Dottorato di Ri er ain Fisi a
Ci loXXVI
Dust Modeling
and
Evolutionary Impli ations
Settore s ientifi odis iplinare di afferenza
FIS/05-Astronomiae Astrofisi a
Presentatada: AlbertoZon a
Coordinatoredottorato: Prof. PaoloRuggerone
From the beginning of the 20th entury astronomers started to move their studies
towardthat parti ular omponentwhi h ouldberesponsibleofthereddeningandthe
extin tion of stars radiation. They agreedthat this omponent must be represented
byapopulationofmi ros opi parti lesnamedinterstellardust,thatitmustbemixed
withtheinterstellargas. Lateron,inthese ondhalfofthe entury,dustwasre ognized
toplayverymanyimportantrolesintheMilkyWayandothergalaxies. Thesendings
hanged the astronomer's viewof the interstellar spa e, that is now fully onsidered
asan a tiveelementin gala ti evolution. For this reasonto a hievea morereliable
understanding of the nature of interstellar dust is ne essaryto takeinto a ount the
response of dust to the physi al onditions of the regions in whi h dust grains are
embedded.
Overtheyearsastronomersusedthe variationofextin tionrespe t towavelength,
the interstellar extin tion urve, to study osmi dust. In this thesis I elaborate a
physi al and hemi almodelof dust, to investigatetheremarkablevarietyof gala ti
interstellarextin tion urves. InChapter1Iintrodu etherolesoftheinterstellardust,
its pro essing in the interstellarmedium, and thetoolsto addressthe problem of its
nature. In Chapter 2 I present the exploited model in a simplied version, and an
appli ation to somepe uliar lines of sight in our galaxy. In Chapter 3the omplete
versionof the model isput forward,and, together with itssimplied version,applied
to a large sample of Milky Way interstellar extin tion urves. All dierent avours
of observed extin tion urves, ranging from the average gala ti extin tion urve to
very pe uliar proles, an be des ribed by su h a model. I show that a mixture of
oremantlesili ate/ arbongrainstogetherwitharelativelysmallnumber(54spe ies
infour hargestates)ofpoly y li aromati hydro arbons anreprodu ethefeaturesof
theextin tion urveintheultraviolet,dismissinganoldobje tiontothe ontribution
of poly y li aromati hydro arbonsto the interstellar extin tion urve. In Chapter
4 I ompare an evolutionary model of the physi al properties of arbona eousgrain
mantleswith theirdeterminationthroughthettingpro edure outlinedin Chapter3.
The results of su h omparison demonstrates, that in the framework of the adopted
dust model, the whole sample of gala ti extin tion urvesare in striking agreement
withtheproposed evolutionarys enario,requiringphysi al onditionsfully onsistent
withstandardvaluesofdensity,temperature,radiationeldintensity,andaverageage
ofdiuse interstellar louds. The resultsofthis thesisshowthat thepresent modelis
ableto re on ilethe great varietyof observedinterstellar extin tion urves within an
uniedevolutionarys enario,withoutmakingdrasti hangesto stru turalproperties
1 Interstellar dustfrom the beginning 1
1.1 Introdu tion. . . 1
1.2 Themanyrolesofdustin theinterstellarmedium . . . 3
1.3 Interstellardustlife y le . . . 5
1.4 Theproblem of osmi abundan es . . . 11
1.5 Dustin ourgalaxy . . . 13
1.6 Dustin externalgalaxies . . . 16
1.7 Dustmodels. . . 20
1.7.1 Draine'smodel . . . 20
1.7.2 TheMessina/Cagliarimodel . . . 21
2 The model and its rst appli ation 23 2.1 Themodelinitssimpliedversion . . . 23
2.1.1 Classi omponent . . . 24
2.1.2 PAHs . . . 28
2.2 Thesamplesele tion . . . 32
2.3 Results. . . 33
2.4 Dis ussion . . . 39
3 Modeling gala ti extin tion 41 3.1 Introdu tion. . . 41
3.2 Thenewsimplied versionofthe
[CM]
2
model . . . 423.3 Thedetailed versionofthe
[CM]
2
model . . . 423.4 Fittingdetails . . . 44
3.5 Results. . . 46
3.5.1 Classi aldustdistribution parameters . . . 48
3.5.2 Comparisonbetweensimplied anddetailedmodel results . . . . 49
3.5.3 The hargeofPAH mixture . . . 52
3.5.4 DetailedPAH omposition: un onstrained. . . 58
3.6 Dis ussion . . . 60
4 The evolvingdust 67 4.1 Introdu tion. . . 67
4.2 Themodelofinterstellardustevolution . . . 69
4.2.1 Outline . . . 69
4.2.2 Propertiesofthesolutions . . . 70
4.3 Observationaldata . . . 73
4.5 Con lusionsandfutureworks . . . 78
A The inverse problem 83
B Analyti approximations 87
Interstellar dust from the beginning
1.1 Introdu tion
Intheeighteenth enturyastronomershadnoti edwideregionsin theskywherethey
ouldnotobservestars. Therstastronomerwhodis ussedthiseviden ewasWilliam
Hers hel: he alled lö her in den himmel (holes in the sky)these regions apparently
withoutstars. Manyyearslater,WilhelmStruverealizedthatstarlightsuers
absorp-tion in proportion to the distan e travelled. In 1930, studying open lusters, Robert
J.Trumplerdis overedananomalyintheopen lusterssize;sohededu edthat
some-thingwaspresentbetweentheobservingastronomerandtheopen lusters,mi ros opi
solids,lateron alledinterstellardust, ouldae t themeasure ofstarsdistan es.
Astronomersmeasurestellardistan es omparingtheapparentmagnitudewiththe
absolutemagnitude,obtainedwiththespe tral lassi ation
m
V
− M
V
= 5 log d
′
− 5
(1.1)where
m
V
andM
V
areapparentandabsolutevisualmagnitudes,respe tivelyandd
′
is
theapparentmean lusterdistan einparse s 1
. Knowing
d
′
,Trumpler oulddedu ethe
lineardiameterof ea h lustergeometri allyfrom themeasuredangular diameter. He
derivedaremarkabletrend: lustersdiametersappearedtoin reasewithdistan efrom
theSolarSystem. Tomakesurethatthe lusterdiameterswereindependentofdistan e,
heintrodu edadistan edependent orre tioninthelefthandsideofequation(1.1)
m
V
− M
V
− A
V
= 5 log d
− 5
(1.2)where
d
is now the real distan e of the luster andA
V
is the orre tion. To shed light on the nature of the orre tion, Trumpler studied the olours of stars involvedin the measurements, nding that their spe tra appeared to be shifted towards the
red,orreddened: learly
A
V
shouldbemoree ientforwavelengthsshorterthanin thevisible. Inparti ular, having omparedthe apparentluminosities of similarstars,Trumplerinferred that
A
V
wasalinearfun tion ofthewavenumber,A
V
∝ λ
−1
. This
resultindi atedthatinterstellarextin tionmustbeduetothepresen eofsolidparti les
1
This equation is derived by the denition of magnitude m from the brightness F:
m(λ) =
−2.5 log F (λ) + C
inwhi hC isa onstantdependentbythewavelength. Theabsolutemagnitudeis denedliketheapparentbrightnesswhi hastarwouldhaveifitwerepla edatastandarddistan eof10p away;thenexploitingtherelationbetweenluminosityandbrightness:
L = 4πd
2
·
F
,inwhi h
d isthetruedistan eofthesour efromtheobserver,Iobtain
M − m = −2.5 log(4πd
2
F/4π10
2
F ) =
withdimensions omparabletovisualwavelengths. Thiswastherstdeterminationof
a toolthat astronomersare now routinely using to des ribe the intera tion of stellar
radiationwithdustalongaparti ularlineofsight: theextin tion urve,arelationthat
links theextin tion ofstellarradiationto thewavenumber.
Theknowledgeoftheexisten eofdispersedsolidsinspa e ouldhavehelpedthose
astronomerswho, in 1920, had dis overedthat many stars loseto thegala ti plane
appeared redderthen expe ted onthe basis oftheir spe tral types. There wasa
dis- repan y in stellar temperature derived by spe tros opy and photometry. By means
ofspe tral lassi ationstartemperaturesmaybeevaluatedthroughthepresen eand
relativeintensities ofspe trallines in thestellar photosphere. Instead olour indi es,
like (
m
B
− m
V
) 2, are indi ators of temperature based on the ontinuum slope and
its equivalent bla kbody temperature. Many stars, that showed spe tral features of
early-typestars, had olour indi es moreappropriate to late-type stars. Su h results
are easilyexplainedbythereddening ausedbyforegroundinterstellardustalongthe
line of sight (Whittet, 2002): stellar spe tra in a limited range do not hange with
reddening, instead olourindi esdependontemperatureandreddening.
Usually the extin tion is reliably determined using the pair method, omparing
spe trophotometry of two stars of the same spe tral lass: if one star has negligible
foregrounddust,whilethese ondstarisreddened, omparisonoftwospe tra,withthe
assumption that extin tion goesto zeroat verylongwavelength, allowsto determine
theextin tionas afun tion ofwavelength(Draine,2003).
