Enhanced dust emissivity power-law index along the western H α filament of NGC 1569
Toyoaki Suzuki, Hidehiro Kaneda, Takashi Onaka, Mitsuyoshi Yamagishi, Daisuke Ishihara, Takuma Kokusho, Takurou Tsuchikawa
aa r X i v : . [ a s t r o - ph . GA ] M a r MNRAS , 000–000 (0000) Preprint 9 November 2018 Compiled using MNRAS L A TEX style file v3.0
Enhanced dust emissivity power-law index along thewestern H α filament of NGC 1569 T. Suzuki ⋆ , H. Kaneda , T. Onaka , M. Yamagishi , D. Ishihara , T. Kokusho and T. Tsuchikawa Graduate School of Science, Nagoya University, Furo-cho Chikusa-ku, Nagoya, 464–8602, Japan Department of Astronomy, Graduate School of Science, The University of Tokyo, 7-3-1 Hongo Bunkyo-ku, Tokyo, 113–0033, Japan Institute of Space and Astronautical Science, Japan Aerospace Exploration Agency, Sagamihara, Kanagawa 252–5210, Japan
ABSTRACT
We used a data set from
AKARI and
Herschel images at wavelengths from 7 µ mto 500 µ m to catch the evidence of dust processing in galactic winds in NGC 1569.Images show a diffuse infrared (IR) emission extending from the galactic disk intothe halo region. The most prominent filamentary structure seen in the diffuse IRemission is spatially in good agreement with the western H α filament (western arm).The spatial distribution of the F / F map shows high values in regions around thesuper-star clusters (SSCs) and towards the western arm, which are not found in the F / F map. The color-color diagram of F / F – F / F indicates high valuesof the emissivity power-law index ( β c ) of the cold dust component in those regions.From a spectral decomposition analysis on a pixel-by-pixel basis, a β c map showsvalues ranging from ∼ to ∼ over the whole galaxy. In particular, high β c valuesof ∼ are only observed in the regions indicated by the color-color diagram. Sincethe average cold dust temperature in NGC 1569 is ∼ K, β c < . in the far-IR andsub-mm region theoretically suggests emission from amorphous grains, while β c = . suggests that from crystal grains. Given that the enhanced β c regions are spatiallyconfined by the HI ridge that is considered to be a birthplace of the SSCs, the spatialcoincidences may indicate that dust grains around the SSCs are grains of relativelyhigh crystallinity injected by massive stars originating from starburst activities andthat those grains are blown away along the HI ridge and thus the western arm. Key words: galaxies: dwarf – galaxies: halos – galaxies: ISM – galaxies: individual:NGC 1569 – infrared: galaxies.
Understanding of galaxy evolution remains a key subject inmodern astrophysics. Circulation of gas and dust on galacticscales has a large influence on the evolution of both the inter-stellar medium (ISM) and the intergalactic medium (IGM),and thus serves a vital role in galaxy evolution.One of major drivers to eject the ISM into the IGM isgalactic winds driven by mechanical energy and momentumfrom supernovae and stellar winds (e.g., Aguirre et al. 2001;Zu et al. 2011). Galactic winds are observationally foundat all cosmic epochs and are very prominent and ubiqui-tous especially in the redshift range of z ∼ ⋆ E-mail: [email protected] and consequently can change the properties of the galax-ies (e.g., Cazzoli et al. 2014). On the other hand, the pres-ence of metal-enriched gas and dust grains in the IGMdirectly affects the thermal balance of gas. In particular,Montier & Giard (2004) show that infrared (IR) emissionfrom dust grains can be considered as a dominant cool-ing agent of gas with temperatures of – K, whichare the typical range in the IGM; the cooling efficiency bydust grains strongly depends on a dust-to-gas mass ratio(DGR) and a dust size distribution. The efficient coolingby dust grains induces an increase of star and dust for-mations, and may impact on hierarchical clustering of thelarge-scale structures. Despite the importance of the roleof dust grains in galaxy evolutions, comprehensive under-standing of the fate of dust due to galactic winds is notobservationally established yet, although recent optical, IRand sub-millimeter (mm) observations of nearby starburst c (cid:13) T. Suzuki et al.
Figure 1.
