Online material

Appendix A: Notes on individual galaxies

Appendix A.1: Galaxies with a regularly rotating H I disk

NGC 1705 has a strongly warped H I  disk. Our rotation curve rises more steeply than those of Meurer et al. (1998) and Elson et al. (2013) because we applied a beam-smearing correction to the inner velocity-points using 3D disk models (see Fig. 5). Meurer et al. (1998) and Elson et al. (2013) decomposed their rotation curves into mass components and found that DM dominates the gravitational potential at all radii. We did not build a detailed mass model because the optical and kinematic centers differ by ~550 pc, while PA_opt and PA_kin differ by ~45°. NGC 2366 has an extended H I  disk with a strong kinematic distortion to the North-West (see its velocity field in Appendix C). Our rotation curve is in overall agreement with previous results (Hunter et al. 2001; Thuan et al. 2004; Oh et al. 2008; Swaters et al. 2009; van Eymeren et al. 2009b), but we do not confirm the declining part of the rotation curve found by Hunter et al. (2001) and van Eymeren et al. (2009b) at R ≳ 5′. This latter result appears to be caused by an anomalous H I  cloud that lies at V_los ≃ 130 km s^-1 along the major axis (~7′ from the galaxy center to the North, see the PV-diagram in Fig. 3 and Appendix C). NGC 4068 has a H I  distribution characterized by a central depression and several shell-like structures. The H I  kinematics is slightly lopsided. Our rotation curve agrees with the one of Swaters et al. (2009) within the errors. NGC 4214 has a H I  disk with multiple spiral arms. Intriguingly, the optical and H I  spiral arms wind in opposite directions (clockwise and counter-clockwise, respectively). The H I  disk is close to face-on and strongly warped, thus the rotation curve is uncertain. In the inner parts, our rotation curve rises more steeply than the one derived by Swaters et al. (2009); the difference seems to be due to a different choice of the dynamical center (see the PV-diagram in Swaters et al. 2009). NGC 6789 has a compact H I  disk that extends out to only ~3.5 optical scale lengths. The inclination is uncertain: we derived i = 43° ± 7° using 3D disk models. UGC 4483 has been studied in Lelli et al. (2012b). I Zw 18 has been studied in Lelli et al. (2012a). I Zw 36 has an extended and asymmetric H I  distribution (see Ashley et al. 2013), but in the central parts the H I  forms a compact, rotating disk. The optical and kinematic centers are offset by ~12′′ (~340 pc), while PA_opt and PA_kin differ by ~36°. SBS 1415+437 is a prototype “cometary” BCD, as the starburst region is located at the edge of an elongated stellar body. Remarkably, the kinematic center does not coincide with the optical one but with the starburst region to the South (see Appendix C; the object at RA ≃ 14^h17^m00^s and Dec ≃ 43°29′45′′ is a foreground star). The lopsided H I  distribution and kinematics may be due to a pattern of elliptical orbits centered on the starburst region (cf. Baldwin et al. 1980).

Appendix A.2: Galaxies with a kinematically disturbed H I disk

NGC 625 has been previously studied by Côté et al. (2000) and Cannon et al. (2004). Côté et al. (2000) suggested that the complex H I  kinematics is due to an interaction/merger, whereas Cannon et al. (2004) argued that it is best described by a gaseous outflow superimposed on a rotating disk. We find it difficult to distinguish between these two possibilities. It is clear, however, that the galaxy has a inner, rotating disk with V_rot ≃ 30 km s^-1 (see PV-diagram in Appendix C). NGC 1569 has been previously studied by Stil & Israel (2002) and Johnson et al. (2012). Both studies derived a rotation curve by fitting the H I  velocity field with a tilted-ring model. The PV-diagram along the major axis, however, does not show any sign of rotation in the inner parts (R ≲ 1′, see Appendix C). Moreover, the H I  line-profiles are very broad and asymmetric, likely due to strong non-circular motions. For these reasons, we restrict our analysis to the rotation velocity in the outer parts (~50 km s^-1). NGC 4163 shows a very small velocity gradient of ~10 km s^-1. The complex H I  kinematics may be due to the low V_rot/σ_{H I} ratio. The PA of the stellar body and of the H I  disk significantly differ by ~40°. NGC 4449 has been previously studied by Hunter et al. (1998, 1999), who found that the H I  distribution forms 2 counter-rotating systems. For the inner H I  disk, we find a rotation velocity of ~35 km s^-1. It is unclear whether the outer gas system is really a counter-rotating disk or is formed by two or three H I  tails wrapping around the inner disk (similarly to I Zw 18, see Lelli et al. 2012a). NGC 5253 has been previously studied by Kobulnicky & Skillman (2008) and López-Sánchez et al. (2012), who discussed the possibility of gas inflows/outflows along the minor axis of the galaxy. The data are, indeed, consistent with a H I  disk with V_rot < 5 km s^-1 and V_rad ≃ 25 km s^-1 (see Fig. 7). Shadowing of the X-ray emission indicates that the southern side of the galaxy is the nearest one to the observer (Ott et al. 2005a), suggesting that the radial motions are an inflow. UGC 6456 has been previously studied by Thuan et al. (2004) and Simpson et al. (2011). Simpson et al. (2011) derived a rotation curve using low-resolution (C+D array) observations. They assumed different values of the PA for the approaching and receding sides, which would imply an unusual, asymmetric warp starting within the stellar component (see their Fig. 13). Our 3D models show that the H I  kinematics may be simply explained by a disk with V_rot ≃ V_rad ≃ σ_{H I} ≃ 10 km s^-1 (see Fig. 6). UGC 9128 has a H I  disk that rotates at ~25 km s^-1, but the VF is very irregular and the H I  line profiles are broad and asymmetric, possibly due to non-circular motions. The optical and kinematic PA differ by ~30°.

