Internal Kinematics of Galaxies at z = 0.25−0.

45

Luc Simard1 and Christopher J. Pritchet1

Department of Physics and Astronomy, University of Victoria, P.O. Box 3055, Victoria, Canada, V8W 3P6,
E-mail: (simard, pritchet)@uvastro.phys.uvic.ca
arXiv:astro-ph/9606006v1 3 Jun 1996

Submitted to Astrophysical Journal Letters

ABSTRACT
Low-mass starbursting galaxies have been proposed as the explanation of the excess
of faint galaxies observed at intermediate redshifts. If this hypothesis is correct, then
intermediate redshift galaxies should rotate more slowly than nearby galaxies with the
same rest-frame luminosity. We present the results of a survey of the internal kinematics
of intermediate redshift (z = 0.25−0.45) field galaxies to search for this effect. Using the
Canada-France-Hawaii Telescope, spatially-resolved spectra of the [O II] λλ 3726−3729
Å doublet emission line have been obtained for 22 galaxies. Vrot sin i and [O II] disk
scale lengths have been extracted from each galaxy spectrum using a Bayesian fitting
technique.
Galaxies in the sample are found to be ∼1.5−2.0 mag brighter than expected from
their rotation velocity and the local Tully-Fisher (TF) relation. Low-mass galaxies ex-
hibit a wider range of evolution relative to the TF relation than high-mass galaxies. The
main source of uncertainty in this result is the large scatter in the local TF relation for
late-type galaxies. Luminosity-dependent luminosity evolution neatly reconciles the lack
of evolution seen in other works with the results of our survey. It is also found that the
overall properties of [OII] kinematics at intermediate redshifts are varied. For example,
25% of the field galaxies in the sample have [OII] kinematics unrelated to rotation; [OII]
emission is confined to the nucleus in most of these galaxies. Anomalous kinematics
is found to be related to the presence of companions — i.e. minor merger events. A
Doppler ellipse similar to those found in local dwarf galaxies has been observed in a z =
0.35 galaxy, and may be interpreted as a supernova-driven supershell.

Subject headings: galaxies: kinematics and dynamics — galaxies: evolution — methods:

data analysis

1 Visiting
Astronomer, Canada-France-Hawaii Telescope, which is operated by the National Research Council of Canada, le Centre
National de la Recherche Scientifique of France, and the University of Hawaii.