Thedegreeofreddeningorsele tiveextin tionisquantiedthroughthe olorex ess
E
B−V
= (m
B
− m
V
)
− (m
B
− m
V
)
0
,
(1.3)where
m
B
− m
V
and(m
B
− m
V
)
0
are observed and intrinsi valuesof olour index, respe tively. Sin e the extin tion is greater in the B band than in V one,E
B−V
is a positive quantity for reddened stars and zero for unreddenedstars. An importantrelationshipbetweentotalextin tionand olorex essis
R
V
=
A
V
E
B−V
.
(1.4)Inequation(1.4)
E
B−V
isdire tlymeasurable,whereasA
V
isevaluatedonlyifR
V
an bedetermined. Asextin tionde reasesrapidlyaswavelengthin reases,IobtainR
V
=
− lim
λ→∞
E
λ−V
E
B−V
.
(1.5)
Theoreti ally,
R
V
is expe ted to depend on the omposition and size distribution of dust grains. Inthelowdensityinterstellarmedium,R
V
isvirtually onstantR
V
≈ 3.05
(1.6)(Fitzpatri kandMassa,2007).
Priortothemid-1960's,interstellarextin tionwasa essibletostudymainlyinthe
visible part of the spe trum, where it was known to in rease linearly for de reasing
wavelength,andwasthereforebelievedto beessentiallydue tos atteringand
absorp-tion bysomeform of small dustparti les. With the openingof theUltraViolet (UV)
2
m
B
istheapparentluminosityinbluerangeofele tromagneti spe trum;therelativeextin tion atageneri wavelengthisexpressedasm
λ
.and InfraRed (IR) portions of the spe trum to astronomi al observations, Whitford
(1948)produ edanInterStellarExtin tionCurve(ISEC)thatshowed urvatureatthe
near-UV and IR regions. As pointed out remarkablyby Greenberg and Shen (1999)
"things were beginning to make somephysi al sense from the point of viewof small
parti les attering". Indeed,atthattimevandeHulst(1949)putforwardtherst
om-prehensivemodel of interstellar dust, and having arealisti dust model he developed
thes atteringtoolstohandletheproblem(vandeHulst,1957). Sin etheninterstellar
dusthasbe omeoneofthesubje tsin theforefrontofastrophysi s.
1.2 The many roles of dust in the interstellar medium
Sin eitsdis overy,theastronomersregardedthedustasadisturbfortheobservationsof
stars,anirritatingfogwhi hpre ludeda learviewofstellarspe trum. Forthisreason
therststudiesofinterstellardustweremotivatedbythedesireto orre tphotometri
data for its presen e. Nevertheless, as it was later understood, dust has very many
importantrolesto playwithintheMilkyWayGalaxy(MWG)andothergalaxies,and
itisa ru ial omponentintheevolutionofgalaxies.
An important role of the dust is the shielding of nebular regions from starlight,
favouring the build-up of a omplex hemistry. Mole ules formed in this hemistry
are important be ause they anbe used by astronomersto tra e the presen eof gas
anditsphysi al onditionthroughIRand radioemissionswi hareunae tedbydust
extin tion. DustgrainsaretheprimaryrepositoryintheInterStellarMedium(ISM)for
many hemi alelements,and atalysetheformationof
H
2
,themostabundantmole ule intheISM.Hydrogenexistspredominantlyinits1s
groundstate,and ollisionsbetween hydrogenatoms anpro eedalongtwopotentialenergy urves,inwhi htheele tronispins are either parallel (triplet state) or anti-parallel (singlet state). As the atoms
are initially unbound, their total energy is positive (the zero of the total energy is
takenat the mole ulardisso iationlimit). Inorder to stabilize, the systemmustlose
energy and its total energy be ome negative. In the gaseous phase this may o ur
eitherbymeansofthree-body ollisions,thethirdbodytakingawaytheex essenergy,
orbymeansofradiativepro esses. Three-body ollisionsareextremelyimprobableat
interstellardensities,and sotheonlywayin whi hthesystem anstabilizeisthrough
the emission of a photon. Unfortunately, transitions between the triplet and singlet
ele troni potential energy urves are forbidden to ele tri dipole radiation as they
involve a hange in the total spin quantum number. Radiative transitions involving
the nu leardegrees offreedom (rotation and vibration) are also forbidden, asthe H
2
mole uleishomonu learanddoesnotpossessapermanentdipolemoment. Itiswidely
a eptedthatH
2
formsongrainsurfa es: Hatomssti ktograinsandbe ometrappedatsurfa edefe tsinthegrainstru ture. Duringthere ombinationtheH
2
bindingenergy,4.476 eV, must be partitioned between the rovibrational ex itation and translational
energyofthenas entmole uleandheatingofthedustgrain(ParneixandBre higna ,
1998). Iftheinternalandtranslationalenergiesofnewlyformedmole ulesarerelatively
small,thensigni antgrainheatingmusttakepla e,whi hmayleadtothedesorption
of volatile mole ules from the dust grain surfa e(Duley and Williams 1993, Roberts
et al. 2007). The H
2
internal energy distribution ould have asigni ant impa t onthe hemistryo urring in theISM be ausevibrationally ex ited
H
2
will in reasethe overall energy budgetof gas-phase pro esses. There havebeen many theoreti aland(e.g., Sizun etal. 2010,Martinazzoand Tantardini2006 andreferen estherein). It is
possiblethatthisformationpumpingmaybeobservableintheIRspe traofparti ular
astronomi alregions(Islamet al.,2010).
Surfa erea tionsmayleadtotheformationofspe iesheavierthanH
2
. Atthelowtemperatures of dense mole ular louds gas-phase spe ies ondense onto dust grains
forming i y mantles. During the pro ess, when spe ies su h as e.g., OH and CO
freeze-out on dust, they be ome hydrogenated via surfa e rea tions and form H
2
Oandmethanol. Therateatwhi htheserea tionso urin reaseswiththedensityofthe
loud. I y mantlesexposed to sho ksorheated nearbyanas entstarbe omesour es
of omplexmole ules. Thus, at alllevelof density,the ex hange ofmaterial between
interstellar gas andgrains isimportant forthe hemi al omposition and evolutionof
theISM asawhole.
Dust is involved in the formation and evolution of stars and planetary systems.
Indeed stars are born within very dense loud of dust and gas and the presen e of
dust, beingane ientradiatorofex essheatenergy,favoursgravitational ollapseto
form protostars. As protostar ontra ts, rotation attens the ir umstellar envelope
in adisk-likestru ture. Proto-planetarydiskshavebeendete tedaround youngstars
throughextin tion and emissionarisingfromthe dustthey ontain. Theee t ofthe
radiationpressureongrainsmayinuen ethedynami sofstellarenvelopesurrounding
bothyoungand evolved stars. Grains ondensing in theenvelopesof late-typegiants
maydrivetheoutowsofevolvedstars,and ontributesigni antlytotheenri hment
oftheISM withheavyelements.
Sin e dust obs ures stellar light, itmustre-emit at longer wavelengths (Witt and
Gordon, 2000). Thus, dust manifests itself through ontinue emission at
λ > 1 µm
. Su h emission beyond afewµm
and up to 100µm
is mainly due to dis rete photon heating of verysmall dust grains;dust heated bymassivestars to temperature largerthan50Kalso ontributetotheuxinthewavelengthregionbelow
∼ 50 µm
. Beyond40
− 70 µ
m,theemissionisduetodustgrainsinnearlysteadybalan ewiththeaverage heating by starlight. Dust shows also emission and absorption features, that, in thenear/mid-IR, in lude absorption bands su h ase.g., those of sili ate materials at 9.7
and 18
µm
, frozenH
2
O
andCO
2
at 3.0 and 4.3µm
(Öberg et al., 2011), and the emissionbands at3.3,6.2,7.7, 8.6and11.3µm
(Tielens,2013).Inadditiontoextin tion,dustparti lesprodu epolarization,mainlylinear,ofstellar
light. To explain interstellar polarization, dust grains are represented as spheroids,
des ribingarotationaboutthemajoraxisofanellipse(prolatespheroid)ortheminor
axis(oblatespheroid);grainshapesare hara terisedbytheelongation,dened asthe
ratiobetweenmajorandminorsemiaxes. Itisgenerallyfoundthatstellarradiationis
polarizedbypartially alignedspheroidaldust grains whi h u tuateand rotateabout
the axis of greatestmoment of inertia. Thedegree of linear polarization depends on
thetypeofthespheroid,prolateoroblate,itselongationandthealignmente ien y.
When dete table, ir ular polarization an also provide informations on grain shape
andelongation.
Finally,dustindu esthereddeningofextragala ti ba kgroundlight,the umulative
radiativeoutputfromallenergysour esintheuniversesin etheepo hofre ombination.
ThesubsequentemissionatIRwavelengthsprodu esthe osmi IRba kground(Dwek,
2001).
The properties of the dust are quite well known in general terms for the MWG
Sin e thespe traldistribution depends on the omposition and geometry of grains,I
an on ludethat dust grains evolvein response to regionalphysi al onditions, su h
ase.g., hemi alenri hment,dynami s,irradiation, andenergy ontent.
