Enhanced Warm H2 Emission in the Compact Group Mid-Infrared "Green Valley"

2 MB·English

by  M. E. Cluver

Accepted to ApJ:17 January2013
PreprinttypesetusingLATEXstyleemulateapjv.03/07/07
ENHANCED WARM H EMISSION IN THE COMPACT GROUP MID-INFRARED “GREEN VALLEY”
2
M.E. Cluver1,2, P.N. Appleton3, P. Ogle1, T.H. Jarrett4, J. Rasmussen5, U. Lisenfeld6, P. Guillard1, L.
Verdes-Montenegro7, R. Antonucci8, T. Bitsakis9, V. Charmandaris9,10,11, F. Boulanger12, E. Egami13, C.K. Xu2,
M.S. Yun14
Accepted to ApJ: 17 January 2013
ABSTRACT
3
WepresentresultsfromaSpitzer,mid-infraredspectroscopystudyofasampleof74galaxieslocated
1
in 23 Hickson Compact Groups, chosen to be at a dynamically-active stage of Hi depletion. We
0
find evidence for enhanced warm H emission (i.e. above that associated with UV excitation in
2 2
star-forming regions) in 14 galaxies (∼20%), with 8 galaxies having extreme values of L(H S(0)-
2
n S(3))/L(7.7µmPAH),inexcessof0.07. Suchemissionhasbeenseenpreviouslyinthe compactgroup
a HCG92(Stephan’sQuintet),andwasshowntobeassociatedwiththedissipationofmechanicalenergy
J
associatedwithalarge-scaleshockcausedwhenonegroupmembercollided,athighvelocity,withtidal

9 debrisinthe intragroupmedium. Similarly,shockexcitationorturbulentheating islikely responsible
1
forthe enhancedH emissioninthe compactgroupgalaxies,since othersourcesofheating (UVorX-
2

rayexcitationfromstarformationorAGN)areinsufficientto accountforthe observedemission. The
]
O group galaxies fall predominantly in a region of mid-infrared color-color space identified by previous
studies as being connected to rapid transformations in HCG galaxy evolution. Furthermore, the
C
majority of H -enhanced galaxies lie in the optical “green valley” between the blue cloud and red-
2
h. sequence, and are primarily early-type disk systems. We suggest that H2-enhanced systems may
p represent a specific phase in the evolution of galaxies in dense environments and provide new insight
- into mechanisms which transform galaxies onto the optical red sequence.
o
Subject headings: galaxies: groups : general – galaxies: evolution – galaxies: interactions – galaxies:
r
t ISM – galaxies: intergalactic medium – infrared: galaxies
s
a
[
1. INTRODUCTION ter galaxies have been “pre-processed” in groups, and

1 then subsequently assimilated into larger systems (e.g.
Compact Groups are key laboratories for studying
v Cortese et al. 2006). Simulations show that spirals in a
morphological transformations as they represent the
9 group environment are strongly influenced by repetitive
highest density enhancements outside of clusters, and
4
slow encounters, building bulge mass as gas is funnelled
5 their relatively low velocity dispersions prolong grav-
intothecentralregions,transformingthemintoS0galax-
4 itational interactions (Hickson et al. 1992). Disentan-
ieswithyoung,metal-richstellarpopulationsintheirin-
. glingthemechanismsthatinfluencegalaxyevolutionare
1 ner bulges (Bekki & Couch 2011). In addition, galaxy
made more challenging with growing evidence that clus-
0 interactionsingroupsmaydifferfromisolatedbinaryin-
3 1Spitzer ScienceCenter, IPAC, CaliforniaInstitute ofTechnol- teractionsbecausetheycanexhibitabroaderrangeofbe-
1 ogy,Pasadena, CA91125,USA haviours, including tidal stripping and interactions with
v: 2ARCSuperScienceFellow,AustralianAstronomicalObserva- theintra-groupmedium(IGM;seee.g. Allen & Sullivan
tory,POBox915,NorthRyde,NSW1670,Australia
Xi 3NASA Herschel S Center, California Institute of Technology, 1980).
Hickson (1982) identified a uniform sample of 100
Pasadena,CA91125,USA
r 4Department of Astronomy, University of Cape Town, Private nearbycompactgroups(HicksonCompactGroups;here-
a BagX3,Rondebosch, 7701,SouthAfrica after HCG) using the Palomar Sky Survey, and applied
5Dark Cosmology Centre, Niels Bohr Institute, University of
thecriterionof4ormorememberswithina3magnitude
Copenhagen, Juliane Maries Vej 30, DK-2100 Copenhagen, Den-
mark range (δmB) that also satisfied an isolation constraint.
6DepartmentodeF`ısicaTeo´ricaydelCosmos,FacultaddeCien- These groups have been the subject of extensive follow-
cias,UniversidaddeGranada,Spain up and study. Radial velocities (Hickson et al. 1992)
7Instituto de Astrof´ısica de Andaluc´ıa (IAA/CSIC), Apdo.
haveledtotheidentificationoftrueassociationsofgalax-
3004,18080Granada,Spain
8UniversityofCaliforniaSantaBarbara,DepartmentofPhysics, ies, with 92 groups consisting of at leastthree accordant
SantaBarbara,CA93106, USA members. That many of the groups are real physical
9DepartmentofPhysics,UniversityofCrete,GR-71003,Herak- associations is further attested by the presence of hot
lion,Greece
10IESL/Foundation for Research & Technology-Hellas, GR- intragroup gas in many of them (Ponman et al. 1996;
71110,Heraklion,Greece Desjardins et al. 2012). Signs of interactions within
11Chercheur Associ´e, Observatoire de Paris, F-75014, Paris, these groups include peculiar rotation curves and dis-
France turbed morphologies of group members (Rubin et al.
