A HIGH SPATIAL RESOLUTION MID-INFRARED SPECTROSCOPIC STUDY OF THE NUCLEI AND STAR-FORMING REGIONS IN LUMINOUS INFRARED GALAXIES

We obtained N-band, low-resolution (from R ∼ 80 at 8 μm to R ∼ 150 at 13 μm), long-slit spectroscopy of the four LIRGs mentioned above with T-ReCS on the 8.1 m Gemini South telescope. The observations were carried out during 2005 and 2006 under programs GS-2005B-Q-10, GS-2006A-Q-7, and GS-2006A-DD-15 (PI: Packham). T-ReCS has a 320  ×   240 pixel detector with a plate scale of 009, which provides a field of view (FOV) of ∼29'' ×  215. We used a slit width of 072 for NGC 5135, IC 4518W, and NGC 7130, and of 036 for NGC 3256. The slits were placed to cover the nucleus of the galaxies as well as some regions of interest (see Figure 1, orange crosses and dots, respectively).

Zoom In Zoom Out Reset image size

The observations were obtained in a standard chop–nod strategy to remove the time-variable sky background, telescope thermal emission, and the 1/f detector noise (see also Packham et al. 2005). The chop throw was 15'' and perpendicular to the slit. The observations were scheduled to be done in a single night but they were divided in various data sets to avoid observing problems or a sudden change in weather conditions.

We observed two Cohen standard stars (Cohen et al. 1999) for each galaxy to obtain the wavelength and absolute-flux calibration of the spectra. The standard star observations were taken with the same instrument configuration and immediately before and after the target to minimize the uncertainties in the photometric calibration. The standard stars were chosen to have similar air masses as the galaxies at the time of the observations. In order to place the slit along the regions of interest, the instrument was rotated accordingly. The standard stars were observed with the same configuration.

Details about the integration times and the final useful data available for each galaxy after discarding chop/nod pairs affected by noise are given in Table 2.

Table 2. Log of the T-ReCS Spectroscopic Observations

Galaxy	Slit Width	tint	Date	Seeing
Name	('')	(s)		('')
(1)	(2)	(3)	(4)	(5)
NGC 3256	036	1800	2006/3/7	035
IC 4518W	072	1900	2006/4/17	036
NGC 5135	072	633	2006/3/6	030
	072	1267	2006/3/10	029
NGC 7130	072	490	2005/9/18	037
	072	760*	2006/6/4	030
	072	633	2006/8/29	036
	072	1267*	2006/9/16	033
	072	1267	2006/9/25	040

Notes. (1) Galaxy; (2) slit width; (3) useful on-source integration time; in the observations marked with an asterisk, only the spectrum of the nucleus (i.e., not including the H ii region) was obtained; (4) date(s) of the observations (YYYY/MM/DD); (5) mean seeing (FWHM of the reference standard star(s) observed right before and/or after the target) throughout each night of observations.

Download table as:  ASCIITypeset image

2.2.1. Obtaining the Two-dimensional Spectra

The T-ReCS data are stored automatically in savesets, which contain information on the on-source (object + sky) and off-source (sky) frames for each chop/nod position. The savesets can be accessed directly to discard images affected by any type of instrumental noise pattern. A given number of savesets forms a data set. Once the bad savesets of a data set are discarded, the sky-subtracted images were stacked to obtain a single image of the two-dimensional spectrum of the target. Because some of the data sets were obtained in different nights and under different atmospheric conditions (see above), the following procedures were applied to each galaxy–standard pair of data sets individually as they contained the full information to perform their own calibration. Figure 2 shows the two-dimensional spectrum of each LIRG (the different savesets were averaged for obtaining high signal-to-noise ratio (S/N) images but these were not used for scientific purposes). Note that even in these partially reduced T-ReCS spectra (not corrected from the shape of the atmospheric transmission; see below) we can detect the 8.6 and 11.3 μm PAHs and the [S iv]10.51 μm, and [Ne ii]12.81 μm emission lines, together with the 9.7 μm silicate absorption feature (see Figure 2).

Zoom In Zoom Out Reset image size

After applying a flat field to the data sets, we made use of Gemini-based iraf⁷ tasks (nswavelength and nstransform) to calibrate in wavelength the two-dimensional spectra. We used the sky emission features in the reference (sky) stacked spectra to calibrate in wavelength both the galaxy and standard star two-dimensional spectra. For the next final steps of the data reduction, we used our own in-house developed idl routines.

In order to subtract the residual background from the two-dimensional-spectrum image, it was fitted to a two-dimensional plane with a fitting algorithm that iteratively rejects outliers 2.25σ above/below this plane to ensure that only "sky" pixels are used for the fit.

2.2.2. Extraction and Calibration of the One-dimensional Spectra

The four galaxies selected for this work are from the HST/NICMOS survey of a volume-limited sample of local LIRGs of Alonso-Herrero et al. (2006b). The observations were taken with the NIC2 camera (pixel size of 0075 and FOV of 193× 193) using two broadband filters (F110W and F160W) for obtaining NIR continuum imaging, and two narrowband filters (F190N and F187N) for observing the Paα line emission and its associated continuum. We refer the reader to Alonso-Herrero et al. (2006b) for details on the observations and data reduction. Using the fully reduced images, we constructed continuum-subtracted Paα emission, and color maps of the nuclear regions of the galaxies.

A few extra steps were needed to make a meaningful comparison between the NICMOS and the T-ReCS images. Most importantly, rescaling the pixel size and match the resolution of the images (see Díaz-Santos et al. 2008 for details). In addition, we simulated the positions of the T-ReCS slits (see Section 2.1) over the Hubble Space Telescope (HST) NICMOS Paα images and extracted the spatial profiles from them. These profiles were then compared with those obtained for the MIR spectral features (see the Appendix). The Paα and NIR continuum images where also used to estimate the ages and extinctions for each of the star-forming regions studied in this work. The ages were inferred from their Paα equivalent widths (EWs) using Starburst99 models (Leitherer et al. 1999) and are upper limits to the real ages of the regions. The extinctions were calculated using the NIR colors and are lower limits to the real values. For more details about the approach used, see Díaz-Santos et al. (2008).

