Oxygen abundances in the most oxygen-rich spiral galaxies

0.25 MB·English

by  L. S. Pilyugin

Mon.Not.R.Astron.Soc.000,000–000 (0000) Printed5February2008 (MNLATEXstylefilev2.2)
Oxygen abundances in the most oxygen-rich spiral galaxies
Leonid S. Pilyugin1, Trinh X. Thuan2 and Jos´e M. V´ılchez3
1Main Astronomical Observatory of National Academy of Sciencesof Ukraine, 27 Zabolotnogo str., 03680 Kiev,Ukraine
2Astronomy Department, University of Virginia, P.O.Box 3818, UniversityStation, Charlottesville, VA 22903
3Instituto de Astrof´ısica de Andaluc´ıa, CSIC, Apdo, 3004, 18080 Granada, Spain
6 Received2005November28;inoriginalform2005July26
0
0
2
ABSTRACT

n Oxygen abundances in the spiral galaxies expected to be richest in oxygen are esti-
a mated. The new abundance determinations are based on the recently discovered ff
J – relation between auroral and nebular oxygen line fluxes in high-metallicity Hii re-

6 gions. We find that the maximum gas-phase oxygen abundance in the central regions
of spiralgalaxiesis 12+log(O/H)∼ 8.75.This value is significantly lower(by a factor

1 >∼ 5) than the previouslyaccepted value. The centraloxygenabundance in the Milky
v
Way is similar to that in other large spirals.
2
2 Key words: galaxies: abundances – ISM: abundances – Hii regions
1
1
0
6
0 1 INTRODUCTION genabundancesforasampleofspiralgalaxieswhichincludes
/ NGC 3184, NGC 3351, NGC 5194, and NGC 6744. They
h
p The oxygen abundance in spiral galaxies has been the sub- found generally lower oxygen abundances with (O/H)C ∼
- jectofmuchdiscussion foralongtime.Whichspiralgalaxy 9.0.
o istheoxygen-richestone?Andhowhighistheoxygenabun- Here we consider anew the problem of the maximum
r
t danceintheoxygen-richestgalaxies?Duetothepresenceof oxygenabundanceinspiralgalaxiesbyattemptingtoderive
s radialabundancegradientsinthedisksofspiralgalaxies,the moreaccurateoxygenabundances.Thesecanbederivedvia
a
: maximumoxygenabundanceoccursattheircenters.Theno- theclassicalTe–method,Tebeingtheelectrontemperature
v tation (O/H)C = 12+log(O/H)R=0 will be used thereafter oftheHiiregion.Measurementsoftheaurorallines,suchas
Xi to denotethe central oxygen abundance. [OIII]λ4363, are necessary to determine Te. Unfortunately,
r The chemical composition of various samples of spi- they are very faint and often drop below the detectability
a ral galaxies has been discussed in a number of pa- levelinthespectraofhigh-metallicityHiiregions. Pilyugin
pers (e.g. Vila-Costas & Edmunds 1992; Zaritsky et al. (2005) hasadvocated that thefaint auroral linefluxcan be
1994; Garnett et al. 1997; van Zee et al. 1998; Garnett computedfromthefluxesinthestrongnebularlinesviathe
2002). According to those articles, the oxygen-richest ff – relation. Then, using the obtained flux in the auroral
galaxies are: NGC 5194 (M 51) with (O/H) = 9.54 line, accurate oxygen abundances can be derived using the
C
(Vila-Costas & Edmunds 1992), NGC 3351 with (O/H)C classicalTe–method.Wewillestimate(O/H)Cinthespiral
= 9.41 (Zaritsky et al. 1994), NGC 3184 with (O/H) galaxies reported tobetheoxygen-richest ones,NGC3184,
C
= 9.50 (van Zee et al. 1998), NGC 6744 with (O/H) = NGC3351, NGC 5194, and NGC6744. Forcomparison, we
C
9.64 (Garnett et al. 1997). Different versions of the one- will also consider NGC2903, NGC2997, NGC5236 aswell
dimensionalempiricalmethod,proposedfirstbyPagel et al. theMilky Way Galaxy.
(1979) a quarter of a century ago, have been used for oxy- We describe the method used to determine oxygen
gen abundance determinations in those papers. Pilyugin abundances in Hii regions in Section 2. The oxygen
(2000, 2001a,b, 2003) has shown that the oxygen abun- abundances in the spiral galaxies NGC 3184, NGC 3351,
dances in galaxies determined with the one-dimensional NGC 5194, NGC 6744, NGC 2903, NGC 2997, NGC 5236
empirical calibrations are significantly overestimated at andtheMilkyWayGalaxyaredeterminedinSection3.We
the high-metallicity end ( 12 + log O/H > 8.25). This discuss the reliability of the derived abundances and sum-
for two reasons. First, the then-existing calibrating points marize our conclusions in Section 4.
at the high-metallicity end were very few and not reli- For the line fluxes, we will be using the following no-
able. Second, the physical conditions in Hii regions cannot tations throughout the paper: R = I /I ,
2 [OII]λ3727+λ3729 Hβ
be taken into account accurately in one-dimensional cal- R = I /I , R = I /I , R =
3 [OIII]λ4959+λ5007 Hβ [OIII]λ4363 Hβ 23
ibrations. Pilyugin et al. (2004) have used instead a two- R + R . With these definitions, the excitation parameter
2 3
dimensional parametric empirical calibration to derive oxy- P can then be expressed as: P = R /(R +R ).
