1 - Astronomiska observatoriet, Box 515, 751 20 Uppsala, Sweden

2 - Instituto de Astrofísica de Andalucía, CSIC, Apdo. 3004, 18080 Granada, Spain

Abstract
Optical spectra of H II regions of two dwarf irregular galaxies are presented. The objects are IC 4662 and ESO 245-G05. Chemical abundances were derived in all the H II regions where the forbidden oxygen line [OIII] $\lambda$ 4363 Å was detected. For spectra with the highest quality, a study of the spatial distribution of chemical elements in these objects were made. Generally, the chemical composition is largely constant within the H II regions studied. Some differences are found but could be attributed varying physical properties, e.g. small scale fluctuations in the temperature of the region. A comparison of the chemical abundances between different H II regions within the galaxies is made for IC 4662 and ESO 245-G05, where one of the H II regions clearly shows deviating chemical abundances.

Key words: galaxies: evolution - galaxies: irregular - galaxies: stellar content - interstellar medium: H II regions: general - galaxies: individual: IC 4662, ESO 245-G05

A common characteristic uniting the three galaxies is the existence of several bright H II regions which make them ideal targets for a study regarding possible variations in chemical composition. Absolute magnitudes and metallicities previously obtained are similar. However, their environments, distances and morphologies are substantially different. NGC 6822 is a small barred galaxy which belongs to the Local Group (see Paper I); IC 4662 is an isolated galaxy (Hidalgo-Gámez & Olofsson 1998) at an intermediate distance experiencing a major event of star formation (Heydari-Malayeri et al. 1990). Finally, ESO 245-G05 is a large barred galaxy in the outskirts of the Sculptor Group.

IC 4662 is a bright ( $\mu= 21.2$ mag arcsec^-2) galaxy in the direction of the NGC 6300 group (de Vaucouleurs 1975) at a distance of $2.50\pm0.02$ Mpc. The most thorough study regarding chemical abundances and star formation was carried out by Heydari-Malayeri et al. (1990). These authors obtained the oxygen abundance for the brightest regions denoted A1 and A2 (hereafter; A1/A2), which dominate the galaxy in an H $\alpha$ image (Heydari-Malayeri et al. 1990). Observations in the radio band (Becker et al. 1988; Harnett 1987), the infrared (Trinchieri et al. 1989; Sanders et al. 1995) and in the UV (Rosa et al. 1984) have been obtained and no peculiar characteristics were reported for this galaxy. The existence of a total of five bright H II regions, makes it a prime candidate for a study of possible inhomogeneities in the distribution of chemical elements in this object.

ESO 245-G05 (also denoted A 143) is a large ( $r=2.2\pm.2$ kpc), low surface brightness object ( $\mu= 24.0$ mag arcsec^-2) at $4.2\pm.3$ Mpc, with several H II regions along the bar, evident from a continuum-subtracted H $\alpha$ image (Miller 1996). This object is particularly interesting since Miller (1996) reported variations in chemical composition between H II regions within ESO 245-G05.

The main goal of this paper is to derive more information regarding the possible inhomogeneity in chemical abundances in dIs. We present data of higher signal-to-noise ratio compared to previous determinations in order to reassess whether the reported differences really do exist. In addition to new data for ESO 245-G05 we have obtained spectral information for the other three H II regions of IC 4662 (denoted B, C and D), never previously studied. A good quality spectrum was obtained for region A1/A2 which allows a two-dimensional mapping of this region similar to the one carried out for Hubble V and Hubble X in NGC 6822. The distribution of the elemental abundances as well as the extinction, excitation and ionization are studied in regions of 22 pc. The quality of the spectra of ESO 245-G05 allows detection of the weak emission line [OIII] $\lambda$ 4363 Å in two of the regions observed.

In the next section the acquisition of the data is described and the analysis is discussed in Sect. 3. Section 4 is devoted to the extinction determination and in Sect. 5 the chemical abundances of the H II regions in the sample are presented. The spatial distribution throughout the regions A1/A2 in IC 4662 is studied in Sect. 6 and a discussion of possible inhomogeneities of chemical elements is presented in Sect. 7. Conclusions are given in Sect. 8.

2 Observations

Table 1: The log of the observations. The H II regions observed are presented in Col. 1. Column 2 gives the date of the observations while Cols. 3 and 4 present the telescope coordinates. The position angle is given in Col. 5. The total integration time is given in Col. 6 and the air mass in Col. 7
H II region Date $\alpha$ $\delta$ Pos. Angle Int. Time Air Mas.

degrees blue/red blue/red

IC 4661 A1/A2 080897 17 $^{\rm h}$ 46 $^{\rm m}$ 59 $^{\rm s}$ -64 $\hbox{$^\circ$ }$ 38'43'' 289 30 $^{\rm m}$ /20 $^{\rm m}$ 1.24/1.23

IC 4661 D 090897 17 $^{\rm h}$ 46 $^{\rm m}$ 50 $^{\rm s}$ -64 $\hbox{$^\circ$ }$ 39'02'' 310 60 $^{\rm m}$ /100 $^{\rm m}$ 1.26/1.36

ESO 245-G05 nr. 19 080897 01 $^{\rm h}$ 45 $^{\rm m}$ 01 $^{\rm s}$ -43 $\hbox{$^\circ$ }$ 37'20'' 270 40 $^{\rm m}$

ESO 245-G05 nr. 12 090897 01 $^{\rm h}$ 45 $^{\rm m}$ 05 $^{\rm s}$ -43 $\hbox{$^\circ$ }$ 37'21'' 295 90 $^{\rm m}$ /60 $^{\rm m}$ 1.08/1.04

**Table 1:** The log of the observations. The H II regions observed are presented in Col. 1. Column 2 gives the date of the observations while Cols. 3 and 4 present the telescope coordinates. The position angle is given in Col. 5. The total integration time is given in Col. 6 and the air mass in Col. 7
H II region	Date	$\alpha$	$\delta$	Pos. Angle	Int. Time	Air Mas.
				degrees	blue/red	blue/red
IC 4661 A1/A2	080897	17 $^{\rm h}$ 46 $^{\rm m}$ 59 $^{\rm s}$	-64 $\hbox{$^\circ$ }$ 38'43''	289	30 $^{\rm m}$ /20 $^{\rm m}$	1.24/1.23
IC 4661 D	090897	17 $^{\rm h}$ 46 $^{\rm m}$ 50 $^{\rm s}$	-64 $\hbox{$^\circ$ }$ 39'02''	310	60 $^{\rm m}$ /100 $^{\rm m}$	1.26/1.36
ESO 245-G05 nr. 19	080897	01 $^{\rm h}$ 45 $^{\rm m}$ 01 $^{\rm s}$	-43 $\hbox{$^\circ$ }$ 37'20''	270	40 $^{\rm m}$
ESO 245-G05 nr. 12	090897	01 $^{\rm h}$ 45 $^{\rm m}$ 05 $^{\rm s}$	-43 $\hbox{$^\circ$ }$ 37'21''	295	90 $^{\rm m}$ /60 $^{\rm m}$	1.08/1.04

Two different slit positions were used for each galaxy. Table 1 presents the individual H II regions, the coordinates, position angles, total integration times for the blue and the red spectral regions, as well as the air mass for each position. Due to poor weather conditions, only data for the blue spectral region is presented for ESO 245-G05 nr. 19. Two subexposures were taken at each position in order to check for cosmic events. The sky was clear at the zenith position except during the end of the first night. The seeing conditions were stable during the observations, with values smaller than 1 $.\!\!^{\prime\prime}$ 2 for all the positions except that for ESO 245-G05 nr. 19 where the seeing was 1 $.\!\!^{\prime\prime}$ 4. Air masses during the observations of ESO 245-G05 were quite small and no correction for differential refraction was made. In the case of IC 4662 the air masses were higher and, using Table 2 of Fillipenko (1982), a correction was made. The initial slit position for IC 4662 passed in the SE-NW direction through the main body of the galaxy, where the two H II regions, A1/A2, (Heydari-Malayeri et al. 1990), were located. The second slit was positioned in the SSE-NNW direction through the three bright H II regions, B, C and D, located to the southeast of the main body. The oxygen emission line [OIII] $\lambda$ 4363 Å was detected in regions A1/A2 and D only. The latter is the most distant region to the main body.

