Evidence for a warm interstellar medium in the Fornax dwarf ellipticals FCC046 and FCC207

Abstract

We present Hα+[N ii] narrow-band imaging of FCC046 and FCC207, two dwarf ellipticals (dES) in the Fornax cluster. B—R and B—I colour maps clearly show signs of dust absorption in FCC207. FCC046 has a very bright blue nucleus, offset by about 1.1 arcsec with respect to the outer isophotes. FCC046 was classified as a non-nucleated dE4 so the presence of its nucleus came as a surprise. Moreover, FCC046 shows a pronounced lopsided shape. Given that FCC046 is an isolated galaxy, it is unlikely that an encounter caused this asymmetry. The emitting regions differ considerably between the two galaxies. Whereas FCC207 has only one central emission region, FCC046 also contains fainter emission regions. Based on broad-band colours, its disturbed shape and its very bright nucleus, FCC046 is akin to the class of amorphous dwarfs. The central emission regions of both galaxies are barely resolved under seeing conditions of FWHM ≈0.8 arcsec and we estimate their diameters at about 60 pc. Their Hα luminosities can be explained as photoionization by post asymptotic giant branch stars in an old population. Some of the fainter extended emission regions in FCC046 are resolved and have diameters of the order of 50–150 pc and Hα luminosities of the order of 10³⁰ W, comparable to supernova remnants or nebulae around Wolf—Rayet stars. Hence, FCC046 is clearly undergoing star formation while for FCC207 the case is not as clear-cut. We estimate the mass of the H ii gas in FCC046 at M_Hii= 40–150 M_⊙ (for T= 10⁴ K, N_e= 1000 cm⁻³). The ionized-gas content of FCC207 is somewhat higher: M_Hii= 60–190 M_⊙.

Hii regions, supernova remnants, galaxies: dwarf, galaxies: individual: FCC046, galaxies: individual: FCC207

1 Introduction

Dwarf ellipticals (dEs) as a rule are pressure-supported objects, characterized by very low rotation velocities compared to their velocity dispersions; fast-rotating dEs do exist but they are rare, see De Rijcke et al. (2001). There are currently a number of models in vogue that attempt to explain this apparent lack of rotation as a result of significant mass loss. According to the ‘wind model’, proposed by Dekel & Silk (1986), dEs form from average-amplitude density fluctuations. Most, if not all, of the interstellar medium (ISM) is subsequently blown away after it has been heated to velocities that exceed the galaxy's escape velocity by the first burst of supernova explosions. This dramatic mass loss causes a more anisotropic orbital structure and makes the galaxy puff up. A more sophisticated version of this scenario can be found in Mori et al. (1997) who discuss the chemodynamical evolution of a 10¹⁰ M_⊙ dwarf galaxy. The first supernovae expel a supersonic outflow of gas from the centre of the galaxy. Stars form in this expanding shell and subsequent supernova explosions further accelerate the expansion of the shell and enrich it with metals. This model explains the outward reddening of dEs as a metallicity effect and reflects in the characteristic exponential surface-brightness profile.

Other scenarios take into account the fact that dEs are found predominantly in high-density environments such as groups and clusters. Mori & Burkert (2000) argue that ram-pressure stripping is able to completely remove the gas from a dE less massive than 10⁹ M_⊙ within a few 10⁸ yr. More massive dEs might be able to retain some gas in the central region. Moore, Lake & Katz (1998) examine the role of tidal interactions between small spirals and giant cluster members to produce dE-like objects. On its orbit through the cluster, a small galaxy is subjected to collisional shocks induced by other galaxies that tear large tidal tails off it, draining angular momentum from the remaining gas and stars. The perturbations heat the dwarf galaxy, raising the velocity dispersion, while the mass loss makes it inflate. Torques exerted by the collisions drive gas to the centre where it is consumed in a starburst. This process could explain the density spike in so-called nucleated dEs.

