STELLAR POPULATION PROPERTIES AND EVOLUTION ANALYSIS OF NGC 628 WITH PANCHROMATIC PHOTOMETRY

Subtraction of the bias and dark current and flat-field correction were performed with the PIPLINE 1 procedure as described in Fan et al. (1996) once the CCD frames were observed. Astrometric information was appended to the image header during this step. Deep frames of each band were combined after shift and rotation to the same image center and orientation according to the star positions. Cosmic-ray hits and bad pixel effects were fixed up during the image combination. Short exposures for both the object and the standard stars were taken at photometric nights. We use these images to perform the flux calibration for the combined deep image as follows. (1) Four Oke–Gunn standard stars (Oke & Gunn 1983) are extracted in short exposure images of the standard stars and their instrumental magnitudes are measured by a stellar photometric procedure (DAOPHOT; Stetson 1987). (2) Comparing the instrumental magnitudes and calibrated AB magnitudes of these standard stars, we obtain the atmospheric extinction coefficient and instrument zero point. (3) Tens or hundreds of unsaturated bright stars are selected as secondary standard stars in a short exposure image of the object, which is observed under the same photometric condition as the standard stars. (4) According to the extinction coefficient and zero point, the instrumental magnitudes of these stars are converted to flux-calibrated AB magnitudes. (5) The average difference between the instrumental magnitudes of the same stars in the deep combined image and their AB magnitudes are calculated. (6) With this difference, an ADU can be converted to a calibrated flux density unit of erg s⁻¹ cm⁻² Hz⁻¹. The calibration error of each short exposure for the object contains two items: one is from the magnitude estimations of the standard stars and the other is from the differences of those secondary standard stars between the instrumental and AB magnitudes. The total calibration error for each band as shown in the last column of Table 1 is the rms of the calibration errors in all short exposures, the number of which is presented in the seventh column of Table 1.

Sky backgrounds should be removed from the mosaics of all bands in order to acquire the intrinsic flux of the galaxy. For the GALEX mosaics, background images are provided, but the background near the center is overestimated if we check their sky background intensity images. The mosaics from 2MASS, the Bok K_s band, and Spitzer were constructed after removal of the sky background by their own techniques (see references in Table 2). We obtain the smoothing sky background maps for all other mosaics by the polynomial fitting method based on the remaining background pixels after masking the source signals extracted by SExtractor (Bertin & Arnouts 1996) and the galaxy by circling it with temperate apertures for different bands. These fitted sky backgrounds were then removed from the original mosaics.

Due to the diverse pixel scales ranging from 075 to 17 and the different directions of the mosaics, we adjust them to the same scale and direction as the BATC images (i.e., 17 per pixel; north at the top and east at the left). Take the Spitzer mosaic, for example. The pixel size is 075 but that of the BATC mosaic is 17. We first calculate the equatorial coordinate for each pixel of the BATC image and then find the corresponding pixel position in the Spitzer mosaic according to its astrometry. After counting how many Spitzer pixels (these can be decimals) are covered by the area of 1.7 × 1.7 arcsec² centered at this pixel position, we sum all the fluxes of those pixels and subsequently the adjustments of both the pixel scale and direction are completed simultaneously. During the adjustments, we transform the pixel values of all bands into AB top-of-atmosphere fluxes in units of 10⁻³⁰ erg s⁻¹ cm⁻² Hz⁻¹. For the Bok 2.2 μm image from the NASA/IPAC Extragalactic Database (NED), since its astrometry is not the same as ours and the magnitude zero point is uncertain, we do something special: several stars with astrometric positions in the BATC image are used to update the astrometry of this image; mosaic fluxes are recalibrated to the AB magnitude system by some selected bright star in the same field of the 2MASS catalogs.

Figure 2 displays the true color image of NGC 628, combined from three scaled mosaics of the NUV, narrow Hα (from NED), and 8 μm bands, which are chosen to present the distributions of the young stellar population and dust. From this image, we can see that the whole disk is very blue, implying much younger populations in its disk than in the bulge. Numerous H ii regions, identified as luminous spots, mostly lie along the spiral arms. Light in H ii regions comes from both the old galactic background stellar population and young stars, such as O and B type, which have been born recently as a simple stellar population (SSP). These regions are more complex than the galaxy itself and they will be specially treated in our series papers. Pixels belonging to 376 H ii regions in all mosaics are obliterated according to the catalog of Fathi et al. (2007). We also find numerous filaments and substructures generated by PAH emission in the red color of Figure 2, indicating the gas/dust distribution across the disk in addition to the strong PAH radiation along the spiral arms.