3 2 3
(cid:13)c 0000RAS
2 L.S. Pilyugin, T.X. Thuan and J.M. V´ılchez
2 ABUNDANCE DERIVATION Eq.(10)willbehereinafterreferredtoastheflux–fluxorff
–relation. Itwasfoundthatthisrelationship ismetallicity-
2.1 Adopted equations for the T method
e dependentat low metallicities, butbecomes independentof
A two-zone model for thetemperature structurewithin the metallicity (within the uncertainties of the available data)
Hii region was adopted. Izotov et al. (2005) have recently at metallicities higher than 12+logO/H ∼ 8.25, i.e. there is
published a set of equations for the determination of the one-to-onecorrespondencebetweentheauroralandnebular
oxygen abundance in Hii regions for a five-level atom. Ac- oxygenlinefluxesinspectraofhigh-metallicityHiiregions.
cordingtothoseauthors,theelectrontemperaturet within Pilyugin (2005) derived
3
the [Oiii] zone, in units of 104K, is given by the following
logR=−4.264+3.087 logR . (11)
equation 23
1.432 Wecannowmakeuseoftheff–relationtodefinea“discrep-
t = (1)
3 log(R /R)−logC ancy index”, equal to the difference between the logarithm
3 T oftheobservedfluxRobs inthe[Oiii]λ4363lineandthatof
where thefluxRcalofthatlinederivedfromthestrong[Oii]λ3727,
C =(8.44−1.09t +0.5t2−0.08t3)v (2) [Oiii]λλ4959,5007 lines using theff - relation:
T 3 3 3
v= 1+0.0004x3 (3) Dff =logRobs−logRcal. (12)
1+0.044x
3 SinceR =R /P,theff–relationcanbealsoexpressed
23 3
and intheformR=f(R ,P).Wewillconsideranexpressionofthe
3
x =10−4n t−1/2. (4) type
3 e 3
As for the ionic oxygen abundances, they are derived logR=a1+a2 logP +a3 logR3+a4(logP)2. (13)
from the following equations UsingasampleofHiiregionswithrecenthigh-precision
12+log(O++/H+)=log(I /I )+ measurements of oxygen lines fluxes, the values of the co-
[OIII]λ4959+λ5007 Hβ
efficients in Eq.(13) can be derived. Such a sample has
1.251
6.200+ t −0.55logt3−0.014t3, (5) been compiled by Pilyugin (2005). Bresolin et al. (2004)
3 have recently detected the auroral lines [Siii]λ6312 and/or
12+log(O+/H+)=log(I /I )+ [Nii]λ5755anddeterminedTe-basedabundancesin10faint
[OII]λ3727+λ3729 Hβ Hii regions in the spiral galaxy M 51. Unfortunately they
1.676
5.961+ −0.40logt −0.034t + wereunabletodetecttheoxygenauroralline.Usingtheelec-
t 2 2
2 tron temperaturet recommended byBresolin et al. (2004)
3
log(1+1.35x2). (6) and the equations given above for the Te-method, we have
x =10−4n t−1/2. (7) estimated the value of R which corresponds to the t3 for
2 e 2 everyHiiregion.Ithasbeenfound(Pilyugin2005)thatthe
Herene is theelectron density in cm−3. oxygenlinefluxesinsixfaintHiiregionsinthespiralgalaxy
Thetotaloxygenabundancesarethenderivedfromthe M 51 (CCM 54, CCM 55, CCM 57, CCM 57A, CCM 71A,
following equation and CCM 84A) satisfy the ff – relation. These data are in-
cludedinthepresentsampletoenlargetherangesinPand
O O+ O++
= + . (8) R . Thus, oursample consists of a total of 48 data points.
H H+ H+ 3
The values of thecoefficients in Eq.(13) are derived by
Theelectrontemperaturet2 ofthe[Oii]zoneisusually using an iteration procedure. In the first step, the relation
determinedfrom an equationwhich relates t2 tot3,derived is determined from all data using the least-square method.