A slit was positioned along the bar in ESO 245-G05, encompassing four of the H II regions (Miller 1996). A second slit was positioned in the EW direction parallel to the first one but slightly towards the north in order to fully encompass all the regions in the southern part of the bar. A third slit was positioned to pass through the northeast extreme of the bar and the adjacent regions (nr. 1 to 10 in Miller 1996). Due to the poor signal-to-noise ratio in the emission line [OIII] $\lambda$ 4363 Å, only two out of eight regions were finally studied (nr. 19 and nr. 12 in Miller 1996).

The reduction was performed with the software package MIDAS. Bias and flat-fielding were performed with bias-frames and continuum lamps. He-Ar arcs were used for the wavelength calibrations. The flux calibrations were performed using spectra of the standard stars LTT 1788 and LTT 1020, with an accuracy of 5 $\%$ and 8 $\%$ in the blue and red spectral regions, respectively. The tables of La Silla were used for the atmospheric extinction correction. See Paper I for a more detailed description regarding the reduction process.

3 The analysis of data

A study of the spatial distribution in the chemical abundances within the H II regions was made for the high quality spectra of IC 4662 A1/A2. For these regions the two-dimensional spectra were divided, as described in Paper I, into three-row spectra (hereafter; 3r-spectra). At the distance of this galaxy, a 3r-spectrum comprises a physical region of 22 pc. The total number of spectra for the H II regions of IC 4662 are 20 for region A1 and 32 for region A2. The same procedure for region D resulted in 27 spectra. These numbers correspond to total sizes of 448 pc, 717 pc and 330 pc, respectively. As a consequence of the low signal-to-noise ratio, no further analysis was carried out for region D. These numbers are based on different emission lines, mainly [OII] $\lambda$ 3727 Å in the blue and H $\alpha$ in the red spectral regions. The alignment of both the blue and the red spectral regions was carried out with the aid of the continuum emission between 5200 Å and 5400 Å, which is the spectral region where the blue and the red overlap. The overlap is perfect for IC 4662 A1 but a displacement of a few pixels was found for IC 4662 A2. The same problem prevail for ESO 245-G05 nr. 12. To gain insight into the possible existence of variations in the chemical abundance between H II regions in dI galaxies, spectra with high signal-to-noise ratios are desirable. Therefore, all the rows where the [OIII] $\lambda$ 4363 Å line was detected were summarized, which resulted in a one-dimensional spectrum for each H II region. The number of rows where [OIII] $\lambda$ 4363 Å was detected was: 10 for IC 4662 A1, 11 for IC 4662 A2, 6 for IC 4662 D, 9 for ESO 245-GO5 nr. 19 and 16 for ESO 245-G05 nr. 12, corresponding to 6.1, 6.7, 3.7, 5.7 and 9.8 arcsec, respectively.

The intensities of the spectral lines were measured with software developed at Uppsala Astronomical Observatory. Consult Paper I for more details regarding the analysis. The spectral resolution of the configuration (8 Å nominal but 16 Å considering the seeing conditions) was not sufficiently high to resolve the doublet [SII] $\lambda$ $\lambda$ 6717,6731 Å. In this case, a special routine in the software, described in Paper I, was used for handling blended lines.

A set of intensities for each 3r and spatially-averaged spectrum was obtained. Spectra with a perfect alignment were normalized to the intensity of the H $\beta$ line. In the case of small displacements, the lines in the blue part of the spectrum were normalized to the intensity of the H $\beta$ line. Correspondingly, lines in the red were normalized to the intensity of the H $\alpha$ line.

The one-dimensional spatially-averaged spectra are shown in Fig. 1 (IC 4662) and Fig. 2 (ESO 245-G05).

3.1 Errors in the continuum and line intensities

As discussed in Paper I, three different sources of uncertainty have been considered for the intensity of the spectral lines: uncertainties introduced by the reduction process, uncertainties due to the extinction corrections and uncertainties in the level of the spectral continuum, with respect to the line.

The latter was found to be the major contributor. Typical values of the uncertainty introduced by the extinction correction were only 30% of the uncertainties in the continuum. The corresponding values due to the flux calibrations are 5 $\%$ in the blue and 8% in the red spectral region.

Two more terms were added for some lines. In the case of blended lines, an additional term was added which takes into account the goodness of the deblending procedure. The uncertainties are 11% for [SII] $\lambda$ $\lambda$ 6717,6731 Å, 15 $\%$ for [NII] $\lambda$ 6548-H $\alpha$ and 18 $\%$ for [OI] $\lambda$ 6300-[SIII] $\lambda$ 6312 Å for IC 4662 A1/A2. The corresponding values for IC 4662 D were 22 $\%$ , 30 $\%$ and 59 $\%$ .

For IC 4662 a correction term, corresponding to the differential refraction, was added to spectral lines affected. It was found that a correction of 15 $\%$ was necessary for the line [OII] $\lambda$ 3727 Å. No other lines were affected by differential refraction.

A total uncertainty was obtained adding together all the terms considered for each line. These are shown in Table 2 for the 3r-spectra of IC 4662. These values correspond to spectra with the lowest signal-to-noise ratios in the emission line [OIII] $\lambda$ 4363 Å and should therefore be considered as the maximum statistical errors.

Table 2: The maximum uncertainties in the intensity of the emission lines. These values correspond to the 3r-spectra with the lowest signal-to-noise ratios in the line [OIII] $\lambda$ 4363 Å and reflect the quality of spectra
Line IC 4662 A1 IC 4662 A2 IC 4662 D

[OII] $\lambda$ 3727 18 60 43

[NeIII] $\lambda$ 3869 24 60 68

HeI $\lambda$ 3889 0 0 26

[NeIII] $\lambda$ 3967 59 62 25

H $\delta$ $\lambda$ 4102 0 0 68

H $\gamma$ $\lambda$ 4340 0 0 34

[OIII] $\lambda$ 4363 83 51 88

HeI $\lambda$ 4471 97 37 61

H $\beta$ $\lambda$ 4861 0 0 43

[OIII] $\lambda$ 4959 11 7 65

[OIII] $\lambda$ 5007 11 7 45

HeI $\lambda$ 5875 21 26 62

[OI] $\lambda$ 6300 0 0 90

[SIII] $\lambda$ 6312 0 0 90

[OI] $\lambda$ 6363 0 - 93

[NII] $\lambda$ 6548 44 94 94

H $\alpha$ $\lambda$ 6563 0 0 51

[NII] $\lambda$ 6583 26 34 51

HeI $\lambda$ 6678 78 59 55

[SII] $\lambda$ 6716 49 64 76

[SII] $\lambda$ 6730 48 69 72

**Table 2:** The maximum uncertainties in the intensity of the emission lines. These values correspond to the 3r-spectra with the lowest signal-to-noise ratios in the line [OIII] $\lambda$ 4363 Å and reflect the quality of spectra
Line	IC 4662 A1	IC 4662 A2	IC 4662 D
[OII] $\lambda$ 3727	18	60	43
[NeIII] $\lambda$ 3869	24	60	68
HeI $\lambda$ 3889	0	0	26
[NeIII] $\lambda$ 3967	59	62	25
H $\delta$ $\lambda$ 4102	0	0	68
H $\gamma$ $\lambda$ 4340	0	0	34
[OIII] $\lambda$ 4363	83	51	88
HeI $\lambda$ 4471	97	37	61
H $\beta$ $\lambda$ 4861	0	0	43
[OIII] $\lambda$ 4959	11	7	65
[OIII] $\lambda$ 5007	11	7	45
HeI $\lambda$ 5875	21	26	62
[OI] $\lambda$ 6300	0	0	90
[SIII] $\lambda$ 6312	0	0	90
[OI] $\lambda$ 6363	0	-	93
[NII] $\lambda$ 6548	44	94	94
H $\alpha$ $\lambda$ 6563	0	0	51
[NII] $\lambda$ 6583	26	34	51
HeI $\lambda$ 6678	78	59	55
[SII] $\lambda$ 6716	49	64	76
[SII] $\lambda$ 6730	48	69	72

4 The correction for underlying absorption and extinction

4.1 Underlying stellar absorption correction

All the intensities of the lines in the red spectral range, except those of IC 4662 A1/A2, were normalized to the H $\alpha$ intensity. Subsequently, a correction for the underlying stellar absorption in the Balmer lines was performed (consult Paper I for details). The equivalent width of the absorption feature in the Balmer line H $\beta$ was found to be 3 Å, for the H II regions in ESO 245-G05. This feature was not measurable in spectra of IC 4662.

