Metallicity of galaxies at high redshift

系外銀河における微細構造輝線の観測とその理解 2014年12月2-3日 @国立天文台
高赤方偏移における星形成銀河の微細構造輝線
観測とその理解 (光赤外からのコメント)
矢部清人 (国立天文台)
系外銀河における微細構造輝線の観測とその理解 2014年12月2-3日 @国立天文台
Cosmic Star-formation history：
Redshift
Hopkins & Beacom 2006
SFR density
z=5
Peak epoch at z~1-3
系外銀河における微細構造輝線の観測とその理解 2014年12月2-3日 @国立天文台
Metallicity of galaxies at high redshift：
The Astrophysical Journal Letters, 771:L19 (6pp), 2013 July 10
Zahid et al.
metallicity
Zahid+13, ApJ, 771, L19
stellar mass
Figure 1. MZ relation at five epochs ranging to z ∼ 2.3. The curves are fits to the data defined by Equation (4). The solid curves indicate metallicities determined
using the KK04 strong-line method and the dashed curves indicate metallicities converted using the formulae of Kewley & Ellison (2008). Data presented in this figure
can be obtained from H.J.Z. upon request.
(A color version of this figure is available in the online journal.)
The mass-metallicity relation and its evolution
系外銀河における微細構造輝線の観測とその理解 2014年12月2-3日 @国立天文台
– 9 – redshift：
Metallicity gradient at high
positive gradient
Cresci+2010
(220km/s, 13.5G, 0.03)
(220km/s, 13.5G, 0.05)
(220km/s, 13.5G, 0.07)
(220km/s, 3G, 0.03)
(220km/s, 3G, 0.05)
(220km/s, 3G, 0.07)
(150km/s, 3G, 0.03)
Yuan+2011
negative gradient
•
Steeper
metallicity
gradient
at Red lines are the measurements for Sp1149 at
Fig. 3.—
Left: Metallicity
vs. galactocentric
radius.
highfrom
redshift
z=1.49
this work. The gradient within the central 4.5 kpc is -0.16+/-0.02 dex kpc−1 . Vertical
→
red
dottedInside-out
lines show thegrowth?
annulus used to average/sum the spectra. Purple dashed lines show
the typical gradients of local isolated late-type galaxies, using the control sample of Rupke et al.
Positive gradient
(2010b). The orange dotted line represents the mean gradient of local early-type galaxies, which
Infall
of pristine
gas?
is →
typically
∼ 3 times
shallower than
local late-type galaxies (Henry & Worthey 1999). Blue lines
•
系外銀河における微細構造輝線の観測とその理解 2014年12月2-3日 @国立天文台
Morphology of galaxies at high redshift：
SXDS/CANDELS
Color composites of galaxies at z~1.4 with HST/ACS+WFC3 images
Large variety of galaxy morphology at z>1
系外銀河における微細構造輝線の観測とその理解 2014年12月2-3日 @国立天文台
Kinematics of galaxies at high redshift：
No. 2, 2009
SINS SURVEY OF HIGH-REDSHIFT GALAXIES
1403
Forster Schreiber+09
Figure 17. Velocity fields for 30 of the 62 galaxies of the SINS Hα sample. The velocity fields correspond to that derived from the Hα line emission as described in
Section 5.1 (the exception is K20–ID5 for which it was obtained from the [O iii] λ 5007 line instead). The color coding is such that blue to red colors correspond to the
blueshifted to redshifted line emission with respect to the systemic velocity. The minimum and maximum relative velocities are labeled for each galaxy (in km s−1 ).
All sources are shown on the same angular scale; the white bars correspond to 1"" , or about 8 kpc at z = 2. The galaxies are approximately sorted from left to right
according to whether their kinematics are rotation-dominated or dispersion-dominated, and from top to bottom according to whether they are disk-like or merger-like
as quantified by our kinemetry (Shapiro et al. 2008). Galaxies observed with the aid of adaptive optics (both at the 50 and 125 mas pixel−1 scales) are indicated by the
yellow rounded rectangles.
Large variety of galaxy kinematics at z~2
系外銀河における微細構造輝線の観測とその理解 2014年12月2-3日 @国立天文台
Connection between galaxies and AGN：
BPT
★☆: Stacking analysis
Kewley+01
Yabe et al. 2014
AGN contribution? Different ISM condition?
6
系外銀河における微細構造輝線の観測とその理解 2014年12月2-3日
@国立天文台
[N II]/Hα and [O III]/Hβ ratios seen at high redshift.
Connection between galaxies
and AGN：
(3)
(4)
AGN
AGN
1.5
0.5
0.0
z~0
HII
-0.5
-1.0
LOG ([OIII]/Hβ)
1.0
0.5
z~0.8
0.0
-0.5
larger
q
0.5
LOG ([OIII]/Hβ)
1.0
0.5
z~1.5
0.0
-1.0
higher
n
e
larger q
-2.0 -1.5 -1.0 -0.5 0.0 0.5
LOG ([NII]/Hα)
-0.5
-1.0
1.0
LOG ([OIII]/Hβ)
harder
EUV
0.0
-0.5
-1.0
Kewley et al. 2013
1.0
HII
LOG ([OIII]/Hβ)
LOG ([OIII]/Hβ)
1.0
0.5
z~2.5
0.0
-0.5
-1.0
-1.5 -1.0 -0.5 0.0 0.5
LOG ([NII]/Hα)
-1.5 -1.0 -0.5 0.0 0.5
LOG ([NII]/Hα)
Figure 2. An illustration of the effect of varying different galaxy
parameters on the star-forming galaxy abundance sequence in the
[N II]/Hα versus [O III]/Hβ diagnostic diagram. The original SDSS
star-forming galaxy sequence is well-fit by the red theoretical curve.