by fitting Hii region models. Several versions of this t2 – Then, the point with the largest deviation is rejected, and
t3 relation have been proposed. A widely used relation has a new relation is derived. The iteration procedure is pur-
been suggested by Campbell et al. (1986) (see also Garnett sueduntiltwosuccessiverelations havealltheircoefficients
(1992))basedontheHiiregionmodelsofStasin´ska(1982). differing by less 0.001 and the absolute value of the largest
Campbell et al.(1986)hasfoundthatthet2–t3relationship deviationislessthan0.1dex.Thefollowingff–relationwas
can beparameterized as obtained
t2 =0.7t3+0.3. (9) logR = −4.151−3.118 logP +2.958 logR
3
It will be used here. − 0.680(logP)2. (14)
2.2 The ff – relation
2.3 Characteristics of the ff – relation
The fluxes R in the auroral lines are necessary to derive
theoxygenabundancesinHiiregionsusingtheT method. The top panel in Fig. 1 shows the flux R in the oxygen
e auroral line as a function of the flux R in strong nebu-
Buttheyarefaintandoftenundetectableinspectraofhigh- 3
metallicity Hii regions. It was shown (Pilyugin 2005) that lar [Oiii]λ4959+λ5007 lines. The open and filled circles
show individual Hii regions with 12+logO/H > 8.25. The
thefluxRintheaurorallineisrelatedtothetotalfluxR
23 relationscorrespondingtoEq.(14)fordifferentvaluesofthe
in thestrong nebular lines through a relation of the type
excitationparameterareshownbythesolid(P=1.0),long-
logR=a+b×logR . (10) dashed(P= 0.7),short-dashed(P=0.3), anddotted(P=
23
(cid:13)c 0000RAS,MNRAS000,000–000
Oxygen abundances in the most oxygen-rich spiral galaxies 3
-1.0 0.6
0.4
0.2
f
f 0.0
-2.0 D
-0.2
-0.4
)
R -0.6
( -3.0 8.2 8.3 128.+4log(O/H8).5 8.6 8.7
g T
o e
l 0.6
0.4
-4.0 0.2
f
f 0.0
D
-0.2
-0.4
-5.0
-0.9 -0.5 -0.1 0.3 0.7 1.1 -0.6
0.0 0.2 0.4 0.6 0.8 1.0
log(R )
3 P
Figure 2.Deviations fromtheff–relationasparameterized by
-1.0
thediscrepancyindexDff asafunctionofoxygenabundance(top
panel) and of excitation parameter P (bottom panel). The filled
-1.5 circlesareHiiregionsusedinderivingtheff–relation.Theopen
circlesareotherHiiregions.
-2.0
) donotshowacorrelationeitherwithmetallicityorwiththe
R -2.5
( excitation parameter.
g
Let us compare the two ff - relations given by Eq.(11)
o -3.0
l andEq.(14).ThebottompanelinFig.1showsthefluxRin
theoxygenaurorallineasafunctionofthetotalfluxR in
-3.5 23
strongnebular[Oii]λ3727+λ3729and[Oiii]λ4959+λ5007
lines. The filled circles are the Hii regions used in deriving
-4.0 Eq.(14). The open circles are the other Hii regions. The
relationscorrespondingtoEq.(14)fordifferentvaluesofthe
-4.5
excitationparameterareshownbythesolid(P=1.0),long-
-0.1 0.1 0.3 0.5 0.7 0.9 1.1
log(R ) dashed(P= 0.3),short-dashed(P=0.2), anddotted(P=
23
0.1) lines. The relation corresponding to Eq.(11) is shown
Figure 1. Theff–relationsforHiiregions.Top panel. FluxR by the plus signs. Inspection of the bottom panel of Fig. 1
in the oxygen auroral line as a function of the flux R3 in the shows that the relations given by Eq.(14) for values of the
strong nebular [Oiii]λ4959+λ5007 lines for Hii regions. The excitation parameter P ranging from 1 to ∼ 0.3 are close
circles (open and filled) show Hii regions with 12+logO/H > each to other. That led Pilyugin (2005) to conclude that
8.25. The relations corresponding to Eq.(14) for different values there appears to be no third parameter in the ff – relation.
of the excitation parameter are shown by the solid (P = 1.0), Indeed, examination of the bottom panel in Fig. 1 shows
long-dashed(P=0.7),short-dashed(P=0.3),anddotted(P= thattheff–relationsgivenbyEq.(11)andEq.(14)arevery
0.1) lines. Only filled circles are used in deriving Eq.(14). Bot-
similar to each other and reproduce the observational data
ttohme topatanlefll.uFxlRux23Rininthtehsetrooxnyggennebauularorr[aOl ilii]nλe37a2s7a+fuλn37ct2i9onanodf wellathighvaluesoftheexcitationparameter,P >∼ 0.3.At
[Oiii]λ4959+λ5007 lines. The open and filled circles have the the same time there is an appreciable divergence between
thosetworelationsatlowvaluesoftheexcitationparameter,
samemeaningasinthetoppanel.Therelationscorrespondingto
Eq.(14)fordifferentvaluesoftheexcitationparameterareshown P <∼ 0.3.Fordefiniteness’ssake,wewillusetheff–relation
by the solid (P = 1.0), long-dashed (P = 0.3), short-dashed (P given by Eq.(14) for all values of P.
=0.2),anddotted(P=0.1)lines.Therelationcorrespondingto It should be noted however that the particular form
Eq.(11)isshownbyplussigns. of the analytical expression adopted for the ff – relation
maybequestioned.Wehavechosenasimpleform,Eq.(13),
but perhaps a more complex expression may give a better
fit to the auroral – nebular oxygen line fluxes relationship.
