2 s, 37.7 s, and 25.2 s for 24 μm, 70 μm, and 160 μm, respectively. A summary of the Spitzer observations is given in Table 1.

We used the MOPEX²¹ software package to process the IRAC and the MIPS images, correcting the background, aligning, resampling the images, removing bad pixels, combining into mosaics, and clipping cosmic-ray events. Final mosaics were constructed with a pixel size of 06 pixel⁻¹ for the IRAC images, 18 pixel⁻¹ at 24 μm, 40 pixel⁻¹ at 70 μm, and 80 pixel⁻¹ at 160 μm for the MIPS images.

The World Coordinate System of the images is limited to ∼04 by the telescope pointing accuracy. However, alignment of the IRAC images has been refined with the Two Micron All Sky Survey (2MASS) to achieve a final accuracy (⩽03). For the MIPS images, the positional error is approximately 07 based on the startracker-to-boresight uncertainty and the MIPS scan mirror uncertainty (M. Lacy et al. 2010, in preparation). A pointing refinement with 2MASS cannot be done because of the large wavelength difference between the MIPS and 2MASS bands.

All photometric fluxes were derived using the IDL APER routine. For the IRAC images, we used 14 and 29 radii circular apertures, using the aperture corrections from Surace et al. (2004). For the MIPS images, we fit the point-spread function (PSF) to the peak of emission to estimate total flux.

The HST FUV and optical images were obtained with the Solar Blind Channel (SBC) and the Wide Field Camera (WFC) on the Advanced Camera for Surveys (ACS) on 2008 May 1 (PID 11196; PI: A. Evans) and 2006 April 15 (PID 10592; PI: A. Evans), respectively. The data were taken in the ACCUM mode using the LINE dither pattern for the WFC and the PARALLELOGRAM dither pattern for the SBC. The near-infrared images were taken using the NIC2 camera on the Near-Infrared Camera and Multi-Object Spectrometer (NICMOS) with the MULTIACCUM mode and the SPIRAL dither pattern on 2008 June 1 (PID 11235; PI: J. Surace). The following filters were used: ACS/SBC F140LP (FUV), ACS/WFC F435W (B band), F814W (I band), and NICMOS/NIC2 F160W (H band) for 2528, 2520, 1440, and 2304 s, respectively. We summarize the HST imaging observations in Table 1.

The HST data reduction began with the calibrated data products produced by STScI. The ACS data were reduced with the PyDrizzle software included in IRAF/STSDAS provided by STScI, to identify and reject cosmic rays and bad pixels, to remove geometric distortion, and to combine the images into mosaics. The NICMOS images were reduced using IRAF and IDL tasks designed to remove cosmic rays and bad pixels, and correct for the different bias levels across the quadrants due to the four amplifiers. The alignments of the HST images were improved based on the 2MASS astrometric solution. IRAF was used to construct the final mosaic, and the IDL routine APER was used to measure the photometry. An aperture size of 025, with corresponding aperture corrections of 1.18, 1.16, 1.16, and 1.39, were used for photometric measurements of point sources in the final ACS/SBC F140LP, ACS/WFC F435W, F814W, and NICMOS F160W images, respectively.

X-ray observations of II Zw 096 were obtained with the AXAF CCD Imaging Spectrometer (ACIS) on the Chandra X-ray Observatory with VFAINT mode on 2007 September 10 (PID 8700551; PI: D. Sanders). The background counts for total integration times 14.56 ks are 0.0025 counts pixel⁻¹ and 0.004 counts pixel⁻¹ in the soft (0.5–2 keV) and the hard (2–7 keV) bands, respectively. The 5σ depths of the images are 1 × 10⁻¹⁵ erg s^-1 cm⁻² in the soft band and 5 × 10⁻¹⁵ erg s^-1 cm⁻² in the hard band. The fundamental data reduction was done by the Chandra data analysis software CIAO version 3.4 and we have verified that there are no anomalous objects left in the data. Then we selected soft (0.5–2 keV) and hard (2–7 keV) X-ray energy ranges for making images and smoothed the data using a Gaussian filter with a σ of 1.5 pixels (FWHM ∼18).

II Zw 096 was observed with all modules of the Spitzer Infrared Spectrograph (IRS; Houck et al. 2004) in staring mode on 2006 November 12 (PID 30323; PI: L. Armus). The slits were centered on the peak of the 8 μm emission. The slit positions for all four modules are overlaid on the IRAC 8 μm image in Figure 1. The number of nod cycles and the integration time in each nod position are listed in Table 1.

