3D structure of H ii regions in the star-forming complex S254-S258

ABSTRACT

The S254-258 star-forming complex is a place of massive star formation where five OB-stars have created H ii regions, visible as optical nebulae, and disrupted the parental molecular gas. In this work, we study the 3D structure of these H ii regions using optical spectroscopy and tunable-filter photometry with the 6- and 1-m telescopes of the Special Astrophysical Observatory of the Russian Academy of Sciences. We construct maps of the optical extinction and find that the H ii emission is attenuated by neutral material with 2 ≤ A_V ≤ 5 mag. The typical electron density in S255, and S257 is ≈100 cm⁻³, with enhancements up to 200 cm⁻³ in their borders, and up to 400 cm⁻³ toward the dense molecular cloud between them, where active star formation is taking place. We show that either a model of a clumpy dense neutral shell, where UV photons penetrate through and ionize the gas, or a stellar wind, can explain the shell-like structure of the ionized gas. S255 is surrounded by neutral material from all sides, but S257 is situated on the border of a molecular cloud and does not have dense front and rear walls. The compact H ii regions S256 and S258 are deeply embedded in the molecular clouds.

techniques: imaging spectroscopy, stars: massive, stars: winds, outflows, dust, extinction, ISM: H ii regions

1 INTRODUCTION

The appearance of H ii regions on optical images of the night sky is one of the most easily observed manifestations for the feedback from the young massive stars on the interstellar medium. Since at least the 1970s, it was observed that H ii regions can appear not only as sphere-like structures, as was previously considered in many theoretical works (e.g. Strömgren 1939; Spitzer 1978; Hosokawa & Inutsuka 2006; Kirsanova, Wiebe & Sobolev 2009; Bisbas et al. 2015). Instead, blister-like, bipolar, and irregular structures were found and modelled later by many authors (e.g. Bodenheimer, Tenorio-Tagle & Yorke 1979; Tenorio-Tagle 1979; Franco, Tenorio-Tagle & Bodenheimer 1990; Garcia-Segura & Franco 1996; Redman, Williams & Dyson 1998; Arthur & Hoare 2006; Steggles, Hoare & Pittard 2017). To date, the three-dimensional structure of H ii regions and surrounding medium has been studied mostly using two-dimensional maps of dust or molecular emission, using radial velocities of various atomic or molecular tracers (e.g. Emprechtinger et al. 2009; Anderson et al. 2015, 2019; Pabst et al. 2020; Beuther et al. 2022, and many others). In some cases for nearby objects, it is possible to use three-dimensional maps of dust extinction, complemented by 6D data from Gaia, to reconstruct not only the structure, but also the star-formation history in molecular clouds (e.g. Foley et al. 2023).

The present study continues the idea of using combinations of various dust tracers to reconstruct the three-dimensional structure of H ii regions, as in our previous work using archival data on far-infrared (IR) dust emission and optical wide-field images for the H ii region Sh2-235 (Kirsanova et al. 2020a, Paper I hereafter). Here, we develop the ideas used in Paper I for exploring the three-dimensional structure for more extensive work on several H ii regions in the star-forming complex S254-258 (see Appendix  A for a review of the appearance and estimated physical parameters of the individual H ii regions). This complex appears as six bubbles in the mid-infrared (see Fig. A1), with five of them associated with the optical H ii regions Sh2-254, Sh2-255, Sh2-256, Sh2-257, and Sh2-258 (S254, 255, 256, 257, and 258 hereafter), as well as a faint region in the north-east that has not been catalogued yet. The complex is situated in Perseus spiral arm, which has a distance from the Sun of ≈2 kpc in the direction of the anticentre of the Galaxy (Xu et al. 2006; Choi et al. 2014). The gas and dust temperatures in the S254-258 star-forming complex are mostly maintained by massive B-type stars (Evans, Blair & Beckwith 1977; Sargent et al. 1981; Avedisova & Kondratenko 1984; Heyer et al. 1989; Dors & Copetti 2003). As the majority of the massive stars that generate infrared ‘bubbles’ in the Galaxy have spectral classes from O to B2/B3 (e.g. Churchwell et al. 2006; Deharveng et al. 2010), our complex can be considered as a typical example of such structures. The parameters of the H ii regions already measured in the literature are presented in Table A1.

Locally, the energetics of the complex are supported by young embedded stellar objects surrounding the H ii regions, with the most prominent star-forming region located between S255 and S257, as well as to the south of them (see large-scale maps by Chavarría et al. (2008), Bieging et al. (2009), and Fig. A1). These authors suggest that a sequential star-formation process in the complex took place in the intersection of the bubbles associated with the H ii regions (see also Ojha et al. 2011; Chavarría et al. 2014; Kohno et al. 2022). In particular, star formation in the most massive and prominent region at mm and infrared wavelengths is a molecular ridge at the interface of S255 and S257 (e.g. Wang et al. 2011a; Zinchenko et al. 2015; Zemlyanukha et al. 2018), and was probably induced by the expansion of these H ii regions (Bieging et al. 2009; Mucciarelli, Preibisch & Zinnecker 2011; Ojha et al. 2011; Wang et al. 2011b; Zinchenko et al. 2012; Ladeyschikov et al. 2021). Samal et al. (2015) and Ryabukhina et al. (2018) found a 20-pc long star-forming molecular filament to the south of the bright optical nebulae, where the compact H ii region S258 is embedded. Finally, the compact H ii region S256 is situated to the south-west of S257, on the edge of an extended molecular cloud.

2 OBSERVATIONS AND DATA REDUCTION

We conducted optical observations with both the 6- and 1-m telescopes of the Special Astrophysical Observatory of the Russian Academy of Sciences (SAO RAS), using a similar technique described in Paper I. Here, we present a detailed description of the observations and data reduction steps, with special attention paid to differences in the detectors used and steps of the data reduction.

