Astron. Astrophys. 361, 465-479 (2000)

3. Results

3.1. Underlying population

In this section we describe the results obtained after comparing the colors measured and those predicted by the evolutionary synthesis models.

We compared the model predictions with the optical-near-infrared colors measured without correcting for internal extinction (=) and using an extinction correction factor of in the B-band. The latter value corresponds approximately to two times the extinction of a Galactic-type disk with an inclination of 40^o relative to the plane of the sky (see Gil de Paz et al. 1999, GZG hereafter; see also Gil de Paz 2000). Therefore, the results obtained applying these extinction correction factors can be taken as upper and lower limits for the age and mass-to-light ratio of this population, respectively.

The same metallicity, 2/5 , was obtained at any galactocentric distance in the interval 0.9-1.6 kpc. Since these results, basically age and mass-to-light ratio, were obtained using only aperture colors, they are not affected by the distance uncertainty described in Sect. 2.1 of Paper I.

The ages derived range between 5.0 Gyr at galactocentric distances of about 0.9 kpc and 9.5 Gyr at distances larger than 1.3 kpc (for an instantaneous burst and =). The corresponding interval in K-band mass-to-light ratio was 0.62-0.87 / (see Table 1). The change in the age and mass-to-light ratio of the stellar population at distances shorter than 1.3 kpc is due to contamination from the plateau component (see the B-band profile decomposition given in Papaderos et al. 1996a). Since the spatial extent of this component coincides with the H emitting zone (see Fig. 7 in Paper I), this contamination is probably related with a progressively higher contribution of the recent star-forming regions to the total emission.

Table 1. Age and mass-to-light ratio for the underlying stellar population. The best-fitting model in all these solutions is two-fifths solar metal abundant. Solutions for an instantaneous, 1, 3 and 7 Gyr duration burst and constant star formation rate models are shown.

The outer region of the galaxy color profiles seems to indicate that no significant age gradients are present. However, a small positive metallicity or extinction gradient could compensate the existence of a negative age gradient, or vice versa , reproducing the observed color profiles.

The results shown in Table 1 also suggest that, although the age of the underlying stellar population (d1.3 kpc) could range between 5 and 13 Gyr depending on the star formation history considered, the mass-to-light ratio in the K-band is very well constrained for a given extinction correction. Thus, the K-band mass-to-light ratio for = is approximately 0.87  and 0.65  for =.

The small differences (30 per cent) obtained in the maximuum likelihood estimator after comparing our data with models with burst duration shorter than 7 Gyr prevent us to infer the star formation history and internal extinction of the underlying stellar population. Therefore, we are not able to determine if this stellar population has effectively formed in a instantaneous burst or during long (several Gyr) periods of time as it has been observed in I Zw 18 (Aloisi et al. 1999). Only continuous star formation models (or with a duration for the burst longer than 7 Gyr) can be rule out since their very low maximuum likelihood estimators and too old ages derived.

The evolutionary synthesis models developed for the analysis of the star-forming regions properties only depend on the observed colors and mass-to-light ratio of the underlying stellar population. Therefore, the conclusions given in Sect. 3.3 for the study of these regions are not affected by our ignorance on the past star formation history of the galaxy.

In order to build these models (see Paper I and Sect. 3.3) we adopted a mass-to-light ratio for the underlying stellar population of 0.87 / in the K-band, which corresponds to a null internal extinction value. However, if the internal extinction was relatively higher, e.g. =, the mass-to-light ratio could be a 25 per cent lower yielding slightly different properties for the most recent star-forming regions (see Sect. 3.3).

3.2. Central starburst

After applying the comparison procedure described in Sect. 2.2, we obtained the three clusters of solutions in the age, burst strength, metallicity and color excess four-dimensional space. Two of these three solutions show probabilites lower than 1 per cent. In Fig. 1 we show the distribution of the total number of solutions obtained within the remaining solution cluster which has a probability of 98 per cent.

Fig. 1a-d. Frequency histograms for the central starburst a  age, b  burst strength, c  continuum color excess and d  stellar mass.

The age obtained for the starburst component is about 30 Myr, and the burst strength is 20 per cent. Fig. 1c shows that the continuum color excess is very well constrained between 0.06 and . The starburst age derived agrees with the absence of H emission for this component. The expected H equivalent width in emission at ages older than 30 Myr and 20 per cent burst strength is lower than 2 Å for any stellar metallicity ².