Zoom In Zoom Out Reset image size

Mosaics are smoothed for the following reasons (1) different astrometric precisions for telescopes lead a single pixel to represent adjacent slightly different parts of the galaxy, (2) transformations and rotations induce position errors, (3) different observational conditions cause diverse image qualities (e.g., CCD noises and seeing), and (4) S/N varies from the center to the edge of the galaxy so that the photometric accuracy decreases in low S/N areas. Smoothing can improve S/N but can in return reduce the spatial resolution of the galaxy. We used the boxcar averaging method to smooth all the mosaics as expressed in the following formula:

$$\begin{displaymath} M_{I,J} = \frac{1}{w^2}\sum _{K=I-\frac{w}{2}}^{I+\frac{w}{2}} \sum _{L=J-\frac{w}{2}}^{J+\frac{w}{2}} F_{K,L}, \end{displaymath}$$

where I and J specify the pixel position in the image, M_{I, J} is the smoothed flux of this pixel, F_{K, L} is the original flux of any given pixel (K, L), and w is the smoothing window width (the so-called boxcar width), depending on the S/N of the specified pixel in the BATC j-band image. The S/N (R) of a pixel is defined as

$$\begin{displaymath} R = \frac{S}{N} = \frac{S}{\sqrt{S+\delta ^2}}, \end{displaymath}$$

where S is the pixel flux, N is the noise, and δ is the standard deviation of the global sky background. Here, the S/N is similar to the nominal definition in a CCD image, and we only use it to calculate the boxcar width and estimate relative photometric errors for different bands. The window size for each pixel is determined by $w = 2{\min }\lbrace [\frac{R_m}{2R}], 5\rbrace + 1$ where $[\frac{R_m}{2R}]$ is the minimum integer larger than $\frac{R_m}{2R}$ and R_m is set as 10. The formula promises that w will be an odd number to obviate the smoothing bias due to asymmetry. This means that if the S/N of the pixel is equal to or larger than 10, the minimum smooth width is 3 pixels and if the S/N is equal to or less than 1, the maximum width is 11.

After smoothing the mosaics, we extract the observed SED for each pixel. The observed SED includes AB magnitudes of 23 bands from UV to IR, which are calculated as

$$\begin{equation} m^{\rm obs} = -2.5{\log }_{10} S -48.6, \end{equation} \tag{ 1 }$$

as long as the flux signal S is larger than 3δ. The magnitude error (σ) is determined by

$$\begin{equation} \sigma = 2.5{\log }_{10}\left(\frac{S+N}{S}\right)=2.5{\log }_{10}\left(1+\frac{1}{R}\right). \end{equation} \tag{ 2 }$$

At least 15 out of 23 bands with valid magnitudes are required for each SED to fit with the model ones as explained in the next section. As a result, we get 33,242 SEDs and their errors for the whole galaxy (see Table 3 for the SEDs of six randomly sampled pixels).

PEGASE is a spectrophotometric evolution model for starburst and evolved galaxies of the Hubble sequence attributed to the extension to the NIR atlas of synthetic spectra with revised stellar libraries to include cold stellar parameters, stellar tracks of the asymptotic giant branch (AGB), and post-AGB phase (Fioc & Rocca-Volmerange 1997). The stellar tracks they adopt are mainly from the "Padova" group, which contains a wide range of chemical abundances, Z = 0.0001, 0.0004, 0.004, 0.008, 0.02, 0.05, and 0.1 with Y = 2.5Z + 0.23 (Z_☉ = 0.02, where Z is the metallicity abundance and Y is the helium abundance). The initial stellar mass in the tracks ranges from 0.6 M_☉ to 120 M_☉ where, in the track of Z = 0.1, pseudotracks with masses larger than 9 M_☉ are generated from the corresponding masses in the Z = 0.02 and Z = 0.05 tracks.

The stellar spectral libraries for optical wavelength (3130–10800 Å) come from the stellar spectrophotometric atlas of Gunn & Stryker (1983) which includes 175 stars with complete ranges of spectral type and luminosity class. For the FUV (1230–3200 Å), stellar spectra are extracted from the International Ultraviolet Explorer ESA/NASA libraries (Heck et al. 1984). In the extreme-UV (220–1230 Å), spectra for the effective temperature T_eff < 50,000 K are calculated using the models of Kurucz (1992). The models of Clegg & Middlemass (1987) are used at all wavelengths for hotter stars (T_eff > 50,000 K). For cold stars dominating the NIR (1–5 μm) of the spectra, observational libraries are adopted when possible and otherwise synthetic spectra are adopted. Blackbody radiations supplement the NIR wavelength range of hotter stars. In the mid-infrared (>5 μm), the analytic extension of Engelke (1992) is used for stars colder than 6000 K.