1
1. Introduction to those of local galaxies in both form and ampli-
tude. The kinematics of the Keck galaxies showed ev-
The observed number counts of faint galaxies ex- idence for only a modest increase in luminosity (∆MB
ceed the predictions of standard models in which evo- ≤ 0.6 mag) compared to the local Tully-Fisher rela-
lutionary effects are ignored. Yet the faint galaxy red- tion. This is in apparent disagreement with the strong
shift distribution appears to be well modelled by these evolution in galactic disks observed by Schade et al.
same no-evolution models (Lilly et al. 1995, Lilly (1996) over the range 0.1 < z < 0.6.
1993, Koo and Kron 1992, and references therein).
In this paper, we present the internal kinemat-
The amount of evolution that faint galaxies have un-
ics of a sample of intermediate redshift galaxies,
dergone (relative to galaxies at the present epoch)
and demonstrate how luminosity-dependent luminos-
is clearly a critical parameter for understanding the
ity evolution can reconcile observations of surface
Universe at intermediate redshifts. Here we focus on
brightness and internal kinematics.
the novel technique of internal kinematics for study-
ing galaxy evolution.
2. Observations
Internal kinematics is directly related to a sin-
gle fundamental property of galaxies: mass. Obser- Twenty field galaxies and two cluster galaxies in
vations of the velocity field of intermediate redshift the redshift range 0.25−0.45 were selected from the
galaxies can therefore be used to quantify the amount CNOC cluster survey database (Yee et al. 1996).
of luminosity evolution that these galaxies have un- This database contains position, Gunn r magnitude,
dergone. As an illustration, consider a galaxy with g − r color, and redshift for thousands of galaxies.
rest-frame absolute luminosity equal to that of the Elongated objects were preferentially selected to min-
Milky Way. If this galaxy were also as massive as the imize sin i effects. No size/color selection criteria were
Milky Way (i.e. if it were unevolved in luminosity rel- used. Rest-frame [OII] equivalent widths W3727 were
ative to the Milky Way), then a rotation velocity of chosen to lie between 20 and 50 Å. Target galaxies
∼ 200 km/s would be observed. On the other hand, may be [OII] strong relative to local early-type (Sb
if it were in fact a lower mass object that had been and earlier) spirals, but they have the same W3727 as
boosted in luminosity by ∼10× (as suggested in the many local late-type (Sc and later) galaxies (Kenni-
luminosity-dependent luminosity evolution scenario – cutt 1992). Our range of W3727 is also representa-
Broadhurst et al. 1988), then the TF relation would tive of the excess galaxy population at intermediate
predict a rotation velocity ∼ 100 km/s. redshifts (Broadhurst et al. 1988, Broadhurst et al.
This kinematical approach to measuring evolution 1992).
is direct. It is not affected by uncertainties in mod- Spatially-resolved observations of the [O II] λλ
els based on local luminosity functions. Furthermore, 3726.1, 3728.8 Å doublet emission line were obtained
whereas luminosity functions derived from redshift with the Multi-Object Spectrograph (MOS) and the
surveys constrain the average amount of evolution in Subarcsecond Imaging Spectrograph (SIS) at the Ca-
an entire galaxy population, the present approach can nada-France-Hawaii 3.6-m Telescope (CFHT) in July-
measure luminosity enhancement in individual galax- August 1994. SIS was ideally suited for studying [OII]
ies, and can thus tie luminosity evolution directly to kinematics because SIS “tip-tilt” corrections yielded a
galaxy properties on a galaxy by galaxy basis. seeing FWHM of 0′′ .5−0′′ .6. We used the B600 grism
Most previous kinematical studies (Franx 1993, and a LORAL 2048×2048 CCD (QE = 23%, NR = 8
Vogt et al. 1993, Colless 1994, Koo et al. 1995, e− ). The detector was binned by 2 along both axes to
Forbes et al. 1996) have suffered either from small improve S/N ratio. Image scale and dispersion were
sample size, or from a lack of spatial information; 0′′ .31/pixel and 1.58 Å/pixel (MOS) and 0′′ .17/pixel
but they have clearly demonstrated that current tele- and 0.88 Å/pixel (SIS). Typical total integration time
scopes are adequate to the challenge of measuring the per galaxy was 7200 seconds split in 2−3 exposures
internal kinematics of intermediate redshift galaxies. for cosmic ray removal.
Recently, Vogt et al. (1996) presented a beautiful
set of rotation curves (observed with the Keck Tele- 3. Synthetic Rotation Curve Fitting
scope) for nine faint field galaxies in the redshift range
Flux levels in [O II] were very low (typical S/N
0.1 ≤ z ≤ 1. These rotation curves appeared similar
ratio per pixel ∼ 2−3); it was therefore important