2.1 MaNGaL emission line mapping

Observations with the 1-m Zeiss-1000 telescope of SAO RAS were performed using the Mapper of Narrow Galaxy Lines (MaNGaL; Moiseev, Perepelitsyn & Oparin 2020), which is a tunable-filter photometer based on a piezoelectric scanning Fabry–Perot interferometer (FPI) working in a low order of interference (∼20 at 656 nm). The width of the instrumental profile in the 480–700 nm spectral range was FWHM = 12 ± 1 Å. The scanning system allows setting the central wavelength (CWL) of the FPI transmission peak to the desired wavelength at the centre of the field of view (FOV). A bandpass filter (FWHM was 10 nm in the Hβ and 25 nm in other spectral ranges) is used to select a single peak of the FPI transmission. Compared with the previous study of S235 in Paper I, the current work had the following differences in observation technique. We used the Tektronix 1 × 1 K CCD detector with 24 μm pixel size, operated by the new DINACON-V controller developed at SAO RAS (Ardilanov et al. 2020; Afanasieva et al. 2023), which provides a FOV of 11.5 arcmin (vignetted by a round holder of a bandpass filter) at a sampling of 0.9 arcsec px⁻¹. This is significantly larger than the square FOV (8.7 × 8.7 arcmin) in the observations of S235 in Paper I with the previous detector. In Paper I, we obtained relatively deep images in each emission line using the MaNGaL ‘single image mode’, where the CWL was switched between the emission line wavelength and the neighbouring continuum. Here, the ‘scanning mode’ was used to correct the mismatch between the peak CWL and the emission line barycentre caused both by variations of the object Dopler velocities and instrumental CWL distribution. We quickly scanned the wavelength regions around the emission line with a short exposure and lower spatial resolution with 4 × 4 readout binning. However, the radial change of the peak CWL across this large FOV with the TK1024 detector was quite significant (exceeding 1/2 of the FWHM). For this reason, all observations of S255-257 region were performed only in a scanning mode, with 1 × 1 binning and relatively long exposures. During a scanning cycle, we obtained several (5–12) images with CWL steps of 5–7 Å, covering the spectral range around each emission line (Hβ, [O iii] λ5007) or line system (Hα + [N ii] λ6548, 6583, [S ii] λ6717, 6731). The spectral continuum was sampled at 20–30 Å from each emission line region. The exposure time of each frame was 600 s for Hβ and 120–200 s for the other lines, and each cycle was repeated several times in order to average the contribution of atmospheric parameters and air mass variations. The total exposures, seeing value and number of frames in each scanning cycle (n_z) are given in Table 1.

Table 1.

MaNGaL observations at the Zeiss-1000 telescope.

Date	Sp. range	Exposure, s	|$n_z$|	seeing, arcsec
2020 Mar 01	Hβ	5400	5	2.6
2020 Mar 02	Hβ	4800	5	2.6
2020 Dec 24	[O iii]	4120	7	2.0
2020 Dec 24	Hα + [N ii]	3600	10	1.9
2022 Nov 28	[S ii]	8000	12	1.5
2022 Nov 30	[S ii]	15800	12	1.2

The data reduction was performed using the programs and algorithms described in Paper I and Moiseev et al. (2020). The calibration to an absolute intensity scale was performed using spectrophotometric standard stars observed in the same spectral range immediately before or after the exposures of the nebulae. The products of the data reduction are data cubes containing n_z-channel low-resolution spectra (FWHM = 12 Å, or R = λ/δλ = 400–650) at each pixel. The data cubes were aligned using the astrometric calibration from the astrometry.net software (Lang et al. 2010), and the data cubes in the Hβ and [O iii] emission lines obtained on different nights were co-added.

The continuum-subtracted spectra in each data cube were fit with a Lorentzian profile, providing a good approximation of FPI instrumental profile. We used a one-component Lorentzian for the Hβ and [O iii] lines, double-component model for [S ii], and three-component model for the Hα + [N ii] line systems. The free parameters were the central velocity, amplitude, and FWHM (the same for all lines in the cases of [S ii] and Hα + [N ii]). For the line systems, the wavelength difference between lines were fixed, and in the case of the [N ii] doublet the line ratio (1:3) was also fixed. The result of this procedure is two-dimensional maps of the flux in each emission line. The signal-to-noise (S/N) maps were calculated as a ratio of a line amplitude to photon noises in each pixel. In order to improve the detection limit for the faint structures we also produced data cubes with different binning: 1 × 1, 2 × 2, 4 × 4, and 8 × 8 pixels.

2.2 SCORPIO-2 spectroscopy

For additional study of the ionized gas properties and also to check the calibration of the MaNGaL flux maps, we performed observations in the long-slit mode of the SCORPIO-2 multimode focal reducer (Afanasiev & Moiseev 2011) on the prime focus of the 6-m Big Telescope Alt-Azimuth (BTA) telescope. The slit had a length of 6.3 arcmin and a width of 1 arcsec, which provides a spectral resolution of about 5 Å in the range 365–730 nm. The E2V 261–84 2 × 4 K pixel CCD detector gave a spatial sample of 0.39 arcsec per pixel. The detailed description of this CCD camera is presented in Afanasieva et al. (2023). The spectra were obtained along four positions in the S255–258 regions, and the corresponding position angles and other parameters are listed in Table 2. The preliminary data reduction was performed in a standard way using custom-developed IDL-based software, as described in our previous works (e.g. Egorov et al. 2018). To remove the contribution from airglow lines to the observed spectra where the ionized gas emission was detected along the entire slit, two blank fields were exposed at a distance of ∼1° from the S254-258 complex. In the case of the compact S258 region, the sky-night spectrum was taken from the area free from object emission along the slit. To calibrate the reduced spectra to an absolute flux-density scale, we used the spectra of spectrophotometric standard star observed at a close zenith distance immediately after the observations of the nebular complex. The integrated fluxes of emission lines were obtained via single-component Gaussian fitting after the removal of continuum emission fitting by splines. The uncertainties of the measured fluxes were estimated from Monte Carlo simulations of the synthetic spectra with the given S/N.

Table 2.

SCORPIO-2 observations at the BTA-6 telescope.

Date	Region and PA, (°)	Exposure, s	seeing, arcsec
2023 Jan 16	S255, PA = 109	1200	2.3
2023 Jan 16	S255/257 blank	600	2.3
2023 Jan 16	S256, PA = 170	720	1.9
2023 Jan 16	S256 blank	600	1.9
2023 Jan 16	S257, PA = 87	1200	2.2
2023 Jan 16	S258, PA = 87	1200	1.9

3 METHODS

3.1 Properties of ionized gas

3.1.1 Electron temperature

Unfortunately, the commonly used temperature-sensitive line ratio [O iii] λ4363/(λ4959 + λ5007) could not be applied, as the weaker [O iii] λ4363 emission line was not detected in the integrated spectra of the H ii regions in the S254–258 star-forming complex. Instead, we estimated T_e([N ii]) using the ratio of nebular to auroral nitrogen line intensities Q_2, N = [N ii] (λ6548 + λ6584)/λ5755 and the improved calibration relation between T_e([N ii]) and Q_2, N for the low-density regime (n_e < 100 cm⁻³) according Pilyugin, Vílchez & Thuan (2010). Because the Q_2, N ratio depends weakly on the interstellar extinction A_V, we dereddened the line intensities in the integrated spectra using the Hα/Hβ line ratio, as described in Section 3.1.2. The ratio of the Balmer lines also depends on T_e, so we did several iterations to determine the T_e([N ii]) values for S255, S256, and S257. The integration radii were 7–120 arcsec in S255 and S257 and 7–50 arcsec in S256. The zone of r < 7 arcsec around the bright central star was ignored to avoid contamination.