12Institute d’Astrophysique Spatiale, Universite Paris Sud 11,
1991; Mendes de Oliveira & Hickson 1994), as well as
Orsay,France
13Steward Observatory, University of Arizona, 933 N. Cherry the presence of intragroup light (Da Rocha et al. 2008).
Avenue,Tucson,AZ85721,USA Hi deficiency in compact group galaxies has
14 Department of Astronomy, University of Massachusetts,
long been suspected (Williams & Rood 1987;
Amherst,MA01003, USA
Huchtmeier 1997) and the Hi study of 72 HCGs
2 Cluver et al.
led Verdes-Montenegro et al. (2001) to propose an galaxiesin the gapregionwith similarities to that found
evolutionary sequence in which compact group galaxies for the Coma Infall region. Accelerated transformation,
become increasingly deficient in neutral hydrogen. possibly preceded by enhanced star formation in some
Multiple tidal interactions, and possible gas stripping galaxies, is also suggested by the significant bimodality
(via interaction with the IGM) may be the cause of in specific star formation rate (sSFR) seen in the HCG
the observed Hi depletion. More recent observations sample of Tzanavaris et al. (2010).
using the Green Bank Telescope, sensitive to extended, Bitsakis et al. (2011) performed a UV to mid-infrared
faint Hi emission, have revealed a diffuse Hi com- analysisof a sample of 32 HicksonCompact Groups and
ponent in all the groups studied (Borthakur et al. find that dynamically “old” groups (containing >25%
2010), explaining in part the “missing Hi” reported early-typegalaxies)aremorecompactanddisplayhigher
by Verdes-Montenegro et al. (2001). Groups con- velocity dispersions compared to dynamically “young”
taining galaxies with the largest Hi deficiencies are systems(containing>75%late-typegalaxies). Late-type
found to have a more massive diffuse-Hi component galaxiesindynamicallyoldgroupsarefound,onaverage,
(Borthakur et al. 2010). These observations have led to have higher stellar mass and lower sSFR, attributed
to the conclusion that Hi-deficient group galaxies lose to a faster build in stellar mass due to past interactions
Hi into the IGM, primarily through tidal interactions. comparedtothedynamicallyyounggroups. Theirstudy
In the most evolved groups, gravitational heating may also finds that the majority (73%) of compact group
eventually create a hot X-ray emitting medium (see galaxies lie in either the optical “green valley” or the
Ponman et al. 1996; Desjardins et al. 2012). “redsequence”,asdefinedbytheirNUV−rcolors. More
Leon et al.(1998)observedthatgalaxieslocatedinthe thanhalfofthe early-typegalaxiesin dynamically“old”
most compact groups within their sample of 45 HCGs groups were found to be located in the “green valley”
have more molecular gas concentratedin their nuclei, as and these are predominantly (>70%) S0/SB0’s.
expectedfromtheeffectsoftidalforcingwithinthedisk. The AGN (Active Galactic Nuclei) activity of galaxies
However,thelinkbetweenmoleculargasandstarforma- in HCGs are a key consideration; although 46% of the
tion properties in these interacting systems is less obvi- sample of Bitsakis et al. (2011) have optically identified
ous. Studies of far-infrared and CO emission of galaxies AGN, from nuclear spectra, they find no evidence of en-
in HCGs indicate no enhanced star formation, but 20% hanced AGN activity at any stage of group evolution.
of spiral galaxies are tentatively found to be deficient in This is consistent with the findings of Mart´ınez et al.
CO emission compared to isolated and weakly interac- (2010) where the median HCG AGN luminosity corre-
tion systems (Verdes-Montenegro et al. 1998). A recent, spondstoalowluminosityAGN(LLAGN),likelycaused
moreextensiveCOstudyofHCGsfindsthatthespecific by gas depletion resulting in relatively low accretion
star formation rate (SFR per unit stellar mass or sSFR) rates, and also Rasmussen et al. (2008), where the fre-
islowerinHi-andCO-deficient(ascomparedtoisolated quencyandstrengthofnuclearX-rayactivityin8groups
systems) HCG galaxies, but that the star formation ef- showed no clear correlation with the dynamical state of
ficiency (SFR per unit cold H mass) in these galaxies the group, as measured by either diffuse X-ray emission
2
appears unaffected (Mart´ınez-Badenes 2012). or Hi-deficiency.
The question of whether galaxies are stripped by in- Our current paper is motivated by a possible new di-
teraction with a dense medium, or by tidal forces, still agnosticofHCGevolutionwhichusesmid-infraredspec-
remains unclear. X-ray observations of 8 HCGs by troscopy from the Spitzer Space Telescope as a probe
Rasmussen et al. (2008) showed no obvious correlation of the warm molecular gas in galaxies. Our team
between the presence of detectable hot intragroup gas (Appleton et al. 2006; Cluver et al. 2010) discovered
and Hi deficiency. Furthermore, in groups where X-ray powerful(L > 1035 W) mid-infrared molecular hydro-
H2
emitting gas was strongly detected, it was shown that it gem (H ) line emission from an intergalactic shock wave
2
wasnotofsufficientdensitytosignificantlystripHifrom in Stephan’s Quintet (SQ; HCG 92). The emission was
the group members, thus calling into question whether found to be spatially associated with a 40kpc-long X-
gasstrippingbyahotX-raymediumisaviablestripping ray and radio-continuum filament believed to be formed
mechanism within compact groups. as a result of a high-speed collision (∼ 1000 kms−1) be-
Observations in the infrared with the Spitzer Space tween a group member and tidal debris from a previous
Telescope have led to potentially new evidence of evo- encounterwithin the group. Inthis case,the highpower
lutionary effects in HCGs. Johnson et al. (2007) ob- of the H relative to both the infrared continuum and
2
serve a correlation in the IRAC color-color diagram of veryfaint PAH (polycyclic aromatichydrocarbon)emis-
HCG galaxies,and argue for an evolutionarysequence – sion, and the close associationwith a known group-wide
fromgroupsdominatedbydustyspiralswith“red”IRAC shockwave,makesastrongcaseforshock-heatingasavi-
colors, to groups containing evolved stellar-dominated ablemechanisminthatcompactgroup(Appleton et al.