4.2 The extinction correction

A correction for the extinction due to possible intergalactic and interstellar dust was applied. It should be emphasized that, in this investigation, the total amount of extinction was considered and no distinction was made between the contribution of various sources of extinction.

The extinction correction, applied to both sets of data (the 3r and the spatially-averaged spectra), was performed using the Balmer decrement with the Whitford modified extinction law (Savage & Mathis 1979).

For spectra where the overlap between the blue and red spectral regions was not ideal, the Balmer line H $\gamma$ was used in the determination of the extinction coefficient, instead of H $\alpha$ . It is well known that the H $\gamma$ line suffers a higher amount of stellar absorption than does the H $\alpha$ line. In order to determine whether the extinction performed here is reasonable, a new extinction coefficient was determined using a spectrum observed with grism nr. 2 which allowed the entire spectral region to be observed, including the H $\alpha$ and H $\beta$ lines. No significant differences in the intensities of the spectral lines were found. This will be discussed in more detail in Sect. 5. The H II region ESO 245-G05 nr. 12 was found to be practically free of dust. Table 3 presents the extinction coefficients, $C_\beta$ , for the H II regions studied in this investigation, all determined from the spatially-averaged spectra. In the case of IC 4662 A1/A2, several Balmer lines were observed and a $C_\beta$ was obtained for each of the lines. Only the H $\gamma$ and H $\delta$ lines were detectable in IC 4662 D, and only the H $\gamma$ line in ESO 245-G05 nr. 19 and nr. 12. The extinction coefficients tend to increase toward shorter wavelengths, indicating an increasing importance of underlying stellar absorption. Table 3 also presents the colour excess determined from the data obtained in this investigation, the extinction coefficients available in the literature for these regions, and the colour excess from the RC3 catalogue (de Vaucouleurs et al. 1992). It should be pointed out that these last values correspond to the whole galaxy, while the E(B-V) determined in this analysis are the values for the H II regions.

4.3 The extinction within the H8pt10ptII regions of IC 4662

Because of the large size of the H II regions, as well as the high signal-to-noise ratio of some of the spectral lines, a study of the extinction over the face of these regions was possible. The extinction coefficients obtained from the 3r-spectra are presented in Fig. 3 as a function of the distance from the geometrical center at zero pc. The distance was obtained from the total emission region in H $\alpha$ for A1/A2 and also from the [OII] $\lambda$ 3727 Å emission region for D. These are the most spatially extended lines in each spectrum.

As pointed out by González-Delgado et al. (1994), a correlation between the H $\alpha$ line intensity and the $C_\beta$ suggests a mixing of the ionized gas with the dust. The H $\alpha$ /H $\beta$ line ratio, not corrected for reddening, is also shown in Fig. 3. $C_\beta$ determined from the H $\alpha$ /H $\beta$ line ratio mimics the behaviour of the H $\alpha$ line independent of the origin of the extinction. Therefore, the $C_\beta$ determined from the H $\gamma$ /H $\beta$ line ratio is preferred for all regions, A1/A2 and D.

Table 3: The extinction coefficients, $C_\beta$ , derived from the non-corrected Balmer line intensities. The coefficients reported in the literature are marked $C_{\beta \rm lit}$ . The color excess obtained with the $C_\beta$ derived in this analysis and the values compiled from the RC3 catalogue (de Vaucouleurs et al. 1991) are also presented. It should be noted that the latter values correspond to the whole galaxy. References: (1) Heydari-Malayeri et al. 1990; (2) Miller 1996
line IC 4662 A1 IC 4662 A2 IC 4662 D ESO 245-05 nr. 19 ESO 245-05 nr. 12

H12 - $7\pm2$ - - -

H10
$6.38\pm.01$ $6.4\pm1$ - - -

H9
$5.9\pm.5$ $6\pm1$ - - -

H $\delta$
$1.42\pm.01$ $1.419\pm.001$ 0.073 - -

H $\gamma$
$2.345\pm.004$ $1.14\pm.02$ $0.23\pm0.03$ $1.35\pm$ $0.03\pm0.02$

H $\alpha$
0.16 $\pm$ .06 0.20 $\pm$ .04 - - - -

E(B-V) 0.11 0.14 0.16 0.95 0.02

$C_\beta$ $_{\rm lit}$ 0.13 (1) 0.13 (1) - 0.35 (2) 0.70 (2)

$E(B-V)_{\rm RC3}$ 0.34 0.34 0.34 0.44 0.44

**Table 3:** The extinction coefficients, $C_\beta$ , derived from the non-corrected Balmer line intensities. The coefficients reported in the literature are marked $C_{\beta \rm lit}$ . The color excess obtained with the $C_\beta$ derived in this analysis and the values compiled from the RC3 catalogue (de Vaucouleurs et al. 1991) are also presented. It should be noted that the latter values correspond to the whole galaxy. References: (1) Heydari-Malayeri et al. 1990; (2) Miller 1996
line	IC 4662 A1	IC 4662 A2	IC 4662 D	ESO 245-05 nr. 19	ESO 245-05 nr. 12
H12	-	$7\pm2$	-	-	-
H10	$6.38\pm.01$	$6.4\pm1$	-	-	-
H9	$5.9\pm.5$	$6\pm1$	-	-	-
H $\delta$	$1.42\pm.01$	$1.419\pm.001$	0.073	-	-
H $\gamma$	$2.345\pm.004$	$1.14\pm.02$	$0.23\pm0.03$	$1.35\pm$	$0.03\pm0.02$
H $\alpha$	0.16 $\pm$ .06	0.20 $\pm$ .04	-	-	-	-
E(B-V)	0.11	0.14	0.16	0.95	0.02
$C_\beta$ $_{\rm lit}$	0.13 (1)	0.13 (1)	-	0.35 (2)	0.70 (2)
$E(B-V)_{\rm RC3}$	0.34	0.34	0.34	0.44	0.44

The behaviour of these two parameters is not exactly the same over the face of the objects. The intensity of H $\alpha$ is constant in regions A1/A2, while the $C_\beta$ , obtained from both the H $\alpha$ and H $\gamma$ intensities, gives smaller values toward the northwestern part of the object. The "bumps'' observed in both H $\alpha$ and $C_\beta$ , at 250 pc, 100 pc, and 175 pc may be indicative of a clumpy origin of the dust. The H $\alpha$ line intensity and the $C_\beta$ seem more correlated in region D. This indicates a high amount of internal dust. Due to small number statistics, these results should be treated with caution.

5 The spatially-averaged chemical abundances

As previously mentioned, a spectrum for each H II region was obtained, adding all the pixels where the emission line [OIII] $\lambda$ 4363 Å was detected. The total number of pixels correspond to total sizes of 224 pc for A1, 246 pc for A2 and 134 pc for D of IC 4662, 113 pc for nr. 19 and 201 pc for nr. 12 of ESO 245-G05. The emission line intensities, corrected for extinction and underlying stellar absorption including the uncertainties, are presented in Table 5. Also, the equivalent width, the flux in the H $\beta$ emission line, as well as the signal-to-noise ratio in the [OIII] $\lambda$ 4363 Å line, are tabulated.

In order to derive the ionic chemical abundances, a five-level, two-zone model was used. The abundances are determined with the use of the temperature-sensitive method (Osterbrock 1989). For more details regarding the model and the derivation of the abundances see Sect. 5.1 in Paper I.