3.1.2 Extinction and electron density

In our analysis, we used only those pixels of the maps and velocity channels in the spectra where the S/N level was >3 and the line intensities and their ratios were not smaller than their uncertainties. Our methods are the same as in Paper I therefore here we only briefly summarize them. The basic property of the H ii regions, electron density n_e, was calculated using the ratio of the [S ii] λλ6716 Å/6731 Å lines and equations (3) and (4) of Proxauf, Öttl & Kimeswenger (2014). We found n_e in the entirety of the S255, 256, S257, and S258 H ii regions, and partially in S254 in the north-west part of the MaNGaL images. No spectroscopy was performed for S254 due to the weak emission. We determined the interstellar extinction A_V comparing the observed Hα/Hβ intensity ratio for each pixel of the MaNGaL images or position on the slits with the intrinsic ratio for Case B conditions given by Osterbrock & Ferland (2006) and the reddening law of Cardelli, Clayton & Mathis (1989).

For the nebular spectra, we determined the extinction also using Hγ/Hβ, Hδ/Hβ, and Hϵ/Hβ values. For the ratio of total to selective extinction, we adopted the value of R_V = 3.1 as our standard model. However, we tried higher values up to R_V = 5.5 for the analysis of the spectra. In order to estimate uncertainties of n_e and A_V, we applied a bootstrap approach using the measured flux densities and their uncertainties for a random sample. In order to study gradients of these values over the nebulae and increase the values-to-uncertainties ratios, we binned the original images and finally used an 8 times larger pixel for the maps of physical parameters and dereddened images of the surface brightness. For the spectra, we used 20 times larger bins to improve the signal-to-noise ratio.

3.2 Dust column density

In order to study the three-dimensional structure of the H ii regions, we used the map of dimensionless equivalent of the optical extinction |$A_{V}^{\rm IR}$| obtained by Ladeyschikov et al. (2021). This map can be transferred into hydrogen column density as |$A_{V}^{\rm IR} = 5.3 \times 10^{-22} \times$| N(H i + H₂) mag (Bohlin, Savage & Drake 1978; Rachford et al. 2009). The map we use in the present study is based on a grey-body fit of the spectral energy distribution (SED), with a dust emissivity κ ∝ λ^−1.8, of the Herschel Hi-GAL data (Molinari et al. 2010) in the 160–500 μm range.

3.3 Properties of ionizing stars and their extinction

We used the stellar spectra in order to estimate the spectral type of the ionizing stars in each of the H ii regions. Furthermore, the shape of the SED of the ionizing stars at optical and near-infrared wavelengths can be used to constrain the value of R_V, i.e. the ratio of the total to selective extinction (A_V/E_{B − V}), in the extinction law, and provide an independent measure of the extinction A_V towards the star.

We fit the continuum-normalized spectra of the ionizing stars of S255, S256, and S257 using the ‘Tlusty’ grids of non-LTE stellar atmospheres for O- (Lanz & Hubeny 2003) and B-stars (Lanz & Hubeny 2007), convolved to the spectral resolution of our observations (FWHM = 5 Å). As a first step, we initially placed no restrictions on metallicity Z/Z_⊙, surface gravity log g or effective temperature T_eff of the model spectra, and found the best fit (i.e. lowest χ²) from the OSTAR2002 or BSTAR2006 precomputed grids without performing any interpolation on the stellar parameters. We specifically excluded wavelengths with nebular emission lines (which might leave artefacts after the background subtraction), DIBs and atmospheric absorption from the fits. For all three regions, we found that a metallicity of Z/Z_⊙ = 0.2 and surface gravity of log g = 4 (typical for main-sequence OB stars) provide the best fits, and we therefore fixed these values for the rest of the fitting process.

After fixing the metallicity and surface gravity, we fit the effective temperature in a more refined manner using the equivalent widths of He i and He ii lines, as well as the C iii 4650 Å line. This approach has the advantage that it is less sensitive to the exact shape of the stellar continuum used for normalization. We interpolated the Tlusty continuum-normalized spectra to arbitrary effective temperatures, as the two grids have spacings of either 2500 K (the OSTAR2002 grid) or 1000 K (the BSTAR2006 grid). The uncertainty in the temperature determination was estimated by a bootstrapping process, where we created synthetic spectra from the observed spectra with flux values drawn from a normal distribution with a width equal to the estimated flux uncertainty of the observations (typically about 10 per cent).

In order to determine the extinction parameters A_V and R_V from the SED of the ionizing stars, we compared the best fit from the Tlusty spectra together with photometric measurements in the optical and near-infrared. Specifically, we used BV measurements from UCAC5 (Zacharias, Finch & Frouard 2017), G, G_BP, and G_RP from Gaia DR2 (Gaia Collaboration et al. 2018), gri from APASS (Henden et al. 2015), and JHK from 2MASS (Skrutskie et al. 2006). The observed (reddened) SED was compared with the theoretical SED using the reddening law of Cardelli et al. (1989), with A_V and R_V as free parameters, and their uncertainties were found using a bootstrapping process. This procedure has the advantage of providing an additional, independent check on the values of A_V derived from the Hα and Hβ images of the nebulae from MaNGaL, and allows us to determine R_V, which cannot be determined independently from A_V from observations of only two Balmer lines.

4 RESULTS OF OBSERVATIONS AND ANALYSIS

The resulting integrated spectra reveal numerous ionized gas emission lines: the hydrogen Balmer series (Hα, Hβ, Hγ, Hδ, Hε), as well as forbidden lines (mostly in S255 and 257): [O i] λ6300, [O ii] λ3727, [O iii] λ4959, 5007, [N ii] λ5199, 5755, 6548, 6583, [S ii] λ6716, 6731, [Ar iii] λ7136. Also in S255 and 257, the helium lines He i λ5876, 6678 were detected. The integrated spectra and relative flux of all emission lines are presented in Appendix  B.

Electron temperatures, T_e, obtained from the integrated spectra, are presented in Table 3. These values are in good agreement with results measured in the literature earlier (Table A1). Also in Table B2, we present T_e([N ii]) values calculated separately for the inner and outer parts of the S255 and S257 regions. A possible radial trend of T_e is not seen. We used a value of T_e = 7940 K, averaged over all the H ii regions, to create maps of the ionized gas parameters based on the MaNGaL images.

Table 3.

Electron temperatures obtained from the integrated spectra (T_e), effective temperature of the ionizing stars (T_eff), and parameters of the absorbing medium toward the stars derived from the observed spectra and SEDs (A_V, R_V).