2
to choose a method of analysis which used all of the their respective error distribution. For each param-
pixels simultaneously to statistically find the best pa- eter, we took the median value of P (ω|D, M ) as the
rameter values and their respective uncertainties. We best value and the 16th and 84th percentile values
adopted a synthetic rotation curve fitting technique, as our “1 σ” error bars. We tested confidence inter-
in which models were compared to the two dimen- vals and parameter value recovery using a set of 342
sional distribution of pixel intensities in the [O II] line simulated rotation curves. We chose input parameter
(the two dimensions being wavelength or velocity, and values in these simulations to yield worse S/N ratios
position along the slit). The parameters of the fitting than seen in the data. No biases were detected. The
model were the projected rotation velocity Vrot sin Metropolis parameter estimates were also more ro-
i of the galaxian disk in km/s, the [OII] exponential bust to noise than least-squares estimates at low S/N
disk scale length rd in h−1 kpc, the [OII] total line ratios.
flux in DU, and the dimensionless [OII] doublet in-
tensity ratio I(3726 Å)/I(3729 Å). (It was necessary 4. Results
to include this doublet intensity ratio in the model
because the [O II] doublet is partially resolved due to 4.1. [OII] Morphologies
(1+z) spectral stretching.) [OII] kinematics at intermediate redshifts is varied.
The 2D model distributions of pixel intensities Seven galaxies have [O II] scale lengths significantly
were constructed assuming that the [OII] emission smaller than their broad-band scale lengths, consis-
was distributed in a thin exponential disk with a flat tent with central unresolved [O II] sources. Two of
rotation curve. These [OII] model disks were con- these “unresolved [O II]” galaxies are serendipitously
volved with a point-spread-function (PSF) extracted observed cluster galaxies (we cut extra slitlets on the
from direct images; the effect of placing a slit in front mask for cluster galaxies whenever possible). Such
of the image was then computed. Finally, the flux discrepancies indicate that the [OII] gas kinematics
passing through the synthetic slit was convolved with is decoupled from galaxy rotation in some galaxies
the spectrograph instrumental line shape, which was and could be confined to the nucleus. Two properties
extracted from comparison arc lines. of kinematically anomalous galaxies have emerged so
The best fitting parameter values were found us- far: (1) All but one have close companions. This sug-
ing the Metropolis algorithm (Saha and Williams gests that enhanced star formation activity may be
1994). This Bayesian algorithm starts with some set the result of merger events. (2) Some of these galax-
of parameter values ω and the associated P (ω|D, M ) ies appear to be of early-type. Kinematically “anoma-
— the probability that ω is the true parameter set lous” field galaxies make up 25% of the field sample.
given the data D and a model M. It then picks a This is similar to the fraction of blue-nucleated galax-
possible change δω in the parameters and computes ies observed in HST images of galaxies at z ∼ 0.6
P (ω + δω|D, M ). If P (ω + δω|D, M ) > P (ω|D, M ), (Schade et al. 1995).
then the change δω is accepted. If P (ω + δω|D, M ) < In one galaxy at z = 0.35, the [OII] line was donut-
P (ω|D, M ), then the change δω is accepted only some shaped. This line shape could be interpreted in two
fraction P (ω +δω|D, M )/ P (ω|D, M ) of the time. Af- ways: either (1) the line is made of two rotation
ter hundreds of iterations, the distribution of accepted curves from two galaxies very close together, or (2)
ω values will converge to P (ω|D, M ) provided that all the line arises from an expanding supershell, presum-
possible ω are eventually accessible. Parameter space ably driven by supernova winds from a massive star-
is thus sampled with a density proportional to the burst. If option (2) is correct, then the supershell has
likehood. a diameter of 2.5 h−1 kpc and a rest- frame expansion
The trial changes δω are chosen at random. If they velocity of about 80 km/s. These characteristics are
are too small (i.e. the parameter search is too “cold”), amazingly similar to those of supershells observed in
then all iterations will accept changes. If the search local dwarf irregular galaxies (Marlowe et al. 1995,
is too “hot”, then none of the iterations will accept Martin 1995). In another galaxy at z = 0.42, there is
changes. The “temperature” of the search is regu- a relatively strong [OII] source unresolved both spec-
lated so that half the iterations accept changes. As trally and spatially superimposed over a fainter ro-
the algorithm proceeds, trial values are stored to si- tational [O II] component. The source could be a
multaneously derive the best parameter values and giant HII region 2.6 h−1 kpc from the center of the