1.3 Interstellar dust life y le
Fromspe tralfeaturesinextin tion,s atteringandemissionastronomers aninferthe
hemi al ompositionofdust. Basedonsu hstudies arbonandsili onare onsidered
asprin ipal omponentsforinterstellardustgrains.
Thestrongest spe tralfeature observablein theextin tion urveis entered at4.6
µm
−1
and it is named bump (see gure 1.2). It suggests the presen e of arbon in
aromati form, sin ethis feature is attributed to
π
∗
← π
absorption. Other features distin tiveofaromati arbonaretheemissionbandsat3.3,6.2,7.7,8.6,and11.3µ
m, alled Aromati InfraredBands(AIBs). These emission features are asso iated to Hatoms atta hed to the edge of a arbon ring skeleton giving rise to the following
vi-brational modes (Allamandola et al., 1989): i) 3.3
µm
CH stret hing mode, ii) 6.2µm
CC stret hing mode, iii) 7.7µm
CCstret hing mode, iv) 8.6µm
CHin plane bendingmode,v)11.3µm
CHoutofplanebendingmodeforthe asein whi hthere arenotH atomsadja ent.Indeed,theonlyse ureidenti ationof arbonindustgrainsisthebroadfeatureat
3.4
µ
mobservableindiuseatomi regions,andindi ativeoftheCHstret hingmode inaliphati hydro arbons.Sili ate mineralsgenerallyhavestrongabsorptionresonan esaround10
µm
dueto SiO stret hing. The observational eviden e of an emission feature near su hwave-lengthobservedinoutowsfrom ooloxygenri hstars(wheresili atematerialswould
ondense) but not in outows from arbonri h stars(where oxygenshould form CO
and notsili ates)supports theidenti ation of sili atesin spa e. Moreover,aweaker
featureat18
µ
misalsoobserved,suggestingthepresen eoftheOSiObendingmode insili ates.Interstellardust is thought to form around evolvedstars (i.e. giant bran h stars,
asymptoti giant bran h stars and supernovae), and then eje ted into the
surround-ing ISM by stellar winds. Here dust grains are subje t to violent pro essing in fast
sho kwavesgeneratedbysupernova. Su hstarsare believed toberesponsibleforthe
formation of dust in high redshift galaxies too. Amounts of dust largerthan
10
8
M
⊙
aredete tedin submillimetregalaxies andquasarsat highredshift fromanumberofinstrumentssu hasSCUBA,MAMBO,MAMBO-2andVLA.TheageoftheUniverse
at z> 6 wasless than 1 Gyr, and early star formation took pla e at redshift
∼
10, approximately500Myrafter theBig Bang(Greifand Bromm,2006). The maximumtime available to build a similar large dust mass was at most
400
− 500
Myr. Con-sideringthat dust is produ ed bystars towardsthe ends oftheir life,in high redshiftgalaxiesonlymassivestarshavesu ientlyshortlifetimewithintheageoftheUniverse
atthis redshift,su hashighmassasymptoti giantbran hstarswithmasses between
3
− 8 M
⊙
havelifetimes of a few10
7
− 10
8
years and they an be potential
ontrib-utors to dust produ tionin high redshift galaxies. Somefeatures in extin tion urves
ofvarious obje tsat highredshifthavebeenattributedto dust ofsupernovaoriginas
theplateauobservedataround
170
− 300
nm,possiblyarisingfromamorphous arbon and magnetite supernova dust (Maiolino et al., 2004). However, the orresponden eof
3
− 8 M
⊙
evolvetoAGB starsand letoutdust throughintensemassloss; ii)stars moremassivethan8M
⊙
explodeassupernova,buttodatethe orresponden eamong thestarmass,thetypeofthesupernovaandthedustmassisnot learyet(Galletal.,2011). Exploiting theHers hel satelliteat far-IRwavelengthsMatsuura et al.(2011)
present the rst dire t eviden e that substantial amounts of dust an be reated in
supernovae. Matsuuraet al.(2011)observations demonstratethe presen e ofdust in
supernova 1987A in a mu h largeramount than previouslyknown, and inferred that
indeedmostoftherefra torymaterialintheeje taofthesupernovahas ondensedinto
dust. Although substantial, the dust produ tion rate appears to be not su ient to
eliminatetheneedforgraingrowthintheISMin theLargeMagellani Cloud(LMC),
aswellas,for omparison, inourowngalaxy(M Kee,2011).
InISMsili atedustappearstobeamorphousduebydustpro essingofnewlyformed
rystallinesili ategrains. Thedierentiala elerationofthegasanddustinsupernova
sho kwavesprodu eshighvelo ityionimpa tsongrainsthat ansputteratomsofthe
grainsand/oramorphise rystallinematerials(Jones,2004). PriortotheInfraredSpa e
Observatory(ISO)missionallsili atesinspa ewerethoughttobeamorphousasaresult
of their rapid gas phase formation in lowdensity ir umstellar shells. Observations
withISO,atwavelengths overingthe
2.5
− 200 µm
range,revolutionisedourthinking ofdust inspa ewiththedis overyofthermalemissionfrom rystallinesili atesinthedustshellsaroundstarsintheirlatestages(Jones,2009). However, rystallinesili ates
wereidentiedindustextra tedfromthe omet81P/Wild2andtheirpresen einother
omets has been revealed by IR spe tralfeatures (Wooden, 2008). Other samples of
rystalline sili ates with anomaloussignatures that indi ate that theyformed around
evolvedstarsin presolarepo hhavebeenextra tedfrom primitivemeteorites. Their
isotopi ompositionsshowvalues veryfar from thesolar systemones. Su h isotopi
anomalies are hara teristi s of the nu leosyntheti pro esses o urring within some
stars at parti ular phasesin their evolution(Jones, 2009). The formationpro ess for
rystalline sili ate is still un ertain,but somespe ulation wasmade. Ábrahám et al.
(2009),observingthemid-IRspe traofthestarEXLupi(ayoungSunlikeeruptivestar
whi hshowsrepetitiveoutburst)intwoseparateepo hs,wereabletore ognizetypi al
feature foramorphoussili atein therstepo h spe trum (thestar wasinaquies ent
phase), and peaks hara teristi of rystalline sili ates superimposed on amorphous
sili atefeature inthese ond epo hspe trum. Thesefeaturesareverysimilarto those
observed in ometary spe tra. A possible interpretation onsiders rystallinizationas
produ edbythermal annealingin thestar's inner diskby outburst. Aproblem ould
be represented by the transport of these rystalline sili ates to be in orporated into
ometsasobserved,sin ethe ometformingzoneisthe old,outerregionofthestar's
disk. A possibleanswer ould begiven by theXwind model, in whi h dust grains
are ballisti ally laun hed above the disk's midplane and transported outward (Shu
etal.,1996). Thismodel ouldhavesomeproblemsinexplainingtheobservedlevelsof
transport. However,itisalsopossiblethatsome rystallinesili atesaremadeinsituin
ometary omae(Li,2009). IntheISM
∼
95%ofsili atematerialisamorphous(Liand Draine,2002),insome ir umstellardisksthe rystallinefra tion appearstobehigherbut the bulk of the sili ates remains amorphous. On the other hand arbona eous
materialsare rystallineintheformofgraphiteordiamond,oramorphousintheform
ofHydrogenatedAmorphousCarbon(HAC).
Dust evolvesin theISM through onstru tivepro esses, and, duringits life y le,
o urinavarietyofinterstellarenvironments,in ludingregionsofintensestellar
radia-tionelds(HIIregions,
γ
raybursts,...),inthehotgasbehindsupernovaremnantsand in interstellar sho k waves. Ingeneraltermadust destru tion pro ess isanyintera -tionbetweenagrainandanotherparti lethatleadstoanetbaryonlossfromthegrain
(Jones,2004),withtheout omeoftheintera tiondependingbythetotalenergyofthe
ollisionandthe ouplingbetweentheparti leandthegrain. Theimportan eofthese
pro essesare evidentfrom thelargevariationsseenin theISEC fromonesightlineto
another: thesedieren esareoriginatedbyinterstellarpro essesbe auseea hsightline
isexpe tedtoaverageoverthestardustprodu edbymanyindividualstars. Themajor
parti legrainintera tionsarethefollowing:
Photongrain intera tions; low energy photons (radio and IR wavelengths) are
weaklyabsorbedbyrefra tory dust materials. Ontheother hand higherenergy
photons anbeabsorbedands atteredleadingtotheheatofgrainsand ausing
thermalemission. Su hkindofintera tions arenotdestru tivepro esses. When
similarpro esseso urformorevolatilematerials,su h asi emantles,the
absorptionofUVphotonsbe omesanimportantdestru tionme hanism(Draine
and Salpeter1979a,Westleyet al. 1995a,b). Muñoz Caro et al.(2010), in their
experiments,measuredCOphotodesorptionfromUVirradiationat7,8and15K.
