Download The Evolution of the Mass-Size relation to z = 3.5 in the GOODS

The Evolution of the Mass-Size relation to z = 3.5
in the GOODS-North Field
Draft version November 12, 2010
Moein Mosleh
Preprint typeset using LATEX style emulateapj v. 5/25/10
THE EVOLUTION OF THE MASS-SIZE RELATION TO Z=3.5 FOR UV-BRIGHT GALAXIES AND SUB-MM
GALAXIES IN THE GOODS-NORTH FIELD
Moein Mosleh1 , Rik J. Williams 2 , Marijn Franx 1 , Mariska Kriek3
Draft version November 12, 2010
Abstract
We study the evolution of the size - stellar mass relation for a large spectroscopic sample of galaxies
in the GOODs North field up to z ∼ 3.5. The sizes of the galaxies are measured from Ks -band
images (corresponding to rest-frame optical/NIR) from the Subaru 8m telescope. We reproduce earlier
results based on photometric redshifts that the sizes of galaxies at a given mass evolve with redshift.
Specifically, we compare sizes of UV-bright galaxies at a range of redshifts: Lyman break galaxies
(LBGs) selected through the U-drop technique (z ∼ 2.5 − 3.5), BM/BX galaxies at z ∼ 1.5 − 2.5, and
GALEX LBGs at low redshift (z ∼ 0.6 − 1.5). The median sizes of these UV-bright galaxies evolve as
Introduction
! Galaxies at high redshift have significantly smaller
sizes than the local galaxies of similar masses
! Quiescent Galaxies are significantly smaller than
star-forming galaxies of similar masses
! These studies are almost bases on Photometric
redshift especially at high z
! Size measurements of Sub-mm or LBG galaxies at
high z are also rare
Williams et al. 2010
m
Λ
0
2. DESCRIPTION OF DATA
The sample of galaxies used here is based on the most
Using most complete spectroscopic catalog of galaxies in
mplete spectroscopic
catalog
of
galaxies
in
the
GOODS
the GOODS-North Field. (Barger et al. 2008)
eat Observatories Origin Deep Survey (Giavalisco
Using U bandfield
(Capak et al.
2004)Barger et al. (2008). This catl. 2004)) "North
by
g gives a compilation
ofF606W
all(V), spectroscopic
observations
" ACS Magnitudes (F435w (B),
F775W (I), F850LP (Z) )
ried out in
this field (e.g. Cowie et al. 2004; Reddy
" Ks-band magnitudes from WIRCam (CFHT, 3.6 m) (6909)
al. 2006; Wirth et al. 2004; Cohen 2001; Cohen et al.
" NUV – FUV from GALEX (1016)
0), where each galaxy sample was selected
in a differRedshift Distribution
way. In addition, Barger et al. (2008) performed specscopic observations for certain subsamples. The cataincludes 2907 sources including stars and is restricted
ources with Ks,AB < 24.5 or F 850LPAB < 26. There
galaxies this large and faint have not been seen at any
redshift.
3. SIZES
The random uncertainties increase with size in all mag20
mag (top panel)
nitude bins.✓ ForUsing
objects
with
KAB
GALFIT (Peng
et al. 2002)
to fit ≤
Sersic
models
the random✓ uncertainties
in 8Msize
recovery are < 5%.
Ks-band image from Subaru
Telescope
FWHM ~ 0.5”with 21 < KAB < 22 the increasHowever, for
✓ galaxies
Pixel scale = 0.12 arcsec/pixel
✓ uncertainties
ing of random
with sizes is significant. The
increase in random uncertainties at larger sizes could be
due to decreasing of surface brightness. However, since
we are mainly concerned about the overall sample properties, the lack of systematic errors is more important.
4. STELLAR MASS ESTIMATES
The stellar masses of galaxies were measured with the
FAST (Fitting and Assessment of Synthetic Templates code) (Kriek et al.
Fitting and✓Using
Assessment
of Synthetic Templates(FAST)
2009)
code (Kriek✓Using
et Salpeter
al. 2009b).
All fluxes from Barger et al.
IMF & Solar Metallicity
(2008) including four optical bands from ACS, the Ks
✓Masses are corrected by a factor of -0.2 dex (for consistency with a Kroupa
(2001)
IMF)
infrared and U
band
were used to find the best-fit galaxy
template SED
toB, V the
, I, Z, Ks broadband photometry using a
✓Bands: U,
χ2 minimization procedure. Bruzual & Charlot (2003)
Simulations
Running simulation for testing the accuracy of parameter
measurements
Evolution of the mass size relation
to z = 3.5 in the GOODS North field
3
Fig. 1.— Left panel : The points are the median of the relative difference between the recovered and input sizes versus magnitude based
on our simulations. Right panel : The same comparison but for the relative difference between output and input Sérsic index. The error
bars illustrate the 68% scatter. The random uncertainties of recovered sizes and Sérsic index increase with magnitude.
4
Mosleh et al.
TABLE 1
Sub-samples
Sample
GALEX/LBG
BM/BX
LBG
SMG
sBzK
Redshift
0.6 < z < 1.4
1.4 < z < 2.7
2.7 < z < 3.5
0.5 < z < 3.