0.1) lines. The filled circles show the data used in deriving Furthermore, the coefficients in the adopted expression are
Eq.(14) (40 out of 48 original data points). derived using calibrating Hii regions which have a discrep-
The deviations from the ff – relation given by Eq.(14) ancyindexD aslargeas0.1dexinabsolutevalue.Perhaps
ff
asparameterized bythediscrepancyindexD areshownin smaller absolute values of D may result in a more precise
ff ff
Fig. 2 as a function of oxygen abundance (top panel) and relation. It can be seen that the calibration curves give a
of excitation parameterP (bottom panel).The filled circles satisfactory fittotheobservationaldata.Atthesametime,
are Hii regions used in deriving Eq.(14). The open circles Fig. 1 shows that the available measurements for calibrat-
are the other Hii regions. Fig. 2 shows that the deviations ing faint Hii regions (those with low R ) are very few in
3
(cid:13)c 0000RAS,MNRAS000,000–000
4 L.S. Pilyugin, T.X. Thuan and J.M. V´ılchez
0.4 9.2
M51
0.2 9.0
)
H
/
∆(logO -00..02 g(O/H) 88..68
o
l
+
-0.4 2 8.4
8.2 8.3 8.4 8.5 8.6 8.7 1
12+log(O/H)
Te 8.2
0.4
8.0
0.2 0 2 4 6 8 10 12 14
H) R (kpc)
/
O
g 0.0
o Figure 4.Radial distributionof oxygen abundances inthe disk
l
( of the spiral galaxy M 51 (NGC 5194). The open circles denote
∆ -0.2
(O/H)ff abundances.Thesolidlineisthelinearleast-squarebest
fittothosedata.Thefilledcirclesshow(O/H)Te abundancesfrom
-0.4 Bresolinetal. (2004). The dashed line is their linear regression
0.0 0.2 0.4 0.6 0.8 1.0
P fit.
0.4
genabundances,i.e. negative∆log(O/H),just theobserved
) 0.2 trend. In general, the (O/H)ff abundances agree satisfac-
H torily with the (O/H) abundances and do not show any
/ Te
O
g 0.0 systematic trend.
o
l In summary, the flux in the faint auroral line
(
∆ -0.2 [Oiii]λ4363canbeestimatedreasonablywellfromthemea-
sured fluxes in the strong nebular lines [Oii]λ3727+λ3729
-0.4 and [Oiii]λ4959+λ5007 via the ff – relation. Then, using
-0.3 -0.2 -0.1 0.0 0.1 0.2 0.3
D the derived flux in the auroral line, the oxygen abundance
ff
can befound through theclassic T – method.
e
Figure3.Thedifference∆log(O/H)=log(O/H)Te –log(O/H)ff
asafunctionoflog(O/H)Te (top panel), ofexcitation parameter
P(middle panel), andofdiscrepancyindexDff (bottom panel).
3 OXYGEN ABUNDANCES IN SPIRAL
GALAXIES
number.Furthermore,theRvaluesforthefaintHiiregions
Here the derived ff – relation will be applied to derive the
from Bresolin et al. (2004) are not measured but estimated
oxygen abundances in a number of spiral galaxies. First,
from the electron temperatures t and, consequently, they
3 M51(NGC5194),theoxygen-richestspiralgalaxy((O/H)
C
are not beyond question. Clearly, high-precision measure-
=9.54)inthesampleofVila-Costas & Edmunds(1992)will
mentsof oxygen lines fluxesin faint Hii regions are needed
be considered. The value of (O/H) in M 51 was derived
C
to check thederived ff – relation.
recently from determinations of (O/H) abundances in a
Te
number of Hii regions by Bresolin et al. (2004). Compari-
sonbetweentheradialdistributionsof(O/H) and(O/H)
Te ff
2.4 The (O/H) abundances
ff abundancesinthediskofM51providesanotherpossibility
Using the Te method, both (O/H)Te abundances based to test the reliability of the latter. Second, (O/H)ff abun-
on the measured line fluxes Robs, and (O/H) abundances dancesinNGC3184,NGC3351,andNGC6744whichhave
ff
based on the line fluxes Rcal determined from the ff – rela- beenreportedtobethemostoxygen-richspirals,willbede-
tion, can be derived for our sample of Hii regions. Com- termined.Third, we will consider threewell-observed spiral
parison of their values gives us a check on the (O/H) galaxies NGC2903, NGC2997, andNGC5236 forcompar-
ff
abundances. Fig. 3 shows the difference ∆log(O/H) = ison.Fourth,theff–relationwillbeappliedtoGalacticHii
log(O/H) – log(O/H) as a function of log(O/H) (top regions toselect high-precision measurements.