Raising the hardness of the ionizing radiation field (orange dashed
line) moves the abundance sequence towards larger [N II]/Hα and
[O III]/Hβ ratios. A similar effect is seen when the electron density of the gas is raised (green dot-dashed line). The relationship
between ionization parameter, metallicity and the [N II]/Hα and
[O III]/Hβ line ratios is more complex. At high metallicities, raising the ionization parameter causes the [N II]/Hα ratio to become
smaller, while [O III]/Hβ is largely unaffected. At low metallicities,
Different ISM condition
at high redshift?
系外銀河における微細構造輝線の観測とその理解 2014年12月2-3日 @国立天文台
Galaxies at high-redshift is dusty!!：
T. Goto et al.: Infrared luminosity functions with the AKARI
Goto et al. 2010
Fig. 16. Evolution of TIR luminosity density based on TIR LFs (red circles), 8 µm LFs (stars), and 12 µm LFs (filled triangles). The blue open
squares and orange filled squares are for only LIRG and ULIRGs, also based on our LTIR LFs. Overplotted dot-dashed lines are estimates from
the literature: Le Floc’h et al. (2005), Magnelli et al. (2009), Pérez-González et al. (2005), Caputi et al. (2007), and Babbedge et al. (2006) are in
cyan, yellow, green, navy, and pink, respectively. The purple dash-dotted line shows UV estimate by Schiminovich et al. (2005). The pink dashed
line shows the total estimate of IR (TIR LF) and UV (Schiminovich et al. 2005).
Contribution to SFRD from LIRG/ULIRG increases
with increasing redshift
系外銀河における微細構造輝線の観測とその理解 2014年12月2-3日 @国立天文台
Metallicity of galaxies at high redshift：
The Astrophysical Journal Letters, 771:L19 (6pp), 2013 July 10
Zahid et al.
metallicity
Zahid+13, ApJ, 771, L19
stellar mass
Figure 1. MZ relation at five epochs ranging to z ∼ 2.3. The curves are fits to the data defined by Equation (4). The solid curves indicate metallicities determined
using the KK04 strong-line method and the dashed curves indicate metallicities converted using the formulae of Kewley & Ellison (2008). Data presented in this figure
can be obtained from H.J.Z. upon request.
(A color version of this figure is available in the online journal.)
Evolution of mass-metallicity relation is real?
high-redshift sample, although without binning them with the rest
系外銀河における微細構造輝線の観測とその理解 2014年12月2-3日
@国立天文台
of the galaxies.
Metallicity of galaxies at
redshift：
2.3 zhigh
= 3–4
Abundances of Luminous Infrared Galaxies
A significant sample of 16 galaxies at redshift between 3 and 4 was
observed by Maiolino et al. (2008) and Mannucci et al. (2009) for the
Mannucci+2010
lower SFR
ULIRG
Seeing only surface metallicity?
Rupke+2008
metallicity
higherSFR
stellar mass
Left-hand panel:
mass–metallicity
of local
galaxies.
g. 11.— Comparison of the mass-metallicity relationFigure
from 1.the
Fig.the
12.—
Differencerelation
between
theSDSS
observed
a
the
thick
central
line
showing
the
median
relation.
The
coloured
lines
show
the
SS (Tremonti et al. 2004) with LIRG and ULIRG abundances
and ULIRGs and the L − Z and M − Z rela
values
of
SFR.
Right-hand
panel: median
metallicity
as a(red)
function
of SFR
for gal
stellar masses. The average LIRG and ULIRG are significantly
of infrared
luminosity.
The
thick
open
d
with
increasing
SFR
at
constant
mass.
er-abundant, as they are when compared to the L − Z relation.
deviations from the M − Z relation. There is
dotted lines show 1σ scatter on either side of the mean SDSS
agreement between the two, though compariso
tion, which has been shifted upward by 0.1 dex to account for
lation yields a slightly larger under-abundance
Low metallicity in dusty and high SFR galaxies?
ratio continues
to increase,
the to
decreasing
Charlot
& Longhetti
(2001). despite
We prefer
estimate
the ionof these line ratios,
we have
summarized
in Figure(1987).
9 a logical highest masses, despite the fact that both methods are predomP05. electron
semiempirical
method
of Veilleux
& Osterbrock
temperature.
Eventually,explicitly,
at still higher
metallicities,
ization
parameter
based
upon nitrothe
process
whereby asofmany
estimates
of both
the chemical
Our
calibration
this
ratio
is
shown
in
Figure
7, with inantly based on H ii regions with Te metallicities. At the lowest
method
is
available
for
only
546/27,730
(2%)
of
The
direct
T
gen becomes
the dominant coolant in the nebula, and the e abundance and ionization parameters as are
系外銀河における微細構造輝線の観測とその理解 2014年12月2-3日
@国立天文台
theoretical
models.
compatible
polynomial
fits as in
equation
(4)
and coefficients
given in stellar masses, this difference disappears. The difference between
electron temperature falls sufficiently to ensure
that thein our
the galaxies
SDSS
sample.