galaxieswith“blue”IRACcolors. Thesecolorsappearto 2006;Cluver et al.2010). Modelsdemonstratethatdriv-
correlatewiththedegreeofHidepletion,supportingthe ing a shock into a multi-phase medium, such as a pre-
idea of group evolution. Johnson et al. (2007) discuss existing Hi tidal arm are capable of explaining many of
a “gap” in the IRAC colour space between dusty/gas- the observed properties of the warm H emission in SQ
2
rich and gas-poor galaxies; this apparent absence of in- (Guillard et al.2009). Thegroupmembersareknownto
termediate mid-infrared colours suggests a rapid evolu- be Hi and CO depleted (see Yun et al. 1997; Gao & Xu
tion from gas-richto Hi-depleted systems (Walker et al. 2000), and may be one of the best candidates for hy-
2010). The role of environment is further investigated drodynamic stripping effects as much of the molecular
in Walker et al. (2012) using a sample of 49 compact gas appears to reside in the IGM (Guillard et al. 2012a)
groups. They find a statistically significant deficit of based on deep IRAM CO observations. The discovery
HCG MOHEGs 3
of Hi and molecular gas between galaxies in HCG 92, 40, on the grounds that its X-ray morphology was quite
as well as in the tidal bridge between the Taffy galaxies similarto thatofStephan’s Quintet, althoughits Hide-
(Peterson et al. 2012) – a system which has recently ex- ficiency is higher (log[M(Hi) ]−log [M(Hi) =0.97)
pred obs
periencedahead-oncollisionsimilartothoseexpectedin than most of the groups selected. In addition we added
dense environments–ledto ourpresentstudy ofamuch HCG 55 and 75 as groups with close associations that
larger sample of 23 HCGs with Spitzer. contain signatures of interaction. The final sample of 23
The discovery of a class of powerful H -emitting ra- groups is listed in Table 1; galaxies that are not group
2
dio galaxies, with similar Spitzer IRS spectra to SQ, led members (i.e. with discordant redshifts) are indicated
to the coining of the term MOHEG (MOlecular Hydro- and not included in this analysis.
gen Emission-line Galaxies) and are defined as having
large H to 7.7µm PAH emission ratios, ≥ 0.04, indi-
2 TABLE 1
cating excitation above that expected from UV heating
HCGSample
(Ogle et al. 2010).
In this paper we focus on the excited H properties
2
of a sample of 74 HCG galaxies, in particular those
that show H -enhancement as defined for a MOHEG. Group za HiDeficiencyb Galaxies Designationc
2 Sampled
A subsequent paper will focus on the cold molecular gas
properties of H -enhanced systems through IRAM CO
2
observations and comparisons of the cold versus warm HCG6 0.0379 0.33 4 A,B,C,D⋄
HCG8 0.0545 >0.04 3 A,C,D
H masses (and temperatures of the excited H ). The
2 2 HCG15 0.0228 0.62 3 A,C,D
paper is organised as follows: in Section 2 we outline HCG25 0.0212 0.26 4 B,C†,D,F
the sample chosen for this study, in Section 3 we sum- HCG31 0.0135 0.18 2 A+C•,B
marise the observations and data reduction procedures, HCG40 0.0223 0.97 4 A,B,C,D
HCG44 0.0046 0.69 3 A,B,D
and in Section 4 present results of the excited H2 line HCG47 0.0317 0.28 3 A,B,D
emission survey. Section 5 explores potential sources of HCG54 0.0049 0.49 3 A,B,C
H excitation and in Section 6 we discuss possible links HCG55 0.0526 – 5 A,B,C,D,E†
2 HCG56 0.0270 0.73 4 B,C,D⋆,E
to evolution within compact groups. Section 7 presents HCG57 0.0304 0.86 5 A,B,C⋄,D⋄,E
our discussion, with conclusions summarised in Section HCG62 0.0137 >0.46 3 A,B,C
8. Throughout this paper we assume a cosmology with HCG67 0.0245 0.27 3 A,B,D⋄
Hubble constant Ho = 73kms−1Mpc−1, matter density HHCCGG6785 00..00048106 0.4–8 43 A,AC,,BD,⋄C,E
parameter Ω=0.3, and dark energy density ΩΛ =0.7. HCG79 0.0145 0.41 4 A,B,C,E†
HCG82 0.0362 >0.76 3 A,B,C
2. THESAMPLE HCG91 0.0238 0.24 3 A,C,D
HCG95 0.0396 >0.21 3 A,B‡,C
We have selected 23 compact groups from the HCG HCG96 0.0292 >0.17 3 A,B,C
catalogofHickson (1982),representing25%ofthephys- HCG97 0.0218 0.89 3 A,C,D
ically associated groups, as targets for probing active HCG100 0.0170 0.5 3 A,B,C
transformation. The aim of the project was to search
forevidenceofextendedmolecularhydrogenemissionby a FromNED
selecting compact groups with intermediate Hi deficien- b log[M(Hi)pred]−log[M(Hi)obs];Verdes-Montenegroetal. (2001)
c FromHickson(1982)
cies (like SQ) which were reasoned more likely to be in
† Discordantredshift(fromHicksonetal. 1992)
an active phase of gas stripping. ‡ Discordantredshift(fromIglesias-Pa´ramo&V´ılchez1998)
The sample of compact groups studied by • MergingObject(Gallagheretal.2010)
Verdes-Montenegro et al. (2001) contained 72 ⋆ SLcoverageonly
⋄ LLcoverageonly
groups with Hi deficiencies ranging from -0.8 <
log [M(Hi) ]−log [M(Hi) ] < 1.56. From this
pred obs
list, we selected galaxy groups with an “intermediate
HI deficiency” i.e. defined as having deficiencies in
the range 0 < log[M(Hi) ]−log[M(Hi) ] < 0.9. Although our primary selection criterion is based on
pred obs
This resulted in 50 out of 72 groups with a median Hi the groups exhibiting intermediate Hi-depletion (based
deficiency of 0.48, which is close to the value of 0.49 for on the original definition of Verdes-Montenegro et al.