Zoom In Zoom Out Reset image size

The data were reduced using the S15.3 IRS pipeline at the Spitzer Science Center. The pipeline software removes bad pixels and droop, subtracts the background, corrects linearity, and performs a wavelength and flux calibration. The backgrounds were subtracted with dedicated sky observations for the high-resolution data, and with off-source nods for the low-resolution data. One-dimensional spectra were extracted with the SPICE²² software package, using the standard point-source extraction apertures for all slits. For the high-res slits, this is equivalent to a full-slit extraction. The slit widths of Short-Low (SL), Long-Low (LL), Short-High (SH), and Long-High (LH) are 36, 105, 47, and 111, respectively. The extraction size for SL is 95, LL is 383, SH is 117, and LH is 228.

The near-infrared spectra of II Zw 096 were taken with AKARI on 2009 May 12 and 13 (PID GOALS; PI: H. Inami) during the post-helium (warm phase) mission using the Infrared Camera (IRC; Onaka et al. 2007). The 2.5–5 μm spectra were obtained using the 1' × 1' aperture (the Np aperture) using a moderate-resolution grism. The spectral resolving power is ∼120 at 3.5 μm for a point-like source (∼5'' FWHM). The use of the Np aperture is intended to avoid the confusion of spectra but multiple sources in this aperture may interfere each other (see Section 3.5). The observations used the IRCZ4 Astronomical Observation Template, taking four spectra, a reference image, and then five spectra at the end of the sequence. The exposure time for each spectrum was 44.41 s. Two pointings were used (in total 2 pointings × 9 frames) for the observations and the total exposure time was 799 s.

The data reduction was performed with the standard software package provided by JAXA, the IRC Spectroscopy Toolkit for Phase 3 data Version 20090211.²³ This software performs dark subtraction, linearity correction, flat correction, background subtraction, wavelength calibration, spectral inclination correction, and spectral response calibration. We extracted the one-dimensional spectra from the two-dimensional spectral image using an aperture 3 pixels wide (44). The number of bad pixels is increasing dramatically in the post-helium mission due to the elevated detector temperatures. A 3 × 3 moving window was used to isolate and replace bad pixels in each spectral image before co-adding the data and extracting the one-dimensional spectrum. The wavelength accuracy is evaluated to be 0.01 μm and the uncertainty in absolute flux calibration is estimated to be ∼10% (Ohyama et al. 2007).

In Figure 1, the Spitzer images from IRAC 3.6 μm through MIPS 70 μm are presented in order of increasing wavelength. Three distinct objects are seen in the IRAC data. The nuclei of the merging galaxies are to the north (source B) and south (source A; Goldader et al. 1997). In addition, there is a bright, red source to the east (sources C and D), which is resolved and elongated at an angle of roughly 60°. This resolved IRAC source is coincident with the near-infrared sources "C" (R.A. = 20$\mathrm{^h}$57$\mathrm{^m}$2447, decl. = +17°07'39 9'') and "D" (R.A. = 20$\mathrm{^h}$57$\mathrm{^m}$2434, decl. = +17°07'39 1'') from Goldader et al. (1997). We estimate the flux density of source D at 3.6 μm, 4.5 μm, 5.8 μm, and 8.0 μm, as 2.85 ± 0.08, 4.67 ± 0.13, 12.7 ± 0.2, and 41.9 ± 0.8 mJy, respectively, using a circular aperture size of 14 radius.

The MIPS 24 μm image shows a strong point source coincident with the position of source D, along with a lot of emission to the southwest, coincident with source A. Although the MIPS 70 μm image is unresolved, the position of the peak is consistent with source D. PSF fitting to source D at 24 μm and 70 μm suggests that it produces 67.2% ± 10.6% (1.37 ± 0.18 Jy) and 87.9% ± 13.8% (10.2 ± 1.2 Jy) of the total galaxy flux at 24 μm and 70 μm, respectively. The angular separation of sources C and D is ∼20 (see Figure 2 or 4). Although this small separation cannot be resolved with MIPS, the centroid of source D can be determined much more accurately than the FWHM of the beam. In addition, we have simulated the MIPS image using two point sources separated by 20, determining that a flux ratio of 4:1 (D:C) is required to match the data. The photometry is summarized in Table 2.