Te ff Te
panel), of excitation parameter P (middle panel), and of
discrepancy index D (bottom panel). Inspection of Fig. 3
ff 3.1 Oxygen abundances in M 51
shows that the differences ∆log(O/H) do not show an ap-
preciable correlation with metallicity orwith excitation pa- ThecompilationofpublishedspectraofHiiregionsinM51
rameter. There is however an anticorrelation of ∆log(O/H) is taken from Pilyugin et al. (2004). The Hii regions with
with D . This can be easily understood. The discrepancy measuredR andR fluxesfromBresolin et al.(2004)(their
ff 3 2
index appears to be an indicator of the error in the auro- Tables 1 and 6) have been added. We determine (O/H)
ff
rallineRmeasurements.Inthatcase,positivevaluesofD abundances for every Hii region using the derived ff – re-
ff
imply overestimated R fluxes, which result in turn in over- lation. The (O/H) abundances in the Hii regions of M 51
ff
estimated electron temperatures and underestimated oxy- are shown as a function of galactocentric distance in Fig. 4
(cid:13)c 0000RAS,MNRAS000,000–000
Oxygen abundances in the most oxygen-rich spiral galaxies 5
byopencircles.Thesolid lineisthelinearleast-squarebest 9.2
fit to those data: 9.0 NGC3184
)
12+log(O/H)=8.74(±0.03)−0.025(±0.004)×R. (15) /H 8.8
(O 8.6
Thedistance of M 51 (d= 7.64 Mpc) and theisophotal ra- g
o 8.4
dius (R = 5.61 arcmin) were taken from Pilyugin et al. l
25 +
(2004). The filled circles are (O/H) abundances deter- 2 8.2
Te 1
mined by Bresolin et al. (2004). The dashed line is their 8.0
linear regression:
7.8
0 2 4 6 8 10 12
12+log(O/H)=8.72(±0.09)−0.02(±0.01)×R. (16) R (kpc)
Fig. 4 shows that the radial distribution of (O/H)
ff 9.2
abundances agrees well with that of the (O/H) abun-
Te 9.0 NGC3351
dances(comparealsoEq.(15)andEq.(16)).Thisisastrong
)
argument in favor of the reliability of the (O/H) abun- H 8.8
ff /
dances. (O 8.6
SomeHiiregionsshowlargedeviationsfromthegeneral og 8.4
l
radialtrendinoxygenabundances.Someofthosedeviations +
2 8.2
can be real, but some are probably caused by uncertainties 1
in the oxygen abundance determinations. It should be em- 8.0
phasized that the uncertainties in the (O/H)ff abundances 7.80 2 4 6 8 10 12
are not necessarily caused by uncertainties in the line flux
R (kpc)
measurements. It has been noted (Pilyugin 2005) that the
ff – relation gives reliable results only if two conditions are 9.2
satisfied; i) the measured fluxesreflect theirrelative contri-
9.0 NGC6744
butionstotheradiationofthewholenebula,andii)theHii )
H 8.8
regionisionization-bounded.Ifthesetwoconditionsarenot /
met, then the derived (O/H)ff abundances may be signifi- g(O 8.6
cantly in error. o 8.4
l
+
2 8.2
1
3.2 Oxygen abundances in the most oxygen-rich 8.0
spiral galaxies 7.8
0 4 8 12 16 20 24
The compilation of published spectra of Hii regions in the R (kpc)
spiral galaxies NGC 3184, NGC 3351, and NGC 6744 was
taken from Pilyugin et al. (2004). The (O/H) abundances Figure5.Radialdistributionsof(O/H)ff abundancesinthespi-
ff ral galaxies NGC 3184, NGC 3351, and NGC 6744. The open
in the Hii regions of NGC 3184 is shown as a function of circlesshowindividualHiiregions.Thelinesaretheleast-square
galactocentric distancein thetop panel of Fig. 5. Thesolid fitstothosedata.
line is the linear least-square best fit to those data:
12+log(O/H)=8.73(±0.03)−0.035(±0.004)×R. (17) regions in the spiral galaxies NGC 2903, NGC 2997, and
The same data are shown for the Hii regions of NGC 3351 NGC 5236 was taken from Pilyugin et al. (2004). Recent
measurements from Bresolin et al. (2005) havebeen added.
in the middle panel of Fig. 5. The solid line is the linear
Oxygen abundances were calculated for every Hii region
least-square best fit to those data:
using thederived ff – relation.
12+log(O/H)=8.74(±0.02)−0.023(±0.004)×R. (18) The (O/H)ff abundances in the Hii regions of
NGC 2903 are shown as a function of galactocentric dis-
Similarly, the data for the Hii regions of NGC 6744 are
tance in the top panel of Fig. 6. The solid line is the linear
showninthebottompanelofFig.5.Thelinearleast-square
least-square best fit to those data:
best fit to those data is given by:
12+log(O/H)=8.68(±0.02)−0.026(±0.003)×R. (20)
12+log(O/H)=8.75(±0.04)−0.022(±0.003)×R. (19)
The middle panel of Fig. 6 shows the same for NGC 2997.
The distances of the galaxies and the galactocentric dis-
The linear least-square fit tothose data is:
tances of the Hii regions were taken from Pilyugin et al.
(2004). 12+log(O/H)=8.59(±0.05)−0.021(±0.007)×R. (21)
Figs. 4 and 5 (see also Eqs.(15,17, 18,19)) show that
The data for NGC 5236 are shown in the bottom panel of
thevalues of (O/H) in thesespiral galaxies are ∼ 8.75.
C Fig. 6. The linear least-square fit to those datais:
12+log(O/H)=8.62(±0.01)−0.022(±0.005)×R. (22)
3.3 Oxygen abundances in other spiral galaxies
ThedistancesofthegalaxiesweretakenfromPilyugin et al.
Wewill considerherethreewell-observed spiralgalaxies for (2004).
comparison.ThecompilationofthepublishedspectraofHii Fig. 6 (see also Eqs.(20,21,22)) shows that the(O/H)
C
(cid:13)c 0000RAS,MNRAS000,000–000
6 L.S. Pilyugin, T.X. Thuan and J.M. V´ılchez
9.2 9.2
9.0 NGC2903 Te9.0
12+log(O/H) 8888....2468 12+log(O/H) 88888.....02468
8.0 7.8
0 2 4 6 8 10 12 14 16
7.8 R (kpc)
0 2 4 6 8 10 12 14 16
R (kpc)
9.2 9.2
9.0 NGC2997 Te9.0
)
H) 8.8 /H 8.8
/ O 8.6
g(O 8.6 og( 8.4
o 8.4 l
l + 8.2
+ 2
12 8.2 1 8.0
8.0 7.8
0 2 4 6 8 10 12 14 16
7.8 R (kpc)
0 2 4 6 8 10 12
R (kpc)
Figure 7. Radial distribution of (O/H)Te abundances in the
diskoftheMilkyWayGalaxy.Toppanel. Theopencirclesshow
9.2
(O/H)Te for all Hii regions. The solid line is the linear least-
9.0 NGC5236 square fit to those data. Bottom panel. The filled circles show
H) 8.8 (O/H)Te abundances for Hii regions with an absolute value of
/
(O 8.6 thediscrepancyindexDff lessthan0.05dex.Thesolidlineisthe
g linear least-square fit to those data. The open square shows the
lo 8.4 oxygenabundanceintheinterstellargasatthesolargalactocen-
+
2 8.2 tricdistance.