Stephan’s Quintet. We note that the median deficiency 2001), the sample spans a large range of galaxy proper-
for the full (original) sample is 0.27. However, as shown tiessharedbythemorecompletesamplesofHCGgroups
by Verdes-Montenegro et al. (2001), the spread in the (e.g. Bitsakis et al. 2011; Walker et al. 2012). As shown
distribution is quite large, and our sample selection has in the IRAC color-color diagram (log[f /f ] vs
8.0µm 4.5µm
resulted in effectively clipping the extreme ends of the log[f /f ]; Figure 1a), the intermediate Hi-
5.8µm 3.6µm
distribution for all groups. The term “intermediate Hi depletion in these groups is not biased towards mid-
depletion” should therefore be seen in that context. In infrared blue or red populations, but rather spans the
ordertoformapracticalsampleforobservationwiththe entire range of mid-infrared colour seen in the studies of
IRS on Spitzer, and to further maximize the chances of Bitsakis et al. (2011) and Walker et al. (2012).
finding dynamically active systems like Stephan’s Quin- This color-color space has been shown to separate
tet, we only consideredthose groups that showedvisible late-type, star-forming galaxies (top right) from early-
signs of tidal interaction, specifically disturbed optical type galaxies (bottom left). The dashed box shows
disks and tidal tails, in two or more members. This the underpopulated region found in the smaller sam-
resulted in 20 groups. We added an extra group, HCG ple of HCGs studied by Johnson et al. (2007) and
4 Cluver et al.
(a) (b)
Fig. 1.— a) The IRAC color-color diagram of the galaxies inour sample showing the mid-infraredcolours of the chiefly intermediate
Hi-deficiency groups. The shaded grey region indicates the AGN locus, as defined by Lacyetal. (2004), and the black dashed region
showstheJohnsonetal. (2007)“gap”region. b)AhistogramcomparingtheIRAC[f5.8µm/f3.6µm]colorswithinoursample(blackfilled;
z<0.035galaxiesonlytoavoidcolorsaffectedbyredshiftedspectralfeatures),versusthelargerHCGsampleofBitsakisetal.(2011,grey
filled; also excluding galaxies at z >0.035), as well as the distribution from the Local Volume Limited (LVL) sample of nearby galaxies
of (Daleetal. 2009, red unfilled) after applying a luminosity cut (see text). The black dashed lines indicate galaxies with intermediate
mid-infraredcolours,asdeterminedfromthedistributionofBitsakisetal.(2011).
Walker et al. (2010). Bitsakis et al. (2011) observe a evolved through this intermediate color region.
similar lowering of density of galaxies with intermedi-
3. OBSERVATIONSANDDATAREDUCTION
ate mid-infrared colors; they attribute this distribution
to the natural result of morphologicaltransformation as 3.1. Spitzer IRS Spectroscopy
galaxies evolve from star forming to passively evolving
ThegalaxiesandgroupslistedinTable1weretargeted
systems. We will show later in this paper that galaxies
by the Spitzer IRS instrument (Houck et al. 2004) using
whichfallwithinthisintermediateregionofmid-infrared the low-resolution Short-Low (R ∼ 60−127; 5.2−14.5
color preferentially show signs of shocked molecular hy- µm)andLong-Low(R∼57−126;14.0−38.0µm)mod-
drogen emission. This may support the idea that the ules. ObservationswerecarriedoutaspartofGO-5PID
“gap galaxies” represent a transitional population.
50764 and taken between 2008, June 29 and 2009, Jan-
The mid-infrared color properties of our sample, both
uary 19.
in the context of the larger HCG group population and
Sinceitwasnotknownaprioriwhereinthegroupenvi-
local galaxy populations, is shown in Fig. 1b. Here
ronment shock-excited H emission may be located, the
2
we compare the IRAC colors of our intermediate Hi- primary observations employed a sparse mapping strat-
deficientsample,tothelargerBitsakis et al.(2011)HCG egy–centeringa3-leggrid15onthemostdisturbedmem-
sampleof32groupswhichwasnotselectedfordeficiency. ber of the group (see Figure 2). The scale was adjusted
Also shown are galaxies taken from the Local Volume-
for each group to provide good coverage of the inner
Limited (LVL) sample of Dale et al. (2009), as a com-
group in both modules. The typical coverage was lin-
parison“field” controlsample. The LVLsample consists ear ∼70kpc for SL and ∼180kpc for LL. In addition, a
of256galaxieswithin11Mpcandisdominatedbyspirals
furthertwomembergalaxiesweretargetedinIRS“Star-
and irregulars; we apply a luminosity cut (as motivated ing Mode”, where the target center is placed at the 1
by Walker et al. 2012) of log(L4.5µm[erg/s/Hz])> 27.5 and 2 position along the length of the slit. Because o3f
to compare galaxies with similar characteristics, which 3
the way the IRS performs a staring mode observation
leaves 65 galaxies. As expected from the morphology-
(by sampling first the 2nd order spectrum followed by a
densityrelation(e.g.Dressler1980),theLVLfieldsample
similarobservationinfirst order),this increasesthe cov-
contains very few early-type systems.