Independent of which scenario is correct, we would expect that not much of an ISM, if any, is present in dEs and hence that they are not actively forming new stars. None the less, there is a growing amount of data that demonstrates that at least some dEs have been able to retain a sizable amount of dust and both cold and warm gas, and that some are forming stars. We give a few examples of recent detections of star formation, or of an ISM in dEs, without attempting completeness. Young & Lo (1997a,b,c) have detected H i 21-cm emission and CO emission in virtually all Local Group dwarf spheroidals they have examined (Leo A, NGC 147, NGC 185, NGC 205, Sag DIG, LGS 3 and Phoenix). NGC 185 shows Hα emission in the form of a central extended emission region of 50 pc in diameter, probably a supernova remnant (SNR). On the other hand, NGC 205 is devoid of emitting regions. Hence, the presence of an ionized ISM in dEs should not be taken for granted. NGC 205 and 185 also show a few dust patches. The dE A 0951+68, in the M81 group, possesses a high-excitation H ii region (Johnson et al. 1997). The observed extended blue light is regarded as evidence of a recent star-formation event. NGC 4486A, a relatively bright dE seen almost edge-on, contains a stellar and dust disc (Kormendy et al. 2001), reminiscent of the nuclear discs of spiral galaxies.

Sandage & Brucato (1979)— see also, for example, Quill, Ramirez & Frogel (1995), Noreau & Kronberg (1986), Marlowe, Meurer & Heckman (1997, 1999)— have coined the name ‘amorphous dwarfs’ for dwarf galaxies that have a disturbed appearance due to recent star formation and the presence of dust but are not irregular enough to be classified as Im (Magellanic Cloud type irregular). Marlowe et al. (1999) argue that blue compact dwarfs (BCDs), H ii galaxies and amorphous galaxies are actually all members of the same class of star-forming dwarfs and owe their names mostly to the selection criteria involved. Most amorphous dwarfs in the sample of Marlowe et al. (1997) have a two-component surface brightness profile: an exponential envelope and a bluer core component. Their amorphous dwarfs show strong Hα emission (L_Hα≈ 10³³ W). These authors argue that it is possible — at least in principle — that the cores and envelopes of BCDs and amorphous dwarfs will fade and reach an end state similar to present-day nucleated dEs after they have used up their gas supply and star formation has ended. However, star formation in dwarf galaxies probably takes place in a series of mild starbursts that deplete the gas rather slowly. Hence, dEs must have had ancestors that evolved more rapidly. The fact that dEs are found predominantly in clusters while BCDs are remarkably scarce in these high-density environments (Salzer 1989) might hold a clue; repeated encounters with giant galaxies and ram-pressure stripping may have sped up the gas-depletion process. It is clear that the present-day star-formation rate (SFR) and ISM content hold important clues to understanding the origin of dEs.

FCC046 (Fig. 1) and FCC207 (Fig. 2) (Ferguson 1989) were selected as targets for an ongoing ESO Large Programme to study the structure and dynamics of dEs. These were the only galaxies in our sample with published evidence of the presence of an ISM and recent star formation. Ionized hydrogen was detected by Drinkwater et al. (2001) in both galaxies. These authors interpret this as photoionization by young stars and they use the calibration of Kennicutt (1983, 1992) between the total SFR and the Hα+[N ii] equivalent width (EW)  

1

where L_B and L_B,⊙ are the B-band luminosity of the galaxy and the sun, respectively, to estimate the SFRs in these galaxies at 1–2 × 10⁻³ M_⊙ yr⁻¹. Held & Mould (1994) present UBV colours and metallicities of, amongst others, FCC207. They conclude that FCC207 is too blue in U—B (U—B= 0.15) and too metal-poor for its B—V (B—V= 0.78) and they interpret this as a consequence of the presence of a young stellar population. This motivated us to investigate both objects more closely using BRI broad-band and Hα+[N ii] narrow-band imaging. In Section 2, we discuss the details of the observations and data reduction. The B—R colour maps are presented in Section 3 and the results of the Hα+[N ii] narrow-band imaging are shown in Section 4.