The
[O
iii]
k4363
line
is
weak
with the
Table
3. data set can be obtained using the techniques
nitrogen line4.4.
weakens
increasing
metallicity.
Since only
described
inlow
this
section.
Thethis
process
canscales
beThe
automated,
and is usually
observed
inmetallicity,
metal-poor
galaxies.
SDSS
Diagnostic
Diagram
The S23with
At
very
ratio
simply
asand
the the empirical methods may be attributed to the different H ii rethe [N ii] line is produced in the low-excitation zone of
an
IDL
script
to
do
this
is
available
on
request
from
the
first
contains
very
fewabundance,
metal-poor
they
nitrogen
to galaxies
first order.because
However,
it isare
known gion samples used to derive the calibrations. At the highest metOur
of the
ratiois ([S
!!6717,catalog
6731
+[S
iii]
the
H calibration
ii region, [N
ii]/H!
alsoii] sensitive
to
ionization
author (L. J. K.). However, we will describe a more direct
in thisand
metallicity
regime,
the nitrogen
abundance allicities, the PP04 methods utilize four H ii regions with detailed
compact,
faint (e.g.,
Terlevich
et al.in1991;
!!9069,
9532)/H" (popularly known as S23intrinsically
) is shown in rare, that
parameter.
method
for the derivation
of these
parameters,
x 7 below.
No. 1, 2002
ABUNDANCES
IN EXTRAGALACTIC
H
REGIONS
41
Masegosa
et al. 1994;
van
Zeeii 2000).
Panel
(10)
ofauthors’
Figuretechni1
From the
comparative
analysis
with
other
4.6. The ½N ii"=½O iii" Diagnostic Diagram
ques
which
follows,
we
are
confident
that
this
new
technique
shows that a total of 477 Te metallicities is insufficient to obtain a
1.0
1.5
Figure
6. Again,
we (7)fit fourth-order
defined as
q=3e8
(7)
will
provide
the most
abundancepolynomials
estimates currently
The
advantage of
using
[N ii] and [O iii] lines is that they
q=3e8reliable
1.5e8 (6)
clear
M-Z relation.
Because
we
are
unable
to fitZ an
M-Z
relation
1.5e8 (6)
=are
0.5 solar
in
equation
(4),
and
the
coefficients
given
in
Table
3.
8e7 (5)
possible.
are unaffected
by
absorption
lines
originating
from
underly8e7 (5)
Kewley
& they
Dopita
02 usinglines
4e7 (4)
do not
Te of
method
further
in
Te metallicities,
4e7 all
(4) suchthe
As
is the
caseconsider
in
ratios
forbidden
to recombina0 we
2e7 (3)
ing1.0stellar populations,
lie close to Balmer
that
2e7 (3)
1e7 (2)
has a maximum at a certain metaltion lines, the S23
1e7 ratio
(2)
can be used to eliminate
errors due to dust reddening,
this work.and
5e6 (1)
5. COMPARISON
WITH OTHER
5e6 (1)
7
0.5
licity,
and
therefore
itlarge
is twofor
valued
at all other metallicities.
they are
both strong and easily observable in The
the optical.
6
scatter
in
the
M-Z
relation
is
all metallicity
calBRIGHT-LINE
TECHNIQUES
0.5
5
For
this
particular
ratio
the
maximum
occurs at a somewhat
Both empirical
and
theoretical
relationships
for
the
[N
ii]/
4
ibrations;
the
rms
residual
about
the
line
of
best
fit
is
0.08–0.13.
Comparison
[O iii] 3ratio as a function of oxygen abundance currently
higher abundance5.1.
than
for the RData
23 ratio; at metallicities of
2
-1
The cause of the roughly
scatter
in
the
M-Z
relation
is
unknown.
Our
exist
Considère et al. 2000).
solar
[log
ðO=HÞ
þ
12
$
8:8].
Again, to
1
0.0 (e.g.,
Most of the data sets previously used
to compare
andraise
cali- the
Our
calibration
of
the
[N
ii]/[O
iii]
ratio
is
shown
in
degeneracy
in
the
solutions,
an
initial
guess
of
the metalcomparison between
differentdiagnostics
metallicity
0.0
bratethe
abundance
hadcalibrations
been selected shows
in different
Figure 8. The fourth-order polynomial fit coefficients
are the and
licity
must first
be obtained
from
angalaxy
alternative
that differing
ionization
parameter
doesdiagnostic.
notH ii
heterogeneous
ways:among
some
bygalaxies
(brightest
-0.5 in Table 3.
given
dependent
on
ionization
parameter
S232 1isorquite
regions,
brightest
disk
H
ii
regions),
some
by objective
cause
or
contribute
to
the
scatter.
The
ionization
parameter
is for all
Because
the
two
ions
have
quite
different
ionization
3
metallicities,
therefore
the ionization
1
prism
searchesand
(which
are biased
towardparameter
strong [Oderived
iii]
..
-2
explicitly
calculated
and
intoii]by
account
some
metallicity
potentials,
the [N ii]/[O iii] ratio depends strongly
on the
4thetaken
4
from
[S iii]/[S
diagnostic
should
toirregueliminate
""4959,
5007),
some
Galaxy in
type
(suchbe
asused
dwarf
-1.0
-0.5
ionization
parameter. Thus, if this diagnostic
is
to
be
use5
5 a free
diagnostics ( KD02;
M91),
but
see ainreduction
thisKK04;
as
variable.
lars).