TheyfoundthephotodesorptionyieldofCOi eis onstantduringirradiationand
independentofthei ethi kness. Theyfoundalsothatthephotodesorptionyield
at15K(
∼
3.5±
0.5×10
−2
COmole ulesphoton(
7.3
− 10.5eV
)−1
)isaboutone
order ofmagnitudehigherthanthepreviousestimate. Onlyhigh energyphoton
intera tions anresultin graindestru tionviadire t heating(Guhathakurtaand
Draine,1989)andviaextremegrain hargingee ts(DraineandSalpeter1979b,
Waxman and Draine 2000, Fru hter et al. 2001). The intera tion of grain with
UV photonsand thegrain heating are responsible for thelossof H atoms from
the arbona eousmaterial in the polymeri form (
sp
3
)and its transition to the
graphiti form (
sp
3
)(Jones et al., 1990). Be ause of irradiation, in arbon
ma-terialsexposed toabove andnearbandgapradiationtheabsorption oe ient
overabroadrangeoffrequen iesin reases. Theamountof in reasedepends on
the wavelength and the intensityof the indu ingradiation, and the durationof
the exposure, leading to a shift of the opti al absorption edge to lower energy
and an in rease in the band tail absorption. The physi al and hemi al
trans-formation is omplex, involving sele tive photodisso iation of hemi al bonds,
sele tivesputteringwhi hredu estherelativeH,N,andO ontent,restru turing
of hemi al bonds to form in reasingly largearomati platelets (see e.g. Jones
2012a,b). However,thenetapparentee t isa hange in olourofthepro essed
arbona eous material, with a progressive darkening, i. e. from transparent to
yellow stu and to brown stu (Greenberg, 1984). Su h a pro ess is alled
photo-darkening.
Ele trongrainintera tions;generallythis kindofintera tionsarenotimportant
destru tive pro esses. These intera tions produ e ele tron sti king leading to
negativegrain harging. A ordingtoDraineandSalpeter(1979b)positivegrain
hargingis also possiblewhen energeti ele tronsimpa ts ausethe emissionof
se ondaryele trons.
Whenthe parti le energy is
≥
MeV it is lassiedas osmi ray and the inter-a tionregardsessentiallyCoulombintera tionsbetween osmi raysand hargeddust grains. The ee t de reaseswith thein reasingrelativespeedbetweenthe
grainandthe hargedparti le. Thes atteringfrequen y
ν
0
ofthisintera tion an beapproximatedastheprodu toftheRutherford rossse tion,proportionaltov
−4
cr
(v
cr
is the speed of the osmi ray) with the density of s atterers and the speed of osmi ray; the s attering frequen yresultsν
0
∝ v
−3
cr
. It is learthat these intera tions have a relevant role if there is alarge energy density in lowenergy(
≤
100MeV) osmi rays(Byleveld et al.,1993). These intera tions an produ e grain heating, harging and ionization ee ts; regardingthis last ee tosmi rays whi h pass through the dust grain may be more important than
Coulomb intera tions, due to the net ee t of all osmi rays that passwithin
aDebyelength 3
(oforder hundredsofmeters) ofthedust grain. Unfortunately,
low energy osmi rays annot be observed into the solar system be ause solar
osmi raysdominatethe osmi rayspe trumat energieslowerthan 100MeV,
so urrentlywehave s ar e information about theabundan e and energy
spe -trum ofgala ti osmi -rays. Insteadat lowerenergies intera tions anprodu e
dierentee ts. Forenergies>10eVpernu leontheseintera tions anprodu e
sputteringofatomsfromthegrain. Thispro esso ursinsho kwavesgenerated
bysupernovaeexplosions;it learlydependsbythein identatom/ionenergyand
dierent sputtering yields are obtained with dierent energies. The sputtering
yield, in addition to the kineti energyof the in ident atom/ion,depends upon
theangleofin iden ea ordingtherategivenbyDraineandSalpeter(1979b). If
thesputteringe ien yislessthanunity(it anhappenforrefra torymaterials)
atom/ionsputteringalsoprodu esan implantationinto thegrain in reasingthe
grainmass. Hen e,thereisadestru tionpro essthat isa ompaniedbyagrain
growthpro ess.
Graingrain intera tions; low energygraingrain ollisions withvelo ities lower
than0.02km/s,produ egrainsti kingand oagulationofgrainsintoaggregates,
resulting in anet in rease in the average grain size (Chokshi et al., 1993). At
velo itiesoftheorderof1km/sgrainfragmentationordestru tion ano ur: i)
forsolidhomogeneousparti lesthispro ess onsistsintheshatteringofthesolid
parti le in a multitude of smaller fragments, ii) for aggregates of parti les the
pro ess resultsin the disaggregationofthe grain into its omponent subgrains
(DominikandTielens,1997). Finallyathighvelo ities(
v
≥ 20
km/s)graingrain ollisionprodu esthepartialortotalvaporizationofthe ollidinggrains.Supernovasho kwaveshaveafundamentalroleindustpro essinganddust
destru -tion,throughsputtering,vaporization,shatteringanddisaggregation,ifthesho kwave
velo ityatthepointofintera tionisfastenough(>50km/s). Followingthedis ussion
of M Kee (1989), the times ale for supernova sho k waves to destroy all dust in the
gala ti ISM is
t
d
=
9.7
× 10
7
R ǫ(v
s7
)v
−3
s7
dv
s7
yr,
(1.7)where
v
s7
isthesho kvelo ityinunitsof10
7
m/sand
ǫ(v
s7
)
isthegraindestru tion e ien y for a sho k wave with velo ityv
s7
. Jones et al. (1996), onsidering grain3
ele -shatteringee tsongraingrain ollisions(thispro essallowstheredistributionofgrain
massthroughthefragmentationoflargegrainsintosmallgrains),foundalifetimeofthe
order
4
× 10
8
yearsforsili ategrainsand
6
× 10
8
yearsfor arbongrains. Thislifetime
maybe shorter by aboutan order of magnitude in the ase of large grains (
a
≥ 100
nm). ComparingthislifetimewiththeformationrateIobtaint
f
≈ 3×10
9
yr,assuming
adusttogasratiois0.01,typi alsupernovamassinje tionofheavyelementsintoISM
∼ 4M
⊙
,dustformatione ien yof50%inwhi hitisin orporated25%oftheavailable Ointosili atesoroxides,andasupernovarateof1/(30yr)(JonesandTielens,1994).Clearly dustdestru tion seemsto befaster thandust formationbyaboutan orderof
magnitude. Thus, to explainthe presen e of dust in the ISM,I needto assumethat
dust anberea retedandre ondensedintheISMassuggestedbyDwek(1998)and
WeingartnerandDraine(1999). Consideringtheaverageresiden e timeofanatomin
agrain,thetotalmassofISM andthestar formationratefor theMWG, oneexpe ts
that only 20% of Si atoms would be found in the original stardust parti le. On the
otherside, astronomersinferthat the90%of Siismissing fromgasphase. Therefore,
mostoftheSiatomsininterstellargrainsshould bein orporatedintodustintheISM:
mostofinterstellardustisnotofstellaroriginbutitisformedfrom physi alpro esses
intheISM (Draine,2003).
For galaxies like the MWG a possible interstellar formation pro ess is a retion.
Elaboratingaparti ulargalaxyevolutionmodeltakingintoa ountthemetalli ityand
agedependen eonthevariousdustsour es,Asanoetal.(2013)foundthatthea retion
time-s ale depends bythe gasmetalli ity. If metalli ityex eedsa given riti al value
thedustmassgrowthbe omesa tive,withdustgrainsrapidly in reasingtheirmasses
untilmetals aredepletedfrom theISM.This riti almetalli ity hangeswith thestar
formationtimes ale: itislargerforafaststarformationrate(seegure1.1). Thenthe
dustmassgrowthin theISM ouldbethedominantsour eofdustingalaxieslikethe
MWG.
Animportantformofdustmassa retionistheformationof arbona eousmantles
onto sili ate ores. Sili ate materials are refra tory and then more long lived than
arbona eousmaterials. Inhydrogen/ arbonri h,lowtemperatureenvironmentssolid
arbon an be deposited on the surfa es of sili ate ores with the following rate of
in reaseforthedustgrainmass
dm
d
dt
= ξπa
2
n
C
(2.5kT
k
m
C
)
1
2
(1.8)in whi h
m
d
isthemassofthegrain,ξ
isthesti king oe ient,aisthegrainsize,nC
is thenumeri aldensityfor a reting arbonparti les, mC
isthe atomi arbonmassand
T
k
is thekineti temperatureof the gas. Thenif the mantle hasadensityρ
the rateofgrowthofgrain isgivenbyda
dt
=
1
4πa
2
ρ
dm
d
dt
= 0.4ξn
C
ρ
−1
(kT
k
m
C
)
1
2
.
(1.9)The equation 1.9 shows that the mantle growth is independent by the value of
the grain size;in atime interval
∆t
mantles with thesamewidth are formed overall grains. Other fa tors aninuen e themantle formation: i) mantle andesorb morerapidlyfromsmallestgrainsifasigni antUVradiationeldispresent;ii) oagulation
may operate together with the mantlegrowth. The ondition and the me hanism of
deposition suggestthat arbonwhi h form themantle isin theHAC polymeri form
(Jonesetal.,1990). From onsiderationsbasedontheRandomCovalentNetworkmodel
andtheConstraintCountingMethod Jonesetal.(1990)obtainedthe
sp
3
/sp
2
ratiofor
arbona eousmaterials omposed by arbonwith this bonds andhydrogen in atomi
fra tion
f
H
n
sp
3
n
sp
2
=
6f
H−1
8
− 13f
H
.
