Exploring Infrared Properties of Giant Low Surface Brightness Galaxies

A basic concern about faint extragalactic sources with highly diffuse disk structure is confusion by foreground Galactic “cirrus” emission. This is especially critical in the case of far-IR cool sources (defined below) such as LSB galaxies. The far-IR ratio of typical local cirrus is $S_{60\mu m}/S_{100\mu m}\leq 0.2$ (Gautier 1986). UGC 6614 and UGC 6879 are well above this limit (Fig. 8c). From its Galactic latitude $b\sim+22^{o}$ and the observed $S_{70\mu m}/S_{160\mu m}$ ratio, it is reasonably safe to assume that UGC 9024 has not been affected by cirrus emission. In the absence of far-IR information it is uncertain for Malin 1. However, its high galactic latitude $b\sim+14^{\circ}$ can be used to argue against any cirrus contamination.

We should also stress the fact that whether there is cirrus in the foreground is less significant as compared to whether the cirrus in the foreground varies on scales of the IRAC and MIPS field of view. If its spatial variation is negligible across the IRAC and MIPS mosaic, it gets subtracted out as “sky” in the data reduction process.

The IR emission beyond $\sim$25 $\mu$m is dominated by the inter-stellar dust under various heating conditions. On the other hand, mid-IR ($\sim$5-25 $\mu$m) emission marks the transition from stellar photospheres to inter-stellar dust dominating the emission. Whereas the morphology of a galaxy at 3.6 and 4.5 $\mu$m represents the stellar disk, at $\sim$5 $\mu$m and beyond it shows the structure of the ISM. The IRAC 3.6 and 4.5 $\mu$m bands are sensitive to the underlying stellar populations typically consisting of red giants and old stars. In some galaxies 3.6 $\mu$m band is also known to contain emission from a hot dust component (Bernard et al. 1994; Hunt et al. 2002; Lu et al. 2003; Helou et al. 2004). The hot dust is also visible in the other IRAC bands. While 3.6 $\mu$m is only sensitive to hot dust near the sublimation temperatures ($\sim$1000 K), the longer wavelength IRAC bands can detect dust at lower temperatures down to several hundred Kelvin.

The IRAC 5.8 and 8 $\mu$m bands are primarily sensitive to the PAH emission at 6.2, 7.7, and 8.6 $\mu$m (Puget & Leger 1989; Helou et al. 2000; Lu et al. 2003). The PAH is the hot component of the inter-stellar dust with effective temperature $\rm T_{d}>100$ K stochastically excited to high energy levels by stellar photons. Stellar photospheric emission also contributes to these two IRAC wavebands. The fraction of stellar emission at 5.8 and 8 $\mu$m, respectively, are $\sim$40% and $\sim$20% (Helou et al. 2004; Pahre et al. 2004).

The emission detected by the MIPS bands are coming from dust grains with different size distributions. Very small grains ($\sim$1-10 nm) emit in the mid-IR region ($\gtrsim$15 $\mu$m), intermediate between thermal equilibrium and single-photon heating. Large, classical grains ($\sim$100-200 nm) are in thermal equilibrium with the radiation field and responsible for far-IR emission (Desert et al. 1990). To begin with we should bear in mind that the demarcation lines among various heating environments are ad-hoc and we assume the following effective color temperature ranges for large grains in thermal equilibrium: warm (40 K $\lesssim\rm T_{d}\lesssim$ 100 K), cool (20 K $\lesssim\rm T_{d}\lesssim$ 30 K), cold (10 K $\lesssim\rm T_{d}\lesssim$ 20 K), and very cold (10 K $\lesssim\rm T_{d}$).