Figure 1

450-s B-image of the dE4 FCC046. The nucleus is offset by 1.1 arcsec to the south-west of the centre of the outer isophotes.

Figure 2

450-s B-image of the dE2 FCC207.

2 Observations and Data Reduction

The observations were carried out on 2001 November 18 and 20 with Yepun (VLT-UT4) using FORS2. We took 20-min exposures of FCC046 and FCC207 with the Hα/2500+60 filter centred on 6604 Å and with a FWHM = 64 Å. R-band images obtained during a previous run (2000 November 1–8) served as off-band images. Two Hα images of the spectrophotometric standard star LTT9239 were taken for flux calibration. During these observations, the seeing (determined from the stars on the images) typically was 0.7–0.8 arcsec FWHM. The standard data reduction procedures (bias subtraction, flat-fielding, cosmic removal, interpolation over bad pixels, sky subtraction) were performed with the Munich Image Data Analysis System (MIDAS).¹ All science images were corrected for atmospheric extinction (using the R-band extinction coefficient, k_e= 0.13) and interstellar extinction (we used the Galactic extinction estimates from Schlegel, Finkbeiner & Davis (1998): A_R= 0.050 for FCC046 and A_R= 0.039 for FCC207). The images were finally converted to units of electrons/second/pixel.

In order to find the correct scaling for the R-band images we adopted the following strategy. The pure emission ‘Em’ can be recovered from a narrow-band image ‘Nb’ and an R-band image ‘Rb’ as  

2

where c is the proper scaling constant and δ is a correction for possible faulty sky-subtraction. To find the best values for c and δ, we first fitted the isophotes of the narrow-band and R-band images in an annulus between m_R= 24.5 and m_R= 26.5 mag arcsec⁻², which in retrospect did not contain any emission (hence Em = 0), using the standard MIDAS FIT/ELL3 command. Thus, a smooth version of this annulus could be constructed for both images. The optimal values of c and δ can be found by minimizing the expression | Nb − (c× Rb +δ)| with Nb and Rb being the smoothed versions of the annulus. With these values in hand, the pure-emission image can be obtained using relation (2). δ was very small for both FCC046 and FCC207, which makes us confident that the sky was properly subtracted in all images. Since the Hα and R-band overlap, subtracting an R-band image in lieu of a continuum image entails a partial removal of some Hα+[N ii] light. The error thus introduced is of the order of the ratio of the effective widths of the filters (R-band, W= 165.0 nm; Hα, W= 6.4 nm), i.e. less than 4 per cent. Since this effect is negligible in comparison to other possible sources of error, we did not correct for it.

A pixel value in the pure-emission image (corrected for both atmospheric and interstellar extinction), denoted by N_g, expressed in electrons per second, can be converted to flux units, F′_g, using the formula  

3

Here, is the spectrum of a flux-calibration standard star and N_* is the measured flux of that star, expressed in electrons per second. ϕ_f(λ) is the transmission of the Hα filter and ϕ_o is the transmission of the optics (which is basically constant for a narrow-band filter). The prime on F′_g indicates that this is the flux incident on the CCD, after going through the telescope and instrument optics and the narrow-band filter. This can also be written as  

4

where F_Hα, and and F_{[N ii}]₂ are the incoming fluxes — i.e. before going through the telescope and instrument optics and the narrow-band filter — of, respectively, the Hα 6563 Å, the [N ii] 6548 Å and the [N ii] 6583 Å emission lines (approximated as δ-functions). This allows us to obtain the true incoming flux of the Hα emission line as  

5

The total incoming Hα+[N ii] flux is simply  

6

Since the Hα filter is relatively flat-topped and the Hα and [N ii] lines are well inside the filter transmission curve, the total flux is rather insensitive to the adopted relative line-strengths. In the following, we assume that the mean value for the ratio of the line-strengths of the two nitrogen lines (Macchetto et al. 1996; Phillips et al. 1986). The ratio is not known and is treated as a free parameter, varying between 0 and 2.