H ii region	Te	Teff	AV	RV
	(K)	(K)	(mag)
S255	8141 ± 90	31300 ± 120	3.84 ± 0.12	3.10 ± 0.32
S256	8222 ± 762	22750 ± 270	5.61 ± 0.15	4.66 ± 1.33
S257	7459 ± 211	27550 ± 130	2.76 ± 0.13	3.30 ± 0.45

A colour composite image using the surface brightness of three emission lines without continuum is shown in Fig. 1. Separate images of the surface brightness of the hydrogen recombination lines Hα, Hβ, and forbidden lines of metals [S ii], [N ii], and [O iii] are shown in Fig. 2. The images are shown scaled to the values of their surface brightness, where the Hα lines reach 5 × 10⁻⁴ erg s⁻¹ cm⁻² sr⁻¹ and the [O iii] line reaches a factor of ten less in the brightest parts. Besides the three H ii regions S255, S256, and S257 situated completely within our field of view, we also detect diffuse emission from the east edge of the H ii region S254. The spatial distribution of the surface brightness is inhomogeneous over the images, with the brightest values to the north-east of the ionizing star LS 19 in S255. There are several elongated structures, indicating that the absorbing material is concentrated in the central part of the image. Images of S256 in the Hα, [N ii], and [S ii] lines appear as rings around the ionizing star. The surface brightness of the [S ii] line in S255 also has a semi-ring distribution. The [O iii] line is detected only to the north-east of LS 19 in S255 and appears as a compact, bright spot.

Figure 1.

Colour composite image using the surface brightness of three emission lines, with red: Hβ, green: [S ii], and blue: [O iii].

Figure 2.

MaNGaL images of S254-258 star-forming region. Only pixels with S/N ratio >3 are shown. The red star symbols show the positions of the ionizing stars. The names of the H ii regions and their ionizing stars are shown by black and red, respectively, on the top panel. The IR sources S255 IRS1 and IRS3 are shown in magenta, where we used the coordinates from Howard, Pipher & Forrest (1997). The white dashed lines show positions of the slits for spectra. A physical scale bar is shown by black line on the bottom panel.

The spatial distribution of A_V is shown in Fig. 3. The optical extinction reaches 4 mag in the dark dust fibres in front of the south-west part of S255. The north-east part of S255 is less obscured, with A_V ≈ 1.5–2 mag. Absorbing material is concentrated to the east in S257. The A_V value decreases from 4 to 1 mag to the west and north-west of the nebula. S256 is obscured up to A_V = 4 mag, with the extinction peak located in the direction of the proposed ionizing star. The spatial distribution of the extinction with the enhancement between S255 and S257 is also visible in Fig. 4, where we show results from the analysis of the long-slit spectra; the slit spectra allow determining the A_V values even in positions where the signal on the MaNGaL images was too weak. A_V reaches 5 mag on the edge of the molecular cloud between S255 and S257, and has a similar value to the south of S256. S258 is deeply embedded in neutral absorbing material, with A_V = 9–11 mag, but the optical emission around the ionizing star is weak, and we did not obtain physical parameters in the rest of the positions along that slit.

Figure 3.

Physical parameters in S255-258: A_V (top) and n_e (bottom). The images were rebinned to a pixel size 8 times larger than the original pixel in Fig. 2. The red star with black border symbols show the positions of the ionizing stars. The black dashed lines show the positions of the slits for the spectra. Pixels with values determined for n_e but without A_V value were removed from the bottom map to simplify comparison. A physical scale bar is shown by black line on the bottom panel.

Figure 4.

Extinction (left) and electron density (right) in the H ii regions along the slits. Vertical dashed lines show the location of the ionizing stars. The orientation of the slits in Fig. 2 is shown by ‘N’,‘S’, ‘E’, or ‘W’ symbols. Extinction values obtained with different Balmer lines are shown by colours.

The extinction, obtained using other lines of the Balmer series: Hγ, Hδ, and Hϵ, is also shown in Fig. 4. Using the same reddening law, we find the A_V values systematically lower than the extinction from the ratio of Hα/Hβ in S255 and S257, while high uncertainties do not allow concluding about S256. Using another total-to-selective extinction ratio R_V value, we were able to bring the A_V values from different lines into agreement at R_V = 4. However, we do not exclude that the described effect is related to deviations of the observed line fluxes from the intrinsic values given by theory due to e.g. inhomogeneity of the absorbing material. Our suggestion that the R_V value can be higher than commonly used for diffuse medium is marginally supported by analysis of the spectra of the ionizing stars. The upper limits of R_V towards them agree with the Balmer series analysis (Table 3). As the present study will be followed by a more extensive work on numerous H ii regions with various properties, we plan to provide a more general analysis of the R_V value towards H ii regions in the following study.

The distribution of the electron density in the observed H ii regions is not homogeneous. The ionizing stars of S255 and S257 are surrounded by gas with n_e ≈ 100–150 and 50–75 cm⁻³, respectively (Fig. 3). We find that the southern part of S255 has systematically higher gas density compared to the northern part, where n_e drops to ≈75–90 cm⁻³. In S257, we find an east–west gradient of n_e from >100 to <70 cm⁻³. The largest n_e value up to 400 cm⁻³ is found on the borders of the H ii regions (Fig. 4).

As is known from previous studies, there is a molecular cloud between S255 and S257. The central enhancement of the n_e close to the molecular cloud is accompanied by a rise of the A_V value (see results from spectra in Fig. 4). This effect is less visible in Fig. 3, as we removed all pixels where n_e is known but A_V is not, in order to make these maps more convenient for comparison. Therefore, these H ii regions irradiate the cloud from opposite sides and ionize the dense material. The enhancement of n_e up to 200 cm⁻³ in the outskirts of the H ii regions can also be related to the ionization of the surrounding dense neutral envelopes (see Fig. A1). Therefore, the H ii regions S255 and S257 are reliable examples of media where UV photons penetrate through the dense medium. The widths of the dense ionized walls of the H ii regions are 20–30 arcsec, which is equivalent to 0.2 pc for the distance of the S254-258 complex.

In the compact H ii region of S256, we found n_e ∼ 100 cm⁻³. It seems to still be embedded in the parental molecular cloud. There is a north-east–south-west gradient of the n_e and A_V values from 100 to 50 cm⁻³ (the limit for the diagnostic using the [S ii] lines) and from 4–5 to 2 mag, respectively. Another compact H ii region S258 demonstrates the highest n_e up to 600 cm⁻³, and A_V up to 9–11 mag. The optical spectroscopy allowed us to estimate the physical conditions only towards the ionizing star due to the faintness of the lines.