Composite image of NGC 1569: 0.66 µ m (H α ) (magenta), 0.837 µ m (green) and 70 µ m (cyan). Martin (1998) defines thefour-superbubble (A, B, E and F) regions denoted by the dashed line. H α filaments inside the superbubbles were identified by Martin(1998) and Westmoquette et al. (2008). The other features were identified by Hodge (1974), Arp & Sandage (1985), Israel & van Driel(1990), M¨uhle et al. (2005), and Hunter et al. (1993). galaxies show extended emissions from dust grains and poly-cyclic aromatic hydrocarbons (PAHs) and revealed kine-matics of PAHs in galactic winds (e.g., Kaneda et al. 2010;Roussel et al. 2010; Yoshida et al. 2011).The dwarf-irregular galaxy NGC 1569 is an ideal lab-oratory to investigate dust processing in galactic windsand to insight into it in the early universe. Dwarf galax-ies are believed to be the ‘building blocks’ of larger galax-ies in the early universe. In particular, starburst dwarf ir-regular galaxies are expected to have experienced grav-itational interactions or mergers. NGC 1569, a memberof the IC 342 group of galaxies, may have interactedwith a companion galaxy (Johnson 2013) and has under-gone at least three major starburst phases; the most re-cent and strongest starburst ( ∼ ⊙ yr − kpc − ) started ∼
40 Myr ago and ended ∼
10 Myr ago (Angeretti et al.2005). During this phase, the galaxy witnessed ∼ su-pernova explosions, and the formation of star clusters in-cluding the two prominent super-star clusters (SSCs) A andB (Arp & Sandage 1985). These activities have had dra-matic effects on both the ISM and the halo of NGC 1569:(i) a cavity shown in the H i , H α and far-IR distributionscentered on SSC A, most probably due to stellar windsfrom SSC A (Israel & van Driel 1990; Hunter et al. 2000;Lianou et al. 2014), (ii) chimneys in southward and north-ward directions from SSC A, which are considered to be ma-jor pathways to eject the ISM into the halo (Hunter et al.1993; M¨uhle et al. 2005; Westmoquette et al. 2008), (iii) abipolar metal-enriched outflow comprising kpc-scale expand-ing superbubbles on the southern and northern sides ofthe galactic disc (Hunter et al. 1993; Heckman et al. 1995;Martin 1998; Martin et al. 2002; Westmoquette et al. 2008).As shown in Fig. 1, H α observations show that thehalo contains four expanding superbubbles (A, B, E and F;Martin 1998), which include a number of cellular filamentsexcited by shocks; their dynamical ages are roughly esti- mated to be 10–25 Myr from their expanding velocities (80–100 km s − ) and diameters ( ∼ α fila-ments, while the harder X-ray component (0.7–1.1 keV) islikely spatially correlated with the centers of the superbub-bles. The softer and brighter X-ray component is associatedwith shocks caused by galactic wind-halo interactions.At a position in the most prominent H α filament(western arm), Onaka et al. (2010) found the presence ofthe unidentified infrared (UIR) bands at 3.3, 6.2, 7.7 and11.3 µ m by AKARI /IRC spectroscopic observations. Un-der the shock environment, the destruction timescale ( ∼ yr) of the UIR band carriers, which have been possi-bly attributed to PAHs, is much shorter than the dy-namical timescales of the galactic wind in superbubble A(cf. Micelotta et al. 2010). The result may indicate thatthe band carriers are produced by fragmentation of largergrains in shocks and that dust processing took place overa wide area of the halo region. Despite the possibility, ex-tensive investigations on dust processing in galactic windsfrom NGC 1569 have never been reported because of a lackof spatially resolved far-IR images before the Herschel era.In this paper, using a data set from
AKARI (Murakami et al. 2007) and
Herschel (Pilbratt et al. 2010)imaging observations at wavelengths from 7 µ m to 500 µ m,we report spatial distributions of dust grains entrained bygalactic winds from NGC 1569 based on pixel-by-pixel spec-tral energy distribution (SED) fitting to catch the evidenceof dust processing in galactic winds. MNRAS , 000–000 (0000) ust properties in galactic winds from NGC 1569 Table 1.
Summary of the IR and sub-mm imaging data set taken from the
AKARI and
Herschel dataarchives.Telescope/Instrument λ ( µ m) FWHM ( ′′ ) Observation IDs 1- σ background noise (MJy sr − ) AKARI /IRC 7 5.1 1400423 . × − AKARI /IRC 11 4.8 1400423 . × − AKARI /IRC 15 5.7 1400424 . × − AKARI /IRC 24 6.8 1400424 . × − Herschel /PACS 70 5.6 1342243816–1342243821 1.0
Herschel /PACS 160 11.4 1342243816–1342243821 1.6
Herschel /SPIRE 250 18.4 1342193013 1.3
Herschel /SPIRE 350 25.2 1342193013 1.5
Herschel /SPIRE 500 36.7 1342193013 0.4
NGC 1569 was observed as part of the
AKARI missionprogram ‘ISM in our Galaxy and Nearby Galaxies’ (IS-MGN – P.I. Kaneda H.; Kaneda et al. 2009a) and aspart of the
Herschel programs ‘the Dwarf Galaxy Sur-vey’ (DGS – P.I. Madden S.; Madden et al. 2013) and ’Ex-ploring the Dust Content of Galactic Winds with Her-schel: The Dwarf Galaxy Population’ (PI: Veilleux S.). Weused
AKARI /IRC (Onaka et al. 2007) and
Herschel /PACS-SPIRE imaging data (Poglitsch et al. 2010; Griffin et al.2010), which were taken from data archives: the
AKARI pointing data archive through the data archive and trans-mission system (DARTS) and the
Herschel
Science Archive(HSA). Table 1 lists the summary of the data used in thisstudy.The IRC data obtained with the IRC02 observationmode provide mid-IR images with four bands (7, 11, 15and 24 µ m), which are finely allocated for probing emissionfrom PAHs and very small grains. Each field-of-view has asize of about ′ × ′ . The details of the observations aredescribed in Onaka et al. (2010); the FWHM of the pointspread function (PSF) ranges from ∼ ′′ to ∼ ′′ . The mid-IR images with a pixel size of 2 . ′′ µ m image, mid-IR emission in the halo region of interest is affected by anextended component of the PSF due to diffraction and scat-tering when bright sources such as SSCs are observed. Tocorrect for the extended PSF effects, the image reconstruc-tion method proposed by Arimatsu et al. (2011) was appliedfor the four-band images; the method consists of the decon-volution with the intrinsic PSF of an input image and of theconvolution with a Gaussian pattern with the same FWHMof the PSF. To apply the convolution, the pixel size of thefour-band images was reduced by a factor of 2 (1 . ′′ µ m), 7.5% (IRC11 µ m), 18% (IRC 15 µ m) and 16% (IRC 24 µ m).Cross-linked observations were performed by the PACS,and simultaneously provide 70 and 160 µ m-band imageswhich cover an area of ∼ ′ × ′ around NGC 1569. Both datawere taken from the HSA and are level-3 products (SPGversion 14.2.0) providing Unimap maps. The FWHM of thePSF is 5 . ′′ µ m and 11 . ′′ µ m.The systematic flux calibration uncertainty is 5% for bothbands (PACS Observer’s Manual version 2.5.1, 2013).SPIRE large cross-scan observations provide 250, 350and 500 µ m-band images at the same time and cover an areaof ∼ ′ × ′ around NGC 1569 in each band. The details ofthe observations are described in R´emy-Ruyer et al. (2013).The SPIRE data were taken from the HSA and are level-2products (SPG version 14.1.0) providing Naive maps withthe pixel sizes of 6 ′′ for SPIRE 250 µ m, 10 ′′ for SPIRE 350 µ mand 14 ′′ for SPIRE 500 µ m. The beam size has the FWHMsof 18 . ′′
4, 25 . ′′ . ′′ µ m, SPIRE 350 µ mand SPIRE 500 µ m, respectively. The systematic flux cali-bration uncertainty of the three SPIRE bands is applied tobe ∼
5% (SPIRE Observer’s Manual version 2.5, 2014).A background level was estimated by averaging the val-ues from multiple apertures placed around 2 ′ –3 ′ away fromthe galaxy center without overlapping with extended emis-sion and was subtracted from each image. For multi-bandanalysis, the spatial resolutions of background-subtracted IRimages were reduced to match the PSF of the SPIRE 500 µ mdata by convolving PACS and SPIRE-band images with ker-nels provided by Aniano et al. (2011). Because such kernelsfor the AKARI bands are currently not available in the li-brary , the AKARI images were convolved by a Gaussiankernel to approximate the SPIRE 500 µ m PSF by the mod-erate Gaussian FWHM of 41 ′′ (Aniano et al. 2011). Then,the convolved images were regridded with a pixel size of 36 . ′′ µ m data. ∼ ganiano/Kernels.htmlMNRAS , 000–000 (0000) T. Suzuki et al.