The rms scatter in the final pure-emission images is about 0.035 electrons/pixel/second (or 2.5 × 10⁻²⁰ W m⁻² for , the average value found by Phillips et al. (1986) for a sample of normal ellipticals).

3 B—R COLOUR MAPS

The B, R and I images were used to extract surface brightness, position angle (PA) and ellipticity profiles (see Fig. 3). The deviations of the isophotes from a pure elliptic shape were quantified by expanding the intensity variation along an isophotal ellipse in a fourth-order Fourier series with coefficients S₄, S₃, C₄ and C₃:  

7

All photometric parameters were fitted by cubic splines as functions of the semimajor axis distance. The galaxy nucleus (i.e. the brightest pixel) was used as zero-point for both a and b, the semiminor axis distance. This allowed us to reconstruct the surface brightness at a given point on the sky and to construct colour profiles (e.g. B—R as a function of radius).

Figure 3

Photometric properties of FCC207 (left) and FCC046 (right), derived from the I-band image, versus the geometric mean of the semimajor and semiminor distances a and b. From top to bottom: the I-band surface brightness μ_I (the dotted line corresponds to a surface brightness equal to 1 per cent of the sky level); the deviation in declination Δδ and right ascension Δα of the centres of the isophotes with respect to the brightest point; the PA; the ellipticity ε= 10(1 −b/a); and the Fourier coefficients S₄, S₃, C₄ and C₃ which quantify the deviations of the isophotes from ellipses.

3.1 FCC207

FCC207 has de-reddened magnitudes m_I= 14.39, m_R= 14.86 and m_B= 16.19 mag (hence B—R= 1.33 and R—I= 0.47 mag). Its nucleus has a distorted shape: it is more elongated than the bulk of the galaxy (E3 versus E2) and is somewhat kidney-shaped. This is probably due to dust absorption to the north of the nucleus, noticeable in the B—R colour map (Fig. 4) as a patch that is ≈0.2 mag redder than its surroundings. The nucleus (B—R= 0.90 mag) is significantly bluer than the bulk of the galaxy (B—R= 1.25 mag). This behaviour is similar to what, for example, Bremnes, Binggeli & Prugniel (1998) find in dwarf galaxies in nearby groups. A small, slightly east—west elongated blue object (B—R= 1.10 mag) can be seen to the west of the nucleus. It is also visible in the Hα image. Its elongation rules out the possibility that it is a faint foreground star. As can be seen in Fig. 5, the B—R, B—I and R—I colours remain essentially constant outside the nucleus. If a young stellar population is present outside the nuclear region of FCC207 (the inner 2 arcsec) then these stars are apparently well mixed with the older population.

Figure 4

B—R colour map of FCC207. The nucleus is rather blue (B—R≈ 0.9 mag) compared to the bulk of the galaxy (B—R≈ 1.25 mag). The inset shows the central 5 × 5 arcsec² region with a different grey-scale. To the north of the nucleus, a signature of dust absorption is visible (Δ(B−R) = 0.2 mag).

Figure 5

B—I, B—R and R—I profiles of FCC207 as a function of the geometric mean of semimajor axis a and semiminor axis b distance. Outside the nucleus, the colours are essentially constant.

3.2 FCC046

FCC046 is a rather blue object, with de-reddened magnitudesm_I= 14.43, m_R= 14.88 and m_B= 15.99 mag (hence B—R= 1.11 and R—I= 0.45 mag). The nucleus, a round (E0) and blue (B—R= 0.10 mag) object (see Fig. 6), is offset by 1.1 arcsec to the south-west of the centre of the outer isophotes (see Fig. 3). B—R, B—I and R—I colour profiles are presented in Fig. 7 and show a very different behaviour than those of FCC207. The colours of the stellar population become redder towards larger radii. The nucleus of FCC046 is much bluer than those of nucleated dwarfs presented by, for example, Bremnes et al. (1998). These authors typically find B—R≈ 0.5 for the nucleus. The nucleus is resolved in the B-band image. This implies that the nucleus is much larger than would be expected for a typical dE. Even with the superior resolving power of the Hubble Space Telescope (HST), Lauer et al. (1995) could not resolve the nuclei of five nucleated Virgo dEs. The diameter (FWHM) of the nucleus was estimated using the relation  