1.4 < z < 3.2
log(M∗ /M" )
8.8 - 11.0
9.8 - 10.8
10.2 - 10.8
10.0 - 11.7
9.8 - 11.7
Simulations
No. of Sources
105
41
5
14
70
Samples of star-forming galaxies studied in this paper.
lected from Steidel et al. (2003, 2004). For this paper,
41 BM/BX galaxies to KAB = 23 are selected from their
sample with spectroscopic redshift z > 1.4.
Galaxies analogous to the LBGs at 0.6 ≤ z ≤
1.4 (hereafter GALEX/LBGs) are selected with the
GALEX/HST (F U V − N U V )AB versus (N U V −
F 435W )AB color-color diagram. Following the selection
criteria used in Barger et al. (2008) and requiring z > 0.6,
we include 105 GALEX/LBGs with KAB ≤ 23.
In addition to the UV-bright galaxies in GOODS-N
field, we also study sizes of the Sub-millimetre galaxies (SMGs) as a population of star forming galaxies at
high redshifts. The subsample of SMGs for our studies
is drawn from Chapman et al. (2005) and Pope et al.
(2006). The catalog of SMGs provided by Pope et al.
(2006) contains 35 candidates from GOODS-N field with
21 secure optical counterparts. Of these, only 8 galaxies
in our observed images have spectroscopic redshifts. We
further include 6 SMGs from the Chapman et al. (2005)
HDF-North study. Therefore, our final SMG sample
studied here contains 14 sources spanning 0.5 ≤ z ≤ 3.
Besides these two populations, star forming galaxies
at z > 1.4 can also be identified using BzK technique
(Daddi et al. 2004). This technique is designed to cull
galaxies from K-selected samples. Star forming BzK
galaxies (sBzK ) tend to be more massive and have higher
reddening than UV-selected galaxies. With this method,
70 sBzK galaxies at z > 1.4 and KAB ≤ 23 are selected
5. SUB-SAMPLES
# UV- Bright Galaxies :
e ability to identify and study distant galaxies h
๏ Lyman Break Galaxies (LBGs) ( 2.7 < z < 3.4 )
ved dramatically during the last two decades. Va
election criteria
< z < 2.7) been designed to select hig
๏ BM/BX (1.4have
Selected from Reddy et al. 2006
Selected from Reddy et al. 2006 Catalog
๏
GALEX/LBGs (0.6 < z < 1.4)
Selected by using (FUV-NUV) versus (NUV-F435W ) color- color diagram
#
Sub-mm Galaxies (SMG)
Selected from samples of Chapman et al. 2005 & Pope et al. 2006
#
Mosleh BzK
et al. (sBzK) (1.4 < z < 3.2)
Star-forming
TABLE 1
Sub-samples
Sample
GALEX/LBG
BM/BX
LBG
SMG
sBzK
Redshift
0.6 < z < 1.4
1.4 < z < 2.7
2.7 < z < 3.5
0.5 < z < 3.
1.4 < z < 3.2
log(M∗ /M" )
8.8 - 11.0
9.8 - 10.8
10.2 - 10.8
10.0 - 11.7
9.8 - 11.7
No. of Sources
105
41
5
14
70
Samples of star-forming galaxies studied in this paper.
6. RESULTS
6.1. Size Evolution
Evolution of the mass size relation to z = 3.5 in the GOODS North field
Fig. 8.— Left panel : The correlation between color (V606 − z850 ) and surface density for galaxies in the redshift range of 0.5 < z < 1.5
and stellar masses 1010 < M/M! < 1011 in GOODS-N field. The blue stars are GALEX (LBGs) from GOODS-N and the brown circles
are from CDFS. The gray symbols are all galaxies from CDFS and the black triangles are galaxies from GOODS-N. The plots indicates
that blue galaxies having lower surface density. Right panel : The color (z850 − K) versus surface density for galaxies at 1.5 < z < 2.5, as
the green squares are the BM/BX galaxies from GOODS-N and the brown circles are pseudo BM/BX from CDF-S.
like, one would expect galaxies
with
AG
Figure 3 we show the size distributions
for
UV
have smaller half-light radii. Their “norm
optical sizes are thus perhaps not strong
galaxies between 10 < log(M
< 11.
Th
∗ /Mdue
" )to possible
Nonetheless,
effects on
the
estimates, X-ray detected galaxies should
istogram shows the measured
half-light
radii o
low-confidence
points.
6.2. Stellar Mass-Size
X/LBGs at z ∼ 1 while the distribution
of Relatio
size
We now investigate how mass-size distr
galaxies
compares
to other
g
red for BM/BX galaxies (z UV-bright
∼
2)
are
shown
as
tions. We further check if the relation b
mass and
size ofplot
these galaxies
existsUV
to h
histogram. It can be seen from
this
that
The stellar mass-size distributions of ou
in Figure 5. We have
divided
our sam
galaxies at z ∼ 2 are smallershown
compared
to
simila
redshift bins, 0.5 < z < 1.5, 1.5 < z < 2.5,
UV-bright galaxies are color coded the sam
es at z ∼ 1. Specifically, the
median half ligh
Solid and dotted lines in each panel show
relation from Shen et al. (2003) for star fo
f BM/BX galaxies in this mass
rangegalaxies
is 2.68
0.1
and early-type
at z ∼ 0,±
respectiv
symbols are the K -selected sample of gal
Sizes
of
UV-Bright
galaxies
evolve
by
a
median
factor
ofeffective
0.60±0.08
gnificantly smaller than the median
(Fran
South with
stellar
masses > 10 M rad
between z ~2 to z~1
Comparing our results with the galaxies fr
LEX/LBGs (4.42 ± 0.52 kpc).