Figure 2.
Background-subtracted nine-band images of NGC 1569 with the
AKARI /IRC and
Herschel /PACS-SPIRE in the (a) IRC 7 µ m,(b) IRC 11 µ m, (c) IRC 15 µ m, (d) IRC 24 µ m, (e) PACS 70 µ m, (f) PACS 160 µ m, (g) SPIRE 250 µ m, (h) SPIRE 350 µ m and (i)SPIRE 500 µ m bands. In each image, the color bar is given in units of MJy sr − . The PSF size in FWHM is shown in the lower left-handcorner. For comparison with the spatial distribution of the H α emission shown in the panel (j) (Hunter & Elmegreen 2004), the samepanels of (d) and (g) are shown together with the H α contours in panels (k) and (l). Figure 2 shows the background-subtracted images ofNGC 1569 obtained with
AKARI and
Herschel . To comparethem with the spatial distribution of the H α emission shownin the panel (j) (Hunter & Elmegreen 2004), the same panelsof (d) and (g) are shown together with the H α contours inpanels (k) and (l). The 1- σ background fluctuations per pixelare 0.04–0.1 MJy sr − for 7–24 µ m and 0.4–1.6 MJy sr − for70–500 µ m (Table 1). Most of the images clearly show dif-fuse IR emission extending from the galactic disk into thehalo region (hereafter referred to as IR-halo emission). Fil-amentary structures seen in the H α emission are spatially correlated with those seen in the IR-halo emission. In su-perbubbles A and E, a pair of filaments which outline thewestern and eastern sides of each superbubble is observedin mid-IR to sub-mm images; the most prominent IR fila-ment is spatially in good agreement with the western arm. Insuperbubble B, the IR-halo emission extends toward northalong the H α chimney. With regard to superbubble F, whichis formed by the ISM escaping from the galactic disk throughthe H i chimney (Westmoquette et al. 2008), patchy IR-haloemission is observed along H α filaments. Furthermore, theIR-halo emission also shows the component extending in themajor-axis direction over the H α emission.The IRC 7 µ m and 11 µ m images are dominated bythe PAH emission not only from the galactic disk itself but MNRAS , 000–000 (0000) ust properties in galactic winds from NGC 1569 Figure 3.
Color maps of (a) F / F , and (b) F / F with the original pixel size of the SPIRE 500 µ m data (14 ′′ ). Panels (c) and(d) are the same as (a) and (b), but with a pixel size of 36 . ′′
7, respectively. The PSF shown in the panels (a) and (c) is matched to theSPIRE 500 µ m-band PSF. The detection threshold is set to be the 5 σ level for each band image. The contours superimposed on theimages are the same as those in Fig. 2. also from the halo; Onaka et al. (2010) detected the PAHemission from the western arm. Overall spatial distribu-tions in the 7 µ m and 11 µ m images are in good agree-ment with those in the 160–500 µ m images which tracethe spatial distribution of the cold dust component (20–30 K, R´emy-Ruyer et al. 2013; Lianou et al. 2014), associ-ated mainly with the old stellar population. The fact mayindicate that PAHs coexist with cold dust. The IRC 15 µ mand 24 µ m images trace emission from both hot (100–200 K)and warm (40–60 K) dust components (Lianou et al. 2014;Galliano et al. 2003), associated mainly with massive star-forming regions. Therefore, the spatial distributions of coldand PAH components are more extended toward the haloregion than those of hot and warm dust components. F / F – F / F To obtain a global view of the variation in the shape of thecold dust component in the local SEDs, sub-mm colors area simple and useful quantitative tool to avoid potential con- tamination from the warm dust component (Bendo et al.2012, 2015). As demonstrated by Boselli et al. (2012), wecreated F / F and F / F color maps. The errors onthe flux densities ( σ ) were calculated from ( σ + σ ) / ,where σ sys and σ stat are systematic and statistical uncertain-ties, respectively.In Figs. 3a and 3b that were regridded with the originalpixel size of the SPIRE 500 µ m data (14 ′′ ) for a display pur-pose, values of the F / F show almost constant over thewhole region, while F / F values are higher in the south-ern halo region, in particular the western arm, than in thenorthern halo region. Those trends can clearly be confirmedin the sub-mm color-color diagram. The filled circles in Fig. 4represent the observed color-color diagram at each pixel ofthe F / F and F / F maps (Figs. 3c and 3d). Thesolid lines are the model-predicted color-color diagram basedon a single-component modified blackbody model with theemissivity power-law index (hereafter referred to as the emis-sivity index) of 1.0 (leftmost line), 1.5, 2.0, and 2.5 (right-most line). The constant temperature points are connected MNRAS000
7, respectively. The PSF shown in the panels (a) and (c) is matched to theSPIRE 500 µ m-band PSF. The detection threshold is set to be the 5 σ level for each band image. The contours superimposed on theimages are the same as those in Fig. 2. also from the halo; Onaka et al. (2010) detected the PAHemission from the western arm. Overall spatial distribu-tions in the 7 µ m and 11 µ m images are in good agree-ment with those in the 160–500 µ m images which tracethe spatial distribution of the cold dust component (20–30 K, R´emy-Ruyer et al. 2013; Lianou et al. 2014), associ-ated mainly with the old stellar population. The fact mayindicate that PAHs coexist with cold dust. The IRC 15 µ mand 24 µ m images trace emission from both hot (100–200 K)and warm (40–60 K) dust components (Lianou et al. 2014;Galliano et al. 2003), associated mainly with massive star-forming