1
8.0
7.8 Thederived(O/H) abundancesareshownasafunctionof
0 1 2 3 4 5 6 7 Te
R (kpc) galactocentric distance by the open circles in the top panel
ofFig.7.Thelinearleast-squarebestfittothosedata(with
Figure6.Radialdistributionsof(O/H)ff abundancesinthespi- two points with a deviation more than 0.2 dexrejected)
ral galaxies NGC 2903, NGC 2997, and NGC 5236. The open
circlesshowindividualHiiregions.Thelinesaretheleast-square 12+log(O/H)=8.79(±0.05)−0.034(±0.005)×R (23)
fitstothosedata.
is shown by thesolid line in thetop panel of Fig. 7.
There is a relative large number of measurements of
abundances in NGC 2903, NGC 2997, and NGC 5236 are abundances in the Milky Way Galaxy. They show a large
smaller than 8.70, i.e. lower than in NGC3184, NGC 3351, scatter since the quality of the data obtained over a pe-
NGC 5194, and NGC 6744. riod of more than thirty years is necessarily heterogeneous.
Thus,themaximumgas-phaseoxygenabundanceinthe Some measurements have a low accuracy. The ff – relation
oxygen-richest spiral galaxies is 12+log(O/H) ∼ 8.75. This providesawaytoselect outtheHiiregionswithhighqual-
valueissignificantlylower(byafactorof5ormore)thanthe ity measurements (Pilyugin & Thuan 2005). The bottom
previousvalue based on the one-dimensional empirical cali- panelofFig.7showsthe(O/H)Te abundancesasafunction
brations (e.g. Vila-Costas & Edmunds 1992; Zaritsky et al. ofgalactocentricdistanceforobjectswithanabsolutevalue
1994;van Zee et al. 1998; Garnett 2002). of the discrepancy index Dff less than 0.05 dex. The linear
least-square best fit to those data
12+log(O/H)=8.70(±0.04)−0.022(±0.004)×R (24)
3.4 Oxygen abundances in the Milky Way Galaxy
isshownbythesolidlineinthebottompanelofFig.7.The
We now consider the Milky Way Galaxy. A compilation
open square is theoxygen abundanceof theinterstellar gas
of published spectra of Galactic Hii regions with an au-
at thesolar galactocentric distance (see below).
roral [Oiii]λ4363 line has been carried out by Pilyugin et
The central oxygen abundance in the Milky Way ob-
al. (2003). That list contains 69 individual measurements tained here, (O/H) ∼ 8.70, is significanly lower than the
of 13 Hii regions in the range of galactocentric distances C
widelyused(O/H) =9.38ofShaveret al.(1983).Ouroxy-
C
from 6.6 to 14.8 kpc. Recent measurements of six Galactic
gen abundance radial gradient is in agreement with that of
HiiregionsbyEsteban et al.(2005)wereaddedtothatlist,
Daflon & Cunha (2004). They derived therelation
resulting in a total of 75 measurements. The T -based oxy-
e
gen abundances were estimated using the equations above. 12+log(O/H)=8.762(±0.105)−0.031(±0.012)×R (25)
(cid:13)c 0000RAS,MNRAS000,000–000
Oxygen abundances in the most oxygen-rich spiral galaxies 7
from abundancedeterminations in young OBstars. istence of theStasin´ska’s effect, theysuggest that thegreat
We note that the central oxygen abundance obtained majorityofHiiregionsingalaxiesareinametallicityrange
herefortheMilkyWayisclosetothatinotherlargespirals. where this effect is not important. Indeed, the Stasin´ska’s
effectbecomesimportantonlyatoxygenabundanceshigher
than 12 + log(O/H) ∼ 8.7. We have found that this ”crit-
ical” value is reached only in the central part of the most
4 DISCUSSION AND CONCLUSIONS
oxygen-rich galaxies. We conclude that, in the light of our
The(O/H) abundancesdeterminedhererelyon thevalid- present results, Hii regions with 12+log(O/H) > 8.7 are
ff
ityoftheclassicT –method.Thishasbeenquestionedfor rare, if they exist at all. It should be noted that Stasin´ska
e
thehigh-metallicityregimeinanumberofinvestigationsby also questionedtheexistenceofsuch metal-rich Hiiregions
comparison with Hii region photoionization models. One (Section 2.2 in Stasin´ska 2005).
should keep in mind however that the existing numerical Anotherfactor thatcan affect Te-basedabundancede-
models of Hii regions are far from being perfect (Stasin´ska terminations is the temperature fluctuations inside Hii re-
2004), and, hence, the statement that Hii region models gions (Peimbert 1967). If they are important, the (O/H)Te
provide more realistic abundances as compared to the T abundance would be a lower limit. In most cases the effect
e
– method should not be taken for granted (Pilyugin 2003). appears to besmall, probably under0.1 dex (Pagel 2003).