3
galaxy. More details on these unusual sources will The luminosity-dependent luminosity evolution sce-
appear elsewhere (Simard and Pritchet 1996). nario neatly reconciles the various amounts of lumi-
nosity evolution seen in the surface brightness and
4.2. An Intermediate Redshift Tully-Fisher internal kinematics studies. At the low end of the
Relation galaxy mass spectrum, Koo et al (1995) looked at
compact, unresolved galaxies with narrow emission
Figure 1 shows kinematical evidence for luminosity
line widths. The line widths of these galaxies were
evolution at intermediate redshifts. The locus of the
2−3× lower than expected from the TF relation of
local Hα rotation velocity−B band Tully-Fisher (TF)
normal spiral galaxies, implying large amounts of
relation for all morphological types (from the data of
evolution (∼ 3 mag). These low-mass galaxies had
Mathewson et al. 1992) is used as a reference. Solid
linewidths similar to those of local HII galaxies. At
circles are the rest-frame Vrot sin i ’s versus rest-frame
the high end of the galaxy mass spectrum, the Keck
B magnitude for the kinematically normal galaxies in
study of Vogt et al. (1996) selected galaxies that were
our sample — i.e. galaxies with [OII] scale length
intrinsically large (rd ≥ 3.0 kpc) and bright (MB ≤
consistent with their broad- band scale length. B-
−20.7). These galaxies were also intrinsically massive,
band k-corrections were computed using the tables of
with typical rotation velocities of 200 km/s. The in-
Frei and Gunn (1994). The upper long dashed line
crease in B luminosity with respect to the local TF
is an unweighted linear fit to all local morphological
relation was less than 0.6 mag.
types. This linear fit was then shifted by ∆MB0 =
−1.0 mag (middle dashed line) and ∆MB0 = −2.0 Our CFHT sample occupies a niche in size and
mag (lower dashed line) to represent various degrees mass between the above two Keck samples. The
of luminosity evolution. CFHT galaxies are typically a full magnitude fainter
than the objects in the Vogt et al. study. They
Galaxies in our sample are ∼1.5−2.0 magnitudes
are also intrinsically smaller with typical disk scale
brighter than expected from their rotation velocity,
lengths less than 2.0 kpc, and some of them are barely
given the local TF relation defined by the Mathew-
resolved. They are also less massive, with rotation
son et al. sample. Such a large local sample clearly
velocities ∼ 100 km/s, but they are nonetheless more
shows the scatter in the local TF relation. Locally,
massive than the galaxies observed by Koo et al.. If
there are no systematic shifts between loci of differ-
mass is taken as an indicator of the luminosity that a
ent morphological types, but it is obvious that certain
galaxy would have had in a quiescent phase, then all
types (e.g. T = 6) display larger dispersions.
three internal kinematics studies can be understood
with luminosity-dependent luminosity evolution.
5. Discussion
The B-band surface brightness µ0 (B) of field disk
The scatter in the local TF relation has a direct galaxies undergoes strong evolution over the redshift
impact on the exact amount of luminosity evolution range 0.1 < z < 0.6 compared to two different local
derived from our observed TF relation at intermedi- log rd −MAB (B) relations: the Freeman law and an
ate z. This is because scatter in the local TF re- empirical z = 0.06 relation for galaxies in Abell 2256
lation may correlate with star formation rates, and (Schade et al. 1996). At a redshift of 0.5, ∆µ0 (B) is
hence emission line strengths. In order to accurately equal to −1.1 mag. This is consistent with or perhaps
measure magnitude offsets relative to the local Tully- slightly less than the evolution seen in our CFHT sam-
Fisher relation, each galaxy in our CFHT sample (or, ple, but it is certainly more than the amount of evo-
for that matter, any other sample) should be com- lution seen in the Keck sample of Vogt et al.. Looking
pared to the local Tully-Fisher relation for galaxies at Figure 1 of Schade et al., there is a hint that sur-
with similar [OII] emission line strengths. This is not face brightness evolution depends on disk scale length:
possible at present. A large, homogeneous sample of smaller galaxies evolve more drastically than large
local emission-line strengths, rotation velocities, mor- galaxies. The effect is particularly noticeable in the
phologies and absolute magnitudes is required to set- highest redshift bin where the log rd −MAB (B) rela-
tle this issue. This will remain a major limitation of tion clearly curves “down”. Taking MB ≃ −21 and
internal kinematics studies at high redshifts as long as <rd > = 4.3 kpc (H0 = 75), one can see on the Schade
technical requirements restrict such studies to strong diagram for 0.45 < z < 0.65 that a number of bright,
emission-line objects. large galaxies at (log rd = 0.8, MAB (B) = −21) show

4
little or no evolution – similar to what is observed in
the Keck sample!
The diversity of [OII] kinematics and the signifi-
cant amount of luminosity evolution seen in our sur-
vey illustrate the power of internal kinematics as a
probe of the nature of intermediate redshift galax-
ies. Although samples are small compared to those of
redshift surveys, they already provide exciting direct
evidence that luminosity-dependent luminosity evo-
lution may indeed be at the root of the faint galaxy
excess problem. Tying internal kinematics to other
galaxy properties on a galaxy by galaxy basis will play
a key role in our understanding of galaxy evolution at
intermediate redshifts.