erage of the group, providing further information about
By comparisonto the LVL sample, HCG group galax-
potential extended emission in the groups, as well as of-
ies appear to have a well-defined red and blue sequence
tenintersectingbychance(dependingontheroll-angleof
in this color space,with a deficiency of galaxies at inter-
thefocalplaneatthetimeoftheobservations)additional
mediate color not seen in the volume-limited sample of
group members. As a result, the SL and LL slits typi-
nearby galaxies, i.e. between
cally sampled ∼800kpc2 and ∼6000kpc2, respectively,
−0.35≤logf5.8µm/f3.6µm ≤−0.05, (1)
15 In most cases this was a 2×3 sparse map with typical step
This deficiency, although not a complete “gap”, is what sizes of 30′′ parallel (to the slit) steps and 35′′ perpendicular for
ledWalker et al.(2010)tosuggestHCGgalaxiesrapidly SL,and70′′ and35′′ parallelandperpendicularintheLLmodule.
HCG MOHEGs 5
covering the IGM and the galaxies themselves. detections ( <2.5σ). PAHFIT is a spectral decomposi-
tionpackage(Smith et al.2007b)thatfitsemissionlines,
bands and dust continua to stitched LL and SL spectra.
The spectra RMS (root mean square) for determining
upper limits were measuredusing ISO Spectral Analysis
Package (ISAP)16.
3.2. Spitzer IRAC and MIPS Photometry
The Spitzer IRAC (Fazio et al. 2004) and MIPS
(Rieke et al. 2004) instruments were used to obtain
imaging at 3.6, 4.5, 5.8 and 8.0µm, and 24µm, respec-
tively, of our sample and observed as part of PID 50764
(P.I. Appleton), PID 40459 (P.I. Le Floc’h), PID 631
(P.I. Mazzarella) and PID 101 (P.I. Kennicutt).
Fig. 2.—TheIRSsampling(Long-Low;∼10.5′′andShort-Low; IRAC photometric measurements (to obtain colours)
∼3.6′′slits)shownonHCG40(IRACfour-colorimageof∼4′×4′).
for 21 galaxies in our sample are taken from
1storderspectralcoverageisshowninyellowwith2ndorderinred.
Bitsakis et al. (2011) as indicated in Table 4. Fluxes for
It should be noted the mapping combined with staring strategy
probestheintragroupregion,aswellasthegalaxiesthemselves. theremainderofgalaxieswerecarefullymeasuredtoim-
prove the deblending systematics where contamination
from nearby stars and galaxies may affect the photome-
PrimarydatareductionswereperformedbytheSpitzer try. For these systems, the data was reduced using the
Science Center (SSC) pipeline, version S18.0.2-18.7.0, SSC science pipeline version S18.5.0 and 18.7.0. Galaxy
which performs standard spectral reductions such as photometry was performed using a matched elliptical
wavelength and flux calibration, ramp fitting, dark cur- aperture,determinedbythe1σisophoteinIRAC3.6µm,
rent subtraction and detector droop and non-linearity after foreground contaminating stars were masked from
linearity corrections. Basic Calibrated Data (BCDs) all images and replaced by the corresponding isophotal
frames, output from the pipeline, were combined within value ofthe source. Nearby contaminatinggalaxieswere
the SSC tool CUBISM (Smith et al. 2007a), optimised similarlymasked. The localbackgroundwas determined
for extended sources, with each AOR forming one spec- from the median pixel value distribution within a sur-
tral cube. This was done to ensure proper background rounding annulus. Aperture corrections were applied as
subtraction,usingbackgroundstakencloseintimetothe specified by the IRAC Handbook. The formal photo-
observations. This was achieved using either dedicated metric uncertainties are ∼5% for the IRAC calibration
backgrounds or from the “off” position BCDs with cov- error.
erage outside of the group. Potentially saturated sources, particularly in
Pixeloutlierrejectionwasdoneusing the CUBISMal- IRAC3.6µm and 4.5µm, namely HCG 56B, 91A,
gorithm (with a conservative 8σ clipping) and then by 96A and 100A were investigated. HCG 96A was
visual inspection of the spectral cubes to ensure that no saturated in the cBCDs at IRAC3.6µm, 4.5µm and
weak signals are lost. The cubes were inspected for any 5.8µm. For this system the 1.2s HDR (High Dynamic
H emission in the IGM and spectra extracted for each Range) exposures were used to determine the IRAC
2
galaxy with coverage. Since we are particularly inter- fluxes. For HCG 100A the source counts for all bands
ested in emission from the disk, we were careful to ex- was nominal, although at the full well capacity. For
tractspectraalongtheslitwithanaperturelargeenough 56B and 91A the source counts for IRAC 5.8µm and
to capture mostof the source’slight while still maximis- 8.0µm were nominal and below the saturation limit at
ing the signalto noiseofthe spectrum. Extractionareas 3.6µm and 4.5µm. However, since the peak pixel flux
for SL and LL for eachgalaxyare listed in Table 8. Due is within the non-linear regime at 3.6µm and 4.5µm,
to the coverage obtained through the sparse mapping the integratedfluxes may be slightly underestimated (as
mode,severalgalaxieshave“off-nuclear”coveragewhere indicated in Table 4), but the mid-infrared colors are
the slits have not been centred on the nucleus; these are likely unaffected.