**Table 4:** The extinction coefficient, the equivalent width of the $H\beta$ emission line, $E_v(H\beta )$ (in Å), the flux in the $H\beta$ line (in erg s^-1cm^-2) as well as elemental abundances from previous investigations. The regions are: IC 4662 A1 (first column) and IC 4662 A2 (second column) of Heydari-Malayeri et al. (1990). ESO 245-G05 nr. 12, (third column) and ESO 245-G05 nr. 19 (fourth column) of Miller (1996)
	1	2	3	4
$C_\beta$	0.13	0.13	0.35	0.7
$E_v(H\beta )$	146	122	206	70
$\log F(H\beta$ )	-12.37	-12.41	-14.1	-14.37
$T_{\rm e}$ (O⁺⁺)	-	-	12100	-
12+log(O/H)	8.11	8.04	7.90	8.20
N(He)/N(H)	0.101	0.099	-	-
12+log(Ne/H)	7.35	7.35	-	-

The atomic chemical abundances were obtained from the ionic ones and the ionization correction factors (ICF) for nitrogen and neon. No ICF for oxygen is necessary since all the important ionization stages are within the wavelength range studied. No ICF was used for helium, following Izotov et al. (1999). Except for ESO 245-G05 nr. 19, where only the blue spectral region was available, the helium abundances were determined using a weighted average value of the three lines, HeI $\lambda$ 4471 Å, HeI $\lambda$ 5875 Å and HeI $\lambda$ 6671 Å, which are less affected by collisional excitation (Izotov et al. 1997). These atomic abundances are presented in Table 6.

The helium, nitrogen, oxygen and neon abundances, as well as their ICF, when determined, are tabulated. The log(N/O) and log(Ne/O) are also compiled, as well as the electron temperature, $T_{\rm e}$ , and the temperature of the ionizing radiation, $T_{\rm ion}$ . From viewing Table 6 it can be concluded that there are some variations in the abundances of helium, nitrogen, oxygen and/or neon.

5.1 Comparison with previous results

Two previous investigations of the same objects were selected for comparison with the data presented in this analysis. The data are presented in Table 4.

5.1.1 IC 4662

Table 5: Line intensities of the spatially-averaged spectra, normalized to the H $\beta$ emission line, of the H II regions studied in this investigation. The equivalent width and the flux in the H $\beta$ line (given in units of erg s^-1cm^-2), as well as the signal-to-noise ratio in the line [OIII] $\lambda$ 4363 Å, denoted by $\sigma$ (l), are also presented
line $\lambda$ (Å) IC 4662 A1 IC 4662 A2 IC 4662 D ESO 245-G05 19 ESO 245-G05 12