An SSP is regarded as a simple system whose stars are born at the same time and with the same initial chemical compositions (typical example: globular clusters). SSPs can be calculated in PEGASE with initial mass functions (IMFs), stellar libraries, and stellar evolutionary tracks. PEGASE provides several commonly applied IMFs, and the Salpeter (1955) law with stellar masses larger than 0.1 M_☉ and less than 120 M_☉ is used to generate SSP models in this paper.

Galaxies are much more complex systems than globular clusters. An evolutionary composite stellar population (CSP) can more accurately describe the evolution history of galaxies. CSP is considered to be a superimposition of SSPs of different ages. The integrated spectrum of CSP is synthesized with an IMF and an SFR for every metallicity. We assume each pixel of the mosaics to be such a CSP system that is composed of stars formed in different periods. There are three kinds of SFRs in PEGASE and the exponentially decreasing SFR with τ = 15 Gyr is used in our paper. Here, τ = 15 Gyr is typical for Sc galaxies (Bolzonella et al. 2000).

Another feature of PEGASE is that the model takes into account the nebular emission (continuum and lines) generated by the ionized gas in the star-forming regions. Nevertheless, the regimes used to synthesize the SED for H ii regions are very outdated. We do not consider the nebular emission in the model spectra.

For model spectra, both age (from 0 to 20 Gyr) and metallicity (from 0.0001 to 0.1) are interpolated by the PEGASE code in logarithmic space with steps of 0.01. The Salpeter (1955) IMF and exponentially decreasing SFR are used to convolve these spectra to create CSPs of the specified metallicity. We use the dust extinction model of Cardelli et al. (1989) to redden the CSP spectra with a reddening step of 0.01 ranging from 0 to 1.0. By convolving the reddened CSP spectra with the filter transmission curves of all bands, we can obtain the model SEDs set in the ranges of different ages, abundances, and reddenings. The model SED contains convolved AB magnitudes of 23 bands computed as

$$\begin{equation} m_i^{\rm csp} = -2.5{\log }_{10}\frac{\int _\lambda {F_\lambda }(t, Z, E){T_i(\lambda)}d\lambda }{\int _\lambda {T_i(\lambda)}d\lambda } - 48.6, \end{equation} \tag{ 3 }$$

where i is the specified band index, T_i is the corresponding filter transmission curve, F(t, Z, E) is the CSP spectrum of the specified age t, metallicity Z, and reddening value E in E(B − V), and m^csp_i is the resultant synthetic AB magnitude. By comparing the observed SED (see Section 3.3) with the synthetic model SEDs, we can simultaneously derive those three parameters (t, Z, and E) for each pixel of NGC 628 via the χ² minimum fitting method as described in Kong et al. (2000) and Ma et al. (2009b):

$$\begin{displaymath} \chi ^2 (t,Z,E) = \sum _{i=1}^{23}\frac{\big[m^{\rm obs}_i - m_i^{\rm csp}(t, Z, E)\big]^2}{\sigma _{i}^2}, \end{displaymath}$$

where m^obs_i is the observed magnitude of the specified filter in the observed SED as shown in Equation (1), σ_i is the observed magnitude errors as given in Equation (2), and m^csp_i(t, Z, E) is the corresponding synthetic magnitude in the model SED as shown in Equation (3).

The degenerate effect of age, metallicity, and dust reddening exists when we match the observed SEDs with the model ones predicted by EPS, since the ways in which these three parameters affect spectra are similar. However, dust and gas absorb the UV radiation of starlight and reradiate in the infrared region. The multiband photometric data from ultraviolet to infrared may aid in degrading this degeneracy. Before being fitted by the above method, the observed SEDs are corrected with the foreground reddening of the Galaxy by using the same reddening law of Cardelli et al. (1989). NGC 628 is located far away from the Galactic disk as the Galactic latitude of its center is b = −457 (R.A.: 2417 and decl.: 1578). Thus the foreground Galactic reddening is small, about 0.07 mag in E(B − V) (Schlegel et al. 1998).

We randomly sample six pixels from different parts of NGC 628 to illustrate the SEDs, fitted parameters, and the fitting goodness: No. 1 near the galactic core, No. 2 from the inner region of the disk, No. 3 from the outer region of the disk, No. 4 from the inner spiral arm, No. 5 from the outer spiral arm, and No. 6 close to an H ii region (see Table 3). The positions of these six pixels are marked in the true color map of Figure 2 and the observed SEDs, matched model SEDs, and the corresponding model spectra are plotted in Figure 3. An excellent match to the model spectrum can be seen in this figure. We might coarsely conclude that the core is oldest, age becomes younger farther from the center, and spiral arms are younger than the rest of the components except for the places near the H ii regions with much younger ages and richer abundances.