lows us toThis
see where means
our galaxies lie tha
relati
mass-selected sample. Brown solid lines
TABLE 3
Median sizes of different samples
Sample
GALEX/LBG
BM/BX
LBG
CDFS
CDFS
CDFS
Quiescent(CDFS)
Quiescent(CDFS)
Quiescent(VD08)
SMG
Redshift
0.6 < z < 1.4
1.4 < z < 2.7
2.7 < z < 3.5
0.5 < z < 1.5
1.5 < z < 2.5
2.5 < z < 3.5
0.5 < z < 1.5
1.5 < z < 2
2 < z < 2.5
0.5 < z < 3.
Size
4.42 ± 0.52
2.68 ± 0.19
2.22 ± 0.61
2.33 ± 0.1
2.40 ± 0.13
1.73 ± 0.18
1.32 ± 0.07
1.12 ± 0.32
0.9 ± 0.21
2.90 ± 0.45
median(log(M∗ /M! )
10.2 ± 0.3
10.4 ± 0.2
10.6 ± 0.2
10.4 ± 0.3
10.4 ± 0.3
10.4 ± 0.3
10.5 ± 0.3
10.7 ± 0.2
11.23
10.8 ± 0.5
Notes. The median sizes of samples are for galaxies with stellar
masses between 1010 − 1011 M! except for the quiescent(VD08)
(van Dokkum et al. 2008b) and SMG galaxies. The CDFS sample
is all galaxies at the same stellar mass range in CDF-S field.
for galaxies in our sample. We have confirmed that there
is a relation between stellar mass and size for UV-bright
galaxies at 0.5 < z < 3.5. The relations are consistent
for both spectroscopic and photometric redshift samples.
Our results verify that there is no significant evolution
of the slope of the size-mass relation to redshift ∼ 3.5.
Fig. 3.—
distribution
of UV-bright
galaxies
with
Fig.Size
9.— Top
panel: Histograms
show the distribution
of sizes
of stellar
The evolution of the relation from z = 1 to z = 0 is
galaxies
(solid
andat
BM/BX
(dotted
lines).
mass 10 Submm
< log(M
< 11
z ∼ galaxies
1 (blue
histogram)
and
∗ /M
" ) line)
not well established, as it is not straight forward to9.8
seThe
gray
region
indicates
apparent
effective
radii
below
0.1”(∼
1
"
∼ 2 (green
histogram).
The
half
light
radii
of
UV-bright
galaxlect Lyman Break galaxies at z ∼ 0. Some authors find
kpc at z ∼ 2) which size measurements have large uncertainties.
es (GALEX(LBGs))
at z ∼ 1distribution
are larger
than for
BM/BX
galaxiesevolution
at
Lower panel: Cumulative
function
SMGs (pink
(e.g. Franx et al. 2008; Williams et al. 2010)
∼ 2. dimonds) and BM/BX galaxies (green squares). As the plot shows,
in the stellar mass-size relation, however, Barden et al.
the distributions of half-light radii of these two populations are
(2005) have reported that the stellar mass-size relation
comparable with SMGs being slightly larger.
for disk galaxies remains constant from z ∼ 1 to the
6
Mosleh et al.
6.1. Size Evolution
Figure 3 we show the size distributions for UV
galaxies between 10 < log(M∗ /M" ) < 11. Th
istogram shows the measured half-light radii o
X/LBGs at z ∼ 1 while the distribution of size
S
red for
BM/BX
galaxies
(z
∼
2)
are
shown
as
i
histogram.
It can be seen from this plot that UV
z
galaxies
at z ∼ 2 are smaller compared to simila
e
es at z ∼ 1. Specifically, the median half ligh
f BM/BX galaxies in this mass range is 2.68 ± 0.1
gnificantly smaller than the median effective rad
LEX/LBGs (4.42 ± 0.52
kpc).
This
means
tha
Redshift
f UV-bright galaxies evolve by a median factor o
Fig. 4.— Evolution of sizes of galaxies as a function of redshift. Each panel shows the size evolution for a narrow stellar mass bin. The
gray symbols are star forming galaxies with spectroscopic redshifts in GOODS-N field. Blue stars are GALEX/LBGs (z ∼ 1) while the
nfidence points.
6.2. Stellar Mass-Size Relation
Evolution of the mass size relation to z = 3.5 in the GOODS North field
now investigate
how mass-size distribution of the7
z~1
z~2
z~3
ight galaxies compares to other galaxy populaWe
S further check if the relation between stellar
andi size of these galaxies exists to high redshift.
stellar
mass-size distributions of our samples are
z
ineFigure 5. We have divided our sample into three
ft bins, 0.5 < z < 1.5, 1.5 < z < 2.5, 2.5 < z < 3.5.
ight galaxies are color coded the same as Figure 4.
and dotted lines in each
panel
Stellar
Mass show the mass-size
n from Shen et al. (2003) for star forming galaxies
rly-type galaxies at z ∼ 0, respectively. The gray
Fig. 5.— Stellar mass size distribution for UV-bright galaxies in different redshift bins compared to the galaxies from CDF-south (Franx
et al. 2008) (gray dots). The color symbols represented here are the same as Fig. 4. The solid and dotted black lines are the size-mass
relations for star forming and quiescent galaxies respectively at z ∼ 0 from Shen et al. (2003). The solid brown lines show the median sizes
in narrow mass bins for all galaxies from CDF-S. The UV bright galaxies are in general larger than normal field galaxies at the same mass.
uiescent galaxies respectively at z ∼ 0 from Shen et al. (2003). The solid brown lines show the median sizes
xiesFig.
from5.—
CDF-S.