Applying the A_V map to the images of the line emission, we obtain the dereddened flux |$F\left(\lambda \right)=F_\mathrm{obs}\left(\lambda \right) \times 10^{0.4 A_\lambda }$| in each filter, where the functional form of A_λ/A_V was taken from the extinction law of Cardelli et al. (1989), and we used the A_V value derived at each pixel from the Hα/Hβ line ratio. The maps of dereddened line emission are shown in Fig. 5. We find three different types of surface brightness distribution among them. The dereddened Hα emission is distributed more uniformly in S255 than on the uncorrected maps (Fig. 2); now it has a peak of 3 × 10⁻³ erg s⁻¹ cm⁻² sr⁻¹, corresponding to the position of LS 19 and the bright area at the south-west. The dereddened Hβ and [N ii] images have qualitatively similar intensity distributions. The [S ii] line emission is distributed in a semi-ring way, also with a bright region of 8 × 10⁻⁴ erg s⁻¹ cm⁻² sr⁻¹ at the south-west. Due to the re-binning of the original line emission images, the signal-to-noise level is higher. Therefore, we see larger area of the [O iii] emission than in Fig. 2, and find the peak of the [O iii] emission of 4 × 10⁻⁴ erg s⁻¹ cm⁻² sr⁻¹ towards the ionizing star. In S257, the Hα and [S ii] are distributed more uniformly at the levels of 7 and 2 × 10⁻⁴ erg s⁻¹ cm⁻² sr⁻¹, respectively, with moderate enhancements around the ionizing star and on the eastern border.

Figure 5.

Dereddened images of the nebulae. Red star with black border symbols show the positions of the ionizing stars. The images are rebinned to a pixel size 8 times larger than the original pixel. A physical scale bar is shown by black line on the bottom panel.

5 3D STRUCTURE OF THE H ii REGIONS

Our conclusions about the clumpy and non-uniform medium are supported by early works in the infrared. Evans et al. (1977) found that the spatial distribution of the neutral material around the H ii regions is inhomogeneous. Radio-continuum images by Israel (1976); Snell & Bally (1986); Fich (1993) show a centrally-peaked distribution of the radio continuum emission for S255 and S256, but diffuse emission for S254 and a peak shifted to the east for S257. They also found that the radio brightness in S255 is higher than in S257 by a factor of 2–3. Here, we confirm a similar effect in the dereddened optical maps.

As shown above, distribution of the foreground material in S255 and S257 is inhomogeneous, with elongated structures in S255. These structures are visible also on the near-infrared images of S 255 obtained by Kirsanova et al. (2023) but are not visible on the images in longer wavelength, e.g. in mid-infrared or millimetre waves by Chavarría et al. (2008); Bieging et al. (2009), respectively. A gradual decrease of extinction is observed in S257. Only in S256 we found the peak of A_V coincident with the location of the ionizing star.

In order to study the 3D structure of the H ii regions, we plot the neutral hydrogen column density in Fig. 6 in dimensionless |$A_{V}^{\rm IR}$| values. This procedure allows separating the neutral material of the front and the rear walls of the H ii regions. The difference between the A_V and |$A_V^{\rm IR}$| values is that the first defines the the foreground extinction between the H ii region and the observer, but the second depends on three components: the foreground extinction, absorbing dust inside of H ii regions, and the neutral material behind the ionized region.

Figure 6.

Column density of neutral hydrogen in dimensionless equivalent of extinction values |$A_{V}^{\rm IR}$|⁠. Colour wedge has a square-root stretch to emphasize lower intensity features. Red star with black border show the positions of the ionizing stars. Black contours show the region with foreground A_V ≥ 3 mag. The plus symbols show positions if infrared sources IRS 1 and IRS 3. A physical scale bar is shown by black line.

The most prominent feature in the distribution of the |$A_V^{\rm IR}$| value corresponds to the molecular cloud between S255 and S257, where |$A_V^{\rm IR} \ge 50$| mag and where active star formation is taking place (see relevant references in Section 1). No optical emission is found towards regions where |$A_V^{\rm IR} \ge 20$| mag, and our contours of A_V ≥ 3 mag avoid that area. Comparing the |$A_V^{\rm IR}$| and A_V values in S255, we find inhomogeneities both in the front and rear neutral walls. Namely, there is an area to the south of the ionizing star where |$A_V^{\rm IR} \approx A_V$|⁠, and we suggest that the rear wall in that direction is absent. The rear wall of this H ii region is visible as a continuous curved line from the west, north and east of LS 19 star. Thus, S255 appears as an example of an H ii region surrounded by neutral walls from all directions, which makes this object a promising target for comparison with theoretical models of H ii regions.

Both S255 and S257 appear similarly, namely, they look as round H ii regions at optical wavelengths. The ionized gas is surrounded by neutral envelopes visible through the mid-infrared emission of PAH particles (see studies by e.g. Pavlyuchenkov, Kirsanova & Wiebe 2013; Mackey et al. 2016; Salgado et al. 2016; Stock et al. 2016; Boersma, Bregman & Allamandola 2018; Murga et al. 2022; Kirsanova & Pavlyuchenkov 2023, about the appearance of bright PAH emission in the neutral envelopes of H ii regions, visible e.g. by Spitzer). However, the morphology of S257 is different from those found in S255. There are no pronounced front or back walls in S257; this region is situated on the edge of the dense molecular cloud. We see neutral borders between S257 and S255, as well as between S257 and S254, in the Spitzer images, which implies that S257 is still bounded by neutral gas from at least the east and west sides. These observations are consistent with previous results by Bieging et al. (2009), who observed CO lines in the star-forming complex and found a good correlation between CO and mid-IR emission of warm dust. These authors suggested that S255 is probably on the near side of the CO molecular ridge, while S257 is carving out part of the molecular cloud.

Relating the measured surface brightness to the line-of-sight depth using the emission measure, as in Paper I, we find the depth of S257 is two times larger than in S255. The values of the depth are 10–20 pc, while the radii are ≈1 pc. In actuality, the ionized volume of S257 can be smaller due to the escape of the gas from S257, as it is located on the border of the dense cloud. S255, on the other hand, is surrounded by dense neutral walls from all sides. We propose that the depth values can be overestimated with this simplified approach, probably due to the inhomogeneous distribution of the ionized gas along the line of sight. Alternatively, as the front and rear walls are non-uniform (S255) or almost absent (S257), the ionized gas can escape from the 1 pc vicinity of the ionizing stars, where 1 pc is the radius of the H ii regions in the plane of the sky (see the studies already mentioned in Section 1 and also recent work by Kirsanova et al. 2023).