Zoom In Zoom Out Reset image size

Table 2. Spitzer, HST, and Chandra Photometries of II Zw 096

Observatory	Instrument	Filter or Module	Source A	Source C	Source D
			20$\mathrm{^h}$57$\mathrm{^m}$2409, +17°07'35 3''	20$\mathrm{^h}$57$\mathrm{^m}$2447, +17°07'39 9''	20$\mathrm{^h}$57$\mathrm{^m}$2434, +17°07'39 1''
Spitzer	IRAC	3.6 μm	3.29 ± 0.10 (29)	2.26 ± 0.08 (14)	2.85 ± 0.08 (14)
		4.5 μm	2.68 ± 0.08 (29)	3.33 ± 0.08 (14)	4.67 ± 0.13 (14)
		5.8 μm	8.43 ± 0.18 (29)	9.36 ± 0.22 (14)	12.7 ± 0.2 (14)
		8.0 μm	28.1 ± 0.5 (29)	27.2 ± 0.4 (14)	41.9 ± 0.8 (14)
	MIPS	24 μm	...	...	1.37 ± 0.18a
		70 μm	...	...	10.2 ± 1.2a
		160 μm	...	...	...
	ACS/SBC	F140LP	(6.59 ± 0.30) × 10−15 (29)	...	...
HST	ACS/WFC	F435W	(3.03 ± 0.02) × 10−15 (29)	...	(3.43 ± 0.34) × 10−17 (14)
					(4.63 ± 0.67) × 10−18 (025)
		F814W	(1.28 ± 0.01) × 10−15 (29)	(5.57 ± 0.19) × 10−17 (14)	(4.85 ± 0.20) × 10−17 (14)
				(4.64 ± 0.39) × 10−18 (025)	(2.85 ± 0.08) × 10−17 (025)
	NICMOS/NIC2	F160W	5.36 ± 0.05 (29)	0.85 ± 0.02 (14)	0.74 ± 0.02 (14)
				0.13 ± 0.01 (025)	0.24 ± 0.01 (025)
Chandra	ACIS	0.5–2 keV	6.6 × 1040	1.8 × 1040 (Source C+D)
		2–7 keV	1 × 1041	1 × 1041 (Source C+D)

Notes. The values in the brackets are the circular aperture sizes in radii. The units are mJy for IRAC and NICMOS/NIC2, Jy for MIPS, erg s^-1 cm⁻² A⁻¹ for ACS/SBC and ACS/WFC, and erg s^-1 for ACIS. ^aMeasured by the PSF fitting (see Section 3.1).

Download table as:  ASCIITypeset image

We estimate the total infrared (TIR, 3–1100 μm) luminosity of source D to be TIR = 5.85 × 10¹¹ L_☉ from the Spitzer 24 and 70 μm images, using Equation (4) in Dale & Helou (2002), assuming that the 160 μm flux of source D is 87.9% of the total at 70 μm. This is approximately 80% of the TIR of the system. II Zw 096 has an infrared luminosity (LIR, 8–1000 μm) of 10^11.94 L_☉ (Armus et al. 2009). Assuming source D accounts for ∼80% of the total LIR of the system (the same fraction as TIR), we estimate an LIR of 6.87 × 10¹¹ L_☉.

HST/ACS and NICMOS images of II Zw 096 are shown in Figure 2, from the upper left to the lower right, ACS/SBC F140LP (FUV), ACS/WFC F435W (B band), F814W (I band), and NICMOS/NIC2 F160W (H band). The contours of the Spitzer 24 μm image are overlaid in each panel.

The I-band and the H-band images with the 24 μm contour overlayed in Figure 2 suggest that source D is the origin of the bulk of the far-infrared emission in II Zw 096. The contour of the MIPS 70 μm image is also consistent with the position of source D.

The HST NICMOS images resolve the structure of the red sources into at least 10 "knots" (see Figure 3). Most of these knots are not seen in the visible image due to heavy extinction. In the H band, the absolute magnitudes of sources C and D are M_H = −18.7 ± 0.1 mag and −19.5 ± 0.1 mag, respectively, measured with a 025 radius aperture. We estimate that approximately 0.81% ± 0.04% and 1.62% ± 0.06% of the H-band flux of the entire system comes from sources C and D alone. Sources C and D provide about 7.78% ± 0.39% and 15.5% ± 0.6% of the H-band emission for the red complex in Figure 3, respectively.

Zoom In Zoom Out Reset image size

In Figure 4, we show a false-color RGB image of II Zw 096 made from the NICMOS H band, ACS B band, and ACS FUV images. The extremely red sources C and D are obvious in this figure, as are a number of unresolved blue sources to the west. These are presumably young star clusters in the northern spiral arm of the western galaxy and the overlap region.

Zoom In Zoom Out Reset image size

The Chandra X-ray images are shown in Figure 5. In the full (0.5–7 keV) and soft (0.5–2 keV) bands, source A dominates the emission. In the hard-band (2–7 keV) image, sources C and D become much more prominent. The 0.5–2 keV luminosities of source A and source C+D complex are 6.6 × 10⁴⁰ erg s^-1 and 1.8 × 10⁴⁰ erg s^-1, respectively. In the hard band, the 2–7 keV luminosity of source A is 1 × 10⁴¹ erg s^-1 and the total of sources C and D is 1 × 10⁴¹ erg s^-1.

Zoom In Zoom Out Reset image size

The C+D complex is clearly extended (see Figure 5). Source C dominates the soft-band emission from the complex, with the fraction of the X-ray emission coming from source D apparently rising toward the hard band. This is evidenced by a gradual shifting of the peak from NE to SW in Figure 5. However, given the low number of total counts (∼20) from this region and the strong overlap of source C and D, it is difficult to derive an independent measure of the hardness ratio of C and D, separately. Together, the hardness ratio (Hard − Soft/Hard + Soft) of source C+D is 0.05 ± 0.14.