For NGC 3256, NGC 5135, and NGC 7130, we compare our T-ReCS data to the Spitzer IRS spectra from Pereira-Santaella et al. (2009). The spectra were obtained with the short-wavelength, low-resolution module (SL), which ranges from ∼5.5 μm to ∼13.5 μm and has a spectral resolution of R∼60–130, similar to that of T-ReCS spectra. The extraction of the Spitzer IRS one-dimensional spectra for each LIRG was made using various fixed apertures, from 2× 2 pixel to 4× 4 pixel. The smaller aperture is approximately the size (FWHM) of a point source at the end of the wavelength range covered by the module, which is about 37× 37 for the SL module. The flux-calibration was performed assuming the sources were extended. The extracted spectra of the galaxies are shown in Figure 5.

Figure 3 shows an example of some individual spectra of NGC 3256 extracted at different positions along the slit (see Section 2.2.2). The main emission features seen are the 8.6 and 11.3 μm PAHs, and the [S iv]10.51 μm and [Ne ii]12.81 μm lines.

Zoom In Zoom Out Reset image size

To measure each feature, we first fitted each one-dimensional spectrum with a polynomial function using a χ²-minimization method (with a weight ∝ 1/σ², where σ is the total uncertainty at each wavelength) after masking out the most prominent features in the spectra. A few examples of these fits are shown in Figure 3. Next, we (1) measured the fluxes of the PAHs and emission lines, (2) calculated their EWs, and (3) obtained the continuum flux at two reference wavelengths: 8.2 and 12.2 μm. The fluxes of the features were calculated by fitting them with Gaussian functions with fixed widths and varying intensities and positions. Although a Gaussian profile does not represent perfectly the shape of the PAH bands, it is still a reasonable choice and a good approximation to their enclosed flux. We considered a feature as detected when its peak was 2.25σ above the continuum. This threshold was chosen as a good compromise between being able to detect low surface brightness emission in diffuse regions and not including unreliable measurements.

We note that the [Ne ii]12.81 μm line may be contaminated with emission from the 12.7 μm PAH feature, since the T-ReCS spectral resolution is not sufficient to separate both components. In addition, due to the distance to IC 4518W, NGC 5135, and NGC 7130, the [Ne ii]12.81 μm emission line of these LIRGs is redshifted almost outside of the T-ReCS N-band filter. This causes the red wing of the line to be significantly affected by the filter and atmospheric transmissions; the latter starts decreasing significantly at λ ≳ 13 μm. We therefore doubled the uncertainties in the fluxes measured of [Ne ii]12.81 μm line for these galaxies because the fits to this line were made using only the half of the line not affected by the filter and atmospheric transmissions. The uncertainties of the ratios and other quantities obtained using this line were calculated accordingly, including this additional uncertainty.

The fluxes and EWs of the different features measured in the nuclear and integrated T-ReCS spectra are given in Tables 3–6.

An important quantity related to the Si _9.7 μm feature is its depth or strength. This is calculated as

$$\begin{equation} S_{\rm Si\,9.7\,\mu m}=\ln \frac{F^{\rm obs}_{\lambda }}{F^{\rm cont}_{\lambda }}, \end{equation} \tag{ 1 }$$

where F^obs_λ is the observed flux density of the feature and F^cont_λ is the continuum, both evaluated at wavelength λ (usually, 9.7 μm). A negative value indicates absorption, while a positive value indicates emission. The S_{Si 9.7 μm} can be associated with an optical depth (and therefore with a visual extinction) via an extinction law and a dust obscuration model or geometry. For a dust screen configuration, the relation between the S_{Si 9.7 μm} (or equivalently in this case: apparent τ_Si9.7 μm =−S_{Si 9.7 μm}) and the apparent τ_V is unique and it is given by the shape of the absorption curve of the adopted extinction law.

The limited spectral range afforded by ground-based N-band observations in conjunction with the broad silicate feature, and the prominent PAH features in our galaxies make it difficult to measure the MIR continuum. To circumvent this problem, we made use of the SL Spitzer IRS spectra which has a larger spectral range. We stress that the Spitzer spectra are extracted with much larger apertures (see Section 3.2) than the T-ReCS spectra, and thus the main assumption here is that the shape of the continuum over the Spitzer (kpc) spatial scales is the same as that over the T-ReCS (hundred pc) scales. That is, for a given galaxy, we used the same Spitzer continuum for all the T-ReCS spectra extracted along all spatial steps. This may have some implications in cases were the Spitzer spatial resolution cannot separate AGN and SF sites.

We fitted the IRS spectrum to a simple linear function with anchors at 5.5 and 13.2 μm (see Figure 4, top). This method yielded similar results to the method of Sirocky et al. (2008), which uses a spline interpolation. We then measured the ratio between the measured flux at 12.2 μm in the IRS spectrum, and the fitted value at the same wavelength. We used this ratio to scale the fit to the T-ReCS spectra at 12.2 μm. We have to note that because of the wavelength selected to scale the spectra is still inside the absorption feature (though in the more outer part of it), the measured silicate strengths might be slightly underestimated. In any case, relative comparisons are not affected by this.

Zoom In Zoom Out Reset image size

We found that the maximum of the Si _9.7 μm absorption feature was not always centered at 9.7 μm (see Sirocky et al. 2008, and reference therein). It varies between 9.4 and 10 μm and sometimes is even located at shorter wavelengths down to ∼9 μm (see, e.g., the spectra of NGC 3256 in Figure 3). Because of these displacements, we computed the S_{Si 9.7 μm} of the T-ReCS spectra using Equation (1) but evaluating it at the maximum of the absorption, rather than exactly at λ_rest =  9.7 μm.