These
different
datawe
setsdo
arenot
reflected
the differences
ful for6 abundance
it must be used in
q=3e8determinations,
(7)
6 the various
between
calibrations
of abundance.
Care must
1.5e8 (6)
in
scatter
for
these
methods.
A
full
investigation
into
the
scatter
combination
with8e7an
(5) independent ionization parameter
-1.5
bewill
taken
therefore
when
comparing
different
abundance
4e7 (4)
4.5.
The
½N
ii&=H#
Diagnostic
Diagram
7
in
the
M-Z
relation
be
presented
in
S.
L.
Ellison
et
al.
(2008,
in
diagnostic.
However,
2e7 (3) if the [O ii] or [S ii], and [S iii] lines
7
diagnostics
to take into account biases (if any) introduced
-3
1e7 (2)
Primary Nitrogen
Secondary
Nitrogen
-1.0
5e6 (1)
In
the
absence
of
other
emission
lines,
the
[N
ii]/H#
line
preparation).
by the comparison data.
-2.07.5
8.0
8.5
9.0
9.5
ratio
canthe
betoused
a M-Z
crude8.5
estimator
of9.0the
metallicity.
Note
LOG (O/H)
8.0
8.5 + 12
9.0 We
9.5
40 7.5
KEWLEY
&compare
DOPITA
Vol. 142
directly
best-fit
curves
for
nine
7.5 chose
8.0 as
9.5
We
use
observations
of
H
ii
regions
available
LOG (O/H) + 12
+ 12
that the
thelarge
[N ii]/H#
ratioLOG
is (O/H)
particularly
from
and2,
homogeneous
data set
van P05.
Zeeto
et shock
al.
strong-line
calibrations
in Figure
including
both
P01ofsensitive
and
4
1.5
Fig.
6.—The
log
(([S
ii]
!!6717,
6731 +[Satiii]high
!!9069,
9532)/H") (S
q=3e8 (7)
23 )
excitation
orauthors
the q=3e8
presence
of a185
hard
ionizing
radiation
field,
very difficult to observe.
Second,
metallicity,
the
(1998).
These
observed
H
ii
regions
in
13
spiral
(7)
1.5e8 (6)vs. metallicity. Curves for each ionization
Fig.
7.—The
log
([N
ii]
!6584/H#)
diagnostic
for
abundance
vs.
metaldiagnostic
for
abundance
1.5e8in
(6) metallicity
The
top
theanrms
scatter
the
mean
Fig. 5.—The
logtemperature
(([O8e7ii](5)!3727 gives
þ ½O iii] !!4959,
5007)/H")
(R23panel
) diagZ =about
0.5on
solar
from
AGN.
The
of between
strong
shock
excitation
lower
electron
thermal
electrons
of shows
galaxies
with
the
Double
Spectrograph
the
Palomar
5
m
8e7 (5)presence
)1
6 to 3 ( 108fewer
6 to 3 (
8 or
Z
=
0.5
solar
cm
s
are
shown.
Filled
circles
10
licity.
Curves
for
each
ionization
parameter
q
¼
5
(
10
parameter
between
q
¼
5
(
10
4e7metallicity.
(4)
nostic for abundance vs.
Curves for each ionization parameter
4e7 (4)
)
¼
0:2.
The
major
difference
in
mass
bins
of
width
(M
/M
an
AGN
will
increase
the
[N
ii]/H#
ratio
and
cause
the
high
energy,
leading
to
a
strong
decrease
in
the
number
of
)1!log
2e7from
(3)
!
telescope.
These
data
have
the
additional
advantage
of
cov)1
6
8
2e7
(3)
7
cm
s
are
shown.
Filled
circles
represent
the
data
points
from
our
models
represent
the
data
points
our
models
at
metallicities
from
left
to
right
3(
are shown. Filled circles represent
between q ¼ 5 ( 10 to1e7
(2) 10 cm s
1e7 (2)
collisional
excitations
theat3.0
blue
ii] the
lines
(which
has aofofthe
relabundances
determined
be
artificially
high.
IfThe
the
. [See
electronic
the M-Zat
.
metallicities
left5e6
to(1)metallicity
right to
of 0.05,
0.1, 0.2,
0.5,
1.0,
1.5,
2.0,
3.0[N
Z*ii]/
of 0.05,
0.2,
0.5, 1.0,
Z*[O
ering
range
in
and
ionization
parameter,
5e6 1.5,
(1)of 2.0,
1.0a large
between
curves
is from
their
position
along
the
y-axis.
the
data 0.1,
points
from
our
models
metallicities
from
left
toedition
right
0.05,
6
2 0.5,
Journal
for
a1.0,
color
version
of this
[See
theratio
electronic
edition
of theet
Journal
for aoptical
color version
of this figure.]
atively
high
threshold
energy
forelectronic
excitation)
relative
to
thefor
is to
be
used,
diagnostic
diagrams
[See the
edition
of the Journal
0.1,
0.2,
1.5,
2.0, 3.0
Z*.figure.]
asH#
was
shown
in
Dopita
al.photoionization
(2000).
curves
with
the largest
y-intercept
are standard
all
model
5
alower
color version
thisii]
figure.]
energyof[N
lines.
should
first
applied to rule
possibility
presSince [S
iii]bemeasurements
areout
notthe
available
for of
thethe
van
4
based
(KK04;
Z94;
KD02;
T04;
M91).