(1.10)where
n
sp
2
andn
sp
3
arethedensitiesofsp
2
and
sp
3
bondingsites,respe tively.
The hydrogenfra tional abundan e is thus in the range
1/6 < f
H
< 8/13
, whi h in per entage results between 17% and 62%. From equation 1.10 I an infer thatde reasingthehydrogenfra tionalabundan ein reasethequantityof arbonhybridized
sp
2
andin reasingthehydrogenabundan ein reasethequantityof arbonhybridized
sp
3
. It is known that in the hydrogen/ arbonri h and lowtemperature environment
of the ISM solid arbon will be deposited on grains ore. At low temperatures, the
intera tions of H atoms with solid arbon produ e
H
2
and the me hanism by means the arbon is deposited is the hemi al me hanism of arbon insertion, parti ularlye ient at low temperature for both C atoms and
C
+
ions fa ilitating the grow of
hydrogenated arbon hains. ThereforewhenHAC materialis deposited onthe grain
surfa eit ishydrogenri hand, from theequation (1.10), itisessentially
sp
3
bonded.
Due to thermal annealing orexposure to UV radiation hydrogen atoms maydesorbe
from themantle,leadingto anin reaseinthe
sp
2
bonddensity. Thenetresultisthe
formation ofadouble layerof arbona eousmaterials: an outerfreshly deposited
sp
3
layer embedding an internalsp
2
zone. Clearly, during the grain lifetime, the inverse
pro essisalsopossible,inwhi hthegraphitizedmaterialabsorbs(hot)hydrogenatoms
in itsmatrixand
sp
3
bondsin rease(Mennellaetal.,2001).
Themodel exploited in thisthesisis inspiredbythiss enarioin whi hthe arbon
is deposited in the aliphati (
sp
3
bonded) form and it is pro essed by annealing an
radiationUV inthearomati (
sp
2
1.4 The problem of osmi abundan es
An important and open question in our understanding of the ISM is the amount of
metals in orporatedinto dust grains,aquantity that annotbederiveddire tly from
observations. The hemi al omposition of thedust is found by omparing the
abun-dan es of the various elements in the gas phase alone derived from highly sensitive,
high-resolutionUVabsorptionmeasurementswithanISreferen eabundan eset(e.g.,
Milleret al. 2007). Additional information omesfrom X-rayabsorption and
s atter-ingmeasurements(Valen i andSmith,2013),analysisof omponentsofinterplanetary
dustparti les(Bradleyetal.,1999),and,indense louds,fromIRobservationsof
solid-stateabsorption bands (Öberg et al.,2011). Thederiveddust metalabundan esare,
however,veryun ertain, be ause su h inferen erelies onthe assumed ompositionof
thetotal,gas+dust,abundan esintheISM.
Hydrogen and helium onstitute about 98% of the matter in the Universe. The
elements that are mostly lo ked in dust grains are C, O, N, Mg, Si and Fe. The
determinationoftheirreferen eabundan eshasalonghistory. Be ausetheSunisthe
loseststar,its hemi alabundan es anbedeterminedwithhigha ura y. Moreover,
the hemi al omposition of the Sun wasthought to be typi al for our Galaxy, and,
indeed, solar omposition represents a good referen e for several astrophysi s studies
asGala ti hemi alevolutionofolderstarpopulationsandsolar-typestars. About30
yearsago,Greenberg(1974)publishedaremarkablepaperentitledin partwhere have
thoseatomsgone?: thiswastherst riti allookattheproblemof hemi alabundan es
intheISM.Greenbergpointedoutthat,onthebaseoftheCoperni ussatellitedata,the
abundan esofelementssu h as arbon,oxygen,andnitrogen,mu hmorewasmissing
fromthegasthat ouldbein orporatedintodustunderanyreasonablemodels. Twenty
years later Snow and Witt (1996) gave an answer to the Greenberg's question: the
una ountableatomswerenevermissingafterall. Inotherwords,solarabundan esare
notaproxiefortheISM.Basi ally,the ru ialproblemistounderstandtowhatextent
thelo al ISMis hemi allyhomogeneous,andthetypesofstarsto beusedasproxies
forISM abundan es: theSun,younghotstarsoryoung oolstars? All of these stars
havebeenexploitedbutthederivedabundan esoftendisagree,andmoreover,theyare
rather errati (Soa, 2004). From the paper put forward by Snow and Witt (1995),
osmi abundan es posed severe onstraints on dust models: the wasno longer any
needtondhiddenreservoirsofinterstellarC,N,andO.Moreover,theformersurplus
of rawmaterials for theinterstellardust wasturned into ashortage. Snow andWitt
(1995)determineda arbonabundan ein theISMof225
±
50ppM,againsta urrent value(atthat time)of335ppM for arbonabundan ein solarenvironment(Grevesseetal.,1994). Thisproblemopenedtheso alled arbon risis,inwhi hexistingmodels
of osmi dustneededto bemodiedtaking intoa ounttheratherlimitedelemental
budgetavailable.
Now, theelemental abundan esrepresent anadditional onstraint to introdu e in
modelsreprodu inginterstellarextin tion. Theverytightlimitsimposed bytheSnow
and Witt (1995) observations were fortunately soondismissed. From newer
observa-tions,itwasdis overedthat itwasquitehardtodene astandardreferen efor
inter-stellarabundan es,withobservationstowardsstarsofdierenttypesshowingdierent
amountsofmetals in orporatedinto dust (Zubkoet al.,2004). The on entrationsof
theelements hangesigni antlyalongdierent linesof sight,with awidedispersion.
phaseistheindire tpro edure,thatis, tomeasurethegasphaseabundan eand
sub-tra tthis fromthetotal(gas+dust)abundan e. Inthesolarneighbourhood, hemi al
abundan es anbeinferredfrom absorptionlinestudiesof oldandwarmISMorfrom
emissionline spe tros opyalongtheOrionnebula. Overlargedistan es,evenin other
galaxies, region HII an be onsidered good sites for the determination of hemi al
abundan es; but the omposition ofionized gaspresentsimportant ompli ations due
to the dependen e of the derived abundan es on the indi ators used in the analysis
(Simón-Díaz andStasi«ska,2011),totheionization orre torfa torsandtolarge
sys-temati errors from temperature u tuations in the nebula (Mathis, 1995). A good
alternativeto nd areferen e for the hemi al omposition of ISM is representedby
unenvolvedearlyBtypestars. Thesestarshaveamassabout
8
− 18 M
⊙
andthey an beobservedinthesolarneighbourhoodandalsoatlargerdistan esintheMWGandinothergalaxies. Theirlifeisveryshort,sotheyhavenotthene essarytimetotravelfar
awayfromtheirbirthsiteandspe tros opi studiesofearlyBtypestarsaresimple
be- ausetheirphotosphereisnotae tedbystellarwinds. Butdespitethissimplephysi s,
spe tralanalysesofmainsequen eBtypestarsprodu edun learresults,summarized
in a tenden y towards a metal poor omposition in omparison with olderstars like
theSunandalargerangeinelementalabundan es. A ordingto somestudiesBstars
annotbe onsideredana urateindi atorofISMabundan es,be auseelement
strati- ationduetodiusionis ommoninthesestars(HempelandHolweger,2003). Thus,
theabundan esmeasuredattheirsurfa esdonotne essarilyree ttheabundan esof
the loudsfrom whi h theyare formed. Thelatestworkson Bstarabundan esseem
to easedieren esbetweenthemand solarvalues(NievaandSimón-Díaz 2011,Nieva
and Przybilla 2012). Re ently, studying nearby B stars, Nieva and Przybilla (2012)
foundanhighdegreeof hemi alhomogeneityofthestarsin thesample,sothey ould
deneaCosmi Abundan eStandard(CAS)forthea tual hemi al ompositionofthe
osmi matterinthesolarneighbourhood. IntheTable1.1itishighlightedtheredu ed
CASmetalli ity,
Z = 0.014
obtainedwithrespe ttothe anoni alsolarone,Z
⊙
= 0.02
(AndersandGrevesse,1989).Hydrogen Helium Metals
Massfra tion 0.710 0.276 0.014
Table1.1: Massfra tionsforhydrogen,heliumandmetalsa ordingtoNievaandPrzybilla(2012).
In thetable (1.2)abundan es, derivedobserving along dierent type of stars, are
displayed-Thegasphase omponentofthelo alISMiswellknownbe ausedetermined
bymanyUVopti alabsorptionstudies. Su h omponentappearsindependentbythe
lineofsighttowardstheassumedreferen estaranditsuggeststhatthereisastandard
lo al ISM abundan e. Parti ularlyyoungstars should provide agood referen esin e
little hemi alevolutionaso urredsin etheirformation. ForthisreasonBstarsand
younglatertypestars(asFandGstars)arepopularproxiesfortheISM omposition.