Since no IRS spectrum is available (to the date) for IC 4518W, we used a second approach to measure the S_{Si 9.7 μm}. In this case, we measured the continuum emission above the Si _9.7 μm feature using directly the T-ReCS spectra. Note, however, that the PAHs and in particular the 8.6 μm feature, in this galaxy are not very strong (see Figure 5). For each step of the spatial profiles, we fitted the spectra between 8.2 and 12.2 μm (at the edges of the silicate feature) with a linear function and interpolated the fitted continuum to the maximum of the absorption. We also applied this alternative method to the other LIRGs to compare these values with those obtained using the IRS spectra. We found that both methods yielded similar results, as can be seen for NGC 3256, NGC 5135, and NGC 7130 in Figure 6.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Tables 3 and 4 give the Si _9.7 μm strengths measured in the nuclear and integrated spectra of the galaxies, respectively. In Section 6, we explore in detail the spatial profile of the 9.7 μm silicate feature in each LIRG.

NGC 3256 has two nuclei separated by ∼5'' (∼1 kpc) along the north–south direction. The northern nucleus (see Figure 1, orange cross) has been clearly classified as a star-forming region (Lípari et al. 2000; Lira et al. 2002) and is resolved in the T-ReCS MIR image (FWHM of 050 ≃ 100 pc; Díaz-Santos et al. 2008). The southern nucleus is heavily obscured (A_V≳ 12–15 mag; Kotilainen et al. 1996; Lira et al. 2002; Alonso-Herrero et al. 2006a; Díaz-Santos et al. 2008), and it is classified as H ii-like. The T-ReCS (nuclear and integrated) and IRS spectra of the northern nucleus of NGC 3256 are all very similar (see Figure 5). All of them show prominent 8.6 and 11.3 μm PAH features together with a conspicuous [Ne ii]12.81 μm emission line. Therefore, despite the nucleus is resolved, the nuclear spectrum is representative of the whole region, even at kpc scales (see IRS spectrum).

The IRS spectrum of the southern H ii-like nucleus of NGC 3256 shows a larger absorption (S_{Si 9.7 μm} ≃−1.4) than that of the northern nucleus (S_{Si 9.7 μm} ≃−0.5; Alonso-Herrero et al. 2010; Pereira-Santaella et al. 2009). However, our T-ReCS spectrum of the southern nucleus shows that the IRS spectrum underestimates the true depth of the silicate feature. In fact, the T-ReCS spectrum is so absorbed that even the intense 8.6 and 11.3 μm PAHs (clearly seen also in the spectrum of Martín-Hernández et al. 2006) cannot be measured because they are totally extinguished. The [Ne ii]12.81 μm emission line is also likely to be somewhat affected by the absorption. All these features are, however, clearly detected in the IRS spectrum. Thus, the T-ReCS spectrum effectively separates the emission arising from the heavily absorbed southern nucleus of NGC 3256 from that of the surrounding regions. In turn, the IRS spectrum is a combination of both.

The central 5'' × 5'' region of this galaxy is known to host a Compton-thick Seyfert 2 nucleus as well as a number of bright H ii regions (Levenson et al. 2004; González Delgado et al. 2001; Bedregal et al. 2009). Our MIR imaging data showed that the AGN is only contributing about 25% of the MIR emission in this region (Alonso-Herrero et al. 2006a). The IRS spectrum (Figure 5) shows intense 8.6 and 11.3 μm PAH features, indicating the presence of SF, as well as [S iv]10.51 μm and [Ne ii]12.81 μm line emission, and a hint of the [Ar iii]8.99 μm emission line. Outside the ground-based N-band spectral range the IRS spectra display evidence for the presence of the AGN in the form of high-excitation emission lines and a strong dust continuum at 6 μm (see Alonso-Herrero et al. 2010; Pereira-Santaella et al. 2009). From the IRS spectrum alone it is not clear whether the spectral features are being emitted by the same source (see Figure 1). The T-ReCS nuclear spectrum is almost featureless, except for the [S iv]10.51 μm and the [Ne ii]12.81 μm emission lines. The comparison between the line fluxes from the IRS and the T-ReCS spectra shows that the AGN of NGC 5135 is responsible for a large fraction, at least 50% (this is a lower limit, since we are not correcting for aperture), of the [S iv]10.51 μm line emission measured from the IRS spectrum (see Table 7). The [Ne ii]12.81 μm emission on the other hand is coming mostly from extra-nuclear regions. It is also worth noting that the T-ReCS nuclear spectrum does not display PAH emission (although the 11.3 μm PAH may be marginally present) or [Ar iii]8.99 μm line emission. We can conclude from the T-ReCS spectroscopy that the SF in this galaxy is mostly circumnuclear (≳150 pc) rather than nuclear.

In contrast, both the integrated T-ReCS spectrum of NGC 5135 and the kpc-scale IRS spectrum show the same features, including the PAHs. The similarity between both spectra seems to indicate that the PAH emission does not vary strongly from region to region within the central kpc of NGC 5135. The T-ReCS nuclear spectrum of NGC 5135 has a shallower Si _9.7 μm feature than that of the integrated spectrum (see below).

NGC 7130 also hosts a Compton-thick Seyfert 2 nucleus, as well as SF within the central ∼150 pc (Levenson et al. 2005; González Delgado et al. 1998). The IRS spectrum shows both AGN and SF features (Pereira-Santaella et al. 2009). However, unlike NGC 5135, our T-ReCS imaging data of NGC 7130 show a resolved nucleus (FWHM ∼ 045) suggesting that in this region the SF and the AGN emissions are still mixed (Alonso-Herrero et al. 2006a). Indeed, Figure 5 shows that the T-ReCS nuclear spectrum of NGC 7130 displays the [S iv]10.51 μm and [Ne ii]12.81 μm emission lines as well as the 8.6 and 11.3 μm PAH features. The presence of PAH features in the nuclear spectrum clearly indicates that there is SF in the very nuclear region (≲100 pc) of NGC 7130, as well as in regions surrounding it (at distances of several hundreds of pc). Nevertheless, at least half of the [S iv]10.51 μm emission measured in the Spitzer spectrum is contained within the inner 036× 072 probed by the T-ReCS nuclear spectroscopy (see Table 7). In contrast, most of the [Ne ii]12.81 μm emission is steming from regions outside the central 036× 072 region of this galaxy.