Themass
UV bright
galaxies arefor
in UV-bright
general larger
than normal
field redshift
galaxies bins
at thecompared
same mass.
Stellar
size distribution
galaxies
in different
to the galaxies from CDF-south (Franx
et al. 2008) (gray dots). The color symbols represented here are the same as Fig. 4. The solid and dotted black lines are the size-mass
relations for star forming and quiescent galaxies respectively at z ∼ 0 from Shen et al. (2003). The solid brown lines show the median sizes
in narrow mass bins for all galaxies from CDF-S. The UV bright galaxies are in general larger than normal field galaxies at the same mass.
6.2. Stellar Mass-Size Relation
We now investigate
how mass-size
distribution
z~1
z~2
z ~ of
3 the
V-bright galaxies compares to other galaxy populaons.
S We further check if the relation between stellar
assi and size of these galaxies exists to high redshift.
The stellar mass-size distributions of our samples are
z
hown in Figure 5. We have divided our sample into three
e
dshift bins, 0.5 < z < 1.5, 1.5 < z < 2.5, 2.5 < z < 3.5.
V-bright galaxies are color coded the same as Figure 4.
olid and dotted lines in each panel show the mass-size
Stellar for
Mass
lation
from
Shen
et
al.
(2003)
star
forming
galaxies
axies (r ∝ M ). In the highe use UV-bright
galaxies fromgalaxies at z ∼ 0, respectively. The gray
nd
early-type
(α = 0.32 ± 0.06). Table 2 lists
meter to the mass size relation
ymbols
are
sampleα of galaxies in CDFSample
Redshift
highofmass
late typethe
galaxiesK
(r -selected
∝M
). In the highe for
slopes
the mass-size
relaGALEX/LBG
< z <9.8
1.4 0.19 ± 0.05
bin and
(z ∼there
3), we
galaxies0.6
from
insest
areredshift
consistent
is use UV-bright
(Franx et al. 2008).
outh
with
stellar
masses
>21.4lists
10
< z < 2.7 M
0.30 "
± 0.06
find thethe
best
fit (α = 0.32
±BM/BX
0.06). Table
OurCDF-S
resultsto
confirm
persisCDFs-UV
Bright
2.5 < z < 3.5 0.32 ± 0.06
thefor
best-fit
power law
parameter to
the mass
size relation
tion
star forming
galaxies
Sample
Redshift CDF-S
α
omparing
our
with1.4relathe
from
alsBzK
<
z < 3.2galaxies
0.31
± 0.09
of UV-bright galaxies.
The results
slopes of the
mass-size
GALEX/LBG
0.6 < z < 1.4 0.19 ± 0.05
tion or
at not
different
bins are consistent and there is
ether
other redshift
star formBM/BX
1.4 < z <to
2.7 0.30
0.06
dshift
have
same
size-mass
no significant
evolution.
Our results confirm
persisws
ustheto
see
where
ourthegalaxies
lie
relative
a ±purely
stellar mass-size relation for the UV-bright galaxies from GOODS-N field and CDF-South in three redshift
the size-mass relation for star forming galaxies at z ∼ 0 from Shen et al. (2003). The color symbols are
n narrow mass bins and the error bars show one σ (68%) dispersion. The dashed-dotted lines are the best
to the individual UV-bright galaxies. The dotted lines are the median sizes in narrow mass bins for all
Fig 5). This plot shows that the derived half-light radii of UV-bright galaxies from both fields are in place
Fig. 6.— Comparison of the stellar mass-size relation for the UV-bright galaxies from GOODS-N field and CDF-South in three redshift
up to z ∼ 3.
bins. The black solid lines are the size-mass relation for star forming galaxies at z ∼ 0 from Shen et al. (2003). The color symbols are
0.4
median of UV-bright
galaxies in narrow mass bins and the error bars show one σ (68%) dispersion. The dashed-dotted lines are the best
TABLE
2 dotted lines are the median sizes in narrow mass bins for all
powere law fits using re ∝ M α to the individual UV-bright galaxies.
The
Best
fits
power
law
parameter
for
the stellar
- size galaxies from both fields are in place
galaxies from CDF-S (Same as Fig 5). This plot shows that the derived half-light
radii of mass
UV-bright
relation
and agreement with each other up to z ∼ 3.
e
0.4
TABLE 2
Best fits power law parameter for the stellar mass - size
relation
Note. Power law parameter α is defined as re ∝ M α .
8 mass size relation for star forming Mosleh
al.