In order to confirm these results, we will compare the H equivalent width and D₄₀₀₀, Mg2, Fe5270 and Fe5406 spectroscopic indexes measured with the values predicted by the evolutionary synthesis models. Unfortunately, we only dispose of spectroscopic index predictions for the case of pure burst models. Therefore, we will derive the spectroscopic indexes of this component using the predictions from pure burst models and the burst strength given above. In this way, a molecular index like Mg2 (see Gorgas et al. 1993), can be written for a composite stellar population as

where, and are the spectroscopic indexes for the underlying stellar population and the young starburst, and are the mass-to-ligth ratios at the continuum and b is the burst strength in mass. The mass-to-ligth ratios and are, respectively, 3.436 and 0.084 / in the B-band (Bruzual & Charlot priv. comm. for Z=2/5 ), using =5.51 (Worthey 1994). In the case of an atomic index (Fe5270 and Fe5406) or the equivalent width of H, it can be derived using

where and are the H equivalent widths (or Fe5270 and Fe5406 indexes) of the underlying and young stellar populations.

Finally, we define the D₄₀₀₀ index (Bruzual 1983; Gorgas et al. 1999) as

Then, assuming

where and are the fluxes per unit wavelength of the underlying and starburst populations (=+), we obtain the following expression for the D₄₀₀₀ index,

The continuum mass-to-ligth ratios used were those predicted for the B-band in the Mg2, EW(H) and D₄₀₀₀ cases and for the V-band in the case of the iron indexes (=2.80 / and =0.0147 / for Z=2/5 ). The latter mass-to-light ratios were obtained using =4.84 (Worthey 1994).

Then, using these expressions and the index values for the underlying and starburst populations -from the predictions of the SSP Bruzual & Charlot (priv. comm.) models-, we obtained the results shown in Table 2. The indexes measured (Column 5 in Table 2) were corrected in order to take into account the different spectral resolution between our spectra and those where the Lick indexes were originally defined (see Gorgas et al. 1993and references therein) and also the fact that our spectra are flux-calibrated.

Table 2. Spectroscopic indexes for the underlying population (2/5  metal abundant), starburst component (2/5  and 1/5  metal abundant) and for the composite stellar population using a 30 Myr old burst with burst strengths 20, 10 and 5 per cent. EW(H) and Fe5270 and Fe5406 indexes are expressed in Å.

Thus, the Mg2 index in the Lick system should be 0.02 magnitudes higher than that measured on flux-calibrated spectra (J. Gorgas, priv. comm.). On the other hand, small differences in the spectral resolution relative to the Lick library spectra yield significant changes in the atomic index values. The Lick library spectra show resolutions () of about 200 km s^-1 for the Fe5270 and Fe5406 indexes (J. Gorgas, priv. comm.), being for our spectra 116 km s^-1. The corrected indexes are shown in Column 6 of Table 2.

Despite of the results shown in this table are compatible with those obtained from the optical-near-infrared colors analysis, the burst strength derived seems to be slightly lower. This difference is probably due to the difference in size between the region covered by the slit #4b and the aperture used to measure the starburst colors.

Therefore, we can conclude that our data, both colors and spectroscopic indexes, well agree with a scenario constituted by a 30 Myr old burst superimposed on a several Gyr old stellar population.

Using the mass-to-light ratio predicted for the composite stellar population, the burst strength and the absolute magnitude measured within the aperture we obtained the stellar mass for the starburst. This mass was corrected using the f ratio between the knot continuum emission in the aperture and its total continuum emission. The total mass derived for this component was 910⁶  with an f factor of 0.619 (61.9 per cent). As we commented in Sect. 3.1 for the underlying population, the age, burst strenght and color excess deduced for the starburst component are not affected by the uncertainty in the distance to Mrk 86. However, due to this distance uncertainty, its stellar mass is not known with a precision better than 40 per cent (for a 20 per cent distance uncertainty; M. E. Sharina, priv. comm.).

3.3. Star-forming regions

Using the maximization procedure described in Sect. 2.2 we derived ages, burst strengths and stellar masses for those regions showing H emission (see Table 4 in Paper I). The , , , and colors and H fluxes were compared with evolutionary synthesis models. In this study, models 1) with metallicity in the range 1/50 Z2  and 2) lower than solar, and 3) with 15 per cent and 4) a null fraction of escaping Lyman photons were explored.

In Table 3 we show the mean age, burst strength, metallicity and stellar mass and their corresponding standard deviation values for all the star-forming regions studied. All those clusters of solutions with probability higher than 20 per cent are shown. This probability has been computed by dividing the number of Monte Carlo particles within a given cluster relative the total number of particles (10³). Using the mean value for the highest probability solution cluster of each star-forming region we obtained the frequency histograms shown in Fig. 2.