Comparing the parallax-based distances of LS 19 (2060|$^{2181}_{1951}$| pc by Gaia Collaboration 2020) to the water maser in S255 IRS1 (1780|$^{1900}_{1670}$| pc by Burns et al. 2016) or to the methanol maser in the same direction (1590|$^{1660}_{1530}$| pc by Rygl et al. 2010), we find that the ionizing star of S255 is at least 50 pc farther from the observer than the masers (compare the low limit 1951 pc for LS 19 and the high limit for the water maser of 1900 pc). Therefore, the star-forming complex appears to be extended along the line of sight. For HD 25332, the Gaia DR3 distance has a quality much less than for LS 19, and we consider a spectro-photometric measurement 2.46 ± 0.16 kpc by Russeil et al. (2007) as the best measurement for now. This distance value supports our suggestion that S254-258 is extended perpendicular to the plane of the sky.

In S256, there is enhancement of the |$A_V^{\rm IR}$| in the south in the direction of the molecular cloud, which is accompanied by a rise of the A_V and n_e values. This H ii region, as well as S257, is apparently an example of a blister-type H ii region, which was formed by a star on the border of the dense molecular cloud. In these types of H ii regions, density of all the neutral as well as ionized gas components decreases away from the molecular cloud, as it was shown by simulations and observations e.g. by Israel (1978); Tenorio-Tagle (1979); Whitworth (1979); O’Dell (2001); Poppi et al. (2007); O’Dell et al. (2008); Gendelev & Krumholz (2012), and many others for the Orion Nebulae and another H ii regions. We note that subsequent analysis of the ionized and neutral gas kinematics can be useful to confirm this suggestion, which will be done in forthcoming studies.

Having only two measurements of the A_V value in S258, we compared them with |$A_V^{\rm IR}$| (not shown on the figures) and found |$A_V^{\rm IR} < A_V$|⁠. Spatial resolution, which is more than two times higher in the optical data versus the infrared, possibly explains this finding. Fig. A1 shows that the mid-IR image of S258 has only a compact and bright spot near the ionizing star. The 3D structure of this H ii region can not be studied with the available data for now.

The spatial distribution of the dereddened [S ii] emission and n_e demonstrate that S255 is not a pure ‘classical’ spherical H ii region, where electrons are distributed uniformly, but resembles a ring-like structure, at least in the plane of the sky, as is sometimes observed for planetary nebulae (e.g. Turatto et al. 2002; O’Dell, Sabbadin & Henney 2007). The presence of the front and back neutral walls shows that the H ii region has no cavity free of electrons along the line of sight and does not resemble a torus. This object resembles an ionized ellipsoid with non-uniform density distribution.

We applied a non-stationary model of an H ii region surrounded by a PDR (Kirsanova et al. 2009, 2019) to the ionizing stars of S255 and S257 to simulate them and confine their physical conditions. Simulations with uniform, as well as with non-uniform, distribution of n_init, where the density rises with distance, always produces uniform H ii regions. Therefore, we again suggest that the high electron density on the borders of H ii regions can be related to the penetration of diffuse UV photons through clumpy neutral envelopes. Similar conclusions about clumpy PDRs were made e.g. by Hogerheijde, Jansen & van Dishoeck (1995); Lis & Schilke (2003); Andree-Labsch, Ossenkopf-Okada & Röllig (2017) for the nearby Orion Bar PDR, and by Ciurlo et al. (2019); Kirsanova et al. (2020b); Schneider et al. (2021) for more distant regions, where the clumps were not resolved in the observations.

Another possibility to obtain shell-like ionized volumes is to include stellar wind in our model. Many authors demonstrated how stellar wind changes the structure of uniform H ii regions and produce central cavities, e.g. Mackey et al. (2016); Katushkina & Izmodenov (2019); Whitworth, Priestley & Geen (2022); Kirsanova & Pavlyuchenkov (2023). We do not see such signatures of the wind as broad-line wings in the stellar spectra. We can exclude wind-only inspecting stellar spectra, looking for the lines of the high excitation in the ultraviolet (UV), which are not currently available. Future UV missions such as Spektr-UV (Boyarchuk et al. 2016) may help to solve the wind question. Diffuse X-ray emission, which also traces stellar wind in H ii regions (e.g. Pabst et al. 2019; Luisi et al. 2021), has not been investigated in S255-257. Therefore, the question about the stellar wind remains for future studies.

The photodissociation regions (PDRs) surrounding the S254-258 complex are irradiated by a moderate UV field G₀ < 100 Habings, see Kirsanova et al. (2023), comparable with the Horsehead and S235 PDRs (Philipp et al. 2006; Kirsanova et al. 2021). Using simulations of an expanding H ii region for a star with effective temperatures of 30 × 10³ K (Kirsanova et al. 2020b), we found values of G₀ and compared them with the observed ones in S255 and S257. A PDR with G₀ > 20 Habings spreads far over 2 pc from the ionizing stars for n_init = 10² cm⁻³. Therefore, this model does not work for S255 and S257. We found models with n_init = 10³ and 10⁴ cm⁻³ are more relevant for these regions as they produce PDRs with 5 < G₀ < 300 Habings.

These indications about the non-uniform gas distribution are also supported by simulations. Kirsanova et al. (2023) found that the widths of the PDRs in S255 and S257 range from 0.1 up to 0.5 pc for different values of the uniformly distributed gas density in the range 10² ≤ n_init ≤ 10⁴ cm⁻³, and these widths are too thin compared with the observed values of 0.3–0.4 pc. Simulations with n_init = 10³–10⁴ cm⁻³ also give n_e values of 50–70 cm⁻³, which agree with our observations within a factor of 2. The regions with the high n_e on the borders of the simulated H ii regions have widths ≤0.05 pc, which is 3–5 times less than those observed. All these discrepancies can be overcome considering a non-uniform clumpy medium where expansion of the H ii regions takes place.

In spite of the importance of the clumpy and non-uniform medium for the study of the three-dimensional structure, another property of the absorbing material along the line of the sight, namely the enhancement of the total-to-selective extinction up to R_V = 4, can be explained by numerical models with uniformly distributed material. Destruction of small dust grains by UV photons, or dust drift under radiation pressure in H ii regions, results in a decrease in the small dust grain abundance, as was found observationally by Marconi et al. (1998); Witt et al. (2006), and subsequently shown by the simulations of Pavlyuchenkov et al. (2013) and Akimkin et al. (2015).