The Solan Digital Sky Survey (SDSS) images of target galaxies are shown in Fig. 1. The Spitzer images of the LSB galaxies are shown in Fig. 2 (Malin 1), Fig. 3 (UGC 6614), and Fig. 4 (UGC 9024). UGC 6879 is shown in Fig. 5. Galaxy images are shown using Gaussian equalization (Ishida 2004). The images are oriented such that north is up and east is to the left. Malin 1 is too faint to be detected by the MIPS 70 and 160 $\mu$m channels. The other three galaxies were detected by all IRAC and MIPS bands. We do not show 5.8 $\mu$m images in these figures since the morphological appearances of each of these galaxies at 5.8 $\mu$m closely follow that at 8 $\mu$m. For all galaxies, the IRAC 4.5 and 8 $\mu$m images are shown without subtracting stellar photospheric emission. The contours represent surface brightness with intervals of $\sqrt{10}$ where the lowest level is 4$\sigma$ above the background. The lowest level of contours at different bands are 0.04 (3.6 $\mu$m), 0.05 (4.5 $\mu$m), 0.30 (8 $\mu$m), 0.08 (24 $\mu$m), 1.6 (70 $\mu$m), and 2.1 (160 $\mu$m) expressed in MJy/Sr.

The giant optical disks in Malin 1 and UGC 9024 appear as point sources in the IRAC images. While we detect the stellar bulges of these galaxies, their optically diffuse disks remain undetected long-ward of the IRAC bands. This suggests that the low surface brightness structures at larger radii might be photometrically distinct components rather than smoothed extensions of the normal inner disks (see Barth 2007 for a discussion on Malin 1 based on Hubble data). That the disks appear in the $B$-band but are undetected at 3.6 $\mu$m suggest that these disks have a small population of young stars rather than a large population of old stars. The bulge spectrum of Malin 1 is consistent with a predominantly old stellar population (Impey & Bothun 1989). For both of these galaxies, the mid-IR emissions at 8 and 24 $\mu$m are concentrated in the central few kpc, within a region of 12″ (20 kpc) radius for Malin 1 and 24″ (5 kpc) radius for UGC 9024.

Undetected far-IR emission from the disk of Malin 1 implies that it contains either very cold ($\rm T_{d}<10$ K) cirrus-like dust emitting in the submillimeter and millimeter wavelengths or it lacks cold dust altogether and contains only neutral gas. For UGC 9024, 70 $\mu$m emission comes from the central region but 160 $\mu$m emission is very hard to measure because of large scale diffuse emission in the field. This result in an upper limit of $S_{160\ \mu m}<268$ mJy.

The optical morphology of UGC 6614 shows a massive bulge and spiral structure. A thin ring $\sim$40″ from the core is prominent in H$\alpha$ (McGaugh et al. 1995). The 3.6 and 4.5 $\mu$m emission is spread over the entire disk of this galaxy. The 3.6 $\mu$m image shows a discernible spiral arm pattern closely resembling the optical morphology. At 4.5 $\mu$m this feature disappears and the disk shrinks in radius showing only its inner region. This galaxy appears markedly different at 5.8 and 8 $\mu$m compared to the other IRAC bands. The 8 $\mu$m morphology suggests that the PAH emission is coming from two distinct regions: the central bulge and an outer ring surrounding the bulge. The 8 $\mu$m morphology closely traces the H$\alpha$ image. The MIPS 24 $\mu$m morphology is similar to the 8 $\mu$m PAH emission although the outer ring appears a bit more disjointed at this band. The dust emission at 24 $\mu$m is coming mostly ($\sim$70%) from the central disk. The lower resolution image at 70 $\mu$m indicates that only $\sim$25% of the dust emission is coming from the central part of the galaxy, with the remaining $\sim$75% emission co-spatial with the ring of radius $\sim$40″. Surprisingly a dumbbell shaped region is the dominant source ($\sim$90%) of 160 $\mu$m emission; these two peaks are located on the NE and NW sides of the ring. The far-IR images of this galaxy show a small, localized region within the ring, SW from the center. Whether or not this region coincide with the spatial location showing the CO emission in this galaxy disk (Das et al. 2006) is not entirely clear. We will investigate this in the forthcoming paper.