[OII]
3727 $1.1\pm.2$ $1.1\pm.2$ $2.31\pm.03$ $2.9\pm.2$ $2.5\pm.1$

H12
3750 $.006\pm.01$

H10
3798 $0.017\pm.003$ $.017\pm.005$

HeI
3820 $0.02\pm.01$ $0.02\pm.01$

H9
3835 $0.03\pm.01$ $0.03\pm.02$

[NeIII]
3869 $0.40\pm.06$ $0.40\pm.02$ $0.3\pm.1$ $0.27\pm.01$ $0.27\pm.01$

HeI
3889 $0.15\pm.03$ $0.14\pm.04$ $0.18\pm.04$ $0.27\pm.01$ $0.149\pm.004$

CaII 3934 $0.010\pm.01$ $0.008\pm.005$

[NeIII]
3967 $0.21\pm.03$ $0.21\pm.03$ $0.36\pm.02$ $0.3\pm.1$ $0.20\pm.03$

HeI
4026 $0.013\pm.002$ $0.014\pm.002$

[SII]
4068 $0.006\pm.004$ $0.01\pm.02$

+4076

H $\delta$
4102 $0.20\pm.01$ $0.204\pm.001$ $0.20\pm.01$ $0.21\pm.01$

H $\gamma$
4340 $0.35\pm.01$ $0.41\pm.01$ $0.47\pm.1$ $0.47\pm.01$ $0.5\pm.1$

[OIII]
4363 $0.07\pm.01$ $0.07\pm.02$ $0.09\pm.01$ $0.046\pm.005$ $0.09\pm.01$

HeI 4471 $0.04\pm.01$ $0.04\pm.01$ $0.10\pm.02$ $0.08\pm.01$ $0.038\pm.003$

[FeII]
4570 $0.007\pm.002$ $0.008\pm.005$

HeII
4650 $0.04\pm.01$ $.04\pm.02$

ArIV]
4711 $0.016\pm.001$ $0.015\pm.001$

ArIV]
4740 $0.007\pm.001$ $0.007\pm.002$

H $\beta$
4861 $1.00\pm.02$ $1.00\pm.02$ $1.00\pm.01$ $1.00\pm.1$ $1.00\pm.1$

HeI
4921 $0.008\pm.002$ $0.008\pm.001$

[OIII]
4959 $2.1\pm.1$ $2.1\pm.1$ $1.1\pm.1$ $0.82\pm.02$ $0.93\pm.04$

[OIII]
5007 $6.4\pm.2$ $6.3\pm.2$ $3\pm1$ $2.4\pm.1$ $2.8\pm.1$

[ClIII]
5517 $0.005\pm.005$ $0.012\pm.004$

[ClIII]
5530 $0.005\pm.001$ $0.011\pm.004$

HeI
5875 $0.11\pm.01$ $0.11\pm.01$ $0.13\pm.03$ $0.12\pm.01$

[OI]
6300 $0.004\pm0.003$ $0.005\pm.001$ $0.09\pm.02$ $0.023\pm.001$

[SIII]
6312 $.02\pm.01$ $0.03\pm.01$ $0.04\pm.01$ $0.010\pm.002$

[OI]
6363 $0.004\pm.001$ $0.004\pm.002$ $0.04\pm.01$ $0.010\pm.001$

[NII]
6548 $0.02\pm.01$ $0.02\pm.01$ $0.03\pm.01$ $0.023\pm.001$

H $\alpha$
6563 $2.86\pm.30$ $2.86\pm.20$ $2.79\pm.06$ $2.86\pm.10$

[NII]
6583 $0.05\pm.01$ $0.05\pm.01$ $0.07\pm.01$ $0.15\pm.01$

HeI
6678 $0.03\pm.01$ $0.03\pm.01$ $0.04\pm.01$ $0.043\pm.004$

[SII]
6716 $0.10\pm.05$ $0.09 \pm.04$ $0.2\pm.1$ $0.3\pm.1$

[SII]
6730 $0.06 \pm.03$ $0.06 \pm.03$ $0.1\pm.1$ $0.2\pm.1$

E_v(H $\beta)$ Å 91 91 46 193 181

$\log F$ (H $\beta$ ) -12.14 -12.17 -12.45 -13.17 -13.06

$\sigma$ (l) $_{\lambda4363}$ 19 19 6.4 2.5 3.9

**Table 5:** Line intensities of the spatially-averaged spectra, normalized to the H $\beta$ emission line, of the H II regions studied in this investigation. The equivalent width and the flux in the H $\beta$ line (given in units of erg s^-1cm^-2), as well as the signal-to-noise ratio in the line [OIII] $\lambda$ 4363 Å, denoted by $\sigma$ (l), are also presented
line	$\lambda$ (Å)	IC 4662 A1	IC 4662 A2	IC 4662 D	ESO 245-G05 19	ESO 245-G05 12
[OII]	3727	$1.1\pm.2$	$1.1\pm.2$	$2.31\pm.03$	$2.9\pm.2$	$2.5\pm.1$
H12	3750		$.006\pm.01$
H10	3798	$0.017\pm.003$	$.017\pm.005$
HeI	3820	$0.02\pm.01$	$0.02\pm.01$
H9	3835	$0.03\pm.01$	$0.03\pm.02$
[NeIII]	3869	$0.40\pm.06$	$0.40\pm.02$	$0.3\pm.1$	$0.27\pm.01$	$0.27\pm.01$
HeI	3889	$0.15\pm.03$	$0.14\pm.04$	$0.18\pm.04$	$0.27\pm.01$	$0.149\pm.004$
CaII	3934	$0.010\pm.01$	$0.008\pm.005$
[NeIII]	3967	$0.21\pm.03$	$0.21\pm.03$	$0.36\pm.02$	$0.3\pm.1$	$0.20\pm.03$
HeI	4026	$0.013\pm.002$	$0.014\pm.002$
[SII]	4068	$0.006\pm.004$	$0.01\pm.02$
	+4076
H $\delta$	4102	$0.20\pm.01$	$0.204\pm.001$	$0.20\pm.01$		$0.21\pm.01$
H $\gamma$	4340	$0.35\pm.01$	$0.41\pm.01$	$0.47\pm.1$	$0.47\pm.01$	$0.5\pm.1$
[OIII]	4363	$0.07\pm.01$	$0.07\pm.02$	$0.09\pm.01$	$0.046\pm.005$	$0.09\pm.01$
HeI	4471	$0.04\pm.01$	$0.04\pm.01$	$0.10\pm.02$	$0.08\pm.01$	$0.038\pm.003$
[FeII]	4570	$0.007\pm.002$	$0.008\pm.005$
HeII	4650	$0.04\pm.01$	$.04\pm.02$
ArIV]	4711	$0.016\pm.001$	$0.015\pm.001$
ArIV]	4740	$0.007\pm.001$	$0.007\pm.002$
H $\beta$	4861	$1.00\pm.02$	$1.00\pm.02$	$1.00\pm.01$	$1.00\pm.1$	$1.00\pm.1$
HeI	4921	$0.008\pm.002$	$0.008\pm.001$
[OIII]	4959	$2.1\pm.1$	$2.1\pm.1$	$1.1\pm.1$	$0.82\pm.02$	$0.93\pm.04$
[OIII]	5007	$6.4\pm.2$	$6.3\pm.2$	$3\pm1$	$2.4\pm.1$	$2.8\pm.1$
[ClIII]	5517	$0.005\pm.005$	$0.012\pm.004$
[ClIII]	5530	$0.005\pm.001$	$0.011\pm.004$
HeI	5875	$0.11\pm.01$	$0.11\pm.01$	$0.13\pm.03$		$0.12\pm.01$
[OI]	6300	$0.004\pm0.003$	$0.005\pm.001$	$0.09\pm.02$		$0.023\pm.001$
[SIII]	6312	$.02\pm.01$	$0.03\pm.01$	$0.04\pm.01$		$0.010\pm.002$
[OI]	6363	$0.004\pm.001$	$0.004\pm.002$	$0.04\pm.01$		$0.010\pm.001$
[NII]	6548	$0.02\pm.01$	$0.02\pm.01$	$0.03\pm.01$		$0.023\pm.001$
H $\alpha$	6563	$2.86\pm.30$	$2.86\pm.20$	$2.79\pm.06$		$2.86\pm.10$
[NII]	6583	$0.05\pm.01$	$0.05\pm.01$	$0.07\pm.01$		$0.15\pm.01$
HeI	6678	$0.03\pm.01$	$0.03\pm.01$	$0.04\pm.01$		$0.043\pm.004$
[SII]	6716	$0.10\pm.05$	$0.09 \pm.04$	$0.2\pm.1$		$0.3\pm.1$
[SII]	6730	$0.06 \pm.03$	$0.06 \pm.03$	$0.1\pm.1$		$0.2\pm.1$

E_v(H $\beta)$ Å		91	91	46	193	181
$\log F$ (H $\beta$ )		-12.14	-12.17	-12.45	-13.17	-13.06
$\sigma$ (l) $_{\lambda4363}$		19	19	6.4	2.5	3.9

Table 6: The spatially-averaged parameters derived for the H II regions in IC 4662 and ESO 245-G05. The bottom row refers to a determination of the oxygen abundance with the use of the semi-empirical method (McGaugh 1994)
line IC 4662 A1 IC 4662 A2 IC 4662 D ESO 245-G05 nr. 19 ESO 245-G05 nr. 12

$T_{\rm e}$ (O⁺⁺)
$12\,000\pm500$ K $12\,000\pm1000$ K $17\,700\pm400$ K $14\,800\pm500$ K $19\,200\pm750$ K

$T_{\rm ion}$
$44\,000\pm4000$ K $44\,000\pm3000$ K $47\,000\pm5000$ K - $49\,000\pm3000$ K

12+log(O/H)
$8.17\pm.03$ $8.17\pm.05$ $7.7\pm.1$ $7.80\pm.01$ $7.59\pm.02$

ICF(N)
$7.8\pm.2$ $8\pm1$ $2.3\pm.9$ - $2.1\pm.2$

12+log(N/H)
$6.78\pm.07$ $6.66\pm.07$ $6.1\pm.3$ - $6.32\pm.05$

log(N/O)
$-1.50\pm.04$ $-1.51\pm.02$ $-1.58\pm.01$ - $-1.27\pm.01$

ICF(Ne)
$1.15\pm.01$ $1.15\pm.02$ $1.8\pm.9$ $2.11\pm.03$ $1.9\pm.1$

12+log(Ne/H)
$7.34\pm.02$ $7.35\pm.02$ $7.2\pm.3$ $7.28\pm.01$ $7.08\pm.03$

log (Ne/O)
$-0.83\pm.01$ $-0.83\pm.03$ $-0.5\pm.1$ $-0.52\pm.01$ $-0.52\pm.01$

N(He)/N(H)
$0.109\pm.004$ $0.11\pm.04$ $0.17\pm.03$ - $0.13\pm.01$

12+log(O/H) $_{\rm sem}$ 8.1 8.1 8 8.0 8.0

**Table 6:** The spatially-averaged parameters derived for the H II regions in IC 4662 and ESO 245-G05. The bottom row refers to a determination of the oxygen abundance with the use of the semi-empirical method (McGaugh 1994)
line	IC 4662 A1	IC 4662 A2	IC 4662 D	ESO 245-G05 nr. 19	ESO 245-G05 nr. 12
$T_{\rm e}$ (O⁺⁺)	$12\,000\pm500$ K	$12\,000\pm1000$ K	$17\,700\pm400$ K	$14\,800\pm500$ K	$19\,200\pm750$ K
$T_{\rm ion}$	$44\,000\pm4000$ K	$44\,000\pm3000$ K	$47\,000\pm5000$ K	-	$49\,000\pm3000$ K
12+log(O/H)	$8.17\pm.03$	$8.17\pm.05$	$7.7\pm.1$	$7.80\pm.01$	$7.59\pm.02$
ICF(N)	$7.8\pm.2$	$8\pm1$	$2.3\pm.9$	-	$2.1\pm.2$
12+log(N/H)	$6.78\pm.07$	$6.66\pm.07$	$6.1\pm.3$	-	$6.32\pm.05$
log(N/O)	$-1.50\pm.04$	$-1.51\pm.02$	$-1.58\pm.01$	-	$-1.27\pm.01$
ICF(Ne)	$1.15\pm.01$	$1.15\pm.02$	$1.8\pm.9$	$2.11\pm.03$	$1.9\pm.1$
12+log(Ne/H)	$7.34\pm.02$	$7.35\pm.02$	$7.2\pm.3$	$7.28\pm.01$	$7.08\pm.03$
log (Ne/O)	$-0.83\pm.01$	$-0.83\pm.03$	$-0.5\pm.1$	$-0.52\pm.01$	$-0.52\pm.01$
N(He)/N(H)	$0.109\pm.004$	$0.11\pm.04$	$0.17\pm.03$	-	$0.13\pm.01$
12+log(O/H) $_{\rm sem}$	8.1	8.1	8	8.0	8.0

5.1.2 ESO 245-G05

This object was studied by Miller (1996). The chemical abundances presented by Miller were derived with the use of the so-called semi-empirical method (e.g. McGaugh 1994). Despite the different approaches taken, the results reached by Miller (1996) are close to those obtained in this analysis (see Sect. 5.3) Moreover, comparable values of the flux in H $\beta$ are obtained in both investigations. One should note that Miller (1996) derived a smaller amount of extinction than derived in the present investigation. As evident from Table 3, a high extinction agrees with the color excess found for this galaxy.

5.2 Spectral characteristics and chemical abundances of IC 4662 A1/A2

As a consequence of the high signal-to-noise ratio of spectra of A1/A2, many lines are clearly detected. Comparing the columns corresponding to IC 4662 A1 and IC 4662 A2, both H II regions have almost identical intensities of the lines, as well as the equivalent width and flux in the H $\beta$ emission line. The main reason could be that they are physically connected, which is supported by the H $\alpha$ image. On the other hand, Heydari-Malayeri et al. (1990), obtained H $\beta$ images of the same regions and claimed that the two regions were clearly distinct. In order to compare the results from this investigation with that of Heydari-Malayeri et al. (1990), two H II regions will be considered: A1 to the northwest and A2 to the southeast.