At a given metallicity, the hardness of the radiation field is a strong function of the age of the ionizing populations. The radiation field decreases with age as stars evolve off the main sequence, but it also shows a temporary increase when the most massive stars enter the Wolf–Rayet phase (see, e.g., Snijders et al. 2007). In this section, we use the age of the ionizing stellar populations, as probed by the Paα EW, as an approximate proxy for the hardness of the radiation field.

Despite the relatively narrow range of ages (∼3.5–7 Myr) of our star-forming regions, Figure 9 clearly shows that the 11.3 μm PAH/Paα ratio depends on the age of the ionizing stellar populations in the sense that more evolved regions show higher 11.3 μm PAH/Paα ratios. This tendency is mainly (but not only) caused by the decreasing of the Paα emission during this period (as seen in Díaz-Santos et al. 2008). In addition, we can also see that the trend is common to all star-forming regions in our sample of LIRGs. This is in agreement with our results in Díaz-Santos et al. (2008) that showed that the 8 μm luminosity (8.6 μm PAH + continuum emission) of LIRG H ii regions is correlated with their age. This may suggest that PAHs may trace B stars (i.e., recent) rather than O stars (i.e., current SF; see, e.g., Peeters et al. 2004) and/or that at the earliest stages of the SF, PAHs may be destroyed by the hard radiation fields of the youngest massive stars. However, the later explanation is unlikely since in our sample of LIRGs we do not see evidence for radiation fields hard enough (Pereira-Santaella et al. 2009) to explain the low 11.3 μm PAH/Paα ratios observed in the youngest regions in terms of the destruction of the PAH carriers (at least at the spatial scales probed in this study).

Zoom In Zoom Out Reset image size

We conclude then, as we did for the 8 μm monochromatic (PAH + continuum) luminosity (Díaz-Santos et al. 2008), that the 11.3 μm PAH luminosity, unlike the [Ne ii]12.81 μm emission line, is not a good tracer of the current SF when measured on scales of a few hundreds of pc. Thus, in order to use the PAH emission as a measure of the global SFR in LIRGs (or in galaxies in general), the calibration must be done using integrated emission of galaxies, where all these local physical properties are averaged out (see also Wu et al. 2005; Alonso-Herrero et al. 2006b; Brandl et al. 2006).

In Figure 9, we showed the evolution of the 11.3 μm PAH/Paα ratio as a function of the hardness (age) of the starburst only. In this section, we study the variation of the 11.3 μm PAH/[Ne ii]12.81 μm ratio (equivalent to the 11.3 μm PAH/Paα ratio) as a function of the [Ne ii]12.81 μm LSD, i.e., as a function of the density of the radiation field, which takes into account not only the hardness (age) but also the intensity (mass density) of the starburst. Figure 10 shows that star-forming regions in LIRGs with high values of the [Ne ii]12.81 μm LSD tend to display lower 11.3 μm PAH/[Ne ii]12.81 μm ratios. This tendency is further supported by the observations of the H ii-like nuclei in the sample of Roche et al. (1991).¹⁰ In fact, the trend seen in Figure 10 is connected to the age effect we discussed in Section 8.1 since, as it is said above, the density of the radiation field depends on the hardness (age) of the starburst.

Zoom In Zoom Out Reset image size

On the other hand, despite the star-forming regions in the inner ≲1'' ≃ 200 pc of NGC 3256 have larger [Ne ii]12.81 μm LSDs than those in NGC 5135 and NGC 7130 (see Figure 10), they do not show significantly lower 11.3 μm PAH/[Ne ii]12.81 μm ratios but instead rather constant values. Given that the age range of these regions of NGC 3256 is quite narrow (see Figure 9), we can attribute the enhancement of the [Ne ii]12.81 μm LSD and the constantness of the 11.3 μm PAH/[Ne ii]12.81 μm ratio to a mass escalation of the starburst with the [Ne ii]12.81 μm LSD (through its dependency on the mass density), which increases as we move toward the nucleus. Therefore, for the case of these regions, we interpret LSD units as a measure of the intensity (not hardness) of the radiation field in the sense of luminosity (or, equivalently, number of ionizing photons) per unit of physical area, that is, as a measure of the mass density of the starburst. In fact, there exists a correlation between the [Ne ii]12.81 μm LSD and the Paα EW for the star-forming regions in our LIRGs that holds except for the nuclear regions of NGC 3256. In turn, these present an enhanced [Ne ii]12.81 μm LSD without increasing their Paα EW. This is in agreement with our interpretation and supports the idea of that it is the mass density of the starburst which is larger toward the nuclei of NGC 3256, and that the age does not play any role in this case. This behavior is not unexpected since a more massive (denser) starburst will have a higher [Ne ii]12.81 μm LSD but the same 11.3 μm PAH/[Ne ii]12.81 μm ratio than a less massive one because the hardness of the radiation field is not modified (only its intensity). Moreover, in general, the scatter of the data in the horizontal direction in Figure 10 could be explained as an effect of the mass density of the starburst (even for those regions that present a dependence of the 11.3 μm PAH/[Ne ii]12.81 μm ratio with the age). Unfortunately, the nuclear regions of both NGC 5135 and NGC 7130 host an AGN, and thus cannot be used to explore the effects of higher density radiation fields in star-forming regions in LIRGs. In any case, our data suggest that it is only the age (hardness), but not the mass (intensity) of the starburst that modifies the 11.3 μm PAH/[Ne ii]12.81 μm ratio as a function of the [Ne ii]12.81 μm LSD (or equivalently Paα LSD; see Figure 10).