The Spitzer IRS SL, LL, SH, and LH spectra of sources C and D (C+D) are shown in Figure 6, and the fluxes of the detected lines are shown in Tables 3 and 4. Note that the LL and LH spectra cover not only source C+D but also source A due to their limited resolutions (see Figure 1(d)). However, the scale factor of LL/SL is 1.09, indicating the continuum is dominated by source C+D only. The IRS low-resolution spectra are dominated by strong PAH emission and silicate absorption. The high-resolution data show strong fine-structure line emissions such as [Ne iii], [Ne ii], and [S iii].

Zoom In Zoom Out Reset image size

From the SL data, we measure 6.2 μm and 11.2 μm PAH equivalent widths (EQW) of 0.26 and 0.14 μm, respectively, and line fluxes of (5.57 ± 0.19) × 10⁻¹⁶ W m^-2 and (3.24 ± 0.32) × 10⁻¹⁶ W m^-2, respectively, using a simple spline fit to the local continuum under the features. The apparent silicate optical depth at 9.7 μm is τ_9.7 ∼ 1 when a screen geometry is assumed. This implies an A_V ⩾ 19 mag toward source C+D (Roche & Aitken 1984). This is comparable with the estimate of 1–2 mag in the K band from the near-infrared colors (Goldader et al. 1997).

We do not see evidence for 14.3 μm [Ne v] emission in the SH spectrum. A 3σ upper limit of the [Ne v] line flux is <0.10 × 10⁻¹⁶ W m^-2. The [Ne ii] line has a flux of (2.74 ± 0.47) × 10⁻¹⁶ W m^-2, implying a 3σ upper limit on the [Ne v]/[Ne ii] line flux ratio of 0.04. The 25.9 μm [O iv] line is not detected, with a 3σ upper limit of <0.29 × 10⁻¹⁶ W m^-2, implying an upper limit on the [O iv]/[Ne ii] line flux ratio of 0.11. The 18.71 μm [S iii] line fluxes are detected in both of SH and LH: (2.16 ± 0.22) × 10⁻¹⁶ W m^-2 in SH and (3.27 ± 0.04) × 10⁻¹⁶ W m^-2 in LH. The large value for 18.71 μm [S iii] in LH implies that there is measurable emission line flux from source A. The flux of the [S iii] line at 33.48 μm is (4.77 ± 0.07) × 10⁻¹⁶ W m^-2. Using the [S iii] 18.71/[S iii] 33.48 line flux ratio in LH of 0.69 ± 0.01, we derive an ionized gas density of 250 cm^-3 at 10⁴ K. The density is an average value over the area enclosed by LH which includes sources C, D, and A.

The fluxes of H₂ S(3), H₂ S(2), and H₂ S(1) lines are (0.45 ± 0.05) × 10⁻¹⁶ W m⁻², (0.29 ± 0.03) × 10⁻¹⁶ W m⁻², and (0.73 ± 0.05) × 10⁻¹⁶ W m⁻², respectively. Taking these lines, and assuming an ortho-to-para ratio of 3.0, we estimate the warm molecular gas has a temperature of 329 K and a total mass of 4.5 × 10⁷ M_☉.

If we fit the entire IRS low-res spectrum using PAHFIT (Smith et al. 2007), we can estimate a total PAH flux (see Table 4) and compare this to the far-infrared flux which we estimate originates from source C+D. When this is done, we derive a ratio of $F_{{\rm PAH}_{\rm tot}}/F_{{\rm IR}} \sim 0.02$. Since the continuum emission at the overlap wavelengths of SL and LL has a difference less than 10%, dust emission in our low-res spectrum is dominated by source C+D.

The spectrum of sources C and D taken with AKARI is well separated from that of the other sources (A and B) in II Zw 096, although they are all in the Np aperture, because the direction of spectral dispersion is nearly perpendicular to the vector separating C, D, and A. The AKARI near-infrared spectrum of source C+D is shown in Figure 7. Strong 3.3 μm PAH emission, 3.4 μm emission from aliphatic hydrocarbon grains, and Brα emission lines are detected. The line fluxes are (2.37 ± 0.14) × 10⁻¹⁶ W m^-2 and (0.50 ± 0.05) × 10⁻¹⁶ W m^-2 for the PAH and the Brα, respectively, after fitting both with a Gaussian function (see Tables 3 and 4). The EQW of the 3.3 μm PAH is 0.14 μm. Because the AKARI spectral resolution between 2.5and5 μm is only R ∼ 120, the Brα recombination line is not resolved.