regions. Therefore, the spatial distributions of coldand PAH components are more extended toward the haloregion than those of hot and warm dust components. F / F – F / F To obtain a global view of the variation in the shape of thecold dust component in the local SEDs, sub-mm colors area simple and useful quantitative tool to avoid potential con- tamination from the warm dust component (Bendo et al.2012, 2015). As demonstrated by Boselli et al. (2012), wecreated F / F and F / F color maps. The errors onthe flux densities ( σ ) were calculated from ( σ + σ ) / ,where σ sys and σ stat are systematic and statistical uncertain-ties, respectively.In Figs. 3a and 3b that were regridded with the originalpixel size of the SPIRE 500 µ m data (14 ′′ ) for a display pur-pose, values of the F / F show almost constant over thewhole region, while F / F values are higher in the south-ern halo region, in particular the western arm, than in thenorthern halo region. Those trends can clearly be confirmedin the sub-mm color-color diagram. The filled circles in Fig. 4represent the observed color-color diagram at each pixel ofthe F / F and F / F maps (Figs. 3c and 3d). Thesolid lines are the model-predicted color-color diagram basedon a single-component modified blackbody model with theemissivity power-law index (hereafter referred to as the emis-sivity index) of 1.0 (leftmost line), 1.5, 2.0, and 2.5 (right-most line). The constant temperature points are connected MNRAS000 , 000–000 (0000)
T. Suzuki et al.
Figure 4. F / F – F / F diagram. The filled circle represents the observed color-color diagram at each pixel shown in Figs. 3cand 3d, while the solid lines are the color-color diagram based on a single-component modified blackbody model defined as ν β c B ν ( T c ) for β c =1.0–2.5 with a step of 0.5. The color of each data point is the same as that of each pixel in Fig. 3d. The constant temperature pointsare connected by the dashed lines for temperatures from 20 (bottom line) to 70 K (top line) with a step of 10 K. A typical error bar isshown in the lower right corner. by the dashed lines for temperatures from 20 K (bottom line)to 70 K (top line) with a step of 10 K. The color-color dia-gram suggests that the plots in and around the disk regiontend to distribute along the direction of the temperaturechange with β ∼ , while those in the IR-halo region tendto distribute relatively along the constant temperature lines(dashed lines) rather than the constant β lines; the westernarm region shows higher β ( ∼ ), while the northern haloregion of the SSCs shows lower β ( < ∼ ). Since it is possiblethat color values are affected by noise as a limitation of thecolor-color analysis, further confirmation of the β variationis required by performing SED fitting. Spectral decomposition analysis on a pixel-by-pixel basisprovides investigations on the spatial distributions of dustproperties such as dust temperature and emissivity index.An individual SED constructed from the nine-band fluxesat each pixel is reproduced by a double-component modifiedblackbody plus a PAH model expressed as F ν , IR = A PAH F ν , PAH + A c ν β c B ν ( T c )+ A w ν β w B ν ( T w ) (cid:26) + (cid:16) ν c ν (cid:17) α e − ( ν c ν ) (cid:27) , (1)where T c , T w , β c , β w , A c , A w , A PAH , and B ν ( T ) are the tem-peratures of cold and warm dust, the dust emissivity in-deces of cold and warm dust, the amplitudes of cold dust,warm dust, PAH components, and the Planck function, re-spectively. The second term in the warm dust componentis the analytic approximation of dust emission with dif-ferent dust temperatures assuming a power-law tempera- ture distribution to take the hot dust component into ac-count in mid-IR wavelengths (Casey 2012). The power-lawturnover frequency ν c defined by Casey (2012) is a functionof the mid-IR power-law slope α and T w . The flux densityof the PAH component, F PAH ( ν ) , is calculated as describedin Suzuki et al. (2010) and is based on the PAH parame-ters taken from Li & Draine (2001) and Draine & Li (2007)by assuming the PAH size distribution ranging from 3.55 to300 ˚A, the fractional ionization and the temperature prob-ability distribution for the typical diffuse ISM with the in-terstellar radiation field in the solar neighborhood. PAHswith sizes larger than 15 ˚A contribute to ∼ µ m contin-uum emission (Draine & Li 2007). Since very small grains(VSGs) which are stochastically heated by absorbed pho-tons contribute to ∼ µ m continuum emission, the PAHcomponent in Eq. (1) takes the VSG emission into account. Since Fig. 4 indicates the β c variation over the disk-IR haloregion, β c is set to be free, while β w is assumed to be constantover the disk-IR halo region. To confirm that the assumptiondoes not affect the result of the β c variation, we performedpixel-by-pixel SED fitting for each of β w =1.0, and 2.0. Ingeneral, when T d and β are treated as fitting parameters,a β – T d anti-correlation is expected (Tabatabaei et al. 2014,and references therein). The anti-correlation mainly comesfrom β – T d degeneracy; for a set of data points, a high good-ness of fit can be obtained for different ( β , T d ) pairs. To avoidthe artificial anti-correlation, one fitting cycle consists of thefollowing two steps: the first fitting step with fixed T c andfree β c , and second fitting step with free T c and fixed β c . Aseries of fitting cycles for each pixel was performed and was MNRAS , 000–000 (0000) ust properties in galactic winds from NGC 1569 Figure 5.