Comparing the emission line ratios derived in this investigation with those from models of photoionization by hot stars (Stasinska 1990), it is evident that any substantial contribution from shock-heating or non-thermal radiation can be ruled out.

The detection of the doublet [ClIII] $\lambda$ $\lambda$ 5517,5537 Å can be explained by photoionization only (Stasinska 1990). The ionic abundance of chlorine can be derived for A1/A2 in the same manner as for Hubble V and X in NGC 6822 (Paper I). Values of N(Cl⁺⁺)/N(H⁺)= 2.0 10^-8 and 4.8 10^-8 were obtained for A1/A2, respectively. These are lower than that of Hubble X but higher than the ionic abundance of chlorine derived for Hubble V. The ratio log(Cl⁺⁺/O⁺⁺) was determined for regions A1/A2 and were found to be -3.8 and -3.4, respectively, which are also intermediate between those of Hubble V and Hubble X.

One can also note the presence of the absorption line CaII $\lambda$ 3934 Å (as well as CaII $\lambda$ 3968 Å although blended with H $\epsilon$ ) in the spectrum of IC 4662 A1/A2. This indicates the presence of an older stellar population underlying the H II region. This in turn emphasizes the importance of correcting for underlying absorption in order to obtain the chemical abundances of the nebular component.

Finally, the line HeII $\lambda$ 4686 Å was detected at various positions along IC 4662 A1/A2. The origin of this line can be nebular, indicating a very hot stellar population, but can also be due to WR stars (Aller 1984). In the latter case one would expect the presence of the broad spectral feature around $\lambda$ 4650 Å, and also another broad feature in the spectral region between 5700-5800 Å, often referred to as the red bump. In A1/A2, in both the spatially-averaged spectra, as well as the 3r-spectra, the equivalent width of the HeII line is small, 3 Å for the spatially-averaged spectrum of A1. The absence of the WR feature in the red spectral region, in combination with the narrowness of the HeII line, the WR origin for this line can be ruled out. Instead, a nebular origin is proposed. The presence of the nebular lines [ArIV] $\lambda$ $\lambda$ 4711,4740 Å supports the nebular origin. The detection of the lines over a large portion of the H II region, about 245 pc, also favours the nebular origin. Even though the spectral characteristics point to a nebular origin, the temperature of the ionizing radiation does not. For both regions the metallicity as well as the $T_{\rm ion}$ is very similar, 43700 K and 44700 K, respectively. The $T_{\rm ion}$ is determined from a single-burst model of spectral evolution (Olofsson 1997), These temperatures are too low to ionize helium twice, which in turn excludes a nebular origin. The situation is obviously somewhat confusing.

5.2.1 The interesting case of IC 4662 D

As evident from Table 6, the chemical abundances derived for IC 4662 D are substantially different from those of regions A1/A2. The largest deviations concern the nitrogen and oxygen abundances, 0.6 and 0.5 dex, respectively. The abundance of neon is similar, within the uncertainties. The difference in the abundances between regions A1/A2 and D cannot be explained only by uncertainties in the measurements.

One possible explanation for these differences could be the extinction correction. As mentioned in Sect. 4.2, the intensities of the lines of IC 4662 D were corrected using the H $\gamma$ line instead of the H $\alpha$ line. In order to determine if the correction performed is erroneous, the following approach was taken. A new set of extinction corrected intensities were obtained with the $C_\beta$ value from spectra obtained with grism nr. 2 (see Sect. 2). The result is that, despite the fact that the extinction coefficients are very different, the resulting oxygen abundances are the same. The difference in the nitrogen abundance using this $C_\beta$ value is larger. The same is true for the neon abundance. Considering the very anomalous values of the log(Ne/O) ratio and the very low log(N/O) ratio obtained with the new set of line intensities, the H $\gamma$ $\slash$ H $\beta$ coefficient was used to derive the line intensities. To conclude, the difference in the abundances, between the regions A1/A2 and region D, cannot be an artifact of the reduction procedure.

Region D appears, in both the spectrum and the H $\alpha$ image, as a two-component region, with a prominent stellar continuum at the southern part of the object. A possible explanation for the difference found in the metallicity could be the large distance, about 780 pc, of region D to the other four components of the galaxy, measured from the nearest edges of regions A1 and D. Moreover, this region seems not to be embedded in a diffuse envelope of gas surrounding the rest of the components (Heydari-Malayeri et al. 1990). These facts, in combination with the small difference in the radial velocity (200 kms^-1), indicate that this region does not belong to IC 4662. Instead it may be a compact, bright object residing slightly behind IC 4662, with two distinct regions of star formation. In order to test this hypothesis, detailed spectroscopy of all the four H II regions are needed to determine the velocity of each component in order to retrieve the dynamical state of the system. Other explanations for the difference in the abundances can also be invoked, such as galactic winds, dynamical disruptions, or a chemically poorly-mixed interstellar medium, due to its large distance from the other H II regions.

5.3 ESO 245-G05

As previously mentioned, it was not possible to use the H $\alpha$ line in the extinction correction for both regions (see Sect. 4). One can ask how much this flaw will influence the final chemical abundances. A similar procedure to that described for region IC 4662 D was performed here. For region nr. 19, a new set of line intensities was obtained corrected with the extinction coefficient obtained from the H $\alpha$ /H $\beta$ ratio. The total oxygen abundances obtained with this new set of data is very close to those presented in Table 6, with a difference of only 0.07 dex. For region nr. 12, a similar approach was taken, but no extinction coefficient could be obtained from the data using grism nr. 2, and a low value ( $C_{\beta}=0.03$ ) was found from the data using grism nr. 6. The latter value was preferred because no significant differences were found in the chemical abundances extracted.

Only the oxygen and neon abundances are compared in regions nr. 12 and nr. 19 in ESO 245-G05. This is due to the absence of the red spectral region for the latter region. For both elements, differences are found which is in agreement with the oxygen abundances obtained by Miller (1996). There is also a difference in the electron temperature, which could reflect a variation in oxygen abundance. The abundance variations could be a result of different star formation histories between regions nr. 12 and nr. 19. This was studied using Cerviño & Mas-Hesse (1994), where the age of the last star-forming event is obtained from the $T_{\rm ion}$ and the equivalent width of the H $\beta$ emission line, $E_v(H\beta )$ . The values of these parameters, extracted for region nr. 12, indicate an event younger than 3 Myr.

One of the many possible explanations for the difference in the chemical abundances could be that the last episode of star formation in this galaxy was initiated before the completion of the mixing of chemical elements.

6 Results on the spatial distribution in IC 4662

Due to the high quality of some of the data, a study of the excitation, the extension of the ionization zones and the spatial distribution in the chemical abundances within the H II regions A1/A2, and D in IC 4662, could be carried out. ESO 245-G05 will not be considered in this section due to the poorer quality of data.

6.1 The ionization structure in IC 4662

The ionization structure of the regions of IC 4662 is presented in Figs. 4a and b, corresponding to regions A1/A2 and D, respectively. The high ionization zone corresponds to physical positions of high [OIII] $\lambda$ 5007/[OII] $\lambda$ 3727 emission line ratios. Correspondingly, regions of low ionization region are those with high [OII] $\lambda$ 3727/[OIII] $\lambda$ 5007 line ratios.

From Fig. 4a, corresponding to the region IC 4662 A1/A2, it is evident that the high ionization region has an extension of about 400 pc, which agrees well with the extension of the excitation region (see the subsequent section). The low and high ionization zones are not so distinctive for region D (Fig. 4b), which indicate a relatively older age of the last episode of star formation. This agrees with the small equivalent width of H $\beta$ (see Table 5).

6.2 The excitation structure of IC 4662

The excitation parameter, [OIII] $\lambda$ 5007/H $\beta$ , for regions A1/A2 and D of IC 4662 is presented in Figs. 5a and b. Despite the results found by Heydari-Malayeri et al. (1990) regarding the existence of two distinct H II regions A1/A2, no such distinction is evident studying the excitation parameter (Fig. 5a). Instead, only one peak, extending almost 500 pc, is clearly visible while the H $\alpha$ intensity is roughly constant throughout the entire region. The large spatial extension of the probable OB associations could indicate a situation more typical of a starburst phenomenon than that of a continuous star formation rate. This seems to be the case for dIs in general.