The intensity of the radiation field, however, does have an effect on the EW of the 11.3 μm PAH feature. Figure 11 shows that there is a trend of the 11.3 μm PAH EW to have lower values for increasing Paα LSD (equivalent to [Ne ii]12.81 μm LSD). This tendency is mostly driven by the star-forming regions of NGC 3256 (and the two H ii regions of NGC 7130) and therefore is related to the mass of the starburst, not the age (see above). Hence, given that the 11.3 μm PAH/[Ne ii]12.81 μm ratio of these regions is not varying with the [Ne ii]12.81 μm LSD we could argue that, for a given age, the more massive (denser) is a star-forming region, the stronger is the continuum at 11.3 μm with respect to its 11.3 μm PAH emission (i.e., the lower is the 11.3 μm PAH EW). Therefore, our data suggest that the PAHs (at least in these regions and at the spatial scales probed here) are not destroyed by the fact of being exposed to more massive starbursts but instead they are being diluted by the enhanced continuum emission. This dilution could be due to smaller surfaces of the PDRs (from where the PAH emission arises) when compared to the volumes of the continuum emitting dust for more massive starbursts.

Zoom In Zoom Out Reset image size

Wu et al. (2006) also established that the density (hardness and intensity) of the radiation field has an impact on the PAH emission of low-metallicity blue compact dwarf galaxies, and now our study reveals that this effect is also present in the relatively high-metallicity star-forming regions of LIRGs. However, our work also suggests that, at least for the age and metallicity ranges of our star-forming regions, a higher radiation field intensity (in the form of a more massive starburst) causes the dilution of the PAHs (i.e., a decreasing of the 11.3 μm PAH EW, in this case) but not their destruction (since the 11.3 μm PAH/[Ne ii]12.81 μm ratio is not modified). Therefore, our data support the PAH dilution scenario also claimed by other studies of low-redshift ULIRGs (Desai et al. 2007) and high-redshift SMGs (Menéndez-Delmestre et al. 2009). The destruction of the PAHs (if it would be taking place) would be related to the age of the star-forming regions (hardness of the radiation field).

Moreover, the diffuse regions located in between the H ii, Paα-emitting regions of our LIRGs show 11.3 μm PAH EWs larger (approximately twice) and 11.3 μm PAH/[Ne ii]12.81 μm ratios higher than those measured in the H ii regions (see, e.g., Figure 17). This is in agreement with the idea of these diffuse regions being less massive (have lower intensities, implying higher 11.3 μm PAH EWs, see above) and older (have milder radiation fields, implying higher 11.3 μm PAH/[Ne ii]12.81 μm ratios) than the H ii regions, as it was suggested in Section 8.1. This also reinforces the idea of the PAH dilution scenario in which the H ii regions would be diluted by the enhanced dust continuum emission that is not otherwise seen in the diffuse regions.

Summarizing, we have shown that the PAH and MIR continuum emitting regions are not always spatially overlapped and display different properties. For example, the MIR dust continuum is mainly associated with massive, young, Paα-emitting H ii regions, i.e., with current SF. The PAH emission on the other hand is also characteristic of more diffuse (less massive) and evolved stellar populations, i.e., of recent SF, that show large 11.3 μm PAH EWs and high 11.3 μm PAH/[Ne ii]12.81 μm ratios. This is very important from the point of view of knowing how recent and where is located the SF in LIRGs (either ≲8–10 Myr or tens of Myr). Depending on the measurement used (MIR dust continuum or PAH emission), different stages in the process of SF and different locations in galaxies will be probed. Therefore, high spatial resolution observations of local and also high-redshift LIRGs are essential to determine in a statistical way: (1) how recent the SF is and (2) where the SF is located in these galaxies (compact H ii regions and/or diffuse emission?). This is also ultimately important in order to know whether local and high-z LIRGs, ULIRGs, and SMGs share the same physical properties and belong to the same galaxy population.

The northern nuclear spectrum of NGC 3256 (see Figure 12; if no specification is given, we will refer to the northern nucleus as the main nucleus of this LIRG) shows prominent PAH features (the 8.6 μm PAH feature and the 11.3 μm PAH complex) which are indicative of intense SF. This is corroborated by the presence of conspicuous [Ne ii]12.81 μm line emission. Furthermore, there is no hint for other lines such as [S iv]10.51 μm that could suggest the existence of an AGN (and/or very young SF). The northern nucleus of NGC 3256 was previously observed by Martín-Hernández et al. (2006) and Lira et al. (2008). Both works also detected intense PAH features and the [Ne ii]12.81 μm emission line. In an aperture of 12× 3'', Martín-Hernández et al. (2006) measured a flux of the [Ne ii]12.81 μm emission line in the northern nucleus of NGC 3256 of 2.1× 10⁻¹² erg s⁻¹ cm⁻², which means that only ∼6% of this flux is contained within our T-ReCS nuclear aperture (about 30 times smaller than theirs). In addition, Martín-Hernández et al. (2006) also measured a 11.3 μm PAH flux from the northern nucleus of 2.5× 10⁻¹² erg s⁻¹ cm⁻²; the T-ReCS nuclear flux accounts for ∼5% of their measurement. The T-ReCS integrated spectrum (not including the southern nucleus) displays the same features as its nucleus, but with both the 8.6 and 11.3 μm PAHs and the [Ne ii]12.81 μm emission line showing larger EWs (see Figure 12 and Tables 5and 6).