UGC 6879 is an edge on spiral with a red central part and a blue (optical) disk. This radial color gradient is perhaps related to greater concentrations of dust in the nucleus than in the overall disk. Being a transitional disk, with a central surface brightness in between HSB and LSB spirals, it is not unexpected to find that UGC 6879 is a strong IR emitting source compared to the LSB galaxies. Both PAH emission and warm dust (24 $\mu$m ) emission show spatial variation along the disk. These emissions peak at the central region and diminish toward the edge.

Panels 8a and 8b show, respectively, mid-IR colors and the well known PAH-metallicity connection. In these diagrams we highlight two extreme classes (in terms of their IR SEDs) of HSB galaxies: the dwarf systems represented by the decimal numerals and the massive elliptical galaxies represented by the roman numerals. All of these galaxies are obtained from the SINGS sample.

Panel 8a compares PAH emission in various classes of galaxies. Metal-rich HSB galaxies preferentially show high ratio in $S_{8\mu m}/S_{24\mu m}$ and low ratios of non-stellar 4.5-to-8 $\mu$m dust emission. Metal poor (e.g. dwarfs and irregulars) and extremely metal poor HSB galaxies such as blue compact dwarfs (BCDs), show the opposite trend and fall within the PAH-deficient box (Engelbracht et al. 2005; hereafter E05). Note that E05 only looked at star-forming galaxies. “Red and dead” ellipticals should have little or no PAH emission, yet are nowhere near the PAH-deficient region. So this region will not necessarily contain all PAH deficient galaxies. Also note that the result of E05 could also arise from selection bias since recent studies have shown that dwarf galaxies are not necessarily PAH defficient systems (Rosenberg et al. 2006).

Panel 8b, on the other hand, illustrates the PAH-$Z$ connection in HSB galaxies. E05 showed that galaxies with low PAH emission have relatively unpolluted ISMs. They noted a sharp boundary between galaxy metallicity with and without PAH emission, although the trend may have been affected by selection bias. We show this trend in panel 8b, where the PAH deficient galaxies reside inside the dashed region and galaxies with higher metallicites avoid two regions in the diagram which are shown by the dotted lines. We are interested to see where the LSB galaxies fit in these diagrams.

Panel 8a shows that the LSB galaxies are different than the PAH-deficient dwarf galaxies in terms of their mid-IR colors. In the color space, LSB galaxies occupy a region similar to HSB galaxies and reside significantly farther ($>3\sigma$, along the horizontal axis) from PAH-deficient galaxies. While UGC 6614 stays right in the middle of the locus, both Malin 1 and UGC 9024 falls in the edge because of the shapes of their SEDs at 8 and 24 $\mu$m. The mid-IR colors of these galaxies closely resemble those of elliptical galaxies which is surprising given that their apparent different star formation histories.

The LSB galaxies are metal poor with $Z\leq 1/3Z_{\odot}$ (McGaugh 1994; de Bloke & van der Hulst 1998b; de Naray et a. 2004). McGaugh (1994) provided an estimate of oxygen abundance for UGC 6614 but it is highly uncertain. Following the general trend shown by the LSB galaxies we assign one-third solar value to Malin 1 and a solar value to UGC 6879. The full range and the median values of published oxygen abundance are shown for UGC 6614 and UGC 9024.Having very limited information it is, therefore, extremely difficult to explore the PAH-metallicity connection for these galaxies. We are interested in the question whether LSB galaxies being low $Z$ systems, will appear close to the PAH-deficient box or will they fall in the region shunned both by HSB dwarfs and HSB spirals with extended disks (dotted regions; Fig 8b)? While it is tempting to give more weight to the latter case, only three data points with large errors are insufficient to derive any trend. A larger sample of LSB galaxies with 8 $\mu$m detections is needed to shed more light on this topic.