Zoom In Zoom Out Reset image size

The spectrum of source A (not shown) also exhibits 3.3 μm PAH and Brα emission. The 3.3 μm PAH has a flux of (1.77 ± 0.12) × 10⁻¹⁶ W m^-2 and an EQW of 0.08 μm. The Brα line flux is (0.33 ± 0.06) × 10⁻¹⁶ W m^-2.

The high-resolution HST imaging data show that a large number of bright star clusters surround the II Zw 096 nuclei and are present throughout the merging disks (see Figure 2). From the B- and I-band images, we identify 128 clusters with apparent magnitudes of 18.3 mag ⩽ m_B ⩽ 26.7 mag (Vega) and 17.9 mag ⩽ m_I ⩽ 25.7 mag (Vega) with 80% completeness limits of 26.0 mag and 25.5 mag, respectively. Of these, 97 and 88 are detected in the FUV and H-band images, respectively. For all clusters, we use a 01 radius circular aperture for the HST photometric measurements. We estimate that the clusters contribute ≃14% of the total FUV luminosity (≈8 × 10³⁶ W) in II Zw 096 (uncorrected for reddening).

In Figure 8, we present an FUV–optical (F140LP−F435W versus F435W−F814W) color–color diagram of the clusters. We additionally show the colors of a number of diffuse emission regions throughout the merger system. For the diffuse emission regions, we use 05–08 radius circular apertures, depending on the location in the FUV image to avoid crowded regions.

Zoom In Zoom Out Reset image size

Two potential explanations for the observed color distribution are considered. First, if all the clusters are coeval, then their different colors result from variable extinction. In other words, those clusters or diffuse emission regions which have colors of F435W−F814W ≳ 1.0 mag or F140LP−F435W ≳ 0.0 mag are heavily reddened (A_V ≳  3 mag). When we de-redden an instantaneous starburst model of Bruzual & Charlot (2003), all of the clusters and the diffuse regions are younger than ∼5 Myr. On the other hand, when we pick an exponentially declining (SFR = 1 M_☉τ⁻¹e^(−t/τ) for τ = 1 Gyr, 0Gyr ⩽ t ⩽ 20 Gyr) or a continuous star formation model (Bruzual & Charlot 2003), they have ages between approximately 1and100 Myr. However, we note that the points of cluster colors are not aligned well with the extinction vector. In particular, there is a significant spread in B − I (F435W−F814W) colors, which is not consistent with a coeval population with similar intrinsic colors and variable reddening.

A more likely explanation is that the colors of the clusters imply at least two populations: one with F435W−F814W <0.5 mag and a redder group with F435W−F814W ⩾1.0 mag. The extinction-corrected ages of the two systems are 1–5 Myr and 20–200 Myr for the instantaneous starburst model (Bruzual & Charlot 2003), respectively. We assume that both have colors consistent with reddening by 1–2 mag of visual extinction (A_V = 1–2 mag). However, many clusters have 0.5 mag ⩽ F435W−F814W < 1.0 mag, implying ages of 200–500 Myr old, if they are un-reddened. If these clusters sit behind 1–2 mag of visual extinction, then they could be as young as 5 Myr old. If we take into account an exponentially declining (SFR = 1 M_☉τ⁻¹e^(−t/τ) for τ = 1 Gyr, 0Gyr ⩽ t ⩽ 20 Gyr), or a continuous star formation history (Bruzual & Charlot 2003), then the ages of the clusters will be between approximately 1 Myr and 1 Gyr with A_V = 1–3 mag. With these models, there are a larger number of older clusters than in the instantaneous starburst model. In particular, the clusters that have colors of both of F435W−F814W ≳1.0 mag and F140LP−F814W ∼−1.0 mag are about 10 times older.

Zoom In Zoom Out Reset image size

In the Antennae Galaxies, the merger with the most extensively studied system of star clusters, Whitmore et al. (1999) find three populations of star clusters with ages of ∼10 Myr, ∼100 Myr, and ∼500 Myr. In the overlap region, the clusters appear to be even younger, possibly less than 5 Myr old based on Hα imaging. The youngest clusters in II Zw 096, therefore, appear to have similar ages as the youngest clusters in the Antennae. In more luminous galaxies (LIRGs and ULIRGs), Surace et al. (1998) and Scoville et al. (2000) find cluster ages between 5and300 Myr, consistent with the spread in ages we find for II Zw 096. In Arp 220, Wilson et al. (2006) similarly find two populations of clusters with ages of <10 Myr and 70–500 Myr.

The regions of diffuse emission in II Zw 096 are typically 0.5–1 mag redder in F435W−F814W than the star clusters. If we assume that the emission of the diffuse regions comes from a mixture of young stars (∼1 Myr) and old stars (∼1 Gyr), then the average F140LP−F435W color of the diffuse regions can be explained by saying that they contain a small fraction (∼5%–10%) of young stars. On the other hand, the F435W−F814W color of the diffuse emissions suggests that nearly 17% of the emission comes from young stars, although the diffuse emissions have a large spread in this color.