Zoom In Zoom Out Reset image size

The southern nucleus of NGC 3256 (the region of interest of this galaxy) is located ∼5'' to the south of the northern nucleus. It is behind large amounts of dust even making the MIR wavelengths to be very affected by a high extinction (A_V≳ 12–15 mag; Kotilainen et al. 1996; Lira et al. 2002; Alonso-Herrero et al. 2006a; Díaz-Santos et al. 2008). This is reflected in the N-band T-ReCS spectrum as a deep absorption of the Si _9.7 μm feature. Figure 12 shows an extremely absorbed continuum whose emission is totally suppressed between 9 and 11 μm. The H ii classification of this nucleus would imply the detection of prominent 8.6 and 11.3 μm PAHs as in the northern one. In fact, Martín-Hernández et al. (2006) and Lira et al. (2008) find both features in their spectrum of the southern nucleus, but we do not. The lack of PAH emission in the T-ReCS spectrum is probably due to that our aperture is much smaller than theirs implying that either there is not PAH emission in the very nuclear region or the extinction is that high (A_V > 12 mag, with A_12 μm ≃  0.037 A_V; Rieke & Lebofsky 1985) that prevents us from detecting the PAHs. In fact, their spatial resolutions and, most important, the size of their extraction apertures (∼2''× 2'' ≈ 390 pc × 390 pc) are >5 times larger than ours (036× 036 ≈ 70 pc× 70 pc). Therefore, their spectrum contains not only the emission from the nucleus but also from the surrounding regions, which includes diffuse continuum and PAH emission. Regarding the [Ne ii]12.81 μm emission line, the flux of the southern nucleus contained in our aperture is only ∼10% that of Martin-Hernandez et al.'s, suggesting that probably much of the [Ne ii]12.81 μm emission they measure is mostly off-nuclear.

We detect 8.6 μm PAH emission at distances of about 1'' (∼196 pc) north and south from the northern nucleus (see Figure 13). The 11.3 μm PAH emission is even more extended, out to 2'' to the north. Because the 8.6 μm PAH detection threshold is higher than that of the 11.3 μm PAH we can only measure it in the central regions. There is no PAH emission in the southern nucleus. The [Ne ii]12.81 μm emission line is detected at almost all positions along the slit and, as with the PAH features, its nuclear emission is more extended than the continuum emission.

Zoom In Zoom Out Reset image size

Moreover, the spatial profile of the [Ne ii]12.81 μm emission is remarkably similar to that of Paα. In particular, the [Ne ii]12.81 μm emission is also detected in the southern nucleus. This agreement is not only qualitative but also quantitative as their ratio is almost constant along the slit. This finding is in agreement with previous results that suggest that the [Ne ii]12.81 μm emission line is directly linked to the total number of ionizing photons of a region, that is, to its SFR (see Section 7, and also Roche et al. 1991; Ho & Keto 2007) as traced by, e.g., the Paα line. The [Ne ii]12.81 μm/Paα ratio is broadly constant (it ranges between ∼ 2and5, not corrected for extinction). However, small-scale variations (less than an arcsecond, i.e., few hundreds of pc) can be found. Variations in the fraction of Ne ⁺ ions, which in turn depends on the age of the ionizing stellar populations (Martín-Hernández et al. 2005), could be responsible for the [Ne ii]12.81 μm versus Paα differences.

The 11.3/8.6 μm PAH ratio is constant along the slit indicating that the ionization conditions (that are suggested to be traced by the amount of neutral to ionized PAHs; Galliano et al. 2008) do not vary significantly throughout the nuclear region of the LIRG on the physical scales probed here (∼100 pc). The 11.3 μm PAH/[Ne ii]12.81 μm ratio (not corrected for extinction) is also almost constant (≃1) within 1'' from the nucleus (see also Figure 10). However, the ratio starts to increase toward the north (positive offset; between 1''and2''). A very young population where all the neon were in the Ne ⁺⁺ state would show a high 11.3 μm PAH/[Ne ii]12.81 μm ratio and would explain this behavior. However, the Paα EW map of the galaxy (see Díaz-Santos et al. 2008) does not display any enhancement of the Paα EW in this area. In addition, we see a very good correlation (slope of 1.00 ± 0.02; see Section 7) between the [Ne ii]12.81 μm and the Paα lines, indicating that all the SF is accounted by the [Ne ii]12.81 μm emission. We refer the reader to Section 8.2 with regards to the discussion on the 11.3 μm PAH/[Ne ii]12.81 μm ratio.

The northern nucleus of the galaxy shows lower EWs of the [Ne ii]12.81 μm emission line and 8.6 and 11.3 μm PAHs than its surrounding regions. This is probably because the features are diluted by the more compact continuum emitted by the hot dust that is concentrated toward the nucleus (see also Sections 6 and 8.2).

Unlike NGC 3256, the nuclear spectrum of IC 4518W displays a rather featureless continuum with the exception of the barely detected [S iv]10.51 μm emission line and the [Ne ii]12.81 μm emission line (see Figure 14). It is not clear whether the lack of PAH emission at both 8.6 and 11.3 μm can be interpreted in terms of absence of SF, since the PAH molecules could have been destroyed by the Sy2 nucleus of the LIRG. As for NGC 3256, the nuclear and integrated spectra of IC 4518W are very similar. The integrated spectrum of the galaxy (see Figure 14) also shows an almost featureless continuum as the nuclear one, as well as the presence of the [S iv]10.51 μm and the [Ne ii]12.81 μm emission lines. An important remark here is that the flux of the [S iv]10.51 μm line is larger for the integrated emission of the galaxy when it is subtracted from that of the nucleus, than for the nucleus itself. That is, for the rest of the LIRGs in which the [S iv]10.51 μm line is detected, the nuclear emission accounts for at least or around half of the flux seen in their integrated spectra (compare Tables 3 and 4; see also Appendices C and D; the nuclear fluxes of IC 4518W and NGC 5135 have not been corrected for aperture effects). In other words, the [S iv]10.51 μm emission line usually stems from the very nuclear region (≲100 pc) of the galaxies except for IC 4518W (see below).