. 7.— Stellar
sBzKetgalaxies
circles). The BM/BX galaxies and SMGs are marked
densitywith
is more ti
sample of ga
and pink circles respectively. The red squares are theusing
median
firms that the colo
he mass-size relation of sBzK sample with one σ dispersion.
surface density an
green dashed-dotted line is the best fit to the BM/BX
galaxshifts.
The sBzK galaxies have a similar size-mass relation to the 6.4.
BX galaxies.
The solid line shows the mass-size relation for
S
Since GOODS-N
orming galaxies at z ∼ 0 from Shen et al. (2003). compare their ma
i
at z ∼ 2. As Figu
sizes of the SMG
z
could be due to the significant overlap between
galaxiesthe
of similar
lightet
radiial.
of all SM
selectede galaxies and sBzK galaxies (e.g. Reddy
agreement with th
BM/BX galaxies
kpc). The media
masses of 1010 −
the BM/BX of at
errors are comput
We further com
galaxies and BM
top panel, the his
5). We note that stellar masses can have significant
ertainties depending on the methods used to calculate
m. We verify that by using the stellar mass estimates
m Reddy et al. (2006) for UV-bright galaxies, the sizeStellar Mass
s distribution
these
same
region
Fig. 7.—of
Stellar
mass sizegalaxies
relation for starcovers
forming sBzKthe
galaxies
(gray circles). The BM/BX galaxies and SMGs are marked with
Sinceoptical
GOODS-N
includes
14UV-bright
SMGs with
the rest-frame
sizes of the
SMGs and
ethods
usedCDFS.
to calculate
are from
The gray symbols are all galaxies from CDFS and the black triangles are galaxies from G
Sizes
Submm
Since
GOODS-N
includes
14their
SMGs
redshifts,
weofUV-sel
that
blue
galaxies
having lowergalaxies.
surface
density.
Right
panel
The with
color
(zGalaxies
−6.4.
K) versus
surface
density
fo
compare
mass
and
sizes
to
the
e stellar
mass
estimates
850
6.4. Sizes
of : Submm
the galaxies,
green squares
are
the BM/BX galaxies from GOODS-N and the brown circles are pseudo BM/BX from
bright
the
sizecompare
their
mass
and
sizes
to
the
UV-selected
galaxies
6.4.
Sizes
of
Submm
Galaxies
at
z
∼
2.
As
Figure
4
illustrates,
the opti
7.
DISCUSSION
Since
GOODS-N
includes
14
Since
GOODS-N
includes
14
SMGs
with
redshifts,
we S
s covers the same region
TABLE
at zpanel
∼ 2.of Figure
Ascompare
Figuretheir
4sizes
illustrates,
the
optical
rest-frame
of and
the
SMGs
(pink
diamonds)
and
compare
their
mass
and
sizes
to t
7.1.
The
growth
of
star-forming
galaxies
middle
mass
sizes
to
the
UV-selected
galaxies
Median sizes
of diff
Since
GOODS-N
includes
14
SMGs
with
redshifts,
we
sizes
of the
SMGs
(pink
diamonds)
UV-selected
sults
are robust
against
galaxies
of
similar
mass
are
comparable.
Th
at z ∼and
2. galaxies
As
Figure
4rest-frame
illustrate
We
study
the
size
evolution
of
with spectroat
z
∼
2.
As
Figure
4
illustrates,
the
optical
compare
mass
and
sizes
to
the
UV-selected
Sample galaxies
Redshift
r masses.
scopic
redshifts
between
z
∼
0.5
−
3.5.
Our
results
are
light
radii
of
all
SMGs
is
2.90
±
0.45
wh
galaxies
oftheir
similar
mass
are
comparable.
The
median
half
sizes
of
the
SMGs
(pink
sizes of the SMGs (pink diamonds)
and UV-selected
GALEX/LBG
0.6 kpc
< z diam
< 1.4
at z radii
∼ 2.ofAs
Figure
4 illustrates,
rest-frame
summarized
in Figuregalaxies
10.the
As optical
this
plot
illustrates,
at
BM/BX
1.4
< zcomp
< 2.7
of
similar
mass
are
agreement
with
the
median
optical
rest-fram
light
all
SMGs
is
2.90
±
0.45
kpc
which
is
in
good
galaxies
of
similar
mass
are
comparable.
The
median
half
e Density
each epoch,
the diamonds)
median sizes of UV-bright
galaxies (color
LBG
2.7 < z < 3.5
sizes
of
the
SMGs
(pink
and
UV-selected
light
radii
of
all
SMGs
is
2.90
light
radii
of
all
SMGs
is
2.90
±
0.45
kpc
which
is
BM/BX
galaxies
over
thesizes
mass
with thesquares)
median
rest-frame
/M"of
) <the
110.5in
are
withoptical
stellar
masses
10
< log(M
CDFS
< good
z rang
<±
1.5 0
of agreement
Franx et al. (2008)
∗whole
galaxies
of
similar
mass
comparable.
The
median
half
CDFS
1.5
<of
zSMG
<
2.5
larger
(byare
athe
median
factor
ofrange
0.45effective
±
0.09)
than
quiescent
agreement
with
the
median
optic
agreement
with
median
optical
rest-frame
sizes
the
en BM/BX
stellar
massgalaxies
surface
kpc).