We show the location of the clusters and the diffuse emission on the ACS F140LP image in Figure 9. There is no apparent correlation of the cluster colors (ages) with position, except for the bluest clusters which are prominent along the spiral arm to the northwest. On the other hand, the diffuse emission regions with F435W−F814W <1.4 mag are concentrated in the east, while those with F435W−F814W >1.4 mag are in the west of II Zw 096.

The F435W−F160W or F814W−F160W colors of the clusters do not show an obvious gradient or pattern with position. However, the clusters to the east side have less blue F814W−F160W colors on average than the clusters to the west.

The NICMOS H-band data can be used to compute the masses of the clusters. In Figure 10, we show an F435W−F814W versus absolute H-band color–magnitude diagram for the clusters. Assuming the clusters have 0–2 mag of visual extinction, as in the color–color diagram (see Figure 8), the H-band data suggest that most of the clusters have masses of 10⁶–10⁸ M_☉.

Zoom In Zoom Out Reset image size

If we compare the masses of the star clusters in II Zw 096 with those in the Antennae Galaxies, the former appear to be ∼100 times larger (Zhang & Fall 1999). However, we note that the resolution of the HST data is about 30 pc for II Zw 096 as opposed to only 5 pc at the distance of the Antennae, and this might account for the apparent differences in the cluster masses between the two galaxies. For comparison, in (U)LIRGs, the cluster masses derived by Surace et al. (1998) and Scoville et al. (2000) lie between 10⁵and10⁷ M_☉, with the youngest clusters dominating the high-mass end. This is also true in II Zw 096, where the youngest clusters (those with F435W−F814W <0.5 mag and A_V = 1–2 mag) appear to have masses above 10⁷ M_☉.

In order to compare the distribution of the luminosity of the clusters, we derive the B-band luminosity function (LF) and fit this with a power law, dϕ/dL_B ∝ L^α_B. In the range of −14.7 mag ⩽ M_B ⩽ −10.7 mag with the detection limit of −9.9 mag, we obtain α = −1.1  ±   0.3. To avoid confusion due to limited spatial resolution, we chose for comparison merging infrared luminous galaxies at similar distances to II Zw 096. The LF of the clusters in II Zw 096 is found to have a flatter slope (α = −1.1 ± 0.3) than that of Mrk 231 (α = −1.80  ±   0.2) at z = 0.042 or Mrk 463 (α = −1.63 ± 0.11) at z = 0.050, which are two nearby ULIRGs studied by Surace et al. (1998). This suggests that there are more luminous clusters in II Zw 096 than Mrk 231 and Mrk 463. In the Antennae Galaxies at z = 0.006, the V-band LF of the clusters have much steeper slopes, α = −2.6 ± 0.2 and α = −1.7 ± 0.2 in the ranges of M_V ≲ −10.4 mag and −10.4 mag <M_V < −8.0 mag, respectively (Whitmore et al. 1999).

The IRS spectra of source C+D show the presence of PAH emission, and no evidence of [Ne v] or [O iv]. The upper limits on the [Ne v]/[Ne ii] and [O iv]/[Ne ii] emission line flux ratios are 0.04 and 0.11, respectively. These line ratios imply a starburst-dominated system. The EQW of the 6.2 μm PAH feature is 0.26 μm, which is only about 1/2 that seen in pure starburst nuclei (Brandl et al. 2006). This might indicate an excess of hot dust in source C+D compared to the nuclei of most starburst galaxies, or a deficiency of PAH emitters.

If we adopt a total far-infrared flux F_IR = 8.48 × 10⁻¹³ W m⁻² for source D, and a total PAH flux $F_{{\rm PAH}_{\rm tot}} = 1.44 \times 10^{-14}$ W m⁻² including all the PAH features (Table 4), we derive a ratio of $F_{{\rm PAH}_{\rm tot}} / F_{{\rm IR}} \, \sim 0.02$. This is at the low end for pure starburst galaxies, or about 1/2 the median value (Marshall et al. 2007).

The high [Ne iii]/[Ne ii] line flux ratio of ∼1.0 is at the high end of that seen in starburst galaxies (Thornley et al. 2000; Madden et al. 2006; Beirão et al. 2008). There are two possibilities that can explain this high ratio—either a low metallicity, or an unusually hard radiation field, possibly produced by a large number of high-mass stars. According to Rupke et al. (2008), the SE source (source A in this paper) has a metallicity of 0.4 Z_☉ ⩽ Z ⩽ 1.8 Z_☉, or a mean oxygen abundance 〈12 + log(O/H)〉 = 8.6. If we assume solar metallicity for the gas, the implied age is about 3 Myr, and an upper mass cutoff of 50–100 M_☉ is required to match the observed [Ne iii]/[Ne ii] ratio (Thornley et al. 2000).