Zoom In Zoom Out Reset image size

As generally known in spiral galaxies, the bulge and disk exhibit many physical and dynamic differences. The bulge is usually brighter, older, and dynamically hotter than the disk. In order to analyze different components of NGC 628 with respect to the two-dimensional (2D) features, we disassemble the galaxy into two main components. These two components are obtained by fitting the azimuthally averaged surface brightness with the exponential disk profile (Freeman 1970) and the Sérsic (1968) law following the iterative decomposition procedure of Kormendy (1977). In this decomposition method, two separate regions of the disk and bulge are chosen in the beginning. Then, a least-squares fitting of an exponential law to the radial surface brightness in the disk range is performed. The calculated exponential disk contribution is extrapolated to the bulge range and subtracted from the observed profile to get a first estimate of the bulge component, which has been fitted by the Sérsic (1968) law. In the same way, this fitted Sérsic (1968) law is extrapolated to the disk range and subtracted from the observed profile to get an estimate of the underlying disk, which can be fitted by the exponential law again. The procedure repeats until all the parameters of the two laws converge within a specified accuracy (10⁻³).

The radial profile is the corrected azimuthally averaged value, given the disk inclination of 6°, the position angle of the major axis of 25°, and the distance of 8.6 Mpc as previously mentioned. The value at each radius is defined as the average of the pixel values (here, surface brightness) within a suitably specified annulus of that radius centered on the galaxy nucleus. The profile error is calculated as the standard deviation of those pixel values. All radial profiles of different parameters (age, color, metallicity, and reddening) in the rest of the paper are computed in the same way with those adopted parameters (inclination, position angle, and distance).

The left panel of Figure 4 shows the radial surface brightness profile of the BATC d band (filled squares with error bars), whose effective wavelength is close to the broad B band, and the decomposed bulge and disk components as shown in the dashed curve and the dotted line. We use the same regions where the bulge and disk clearly dominate the observed profile as adopted in the paper of Boroson (1981) to iteratively derive the decompositions: between 35 and 230 for the bulge and between 690 and 2302 for the disk. The decomposition procedure gives a central surface brightness (μ^B₀) of 19.64 ± 0.19 mag arcsec⁻², an effective radius (r_e) of 0.41± 0.04 kpc (0.16 ± 0.02 arcmin), a Sérsic index (n_b) of 0.82 ± 0.08 for the Sérsic law of the bulge, a disk central surface brightness of μ^D₀ = 21.19 ± 0.11 mag arcsec⁻², and a scale length of h = 3.25 ± 0.20 kpc (1.30 ± 0.08 arcmin) for the exponential law of the disk. We obtained a much smaller effective radius of the bulge than that of Boroson (1981), because the de Vaucouleurs (1948) law cannot fit the bulge component very well when we refer to Figure 6 of their paper. Despite a large discrepancy for the bulge, the scale length of the disk approximates that of Boroson (1981) corresponding to the adopted distance of 12.2 Mpc. Actually, many exceptions to fitting the bulge with the de Vaucouleurs (1948) law were mentioned by McDonald et al. (2009 and references therein), and it is more acceptable that the projected three-dimensional bulge profile is fitted by the generalized Sérsic (1968) law.

Although blue bands are traditionally used to derive the structural parameters of different components, their light profiles are substantially affected by the recent star formation and the extinction of the gas and dust. On the contrary, the NIR luminosity is a good tracer of the stellar mass and maps the distribution of the old stellar populations in galaxies. It is insensitive to the luminosity of young massive stars and unaffected by the extinction of the gas and dust for its ignorable extinction coefficient. Therefore, we also decompose the surface brightness profile of the NIR K_s band into two components as shown in the right panel of Figure 4. The central surface brightness, the effective radius, and the Sérsic index of the bulge are 16.87 ± 0.18 mag arcsec⁻², 0.52 ± 0.04 kpc (0.21 ± 0.02 arcmin), and 1.31 ± 0.11, respectively. The disk central surface brightness and scale length are 19.35 ± 0.10 mag arcsec⁻² and 2.51 ± 0.12 kpc (1.00 ± 0.05 arcmin), respectively. The effective radius, Sérsic index, and disk scale length are consistent with the structural parameters derived by Ganda et al. (2009) for the Hubble Space Telescope (HST) H-band photometry (Ganda et al. 2009), which give r_e = 0.21 ± 0.002 arcmin, n_b = 1.23 ± 0.01, and h = 1.18 ± 0.02 arcmin. From the fitting parameters, we find that the disk profile decreases more dramatically in the NIR band than in the blue optical band since there are a number of young stellar populations located in the disk and they raise relatively more luminosity in the blue bands. The effective radius of the K_s band is larger than that of the d band, giving a more reliable span of the stellar mass of the bulge. We regard the region of R < 021 as the region of the bulge, the region of R > 05 where the profile starts to deviate from the exponential law as the region of the disk, and the region between these two areas as a transition zone.