RecentlyStasin´ska (2005)hasexaminedthebiasesinvolved Fortunately, there is a way to verify the validity of the
inTe-basedabundancedeterminationsathighmetallicities. Te – method at high metallicities (Pilyugin 2003). High-
She has found that, as long as the metallicity is low, the resolution observationsoftheweakinterstellarOiλ1356ab-
derived(O/H) valueisveryclosetotherealone.Discrep- sorption lines towards stars allow one to determine the in-
Te
ancies appear around 12+log(O/H) = 8.6 - 8.7, and may terstellar oxygen abundance in the solar vicinity with a
become very large as the metallicity increases (Fig. 1a in very high precision. It should be noted that this method
Stasin´ska 2005). The derived (O/H) values are smaller is model-independent. These observations yield a mean in-
Te
than the real ones, sometimes by enormous factors. This terstellaroxygenabundanceof284–390Oatomsper106 H
means that, if such metal-rich Hii regions exist, the classic atoms (or 12+log(O/H) = 8.45 – 8.59) (Meyeret al. 1998;
T –methodwillalwaysleadtosystematicallylowerderived Sofia& Meyer 2001; Cartledge et al. 2004). Oliveira et al.
e
oxygen abundances. (2005) have determined a mean O/H ratio of 345 ±19 O
Can the effect discussed by Stasin´ska be responsible atoms per 106 H atoms (or 12+log(O/H) = 8.54) for the
for the low (O/H) abundances obtained here? Let us as- Local Bubble.Thus, an oxygen abundance12+log(O/H) =
Te
sume that it is the case, and that the true central oxy- 8.50 in the interstellar gas at the solar galactocentric dis-
gen abundances in the spiral galaxies are higher than 12 tanceseems tobeareasonable value.Thevalueof theoxy-
+ log(O/H) = 9.0. What radial distributions of (O/H) genabundanceatthesolargalactocentricdistancetracedby
Te
abundances in the disk of spiral galaxies would one expect (O/H)Te abundances in Hii regions is then in good agree-
in this case? Starting from the periphery of a galaxy, the mentwiththeoxygenabundancederivedwithhighprecision
(O/H) abundances would increase with decreasing galac- from the interstellar absorption lines towards stars (open
Te
tocentricdistance(following thetrueabundances) untilthe square in the bottom panel of Fig. 7). This is strong evi-
(O/H)Te abundance reaches the value of 12+log(O/H) ∼ dencethattheclassicTe–methodprovidesaccurateoxygen
8.7.Afterthatthe(O/H) abundanceswoulddecreasewith abundancesin high-metallicity Hii regions.
Te
decreasing galactocentric distance because of the Stasin´ska The measured oxygen abundance in the solar vicinity
(Fig. 1a in Stasin´ska 2005) effect, although the true abun- provides an indirect way to check our derived central oxy-
dances continue to increase with decreasing galactocentric gen abundances. The oxygen abundance in the solar vicin-
distance. Thus, the radial distribution of (O/H) abun- ityis 12+log(O/H)=8.50, and agas mass fraction µ∼ 0.20
Te
dances should show a bow-shaped curve with a maximum seems appropriate. The simple model of chemical evolution
valueof12+log(O/H)∼8.7atsomegalactocentricdistance. of galaxies predicts that a decrease of µ by 0.1 results in a
Do the actual data (Fig. 5–7) show such a behavior? increaseoftheoxygenabundanceby∼0.12dexintherange
The derived radial distributions of (O/H) abundances in of µ from ∼ 0.75 to ∼ 0.05. When all the gas in the solar
Te
the disks of spiral galaxies do not show an appreciable cur- vicinity will be converted into stars, the oxygen abundance
vature, and the (O/H) abundances increase more or less willincreaseby0.2–0.25dexandwillreachavaluearound
Te
monotonicallywithdecreasinggalactocentricdistance.Only 12+log(O/H)=8.70–8.75.Thisvalueisinexcellentagree-
theHiiregionsintheverycentralpartsofthespiralgalax- ment with the central oxygen abundances obtained above
ies NGC 3351 (the middle panel of Fig. 5) and NGC 2997 fortheMilky Wayand othergalaxies. Thereis noroom for
(themiddlepanelofFig.6)mayshow(O/H) abundances central oxygen abundances as large as 12+logO/H = 9.00
Te
which are slightly lower than expected from thegeneral ra- or greater.
dialabundancetrends,buttheeffectisveryslight andmay Thepresentstudysuggeststhatthereisanupperlimit
not be real if errors of 0.1 dex (Fig. 3) in the calibration totheoxygen abundancesin spiral galaxies. Iftrue,theex-
and in the line intensity measurements are taken into ac- isting luminosity – metallicity relation should be revisited.
count. To settle the question of whether the abundances in One can expect the metallicity to increase with luminos-
themetal-richest HiiregionsareaffectedbytheStasin´ska’s ity up to 12+(O/H) ∼ 8.75, but to remain approximately
effect,manymoreaccuratemeasurementsofHiiregions(in- constant for higher metallicities, resulting in a flattening of
cludingthoseintheverycentralparts)inthemostoxygen- theluminosity–metallicityrelation.Thissuggestionwillbe
rich spiral galaxies are necessary. investigated in a future paper.