The strong 3.3 μm PAH emission feature seen in the AKARI spectrum also supports a powerful starburst in source D (Moorwood 1986; Imanishi et al. 2008). In addition, the typical indicators of a buried active galactic nucleus (AGN), e.g., a rising (power-law) continuum (Armus et al. 2007; Veilleux et al. 2009) or 3.4 μm absorption by aliphatic hydrocarbon grains (Imanishi et al. 2006) are not detected.

The Chandra data for source C+D are also consistent with a buried starburst, although they are slightly weak in X-rays for their far-infrared luminosity. Starburst galaxies show a correlation between their infrared and X-ray luminosities (Ranalli et al. 2003). To convert from 2–7 to 2–10 keV, we have extrapolated the spectral model fit to the data from 7 to 10 keV as in Iwasawa et al. (2009). Together, source C+D lies below the starburst correlation, but within the 2σ envelope of the distribution. In some (U)LIRG nuclei of the GOALS sample, the low ratio of the hard X-ray emission to the far-infrared emission has been taken as evidence for highly obscured AGN (Iwasawa et al. 2009), although this argument hinges on the detection of an Fe K emission line. Given the low number of counts in the spectrum of source C+D, an accurate description of the X-ray properties will require significantly deeper integrations. The other X-ray luminous source, source A, lies on the X-ray selected starburst correlation line of Ranalli et al. (2003), suggesting this source is powered by starburst.

From the infrared luminosity of L_IR = 6.87 × 10¹¹ L_☉, the star formation rate (SFR) of source D is estimated to be 120 M_☉ yr^-1 (Kennicutt 1998). This is approximately a factor of 5 above that estimated by Goldader et al. (1997) based on the near-infrared data alone. Assuming this far-infrared emission originates from a region having a size comparable to that of the projected FWHM of the MIPS 24 μm beam (6''), we estimate a starburst luminosity density of ∼4.0 × 10¹⁰ L_☉ kpc^-2, or an SFR density of ∼6.9 M_☉ yr^-1 kpc⁻². If instead, we take the size of source D estimated from the NICMOS image (∼220 pc radius), we obtain values which are about 2 orders of magnitude larger, namely a luminosity density of ∼4.5 × 10¹² L_☉ kpc^-2 and an SFR density of ∼780 M_☉ yr^-1 kpc⁻². Although large, this value is still below the peak values estimated for clusters in starburst galaxies (Meurer et al. 1997) which are an order of magnitude higher. Similarly, our derived value for source D is well below most of the ULIRG starburst nuclei observed by Soifer et al. (2000), except for IRAS 17208−0014.

The Brα line detected with AKARI provides a direct measure of the SFR. We assume Case B for hydrogen recombination, the effective temperature of ∼10⁴ K, and the density of ∼100 cm^-3. We then estimate an SFR of approximately 45 M_☉ yr^-1 uncorrected for extinction (Kennicutt 1998). Although we do not have an estimate of the extinction from the AKARI spectra, if we apply the line-of-sight value estimated from the depth of the silicate feature in the IRS data (A_V = 19 mag; Smith et al. 2007, Equation (4)), the implied SFR would be about 70 M_☉ yr^-1.

Existing radio observations of II Zw 096 also suggest that source D is a site of intense star formation. At 1.425 GHz, Condon et al. (1996) measure an integrated flux density of 26.6 mJy using the Very Large Array (beam size of 6''). The J2000 position of the radio peak is R.A. = 20$\mathrm{^h}$57$\mathrm{^m}$243, decl. = +17°07'38 8'', well within the positional uncertainties of the position of source D in the MIPS data. With the new Spitzer imaging data, we can, for the first time, estimate the infrared-to-radio ratio q (Helou et al. 1985) for source D as q = 2.93. This value is within the ∼1σ (0.26 dex) scatter of the average q (〈q〉 = 2.64) among star-forming galaxies (Bell 2003), and is thus consistent with our mid-infrared spectra and X-ray data implying starburst activity in source D.

The H-band magnitude (−19.5 mag) and assumed age, extinction, and metallicity are used to derive the mass of source D by the model of Bruzual & Charlot (2003). Goldader et al. (1997) estimate the age of the II Zw 096 starburst to be roughly 5–7 Myr from the CO index and Brγ EQW, using the models of Leitherer & Heckman (1995). Assuming an extinction of 3.3 mag in the H band (A_V  ⩾   19 mag—see Section 3.4) and solar metallicity, we estimate the mass of source D to be approximately 1–4 × 10⁹ M_☉.