Zoom In Zoom Out Reset image size

Figure 5 displays the age distributions of NGC 628 in two dimensions (left panel) and a histogram (right panel). In the left panel of Figure 5, we find that age is oldest near the galactic core and becomes younger with increasing radial distance (the age ranges from about 10 Gyr to 2.0 Gyr). Two apparent spiral arms in the disk, where considerable H ii regions are concentrated, are much younger (∼1.3 Gyr) than any other components. They extend from the center to the outer region of this galaxy. The inter-arm areas are filled with relatively older stellar populations.

Zoom In Zoom Out Reset image size

In order to derive the statistical properties of the stellar populations more physically, the near-infrared luminosity is used as the mass weight for each pixel in the galaxy image. Since the near-infrared luminosity is insensitive to young massive stars, and reddening in this wavelength range can also be ignored, the near-infrared bands, especially the K_s band, are ideal tracers of the stellar mass (Cole et al. 2001; Kochanek et al. 2001). In the following section of this paper, we will derive various statistical features (e.g., average values, histograms, and radial profiles) of age, metallicity, and reddening weighted by the K_s-band flux.

The average age of the whole galaxy, bulge, and disk is about 4.9 Gyr, 7.5 Gyr, and 4.4 Gyr, respectively. Half of the total stellar mass is younger than about 3.5 Gyr as seen in the right panel of Figure 5. In this figure, we can also see that there are quite a few extremely young stellar populations with age less than 1 Gyr. A majority of them are located close to the double spiral arms and some H ii regions as confirmed by their positions on the age map.

Zoom In Zoom Out Reset image size

Sánchez et al. (2011) presented a wide-field Integral Field Spectroscopy survey on NGC 628 and obtained 2D spectra in the field of view of about 6 arcmin in diameter with a spatial resolution of about 2.7 arcsec. A dithering mode was adopted for the central pointing to increase the spatial resolution. They derived a global age of about 8.2 Gyr by both fitting the integrated spectrum of the whole observed galaxy to linearly combined SSPs and averaging the radial age distribution. The discrepancy of the average age with respect to our estimates mainly originates from the different size of the observed galactic area, which in our study is more than two times that of Sánchez et al. (2011). Considerable rather young stellar populations in the outer region of the disk are involved in the calculation of our average age.

The radial age profile in the left panel of Figure 6 shows that there are two distinct components in the disk where we consider the radius range of 05–10 to be the inner region of the disk and the range from 10 to 22 to be the outer region of the disk (separated at the radius of about 10). This radial age profile is very similar to that of Sánchez et al. (2011) within the radius they analyzed regardless of the system uncertainties. The inner region of the disk is older and its age gradient is much steeper than that of the outer region. The mean ages of these two components are about 7.2 Gyr and 2.9 Gyr, respectively, and the slopes are 11.9 ± 1.0 Gyr arcmin⁻¹ (4.8 Gyr kpc⁻¹) for the inner region (solid line in left panel of Figure 6) and 1.2 ± 0.1 Gyr arcmin⁻¹ (0.5 Gyr kpc⁻¹) for the outer region (dashed line in the left panel of Figure 6).

Such distinct disk components in the age distribution should appear in the color profile which is related to the properties of the underlying stellar population, such as age and metallicity. We calculated the radial profile of the BATC d − g color as plotted in the right panel of Figure 6. These two bands are selected according to their effective wavelengths approximating the broadbands of B and V. The variation in this color profile is similar to that of the age profile: a flattening or even inverse tendency within 05 and two different gradients in the region of R > 05.

Although the inner and outer regions are both in the exponential disk, the age of the inner region is much older than that of the outer one. The steeper age gradient of the inner region of the disk indicates that this smaller component should experience a very long evolution history in an age span of about 7 Gyr. The outer region of the disk with a much shallower gradient might have formed 2–3 Gyr ago within a time interval of about 1 Gyr, which implies that NGC 628 might have endured some encounter or accretion events and a large amount of gas fell in to produce new stars and generate the large-scale outer region of the disk in a very short time (see Section 6.3).

In the left panel of Figure 6, contrary to the disk components, an inverse age gradient resides in the bulge and the transition zone (also discovered in the d − g color profile), while the oldest age up to 10 Gyr is at a galactocentric radius of about 05 (1.3 kpc) instead of the galactic center. A relatively younger structure near the galactic core indicates that some processes like gas inflow directly through the galactic disk to the bulge might have occurred in the history and triggered the star formation. This also may be associated with the possible existence of a weak barred potential that induced the circumnuclear star formation as presented in the paper of Seigar (2002).