Among
these
photoFor Z < 0:5 Z! [log ðO=HÞ þ 12 < 8:6], the metallicity
Zee
H of
ii regions,
also used
two additional
data sets
ence
an AGNweorhave
shocked
excitation.
We recommend
the
3
ionization
model for
metallicities,
the agreement
is&optical
#0.2
dex.
This and
dependence of the [N ii]/[O ii] ratio is lost because
nitrogen
Stasińska
(1983)
the
use0.5
of Sthe
Kewley
et Dennefeld
al. (2001b)
diagnostics
2 since
23 diagnostic;
factor
’’ or ‘‘ excitation
parameter
’’ to correct
the observed
1 here
(like 0oxygen)
is predominantly
a primary
nucleosynthesis
Kennicutt
&based
Garnett
(1996).
These
also
cover
a wide
range
these
the same
theoretical
models
used
agreement
is within
theare
margin
ofonerror
typically
cited
for
these
1ionization parameter, such as in Pilyugin
(2000);
R
23 for in
inand
ionization
parameter
and
metallicity.
Using
the previous
ESO
element
this metallicity range. In addition, the nitrogenhave
been
shown
to
be
more
reliable
than
the
calibrations
(#0.1–0.15
dex for each
calibration).
Some
calibraCharlot
ion2& Longhetti (2001). We prefer to estimate
3.6
m telescope,method
Dennefeld
& Stasińska
(1983) observed
to-oxygen
abundance ratio shows large scatter fromthe
object
semiempirical
of Veilleux
& Osterbrock
(1987).
0.0
ization
parameter
explicitly,
based
upon
the
tions
consistently
agree
to
within
0.1
dex
(e.g.,
KK04
and
Z94;
#40Our
H iicalibration
regions in the
Galaxy,
3 In this regime, nitrogen production increases as a
to object.
of Magellanic
this ratio isClouds
shownand
in the
Figure
7, with
theoretical
Kennicutt
(1996)
observed
a similar numKD02
and M91).while
Comparisons
between
metallicities
calculated
function4 ofmodels.
time since the bulk of the star
formation
polynomial
fits&
asGarnett
in equation
(4)
and coefficients
given in
-2
ber
ofmethods,
H3.ii regions
in M101
using and
the 2.1
m telescope
at Kitt
5 (Edmunds & Pagel 1978; Matteucci & Tosi 1985;
occurred
Table
using these consistent
such
as KD02
M91,
are likely
-0.5
6 al. 1997).
Peak.
Dopita et
Therefore,
for a sample
of H ii regions,
Diagram
4.4. The
S23 Diagnostic
At
very
low
metallicity,
this
ratio
scales
simply
Kewley & Ellison 08
to be reliable to within 0.1 dex. However, comparisons between as the
7
the varying
age distribution of the stellar population from
nitrogen
abundance,
to first order.
However,
it is known
Our calibration of the ratio ([S ii] !!6717,
6731
+[S
iii]
5.2.
Comparison
Techniques
for
Deriving
Abundance
that showthat
large
disagreement
(such
as KK04 and
P05)
object to object will cause scatter in themethods
N/O ratios
in thisPrimary
metallicity
regime, the nitrogen
abundance
Nitrogen
Secondary Nitrogen
!!9069,
9532)/H"
(popularly known as S23
) is shown
in
Primary Nitrogen
Secondary
-4
observed.
Good evidence for the primary dependence
of
-1.0
will
beNitrogen
contaminated
by
the large
discrepancy
between
As
discussed
in xsystematic
1, a wide number
of empirical
and semi7.5
8.0
8.5
9.0
9.5
7.5
8.0
8.5
9.0
9.5
nitrogen at low abundances
4 of
empirical approaches already
LOGcan
(O/H) +be
12 found in
LOGexist
(O/H) + for
12 the determination of
theFigure
calibrations.
Considère
et al. (2000), in which log ðN=OÞ derived from a
abundances in H ii regions. We compare three of the most
1.0
The
lowest curves
in Figure
2 q=3e8
are(7)the produced
M-Z relations
derived
(7)
large
sample
iiq=3e8
regions
is iii]
compared
with the
commonly
used
calibrations
by McGaugh
(1991,
Fig.
8.—Theof
logH
([N
ii]
"6584/[O
"5007) diagnostic
for expected
abundance
1.5e8
(6)
Fig. 4.—The
log ([N
ii]
!6584/[S
ii] !!6717,Z6731)
diagnostic for
abunFig. 2.—Robust best-fit M-Z relations calculated using the different metallicity
1.5e8
(6)
6
= 0.5 solar
8e7
(5)
using
the
empirical
methods
(i.e.,
P01,
P05,
and
the
two
PP04
vs.
metallicity.
Curves
for
each
ionization
parameter
between
q
¼
5
%
10
relations for &1
a primary,
secondary, and primary + seconhereafter
M91), Zaritsky,
Kennicutt,
& Huchra
(1994,
here8e7 (5)
dance
vs.
metallicity.
Curves
for
each
ionization
parameter
between
4e7
(4)
8 cm s
calibrations
listed in Table 1, except the Te method. The top panel shows the rms
to
3
%
10
are
shown.