Source D is a powerful starburst not associated with the primary nuclei of the II Zw 096 system (sources A and B). It could be a starburst in the disturbed disk of source A, or even the nucleus of a third galaxy—one that is nearly completely hidden behind A and B. While large-scale mapping and modeling of the gas and stellar dynamics of II Zw 096 will be needed to fully understand the true nature of the system, we favor the former explanation for two reasons. First, simulations have shown (e.g., Barnes 2004) that shock-induced starbursts can occur both in and outside the nuclei of merging galaxies. Second, we have a number of examples of powerful, extranuclear starbursts in the local universe. Below we compare the properties of source D to the extranuclear starbursts in two well-studied merging galaxies NGC 4038/9 (the Antennae Galaxies) and Arp 299.

In the Antennae Galaxies, nearly 1/2 of the total 15 μm emission (about 5 × 10¹⁰ L_☉) comes from the overlap region (5 × 3 kpc) between the merging spiral galaxies, with about 15% coming from a single starburst knot (Mirabel et al. 1998). Recent mid-infrared spectral mapping observations with the IRS on Spitzer have resolved this off-nuclear starburst region into a number of clumps in the mid-infrared (peaks 1, 2, 3, and 5 of Brandl et al. 2009). These clumps (or clusters) have infrared luminosities of 3.86 × 10⁹ L_☉ < L_IR < 1.14 × 10¹⁰ L_☉, corresponding to ∼5%–16% of infrared radiation coming from the system. Brandl et al. (2009) estimate that the SFRs in these clusters range between 0.66and1.97 M_☉ yr^-1.

To make a direct comparison with II Zw 096, we pick the most luminous infrared peak in the Antennae (peak 1 in Brandl et al. 2009) which has L_IR = 1.14 × 10¹⁰ L_☉ and assume the radius of this cluster to be R ∼ 48 pc (the size estimated by Whitmore & Schweizer 1995). We then estimate a luminosity density of 1.6 × 10¹² L_☉ kpc^-2. This value is about three times smaller than the upper limit of the luminosity density of source D (assuming a size of 220 pc radius—the size of D in the NICMOS image). While the luminosity density of peak 1 is extremely large, it only contributes ∼15% of the total infrared emission of the entire system, in contrast to source D in II Zw 096 which dominates the total infrared output of the system.

In the Antennae Galaxies, the silicate optical depths at 9.7 μm of the star clusters measured by Brandl et al. (2009) are τ_9.7 = 0.13–1.03 (A_V = 2.4–19 mag). Although one of the clusters in the Antennae, peak 5, has an extinction comparable to that measured in source D in II Zw 096 (τ_9.7 ∼ 1.03), its L_IR is the smallest in all of the clusters in the overlap region of the Antennae. This is unlike source D which has both high extinction and high L_IR. Peak 1 itself has a τ_9.7 = 0.19, much lower than our estimated value for source D, even though the estimated age of peak 1 (2–4 Myr) is slightly less than that derived for source D (5–7 Myr; Goldader et al. 1997).

Arp 299 (IC 694 + NGC 3690) has substantial extranuclear infrared emission like the Antennae but an infrared luminosity of ∼6 × 10¹¹ L_☉, making this system an LIRG. In particular, regions C and C' in the overlap region between IC 694 and NGC 3690 have all the signatures of a merger-induced starburst (e.g., Alonso-Herrero et al. 2000, 2009; Charmandaris et al. 2002). Region C+C' has an infrared luminosity (4.4 × 10¹⁰ L_☉; Charmandaris et al. 2002), approximately 6.4% of the L_IR of source D in II Zw 096. The fraction of the infrared emission in regions C and C' to that in the Arp 299 system is only ∼10% (Charmandaris et al. 2002), much less than the fractional infrared luminosity in source D (80%). The infrared luminosity densities of regions C and C' are estimated to be ∼1.0 × 10¹¹ L_☉ kpc^-2 and ∼7.5 × 10¹⁰ L_☉ kpc^-2, respectively, if we assume that their sizes are R ∼ 300 pc and R ∼ 250 pc, corresponding to the size of Paα emission estimated by Alonso-Herrero et al. (2009), and that the infrared luminosities of regions C and C' have a ratio of 2:1. These numbers are a factor of 2–3 larger than what we estimate for source D when we assume that the size of the source D is equivalent to the beam size of the MIPS 24 μm. However, if the size of source D is comparable to the NICMOS resolution, then the luminosity density could be significantly larger than that seen in regions C and C' in Arp 299.

Assuming a metallicity of 2Z_☉, Alonso-Herrero et al. (2009) use the IRS spectra of regions C and C' and the models of Snijders et al. (2007) to estimate ages of 4–7 Myr. If we apply the same model and metallicity to source D in II Zw 096, the measured [Ne iii]/[Ne ii] =1.0 and [S iv]/[S iii] =0.33 line ratios suggest that source D also has an age of 4–7 Myr.