Natali et al. (1992) showed the broadband UBVRI surface photometry of NGC 628 and found an obvious blue drop in their B − V profile for R < 1.5 kpc, which can be predicted by models due to infall of gas in the bulge. The submillimeter CO (1−0) observation of Wakker & Adler (1995) and the infrared 2.3 μm CO absorption spectroscopy of James & Seigar (1999) displayed the existence of a circumnuclear star-forming ring in the center of the galaxy. Cornett et al. (1994) used UV surface photometry and discovered that the nuclear region has the overall morphological characteristics of spiral arm materials resembling those of M33 and might have endured a significant star formation in the past few Gyr predicted by the UV-optical colors.

In Sánchez et al. (2011), in addition to the decreasing age gradient in the inner region and the possible shallow gradient in the outer region of the disk, an inverse age gradient also was obtained in a circumnuclear ring at about 25'', which is very close to the position of our result. This kind of ring structure was also detected by Ganda et al. (2006) in their Hβ and [O iii] distributions, obtained by the integral field spectroscopy with the SAURON Integral Field Unit pointing to the galactic core (about 33'' × 41'').

The metallicity Z, derived by the SED fit, is converted to [Fe/H] using the standard chemical composition of the Sun (Grevesse & Sauval 1998). The related histogram and radial distribution also are calculated. In the left panel of Figure 7, we can see that the metallicity is lower near the core than in the outer regions and the abundance close to the spiral arms is much richer than the inter arms due to their poles of the frequently born place of massive stars.

Zoom In Zoom Out Reset image size

Average [Fe/H]s of the whole galaxy, bulge, disk, and the inner and outer regions of the disk are about −0.77, −1.04, −0.78, −0.93, and −0.71 dex, respectively. The derived global metallicity is lower than that of Sánchez et al. (2011), whose spectra mostly lie along the spiral arms and in the bulge. Much richer abundances of H ii regions in the spirals make the global metallicity of Sánchez et al. (2011) larger than ours. In the histogram of [Fe/H] (middle panel of Figure 7), two components dominate this spiral galaxy: an older stellar population with poorer abundance and a younger population with richer metallicity. The poor stellar population mostly lies within the galactic core and inter arms, while the rich one mainly resides close to the spiral arms and the outer region of the disk. The distributions of these two populations take on a bimodal profile. We fit the overall distribution to a double Gaussian function which has two single Gaussian functions as its items. The parameters of these two Gaussian components are summarized in Table 4. The two components and the bimodal curve are plotted with the solid and dashed lines, respectively, in the middle panel of Figure 7.

As presented in H ii observations of Talent (1983) and Belley & Roy (1992), an obvious radial O/H abundance gradient exists in the disk of NGC 628, which is 0.21 dex arcmin⁻¹ and 0.17 ± 0.004 dex arcmin⁻¹, respectively, corresponding to their adopted distances. Equally, a weak decreasing [Fe/H] gradient located in the outer region of the disk is shown in the right panel of Figure 7. Sánchez et al. (2011) also obtained the radial stellar metallicity distribution, the profile of which is very similar to that of ours. The slope of this weak gradient is about 0.11 ± 0.03 dex arcmin⁻¹ (0.05 dex kpc⁻¹). Comparing those gradients with our result of 0.11 dex arcmin⁻¹, we discover that the stellar metallicity gradient is somewhat lower than that of the gas abundance. It is notable that there is a reverse gradient of the inner region of the disk with a slope of 0.48 ± 0.08 dex arcmin⁻¹ (0.19 dex kpc⁻¹). This kind of abundance distribution in the disk provides fundamental constraints on the chemical evolution of NGC 628.

Both dust and gas absorb and scatter starlight and make the galaxy redder. Extinctions are much larger near the UV bands than in the IR bands. Accurate photometry from the UV to infrared can derive reliable intrinsic reddening (E(B − V)) of NGC 628 as shown in the left panel of Figure 8. In this map, regions close to the spiral arms and the bulge hold relatively larger reddenings. The mean reddening of those regions close to spiral arms is about 0.38 and shows no radial gradient, which is consistent with the observational measurements in Belley & Roy (1992). Some inter-arm regions and parts of the outermost disk appear to be less reddened by dust and gas.