Appendix A.3: Galaxies with unsettled H I distribution

UGC 6541 has a very asymmetric H I  distribution. Gas emission is detected only in the northern half of the galaxy. This may be the remnant of a disrupted disk. UGCA 290 has a H I  distribution that is offset with respect to the stellar component. The kinematics is irregular and dominated by a few distinct H I  clouds.

Appendix B: Tables

Appendix B.1: Properties of the H I datacubes

Column (1) gives the galaxy name, following the ordering NGC, UGC, UGCA, Zwicky, SBS. Column (2) gives the radio interferometer used for the 21 cm-line observations. Columns (3)−(5) give the spatial and spectral resolutions of the original cube. This cube is typically obtained using a Robust parameter ℜ ≃ 0. Columns (6)−(8) give the spatial and spectral resolutions of the cube after Gaussian smoothing. Column (9) gives the ratio of the H I  radius R_{H I} (given in Table 3) to the final H I  beam. Column (10) gives the noise in the final cube. Column (11) provides the reference for the original cube.

Appendix B.2: Optical and H I  orientation parameters

Column (1) gives the galaxy name. Columns (2)−(6) give the optical center, ellipticity, inclination, and position angle. These values are derived by interactively fitting ellipses to the outer isophotes. The inclination is calculated assuming an oblate spheroid with intrinsic thickness q₀ = 0.3. Columns (7)−(11) give the kinematical center, systemic velocity, inclination, and position angle. These values are derived using H I  velocity fields, channel maps, PV-diagrams, and building 3D disk models. Column (12) gives the projected offset between the optical and kinematical centers. This is calculated as , assuming the galaxy distance given in Table 1. The error is estimated as FWHM/2.35, where FWHM is the beam of the smoothed H I  datacube (see Table B.1). Projected distances smaller than FWHM/2.35 are assumed to be zero.

Appendix B.3: Mass budget within the optical radius.

Column (1) gives the galaxy name. Column (2) gives the stellar mass. This is calculated by integrating the galaxy SFH and assuming a gas-recycling efficiency of 30%. The SFHs were derived by fitting the CMDs of the resolved stellar populations and assuming a Salpeter IMF from 0.1 to 100 M_⊙. Column (3) gives the molecular mass. This is indirectly estimated using Eq. (4), which assumes that the star-formation efficiency in dwarfs is the same as in spirals. Column (4) gives the H I  mass M_{H I} within R_opt. Column (5)−(7) give the baryonic mass within R_opt assuming, respectively, a Kroupa IMF, a Salpeter IMF, and a Salpeter IMF plus the possible contribution of molecules. Column (8) gives the circular velocity at R_opt. Column (9) gives the dynamical mass within R_opt calculated as . Columns (10)−(12) gives the baryonic fraction within R_opt assuming, respectively, a Kroupa IMF, a Salpeter IMF, and a Salpeter IMF plus molecules. Italics indicate unphysical values >1.

Appendix C: Atlas

In the following, we present overview figures for the 18 starbursting dwarfs in our sample. For each galaxy, we show six panels including both optical and H I  data. Top-left: a sky-subtracted optical image in the R or V band. The cross shows the optical center. Bottom-left: an isophotal map (black contours) overlaid with a set of concentric ellipses (white contours). The value of the outermost isophote μ_out is given in the note; the isophotes increase in steps of 1 mag arcsec^-2. The orientation parameters for the ellipses (ϵ_opt and PA_opt) are given in Table B.2. The cross shows the optical center. For I Zw 18, the isophotal map was derived from a R-band HST image after the subtraction of the Hα emission, as the nebular emission dominates the galaxy morphology (see Papaderos et al. 2002). Top-middle: the total H I  map. The contour levels are at 1, 2, 4, 8, ... ×N_{H I} (3σ), where N_{H I} (3σ) is the pseudo-3σ contour, calculated following Verheijen & Sancisi (2001). The value of N_{H I} (3σ) is given in the note. The cross shows the optical center. The ellipse shows the beam. Bottom-middle: the H I  surface density profile, derived by azimuthally averaging over the entire H I  disk (black line) and over the approaching and receding sides separately (filled and open circles, respectively). In UGC 6541 and UGCA 290, H I  emission is detected only on one side of the galaxy, thus the H I  surface density profile was derived using the optical orientation parameters and averaging over a single side. Top-right: the H I  velocity field. Light and dark shading indicate approaching and receding velocities, respectively. The thick,

