Almudena Alonso-Herrero, Marcia J. Rieke, and George H. Rieke Steward Observatory, the University of Arizona, Tucson, AZ 85721, E-mail: firstname.lastname@example.org, email@example.com, firstname.lastname@example.org
Arp 299 (IC 694 + NGC 3690) is one of the nearest examples of interacting galaxies with a high infrared (IR) luminosity ( , for the assumed distance Mpc), close to the limiting value for ultraluminous IR galaxies. We present HST/NICMOS observations taken with all three cameras. We concentrate on the data reduction and analysis of the images taken with camera 3 through the narrow-band filters F164N and F166N. The [Fe II] m morphology and properties are described in detail. The relative ages of the burst of star formation together with the supernova rate for the bright sources in the interacting system are derived.
galaxies, interacting, stellar content, star formation, infrared, Arp299
The existence of luminous and ultraluminous IR galaxies has long been known (Rieke & Low 1972), but it was with the launch of the IR satellite IRAS that this class of galaxies was detected in large numbers. Since then an intense debate started as to whether there is an evolutionary link between luminous IR galaxies, ultraluminous galaxies and optically selected quasars (see Sanders & Mirabel 1996 for a recent review). Among the ultraluminous IR class ( L, IR luminosity between 8 and m), a large percentage is found to be interacting/merging systems containing active galactic nuclei. Sanders et al. (1988) proposed that the IR luminous phase is the initial stage for the appearance of a quasar. In this context the interacting system Arp 299 (NGC 3690+IC 694 or Mrk 171) is an interesting study case by itself, not only because of its high IR luminosity ( L), close to the limiting value for ultraluminous IR galaxies, but also because it is one of the nearest examples of interacting starburst galaxies (distance Mpc for km s-1 Mpc-1).
A number of bright IR and radio sources have been detected from ground-based observations (see Gehrz et al. 1983, Wynn-Williams et al. 1991 and references therein). Following the notation introduced by Gehrz et al. (1983) for the interacting pair of galaxies Arp 299, the nucleus of IC 694 (eastern component) is referred to as source A, and the sources in NGC 3690 (western component) are called B1, B2, C and C (see Figure 1). One of the most remarkable characteristics of this interacting system (and other luminous IR galaxies) is the high concentration of molecular hydrogen within relatively small regions. From CO maps, Sargent & Scoville (1991) estimated the density of molecular gas pc-2 in IC 694, pc-2 in components B1 and B2 of NGC 3690 and pc-2 at the interface of both galaxies, region C+C'. Numerical simulations of collisions between gas-rich galaxies (see Barnes & Hernquist 1996 and references therein) show that collisions are very efficient at transporting large quantities of molecular gas into the centers of galaxies; such large quantities are similar to those observed in IC 694 and other ultraluminous IR galaxies.
HST NICMOS images of the system together with MMT optical and IR spectroscopy are analyzed to derive the star formation properties of the system.
NICMOS on the HST observations of the interacting galaxy Arp 299 were obtained on November 4 1997 using all three cameras. Images were taken with the following filters and cameras: NIC1 F110M; NIC2 F160W, F222M, F237M, F187N, F190N, F212N and F215N; NIC3 F164N and F166N. The pixel sizes for NIC1, NIC2 and NIC3 are 0.045arcsec pixel-1, 0.076arcsec pixel-1 and 0.20arcsec pixel-1 respectively. In this paper we will concentrate on the data reduction and data analysis of the camera 3 narrow-band filter images. The observational strategy consisted of taking a spiral dither with a 5.5 pixel spacing, with two, three or four positions. The orientation of the images is PA = 90degree. The FWMH of the point sources in the fully-reduced NIC3 F164N image is 0.29. It is important to note that the NIC3 images were not taken during the NIC3 observing campaign. Nevertheless the quality of the images is remarkable and most suitable for scientific purposes.
Part of the reduction of the NICMOS images was performed with routines of the package NicRed (McLeod 1997). This data reduction package works within the IRAF environment. Darks with exposure times corresponding to those of our observations were obtained from other proposals close in time. Usually between 10 and 20 darks were averaged together for a given sample sequence after the subtraction of the first readout. The flatfields for the filters NIC3 F164N and F166N are in-flight flats kindly reduced by Dr. Rodger Thompson. The first steps in the data reduction (done with the task nicfast within Nicred) involve subtraction of the first readout, dark current subtraction on a readout basis, correction for linearity and cosmic ray rejection (using fullfit), and flatfielding. Since our NICMOS images were obtained after August 1997, no correction for the pedestal effect was necessary. The background was measured on blank regions of the flatfielded images and subtracted from each image.