SEDs at pixel (36 . ′′
7) positions of (a) SSCs, (b) south halo, and (c) north halo regions as denoted by open circles in Fig. 6a.At each panel, the solid thick line shows the best-fit model with β w = . , which is described in Eq. (1). The cold dust, warm dust, andPAH components are indicated by dashed, dash-dotted, and dotted lines, respectively. The PSF is matched to the SPIRE 500 µ m-bandPSF. completed when χ converges with the lowest value. Thus,at each pixel, the six-free parameters ( T c or β c , T w , A c , A w , A PAH , and α ) are derived from the nine bands.To minimize the probability of falling into local minimaof the merit function, Σ ( F ν model − F ν obs ) / σ , where σ is theflux uncertainty, we generated uniform samples of ran-dom sets of the six parameters. As for each parameter, therandom value ranges within 5% from the input value whichis obtained from the previous fitting step. For the first fit-ting cycle, input values of T c and T w were set to be the valuesobtained from SED fitting of the whole galaxy with β c = . and β w = . or 2.0. From each random set of the six pa-rameters, F ν , IR at each wavelength was calculated. For eachpixel, the initial parameter set which provides the fluxesmost closest to those observed was chosen by maximizingthe likelihood. Then, using the initial parameter set, pixel-by-pixel SED fitting was performed to minimize χ . Colorcorrections to the IRC 15 µ m, 24 µ m, PACS and SPIRE-band fluxes were iteratively performed with the assumptionof a modified blackbody spectrum for each fitting cycle. Asa result, the six parameters with β w =2.0 better reproducethe observed fluxes with pixel-averaged χ ν ( d . o . f = ) of 0.6.Figure 5 shows SEDs at pixel positions of (a) SSCs, (b) southhalo, and (c) north halo regions, which are denoted by opencircles in Fig. a, together with the best-fit model. β c , T c , and T w Figure 6 shows the spatial distributions of β c (top), T c (mid-dle), and T w (bottom). The panels in the left and right columns show their maps obtained with β w of 1.0, and 2.0,respectively. Moreover, Fig. 7 shows the maps with a higherspatial resolution obtained from SED fitting using the eight-band images (IRC 7 µ m–SPIRE 350 µ m), whose PSFs werematched to the SPIRE 350 µ m-band PSF. Uncertainties in β c , T c , and T w are 4%, 5%, and 6%, respectively. From Figs. 6and 7, it is confirmed that overall spatial distributions of β c , T c , and T w are independent of the β w values.As suggested by the color-color diagram, Fig. 6 showsthat high β c values are observed around the western armand SSCs regions. In fact, the pixel-by-pixel comparison of β c in Fig. 8 shows that the β c values obtained from SEDfitting ( β SED ) are in agreement with those from the F / F color ( β color ), which are obtained with a single-componentmodified blackbody model. The result suggests that β SED isnot influenced by the warm dust component. The β c mapwith a higher spatial resolution (25 . ′′
2, Fig. 7) reveals that β c shows almost constant value of ∼ β c values as high as ∼ are found around the SSCsand in regions towards the western arm. As for the IR-haloregion, enhanced β c values ( ∼ ) are clearly observed alongthe western arm in superbubble A, while such high β c valuesare not observed in other three superbubbles; the lowest β c values are observed in superbubble B.The T w map in Fig. 7 is in agreement with the H α mapin the disk region. High- T w values are observed near currentstar-forming regions such as the SSCs, while the T c mapshows a more uniform distribution than the T w map; high- T c values are found around SSCs. Those results are consistentwith the spatial distribution of T c shown in Lianou et al.(2014). In the IR-halo region, both T c and T w are found to MNRAS , 000–000 (0000)
T. Suzuki et al.
Figure 6.
Spatial distributions of β c (top), T c (middle), and T w (bottom). In each column, left and right panels show their maps with β w = . and 2.0, respectively. The open circles in the panel (a) represent pixel positions of SEDs in SSCs, south halo, and north haloregions (see Fig. 5). The detection threshold is set to be the 5 σ level for each band image. The contours superimposed on the images arethe same as those in Fig. 2. The PSF size in SPIRE 500 µ m is shown in the lower left-hand corner on the panel (a). The pixel size is thesame as the original one of the SPIRE 500 µ m data (36 . ′′ be systematically higher on the northern side than on thesouthern side. Those high-temperature dust grains tend tobe spatially matched adjacent to the H α filaments of super-bubbles B and E.Figure 9 shows the β c – T d relation obtained from Figs. 6band 6d (filled square) and from individual dwarf galaxiesobserved by Herschel (R´emy-Ruyer et al. 2013, 2015). Al-though R´emy-Ruyer et al. (2013) investigated β and T d val-ues for the dwarf galaxy samples, they used the data setsprocessed with an older calibration version than our version.Thus, we reanalyzed β and T d values for the samples us- ing flux densities shown in R´emy-Ruyer et al. (2015), whichwere calibrated with the latest version, by applying the sameSED fitting procedure as described in R´emy-Ruyer et al.(2013). The distribution of our data points overlaps wellwith the β - T d anti-correlation from individual dwarf galax-ies observed by Herschel (R´emy-Ruyer et al. 2013, 2015).Furthermore, our data plots are in agreement with that forthe whole galaxy of NGC 1569 (black-filled circle).To check if the observed β c – T d relation comes from theartificial anti-correlation, we compared input ( β c , in , T c , in )with output sets of ( β c , out , T c , out ) by SED fitting of the mock MNRAS , 000–000 (0000) ust properties in galactic winds from NGC 1569 Figure 7.