The excitation parameter seems to be largely constant in region D (Fig. 5b) although the H $\alpha$ /H $\beta$ ratio shows a minimum at 50 pc from the geometrical center of the H II region. This OB association is also extended, 200 pc, but the excitation is weaker than in region A1/A2. Comparing Fig. 5b with the H $\alpha$ image of this galaxy, one sees two small concentrations which may correspond to the two peaks in the H $\alpha$ /H $\beta$ ratio. These peaks are also noticeable when studying the ionization structure (Fig. 4b).

6.3 The distribution of chemical elements of IC 4662

Figures 6a and b present the spatial distribution of the signal-to-noise ratio determined from the 3r-spectra for regions A1/A2 and D. The signal-to-noise ratio differs by a factor of 10 in the most extreme case, comparing these H II regions. The signal-to-noise ratio of the [OIII] $\lambda$ 4363 Å line determined for regions A1/A2 (Fig. 6a) is considered to be sufficiently high for a reliable analysis, which is not the case for region D (Fig. 6b). It is concluded that the quality of the data of the latter is too poor for a study of possible variations in chemical abundances inside this H II region. In the following, only the locations where the signal-to-noise ratio is higher than $\log(\sigma)=0.9$ , based on the H $\alpha$ intensity, or $\log(\sigma)=0.5$ , based on the [OIII] $\lambda$ 4363 Å intensity, will be considered in the further analysis. The latter value is considered to be the lowest signal-to-noise ratio acceptable for a meaningful analysis (Rola & Pelat 1995).

6.3.1 Electron temperatures and oxygen abundances of IC 4662

The $T_{\rm e}$ over the face of A1/A2 in IC 4662 is presented in Fig. 7. The value obtained from the spatially-averaged spectra of the two regions is also visualized. One can note the largely uniform electron temperatures in the two regions.

Table 7: The maximum deviation in chemical abundances comparing the spatially-averaged spectra and the 3r-spectra for IC 4662 A1/A2. The first column is an order number; the spectrum denoted as number 1 corresponds to the first point to the left in Figs. 7 to 12 (although only rows 10 to 31 are used in the analysis). For the remaining columns, $\Delta$ is defined as $\Delta = { (X_{\rm ave} - X_{\rm 3r}) \slash ( \Delta_{\rm ave} + \Delta_{\rm 3r})}$ , where $X_{\rm ave}$ is the value of the quantity derived from the spatially-average spectra and $\Delta _{\rm ave}$ the corresponding uncertainty. $X_{\rm 3r}$ and $\Delta _{\rm 3r}$ are the corresponding values of the 3r-spectra. Empty spacings mean that the values derived from the two methods coincide
3r-spectrum $\Delta T$ $\Delta$ O $\Delta$ N $\Delta$ Ne $\Delta$ (N/O) $\Delta$ (Ne/O)

10
4900 K 0.45 0.43 0.18 0.19

11
0.18 0.09 0.17

12
0.09 0.05 0.03

13
0.03 0.1 0.04 0.13

14

15
0.06 0.06 0.22

16

17
0.02 0.04

18

19
19 0.04 0.05

20 350 K 0.17 0.18 0.05 0.09

21 0.04

22 0.05 0.01

23

24

25 0.07 0.01

26 0.1 0.17 0.14

27 0.1 0.07 0.2

29 650 K 0.1 0.1

30 0.14 0.15 0.11

31 0.03 0.04 0.13 0.09

**Table 7:** The maximum deviation in chemical abundances comparing the spatially-averaged spectra and the 3r-spectra for IC 4662 A1/A2. The first column is an order number; the spectrum denoted as number 1 corresponds to the first point to the left in Figs. 7 to 12 (although only rows 10 to 31 are used in the analysis). For the remaining columns, $\Delta$ is defined as $\Delta = { (X_{\rm ave} - X_{\rm 3r}) \slash ( \Delta_{\rm ave} + \Delta_{\rm 3r})}$ , where $X_{\rm ave}$ is the value of the quantity derived from the spatially-average spectra and $\Delta _{\rm ave}$ the corresponding uncertainty. $X_{\rm 3r}$ and $\Delta _{\rm 3r}$ are the corresponding values of the 3r-spectra. Empty spacings mean that the values derived from the two methods coincide
3r-spectrum	$\Delta T$	$\Delta$ O	$\Delta$ N	$\Delta$ Ne	$\Delta$ (N/O)	$\Delta$ (Ne/O)
10	4900 K	0.45	0.43	0.18		0.19
11		0.18	0.09			0.17
12				0.09	0.05	0.03
13		0.03	0.1	0.04	0.13
14
15		0.06		0.06	0.22
16
17				0.02		0.04
18
19	19	0.04				0.05
20	350 K	0.17	0.18	0.05		0.09
21			0.04
22			0.05	0.01
23
24
25		0.07				0.01
26		0.1	0.17			0.14
27		0.1	0.07	0.2
29	650 K	0.1	0.1
30		0.14	0.15			0.11
31		0.03	0.04	0.13		0.09

An average $T_{\rm e}$ of 12500 K was determined from the data of the 3r-spectra and agrees well with that from the spatially-averaged spectrum (12000 K). The $T_{\rm e}$ seems constant throughout the regions but some deviations in this quantity seem to be evident, especially at 150, 20, 0, and -40 to -60 pc. From consulting Table 7 one can conclude that, considering the high signal-to-noise ratio in this region, no significant variations in the $T_{\rm e}$ are present. The variation in the $T_{\rm e}$ detected could be real or may reflect small-scale temperature fluctuations (Peimbert 1967) or possible variations in the ionization parameter, U. A more thorough study regarding these parameters is necessary in order to unveil the origin of the variations in the $T_{\rm e}$ . This is beyond the scope of this paper.

It should be pointed out that the average oxygen abundance extracted from the 3r-spectra is slightly lower (8.10) than that derived from the spatially-averaged spectrum (8.17).

6.3.2 The distribution of helium, nitrogen and neon

By studying the distribution of helium it is obvious that the average value derived from the 3r-spectra (N(He)/N(H) = 0.10) is equal to the spatially-averaged helium abundance. Considering the high uncertainties associated with the helium abundance determination no conclusion regarding the possible variation of this element over the face of IC 4662 A1/A2 can be drawn.

The nitrogen distribution is presented in Fig. 9, (corresponding to spectra nr. 20, 26 and 29 in Table 7). A difference is found in the spatially-averaged abundances between regions A1/A2 (see Table 5). This is not particularly evident when the spatial distribution is studied. If restricted to the region of high signal-to-noise ratio only, the locations at 160, 20, and -40 pc (see Table 7) present a real difference in the abundance of nitrogen while a constant value can be inferred for the rest of the locations, considering the uncertainties involved. The nitrogen abundance determined from the 3r-spectra is slightly lower than that obtained from the spatially-averaged spectrum (0.13 dex).

The region around 0 pc (nr. 20 in Table 7) presents overabundances for all elements except nitrogen. One explanation can be the low signal-to-noise ratio in the emission line [OIII] $\lambda$ 4363 Å in this part of the spectrum. The region at 155 pc (nr. 27 in Table 7) reveals underabundances in all elements, and a possible explanation can be that this point marks the frontier between the two H II regions A1/A2. Another explanation for the abnormal values will be discussed in the next section.