As a first try to produce the final [Fe II] m line emission image, the dithered galaxy images for a given filter were registered to a common position using fractional pixel offsets and cubic spline interpolation, and combined to produce the final images through filters F164N and F166N. Once the images were combined, the flux calibration was performed using the conversion factors based on measurements of the standard star P330-E during SMOV (Marcia Rieke 1997 private communication), which are: Jy ADU-1 and Jy ADU-1 for NIC3 F164N and F166N respectively. The combined continuum and line+continuum images were again shifted to a common position and finally the F164N was subtracted from the F166N to produce the [Fe II] m line emission image (note that at the redshift of Arp 299 the [Fe II] m line gets shifted into the F166N filter). The resulting image showed some residuals produced by both the under-sampling of the point spread function (PSF) and the variation of the shape of PSF with wavelength. In addition it appears that for point-like sources and camera 3 there is a significant variation of the shape of the PSF depending on the position along the pixel.
To solve some of these problems we used a different strategy when subtracting the continuum images from the line+continuum images. Instead of combining all the images together prior to the continuum subtraction, we realigned both the F164N and F166N individual images for a given position (usually by less than a few tenths of a pixel), and subtracted the F164N individual images from the F166N individual images. The resulting individual continuum-subtracted F166N images were then shifted to a common position (using the offsets computed with the F166N images before continuum subtraction) and combined to the final line emission image. Even though we first dithered the images for a given filter and then changed filters, this method yielded better results. This is because the pointing of the HST is very accurate, and therefore after changing filter, the telescope goes back to the initial point of the dither sequence with a precision of less than one tenth of a pixel. The method described here would produce even better results for those cases in which for a given position both the continuum and the continuum+line images were taken before moving the telescope to the next position.
Finally we would like to point out that we performed a straight subtraction using the continuum at m, i.e., the continuum has not been converted to m, wavelength of the underlying continuum of the [Fe II] line in Arp 299. For sources in which obscuration is patchy or/and very high, a better approach would be to fit the continuum between two different wavelengths, and construct an image in which the continuum has been interpolated to the line wavelength.
In Figure 1 we show a contour plot of the interacting galaxy Arp 299 through the NIC3 F164N filter which contains the continuum adjacent to the [Fe II] m emission line. Figure 2 is a grey scale map of the [Fe II] m line emission (continuum-subtracted NIC3 F166N filter), and in Figure 3 we overlay the line emission contours on the grey scale map of the continuum emission (NIC3 F164N) for clarity.
The NIC3 F164N continuum image (Figure 1) of the system clearly shows not only the emission from the brightest sources (i.e., A, B1, B2, C and C') already known from ground-based images, but also reveals the spiral nature of IC 694 with a number of H II regions along the spiral arms, along with some compact sources in NGC 3690 surrounding B1, B2, and located south-west of C. In the large scales (better seen in Figure 3) Arp 299 shows the characteristic tidal tails common to interacting pairs extending south and west of NGC 3690.
Most of the [Fe II] m line emission (Figure 2) originates from three regions, that is, source A (nucleus of IC 694), B1 in NGC 3690 and C+C' at the interface of both galaxies. These three regions are not compact but quite extended with linear projected sizes of about 1kpc. In addition sources B1 and C show structure. It is remarkable that very little or no emission seems to be coming from source B2 (the brightest source at visible wavelengths). The same result was found by Fisher, Smith & Glaccum (1991) from their ground-based Br images. In addition to the emission from the bright sources the [Fe II] image traces the emission from H II regions located in the spiral arms south-east of IC 694. There is a number of H II regions northwest of B1, and east of C extending all the way to source C', and east of C. The resemblance in morphology with the our NIC2 Pa ( m) images (not shown here, see Alonso-Herrero et al. 1998) is remarkable, although some differences are found. The [Fe II] emission is more extended in A, B1 and C than the Pa emission. If most of the [Fe II] emission is excited by shocks in supernova remnants (SNR) in the star-forming regions, then it is expected that the [Fe II] emission would be more extended as the shocks propagate outwards the H II regions, whereas the Pa emission will tend to be more concentrated toward the center where the young ionizing stars are located. The [Fe II] m to Pa line ratio is found to vary radially within the star-forming regions, increasing by at least a factor of five for increasing radial distances. In addition, age effects will also contribute to differing line ratios (see next section). In contrast the H2 morphology of Arp 299 is quite different with point-like emission originating from B1 and C in NGC 3690, whereas the nucleus of IC 694 is very bright in H2 and shows some diffuse emission with a beautiful butterfly-like shape (see Alonso-Herrero et al. 1998). The lack of resemblance between the [Fe II] and the H2 emission suggests that the scenario in which both emissions have a common origin may not be that simple, perhaps indicating that the nature of the shocks producing both emissions is different.