Filled
circles
represent
the
data
points
from
4e7
(4)
*1
8 cm s
0Z94),
(3)
and
Charlot
Longhetti
(2001,
C01)
dary origin for 2e7
nitrogen
production (Vila-Costas
&
3 + 10
are
shown. Filled
circles hereafter
represent the
data
qafter
¼5+
106 tomethods
methods). These
empirical
are&
calibrated
predominantly
2e7 (3)
1e7 (2)from left to right of 0.05, 0.1, 0.2, 0.5, 1.0, 1.5,
our models at metallicities
scatter in metallicity about the best-fit relation for each calibration in 0.1 dex bins of
points
from
our
models
at
metallicities
from
left
to
right
of
0.05,
0.1,
0.2,
1e7
(2)
5e6
(1)
with
the
results
produced
by
our
proposed
theoretical
Edmunds
[See the electronic edition of the Journal for a color
2.0, 3.0 Z'. 1993).
via version
fits ofof this
the relationship
between
strong-line
H ii for
re-a color stellar mass. The y-axis offset, shape, and scatter of the M-Z relation differ sub(1)the electronic ratios
[See
edition ofand
the Journal
0.5,
1.0, 1.5, 2.0,
3.0 Z!. 5e6
0.5
diagnostics.
We
conclude that [N ii]/[O ii] provides an excellent abunfigure.]
version
of
this
figure.]
stantially, depending on which metallicity calibration is used.
metallicities.
There
is
considerable
variation
among
the
gion
T
e
be
dance diagnostic for Z > 0:5 Z!, but this ratio cannot
used at lower abundances.
-1
For Z > 0:5 Z!, the curves in Figure 3 can be fitted by a
primarily
to the different nucleogenic status of the two elesimple
quadratic, facilitating abundance determination
ments. At low metallicity, both the elements are primary
0.0
using the [N ii]/[O ii] ratio, i.e.,
and the ratio becomes insensitive to metallicity. This diag1
nostic is2not
as useful as [N ii]/[O ii] for the determination of
log ðO=HÞ þ 12 ¼ log ½1:54020 þ 1:26602 R
Metallicity calibration with optical lines：
[SII]/[NII]
LOG ( [NII] / H" )
[NII]/Hα
N2
LOG ( [NII] / [SII] )
LOG
{ ({ [OII]+[OIII]
LOG
( [SII]+[SIII]) )/ /H!
H!} }
LOG ( [NII] / [OIII] )
[NII]/[OIII]
([OIII] + [OII])/Hβ
R23
O3N2
N2S2
LOG ( [NII] / H" )
G { ( [SII]+[SIII] ) / H! }
Empirical/theoretical metallicity calibration by using
strong optical lines (note that systematics exist)
系外銀河における微細構造輝線の観測とその理解 2014年12月2-3日 @国立天文台
Metallicity calibration with FIR FSLs：
A&A 526, A149 (2011)
Fig. 3. Predicted emission-line flux ratio of [O]51.80/[N]57.21 as a
function of gas metallicity. Blue and red lines denote the models with
U = 10−2.5 and 10−1.5 , and solid and dashed lines denote the models
with nH = 101.0 cm−3 and 103.0 cm−3 , respectively. The flux ratio observed in M 82 and in the Antennae galaxy is shown by black horizontal
line (these two galaxies show very similar [O]51.80/[N]57.21 flux
ratios; see Table 5). The x-range of this horizontal line corresponds to
the inferred metallicity range for M 82 and the Antenna galaxy.
Fig. 3. Predicted emission-line flux ratio of [O]51.80/[N]57.21 as a
function of gas metallicity. Blue and red lines denote the models with
U = 10−2.5 and 10−1.5 , and solid and dashed lines denote the models
with nH = 101.0 cm−3 and 103.0 cm−3 , respectively. The flux ratio observed in M 82 and in the Antennae galaxy is shown by black horizontal
line (these two galaxies show very similar [O]51.80/[N]57.21 flux
ratios; see Table 5). The x-range of this horizontal line corresponds to
the inferred metallicity range for M 82 and the Antenna galaxy.
Fig. 5. Same as Fig. 3 but for the flux ratio of ([O]51.80+
[O]88.33)/[N]57.21. Note the much lower dependence on the gas
density, which makes this ratio particularly suited to measure the gas
metallicity.
instead of [O]51.80/[O]88.33. Since the [N]205.4 emission is very faint and at a very long wavelength, it is very difficult to study with Herschel and SPICA, though its detection
Fig. 5. Same as Fig. 3 but for the flux ratio of ([O]51.80+
should be feasible with ALMA.
[O]88.33)/[N]57.21. Note the much lower dependence on the gas
Metallicity calibration by using
3.3. Dependences
on the stellar
age
FIR
[OIII]52µm,
[OIII]88µm,
All emission-line flux ratios shown in Figs. 3−7 are calculated
[NII]57µm
by adopting constant star-formation SEDs with an age of 1 Myr.
density, which makes this ratio particularly suited to measure the gas
metallicity.
instead
of it[O
]51.80/[O
]88.33.
Since
[N]205.4
However,
should
be verified
whether
thethe
predicted
flux emisratios
sion
is
very
faint
and
at
a
very
long
wavelength,
it
is
very
difdepend on the age of the stellar population, because the adopted
ficult
to
study
with
Herschel
and
SPICA,
though
its
detection
age (1 Myr) seems too young compared to the typical age of
should
be feasible
withinALMA.
star-forming
galaxies
general. Cid Fernandes et al. (2003) investigated stellar populations of nearby starburst galaxies and
reported
that the typical
starburst
age is ∼107.0−7.5 yr. More re3.3.