The
median
radii
of
over
the
whole
mass
(2.68
±
0.25
CDFS
2.5
< z < 3.5
galaxies
in
a2.90
similar
mass
range
selected
from
CDF-S
(red
laxies
to z radii
∼ 3. They
10whole
11
light
of
all
SMGs
is
±
0.45
kpc
which
is
in
good
BM/BX
galaxies
over
the
whole
BM/BX
galaxies
over
the
mass
range
(2.68
±
0.25
−
10
is
2.65
±
0.56
which
masses
of
10
Quiescent(CDFS)
0.5
<
z
kpc).
The
median
effective
radii
of
SMGs
with
stellar
=
circles). The red triangle is the median sizes of quiescent < 1.5
lower surface densities
Quiescent(CDFS)
1.5
< z <rad
2
kpc).
The
median
effective
radii
ofmedian
SMGs
with
stellar
kpc).
Thesame
effective
agreement
with
the
optical
rest-frame
sizes
of
the
10
11median
galaxies
studied
in
van
Dokkum
et
al.
(2008b)
(VD08
the
BM/BX
of
at
the
mass
range
(2.6
dicated
that
the
color
of
− 10of 10is10 2.65
±
0.56
which
is
similar
to
masses of 10masses
Quiescent(VD08)
2 < z < 2.5
11
10
11
11
−
10
is
2.65
±
0.56
which
is
similar
−
10
is
±
masses
of
10
M2.65
entally
with the galaxies
stellar
hearafter)
with
a median
stellar
mass of
1.7 × 10
BM/BX
over
the
whole
mass
range
(2.68
±
0.25
"
SMG
0.5
<resam
z <to
3.
errors
are
computed
using
bootstrap
BM/BXthe
of at
the
same
mass
range
(2.68±0.19).
The
and
median
redshift
of
2.3.
Table
3
lists
the
median
half
ass.the
=
BM/BX
of
at
the
same
mass
range
(2.68±0.19).
The
the
BM/BX
of
at
the
same
Notes.
The
median
sizes mass
of
samp
kpc).
The
median
effective
radii
of
SMGs
with
stellar
We
further
compare
the
size
distribution
10
11
errors
computed
bootstrap
resampling.
− 10
M! ex
massesresampling.
between
lightusing
radii for
UV-bright
and
quiescent
galaxies10using
seen
in boots
tween
color are
and stellar
errors
are
computed
using
bootstrap
errors
are
computed
10
11
(van
Dokkum
et al. time
2008b)
and SMG
galaxies
and
BM/BX
galaxies
in
Figure
Figure
10.
The
growth
of
UV-bright
galaxies
with
is
-forming
and
quiescent
−
10
is
2.65
±
0.56
which
is
similar
to
masses
of
10
We further We
compare
the
size distributions
ofcompare
sub-mm
is all galaxies
at the
stellar
mad
further
compare
the
size
distributions
ofsame
sub-mm
We
further
the
size
in
Figure
8.
In
the
left
also
illustrated
in
this
figure.
The
dashed
line
shows
that
forming
sBzKand
galaxies
top
panel,
the
histograms
show
the
size
d
the
BM/BX
of
at
the
same
mass
range
(2.68±0.19).
The
galaxies
BM/BX
galaxies
in
Figure
9.
In
the
galaxies
and
BM/BX
galaxies
in
Figure
9.
In
the
versus the
galaxies scaled
up their
sizesBM/BX
towards lower
galaxies
and
galaxies
50 ) is plotted
MGs
are marked
with UV-bright
for
galaxies
in
our
sample.
We
BM/BX
galaxies
SMGs.
In
the
botto
−1.11±0.13 and
errors
are
computed
using
bootstrap
resampling.
BzK
galaxies
.
alaxies
at
redshift
∼
1.
redshifts
as
(1
+
z)
top
panel,
the
histograms
show
the
size
distribution
of
top panel,
the
histograms
show the
distribution
of stellar
e relation
forthe
starmedian
forming
sBzK galaxies
top size
panel,
histograms
show
squares
are
is athe
relation
between
ma
normalized
cumulative
distribution
functio
marked
with
Previous
studies
(Franx
et
al.
2008;
Kauffmann
et
al.
alaxies
from
GOODS-N
X BM/BX
galaxies
SMGs
areand
marked
with
BM/BX
galaxies
and
SMGs.
In galaxies
the bottom
panel,
We
further
compare
the
size
distributions
of
sub-mm
galaxies
at 0.5
<
z < 3.5.the
Th
with
one
σ and
dispersion.
galaxies
SMGs.
In
the
bottom
panel,
the
BM/BX
and
SMGs.
In
the
median
2003)
reported
that
there
is
a
correlation
between
color
he
blue
stars
and
brown
ectively.
The red squares
are the
median
for
both
samples.
As
plot
shows,
forthis
both9.
spectroscopic
andthe
phot
to
the
BM/BX
galaxnormalized
cumulative
distribution
functions
are
shown
galaxies
and
BM/BX
galaxies
in
Figure
In
the
normalized
cumulative
distribut
normalized
cumulative
distribution
functions
are atshown
and
stellar
mass surface density
of
galaxies
boththat
low there i
Gs)
and their
analogues
Our
results
verify
ofdispersion.
sBzK
sample
with
one
σ
dispersion.
e-mass
relation
to
the
tions
ofAs
SMGs
are
comparable
with
z ∼rs
for
both
samples.
this
plot
shows,
the
size
distribuM/BX
galaxand
high
redshifts.