We thank the Canadian Time Allocation Com-

mittee for the CFHT for generous grants of observ-
ing time, and the CFHT organization for the tech-
nical support which made this project possible. We
also thank the Canadian Astronomical Data Center
(CADC) for generous allocations of computer time.
We gratefully acknowledge financial support from the
Natural Sciences and Engineering Research Council
of Canada through an operating grant to CJP and a
Postgraduate Fellowship to LS.

5
 Table 1
Target galaxies: Characteristics and Results.

ID z Gunn Gunn W3727 MB0 Vrot sin i rd

r (g − r)obs (Å) (km/s) (h−1 kpc)
(1) (2) (3) (4) (5) (6) (7) (8)

+ 9 +0.02
A2390-101033ac 0.2460 18.90 0.70 55 −20.3 −212 − 11 0.09 −0.02
+ 21 +0.50
A2390-100686n 0.3822 21.22 0.67 21 −19.4 91 − 39 2.3 −0.40
+ 18 +0.40
A2390-350416n 0.2558 20.14 0.27 35 −19.5 62 − 21 1.9 −0.40
+ 26 +0.70
A2390-350471n 0.2559 19.40 0.28 21 −20.3 171 − 29 2.2 −0.50
+ 22 +0.20
E1512-301037An 0.3457 21.41 0.19 41 −19.1 77 − 24 1.8 −0.50
+ 8 +0.40
E1512-301037Bn 0.3457 21.41 0.19 41 −19.1 86 − 18 0.8 −0.20
+143 +0.40
E1512-101526a 0.4026 20.36 1.12 48 −20.2 −222 − 33 0.1 −0.10
+ 13 +0.20
E1512-201429n 0.4231 20.44 0.56 26 −20.4 151 − 11 1.0 −0.10
+ 18 +0.20
E1621-100515 0.3455 20.00 0.42 22 −20.3 −237 − 7 0.8 −0.10
+121 +0.50
A2390-100225 0.3829 21.65 0.89 26 −18.9 61 −107 0.6 −0.40
+ 12 +0.03
A2390-101084ac 0.2302 17.31 0.68 110 −21.7 −271 − 12 0.4 −0.06
+ 13 +0.20
A2390-200928n 0.2645 21.51 0.22 40 −18.2 65 − 15 0.3 −0.10
+ 76 +0.40
A2390-200802n 0.3208 21.16 0.37 25 −19.1 −198 − 65 1.0 −0.50
+ 52 +0.90
A2390-200372n 0.3485 20.15 0.65 16 −20.2 −86 − 18 3.2 −0.60
+ 22 +0.10
E1512-201845a 0.3387 19.60 0.82 36 −20.5 385 − 31 0.4 −0.10
+ 13 +0.10
E1512-201773 0.3383 20.06 0.35 26 −20.3 −116 − 15 0.9 −0.10
+ 7 +0.10
E1512-200730a 0.4266 21.35 0.58 32 −19.5 −1 − 5 0.2 −0.10
+ 46 +0.30
E1512-200334n 0.4142 21.81 0.70 36 −19.0 60 − 31 1.5 −0.30
+ 56 +0.10
E1512-200672a 0.4152 21.85 0.43 29 −19.1 44 − 91 0.3 −0.10
+ 45 +0.30
E1512-201268n 0.3412 21.77 0.39 29 −18.6 33 − 84 1.4 −0.20
+ 39 +0.20
E1512-202096n 0.4252 20.33 0.66 23 −20.5 28 − 33 2.0 −0.30
+ 79 +0.40
E1512-201125a 0.3823 21.69 0.74 15 −18.9 59 − 92 0.5 −0.30

a
Kinematically anomalous galaxy
n
Kinematically normal galaxy
c
Serendipitously observed cluster galaxy

Note.—(1) Galaxy identification code. First part is the CNOC cluster name and the second part is the CNOC
PPP number; (2) CNOC galaxy redshift; (3) CNOC observed Gunn r magnitude; (4) CNOC observed Gunn g − r
color; (5) Rest-frame [OII] equivalent width. Obtained by dividing the observed width by (1+z). Error ∼ 10%; (6)
k-corrected absolute B-band magnitude (H0 = 75 km s−1 Mpc−1 , q0 = 0.5). Error = ±0.1 mag; (7) Rest-frame
projected [OII] disk rotation velocity with 68% confidence interval; (8) [OII] exponential disk scale length with 68%
confidence interval (h = H0 /(100 km s−1 Mpc−1 ), q0 = 0.5). Galaxies with no superscript were rejected because
cosmic rays hit the [OII] line directly, or simulations showed that the [OII] flux was too low to derive reliable
kinematical parameters.