Dependences
on the
stellar age
cently, Rodríguez Zaurín et al. (2010) studied stellar populations
of low-z
ULIRGs,flux
finding
similar
ageinranges.
It isare
therefore
imAll
emission-line
ratios
shown
Figs. 3−7
calculated
portant
to examine
how
the emission-line
in
by
adopting
constant
star-formation
SEDs diagnostics
with an age studied
of 1 Myr.
this work it
depend
age ofwhether
the stellar
used
for
However,
shouldonbethe
verified
the populations
predicted flux
ratios
→ “SPICA” is necessary！How
about ALMA?
Nagao+11
Fig. 4. Same as Fig. 3 but for the flux ratio of [O]88.33/[N]57.21.
系外銀河における微細構造輝線の観測とその理解 2014年12月2-3日 @国立天文台
Metallicity of galaxies at high redshift：
1924
F. Mannucci et al.
metallicity
Mannucci+09, MNRAS, 398, 1915
stellar mass
Figure 5. Evolution of the mass–metallicity relation from z = 0.07 (Kewley & Ellison 2008) to z = 0.7 (Savaglio et al. 2005), z = 2.2 (Erb et al. 2006a)
and z = 3–4 (AMAZE+LSD). All data have been calibrated to the same metallicity scale and IMF (Chabrier 2003) in order to make all the different results
directly comparable. Turquoise empty dots show the AMAZE galaxies, blue solid dots the LSD galaxies. The solid square shows the ‘average’ LSD galaxy,
having average mass and composite spectrum (see Fig. 4). The lines show quadratic fits to the data, as described in the text.
Metallicity measurements are limited up to z~3
系外銀河における微細構造輝線の観測とその理解 2014年12月2-3日 @国立天文台
Metallicity of galaxies at high redshift：
1924
F. Mannucci et al.
metallicity
Mannucci+09, MNRAS, 398, 1915
z>4
stellar mass
Figure 5. Evolution of the mass–metallicity relation from z = 0.07 (Kewley & Ellison 2008) to z = 0.7 (Savaglio et al. 2005), z = 2.2 (Erb et al. 2006a)
and z = 3–4 (AMAZE+LSD). All data have been calibrated to the same metallicity scale and IMF (Chabrier 2003) in order to make all the different results
directly comparable. Turquoise empty dots show the AMAZE galaxies, blue solid dots the LSD galaxies. The solid square shows the ‘average’ LSD galaxy,
having average mass and composite spectrum (see Fig. 4). The lines show quadratic fits to the data, as described in the text.
Metallicity measurements are limited up to z~3
系外銀河における微細構造輝線の観測とその理解 2014年12月2-3日 @国立天文台
Metallicity of galaxies at high redshift：
Strong emission lines
used to calibrate the
metallicity come
through the observable
wavelength (>2.5µm)
系外銀河における微細構造輝線の観測とその理解 2014年12月2-3日 @国立天文台
Metallicity of galaxies at high redshift：
1924
F. Mannucci et al.
Mannucci+09, MNRAS, 398, 1915
metallicity
JWST?
SPICA?
or
z>4
stellar mass
ALMA?
Figure 5. Evolution of the mass–metallicity relation from z = 0.07 (Kewley & Ellison 2008) to z = 0.7 (Savaglio et al. 2005), z = 2.2 (Erb et al. 2006a)
and z = 3–4 (AMAZE+LSD). All data have been calibrated to the same metallicity scale and IMF (Chabrier 2003) in order to make all the different results
directly comparable. Turquoise empty dots show the AMAZE galaxies, blue solid dots the LSD galaxies. The solid square shows the ‘average’ LSD galaxy,
having average mass and composite spectrum (see Fig. 4). The lines show quadratic fits to the data, as described in the text.
Metallicity measurements are limited up to z~3
系外銀河における微細構造輝線の観測とその理解 2014年12月2-3日 @国立天文台
Chemical abundance ratio at high redshift:
J. Köppen and G. Hensler: Chemistry of epi
lg(N/O)
0
-0.5
-1
100
-1.5
-2
1
Mcloud /Mgal
6.5
7
Koppen & Hensler 05
30
7.5
10
3
8
8.5
9
lg(O/H)
9.5
Fig. 2. Tracks in the N/O–O/H-diagram of infall models with various
Effect of gas inflow on chemical evolution
mass ratios of the infalling gas and the galaxy. The infall event starts
Fig. 3.
masses
系外銀河における微細構造輝線の観測とその理解 2014年12月2-3日 @国立天文台
multaFigure 4. SED of A383-5.2 and population synthesis models which provide
3
e CIII]
best fit to the continuum SED and CIII] equivalent width. The observed
SED
is denoted by the diamond data points. The two grey data points at
ar for8
Stark
et al.
<1µm are not included
in the fit because of the uncertainty
with (arXiv:1408.3649)
CIII]
nsider
Starkassociated
et al. 2014
Ly↵ emission contamination and IGM absorption. The data are best fit by a
rburst’
3000
model with a two component star formation history. The UV continuum and
n with
z=6−7, This work
20
CIII] equivalent width
are
powered
by
a
recent
star
formation
episode
(cyan
z=1.5−3.0, Stark
et al. 2014
A383−5.2
stellar2500
z=2.3,
Erb
et
al.
2010
curve), while the optical
continuum
is dominated by an older generation
of
ΔvLyα
= 120 km/s
z~3, Shapley et al 2003
th age
(composite
spec.)
stars (red curve). The composite SED is shown in black. Yellow
diamonds
2000
shift).
show predicted broadband fluxes from the best-fitting model.