8

where FWHM_true is the true dimension, FWHM_obs is the observed FWHM and FWHM_star is the average FWHM of the stars in the image. The seeing, estimated from 10 stars in the B-band image, was 0.82 ± 0.04 arcsec. The measured FWHM of the nucleus is FWHM_obs= 1.1 arcsec or FWHM_true≈ 65 pc (for H₀= 75 km s⁻¹ Mpc⁻¹ and a Fornax systemic velocity v_sys= 1379 km s⁻¹). We fitted a two-component model to the B-band surface brightness of FCC046: an axisymmetric component centred on the outer isophotes, which represents the light of the underlying stellar population, and a round component centred on the position of the nucleus. The results of this decomposition are presented in Fig. 8. The nucleus has a blue magnitude m_B= 18.55 mag (M_B=−12.77) and comprises about 10 per cent of the total B-band luminosity of the galaxy. It should be noted that the nucleus of FCC046 was apparently not visible on the photographic plates on which the catalogue of Ferguson (1989) was based, since it is classified as a dE4 (i.e. as a non-nucleated dwarf). The underlying stellar envelope deviates from an axisymmetric mass model and shows a pronounced lopsidedness, visible in Fig. 3 as the bump in Δα in the region . This asymmetry may be due to an asymmetric distribution of few but bright young stars. This appears to be plausible since the dynamic time-scale, estimated as  

9

for typical values r≈ 0.5 kpc and M(r) ≈ 10⁹ M_⊙, is of the order of the lifetime of the youngest stars, so these would not have had time to disperse all over the face of the galaxy. The cause of persistent m= 1 perturbations, which involve a sizable fraction of a galaxy's mass, is still poorly understood. Interactions are often invoked, especially in bright galaxies, but examples of isolated lopsided galaxies are known; particularly in H i, see Baldwin, Lynden-Bell & Sancisi (1980). Since there is no galaxy detected within a 20 × 20 arcmin² square centred on FCC046, it seems unlikely that an encounter with another galaxy has caused the lopsidedness. Dynamical instabilities have also been invoked — see Merritt (1999) and references therein — but it remains unclear whether such a hypothesis may work for all galaxies.

Figure 6

B—R colour map of FCC046. White is blue, black is red. Obviously, the off-centre nucleus is very blue (B—R≈ 0.1 mag). The inset shows the central 5 × 5 arcsec² region with a different grey-scale. The asterisks mark the positions of the brightest Hα features (see Section 4.3).

Figure 7

B—I, B—R and R—I profiles of FCC046 as a function of the geometric mean of semimajor axis a and semiminor axis b distance. The nucleus is clearly much bluer than the envelope. The colours of the underlying stellar population become redder towards larger radii.

Figure 8

Two-component fit to the surface brightness of FCC046. The galaxy is presented with the origin in the bright nucleus and the x-axis along the major axis. The top left panel shows isophotes of the two-component model (grey) and the observed surface brightness (black). The top right panel shows isophotes of the nucleus alone (grey) and the lower right panel shows isophotes of the axisymmetric envelope alone (grey). In all these panels, the same isophotes are plotted. The lower left panel shows the residue between the model and the data (black is positive, grey is negative). Apart from the off-centre nucleus, FCC046 is obviously also lopsided.