Panel 8c shows the connection between mid-IR and far-IR colors which basically describe the nature of dust emission at these wavelengths. In this panel, the far-IR color of Malin 1 is shown with respect to a far-IR flat SED, i.e. same flux density at 70 and 160 $\mu$m. For the other two galaxies the diagram shows rather low $S_{70\mu m}/S_{160\mu m}$ and high $S_{8\mu m}/S_{24\mu m}$ ratios. While low far-IR color suggest that they are IR cool sources, the mid-IR color may be linked with the destruction of very small grains (but not the PAH molecules) and thus can be used as a parameter which can signal evolutionary stages of an ISM. A higher value of $S_{8\mu m}/S_{24\mu m}$ can mean that $S_{8\mu m}$ is high (large amount of PAH) or that $S_{24\mu m}$ is low (little or no warm dust). Ellipticals are in the latter category (see Pahre et al. 2004) whereas LSB systems are in the first category. It should be noted that the 24 $\mu$m emission is very closely associated with H II regions (Helou et al. 2004). Therefore, the lack of emission at this wavelength is more likely a deficiency in H II region which is quite consistent with the H$\alpha$ images of LSB galaxies (McGaugh et al. 1995; de Naray et al. 2004).

The vertical arrow in panel 8d represents a probable range of IRAS far-IR color for UGC 9024 assuming that Spitzer far-IR ratio is the same as the IRAS far-IR ratio. The sequence of IR colors can be associated with a progression toward greater dust heating intensity and thus with a sequence of star formation activity. The cool end of the color sequence corresponds to cool diffuse H I medium and quiescent molecular clouds, whereas the warm end corresponds to the colors of H II regions, starbursts, and galaxies with higher $\rm L_{TIR}/L_{B}$ ratios. Although the IR nature of these three LSB galaxies are explicit in this diagram, the interesting feature is that they are not the extreme cases in terms of IR SEDs as shown by some of the SINGS galaxies.

Two primary sources have been proposed to explain the heating of the dust which produces IR luminosity in spiral galaxies - massive young (OB) stars and associated H II regions (Helou et al. 1985; Devereux & Young 1990, 1993), and non-ionizing A and later stars (Lonsdale & Helou 1986; Walterbos & Schwering 1987; Bothun et al. 1989). Some authors, however, suggest contribution from both sources (Smith 1982; Habing et al. 1984; Rice et al. 1990; Sauvage & Thuan 1992; Smith et al. 1994; Devereux et al. 1994). The dominance of the heating source is, therefore, governed by the availability of discrete and dense star forming regions in the ISM. While observational evidence suggests that the global IR emission from luminous IR spiral galaxies provides a measure of high mass SFR (Kennicutt 1998), the diffuse IR emission in quiescent galaxies is caused mainly by the thermal radiation of the interstellar dust heated by the interstellar radiation field (ISRF) (Jura 1982; Mezner et al. 1982; Cox et al. 1986; Jura et al. 1987).

The dust color temperature ($\rm T_{d}$) deduced for galaxies from the IRAS 60 and 100 $\mu$m flux densities are typically 25-40 K assuming emissivity index $\beta=2$ (Rahman et al. 2006). This range is similar to the temperature of dust in Galactic star-forming regions (Wynn-Williams & Becklin 1974; Scoville & Good 1989), and considerably greater than the 15-20K temperatures expected for dust heated by the ambient ISRF (Jura 1982; Mezner et al. 1982; Cox et al. 1986; Jura et al. 1987). The color temperature derived from the MIPS 70 and 160 $\mu$m of LSB galaxies ranges $\sim$17-21 K.

From a statistical study of a large sample of optically selected galaxies, Bothun et al. (1989) demonstrated that in the absence of UV radiation, far-IR color ratio of $S_{60\mu m}/S_{100\mu m}\leq 0.3$ can result from dust which is heated by old stars. Galaxies with $S_{60\mu m}/S_{100\mu m}\approx 0.3-0.5$ requires a steadily increasing proportion of UV heated dust, while galaxies with $S_{60\mu m}/S_{100\mu m}\geq 0.5$ are entirely dominated by the UV heated dust. From panel 8d we find that the LSB galaxies have cool effective dust temperatures and therefore lack intense heating from massive stars.