Fig. 2. Frequency histograms for the age (upper panels), burst strength (central panels) and stellar mass (lower panels) of the star-forming regions. These distributions have been obtained considering only one solution for each star-forming region. This solution corresponds to the mean of the highest-probability solution cluster. (see Table 3)

Table 3. Mean value and standard deviation for the age, burst strength, mass and metallicity of each individual cluster of solutions. Only the properties of those clusters with probability higher than 20 per cent for each region in Column 1 are given. Results for models with 15 per cent and a null fraction of Ly photons escaping from the galaxy are shown separately. Probabilities for these clusters of solutions are given in Columns 6 and 11. Metallicity is expressed as log(Z/).

Table 3. (continued)

Table 3. (continued)

The results shown in Table 3 indicate that 50 per cent of the regions under study only show a cluster of solutions with probability higher than 20 per cent. In the remaining regions the mean differences obtained between the several solution clusters are 2.2 Myr, 0.15 dex and 0.14 dex in age, burst strength and mass, respectively, for models with a 15 per cent fraction of escaping photons and any metallicity. These differences are even lower by using other sets of models -1.8 Myr, 0.10 dex and 0.10 dex, respectively, for the subsolar metallicity models-. In any case, these differences are significantly lower than the dispersion observed in Fig. 2.

From Fig. 2 we also deduce that there is no large differences in the properties derived assuming 15 per cent or a null fraction of Lyman photons escaping from the nebula. If we compare the results obtained using subsolar metallicity models and those obtained for the whole range in metallicity, it seems that a higher number of regions older than 10 Myr and with burst strength lower than 1 per cent is obtained in the former case. Since the metallicity of the ionized gas (see Sect. 3.4) is clearly lower than solar, we are more confident with the results obtained using subsolar metallicity models. The Principal Component Analysis performed on the highest probability solution clusters indicate that the direction in the (t,b,Z) space that better reproduces the data variance is (,,)=(+0.707,+0.707,0.000). This fact suggests the existence of a small degeneracy between age and burst strenght.

In the lower panels of Fig. 2 we show the stellar mass distribution. The stellar masses (see also Table 3) have been computed using the K-band absolute magnitudes measured within the apertures and the mean mass-to-light ratio of the highest-probability solution cluster. These stellar masses were corrected for the aperture effect by dividing them by the factors f given in Table 3 (see Sect. 2.2).

Finally, the age, burst strength and mass values obtained for these regions are represented in Fig. 3 using different sized symbols. In Fig. 3a the size of the symbols used is related with the age of the burst, larger symbols represent younger regions. In Fig. 3b the symbol size is proportional to the burst strength, and finally, in Fig. 3c its size is proportional to the burst stellar mass. Fig. 2 and Fig. 3a show that the age of the star-forming regions is well constrained between 5-13 Myr. There is no significant age gradients across the different structures observed in the H image (see Sect. 4).

Fig. 3a-c. Age, burst strength and stellar mass of the star-forming regions using a symbol size code overplotted on the H image.

It should be noticed that the age and burst strength values derived are not affected by the Mrk 86 distance uncertainty, since only aperture colors and equivalent widths have been used in this work.

Since we have adopted in our models a fixed mass-to-ligth ratio and colors for the underlying stellar population, the contamination from intermediate aged populations could yield sistematically higher age, burst strength and stellar mass values in some regions. This could be the case of the #45, #49 and #59 regions, contaminated from the central starburst continuum emission. On the other hand, the ignorance on the actual mass-to-light ratio of the underlying population, as we pointed out in Sect. 3.1, may introduce slight uncertainties in the stellar masses derived. However, although the absolute values for these masses would be quite uncertain, the relative differences should be similar.

3.4. Gas diagnostic for the star-forming regions

In Table 5 of Paper I we gave the emission-line fluxes measured in 4.302.65 arcsec² regions centred in the maximum of the emission knot section covered by the slit.

The gas electron densities have been obtained from the [SII ]6716 Å/[SII ]6731 Å line ratio following Osterbrock (1989). When the latter line ratio was not measurable we adopted an standard density of =100 cm^-3. In the cases where the ([OIII ]4959 Å+[OIII ]5007 Å)/[OIII ]4363 Å line ratio was measurable, we determined the electron temperatures applying the algorithms given by Gallego (1995). The corresponding oxygen, nitrogen and helium abundances were also computed using the algorithms given by Gallego (1995). Since the HeII 4686 Å emission-line fluxes were not measurable we could not determine the HeIII abundance. In addition, since the [SIII ]9069,9532 Å emission lines were not accessible, we assumed a ionization correction factor of 1 and, consequently, y=. A summary of the electron densities, temperatures and chemical abundances deduced is given in Table 4.