4 Hα IMAGING

4.1 The Hα equivalent width

Drinkwater et al. (2001) have measured Hα EWs of 108 confirmed Fornax cluster members, including FCC046 and FCC207 with the FLAIR-II spectrograph on the UK Schmidt Telescope. The effective aperture diameter of this system is at least 6.7 arcsec (the fibre diameter) and could be as large as 15 arcsec (because of image movements due to tracking errors and differential atmospheric refraction). They find  

For comparison, we calculated the EW inside some aperture radius r from our images as  

10

where Δλ= 64 Å is the FWHM of the redshifted Hα filter and F_em (r) and F_cont (r) are the total number of counts inside a circular aperture with radius r of the Hα+[N ii] and the continuum image, respectively. We find  

Given the possible sources of error (photon shot-noise, sky and continuum subtraction) that can affect our measurements, we consider these values in good agreement with the EWs measured by Drinkwater et al. (2001).

4.2 The Hα+[N ii] and Hα luminosities

Pure Hα+[N ii] emission images of FCC046 and FCC207 are presented in Figs 9 and 10. For FCC046, we find F_em (FCC046) = 1.53–1.57 × 10⁻¹⁸ W m⁻², corresponding to a total luminosity L_em (FCC046) = 6.21–6.37 h⁻²₇₅× 10³⁰ W. The range of values is given for /F_Hα= 0–2 (see Fig. 11). The central emission peak comprises about half of the luminosity. It alone has a luminosity of about 3 × 10³⁰ W. The total flux of FCC207 is somewhat higher, F_em (FCC207) = 1.93–2.18 × 10⁻¹⁸ W m⁻², which yields a total luminosity L_em (FCC207) = 7.83–8.84 h⁻²₇₅× 10³⁰ W. These numbers can be compared to those found by Buson et al. (1993), Kim (1989), Phillips et al. (1986) and Shields (1991) for normal elliptical and S0 galaxies. The luminosities of the central emission peaks in FCC046 and FCC207 are compared to those of ellipticals in Fig. 12. Typical emission luminosities for these galaxies lie in the range L_em= 10³³–10³⁵ W, i.e. more than a 1000 times brighter. The fact that the luminosity of the nuclear emission in these dEs agrees fairly well with the trend of normal Es — extrapolated over more than 2 mag — suggests that the ionizing mechanism, at least for the central emission, is the same and therefore somehow related to the stellar population.

Figure 9

The pure-emission image (Hα+[N ii]) of FCC046. The asterisk marks the centre of the outer isophotes. The bright emission feature in the centre coincides with the off-centre nucleus. The six fainter emission ‘clouds’ are labelled Cl1–Cl6.

Figure 10

The pure-emission image (Hα+[N ii]) of FCC207. The asterisk marks the centre of the outer isophotes. A small emission feature can be discerned 2 arcsec to the west of the nucleus.

Figure 11

The logarithm of the total Hα+[N ii] flux (F_em) and the Hα flux (F_Hα) versus the ratio of the strengths of the [N ii] 8584 Å and the Hα line. The total flux is virtually independent of this line ratio.

Figure 12

The total Hα+[N ii] emission-line luminosity of FCC046 and FCC207 versus absolute blue magnitude. The dark-grey lines indicate the linear relation and its 1 −σ deviation observed by Phillips et al. (1986). The Es and S0s observed by Buson et al. (1993) fill the light-grey area. All observations have been converted to the distance scale adopted in this paper.

The total Hα flux of FCC046 is F_Hα= 4.17–15.7 × 10⁻¹⁹ W m⁻², depending on the value of . This translates into a total Hα luminosity L_Hα= 1.69–6.37 h⁻²₇₅× 10³⁰ W, about half of which is emitted by the central peak corresponding to the galaxy's nucleus. The total Hα flux of FCC207 is somewhat higher, F_Hα= 5.95–19.3 × 10⁻¹⁹ W m⁻², corresponding to L_Hα= 2.41–7.83h⁻²₇₅× 10³⁰ W. Binette et al. (1994) propose photoionization by post asymptotic giant branch (AGB) stars as a source for the central emission in elliptical galaxies. Using their prescriptions, we derive central Hα luminosities of the order of 2 × 10³⁰ W, i.e. comparable to what is observed. Hence, blindly interpreting the central Hα emission as evidence for star formation may be somewhat audacious. We can, however, check our results and use the calibration of Kennicutt (1983) between the total SFR and the Hα luminosity,  

11

where E_Hα= 1 mag is the internal extinction factor. We obtain  

12

in good agreement with the estimates based on the EWs given by Drinkwater et al. (2001).