7 Chemically homogeneous or non-homogeneous galaxies?

7.1 Small-scale variations

When small spatial scales are considered, the main conclusion from this investigation is that the abundances are nearly constant throughout the face of A1/A2 in IC 4662, except perhaps at a few locations. The spectral lines [FeII] $\lambda$ 4570 Å, HeII $\lambda$ 4685 Å, and [ArIV] $\lambda$ $\lambda$ 4711,4740 Å are clearly present at those locations where variations in some elements are found, as discussed in the previous section. In Sect. 5.2, it was concluded that the origin of the HeII $\lambda$ 4686 Å line is not completely understood. If this line is due to WR stars or SN type II, the underabundance in helium at -22 pc can be explained as a result of sweeping of chemical elements by winds. Traces of chemical enrichment may be observed at the locations at -44 and 265 pc. A weak point in this argument is the non-symmetric distribution of the nebular edges, with the northeast very close by and the southwest further away. This could only be explained by a lower electron density ( $N_{\rm e}$ ) in the interstellar medium along the northeast direction, as Recchi et al. (2000) have shown. Another problem is the size of the region where the possible WR spectral features are detected. The total size is 245 pc, which is almost one fifth of the total size of A1/A2. From the simulation performed by Recchi et al. (2000), the edge of the wave of an event of a SN explosion will travel a distance of nearly 1.5 kpc, given a luminosity of $L_{\rm b} = 1.5~10^{36}$ ergs^-1, where $L_{\rm b} = 17.9 ~\rho_{\rm o}$ $H_{\rm eff}$ ² $C_{\rm o}$ ³. $\rho$ $_{\rm o}$ is the central density, $C_{\rm o}$ the speed of sound and $H_{\rm eff}$ the effective scale length of the ISM. Supposing that such a wave is detected in the data presented here, the distance from the event is only 125 pc (for a central event), and a $L_{\rm b}$ of 1.2 10³⁵ erg s^-1 is required, Using typical values of these quantities of 100 cm^-3, 4 km s^-1and 100 pc, respectively, a $L_{\rm b}$ of 1.14 10³⁹ erg s^-1 is obtained, which is higher than that required. In principle, two SN events could be the explanation for the "double-bowl shape'' of the distribution of those elements, from 300 to 0 pc; one at 22 pc, the other at 155 pc. The helium abundance was found to be largely constant throughout the H II regions. It should be emphasized that the contribution of singly-ionized helium was not considered in the derivation of the total abundances.

A final point concerning the helium and nitrogen distribution is the expected enrichment of these elements when the WR/SN features are observed. As mentioned, no significant enhancement of helium is found in the locations where the HeII $\lambda$ 4686 Å line is detected and the abundance of nitrogen is constant in the same locations or, perhaps, somewhat depleted. The results presented here are consistent with the suggestion that helium and nitrogen emerging from the present epoch of star formation are only detectable in X-rays due the high temperatures involved (Kobulnicky et al. 1997).

7.2 Large scale variations in chemical abundances

The discussion of possible inhomogeneities in chemical abundances of the ISM within dwarf galaxies has been considered for many years. Part of the data support the idea of a homogeneous chemical composition which is in line with a closed-box model (Tinsley 1980). This model seems to behave well for giant, dynamically stable elliptical galaxies. The situation for dwarf galaxies may be different because of their small gravitational potentials, which could make the galaxy partially unstable to strong interstellar winds or external gravitational influences. Selective loss of chemical elements through winds has been proposed in order to explain the chemical abundances observed in dwarf galaxies (Pilyugin 1992; Matteucci & Tosi 1985). The situation is unclear since the results of the modeling vary from those where chemically enriched gas remain in the galaxy, to those where the metals are partially ejected from the galaxy in a selective manner (D'Ercole & Brighenti 1999).

The typical model of chemical evolution for dwarf irregular galaxies contains self-enrichment (total or partial) of the ISM due to previous events of star formation. These events are spaced out to a few Gyr. During the quiescent phases, elements ejected will cool until reaching the temperature of the ISM and is assumed to mix with the rest of the ISM. The mixing is supposed to be homogeneous and efficient throughout the galaxy by any of the mixing processes proposed: e.g., epicyclic or radial mixing, superbubble expansion (Roy & Kunth 1995). These processes would erase any abundance variations on timescales less than 10⁹ years. Under such circumstances, a new event of star formation can occur when the mixing of elements is completed and the chemical composition throughout the ISM would remain the same. I Zw 18 (Izotov et al. 1999) and NGC 4214 (Kobulnicky & Skillman 1996) seem to be good examples of chemically homogeneous galaxies. However, a few other galaxies do not show a homogeneous chemical composition, e.g. WLM (Hodge & Miller 1995) or IC 10 (Lequeux et al. 1979)

Three dI galaxies have been studied in this series of articles in order to increase our knowledge of this subject. As is evident, no unique answer to whether dIs are homogeneous or not has been obtained. It seems logical that the mechanisms acting in favour of mixing the ISM does not work with the same efficiency in all cases. Actually, some of the mechanics more efficient in the mixing are believed not to be operating in dwarf galaxies. This could simply be a consequence of the physical conditions and morphology of each particular galaxy, or region of a galaxy. Each of the three dIs studied, the two presented in this article and NGC 6822 (Paper I) should be considered independently.

The study of NGC 6822 (Paper I) showed almost no variations in chemical composition between the H II regions, while for the other two, IC 4662 and ESO 245-G05, some deviations are found. As discussed in Paper I, the two H II regions of NGC 6822, Hubble V and Hubble X, are situated in the northern part of the bar. The small distance between the H II regions probably acts in favour of the minor variations in the chemical abundances.

ESO 245-G05 is an exceptional case. All the H II complexes except one, which include regions nr. 5, 6 and 7 in Miller (1996), are located along the bar of the galaxy. It is well known that, as the ellipticity of a bar increases, the smaller the difference in the abundances (Martin & Roy 1994). External gravitational interactions could play a role. A number of explanations could be proposed for the variations in the abundances, e.g. shorter events of star formation, which would act to inhibit complete mixing of the elements. This would especially be the case for large galaxies such as ESO 245-G05. Infall of primordial gas or interaction with HI clouds could be the main processes triggering star formation before the completion of the mixing of the elements. If continuous star formation is invoked for this kind of galaxy (Legrand 1999), the explanation is more reasonable since the mixing timescale would be longer.

IC 4662 is the most striking galaxy in this sample. As a result of the morphology (small, non-barred galaxy), its physical environment (isolated) and the fact that IC 4662 is dominated by a powerful event of star formation, a homogeneous distribution in chemical abundance would be expected for this galaxy. In agreement with this, no differences have been found between the H II regions A1/A2. The chemical homogeneity of A1/A2 could be a consequence of the proximity of the regions. Actually, no clear distinction between these regions could be found studying two-dimensional spectra and one should perhaps regard A1/A2 as a single H II region. However, region D shows completely different chemical abundances, compared to A1/A2, for all elements studied. The large distance between region D and the main body of the galaxy, and the fact that this region seems to be located outside the HI cloud (Heydari-Malayeri et al. 1990), could be the explanation for the different chemical abundances. The homogenization mechanisms which work at large scales are less efficient for dIs due to longer timescales for the mixing of the gas. For the rest of the galaxies with high quality spectra, no inhomogeneities have been detected, except for NGC 5253 (Kobulnicky et al. 1997). Some of these galaxies are larger than 1 kpc, e.g. NGC 4214, which exceeds the distance between region D and the rest of the regions. However, in all these cases there seems to be a dynamical connection between all the H II regions studied, which is not the case for IC 4662.

Finally, two galaxies in this investigation (NGC 6822 and ESO 245-G05) have gravitational companions (consult Hidalgo-Gámez & Olofsson 1998 for their definition of gravitational companion), which could affect the star formation history, as well as the gaseous mixing processes. The largest galaxy in the sample is ESO 245-G05 and the smallest is NGC 6822. The results presented in this investigation support the idea that the larger the galaxy, the more inhomogeneous its chemical composition. Moreover, the H II region which presents the largest differences in chemical composition in IC 4662 is A1/A2, which is the one most distant from the main body of the galaxy.

8 Conclusions

A detailed analysis regarding chemical abundances of three galaxies have been carried out in this series of articles. For two of the galaxies, NGC 6822 studied in Paper I and IC 4662, the quality of spectra were sufficiently high for an analysis of the spatial distribution of some important parameters. The main results concerning the spatial distribution for these objects are: the H II region in IC 4662 A1/A2 may be experiencing SN explosions or powerful WR stellar winds. This could be the reason for the differences found in the distribution the chemical elements across the face of the object. An investigation regarding chemical abundances between different H II regions gave the following results. NGC 6822 shows no significant difference in the abundances between the two regions Hubble V and Hubble X, while important variations in all elements were found in IC 4662 and ESO 245-G05.

The chemical abundances in a sample of dwarf irregular galaxies

II. The case of IC 4662 and ESO 245-G05

1 Introduction