Same as Fig. 6, but the maps obtained from SED fitting using the eight-band images (IRC 7 µ m–SPIRE 350 µ m) to showthe maps with a higher spatial resolution matched to the SPIRE 350 µ m-band PSF (25 . ′′ µ m data (10 ′′ ). data sets. We generated uniform samples of 2000 randomsets of the seven-SED parameters; each parameter value isuniformly distributed within the range of 5% of the originalbest-fit parameter value to construct realistic mock SEDs.Then, each random set of the seven-SED parameters is usedto synthesize flux densities for β w = . given by Eq. (1) ateach wavelength. Finally, the same SED fitting procedureas described in Sect. 3.3.1 was applied to the mock datasets. Figure 10 shows output/input ratio of β c and T c aftertaking average in each β c bin (10 bins for the β c range of 1.0–2.0). In this figure, although the artificial anti-correlation isconfirmed, the range of both β c and T c variations is within ∼ %. Therefore, the observed β c – T d relation is not signifi- cantly influenced by the artificial anti-correlation. Moreover,the enhanced β c values reflect variations in the shape of thecold dust component in observed SEDs. β c values along the western arm H α observations reveal a number of filaments excited byshocks, which are associated with the four expanding su-perbubbles (Westmoquette et al. 2008). Moreover, the su-perbubbles are filled both with dust grains (see Fig. 2) andhot plasma (Martin et al. 2002). Among those H α filaments, MNRAS000
Same as Fig. 6, but the maps obtained from SED fitting using the eight-band images (IRC 7 µ m–SPIRE 350 µ m) to showthe maps with a higher spatial resolution matched to the SPIRE 350 µ m-band PSF (25 . ′′ µ m data (10 ′′ ). data sets. We generated uniform samples of 2000 randomsets of the seven-SED parameters; each parameter value isuniformly distributed within the range of 5% of the originalbest-fit parameter value to construct realistic mock SEDs.Then, each random set of the seven-SED parameters is usedto synthesize flux densities for β w = . given by Eq. (1) ateach wavelength. Finally, the same SED fitting procedureas described in Sect. 3.3.1 was applied to the mock datasets. Figure 10 shows output/input ratio of β c and T c aftertaking average in each β c bin (10 bins for the β c range of 1.0–2.0). In this figure, although the artificial anti-correlation isconfirmed, the range of both β c and T c variations is within ∼ %. Therefore, the observed β c – T d relation is not signifi- cantly influenced by the artificial anti-correlation. Moreover,the enhanced β c values reflect variations in the shape of thecold dust component in observed SEDs. β c values along the western arm H α observations reveal a number of filaments excited byshocks, which are associated with the four expanding su-perbubbles (Westmoquette et al. 2008). Moreover, the su-perbubbles are filled both with dust grains (see Fig. 2) andhot plasma (Martin et al. 2002). Among those H α filaments, MNRAS000 , 000–000 (0000) T. Suzuki et al.
Figure 8.
Relation between β c from the F / F color ( β color ) and β c from the SED fitting ( β SED ) with β w = . (circle) and 2.0 (square)cases. The β c values are obtained based on the SPIRE 500 µ m-band PSF with a pixel size of 36 . ′′
7. A typical error bar is shown in thelower right corner.
Figure 9.
Pixel-by-pixel based β c – T c relation (filled square) overplotted on the relation for individual dwarf galaxies from the DwarfGalaxy Survey with Herschel (filled circles). The filled square is obtained from Figs. 6(b) and (d) with the SPIRE 500 µ m-band PSF. β c > ∼ . is found in the western arm. The black-filled circle denotes the plot for the whole region of NGC 1569. dust grains associated with the western arm clearly show en-hanced β c values. Why are such grains observed along thewestern arm only?Theoretically, the Lorentz model gives β =2.0 in far-IR and sub-mm region for perfect ionic crystal grains suchas crystalline silicate dust. Meny et al. (2007) proposed amodel for dust emission in far-IR and sub-mm region basedon physical properties of amorphous material (e.g., amor-phous silicate). They introduced a disordered charge distri-bution combined with the presence of two level tunneling states to explain a wavelength and temperature-dependent β . The model with standard parameters well reproducedGalactic dust emission that mainly comes from amorphoussilicate grains (Paradis et al. 2011); the model with the pa-rameters predicts β < . for the dust temperature of ∼ K,which is the average T c in NGC 1569.The fact that β c ∼ is continuously observed from theregion around the SSCs to the western arm may attributeto starburst activities in the SSCs. Figure 11 shows the β c map together with HI integrated contours (Walter et al. MNRAS , 000–000 (0000) ust properties in galactic winds from NGC 1569 Figure 10.
Output/input ratio of β c and T c based on SED fitting of the mock data sets with β w = . . Each filled circle shows the averageof ( β c , out / β c , in , T c , out / T c , in ) sets in each β c bin: 10 bins for the β c range of 1.0–2.0. Figure 11.