4.3 H ii masses

The total mass in ionized hydrogen can be written as  

13

where L_Hα is the total Hα luminosity, m_H is the mass of the hydrogen atom and N_e is the electron density in the gas. The hydrogen Hα emissivity j_Hα is given by  

14

in ‘case B’ recombination, i.e. complete re-absorption of all Lyman photons in an optically thick nebula (Osterbrock 1989; Spitzer 1978; Macchetto et al. 1990). Each Lyman photon emitted from a level with n≥ 3 is later on converted to (a) Balmer photon(s) plus one Lyman α photon, thus raising the flux in the Balmer lines. The production coefficient α_Hα (calculated for T= 10⁴ K) is insensitive to the electron density (it changes by only 4 per cent if N_e is raised from 1 cm⁻³ to 10⁶ cm⁻³) and varies as T^−0.8 as a function of temperature. Using equations (13) and (14), the ionized hydrogen mass can be written concisely as  

15

cf. Kim (1989). In the following, we assume the value N_e= 1000 cm⁻³ for the electron density to be in accord with most other authors and to be able to directly compare our ionized hydrogen masses with the literature (however, Spitzer (1978) advocates N_e= 100 cm⁻³ as a typical value for both Galactic H ii regions with diameters of the order of 100 pc and for SNRs). Using equation (15), the mass of the ionized hydrogen gas in FCC046 can be estimated at M_{H ii}≈ 40–150 h⁻²₇₅ M_⊙ and at M_{H ii}≈ 60–190 h⁻²₇₅ M_⊙ in FCC207.

4.4 FCC06: a star-forming dE?

The H ii emission of FCC046 is distributed over a bright central region and six fainter clouds, labelled Cl1–CL6 in Fig. 9. Cl1, Cl2, Cl5 and Cl6 are identifiable in the B—R colour map. Cl1 and CL6 are part of the bluish nebulosity to the north of the nucleus whereas Cl2 and Cl5 show up as individual blue spots, about 0.1 mag bluer than their immediate surroundings. The diameters (FWHM) of these clouds were estimated using equation (8). We fitted Gaussian profiles to 11 stars in the pure-emission image of FCC046 and found FWHM_star= 0.78 ± 0.06 arcsec. Hence, clouds with an observed FWHM smaller than 0.84 arcsec (or a diameter smaller than ≈30 pc) cannot be regarded as resolved. Clouds Cl1, Cl3 and Cl6 are resolved under the given seeing conditions. In Table 1, the diameters and luminosities of the clouds are listed. In the fourth column, we give the emission rate of hydrogen ionizing photons Q_max needed to produce the luminosity L_em if the clouds are H ii regions (i.e. we assume that all the light is in the Hα line to obtain an upper limit for Q)  

16

where α_B= 2.59 × 10⁻¹³ cm³ s⁻¹ is the ‘case B’ recombination coefficient for T= 10⁴ K (Osterbrock 1989). An upper limit for the diameter of a H ii region, D_max, is then given by  

17

The values in Table 1 are calculated for N_e= 100 cm⁻³.

Table 1

The second and third columns show the logarithm of the measured diameters D (pc) and Hα+[N II] luminosities L_em (W) of the six emission clouds in FCC046. The fourth and fifth columns show the upper limits for the logarithm of the ionizing photon emission rates Q_max (s⁻¹) and diameters D_max (pc) if the clouds are HII regions with N_e = 100 cm⁻³.