In thetable (1.2)to inferthe abundan esavailable for elements lo kedin grains it is
su ienttomakethedieren ebetweenthetypi al hemi al ompositionofthestarand
theISMgasphaseabundan es. NievaandPrzybilla(2012)obtainedtheirabundan es
analysing 29 Bstars withsophisti ated models, inferring asili ate ri h and relatively
poor arbonnatureofthelo alISMdust. Otherstudiesseemto onsiderthat Bstars
Carbon Nitrogen Oxygen Magnesium Sili on Iron
Gas − phase
a
91±
6 62±
4 389±
9 1.5±
0.1 2.2±
0.3 0.2±
0.0 B starsTotal
b
214±
20 62±
6 575±
66 36.3±
4.2 31.6±
3.6 33.1±
2.3 Dust 123±
23 0±
7 186±
67 34.8±
4.2 29.4±
3.6 32.9±
2.3 F&
GstarsTotal
c
358±
82 445±
156 42.7±
17.2 39.9±
13.1 27.9±
7.7 Dust 267 56 41.2 37.7 27.7 SunTotal
d
269±
31 68±
8 490±
56 39.8±
3.7 32.4±
2.2 31.6±
2.9 Dust 178 6 101 38.3 30.2 31.4Table1.2: Observed,gas-phase,andinferred,dust-phase,abundan esinthediuseISMpermillions
ofhydrogennu lei.
(a)
Soaetal.(2011)forC,Meyer etal.(1997)forN,Cartledgeetal.(2004)forO,Cartledgeetal.
(2006)forMg,Si,Fe.
(b)
NievaandPrzybilla(2012).
(c)
Soaand Meyer (2001) obtainedas averages fromthe surveys byEdvardssonet al.(1993) and
Tomkinetal.(1995).
(d)
Asplundetal.(2009).
Fand G starswere onsideredgood andidate asanproxiesforthe ISM abundan es
(Soa andMeyer,2001). Abundan es for thesestars provide reasonablevaluesfor all
elements,but thenitrogen(seetab. 1.2).
Aparti ularproblemisrepresentedbytheoxygendepletionintotheISM.Are ent
study ofJenkins (2009)presentedanextensivereanalysisof ar hivaldatafor 17
ele-ments. Inthisworkisunderlinedthatoxygenisbeingdepletedfromdiuseinterstellar
gas at a rate that ex eeds by far that at whi h it an be in orporated into sili ates
and metalli oxides parti les. Studying a wide range of environments, from tenuous
inter loudgasanddiuse loudstodense loudswherei emantlesandgaseousCOare
importantreservoirsofO,Whittet(2010)triedtounderstandwhat ouldbethe
poten-tial arrier. He on ludesthat themostplausiblereservoir anbeaformofObearing
arbona eousmattersimilartotheorgani matterfoundin ometaryparti lesreturned
bytheStardust mission(vanDishoe k,2008). Furtherstudiesarene essaryto assess
if this lass of materials is present in quantities su ientto a ount for asigni ant
fra tionoftheuna ounteddepletedoxygen.
1.5 Dust in our galaxy
In spiral galaxies like the MWG, dust resides mainly in the gala ti dis , suggesting
itsasso iationwiththerelativelystellardis population. Within thedis , mostofthe
material is onnedto the spiralarms. In ourgalaxydust representsonly1% of the
ISMmass(Williams,2000).
Thedistributionofinterstellarmatterinthedis ofourgalaxyisextremelyuneven
with inhomogeneitieson allsize s alesfrom
10
−4
to
10
3
p . Thegeneraltenden y of
the extin tion to in reasewith the distan e, due to the in reaseof traversed olumn
density, is not linearwith the distan e. Usually lumps witha density of dust above
diuse louds are themosttransparent louds, theyare omposed mainly by
atomi gas, although astrikingly mole ular inventory has been dete ted (Liszt
etal.,2008);thedensityis intherange
n
H
∼ 100 − 500 cm
−3
,whilegas
temper-atures are
30
− 100
K (Snow and M Call, 2006). Usually these louds havean extin tionA
V
≤ 2
mag; translu ent louds these obje tsare denser than diuse louds; they havean
extin tionof
2
≤ A
V
≤ 5
andthey anbeinthephaseofgravitational ontra tion whi h istheprelude tostarformation; dense louds theyare thedensestgas ondensations(
10
3
− 10
6
cm
−3
);kineti
temperaturesrangesbetween10and50K;theyaremainly omposedbymole ular
gasand,duetotheirhighextin tionsobs ureba kgroundstars.
OurSolarSystemresidesinarelativelytransparentinter loudregionneartheedge
of theOrion Armof theMWG with little reddening,
E
B−V
≤ 0.03
, for stars within a distan e of 100p . Onaverage,in a distan e of1 kp areddening ofE
B−V
≈ 0.6
o urs. Then,exploiting theequations(1.4)and(1.6)Iobtaintheaverageratioofthevisualextin tionto pathlength
h
A
L
V
i ≈ 1.8 mag kpc
−1
.
(1.11)Thisestimateis validonlyfordistan es uptoafewkiloparse sfromthe Sun,and
forlinesofsightwhi hare losetotheplaneoftheMWG.Atgreaterdistan esitisvery
hard to determinethis ratio be auseeven the most luminousstars be ome too faint;
e.g., themagnitudeof asupergiant,6.5 kp faraway,is about20 magforanaverage
reddening. Theextin tion towardtheIR lusterat thegala ti enter isestimated to
be
A
V
≈ 30
magfor apathof∼
8kp (Ro he,1988). Forsimilar largedistan es IR photometry ouldbeusefulandinformationsaboutvisualextin tion ouldbeinferredfromassumptionsregardingthewavelengthdependen eoftheinterstellarextin tion. In
generalobservationsthatextendbeyond3kp arebasedonlongwavelengthastronomy.
For somelines of sight extin tion is oddlylow, making possible to extend studies at
visible wavelengthsuptodistan es of
∼
10kp .Correlationofdustwithgaswas al ulatedusingUVabsorptionlinespe tros opy
ofreddenedstarswithin1kp oftheSun. (SnowandM Call,2006)foundthefollowing
relation
N
H
E
B−V
= 5.9
× 10
21
cm
−2
mag
−1
(1.12)to be representativeofdust in diuse regions(Gudennavaret al. 2012, olle ting
ab-sorption linedatatoward3008stars,updatethis valueto
6.12
× 10
21
cm
−2
mag
−1
). Inthis relation
N
H
is onsideredasthesumofmole ularandatomi hydrogenN
H
= N (HI) + 2N (H
2
)
(1.13)with thefa tor twoallows for the fa t that
H
2
ontainstwoprotons. Exploiting the equations (1.4) and (1.6) I an onvert reddeningE
B−V
in extin tion in the visual wavelengthA
V
N
H
A
V
= 1.93
× 10
21
cm
−2
mag
−1
.
This equation is representative of dust in diuse louds. Combining theequation
(1.14) with the equation (1.11), I aneliminate
A
V
and write the average hydrogen numberdensityhn
H
i = h
N
H
L
i ≈ 1.04 cm
−3
(1.15)ThisistheaveragedensityofISMinthedis ofourgalaxy,butsubstantialdeviations
(alsoofordersofmagnitude) anbefoundforindividualregions ausethemattertends
tobedistributedinto lumpswith
n
H
>>
hn
H
i
andinter lump gaswithn
H
<<
hn
H
i
. Usingtheequation(1.15)todeterminethemassdensityandanheights aleof100pfortheISMI analsondthesurfa emassdensity:
hσ
H
i = 5.3 M
⊙
pc
−2
.
Awayfromthegala ti dis theextin tionisgenerallylow,butitsentityisimportant
to study intrinsi properties of external galaxies. Be ause the low densityof dust in
the halo of our galaxy, the reddening of stars whi h are observed in high latitude (
|b| > 20
◦
)isindependentfromthedistan e. Consideringthedis asaat,uniformslab
withthesolar systemin the entralplane I aninferadependen e of extin tionfrom
thelatitudeb
A
V
(b) = A
P
cosec
|b|.
(1.16) Thisdependen efollowsa ose antlawandA
P
isthevisualextin tionatthegala ti poles4
. Clearly thisformulationmustbe onsidered justasanapproximationbe ause
the dust distributionis very inhomogeneous, and it annot be representedon small
s ales by asmoothlyvarying fun tion; indeed some louds situated at high latitudes
are denseenough to ontain amole ular phaseand to produ esigni antextin tion:
A
V
∼ 1
(Penprase, 1992). Many of these dense louds are extensions of lo al dark loud omplexes,but other loudsappeartobeisolated. Thenforlinesofsightwhi hpassthroughthese loudstheequation (1.16)will predi talowerextin tionthan the
ee tivelyobservableone.