It should be mentioned here that in the IR color analysis we did not subtract the stellar contributions from the 8 and 24 $\mu$m measurements to exactly reproduce the E05 diagram. Assuming conservatively that the 3.6 $\mu$m emission is coming from the stellar photosphere, the stellar contributions obtained from the empirical stellar SED of Pahre et al. (2004) derived for early type galaxies are $\sim$59%, $\sim$38%, and $\sim$19% at 4.5, 5.8, and 8 $\mu$m, respectively. The stellar contribution at 24 $\mu$m emission is $\sim$5% or less. Applying corrections to the corresponding flux densities do not change our results since these corrections systematically shift the parameters in the color space. Note that we use the same set of numbers for each galaxy ignoring the variation in amount of stellar contributions from galaxy to galaxy. However, error due to this is negligible. To find stellar contributions, preference is given to the empirical SED than any stellar synthesis model because of unknown SFHs and poor metallicity constraints for LSB galaxies.

In summary, from the four panels of Fig. 8 we conclude that: (a) Malin 1, UGC 6614, and the intermediate HSB/LSB object UGC 6879 have mid-IR colors similar to quiescent HSB disk galaxies; UGC 9024 falls in the region of HSB elliptical galaxies in this color plane. (b) There is insufficient data to conclude whether LSB galaxies have PAH emission properties significantly different from HSB galaxies with comparably low metallicities. Observations of many more LSB galaxies are needed to settle this. (c) Available far-IR detections and upper limits indicate that LSB galaxies are far-IR cool sources. The dust temperatures derived from the MIPS 70 and 160 $\mu$m of LSB galaxies ranges $\rm T_{d}\sim$17-21 K, similar to many quiescent HSB spiral and elliptical galaxies.

The LSB galaxies are rich in neutral hydrogen (H I). Molecular hydrogen ($\rm H_{2}$) gas inferred from CO emission has been detected in only a handful of such galaxies (Matthews & Gao 2001; O’Neil & Schinnerer 2004; Matthews et al. 2005; Das et al. 2006). The low rate of CO detection has been attributed to an ISM with low dust content and a low surface density of neutral gas. Dust opacity is crucial for the formation and survival of molecules since it provides them necessary shielding from the ISRF. A larger column density is needed to self-shield the H₂ molecule. A low density and less dusty environment exposes H₂ to UV photons which can easily dissociate these molecules. The low star formation and far-IR cool nature of LSB galaxies implies lower energy density of the radiation field and consequently lower dissociation of CO and H₂ (de Blok & van der Hulst 1998b).

The deficiency of molecular gas detections in LSB galaxies may point to a dynamical condition such as the absence of large-scale instability in the disk preventing formation of giant molecular clouds (Mihos et al. 1997). Local instabilities may lead to cloud condensation resulting in localized star formation which may escape detection by current observations (de Blok & van der Hulst 1998b). Local instability in the disk invokes energetic phenomena such as SN explosions and the frequency of such occurrence in LSB galaxies is low (Hoeppe et al. 1994; McGaugh 1994). The enhanced cooling by molecules is crucial in the onset of instability of molecular clouds. Therefore, the effect of less efficient cooling of the ISM can also prevent local instabilities. Long cooling time leads to higher cloud temperatures and thus makes it difficult for a cold molecular phase to exist (Gerritson & de Blok 1999).