indicated in Table 8 and are treated as separate spectra For galaxies located at z > 0.035 the shifting emis-
to those that cover the nuclear region. sion features, in particular the 6.2µm PAH, affect the
Spectrawereextractedforallbut4galaxiesinthesam- observedcoloursof a galaxy. For the groupsin our sam-
ple (HCG 6A, 6C, 6D and 79C) where no signal above ple affected by this (HCG 6, 8, 55, 75, 82 and 95) we
the noise was detected. Spectra were converted to flux have corrected the IRAC fluxes for redshift (i.e. “k-
densities(mJy)usingtheextractionareas,andSLscaled corrected”)usingtheempiricaltemplatelibraryfromM.
to LL to match continuum levels in cases that required Brown et al. (in prep.). Consisting of 125 galaxy tem-
it (see Table 8). This accounts for beam resolution dif- plates of local, well-studied and morphologically diverse
ferencesbetweenthe longandshortorder;here wemake galaxies(e.g. SINGS,theSpitzerInfraredNearbyGalaxy
the assumption that the emission in eachslit is uniform. Survey), these are generated using optical and Spitzer
The scaling of spectra is further discussed in Section A spectroscopy with matched aperture photometry from
of the Appendix.
PAHFIT was used to characterize our spectra, but 16 TheISOSpectralAnalysisPackage(ISAP)isajointdevelop-
mentbytheLWSandSWSInstrumentTeamsandDataCenters.
each visually inspected one by one to determine which
Contributing institutes are CESR, IAS, IPAC, MPE, RAL and
lines were detected reliably and which were marginal SRON.
6 Cluver et al.
GALEX, XMM UV, SDSS, 2MASS, Spitzer and WISE,
synthesized with MAGPHYS (da Cunha et al. 2008).
For galaxies with recessional velocities < 9000kms−1,
k-corrections do not appreciably affect the mid-infrared
colours, or analysis derived thereof, presented in this
work.
MIPS 24µm data was processed through SSC science
pipeline versions 18.1.0, 18.12.0 and 18.13.0, achiev-
ing a spatial resolution of ∼6′′. The LL spectral
extraction areas output from CUBISM were used to
make matched aperture photometric measurements us-
ingtheIRAF17 task,POLYPHOT.Forstar-formingand
AGN-dominatedspectrashowingcontinuawithastrong
power-law dependence, we applied colour corrections as
recommended in the MIPS Handbook. This correction
is of the order of ∼5%. The MIPS calibration error is of
(a)
the order of ∼10 – 20%. .
In the cases of 31AC, 40C, 44A and 75D we have H
2
detections in LL coverage without matching SL cover-
age. In order to measure the 7.7µm PAH emission, we
use the prescriptionof Helou et al.(2004) andmeasured
matched (to the LL extraction) apertures of the IRAC
3.6µm and IRAC 8µm bands. By subtracting a scaled
version of IRAC 3.6µm from IRAC 8µm we can com-
pensate for stellar light contamination and use this as a
measure of the strength of the aromatic emission within
the band (see, for example, Roussel et al. 2007). This
was also done for the IGM detections discussed in Sec-
tion 4.1.1.
4. RESULTS
4.1. Emission from Warm Molecular Hydrogen
The pure rotational transitions of molecular hydrogen
covered by the IRS SL and LL spectral windows at the (b)
redshifts of the HCGs, are the 0-0 S(0), S(1), S(2), S(3),
S(4) and S(5) lines at 28.22,17.03, 12.28,9.67, 8.03 and Fig. 3.—a)HCG40IRAC3.6µmimage(∼4′×4′)withLLslit
6.91µm,respectively. Thesecanbeexcitedinavarietyof coverageoverlaid;theblueboxshowsthepositionoftheextraction.
The VLA integrated Hi distribution (Jy/beam/s) from Verdes-
astrophysical processes including, UV pumping and col-
Montenegro et al. (private communication) is shown as overlaid
lisional heating in photodissociation regions associated contours; image reproduced with kind permission. North is up
withstarformation(e.g.Hollenbach & Tielens1997),X- and East is left. b) LL Extraction centered on 09h38m53.45s, -
ray heating in XDRs (X-ray dominated regions) partic- 04◦51′47.2′′. Due to a latent-induced flux bias in the first set of
BCDs,thesignal-to-noiseoftheS(1)lineisdiminished.
ularly those associatedwith AGN (e.g. Draine & Woods
1992),cosmicrayheating(e.g.Dalgarno et al.1999)and
heatingbyturbulenceorshocks(e.g.Shull & Hollenbach H2 from UV-heating within photodissociation regions
1978). Wedetect(>2σ)twoormorelinesofwarmH in (PDRs), we use the 7.7µm PAH emission band as a
2
32/74 galaxies in our sample; this includes marginal de- discriminator (described by Ogle et al. 2010). The ra-
tections (i.e. between 2 and 3 σ), but we do not include tio of H luminosity (summed over the 0−0S(0)–S(3)
2
these in the analysis that follows. In the groupIGM, we lines) to the 7.7µm PAH luminosity allows us to de-
find evidence of two locations with tentative detections termine which systems have enhanced H relative to
2
of excited H – both suffer diminished signal to noise of SINGS star-forming galaxies (Roussel et al. 2007) and
2
the S(1) line due to an artifact latent in the first set of thusexhibiting“MOHEG-type”emission. Thisvalue(≥
BCDs. 0.04)ofH2S(0)-S(3)/7.7µmPAH,hereafterreferredtoas
Many of the galaxies located in HCGs are star form- H2/7.7µm PAH for brevity, separating PDR-dominated
ing systems, generating a UV radiation field capa- H heating from other heating sources has been demon-
2
ble of heating very small grains (VSGs) and exciting strated by radiation modeling (Guillard et al. 2012b).