Same as Fig. 7b but different color scale to show regions with β c ∼ . The white contours superposed on the image show theintegrated HI (Walter et al. 2008), while the gray contours are the same as those in Fig. 2. Two cross marks are the positions of SSC-A(right) and B (left). β c ∼ are distributedaround the SSCs (two cross marks), and the region isclearly confined by two dense HI clouds called as the HIridge (Israel & van Driel 1990; M¨uhle et al. 2005). Giventhat the SSCs are located in between these two dense HIclouds, as pointed out by Johnson et al. (2012), the SSCsare likely to be formed out of the HI ridge that used toconnect these two dense HI clouds. Those spatial coinci-dences may indicate that dust grains around the SSCs arecrystalline ones produced by massive stars originating fromstarburst activities as have been found in ultra-luminous in- frared galaxies (Spoon et al. 2006), since the latest starburstphase ended 5–10 Myr ago (Angeretti et al. 2005), whichis shorter than the crystalline-to-amorphous conversiontimescale of 40–70 Myr due to cosmic-ray hits (Spoon et al.2006; Kemper et al. 2011). This might indicate that crys-talline grains injected by massive stars are blown away alongthe HI ridge and thus the western arm. MNRAS000
Same as Fig. 7b but different color scale to show regions with β c ∼ . The white contours superposed on the image show theintegrated HI (Walter et al. 2008), while the gray contours are the same as those in Fig. 2. Two cross marks are the positions of SSC-A(right) and B (left). β c ∼ are distributedaround the SSCs (two cross marks), and the region isclearly confined by two dense HI clouds called as the HIridge (Israel & van Driel 1990; M¨uhle et al. 2005). Giventhat the SSCs are located in between these two dense HIclouds, as pointed out by Johnson et al. (2012), the SSCsare likely to be formed out of the HI ridge that used toconnect these two dense HI clouds. Those spatial coinci-dences may indicate that dust grains around the SSCs arecrystalline ones produced by massive stars originating fromstarburst activities as have been found in ultra-luminous in- frared galaxies (Spoon et al. 2006), since the latest starburstphase ended 5–10 Myr ago (Angeretti et al. 2005), whichis shorter than the crystalline-to-amorphous conversiontimescale of 40–70 Myr due to cosmic-ray hits (Spoon et al.2006; Kemper et al. 2011). This might indicate that crys-talline grains injected by massive stars are blown away alongthe HI ridge and thus the western arm. MNRAS000 , 000–000 (0000) T. Suzuki et al.
In the IR-halo region, spatial distributions of T c and T w areasymmetric with respect to the disk. Two possible heatingsources are considered. One is radiative heating by stars dis-tributed over the disk. In this case, the observed dust tem-peratures should monotonically decrease away from the disk.Therefore, the observed dust temperatures and their distri-butions in the IR-halo region cannot be explained solely byradiative dust heating. This means that an additional heat-ing source is required.The other heating source is collisional heating mainlywith electrons in a hot plasma; the possibility is alsopointed out for dust heating in the halo region of NGC 253by Kaneda et al. (2009b). Dwek (1987) presented a detailedanalysis of collisional heating of dust grains under variousplasma conditions with temperatures above ∼ K. Ac-cording to Dwek (1987), dust temperatures due to collisionalheating were calculated by taking the following conditionsinto account; for the calculation of the collisional heatingrate of dust grains, a collision partner is mainly electrons.As for the cooling rate of dust grains, the dust absorptioncoefficient is applied for silicate and carbonaceous grainswith their sizes of 0.01–1.0 µ m. Since the average numberdensity and temperature of electrons in the hot plasma are0.035 cm − and . × K (Ott et al. 2005), respectively,the expected dust temperature ranges from 15 to 30 K dueto collisional heating. Thus, it is likely that the collisionalheating process significantly contributes to dust heating es-pecially in the northern IR-halo region.
AKARI and
Herschel images at wavelengths from 7 µ mto 500 µ m show a diffuse IR emission extending from thegalactic disk into the halo region. The most prominent fila-mentary structure seen in the diffuse IR emission is spatiallyin good agreement with the western arm seen in the H α . Thespatial distribution of the F / F map shows high valuesin regions around the SSCs and towards the western arm,which are not found in the F / F map. The color-color di-agram of F / F – F / F indicates enhanced β c in thoseregions. From a spectral decomposition analysis on the pixel-by-pixel basis, we obtained β c , T c , and T w maps; the β c mapshows values ranging from ∼ to ∼ over the whole galaxy.In particular, high β c values of ∼ are observed in the re-gions indicated by the color-color diagram. As for the T c and T w maps, those show high temperatures on the northern sidethan on the southern side in the IR-halo region.Since the average cold dust temperature in NGC 1569is ∼ K, β c < . in the far-IR and sub-mm region theo-retically suggests thermal emission from amorphous grains,while β c = . suggests that from crystal grains. Given thatthe enhanced β c regions are spatially confined by the HIridge that is considered to be a birthplace of the SSCs, thespatial coincidences may indicate that dust grains aroundthe SSCs are crystalline ones injected by massive stars orig-inating from starburst activities as have been found in ultra-luminous infrared galaxies and that those grains are blownaway along the HI ridge and thus the western arm.The observed asymmetric temperature distribution with respect to the disk cannot be explained solely by radia-tive dust heating by stars distributed over the disk. Giventhat the presence of the hot plasma in the IR-halo region, itis likely that the collisional heating process significantly con-tributes to dust heating especially in the northern IR-haloregion. ACKNOWLEDGMENTS
This research is based on observations with AKARI, a JAXAproject with the participation of ESA.PACS has been developed by a consortium of insti-tutes led by MPE (Germany) and including UVIE (Austria);KU Leuven, CSL, IMEC (Belgium); CEA, LAM (France);MPIA (Germany); INAF-IFSI/OAA/OAP/OAT, LENS,SISSA (Italy); IAC (Spain). This development has been sup-ported by the funding agencies BMVIT (Austria), ESA-PRODEX (Belgium), CEA/CNES (France), DLR (Ger-many), ASI/INAF (Italy), and CICYT/MCYT (Spain).SPIRE has been developed by a consortium of insti-tutes led by Cardiff University (UK) and including Univ.Lethbridge (Canada); NAOC (China); CEA, LAM (France);IFSI, Univ. Padua (Italy); IAC (Spain); Stockholm Observa-tory (Sweden); Imperial College London, RAL, UCL-MSSL,UKATC, Univ. Sussex (UK); and Caltech, JPL, NHSC,Univ. Colorado (USA). This development has been sup-ported by national funding agencies: CSA (Canada); NAOC(China); CEA, CNES, CNRS (France); ASI (Italy); MCINN(Spain); SNSB (Sweden); STFC, UKSA (UK); and NASA(USA).This work made use of THINGS, ’The HI nearby GalaxySurvey’ (Walter et al. 2008).