If the low density ISM truly lacks H₂ and CO molecules, what other types of molecule can exist in this environment? Are they very cold or very warm molecules of known types? To which dust components do they belong? Along with millimeter wavelength observations, can we use currently available IR data to shed light into these questions? One of the three LSB galaxies, UGC 6614, has been reported to have CO(1-0) emission at a certain localized region in the disk (Das et al. 2006), while the other two galaxies have not been observed in this wavelength. On the other hand, the Spitzer observation of UGC 6614 shows the presence of enhanced PAH emission on the bulge and almost entirely along the outer disk. This 8 $\mu$m emission, however, is concentrated in the central regions of the other two galaxies.

A possible source of the origin of PAH molecules is the dense, high temperature, carbon-rich ([C]/[O] $>1$) environment of circumstellar envelopes surrounding mass-losing AGB carbon stars. PAH formation by stars with more normal, oxygen-rich, photospheric abundances ([C]/[O] $<1$) will be negligible because nearly all the available carbon is bound up in the CO molecules (Latter 1991). Therefore, if the stellar population responsible for the enrichment of the ISM are dominantly old carbon-rich stars, the warm PAH molecules will be ubiquitous resulting in lower abundance of CO molecules. While a detailed investigation is beyond the scope of this study, we believe the observed PAH emission and lack of CO emission holds the potential clue to probe not only the ISM but also to better understand the SFHs in LSB galaxies.

The optical spectra of large LSB disks show an unexpected high occurrence of low-level active galactic nuclei (AGN) type activity (Spraybarry et al. 1995; Schombert 1998). UGC 6614, a optical giant LSB galaxy, is suspected to harbor a weak AGN from the optical spectrum (Schombert 1998) and from excess emissions at millimeter (Das et al. 2006) and radio wavelengths (Condon et al. 2000). Integrated mid-IR photometry provides a robust technique to identify AGN in HSB galaxies where AGN tend to be redder than normal star forming galaxies in the mid-IR (Lacy et al. 2004; Stern et al. 2005).

To determine whether one can use a similar technique to detect AGN signatures in a LSB bulge, we analyze the IRAC colors ([3.6]-[4.5] vs. [5.8]-[8.0]) of UGC 6614. The colors are measured for two regions of radius $\sim$1 and $\sim$5.5 kpc, respectively, encircling the galaxy center (Fig. 9; left panel). We find that in the color space, along the vertical [3.6]-[4.5] axis, the galaxy resides well outside the Stern et al “AGN box”. Although the lower end of the AGN box is within the error bar of this galaxy, its overall mid-IR color put it in a region occupied mostly by star forming galaxies suggesting that broad band colors may not be an efficient tracer for weak AGNs.

The contribution from an AGN to the measured IRAC fluxes can be estimated by combining image subtraction with assumptions about the nature of the SED coming from starlight and AGN emission. We use Pahre et al. (2004) for stellar flux ratios, and a $\nu^{-1}$ power law SED for AGN (Clavel et al. 2000). Following the procedure of Howell et al. (2007), a 4.5 $\mu$m image of the non-stellar emission was constructed (Fig. 9; right panel). Unlike the procedure of E05 which simply measures $S_{4.5}-\alpha S_{3.6}$, this procedure includes an additional factor to account for the contribution of non-stellar emission to the 3.6 $\mu$m image. The non-stellar flux density measured this way indicates that, within a 12″ aperture, the AGN contributes $\sim$12% of the light at 4.5 $\mu$m and $\sim$6% at 3.6 $\mu$m. At 8 $\mu$m, starlight and the AGN each contribute $\sim$35% of the light, with PAHs contributing the remaining. Note that the selected aperture includes only the bulge of the galaxy, excluding the spiral arm/ring structure.

UGC 6614 illustrates that although [3.6]-[4.5] color can identify strong AGN, weaker AGN will not be clearly separated from pure stellar sources. The procedure of E05, measuring $S_{4.5\mu m}-\alpha S_{3.6\mu m}$, will identify regions of non-stellar emission but will not provide quantitative picture of stellar emission. Given reasonable assumptions the procedure of Howell et al. allows a quantitative decomposition of the stellar and non-stellar flux densities.