black line shows the systemic velocity. The velocity interval between approaching (black) and receding (white) contours is given in the note. The cross shows the optical center, while the circle shows the kinematic center. The dashed line indicates the kinematic position angle. The ellipse shows the beam. Bottom-right: Position-Velocity diagram taken through the kinematic center and along the kinematic major axis. Contours are at −3, −1.5 (dashed), 1.5, 3, 6, 12, ... ×σ. The value of σ is given in Table B.1. The vertical and horizontal lines show the kinematic center and the systemic velocity, respectively. For galaxies with a regularly rotating H I  disk, squares show the rotation curve as derived in Sect. 5.1, projected along the line of sight. For galaxies with a kinematically disturbed H I  disk, arrows show the estimated value of V_rot, projected along the line of sight.

	Fig. C.1 Contours: μout = 24.5 R mag arcsec-2; NH I (3σ) = 1.1 × 1020 atoms cm-2; Vlos = 398 ± 10 km s-1.
Open with DEXTER

	Fig. C.2 Contours: μout = 24 V mag arcsec-2; NH I (3σ) = 4.3 × 1020 atoms cm-2; Vlos = −80 ± 20 km s-1.
Open with DEXTER

	Fig. C.3 Contours: μout = 25.5 R mag arcsec-2; NH I (3σ) = 1.1 × 1020 atoms cm-2; Vlos = 635 ± 20 km s-1.
Open with DEXTER

	Fig. C.4 Contours: μout = 24.5 V mag arcsec-2; NH I (3σ) = 2.3 × 1020 atoms cm-2; Vlos = 103 ± 10 km s-1.
Open with DEXTER

	Fig. C.5 Contours: μout = 24.5 R mag arcsec-2; NH I (3σ) = 3.6 × 1020 atoms cm-2; Vlos = 206 ± 10 km s-1.
Open with DEXTER

	Fig. C.6 Contours: μout = 26.5 V mag arcsec-2; NH I (3σ) = 2.7 × 1020 atoms cm-2; Vlos = 158 ± 5 km s-1.
Open with DEXTER

	Fig. C.7 Contours: μout = 24.5 V mag arcsec-2; NH I (3σ) = 1.2 × 1020 atoms cm-2; Vlos = 291 ± 15 km s-1.
Open with DEXTER

	Fig. C.8 Contours: μout = 25 V mag arcsec-2; NH I (3σ) = 1.4 × 1020 atoms cm-2; Vlos = 210 ± 10 km s-1.
Open with DEXTER

	Fig. C.9 Contours: μout = 24 R mag arcsec-2; NH I (3σ) = 2.1 × 1020 atoms cm-2; Vlos = 410 ± 10 km s-1.
Open with DEXTER

	Fig. C.10 Contours: μout = 24.5 R mag arcsec-2; NH I (3σ) = 3.4 × 1020 atoms cm-2; Vlos = −151 ± 5 km s-1.
Open with DEXTER

	Fig. C.11 Contours: μout = 25 R mag arcsec-2; NH I (3σ) = 3.4 × 1020 atoms cm-2; Vlos = 158 ± 10 km s-1.
Open with DEXTER

	Fig. C.12 Contours: μout = 25 R mag arcsec-2; NH I (3σ) = 1.8 × 1020 atoms cm-2; Vlos = −102 ± 5 km s-1.
Open with DEXTER

	Fig. C.13 Contours: μout = 25.5 R mag arcsec-2; NH I (3σ) = 3.4 × 1020 atoms cm-2; Vlos = 250 ± 10 km s-1.
Open with DEXTER

	Fig. C.14 Contours: μout = 26.5 V mag arcsec-2; NH I (3σ) = 2.0 × 1020 atoms cm-2; Vlos = 150 ± 5 km s-1.
Open with DEXTER

	Fig. C.15 Contours: μout = 25 R mag arcsec-2; NH I (3σ) = 3.1 × 1020 atoms cm-2; Vlos = 468 ± 5 km s-1.
Open with DEXTER

	Fig. C.16 Contours: μout = 24 R mag arcsec-2; NH I (3σ) = 6.3 × 1020 atoms cm-2; Vlos = 767 ± 10 km s-1.
Open with DEXTER

	Fig. C.17 Contours: μout = 24.5 R mag arcsec-2; NH I (3σ) = 7.4 × 1020 atoms cm-2; Vlos = 277 ± 10 km s-1.
Open with DEXTER

	Fig. C.18 Contours: μout = 24.5 R mag arcsec-2; NH I (3σ) = 3.7 × 1020 atoms cm-2; Vlos = 616 ± 5 km s-1.
Open with DEXTER

© ESO, 2014