6 CONCLUSIONS

We present observations of the star-forming complex S254-258 using the optical tunable-filter photometer MaNGaL on the Zeiss-1000 telescope at SAO RAS. Surface photometry in Hα, Hβ, [N ii], [S ii], and [O iii] lines is complemented by long-slit spectroscopic observations using SCORPIO-2 at the 6-m BTA telescope at SAO RAS. Our main conclusions are as follows:

• The two extended H ii regions in the complex, S255 and S257, both have a more or less round shape in the plane of the sky. Both regions are attenuated by absorbing material with 2 ≤ A_V ≤ 5 mag, with clear enhancements towards the molecular cloud located between them. The electron density in these regions rises from 100 cm⁻³ near the ionizing stars, to 400 cm⁻³ at the edge of this dense molecular cloud. Moreover, there is another enhancement of the electron density toward the outskirts of S255 and S257. These enhancements may be related to diffuse UV photons that penetrate through the clumpy, dense neutral material and ionize it. Another possibility is the evacuation of ionized gas from the vicinity of the massive stars by stellar winds. In order to test the latter suggestion, additional observations in X-rays or UV wavelengths are needed.

• The three-dimensional structure of S255 differs from that of S257. Specifically, S255 is surrounded by dense neutral gas from all sides, while S257 is situated on the border of a molecular cloud, and does not have dense front and rear walls. We propose that S257 represents a blister-type H ii region, as the density of both ionized and neutral gas components decreases away from the molecular cloud.

• There are two compact H ii regions, S256 and S258, which are deeply embedded into molecular clouds, with A_V ≈ 5 and 10 mag, respectively. S256 probably represents a blister-type H ii region, but at an earlier stage in comparison to S257. The electron density in S256 decreases from 100 to 50 cm⁻³, along with the amount of neutral material at both the front and rear walls of the H ii region. S258 has the highest electron density of all the regions (600 cm⁻³), although we were not able to study the three-dimensional structure of this region due to the weakness of the optical lines.

• We suggest that the total-to-selective extinction R_V towards these H ii regions is higher than in the diffuse ISM, due to the destruction of small dust grains by UV photons or dust drift from radiation pressure.

ACKNOWLEDGEMENTS

We thank P. M. Zemlyanukha and I. I. Zinchenko for fruitful discussions of the star-forming complex. We are also thankful to the unknown referee for his/her very relevant comments.

This study was supported by the Russian Science Foundation, grant 21-12-00373. Observations with the SAO RAS telescopes are supported by the Ministry of Science and Higher Education of the Russian Federation. The renovation of telescope equipment is currently provided within the national project ‘Science and Universities’.

DATA AVAILABILITY

The data underlying this article will be shared on reasonable request to the corresponding author.

APPENDIX A: PARAMETERS OF THE STAR-FORMING S254-258 COMPLEX

Figure A1.

Mid-IR colour image made by Spitzer telescope at 8 μm of the star-forming complex S254-258 with the Spitzer Super-Mosaic Pipeline (Digital Object Identifier (DOI): 10.26131/IRSA433). The black contours show hydrogen column density N(H i + H₂) calculated with the Herschel data the 160–500 μm range by Ladeyschikov et al. (2021). The contours are given for 6, 12, 24, and 48 × 10²⁰ cm⁻². The black dashed lines show the long slits for the optical spectroscopy in the present study. Red star with black border show the positions of the ionizing stars. A physical scale bar is shown by white line. The plus symbols show positions if infrared sources IRS 1 and IRS 3.

Table A1.

Parameters of the exciting stars and H ii regions from the literature used in the present study.

H ii region	Exciting star	Te
		(K)
S254	O9.0V1, O9.0V3	–
S255	B0.0V1, B0.0V3	82004, 8833 ± 1707
S256	B2.5V2, B0.9V3	–
S257	B0.5V1	6900–79005, 7970 ± 2606
S258	B3V2, B1.5V3	–

References:¹ Moffat, Fitzgerald & Jackson (1979), ² Russeil, Adami & Georgelin (2007), ³ Chavarría et al. (2008), ⁴ Fernández-Martín et al. (2017), ⁵ Méndez-Delgado et al. (2022), ⁶ Esteban & García-Rojas (2018), ⁷G192.638-00.008 in Wenger et al. (2019).

APPENDIX B: INTEGRATED SPECTRA OF THE NEBULA

Figs B1–B2 present the integrated of all four regions observed with SCORPIO-2 in the long-slit mode after subtraction of continuum interpolated by cubic spline. The central zone of the star formation regions (r < 7 arcsec) was ignored during flux integration to avoid bright central star contamination in the total spectrum. The Poisson noise grows significantly in the blue part of the spectra because the total quantum efficiency of the used CCD and grism decreases at λ < 5000 Å.

Figure B1.

The integrated continuum-subtracted optical spectra of S255 and S257. The integration range is shown at the top. The main object’s and telluric emission lines are labelled in blue and black, respectively.

Figure B2.

Same as Fig. B1, but for S256 and S258.

The normalized intensities of the emission lines (Gaussian fitting results) are presented in Table B1. For the two largest regions S255 and S257, we present results separately for the inner (r < 60 arcsec) and outer (r = 60–120 arcsec) radial range, because the relative intensity of high-excitation lines ([O iii], He i, [Ar iii]) is significantly lower in the outer regions. Table B2 presents the similar data after interstellar extinction corrected by using Hα/Hβ ratio in the integrated spectra. The corresponding A_V and T_e values are also shown. All errors are 3σ.

Table B1.

Relative intensities of emission lines in the integrated spectra of S255-S258 regions (I(Hβ) = 100) before reddening corrections. Flux errors are 3σ.