Table 4. Ionized gas diagnostic

Finally, we have compared the line ratios measured with the predictions of a grid of model nebulae taken from Martin (1997) and originally calculated with CLOUDY . The nitrogen-to-oxygen and carbon-to-oxygen abundance ratios used were those employed by Martin (1997). The oxygen abundance was 0.2 .

In Fig. 4 the extinction corrected [OIII ]5007 Å/H, [SII ]6717,6731 Å/H, [OII ]3727 Å/[OIII ]5007 Å and [NII ]6583 Å/H line ratios jointly with the models predictions have been plotted. We have drawn models with effective temperatures in the range 40000-50000 K and ionization parameters between log U=-1.91 and log U=-4.60. Solid-lines represent the change in the line ratios for different ionization parameter values between log U=-1.91 and -4.60 at increments of 4.7 in U. Thicker lines mean higher temperatures. The dashed-lines represent the change in the line ratios as a function of the effective temperature for a fixed ionization parameter. In all these diagrams the ionization parameter of the models increases from right to left. In Fig. 4a we also show the change in the line ratios measured along the major axis of the expanding bubble Mrk 86-B (GZG) from North to South (dotted-line ).

Fig. 4a and b. Extinction corrected emission-line ratios. Only those regions with measurable Balmer decrements are shown. Photoionization models computed with CLOUDY are drawn for 50000, 45000 and 40000 K effective temperatures and ionization parameters between log U=-1.90 and -4.60. Solid-lines connect line ratios computed with different ionization parameters and fixed effective temperature. Thicker-lines mean higher temperatures. Dashed-lines connect predictions for different effective temperatures and fixed ionization parameter.

As it was pointed out by Martin (1997), the bulk of the discrepancy of these line ratios with the prediction of photoionization models suggests the existence of an aditional excitation mechanism. This discrepancy will be higher using lower metallicity models. The contribution of this additional mechanism (or mechanisms) is more significant, relative to that produced by photoionization, in the case of the #45, #54 and #70 star-forming regions. In the latter case, the anomalous line-ratios measured are probably related with enhanced shocked gas emission in the Mrk 86-B bubble fronts (see GZG). The contamination from the Mrk 86-B north lobe could be also responsible for the line ratios measured in the #54 region.

3.5. Comments on several individual regions

#9, #10, #12, #22, #55, #57, #79, #84 and #85 : All these regions show photometric H emission, but very faint or undetectable R-band continuum emission. There are two feasible explanations for this very faint continuum emission.

First, these regions could be high gas density clumps photoionized by distant stellar clusters. Then, they should be placed in regions with intense diffuse H emission. This could be the case of the #9, #10, #12, #22, #55 and #57 regions.

On the other hand, at the early evolutionary stages of a starburst the emission-line equivalent widths can be as high as 1000 Å. Thus, star-forming regions with low burst strength will be only detectable by their H or [OIII ]5007 Å emission. This could be the case of the #79, #84 and #85 regions.

#26 & #27 : These regions conform a massive association (see GZG) with very complex structure. The best-fitting model for the #26 region yields an age of about 10 Myr with a high burst strength value. On the other hand, the #27 region is a younger burst (5 Myr) with low burst strength and complex H emission structure (see Fig. 5). The peculiar velocity profile obtained by GZG suggests that this association could belong to an independent stellar system merged with Mrk 86. In Fig. 5 we show the continuum and H structure of this association. From this figure is not clear if the H emission arises from photoionization of the #26, #27 or both stellar clusters. The H fluxes of the #26 and #27 regions given in Table 5 of Paper I are those measured for the north-east and south-west structures shown in Fig. 5 (right panel ), respectively.

Fig. 5. #26 and #27 regions. B, R, Ks and H images are shown. North is up and East is to the left. The scale is  pixel-1. The size of these images is approximately 0.4 kpc0.5 kpc.

#42, #70 & #18 : These regions correspond to the starburst precursors of the Mrk 86-A, Mrk 86-B and Mrk 86-C expanding bubbles, respectively (see Martin 1998, GZG). From Fig. 3 we observe that these regions have extreme properties. In particular, the #70 region shows the highest burst strength (excepting #45 region and the central starburst) and the #18 region is the youngest of the regions analyzed in Sect. 3.3.

© European Southern Observatory (ESO) 2000

Online publication: October 2, 2000
helpdesk.link@springer.de