In theMWGISECsshowagreat varietyof features and shapesfor dierentlines
ofsight. Ingure(1.2)areplotted threegala ti ISECsobservedalongthree dierent
linesofsight. Broadly hara teristi sofatypi alISEC anbesummarizedinthisway:
i) alinear trend in the visible portion of ele tromagneti spe trum; ii) the bump at
4.6
µm
−1
(toexplainthepresen eofthisprominentfeaturemanyhypothesisaboutits
naturewerebeenelaboratedovertheyears);iii)anonlinearriseintheFarUltraViolet
(FUV)rangeof theele tromagneti spe trum.
Cardelliet al.(1989)elaboratedan analyti formulawhi hallowsto derivean
ex-tin tionlawdepending only from oneparameter:
R
V
. This work startedby the on-sideration that dust grains must suered the same hemi al pro esses for the wholedimensionaldistribution,thatis hemi alandphysi alpro essesmustnotbesele tive
andtheya tedwithoutdistin tiononthewholedustgrainspopulation.
Fitzpatri kandMassa(2007)presentedanextensiveset of328Gala ti extin tion
urvesthatshowaremarkablevarietyofshapesandfeatures. These urveshavebeen
des ribed in the UV (
λ
−1
≥ 3.7 µ
m−1
) by means of a 7parameterdes ription. The
ommon features in the Fitzpatri k and Massa (2007) sample are a Drude prole, a
nonlinearriseintheFUV,andalinearbaseline
E(λ
− V )
E(B
− V )
=
c
1
+ c
2
x + c
3
D(x, x
0
, x
1
),
x
≤ c
5
,
c
1
+ c
2
x + c
3
D(x, x
0
, x
1
) + c
4
(x
− c
5
)
2
,
x > c
5
,
(1.17) 40 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 1 2 3 4 5 6 7 8 9
A
λ
A
V
λ
−1
(µm
−1
)
ALS908 CD-424120 COLLINDER46318Figure 1.2: ISECs observed alongthreegala ti dierentlinesof sigth: ALS908, CD-424120and
COLLINDER46318. where
x
≡ λ
−1
inµm
−1
andD(x, x
0
, x
1
) =
x
2
(x
2
− x
2
0
)
2
+ x
2
x
2
1
.
(1.18)These authors produ ed an estimate of the typi al or average wavelength
depen-den eof interstellar extin tion. Whilethe derivation isstraightforward,theextent to
whi hthepropertiesoftheaverageISECree ttheaveragepropertiesoftheGala ti
extin tion is problemati to dene. As a onsequen e, it is di ult to derivea
stan-dard for Gala ti dust. In turn this means that, in thesame way, it is very hard to
quantifyhowpe uliarthenatureofdustisalongalineofsightwhoseextin tiondiers
signi antlyfrom theaverage ISEC.Indeed, the shapes ofthe ISECsmay simply
re-e t manifestationsofdustgrainsinwhi hparti ularphysi al/ hemi alpropertiesare
emphasised, leavingrelativelyunmodiedtheunderlyingglobalpi ture.
1.6 Dust in external galaxies
An important dis overyin extragala ti astronomywasthe identi ationof luminous
IR galaxies: galaxies whi h emit more energy in the mid and far IR than all other
wavelengths. The sour e of su h emission is heated dust. The presen e of dust in
galaxies removeshalf ormoreof thestellarenergyfrom theUVopti alrange oming
galaxies in the nearby Universe ontain dust. Morethan onethird ofthe bolometri
luminosityoflo algalaxiesistransferredbydustinIR.Thefra tionofstellarradiation
repro essedintotheIRAS8-120
µm
windowisaround25%30%;in ludingbolometri orre tionsfromtheIRASwindowtothefullIRrange,theIRenergyfra tionbe omes35%40%ofthetotalbolometri energy within
∼ 100
Mp . This average value ould besubje ttolargevariationsbetweengalaxies,dependingonluminosity,morphologi altypeanda tivitylevel(Calzetti,2001).
Dustlimitsastronomersabilitytoknowthelo alanddistantUniverseposingserious
problems problems in the interpretation of the spe tral energy distribution. Any
a - uratedeterminationsoffundamental parameters,su hasage,stellarpopulationmix,
starformationrates,and stellarInitialMass Fun tionare prevented. Moreoveryoung
stellarpopulationsaremoredeeplyembeddedindusty loudsthanolderstars,implying
thatapoorknowledgeofthestarformationhistoryingalaxies. Finally,measuringthe
fra tion of metals lo ked into dust at dierentepo hs is is possibleto tra e hemi al
evolutionof galaxies. The dust ontentof galaxies isproportionalto both metalli ity
andgas ontent;asstellarpopulationevolvesintimeandmoregasislo kedintostars,
the metalli ityof agalaxy in reaseswhile itsgas ontentsde reases. Galaxiesin the
lo alUniverseareonlymoderatelyopaque,andextremevaluesoftheopa ityareonly
foundinthestatisti allynondominantmorea tivesystems.
Frommultiwavelengthopti al andnearIRimagingofedgeonornearlyedgeon
systemsindi ate thatthes aleheightofthedustisabouthalfthatofthestars,whi h
meansthatthedustismore on entratedneartheplaneofthedisk,anditappearsmore
radiallyextended thanthestellar distribution;thedust massis foundto beaboutan
order of magnitudemorethan previouslymeasured using theIRAS uxes, indi ating
the existen e of a old dust omponent. The gas to dust mass ratio is lose to the
valuederivedforourGalaxy(Xilourisetal.,1999). IRAS100
µm
(Nelsonetal.,1998) andISOlongwavelengthmaps(Radovi hetal.,2001) onrmthat olddustemissionextends beyond the limits of the opti al disks along the radialdire tion, with s ales
thatare
∼
40%largerthanthoseofBbandemittingstars,butstillwellwithintheHI disks.Irregulargalaxieshavealowermetalli itythanspiralgalaxies:in irregulargalaxies
the massin interstellarmetals perunit
L
B
5is roughly3times smaller thanin spiral
galaxies.
Among lo al galaxies there is aspe ial groupof galaxies: starburst galaxies. The
average UV and IR luminosities perunit omoving volume in rease with redshift by
roughlyanorder ofmagnitudeuptoz
∼
1-2. Despitetheheterogeneousset that UV-sele tedstarburstsform,theirdustreddening andobs urationpropertiesareuniform.Inthesesystems,thegeometryofdustthatbestdes ribesthereddeningoftheionized
gas emission is that of a foregroundlike distribution. In starbursts region, internal
dustdoesn'tappeartobepresentinlargeamounts,andthelittlepresentislikelytobe
in ompa t lumps(Calzetti et al., 1995)and the prin ipal sour eof opa ityappears
to be given bydust that is external, or mostlyexternal, to the starburst region, but
still internal to thehost galaxy, similar to a lumpy dust shell surrounding a entral
starburst.
Wehavesomeinformationsaboutthenatureofdustinthenearestgalaxiesbelonging
to the lo al luster. Thelo al luster in ludes
∼
70 galaxies. The twomost massive membersofthe luster,theMWGandtheAndromedaGalaxy(M31)haveasystemofsatellitegalaxies. AmongtheothersarefoundtheSmallandLargeMagellani Clouds
5
(MCs), satellites of our own galaxy. The MCs and the M31 are the only external
galaxieswheresofarithasbeenpossibletostudytheextin tion alonglinesofsightto
individual stars.
LargeMagellani Cloud(LMC):thisgalaxyislo atedatadistan eof50kp and
ithasanalmostfa eonviewingangle. Ithasmetalli ityatlevel0.5oftheMilky
Way ISM (Welty et al.,2001), anda dust to gas ratio is redu ed bya fa tor4
(Fitzpatri k,1986). Thetotalaveragestarformationrateisrelativelylowas0.26
M
⊙
yr
−1
(Kenni uttetal.,1995),whiletheoneoftheMWGis0.68
−1.45 M
⊙
yr
−1
(Robitaille and Whitney, 2010). These physi al onditions are very similar tothoseoftheearlyevolutionofhighredshift andmetalpoorgalaxies. Stars more
than
∼
500p awayfromthe30Doradusregion,astarburstregionofLMC,showR
V
≈ 3.2
with anextin tion law resembling the MWG diuse loud extin tion law(Misseltetal.,1999). Inthe30Doradusregiontheextin tionlawshowsverydierent properties ompared to the gala ti average one. Indeed, asthegure
(1.3)shows,thefeature at4.6
µm
−1
isweakenedandin theFUVastrongerrise
thanthat ommonintheMWGisobservable(Cartledgeetal.,2005).
Small Magellani Cloud (SMC): thisgalaxy islo ated at 61kp awayfrom the
MWG andit is seenalmost fa eonfrom ourgalaxy. It has adust to gasratio
redu ed by a fa tor 10 with respe t to the MWG (Bou het et al., 1985). Its
metalli ityis
0.2
×
theMWGmetalli ity(Bernardetal.,2008). ObservedISECs forthisgalaxyareverydierentfromtheaverage urvefortheMWG(seegure1.3), indeedGordon et al. (1997)dis overedthat UV extin tion in SMCis very
similartothatobservedinstarburst galaxieswhi hhavebeendete tedfromlow
tohigh (
z > 2.5
)redshifts. Thebump strengthisfurther diminishedthanISECs forLMCor absent,asin AzV18,andtheriseintheFUVisfurther steep. Andromeda Galaxy: this isaspiralgalaxylo ated atadistan eof750kp from
theMWGand ithas ametalli ityverysimilarto that of ourgalaxy(Blair and
Kirshner,1985). Forthisgalaxywehavepoorinformations butobservedISECs
are more similar to the Gala ti average one than typi al interstellar urves of
SMCand LMC; animportant dieren e isthat thebump strengthis weakerin
M31thanin MWG(Bian hiet al.,1996).