15
match-1500
1909Å
t comst leftmost
pro-1000panel is centred on Ly↵ emission which is detected with the visible
he
10
iddle
(unsmoothed)
and
rightmost
(smoothed)
panels.
The
[CIII]
1907
emission
he
star
Model fit to A383.5.2
500
mation
log U
1.70+0.49
fitting. 0
0.64
5
+0.04
log (M⇤,young /M⇤,tot )
2.99 0.03
ing of−500
ware (v.2.2.0) in the Recipe Flexible Execution Workbench
+0.27 (RE−200 log (Z/Z
0
200
400
600
800
1000
)
1.33
the al0.20
FLEX) environment to perform
a first
reduction
of
Vel (km
s−1) calibration and
+0.06
log(C/O)
0.58
mass
0.06routines
each exposure. We then applied standard IDL and IRAF
WCIII],0 (Å)
Fλ (10−20 erg cm−2 s−1 Å−1)
Chemical abundance ratio at high redshift:
+0.10
0
Figure
5. Velocity
profile of Ly↵
emission
in the zthe
= 6.027
galaxy A383-Speciflog(age/yr)
for
optimally
combining
and
extracting
15 8.72
exposures.
0.10
5.2. Ly↵ is shifted to the rest-frame using the systemic redshift provided
by
+0.10
−50
0
50
100
150
ically,
we used
the Lyman-↵
emission
line,bywell-detected
in each
log(M
/M
) is shifted
9.50
⇤,tot
n from
CIII] 1909. The
peak flux
of Ly↵
emission
v=120 km
s 1
0.10
WLy ,0 (Å)
+0.08 condiexposure,
to correct
for variations
in1 )seeing and 0.29
atmospheric
from the systemic
redshift.
log(SFR/M
yr
g wide
0.08
tions between the different OBs, and applied a scaling
and weight+0.05
Note that they are gravitational
⌧
ˆ
0.05
model
0.05level ofFigure 6. A comparison of the rest-frame equivalent widths of CIII] 1909
ing of the 2D V
spectra according to the flux and detection
Å and Ly↵ fromlensed
the study of objects
z ' 2 lensed sources by Stark et al (2014,
the
line.
We
also
used
the
spatial
position
of
the
line
edshift
6 Lyman-alpha
DISCUSSION
AND
SUMMARY
orange
points),
Erb
et
al
(2010)
Table 1. Results of fitting procedure for A383-5.2. Details are providedand
in the z' 3 Shapley et al (2003) composite
measured in the reduced spectrum to precisely compute
the off-alongside the two new CIII] detections beyond z ' 6 (this paper, red stars).
nt; 30
Considerable
effortand
has been
placed
spectroscopic
6
§4.1,
results
areindiscussed
in study
§4.2.of z >
⇠
sets
between
each
OB
for optimal
combination.
We used
galaxies.
Yet, in
spite
of significant
allocations
of telescope
timethe
to same
ompooffsets,
scaling factors
andspectroscopic
weights to combine
the exposures
several research
teams, few
redshifts have
been con- in the
icities
near-infrared
corrections
slightly
strengthened
firmed beyondarm.
z 'Applying
7 due to these
the absence
of Ly↵.
Although
the
Detection of CIII]λ1909 from galaxies at z~6
系外銀河における微細構造輝線の観測とその理解 2014年12月2-3日 @国立天文台
Newman et al.
Morphology and Kinematics at high redshift：
nal, 752:111 (19pp), 2012 June 20
Forster Schreiber+09
Clumpy structure in high
redshift galaxies
Strong Galactic winds in
individual clumps of high
redshift galaxies
Dust geometry in high
redshift
galaxies
Newman et al. 2012, ApJ, 752, 111
Figure 2. Emission line profiles and best fit to Hα and [N ii]λ6584 features for six regions of ZC406690: clump A
MIPS
J1428 [O I] 63
系外銀河における微細構造輝線の観測とその理解 2014年12月2-3日
@国立天文台
MIPS J1428
F10214
F10214
F10214
[O III] 88
[O IV] 26
[S III] 33
[O III] 52
205
205
179
185
179
Connection between galaxies and AGN：
BPT
★☆: Stacking analysis
1 342 187 779
1 342 187 779
1 342 186 812
1 342 187 021
1 342 186 812
1
7
10
6
s
Notes. (a) Number of line/range repetitions per nod cycle × number
calibration uncertainty of 30% applies; (c) Correction factor, applied
central spatial pixel (see text); (d) The signal in this PACS wavelengt
of upper limits and RMS unreliable; (e) assuming a Gaussian profile
Sturm+10, A&A, 518, L36
[O IV] (25.91µm) / FIR
10−2
Kewley+01
Yabe et al. 2014
AGN diagnostics by
using FIR fine structure
lines ?
NGC 1068
F10214
10−3
Mrk 231
10−4
Arp 220
10−5
M82
108
109
1010
1011
1012
1013
LFIR [ L ]
Fig. 2. The [O IV]/FIR limit in F10214 compared to template obje
系外銀河における微細構造輝線の観測とその理解 2014年12月2-3日 @国立天文台
What I want to know：
• Star formation rate of galaxies
• Metallicity of galaxies
• Distribution of star-formation
• ISM condition / AGN connection
... from dust-free observations
spatially resolved (if possible) ...