So far we have not discussed the effects of extinction. From the optical and near-infrared spectroscopy obtained for all the bright sources in the system, we have estimated the extinction using hydrogen recombination line ratios. However, the values of the extinction to the gas are quite dependent on the wavelength of the lines involved, the highest values obtained when using the line ratio (indicating that some of the hydrogen lines may be still optically thick in the near-infrared). The observed f([Fe II] m)/ line ratios have to be corrected for extinction by a factor (assuming a simple model of foreground dust screen): where the term 0.04 accounts for the differential extinction between m and m. Note that we analytically fit the extinction curve of Rieke & Lebofsky (1985) for near-infrared wavelengths. The values of the extinction derived for the bright components are between mag and at least mag (assuming a foreground dust screen model).
In Alonso-Herrero et al. (1997) we showed that the f([Fe II] m)/ line ratio in starburst galaxies is understood as transition from pure H II region (such as the Orion Nebula) to an increasing rôle of shock excitation by supernova remnants. The extinction-corrected values of this ratio are well apart from the typical value for pure ionization for all the sources indicating that an important fraction of the [Fe II] emission is produced by supernova remnants. Moreover, as the starburst ages the f([Fe II] m)/ ratio will increase as the number of SNR grows and the ionization from very young stars decreases, making this line ratio a good age indicator. From the predictions of the models presented in Vanzi, Alonso-Herrero & Rieke (1998) we can estimate that the relative difference in ages between the bursts in both components A and B1, and components C and C' is approximately 3 million years assuming a Gaussian burst with FWMH Myr. From these models we derive an age (measured from the peak of star-formation) as young as 4Myr for components C and C' in NGC 3690, indicating that the most recent star-formation is occurring at the interface of the two galaxies. A more detailed discussion of the star-forming properties of this system will be presented in Alonso-Herrero et al. (1998).
Finally the total [Fe II] m fluxes can used to derive the supernova rate (SNr) using the calibration for M82 derived in Vanzi & Rieke (1996). For the (SNr) we measure the [Fe II] m fluxes through a 4arcsec-diameter aperture, except for the H II regions north-west of B1 for which we use the flux through the 1arcsec-diameter aperture. The values of the supernova rate are presented in the last column of Table 1. The two values correspond to the supernova rates obtained from the [Fe II] m fluxes not-corrected and corrected for extinction. Gehrz et al. (1983) estimated the supernova rate from their 20cm radio measurements with a 5arcsec beam. Their values are 4.2, 2.5, 1.9 and 0.9yr-1 for components A, B1+B2, C and C'. Taking into account the uncertainty in the calibration for M82 for the (SNr) in terms of the [Fe II] m flux (a factor of 2, Vanzi & Rieke 1996), the agreement between the two independent estimates is quite good for components A and C. The largest discrepancy occurs for B1. Given the fact that source B2 shows no [Fe II] emission it is possible that the radio emission from B2 is not related to SNR, but still included in Gehrz et al. (1983) calculations.
Most of the [Fe II] m line emission is found to be originating from three bright sources, A (nucleus of IC 694) and B1 and C+C' in NGC 3690. Little or no emission is coming from B2 (the brightest source at visible wavelengths). The resemblance between the [Fe II] m and Pa emission is remarkable, although the [Fe II] m emission is extended to larger scales. The [Fe II] m to Pa line ratio is a good age indicator, since as the starburst ages the number of SNR grows whereas the flux from young ionizing stars decreases. We find that the region C+C' at the interface of the two galaxies is undergoing the youngest star-formation process. Finally the [Fe II] m fluxes are used to compute the supernova rate for each component.
During the course of this work AA-H was supported by the National Aeronautics and Space Administration on grant NAG 5-3042 through the University of Arizona. The work was also partially supported by the National Science Foundation under grant AST-95-29190.