A uratemeasurementsof extin tion urvesare almost ex lusively limited to the
Galaxy,theLMC,andtheSMC,be auseatgreaterdistan esitbe omesimpossibleto
obtainthephotometryorspe tros opyofindividualstarsneededforextin tion
deter-minations. Inthelast de adeobservershavedis overedhugequantities of interstellar
dust near themostdistantquasarsin theveryyounguniverse, only700million years
afterprimordialnu leosynthesis.Su hstudies(e.g.,Jiangetal.2006)indi atethathigh
redshift quasarsystemshavebeenalreadyenri hedwithdust uptoalevel omparable
to nearbydustygalaxies. Thereisavigorousdebateandtwodistin ts enariosforthe
originof dustat high redshifts,in whi h supernovaeand lower-massasymptoti giant
bran hstarsareinvolved(seeSe tion 1.3). Whateverthe asemaybe,dustexists at
thehighestredshiftsprobedsofar(
z
∼ 6
).Abouttwode adesago,quasarswithdampedLy
α
systemsintheforegroundhave beenstudiedbyPeietal.(1991)andwerefoundtobeonaverageredderthanthoseMilkyWayaverage urve AzV18 Sk-68 155
λ
−1
(µm
−1
)
A
λ
A
V
10 9 8 7 6 5 4 3 2 1 0 10 8 6 4 2 0Figure1.3: AverageISECfortheMWG(Fitzpatri kandMassa,2007) omparedtoanISECobserved
inLMCnearthe30Doradusregion(Sk-68155)(Cartledgeet al.,2005)andaninterstellarextintion
urveobservedinSMC(AzV18)(Cartledgeetal.,2005).
extin tion, and fully ompatible withSMC extin tion. Su h ndingswere questioned
byMurphyandLiske(2004)thatfoundnotra eofextin tion. Yorketal.(2006)found
no eviden e of the 217.5 nm bump, and extin tion urves similar to SMC extin tion
in asample of quasars of theSloan DigitalSky Survey. Extin tion urves that dier
from the SMC, LMC, and MWG have been obtained in the ase of A tive Gala ti
Nu lei(AGNs),although mildlyreddenedquasarsappeartofollowanSMC-like
dust-reddeninglaw(e.g.,Hopkinsetal.2004). Inthe aseofstar-forminggalaxiesextin tion
urvesaredi ulttoinferduetothe rowdingofdustwithemittingsour es. Starburst
galaxies anbe hara terisedbythela kingofthe217.5nmabsorptionfeature,likethe
SMC,andasteepfar-UV rise,intermediatebetweenthose intheSMCandtheMWG
extin tion urves. However,forUV luminousgalaxies,theextin tion urvesrange
be-tweenthosetypi aloftheSMCandLMC,withsomeofthemexhibitingweakbumpsas
inLMC(NollandPierini,2005). Ingeneral,asigni ant217.5nmbumpisobservedin
thespe traof star-forminggalaxies at
z
∼ 2
,indi atinganLMC-likeextin tion urve (Nolletal.,2007). Chenet al.(2006)analysedtheextin tion urvefor10GammaRayBurst(GRB)hostgalaxies,ndingrelativelyatandgrayproles,togetherwithothers
1.7 Dust models
Afterrstobservationswhi hhighlightedthepresen eofinterstellardustinourgalaxy,
astronomersbegintodevelopinterstellardustmodelstakingthe uefromtheobserved
interstellar extin tion and theelemental depletions in the ISM. Current dust models
tra etheiran estrytothemodelofMathisetal.(1977)(MRNfrominitialsofauthors'
surname). Those authors assumed the existen e of two distin t types of interstellar
dust, graphitegrains andsili ate grains. All grains were assumedto bespheri aland
to havea ontinuousdistribution ofradii,
a
, from 5nmto 250nm, with thenumber ofgrainswithradiiin therangea
→ a + da
beingproportionaltoa
−3.5
; i.e.,thereare
verymany moresmall grains than large. The al ulations were performedusing Mie
theory(1908)andthismodelwassu essfulinreprodu ingverywelltheaverageISEC
in theopti alandUV. Manyofthedistin tivefeatures oftheMathiset al. modelare
retained in moresophisti ated models. There are too many models dis ussed in the
re entliteratureforustobe omprehensive,soIshallfo usontwoexamplesofmodels
that havesomesimilaritiesandsomedieren es.
1.7.1 Draine's model
Draineandhis many ollaboratorshavemadethemost omprehensive, onsistent,and
impressive developments of the MRN pi ture, and have applied their model to the
widest range of observational tests with onsiderablesu ess. Draine and Lee(1984)
introdu ed opti al fun tions for "astronomi alsili ate",derivedfor the olivinefamily
of sili ates from laboratory data in dierent wavelength regimes, pie ed together in
a manner onsistentwith the Kramers-Kronigrelations but omitting any feature not
observedininterstellardustspe tra. Thisapproa hhasbeenwidelyadoptedbyothers.
TheworkofDraineandLee(1984)wasrevisedandalsoextendedintotheX-rayregime
for both graphiteand astronomi alsili ate byDraine (2003). Wherethe model isfor
spheri algrainsofuniformmaterialthe al ulationswere arriedoutbyMietheory.
Asigni ant hangetotheMRNpi turehasbeentheassumptionbymanymodellers
thatPAHmole ulesformanimportantadditional omponentintheISM,equivalentin
somerespe tstoverysmallgrains. Thepotentialimportan eofthePAH ontribution
to dust-related phenomenahas beenexplored by LiandDraine (2001b),Weingartner
and Draine (2001), Zubko et al. (2004), and more re ently byDraine and Li(2007).
The eviden e supporting the in lusion of aPAH omponent is thewidespread
obser-vation (espe iallybytheSpitzer IR spa emission)of theUnidentied InfraredBands
(so- alledUIBs)between3.3and12
µ
m. Manydierentmodelsnowin orporateaPAH omponent(usinglaboratoryandtheoreti alopa ities)whoseabundan eisadjustedtoreprodu e the observed strength of the UIBs. If they are present at this level, then
PAHs will also ontribute signi antly to the "bump" and to the far UV extin tion.
Weingartnerand Draine(2001)nd that atrimodal arbona eousgrainsize
distribu-tion isneeded to mat hextin tion urve,theUIBs, and IR emission -thus departing
signi antlyfrom thesimplepowerlawdistributionoftheMRNmodel.
ThemodelbyDraineand ollaboratorshasbeenappliedtoextin tionintheMWG
and othergalaxies(DraineandLi,2007),toIR emissionin ludingthedust- orrelated
mi rowaveemissionobservedbyCOBE(DraineandLazarian,1998),andtopolarization
inthefar-IRandsubmillimetreDraineandFraisse(2009),andX-rays attering(Draine
Travis 1994, Mish henko et al. 1997). The model does, however, appear to require
ex essiveamountsofMg,Si,andFe,andalsostrainstheabundan eof arbon,leaving
littleavailableforthegas-phase arbon-bearingspe iesC
+
,C,andCO.
1.7.2 The Messina/Cagliari model
A model with signi ant dieren es from that of Draine and ollaborators has been
developedbyresear hersatMessinaandCagliari. The omponentsofthismodelin lude
"astronomi alsili ates",solid arbon,andPAHs, asintheDrainemodel;however,the
arbonisassumedtobehydrogenatedamorphous arbon ontainingbothsp
2
(H-poor
graphiti )andsp
3
(H-ri hpolymeri ). A avityis allowedtobepresentin thesili ate
ores,tomimi possibleporosityandtorelieveabundan e onstraintsonsili on. Fifty
typesof PAHs are alsoin luded, and these are in neutral, ation, di ation and anion
forms,withopa ities omputedab initiobyMallo ietal.(2007). Themostsigni ant
dieren e is that the arbon (other than the PAHs) is assumed to be deposited on
thesili ate ores,ratherthanasseparate arbongrains. Thisdes riptionissuggested
by the work of Jones et al. (1990)who argued that there is a y le of arbonin the
ISM; arbonis deposited ongrains asH-ri h sp
3
(polymeri ) arbon,annealed to sp
2
(graphiti ) arbonbytheinterstellarradiationeld,andultimatelyremovedinsho ks.A further dieren efrom other models isthat small andlarge grainsare distin tsets
ratherthan ontinuous;however,the
a
−3.5
distributionappliestobothsmallandlarge
grains. This model an t a wide variety of extreme MWG extin tion urves, and
also some unusual extin tion urves from external galaxies. The dieren es between
dierent lines of sight require hanges to arbonlayer, rather thanto the underlying
(refra tory)sili ates; thus theymay beregardedasevolutionary hanges. Themodel