However,
the
results
based
on
photoright
panel,
the
relafor
both
samples.
As
this
plot
top
panel,
the
histograms
show
the
size
distribution
of
of
the
slope
of
the
size-mass
ine
is
the
best
fit
to
the
BM/BX
galaxfor
both
samples.
Asshowthis
plot shows,
the size distribumass-size
relation
for
Fig.
9.—
Top
panel:
Histograms
the
distribution
of
sizes
of
ation
to
the
K)
and
surface
density
metric
redshifts
can
be
uncertain
especially
for 2
star
formgalaxies
and
the
SMGs
the
same
tions of
SMGs
with
zfollow
∼
The evolution
ofUV-bright
the relation
have a similar size-mass
relation
to are
the comparable
g different estimates of stellar masses.
v-z
z-k
We study the size evolution of galaxies with spectroopic redshifts between z ∼ 0.5 − 3.5. Our results are
Surface
Density
ummarized in 6.3.
FigureColor
10. As -this
plot illustrates,
at
ach epoch, the median sizes of UV-bright galaxies (color
Evolution of the mass sizeresults
to z = 3.5 in of
the GOODS
North field
ne
of with
thestellar
surprising
Franx
11
are et al. 9 (2008)
quares)
masses 10 <relation
log(M
∗ /M" ) <
rger
a median
factor of 0.45 ±between
0.09) than quiescent
the(bytight
correlation
stellar mass surface
alaxies in a similar
range
selected from CDF-S (red
2 mass
z
~
1
zto
~ 2 z ∼ 3. They
ity
(M
/r
)
and
color
of
galaxies
rcles). The∗ redetriangle is the median sizes of quiescent
alaxiesthat
studied
in vangalaxies
Dokkum et have
al. (2008b)
(VD08
wed
bluer
lower
surface densities
11
earafter) with a median stellar mass of 1.7 × 10 M"
red
quiescent
They
that the color o
nd
median
redshift of ones.
2.3. Table
3 lists indicated
the median half
ght radii
for UV-bright
and quiescent
galaxies seen in
xies
correlates
more
fundamentally
with the stellar
igure 10. The growth of UV-bright galaxies with time is
than
the mass.
sosurface
illustrateddensity
in this figure.
The dashed
line shows that
scaled up
their sizesbetween
towards lower
eV-bright
showgalaxies
the tight
relation
color
and
stellar
dshifts as (1 + z)−1.11±0.13 .
surface
density
both
Previous
studies
(Franxfor
et al.
2008; star-forming
Kauffmann et al. and quiescent
10(V − z ) and surface
11between
Fig. 8.— Left panel
: The correlation
between
colorcorrelation
density
for Figure
galaxiescolor
in the redshift range
of 0.5
< z < 1.5
003) reported
that
there
is
a
∼
10
−
10
in
8.
In
the
left
xies
with
M
"
and stellar masses 10 <
M/M < 10 in GOODS-N field. The blue stars are GALEX (LBGs) from GOODS-N and the brown circles
Stellar
Mass
Surface
Density
are from CDFS. The gray symbols
are all galaxies from
CDFS and the
black triangles are galaxies
from GOODS-N. The plots indicates
nd stellar
mass
surface
of: Thegalaxies
low
that blue
galaxies having
lower surface density
density. Right panel
color (z
− K) at
versus both
surface density
for galaxies at 1.5 < z < 2.5, as
lndthe
observed
color
(V
−z
)
is
plotted
versus
the
the green squares are the BM/BX galaxies from GOODS-N606
and the brown circles
are
pseudo
BM/BX
from
CDF-S.
850
high redshifts. However, the results based on photoTABLE 3
of different
ar
mass
surface
density
for galaxies
at samples
redshift ∼ 1
etric
redshifts
can be uncertain
especially
forMedian
starsizesform10
!
11
606
850
850
mass and
We study the size evolution of galaxies with spectroscopic redshifts between z ∼ 0.5 − 3.5. Our results are
The ste
summarized10in Figure 10. As this plot illustrates, at
Mosleh et al.
shown in
each epoch, the median sizes of UV-bright galaxies (color
squares) with stellar masses 10 < log(M∗ /M" ) < 11 are
galaxies. redshift
Since BzK b
08)
tend to represent
mas
larger (by a median factor of 0.45 ± 0.09) than quiescent
ace
UV-brigh
that they appear large
galaxies in a similar mass range selected from CDF-S (red
hey
rected forSolid
differingand
ste
circles). The red triangle is the median sizes of quiescent
ties
parison to be made.
galaxies studied in van Dokkum et al. (2008b) (VD08
r of
relation
f
Our
results
also
show
11
lar
hearafter) with a median stellar mass of 1.7 × 10 M"
half light and
radii ofearly
2.90 ±
and median redshift of 2.3. Table 3 lists the median half
sizes of t
Fig. 3.—
distribution
of UV-bright
galaxies
with stellarframe optical
light Size
radii for
UV-bright and
quiescent galaxies
seen in
lar
symbols
is in agreement with a
ass
10
<
log(M
/M
)
<
11
at
z
∼
1
(blue
histogram)
and
∗
"
Figure 10. The growth of UV-bright galaxies with time is
ent
(2010) where
they wi
use
South
∼
2
(green
histogram).