6
REFERENCES Yee, H. K. C., Ellingson, E. and Carlberg, R. G. 1996,
ApJS, 102, 269
Broadhurst, T. J., Ellis, R. S., and Shanks, T. 1988,
MNRAS, 235, 827
Broadhurst, T. J., Ellis, R. S., and Glazebrook, K.
1992, Nature, 355, 55
Colless, M. 1994 in 35th Herstmonceux Conference:
Wide Field Spectroscopy and the Distant Uni-
verse, S. Maddox and A. Aragon-Salamanca (eds.),
World Scientific Publishing Company, Singapore.
Forbes, D. A., Phillips, A. C., Koo, D. C. and Illing-
worth, G. D. 1996, ApJ, 462, 89
Franx, M. 1993, PASP, 105, 1058
Frei, Z. and Gunn, J. E. 1994, AJ, 108, 1476
Kennicutt, R. C. 1992, ApJ, 388, 310
Koo, D. C., Guzman, R., Faber, S. M., Illingworth, G.
D., Bershady, M. A., Kron, R. G. and Takamiya,
M. 1995, ApJ, 440, L49
Koo, D. C. and Kron, R. 1992, ARA&A, 30, 613
Lilly, S. J., Tresse, L., Hammer, F., Crampton, D.
and LeFèvre, O. 1995, ApJ, 455, 108
Lilly, S. J. 1993, ApJ, 411, 501
Marlowe, A. T., Heckman, T. M., Wyse, R. F. G. and
Schommer, R. 1995, ApJ, 438, 563
Martin, C. L. 1995, BAAS, 27, no. 4, 1375
Mathewson, D. S., Ford, V. L. and Buchhorn, M.
1992, ApJS, 81, 413
Saha, P. and Williams, T. B. 1994, AJ, 107, 1295
Schade, D., Lilly, S. J., Crampton, D., Hammer, F.,
Le Fèvre, O. L. and Tresse, L. 1995, ApJ, 451, L1
Schade, D., Carlberg, R. G., Yee, H. K. C., López-
Cruz, O. and Ellingson, E. 1996, ApJ, Submitted,
astro-ph/9604139
Simard, L. and Pritchet, C. J. 1996, In preparation
Vogt, N. P., Herter, T., Haynes, M. and Courteau, S.
1993, ApJ, 415, 95
Vogt, N. P., Forbes, D. A., Phillips, A. C., Gronwall,
C., Faber, S. M., Illingworth, G. D. and Koo, D. This 2-column preprint was prepared with the AAS LATEX
C. 1996, ApJ, Accepted, astro-ph/9604096 macros v4.0.

7
Fig. 1.— Kinematical evidence for luminosity evolu-
tion at intermediate redshifts. The locus of the local
Hα rotation velocity−B band Tully-Fisher relation for
all morphological types as defined by data taken from
Mathewson et al. (1992) is used as a reference. Solid
circles are the Vrot sin i ’s versus rest-frame B mag-
nitude for the kinematically normal galaxies in our
sample. The upper long dashed line is an unweighted
linear fit to all the local morphological types. This lin-
ear fit was then shifted by ∆MB0 = −1.0 mag (middle
dashed line) and ∆MB0 = −2.0 mag (lower dashed
line) to represent various degrees of luminosity evolu-
tion.

8
 3.0
T1 (Sa)
2.8
T3 (Sb)
T4 (Sbc)
2.6 T5 (Sc)
T6 (Scd)
2.4 T7 (Sd)
Log Vrot (km s )

T8 (Sdm)
-1

2.2 data

2.0

1.8

1.6

1.4

1.2

1.0
-23.0 -21.0 -19.0 -17.0 -15.0 -13.0
MB - 5 log h
0