Zoom In Zoom Out Reset image size

In order to confirm the reliability of the reddening value, dust emission from Spitzer 8.0 μm is compared with the reddening distribution. We construct the intensity map from the dust in Spitzer 8.0 μm (right panel of Figure 8) dominated by PAH emissions using the method of Helou et al. (2004), who subtracted the background stellar fluxes from the observed fluxes. In the determination of the stellar fluxes in wavelengths longer than 3.6 μm, Helou et al. (2004) extrapolated the 3.6 μm flux using the stellar population model of StarBurst99 (Leitherer et al. 1999) to obtain the scale factor of 0.232 in 8.0 μm (i.e., D_8.0 = I_8.0 − 0.232I_3.6, where D_8.0 is the intensity of the dust emission and I_3.6 and I_8.0 are the observed intensities in the Spitzer 3.6 and 8.0 μm, respectively). Superposed contours in the right map of Figure 8 show the distribution of 2.6 mm CO (1−0) emission intensity from the Berkeley–Illinois–Maryland Association Survey of Nearby Galaxies (BIMA SONG; Regan et al. 2001). Regan et al. (2006) concluded that the CO and 8 μm emission morphologies are similar as also presented in this map and 8 μm PAH surface brightness can be used as a possible tracer of the interstellar medium. Clearly, strong PAH emissions are preferentially distributed along the spiral arms where a number of H ii regions radiate UV photons to photodissociate the PAHs around the molecule clouds and generate IR emissions. Compared with the reddening map, we discover that many of the regions with large reddening values in the galaxy show luminance peaks in the 8 μm PAH image, implying the existence of abundant gas and dust. Some other regions with large reddenings might be caused by the absorption of dust lanes as checked in the optical observations with respect to the positions of primary dust lanes and reddening feathers, which were investigated by La Vigne et al. (2006).

The histogram in the left panel of Figure 9 shows that E(B − V) varies from 0.1 to 0.6 and the mean reddening is about 0.31. The average reddenings of the bulge, the disk, and the inner and outer regions of the disk are about 0.41, 0.31, 0.32, and 0.31 mag, respectively. The radial distribution in the right panel of Figure 9 presents a very weak gradient of the intrinsic reddening with a slope of 0.015 ± 0.002 mag arcmin⁻¹. Several bumps in this plot are produced by the large reddening values near the spiral arms. We can see in the radial profile that the bulge and the transition zone have larger reddening values than the outside area, supporting the young circumnuclear ring structure holding abundant gas and dust. That is, a considerable volume of dust and gas should be deposited in the core of NGC 628, giving a rise of star formation in its recent history. In contrast, Kong et al. (2000) studied another nearby galaxy (M81) of Sab type using a similar method. Their results show that M81 has a much lower reddening range of 0.08–0.15 mag in the bulge than ours and the disk reddening (∼0.2 mag) is also somewhat smaller, implying that NGC 628 might be a gas-rich galaxy.

Zoom In Zoom Out Reset image size

The Sérsic indices in both the optical and NIR bands that we have derived from the brightness profile fitting are close to 1, which represents the exponential law. It has been increasingly discovered in recent years that bulges with surface brightness profiles close to the exponential law are significantly different from those with larger Sérsic indices resembling the elliptical galaxies (Kormendy 1993; Kormendy & Kennicutt 2004; Fisher & Drory 2008; Ganda et al. 2009; Fisher et al. 2009). Two types of typical bulges are classical bulges, which are dynamically hot and featureless, and pseudobulges, which retain the memory of the disk origin. Pseudobulges are very common in late-type galaxies (75% of 77 Sd–Sm galaxies as given in Böker et al. 2002). They have flatter shapes, kinematics dominated by rotation and corresponding smaller velocity dispersion, active star formation, nuclear bars, nuclear rings, and/or nuclear spirals, and nearly exponential brightness profiles (n_b ∼ 1), while classical bulges are dominated by random motions, contain old stellar populations, and are more similar to E-type galaxies (Fisher & Drory 2008; Ganda et al. 2009; Fisher et al. 2009, and references therein). Specifically, Fisher et al. (2009) identified pseudobulges as those bulges containing nuclear bars, nuclear spirals, and/or nuclear rings and having a Sérsic index less than 2.

The Sérsic indices of NGC 628 are 0.83 for the blue band and 1.31 for the NIR band, both of which approximate the exponential law. Nuclear spiral arms are clearly seen in the UV band of this galaxy (Cornett et al. 1994). Moreover, the young ring structure detected in our age map and cental concentration of rich gas shown in the reddening map declare recent young star formation in the galactic center. Therefore, the bulge of NGC 628 belongs to that so-called disk-like pseudobulge.

Unlike the classical bulges, which are typically merger-built, pseudobulges have an opportunity to be grown via the internal secular evolution of the disk (Kormendy & Kennicutt 2004; Fisher et al. 2009). Disk galaxies can evolve by the rearrangement of mass and angular momentum driven by the nonaxisymmetries of bars, ovals and/or spiral structures, which actuate the gas infall toward the galactic center, trigger the star formation, and build up the central mass concentrations of the pseudobulges (Kormendy & Kennicutt 2004). In barred or oval disk galaxies, gravitational torques and/or shocks near the potential minimum cause gas inflow and outflow, which can bring together the disk gas, hence triggering the star formation and forming outer rings, inner rings, and central mass concentrations (pseudobulges; Regan & Teuben 2003; Fisher et al. 2009). In normal spiral galaxies without bars and ovals, spiral structures are maintained by the density wave that propagates through the disk. Shocks are formed as gas approaches and leaves the arms, making the gas lose energy and sink into the center, forming the disk-like bulge (Kormendy & Kennicutt 2004). NGC 628 is a late-type spiral galaxy without evident bars and ovals, so it is possible that its pseudobulge was formed by the secular evolution of the disk driven by the nonaxisymmetric potential of spiral arms. Actually, Fisher et al. (2009) concluded that the median pseudobulges could have grown the current stellar mass at their present-day SFRs in 8 Gyr and their results are consistent with a scenario in which bulge growth occurs via internal secular star formation. The age of the pseudobulge in NGC 628 which we obtained previously is about 7.5 Gyr, similar to the results of Fisher et al. (2009) and hence supporting secular evolution.