PAH molecules, thus producing distinctive dust fea-
tures in the mid-infrared. We shall focus on systems 4.1.1. H2 in the Group IGM
where the H2 emission is enhanced relative to UV- A key aim of this project was to determine the preva-
excitation. In order to separate mechanical heating of lence of warm H emission in the IGM of HCGs most
2
likely to be in an active stage of transformation. From
17 IRAFisdistributedbythe NationalOptical AstronomyOb- thesampleof23groups,wehavediscoveredtwolocations
servatory, whichisoperated bytheAssociationofUniversitiesfor
showingwarmH detectionsintheIGM.Theseareboth
Research in Astronomy, Inc., under cooperative agreement with 2
theNationalScienceFoundation detected in the LL spectra and appear associated with
HCG MOHEGs 7
theedgesofgalaxydisks. Figure3and4showthedetec- vations (Lisenfeld et al., in prep.) find indications for
tions in HCG 40 andHCG 91,respectively. We consider extended molecular CO emission at the location of the
these preliminary, demanding follow-up observations for IGM detection within the group.
confirmation. In HCG 91, we find a detection at what appears to be
theedgeofthediskof91A,astar-forminggalaxy(Figure
4). TheLL2spectrumsuffersfromasimilarlatenteffect
as above (due to observing a bright target prior to this
observation) and the first BCD is contaminated, result-
ing in a lower signal to noise for the S(1) detection, yet
thereisastrongS(0)detection. Moreover,theestimated
H /7.7µm PAH ratio (calculated as above, with an H
2 2
flux of 1.43×10−17W.m−2 and 7.7µm dust estimate of
3.51×10−16W.m−2) is >0.041 and therefore would be
classified as a MOHEG even without SL coverage.
Barnes & Webster (2001) find two Hi knots centered
aroundHCG91Aand91Cwithaconnectionbetweenthe
twothroughagasbridge(shownascontoursinFig. 4a),
since there is a common velocity between the southern
part of HCG 91C and the northern part of HCG 91A.
(a)
Amram et al.(2003)findthattheHαdistributionshows
a tidal arm pointing from HCG 91A towards HCG 91C.
TheyalsofindadoublegaseouscomponentforHCG91C
strongly suggestive of a past interaction. They propose
a scenariowhere HCG 91C is passing throughthe group
forming the tail of HCG 91A.
Thelocationofthesetwodetectionssuggestapossible
connection with disks interacting with the group IGM.
4.1.2. H in Individual Group Galaxies
2
The mapping strategyemployedin this study resulted
in 74 compact group galaxies with an IRS spectrum (ei-
ther full or partial). In Table 4 of the Appendix, we
indicate whether a warm H detection was made in an
2
individualgalaxy. ThefluxesdeterminedfortheH emis-
2
sion lines are presented in Table 5; we note that several
systems have their SL and LL lines presentedseparately
due to the regions sampled by the IRS not overlapping
(b)
(and therefore not joined together). In addition, spec-
Fig. 4.—a)HCG91IRAC3.6µmimage(∼6′×10′)withLLslit tra that are not centred on the nucleus (and therefore
coverage overlaid; the blue box shows the position of the extrac- dominatedby emissionfromthe disk)areindicated. For
tion. The ATCA Hi distribution from Barnes&Webster (2001) completeness, extraction areas for H -detected galaxies
are shown as contours. Image reproduced with kind permission 2
are listed in Table 8.
from the authors. North is up and East is left. b) LL extraction
centered on22h09m06.20s,-27◦48′09.1′′. Dueto alatent effect in Thestrengthsofthe PAHcomplexesandatomicemis-
thefirstsetofBCDs,thesignal-to-noiseofthe S(1)lineisdimin- sion lines for the H2-detected systems are presented in
ished. Tables 6 and 7, respectively. Upper limits for the H
2
S(0)–S(3) emission for galaxies without H detections,
2
andmeasurementsoftheirPAHfeatures,areincludedin
In HCG 40, we detect the S(1) line (Figure 3) outside the Appendix.
theSbgalaxyHCG40C(thepositionsoftheIRSslitsare In Table 2 we list the summed H fluxes, H /7.7µm
2 2
showninFig. 2).Using the IRAC 3.6µmand 8.0µmcov- PAH ratio and MIPS 24µm fluxes (measured within a
erageto provideanestimate ofthe 7.7µmPAHemission matched aperture). Upper limits and marginal detec-
(4.77×10−16W.m−2) and the combined S(0) and S(1) tions are not included in the summed H fluxes. Our
2
fluxof1.02×10−17W.m−2,wefindaH /7.7µmPAHra- sample contains 13 systems with enhanced H emission,
2 2
tio of >0.021. This canregardedas a lowerlimit due to satisfyingtheMOHEGcriteria(H /7.7µm PAH≥0.04);
2
missing SL coverage, as well as the poor signal to noise herewe include HCG 56Cwithits H /7.7µm PAHratio
2
ofthe S(1)line due tothe latent-inducedflux biasinthe of0.037. Theremainingsystems(19intotal),havenom-
first set of BCDs. inal H emission, consistent with UV photoionization.