Line	Sh2-S255		Sh2-S256	Sh2-S257		Sh2-S258
	(r = 7–60 arcsec)	(r = 60–120 arcsec)	(r = 7–50 arcsec)	(r = 7–40 arcsec)	(r = 40–120 arcsec)	(r = 7–30 arcsec)
[O ii] λ3727	97.8 ± 2.41	122.9 ± 3.20	133.3 ± 14.8	95.23 ± 4.57	107.7 ± 4.66	–
H|$\varepsilon \, \lambda$|3970	7.01 ± 0.32	8.54 ± 0.39	20.0 ± 3.51	5.89 ± 0.58	6.03 ± 0.55	–
H|$\delta \, \lambda$|4102	12.14 ± 0.27	13.11 ± 0.33	15.4 ± 2.44	11.93 ± 0.48	10.90 ± 0.50	–
H|$\gamma \, \lambda$|4340	29.33 ± 0.27	31.64 ± 0.34	31.7 ± 1.87	31.64 ± 0.49	30.89 ± 0.49	–
H|$\beta \, \lambda$|4861	100.00 ± 0.27	100.00 ± 0.32	100.0 ± 1.64	100.00 ± 0.45	100.00 ± 0.44	100.00 ± 22.0
[O iii] λ5007	19.41 ± 0.15	2.65 ± 0.12	–	5.73 ± 0.22	0.40 ± 0.18	–
[N ii] λ5199	1.20 ± 0.09	2.77 ± 0.13	7.4 ± 1.52	2.08 ± 0.22	2.99 ± 0.26	130.92 ± 40.5
[N ii] λ5755	1.54 ± 0.08	1.89 ± 0.09	1.8 ± 0.50	1.15 ± 0.15	1.12 ± 0.15	–
He i λ5876	12.62 ± 0.11	1.44 ± 0.09	–	6.24 ± 0.17	0.79 ± 0.16	–
[O i] λ6300	6.21 ± 0.14	11.79 ± 0.21	–	2.28 ± 0.27	3.99 ± 0.34	–
[S iii] λ6312	2.07 ± 0.09	1.65 ± 0.13	–	0.88 ± 0.17	0.78 ± 0.18	–
N ii λ6548	101.18 ± 0.36	111.28 ± 0.46	119.9 ± 1.46	85.72 ± 0.51	94.27 ± 0.55	200.60 ± 20.8
H|$\alpha \, \lambda$|6563	893.23 ± 2.53	778.28 ± 2.60	987.6 ± 3.16	755.87 ± 3.49	753.80 ± 3.46	2705.08 ± 39.2
[N ii] λ6583	309.35 ± 0.93	341.51 ± 1.20	356.0 ± 2.05	259.84 ± 1.28	286.26 ± 1.39	738.48 ± 28.0
He i λ6678	5.43 ± 0.10	0.37 ± 0.09	1.0 ± 0.36	2.26 ± 0.15	0.20 ± 0.10	–
[S ii] λ6716	80.69 ± 0.30	127.57 ± 0.51	202.4 ± 1.69	106.97 ± 0.61	156.50 ± 0.82	527.75 ± 26.2
[S ii] λ6731	64.19 ± 0.26	100.70 ± 0.43	155.9 ± 1.58	82.49 ± 0.50	119.19 ± 0.66	526.03 ± 26.3
He i λ7065	3.39 ± 0.10	0.16 ± 0.05	–	1.16 ± 0.18	–	–
[Ar iii] λ7136	15.50 ± 0.13	1.77 ± 0.11	–	5.58 ± 0.19	–	30.92 ± 9.27

Table B2.

Reddening-corrected relative intensities of emission lines in the integrated spectra of S255-S258 regions (I(Hβ) = 100). Flux errors are 3σ.

Line	Sh2-S255		Sh2-S256	Sh2-S257		Sh2-S258
	(r = 7–60 arcsec)	(r = 60–120 arcsec)	(r = 7–50 arcsec)	(r = 7–40 arcsec)	(r = 40–120 arcsec)	(r = 7–30 arcsec)
[O ii] λ3727	246.28 ± 6.07	275.85 ± 7.18	364.78 ± 40.47	208.46 ± 10.00	235.34 ± 10.18	–
H|$\varepsilon \, \lambda$|3970	14.47 ± 0.66	16.12 ± 0.74	44.28 ± 7.76	10.92 ± 1.08	11.13 ± 1.02	–
H|$\delta \, \lambda$|4102	22.61 ± 0.50	22.58 ± 0.57	30.43 ± 4.79	20.21 ± 0.81	18.44 ± 0.85	–
H|$\gamma \, \lambda$|4340	45.43 ± 0.42	46.40 ± 0.50	51.16 ± 3.01	45.88 ± 0.71	44.74 ± 0.71	–
H|$\beta \, \lambda$|4861	100.00 ± 0.27	100.00 ± 0.32	100.00 ± 1.64	100.00 ± 0.45	100.00 ± 0.44	100.00 ± 22.04
[O iii]λ5007	17.07 ± 0.13	2.37 ± 0.11	–	5.14 ± 0.20	0.35 ± 0.16	–
[N ii]λ5199	0.90 ± 0.07	2.15 ± 0.10	5.43 ± 1.11	1.64 ± 0.17	2.35 ± 0.20	74.80 ± 23.17
[N ii] λ5755	0.79 ± 0.04	1.05 ± 0.05	0.87 ± 0.23	0.65 ± 0.08	0.64 ± 0.09	–
He i λ5876	5.99 ± 0.05	0.75 ± 0.05	–	3.32 ± 0.09	0.42 ± 0.09	–
[O i] λ6300	2.34 ± 0.05	5.02 ± 0.09	–	1.00 ± 0.12	1.74 ± 0.15	–
[S iii] λ6312	0.78 ± 0.03	0.69 ± 0.05	–	0.39 ± 0.07	0.34 ± 0.08	–
N ii λ6548	33.71 ± 0.12	42.51 ± 0.18	36.17 ± 0.44	33.72 ± 0.20	37.18 ± 0.22	22.89 ± 2.38
H|$\alpha \, \lambda$|6563	295.49 ± 0.84	295.48 ± 0.99	295.45 ± 0.95	295.47 ± 1.36	295.49 ± 1.36	302.83 ± 4.37
[N ii] λ6583	101.36 ± 0.30	128.58 ± 0.45	105.38 ± 0.61	100.75 ± 0.50	111.31 ± 0.54	80.70 ± 3.06
He i λ6678	1.70 ± 0.03	0.13 ± 0.03	0.28 ± 0.10	0.85 ± 0.06	0.08 ± 0.04	–
[S ii] λ6716	24.84 ± 0.09	45.49 ± 0.18	55.99 ± 0.47	39.35 ± 0.22	57.74 ± 0.30	51.07 ± 2.54
[S ii] λ6731	19.63 ± 0.08	35.69 ± 0.15	42.81 ± 0.43	30.17 ± 0.18	43.71 ± 0.24	50.38 ± 2.52
He i λ7065	0.90 ± 0.03	0.05 ± 0.02	–	0.37 ± 0.06	–	–
[Ar iii] λ7136	3.98 ± 0.03	0.54 ± 0.03	–	1.76 ± 0.06	–	2.19 ± 0.66
AV	2.96 ± 0.01	2.81 ± 0.01	3.23 ± 0.01	2.51 ± 0.01	2.51 ± 0.01	–
Te([N ii]), K	8089 ± 133	8218 ± 130	8222 ± 762	7615 ± 290	7357 ± 311	–

APPENDIX C: SPECTRA AND SEDS OF THE IONIZING STARS

Figure C1.

Normalized spectrum of the ionizing star of S255 (black) together with the best-fit stellar atmosphere (red). Photospheric lines are labelled in black, while interstellar and telluric lines are labeled in grey.

Figure C2.

Same as Fig. C1, but for S256.

Figure C3.

Same as Fig. C1, but for S257.

Figure C4.

Fit of the observed (reddened) SED of the ionizing star of S255 (black), together with the dereddened spectrum (grey). Photometric measurements are shown as crosses, with the measurement error indicated for the observed values.

Figure C5.

Same as Fig. C5, but for S256.

Figure C6.

Same as Fig. C6, but for S257.