Zoom In Zoom Out Reset image size

The region of interest of IC 4518W is a region located 05 (≃165 pc) to the north of the galaxy. Its spectrum is very similar to that of the nucleus, not only in the shape of the continuum but also in its spectral features (see Figure 14). In fact, the [S iv]10.51 μm emission line is as intense as in the nucleus. This is interesting because the nucleus of IC 4518W is unresolved (FWHM ≲04; see Table 2) in continuum emission whereas the [S iv]10.51 μm emission is clearly detected out to ∼08 (≃265 pc) north from the nucleus. Indeed, this extended [S iv]10.51 μm emission can explain the difference seen above between the nuclear and integrated spectrum of the galaxy.

Figure 15 shows quantitatively that the [S iv]10.51 μm line emission of the region located ∼ 05  to the north of the nucleus (positive offset) is comparable to that of the nucleus. Although the [S iv]10.51 μm line can be produced in very young star-forming regions, we do not detect significant Paα nor [Ne ii]12.81 μm emissions spatially coincident with the [S iv]10.51 μm extended emission. Moreover, we do not detect any of the two dominant 8.6 and 11.3 μm PAHs in the spectrum of the Sy2 nucleus of this LIRG. Thus, we conclude that the extended [S iv]10.51 μm emission is most likely associated to the central AGN and produced in an extended narrow line region (NLR) excited by the Seyfert nucleus. Because of the intermediate ionization potential of the [S iv]10.51 μm line, it can be ionized by the AGN emission far away from the nucleus. This has already been found in other Sy2 galaxies, such as Circinus (Roche et al. 2006) or NGC 5506 (Roche et al. 2007), where the [S iv]10.51 μm line is more extended than the dust continuum emission and has been associated to high-excitation (coronal or narrow line) regions.

Zoom In Zoom Out Reset image size

The change on the physical conditions between the nucleus and the location identified as the extended NLR is also shown by the [Ne ii]12.81 μm/[S iv]10.51 μm ratio. While it is approximately constant in the nucleus (almost ≃3), the ratio steeply diminishes toward the [S iv]10.51 μm extended emission region. The value measured in the nucleus is in agreement with the limit given by Martín-Hernández et al. (2006), [Ne ii]12.81 μm/[S iv]10.51 μm ≲ 3, for sources with no PAH detection. They use the [Ne ii]12.81 μm/[S iv]10.51 μm ratio as a measure of the hardness of the radiation field. A higher [Ne ii]12.81 μm/[S iv]10.51 μm ratio implies a softer medium, where PAH emission would be detected. Our values are always below this limit along the slit, in agreement with the lack of PAH emission in this galaxy. The EWs of the [S iv]10.51 μm and the [Ne ii]12.81 μm lines show their lowest values in the nucleus, increasing slowly outward. This is again an effect of the continuum emission being more spatially compact than the line emission.

The Seyfert 2 nuclear spectrum of NGC 5135 shows intense [S iv]10.51 μm line emission (see Figure 16), suggesting a relatively high ionized medium in agreement with its classification as Seyfert. There is also faint [Ne ii]12.81 μm line emission probably associated with the central AGN as well. Although quantitatively speaking there is no PAH emission, there seems to be a little bump and a peak on the spectra at the positions where the 11.3 μm PAH should be. Following the evidence above, the PAHs are supposed to be depleted in the vicinity of an AGN but in some cases, the molecules can survive to the radiation field if they are shielded, for example, by the SF itself (Voit 1992; Mason et al. 2007). If this was the case, we could not definitively affirm that there is no SF in the nuclear region (central ∼100 pc) of NGC 5135. In any case, the AGN effectively dominates the MIR luminosity of the nucleus and the SF would account for a very small fraction of it, and if there is any, it must be at a very low level compared with that taking place in the circumnuclear region (see below). Contrarily to NGC 3256 and IC 4518W, the integrated spectrum of NGC 5135 shows significant differences when compared to its nuclear spectrum. In this LIRG, the MIR AGN emission is clearly separated from that of the SF (see Díaz-Santos et al. 2008). The integrated spectrum includes not only the emission from the nucleus but also the emission arising from an H ii region located at about ∼26 southwest, and from the diffuse region between both (see Figure 1).

Zoom In Zoom Out Reset image size

Both the H ii and the diffuse emission regions of NGC 5135 (the two regions of interest of this galaxy; see Figures 1 and 16) exhibit strong 11.3 μm PAH emission. The MIR continuum near the edges of the spectrum is however significantly brighter in the H ii region. Thus, the diffuse ISM of NGC 5135 is characterized by intense PAH emission but by a fainter continuum emission from hot dust when compared with that of the H ii region. The 11.3 μm PAH EW at the diffuse region is two times higher (∼1 μm) than that measured at H ii region (∼0.5 μm). This is expected if the typical stellar populations in the region of diffuse emission are not extremely young (≲10 Myr) neither very massive, which is in agreement with the presence of weak Paα emission. Therefore, the lack of a hot dust continuum (associated with the Paα emission) in the diffuse region would make its 11.3 μm PAH EW larger than in the H ii region, supporting the PAH dilution scenario in star-forming regions (see discussion in Section 8.2).

On the other hand, the extinction could play a role in this situation. If the PAH and continuum emissions are decoupled, that is, arise from different regions, they can be obscured by different amounts of cold dust, thus leading to different extinctions. There is a correlation between the Paα LSD of star-forming regions and their obscuration (brighter regions are affected by higher extinctions; Calzetti et al. 2007; and Díaz-Santos et al. 2008, their Figure 9). Therefore, if the H ii region (that shows bright Paα emission) is more absorbed than the diffuse medium, the strength and shape of the Si _9.7 μm absorption feature would make the continuum of the H ii spectrum deeper at ∼10 μm when compared to the emission at 8 or 13 μm. This is in agreement with what we see in Figure 16 where the diffuse region shows a shallower spectrum. However, this is difficult to test since our data do not enable us to probe the geometry of the regions in detail or how the dust is distributed.

NGC 5135 is a clear example where SF and AGN activity are well separated (see Figure 1) with our T-ReCS spatially resolved spectroscopy. While we find 11.3 μm PAH emission outside (≳05) the nucleus but not within it, the [S iv]10.51 μm emission line is only detected in the central region (≲07), with little overlap between both features (see Figure 17).