REFERENCES
Aguirre A., Schaye J., Quataert E., 2001, ApJ, 561, 550Angeretti L., Tosi M., Greggio L., Sabbi E., Aloisi A., LeithererC., 2005, AJ, 129, 2203Aniano G., Draine B. T., Gordon K. D., Sandstrom K., 2011,PASP, 123, 1218Arimatsu K., et al., 2011, PASP, 123, 981Arp H., Sandage A., 1985, AJ, 90, 1163Bendo G. J., et al., 2012, MNRAS, 419, 1833Bendo G. J., et al., 2015, MNRAS, 448, 135Boselli A., et al., 2012, A&A, 540, A54Casey C. M., 2012, MNRAS, 425, 3094Cazzoli S., Arribas S., Colina L., Piqueras-L´opez J., Bellocchi E.,Emonts B., Maiolino R., 2014, A&A, 569, A14Draine B. T., Li A., 2007, ApJ, 657, 810Dwek E., 1987, ApJ, 322, 812Erb D. K., Quider A. M., Henry A. L., Martin C. L., 2012, ApJ,759, 26Galliano F., Madden S. C., Jones A. P., Wilson C. D., BernardJ.-P., Le Peintre F., 2003, A&A, 407, 159Griffin M. J., et al., 2010, A&A, 518, L3Heckman T. M., Dahlem M., Lehnert M. D., Fabbiano G.,Gilmore D., Waller W. H., 1995, ApJ, 448, 98Hodge P. W., 1974, ApJ, 191, L21Hunter D. A., Elmegreen B. G., 2004, AJ, 128, 2170Hunter D. A., Hawley W. N., Gallagher III J. S., 1993, AJ,106, 1797Hunter D. A., O’Connell R. W., Gallagher J. S., Smecker-HaneT. A., 2000, AJ, 120, 2383 MNRAS , 000–000 (0000) ust properties in galactic winds from NGC 1569 Israel F. P., van Driel W., 1990, A&A, 236, 323Johnson M., 2013, AJ, 145, 146Johnson M., Hunter D. A., Oh S.-H., Zhang H.-X., Elmegreen B.,Brinks E., Tollerud E., Herrmann K., 2012, AJ, 144, 152Kaneda H., Koo B. C., Onaka T., Takahashi H., 2009a,Advances in Space Research, 44, 1038Kaneda H., Onaka T., Suzuki T., Takahashi H., Yamagishi M.,2009b, in T. Onaka, G. J. White, T. Nakagawa, & I. Ya-mamura ed., Astronomical Society of the Pacific ConferenceSeries Vol. 418, AKARI, a Light to Illuminate the Misty Uni-verse. pp 197–+Kaneda H., et al., 2010, A&A, 514, A14Kemper F., Markwick A. J., Woods P. M., 2011, MNRAS,413, 1192Li A., Draine B. T., 2001, ApJ, 554, 778Lianou S., Barmby P., R´emy-Ruyer A., Madden S. C., GallianoF., Lebouteiller V., 2014, MNRAS, 445, 1003Madden S. C., et al., 2013, PASP, 125, 600Martin C. L., 1998, ApJ, 506, 222Martin C. L., Kobulnicky H. A., Heckman T. M., 2002, ApJ,574, 663Meny C., Gromov V., Boudet N., Bernard J.-P., Paradis D.,Nayral C., 2007, A&A, 468, 171Micelotta E. R., Jones A. P., Tielens A. G. G. M., 2010, A&A,510, A36+Montier L. A., Giard M., 2004, A&A, 417, 401M¨uhle S., Klein U., Wilcots E. M., H¨uttemeister S., 2005, AJ,130, 524Murakami H., et al., 2007, PASJ, 59, 369Onaka T., et al., 2007, PASJ, 59, 401Onaka T., Matsumoto H., Sakon I., Kaneda H., 2010, A&A,514, A15Ott J., Walter F., Brinks E., 2005, MNRAS, 358Paradis D., Bernard J.-P., M´eny C., Gromov V., 2011, A&A,534, A118Pilbratt G. L., et al., 2010, A&A, 518, L1Poglitsch A., et al., 2010, A&A, 518, L2R´emy-Ruyer A., et al., 2013, A&A, 557, A95R´emy-Ruyer A., et al., 2015, A&A, 582, A121Roussel H., et al., 2010, A&A, 518, L66Spoon H. W. W., et al., 2006, ApJ, 638, 759Suzuki T., Kaneda H., Onaka T., Nakagawa T., Shibai H., 2010,A&A, 521, A48+Tabatabaei F. S., et al., 2014, A&A, 561, A95Walter F., Brinks E., de Blok W. J. G., Bigiel F., Kennicutt Jr.R. C., Thornley M. D., Leroy A., 2008, AJ, 136, 2563Westmoquette M. S., Smith L. J., Gallagher J. S., 2008, MNRAS,383, 864Yoshida M., Kawabata K., Ohyama Y., 2011, PASJ, 63, 493Zu Y., Weinberg D. H., Dav´e R., Fardal M., Katz N., Kereˇs D.,Oppenheimer B. D., 2011, MNRAS, 412, 1059This paper has been typeset from a TEX/L A TEX file prepared bythe author.MNRAS000