Fig. 13 presents the diameters ( where a and b are the long and short axes FWHM, respectively) and Hα+[N ii] luminosities of M33 SNRs and of the six clouds identified in FCC046 (Long et al. 1990). The properties of the largest clouds are consistent with those of SNRs. The luminosities of the clouds are also compatible with those of H ii regions ionized by the light of single 05-B0 stars but at least Cl1, Cl3 and Cl6 seem too large for this interpretation. Though suggestive, this of course does not mean that they necessarily are SNRs. Nebulae around Wolf—Rayet stars could be a plausible alternative and are found in many irregulars and have appropriate luminosities and diameters (Hunter & Gallagher 1986; Chu & Lasker 1986). It is striking that these emission clouds, whatever their interpretation, are not found predominantly inside the bluish nebulosity to the north of the nucleus, something that would be expected if the blue light were coming from a young population of stars. Their true nature can, of course, only be assessed by spectroscopy.

Figure 13

Hα+[N ii] luminosity of SNRs in M33 (open squares) versus their diameter. The six clouds identified in FCC046 are presented as black dots. Three clouds have diameters that are too small to be measured (i.e. smaller than D≈ 30 pc) under the given seeing conditions.

Figure 14

Extrapolated central B-band magnitude, B0, versus scalelength derived by fitting an exponential to the surface brightness profile: black dots, FCC046 and FCC207; asterisks, BCDs; triangles, dEs; diamonds, dIrrs.

5 Conclusions

The similarities of the broad-band colours of FCC046 to those of star-forming or amorphous dwarfs, its relatively strong core and the presence of emission clouds support the conclusion that FCC046 is actively forming stars, albeit at a very leisurely pace when compared to BCDs and amorphous dwarfs, which are about a factor of 1000 more luminous in Hα. The nuclear emission of FCC046 and FCC207 can be adequately accounted for by photoionization by post-AGB stars, although a contribution of Hα emission from star formation cannot be excluded. Only the emission from the six clouds observed in FCC046 (SNRs, Wolf—Rayet nebulae) can be interpreted as unambiguous evidence for recent or ongoing star formation. The presence of physically different emission regions makes the interpretation of this emission in terms of a SFR cumbersome. High-resolution spectroscopy of a broad wavelength region is required to measure the strengths of Hα, Hβ and of telltale O, N and S emission lines in the visible part of the spectrum. These can be used as diagnostics to probe the physical nature of the different emission clouds.

Drinkwater et al. (2001) find no distinct class of star-forming BCD galaxies but instead observe Hα emission in dwarf galaxies of all sizes and types. Among these, star-forming dEs such as FCC046 may prove to be the descendants of more fiercely star-forming dwarfs such as BCDs which are not (or no longer) present in Fornax. As a check, we fitted exponentials to the surface-brightness profiles of FCC046 and FCC207 and compared the extrapolated B-band central surface brightnesses and the scalelengths with those of the Virgo dEs, BCDs and dwarf irregulars (dIrrs) presented in Drinkwater & Hardy (1991). Both galaxies have scalelengths in between those of BCDs and dEs, and quite high B-band central surface brightnesses compared to the dEs in the sample of Drinkwater & Hardy (1991). These results support our conjecture that dEs which contain ionized gas and possibly ongoing low-powered star formation can be considered as a missing link between BCDs and traditional dEs.

Acknowledgments

This paper is based on observations collected at the European Southern Observatory, Chile (ESO Large Programme Nr. 165.N-0115). This research has made use of the NASA/IPAC Extragalactic Data base (NED) which is operated by the Jet Propulsion Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space Administration. We thank the anonymous referee for helpful comments. SDR wishes to thank Dr Victor Debattista for useful comments on the causes of lopsidedness. WWZ acknowledges the support of the Austrian Science Fund (project P14783) and of the Bundesministerium für Bildung, Wissenschaft und Kultur.

1

MIDAS is developed and maintained by the European Southern Observatory (ESO).