Zoom In Zoom Out Reset image size

As for NGC 3256, the spatial profile of the [Ne ii]12.81 μm emission line follows well that of the Paα line. There is [Ne ii]12.81 μm emission not only in the nucleus and in the H ii region located at 26 southwest (negative offset), but it is also detected in the Paα bump at ∼08. However, unlike NGC 3256, which is classified as H ii, the nuclear [Ne ii]12.81 μm emission of NGC 5135 should be mainly associated to the AGN. Therefore, we have scaled the Paα profile to the [Ne ii]12.81 μm emission at the location of the H ii region as we found that, for star-forming dominated regions (as in NGC 3256), both emission lines correlate quite well (see Section 7). Taking this into account, we can see that the nucleus shows a slightly higher [Ne ii]12.81 μm/Paα ratio than the H ii region. This excess might be attributed to the extinction as the [Ne ii]12.81 μm line is less affected by obscuration than Paα.

Figure 17 shows that the 11.3 μm PAH emission is not only produced by the youngest stellar populations as traced by the Paα line emission, but it is also detected in the diffuse ISM. This confirms that the PAH emission is associated not only to the ionizing stars but also to the more evolved stellar populations and their UV flux (see also, e.g., Peeters et al. 2004; Tacconi-Garman et al. 2005). That is, the spatial profile of the PAH emission profile does not resemble that of the Paα line (see also Section 8), and the diffuse region shows a 11.3 μm PAH intensity similar to that of the H ii region. On the other hand, the [Ne ii]12.81 μm emission is fainter in the diffuse region and bright where there is an enhancement of the Paα line.

The MIR emission of the nucleus of NGC 7130 is resolved (Díaz-Santos et al. 2008) and shows signatures of both SF and AGN activity. Its optical classification as LINER/Sy is in agreement with the detection of the [Ne v]14.32 μm emission line in the Spitzer IRS spectrum of the galaxy (Alonso-Herrero et al. 2010), supporting the existence of an AGN. The T-ReCS nuclear spectrum reveals clear 8.6 and 11.3 μm PAH emission together with the [S iv]10.51 μm emission line (see Figure 18). This is a clear example of the coexistence of PAH emission and an AGN within less than ∼100 pc. There are two explanations for this: (1) the PAH molecules are far away enough from the AGN so that they cannot be destroyed by its radiation field and (2) the PAH molecules are being shielded in some manner. The integrated T-ReCS spectrum shows the same features as the nuclear spectrum since the nucleus of NGC 7130 is quite compact and outside ∼500 pc there is no emission (neither continuum nor feature emission).

Zoom In Zoom Out Reset image size

The nucleus of NGC 7130 is a clear example where SF and AGN activity manifest their characteristic spectral features within the same region (see Figure 19). We find that the [Ne ii]12.81 μm and Paα profiles are as compact as the (resolved) continuum emission. We also detect 8.6 and 11.3 μm PAH emission, but extended within the central 15 of the galaxy. Both, the PAH features, and the [Ne ii]12.81 μm and Paα emission lines clearly indicate that there is SF within the nuclear region. Moreover, Figure 19 shows that in NGC 7130 the PAH molecules can survive within a distance of less than ∼100 pc from the AGN without being destroyed. The intrinsic hard (2–10 keV) X-ray luminosities (L_2–10 keV) of the AGNs of NGC 7130 and NGC 5135 are very similar (≈ 1× 10⁴³ erg s⁻¹; Levenson et al. 2005, and Levenson et al. 2004, respectively). Therefore, the existence of PAH emission in the nucleus of NGC 7130 but not in the nucleus of NGC 5135 suggests that the PAH carriers are not being destroyed in the latter but the absence of PAHs is due to the fact that there is no nuclear SF (or at least it is very weak). The [S iv]10.51 μm line is also detected (although with a high uncertainty) as in the nucleus of IC 4518W and NGC 5135, and is as compact as the continuum emission.

Zoom In Zoom Out Reset image size

The [Ne ii]12.81 μm/Paα ratio is almost constant in the nuclear region of the galaxy (between 3and5 within the inner 1''≃ 340 pc). The 11.3/8.6 μm PAH ratio is also almost constant along the central ∼200 pc with a mean value of ≃2.3. This value is slightly higher than that seen in NGC 3256 (≃1.7), but both are in agreement with the median ratio found by Smith et al. (2007) for the SINGS galaxy sample (1.5 ± ^1.4_0.3). The spatial profile of the 11.3 μm PAH/[Ne ii]12.81 μm ratio has his minimum at the nucleus of the galaxy with a value of ≃1.1 (similar to that of the nucleus of NGC 3256) and increases outward from the central region.

The [S iv]10.51 μm/11.3 μm PAH ratio peaks at the nucleus, while the [Ne ii]12.81 μm/[S iv]10.51 μm ratio has the minimum there. The former is due to the fact that the 11.3 μm PAH emission is more extended than the [S iv] 10.51 μm line emission. In the same manner, the [S iv]10.51 μm emission seems to be more concentrated than the [Ne ii] 12.81 μm emission as the [Ne ii]12.81 μm/[S iv]10.51 μm ratio increases outward from the nucleus. The nucleus of NGC 7130 is resolved in continuum emission so these variations are indeed real. In fact, this would be in agreement with the [S iv]10.51 μm line emission being the signature of the unresolved AGN emission, and the [Ne ii]12.81 μm line emission being mostly associated to the surrounding star-forming ring seen in the UV (González Delgado et al. 2001; also detected in PAH emission). Because the 8.6 and 11.3 μm PAH emissions are more extended than the continuum and line emissions, the spatial profiles of their EWs show again their minima in the nucleus, increasing outward. In contrast, the [S iv]10.51 μm EW is approximately constant (within the large uncertainties) confirming that the region from where the [S iv]10.51 μm line emission arises is very compact (at least as compact as the continuum).