2
The VLAHidistribution(fromVerdes-Montenegroet Thereisoneexceptiontolist: HCG44AhaslimitedIRS
al., private communication) is shown as contours in Fig- coverage,renderingtheH /7.7µmPAHratioasanupper
2
ure 3a and indicates the presence of Hi around 40B and limit. However,fromthe SINGS study,we knowithasa
40C, with a tail towards 40D. Interactions with a tidal H /7.7µm PAH value of 0.042 (Roussel et al. 2007) and
2
tail could account for this emission, and would also ex- is therefore a weak MOHEG. We therefore include it as
plain the (albeit weak) warm H signal from 40B pre- aMOHEGinoursampleandusethisvalueanditsvalue
2
sented in the next section. Follow-up IRAM CO obser- of L(H S(0)-S(3))/L = 0.022 (Roussel et al. 2007) in
2 24
8 Cluver et al.
(a) HCG57A-includingthenuclearregion (b) HCG57A-off-nuclearextraction
Fig. 5.—SpectraofHCG57A.Thematched MIPS24µm photometryisshownasafilledgreycircle.
future analyses. excite the H in the disk of galaxy, offset from the nu-
2
As shown in Section 2, the distribution of cleus.
Bitsakis et al. (2011) indicates a lowering in the In Figures 6, 7 and 8 we plot the spectra of the other
density of galaxies, compared to the LVL galaxy colors, HCGgalaxiesclassifiedas MOHEGsthat lie atinterme-
in the region given by Equation (1). If we separate the diate mid-infrared colors (Equation 1) as used in Table
MOHEGs according to mid-infrared color, we find 12 3). Inparticular,HCG6B(Fig. 6a)andHCG15D(Fig.
locatedbetween −0.35≤log[f /f ]≤−0.05(i.e. 6c)showthedistinctiveH S(1)emissionlinedominating
5.8µm 3.6µm 2
intermediate mid-infrared colors); these systems have the spectrum, with little 6.2 and 7.7µm PAH emission,
enhanced H relative to their star formation, and their reflected in Figure 11.
2
mid-infrared colors reflect that globally they are not HCG 25B has SL coverage of the nuclear region (Fig.
dominated by star formation (or AGN emission); this is 6e)andweseehighsignaltonoiseH2 S(2),S(3)andS(5)
a clear indication that the warm H is not UV-excited. emission(theweakerS(4)transitionisnotseenabovethe
2
This is discussed further in Section 4.1.4. PAH emission at 8µm). Similar to HCG 57A, 25B has
The morphological types of the H -detected sys- coverage of the disk region – it covers a larger region
2
tems are given in Table 3, where de Vaucouleurs et al. comparedto the SLextraction,andisnoisier,yetwesee
(1991) classifications are given for most of the sam- theS(1)clearlyabovethe weakmid-infraredcontinuum,
ple, or Hickson et al. (1989) for those that were ab- againsuggestiveofamechanisminfluencingregionsaway
sent or assigned unknown classification. We include from the nucleus.
the nuclear classifications from Coziol et al. (2004), For HCG 68A and 68B (Fig. 7d and e), we see the
Brinchmann et al. (2004), Mart´ınez et al. (2010), as strong contribution from the stellar continuum (λ <
well as the T-Type from the RC3 catalogue of 18µm), indicative of their early type morphology (S0),
de Vaucouleurs et al. (1992). yet the H2 S(1) and S(3) ortho transitions feature
Weincludespectraforthenon-MOHEGH galaxiesin strongly. The 7.7µm PAH is particularly weak in HCG
2
SectionBoftheAppendix,aswellasanydiscussionper- 68A and only an upper limit exists. In HCG 68B the
taining to individual systems. Spectra for the MOHEG PAH is clearly defined, but the power in the H2 lines is
galaxies are presented in the following section. striking.
By contrast, the galaxy HCG 40B (Fig. 6d) has the
4.1.3. Spectra of MOHEG Sources S(1) line weakly detected with similarly low 7.7µm PAH
emission. The fact that the 11.3µm PAH often fea-
The power in the H emission lines compared to the
2 tures prominently in these spectra is not surprising as
mid-infraredcontinuumis particularlynoticeable insev-
the large, neutral PAHs can be excited by soft radiation
eral MOHEGs. For example, in Figure 5 we show two
from evolved stars (Kaneda et al. 2008) and this would
spectra of HCG 57A, one centered on the nuclear region
be consistent with their early-type morphologies.
(LL only) and a matched, off-nuclear extraction in the
In contrast to the typical MOHEG spectra presented
disk of the galaxy,made possible by the sparse mapping
in this paper, the MOHEG HCG 95C exhibits an ex-
strategyemployed(see Section3.1). The nuclearextrac-
ceptional, likely tidally-induced, star formation spec-
tion shows powerful H S(1) and S(0) emission and, al-
2 trum (Fig. 8). The strong PAH emission and steeply
though we have no coverage of the PAH bands, we do
rising 24µm continuum is indicative of a system dom-
not see a steeply rising mid-infrared continuum typical
inated by star formation. Figure 8b shows that the
of star-forming and AGN-dominated systems.
galaxyishighlydisruptedduetoaninteractionwith95A,
Theoff-nuclearextractionshowspowerfulH emission
2 with its current state classified as an Sm morphology.
with relatively weak 6.2 and 7.7µm PAH emission. This
Iglesias-P´aramo& V´ılchez (1998) find evidence for two
spectrumsuggestsanon-star-formingmechanismableto
HCG MOHEGs 9
(a) HCG6B (b) HCG15A
(c) HCG15D (d) HCG40B
(e) HCG25B-includingthenuclearregion (f) HCG25B-off-nuclearextraction
Fig. 6.—MOHEGswithIntermediateMid-infraredColours. ThematchedMIPS24µmphotometryisshownasafilledgreycircle.
10 Cluver et al.
(a) HCG44A-SL(Off-Nuclear) (b) HCG44A-LL(Nuclear)
(c) HCG56C (d) HCG68A
(e) HCG68B (f) HCG82B
Fig. 7.—(a)−(f)MOHEGswithIntermediate Mid-InfraredColours;matchedMIPS24µmphotometry isshownasafilledgreycircle.