The
half
light
radii
of
UV-bright
galaxeft
also illustrated in this figure. The dashed line shows that
ing and show that th
Comparin
sthe
(GALEX(LBGs))
at zscaled
∼ 1 are
largersizes
than
BM/BX
UV-bright galaxies
up their
towards
lowergalaxies at
SMGs and
UV-bright
−1.11±0.13
.
∼∼1.2. redshifts as (1 + z)
parable. lows
The typical
us th
Previous studies (Franx et al. 2008; Kauffmann et al.
S-N
H -band (rest-frame opt
(2010) is mass-sele
2.8 ± 0.4 wh
2003) reported that there is a correlation between color
wn
surements.
Almaini
e
and stellar
mass of
surface
of galaxies
at)both
low at z ∼ 3
ues
ithin mass
range
10 density
< log(M
< 11
of
each
p
∗ /M"
of submm galaxies wit
and high redshifts. However, the results based on photoelaour
sample.
The
median
half-light
radius
of
these
4
mass
frame optical
and bin
found
ity
metric redshifts can be uncertain especially for star formand
LBG
assiveinggalaxies
kpc,
0.82
± 0.23 smallerof the SMGs
ith
galaxies. is
By 2.22
means±
of 0.61
our large
sample
of galaxies
z
∼
1
and
have
larger
median
ste
res
with
secure
redshifts,
we
have
confirmed
that
the
tight
Redshift
han
z
∼
2
BM/BX
galaxies.
We
perform
a
power-law
radii
com
galaxies (also
mentione
Fig.
10.—
The
median
sizes
of
UV-bright
galaxies
can
correlation between color and stellar mass surface den-(squares)
−1.11±0.13
suggests higher stellar m
as a function of redshifts for galaxies with stellar masses
1010 <
∝
(1
+
z)
taveon the
evolution
and
find
r
For galax
sity size
of
galaxies
that
holds
out
to
high
redshifts;
as
the
11
UV-bright
galaxies
evolve
as
M/M! < 10 in GOODS-N field. eThe red filled circles are quito UV bright galaxies.
escent
galaxiesof
from
CDF-S
with similar
mass
range and
and thestellar
mass
densities
blue
starstudy
forming
galaxies
are
between these
ver
range
0.6
!
z
!
3.5.
fer two
by gala
a
red those
triangleof
is red
quiescent
sample from
Dokkum
et al. (2008b)
ace
smaller than
quiescent
ones.van11This
verifies
by itself allow conclus
nst
Size
Conclusion
dashed linewith
with median
stellar masses
of 1.7 × 10 M . The
The size-redshift
relation
for
star-forming
galaxies
2 res
tween theand
two populat
shows the best fitting size evolution to the UV-bright galaxies
!
Summary
Quantifying the size mass evolution of galaxies without potential
uncertainties of photometric redshift.
UV-Bright galaxies evolve strongly with redshift.
UV-Bright galaxies are significantly larger than quiescent
galaxies. At the same mass, the median difference is 0.45 ± 0.09.
The LBG, BM/BX and GALEX/LBG samples show smooth
evolution with redshift.
The SMGs have half-light radii similar to UV-Bright galaxies of
the same mass.
Draft version November 12, 2010
Preprint typeset using LATEX style emulateapj v. 5/25/10
THE EVOLUTION OF THE MASS-SIZE RELATION TO Z=3.5 FOR UV-BRIGHT GALAXIES AND SUB-MM
GALAXIES IN THE GOODS-NORTH FIELD
Moein Mosleh1 , Rik J. Williams 2 , Marijn Franx 1 , Mariska Kriek3
Draft version November 12, 2010
Abstract
We study the evolution of the size - stellar mass relation for a large spectroscopic sample of galaxies
in the GOODs North field up to z ∼ 3.5. The sizes of the galaxies are measured from Ks -band
images (corresponding to rest-frame optical/NIR) from the Subaru 8m telescope. We reproduce earlier
results based on photometric redshifts that the sizes of galaxies at a given mass evolve with redshift.
Specifically, we compare sizes of UV-bright galaxies at a range of redshifts: Lyman break galaxies
(LBGs) selected through the U-drop technique (z ∼ 2.5 − 3.5), BM/BX galaxies at z ∼ 1.5 − 2.5, and
GALEX LBGs at low redshift (z ∼ 0.6 − 1.5). The median sizes of these UV-bright galaxies evolve as
(1 + z)−1.11±0.13 between z ∼ 0.5 − 3.5. The UV-bright galaxies are significantly larger than quiescent
galaxies at the same mass and redshift by 0.45 ± 0.09 dex. We also verify the correlation between color
and stellar mass density of galaxies to high redshifts. The sizes of sub-mm galaxies in the same field
are measured and compared with BM/BX galaxies. We find that median half-light radii of SMGs is
2.90 ± 0.45 kpc and there is little difference in their size distribution to the UV-bright star forming
galaxies.
Subject headings: galaxies: evolution – galaxies: high redshift – galaxies: structure
arXiv: 1011.3042v1
1. INTRODUCTION
Recent studies provide evidence that sizes of galax-
based on photometric redshifts. Although techniques
to construct galaxy spectral energy distributions (SEDs)