Beside secular evolution, possible effects that build the pseudobulges include extremely gas-rich accretion events, distant gravitational encounters, and gravitational interactions in a cluster (Kormendy & Kennicutt 2004; Fisher & Drory 2008). NGC 628 is an isolated galaxy and no large galaxies are found within close distance that could generate possible tidal interactions (Kamphuis & Briggs 1992). Thus, gravitational encounters and interactions should be unimportant for the formation of the pseudobulge in this galaxy. H i observations by Kamphuis & Briggs (1992) show two symmetrical high-velocity complexes (HVCs) and an extended tail to the southwest. Gas accretion events from an external H i-rich object may form the outer tail, and accretion can also create extreme velocity deviations making those HVCs visible. Kamphuis & Briggs (1992) argued that we might be witnessing the flattening of the disk after accreting nearby companions a long time ago according to the orbit of the tail, disk warp sustained by a steady gas infall, and complicated velocity distortion in the outer disk. However, it is extraordinarily uncertain whether this accretion-induced formation of pseudobulges can retain disk-like properties. They also did not exclude the primordial origin that regards the observed gas distribution as the leftover of the initial formation of the galaxy. Nevertheless, the primordial origin and external accretion of gas disk in NGC 628 do not prevent this galaxy from forming the pseudobulge via processes in secular evolution and secular evolution can still have an effect.

Both the age map (Figure 5) and the reddening map (Figure 8) show the nucleus with young age (about 3.5 Gyr) and rich gas (about 0.5 mag in E(B − V)). Decomposition of the NIR H band in extremely high resolution of the HST image also shows that the radial brightness profile exceeds the Sérsic law of the pseudobulge in the innermost part of the galaxy (Ganda et al. 2009). These kind of compact nuclei host nuclear star clusters and they are very common in late-type galaxies (Böker et al. 2002). The nuclei are obviously distinguished from the ambient (pseudo)bulges and disk in the sense that they have a much smaller effective radius and higher effective surface brightness. Stellar populations of nuclei tend to have blue colors implying a young age for the stars that contribute most of the light (a typical example is M33 in Long et al. 2002). Both the nucleus and the pseudobulge can exist in the same galaxy but they show a great number of differences, making the nucleus harder to understand. Maybe these tiny and dense nuclei were formed via dynamic friction driving the clusters to sink to the center as mentioned in Kormendy & Kennicutt (2004). However, the origin and evolution of nuclei are quite complicated and their discussion is far beyond our ability with the current results of our study.

In our results, we discover that the disk of NGC 628 can be divided into two parts: the inner region and the outer region. These two parts have different properties of stellar populations as far as age and chemical abundance are concerned. Disk dynamics indicate that the inner part of the disk shrinks and the outer part expands due to angular momentum transport caused by differential rotation and a nonaxisymmetric self-gravitating mode, such as spiral arms (Lynden-Bell 1979; Lin & Pringle 1987; Tremaine 1989). This redistribution of mass and angular momentum results from the minimization of the total energy through the total angular momentum conservation. The gas contraction in the inner part of the disk could more easily trigger star formation than in the outer part, and hence evolve more quickly. Merely by this kind of internal secular evolution, the disk can construct two different regions with a large age discrepancy. Certainly, one should confirm whether this process is fast enough to form the structures in Hubble time. Nonaxisymmetry of the potential caused by the spiral arms can provide the engine for rapid evolution as discussed in Kormendy & Kennicutt (2004). On the other hand, H i observations of Shostak & van der Kruit (1984) and Kamphuis & Briggs (1992) show that most of the gas concentrates on the disk and an extensive tail lies to the southwest. Gas accretion events from an H i-rich object with $M_{\rm H\,\mathsc{i}} \sim 9\times 10^8\, M_\odot$ might have occurred a long time ago and they affect only the outer disk loosely bounded by gravitation (Kamphuis & Briggs 1992). Therefore, the accreted gas could rule the following star formation activities in the outer region, forming a low stellar density and young outer region of the disk.