Thus,while ourresultsdonotruleout thepossible ex- The maximum gas-phase oxygen abundance in spiral
(cid:13)c 0000RAS,MNRAS000,000–000
8 L.S. Pilyugin, T.X. Thuan and J.M. V´ılchez
galaxies is 12+log(O/H) ∼ 8.75. The central oxygen abun- Pilyugin, L.S., Ferrini, F., & Shkvarun,R.V., 2003, A&A,
danceintheMilkyWayissimilartothatinotherlargespi- 401, 557
rals. Some fraction of the oxygen is locked into dust grains Pilyugin, L.S.& Thuan,T.X. 2005, ApJ, 631, 231
inHiiregions.Esteban et al. (1998)foundthatthefraction Pilyugin, L.S., V´ilchez, J.M., & Contini, T. 2004, A&A,
of the dust-phase oxygen abundance in the Orion nebula is 425, 849
about 0.08 dex. Then, the true gas+dust maximum value Shaver, P.A., McGee, R.X., Newton, L.M., Danks, A.C.,
of the oxygen abundancein Hiiregions of spiral galaxies is Pottash, S.R.1983, MNRAS,204, 53
12+log(O/H) ∼ 8.85. This value can increase up to ∼ 8.90 Sofia U.J., & Meyer D.M. 2001, ApJL, 554, 221L
–8.95iftemperaturefluctuationsinsideHiiregionsareim- Stasin´ska, G. 1982, A&AS,48, 299
portant. Stasin´ska, G. in: Cosmochemistry. The melting pot of the
elements. XIII Canary Islands Winter School of Astro-
physics.Eds.C.,Esteban,R.J.,Garc´iaL´opez,A.,Herrero,
Acknowledgments F., S´anchez, Cambridge contemporary astrohysics, Cam-
bridge, UK,Cambridge UniversityPress, 2004, p.115
We thank Yuri Izotov and Grazyna Stasin´ska for provid-
Stasin´ska, G. 2005, A&A,434, 507
ing us with a new set of T -equations in advance of pub-
e
Vila-Costas, M.B., &Edmunds,M.G.1992, MNRAS,259,
lication. We thank the anonymous referee for helpful com-
121
ments.Theresearchdescribedinthispublication wasmade
vanZee,L.,Salzer,J.J.,Haynes,M.P.,O’Donoghue,A.A.,
possible in part by Award No UP1-2551-KV-03 of the U.S.
& Balonek, T.J. 1998, AJ, 116, 2805
Civilian Research & Development Foundation for the Inde-
ZaritskyD.,KennicuttR.C.,Jr.,&HuchraJ.P.1994,ApJ
pendent States of the Former Soviet Union (CRDF). L.S.P
420, 87
wasalsopartlysupportedbygrantNo02.07/00132fromthe
Ukrainian Fund of FundamentalInvestigations.
REFERENCES
Bresolin, F.,Garnett,D.R.,&Kennicutt,R.C.2004, ApJ,
615, 228
Bresolin, F., Schaerer, D., Gonz´alez Delgado, R.M., &
Stasin´ska, G. 2005, A&A,441, 981
Campbell, A.,Terlevich,R., &Melnick, J. 1986, MNRAS,
223, 811
Cartledge, S.I.B., Lauroesch, J.T., Meyer, D.M. & Sofia,
U.J., 2004, ApJ, 613, 1037
Daflon, S. & Cunha, K.2004, AJ, 617, 1115
Esteban, C., Peimbert, M., Torres-Peimbert, S. & Es-
calante, V. 1998, MNRAS,295, 401
Esteban,C.,Garc´ıa-Rojas,J.,Peimbert,M.,Peimbert,A.,
Ruiz, M.T., Rodr´ıguez, M., & Carigi, L. 2005, ApJ, 618,
L95
Garnett, D.R.1992, AJ, 103, 1330
Garnett, D.R.2002, ApJ, 581, 1019
Garnett, D.R., Shields, G.A., Skillman, E.D., Sagan, S.P.,
& Dufour, R.J. 1997, ApJ,489, 63
Izotov, Y.I., Stasin´ska, G., Meynet, G., Guseva, N.G., &
Thuan,T.X. 2005, A&A,in press (astro-ph/0511644)
Kennicutt,R.C.,Bresolin, F.,&Garnett,D.R.2003, ApJ,
591, 801
MeyerD.M.,JuraM.,&Cardelli J.A.1998, ApJ,493,222
Oliveira,C.M.,Dupuis,J.,Chayer,P.,&Moss,H.W.2005,
ApJ,625, 232
Pagel,B.E.J.2003,in:CNOintheUniverse,Eds.C.Char-
bonnel,D.Schaerer,&G.Meynett,(ASPConferenceSe-
ries No 304), p.187
Pagel, B.E.J., Edmunds, M.G., Blackwell, D.E., Chun,
M.S., & Smith,G. 1979, MNRAS,189, 95
Peimbert, M. 1967, ApJ, 150, 825
Pilyugin, L.S. 2000, A&A,362, 325
Pilyugin, L.S. 2001a, A&A,369, 594
Pilyugin, L.S. 2001b, A&A,373, 56
Pilyugin, L.S. 2003, A&A,399, 1003
Pilyugin, L.S. 2005, A&A,436, 1L
(cid:13)c 0000RAS,MNRAS000,000–000