Detecting Star Formation in Brightest Cluster Galaxies with GALEX
ABSTRACT We present the results of GALEX observations of 17 cool core (CC) clusters of galaxies. We show that GALEX is easily capable of detecting star formation in brightest cluster galaxies (BCGs) out to $z\ge 0.45$ and 50-100 kpc. In most of the CC clusters studied, we find significant UV luminosity excesses and colors that strongly suggest recent and/or current star formation. The BCGs are found to have blue UV colors in the center that become increasingly redder with radius, indicating that the UV signature of star formation is most easily detected in the central regions. Our findings show good agreement between UV star formation rates and estimates based on H$\alpha$ observations. IR observations coupled with our data indicate moderate-to-high dust attenuation. Comparisons between our UV results and the X-ray properties of our sample suggest clear correlations between UV excess, cluster entropy, and central cooling time, confirming that the star formation is directly and incontrovertibly related to the cooling gas. Comment: 39 pages, 11 figures; accepted for publication in The Astrophysical Journal. Figure quality reduced to comply with arXiv file size requirements
Detecting Star Formation in Brightest Cluster Galaxies with
A. K. Hicks1
We present the results of GALEX observations of 17 cool core (CC) clusters
of galaxies. We show that GALEX is easily capable of detecting star formation
in brightest cluster galaxies (BCGs) out to z ≥ 0.45 and 50-100 kpc. In most
of the CC clusters studied, we find significant UV luminosity excesses and colors
that strongly suggest recent and/or current star formation. The BCGs are found
to have blue UV colors in the center that become increasingly redder with radius,
indicating that the UV signature of star formation is most easily detected in the
central regions. Our findings show good agreement between UV star formation
rates and estimates based on Hα observations. IR observations coupled with our
data indicate moderate-to-high dust attenuation. Comparisons between our UV
results and the X-ray properties of our sample suggest clear correlations between
UV excess, cluster entropy, and central cooling time, confirming that the star
formation is directly and incontrovertibly related to the cooling gas.
Subject headings: galaxies: clusters: general—galaxies: elliptical and lenticular,
cD—galaxies: stellar content—galaxies: clusters: intracluster medium—stars:
formation—ultraviolet: galaxies—X-rays: galaxies: clusters
1Department of Physics & Astronomy, Michigan State University, East Lansing, MI 48824-2320
2Goddard Space Flight Center, Code 662, Greenbelt, MD, 20771
3Department of Astronomy, University of Maryland, College Park, MD 20742-2421
arXiv:1006.3074v1 [astro-ph.CO] 15 Jun 2010
– 2 –
The existence of star formation in “cooling flow” (hereafter referred to as cool core,
or CC) clusters has been a contentious issue for over 25 years (Fabian et al. 1986). A
class of clusters with high central gas densities and theoretically short cooling times was
discovered with the Einstein observatory (e.g., Fabian & Nulsen 1977; Cowie & Binney
1977; Mathews & Bregman 1978; Canizares et al. 1979; Mushotzky et al. 1981) via X-ray
imaging and low resolution X-ray spectroscopy. These objects were also associated with
Hα filaments (e.g., Cowie et al. 1983; Heckman et al. 1989), central radio sources, high
Faraday rotation (e.g., Ge & Owen 1994), extra blue light (e.g., McNamara & O’Connell
1989; Cardiel, Gorgas, & Aragon-Salamanca 1998) and spatial coincidence of the X-ray peak
with the central radio source (e.g., Burns et al. 1981). The simplest physical model was one
in which the gas in the center cooled by radiating away its thermal energy, gradually losing
pressure support, resulting in a flow (e.g., Cowie & Binney 1977; Fabian & Nulsen 1977;
Mathews & Bregman 1978). Cool cores are inferred to be present in more than half of all
clusters at low-redshift (Peres et al. 1998), and nearly as prevalent at moderate-z (30−50%
at 0.15 < z < 0.4; Bauer et al. 2005).
It has become clear with the advent of XMM and Chandra data that almost every
such cluster also shows a temperature drop in the center (e.g., Cavagnolo 2008). However,
measurements with the high spectral resolution XMM RGS spectrometer (e.g., Peterson et
al. 2003; Kaastra et al. 2004; Piffaretti & Kaastra 2006) show that the X-ray spectra of
the cooler gas has major differences from the theoretical cooling flow model, with a marked
absence of gas at temperatures below ∼ 1/3 of the average cluster temperature. Thus it
remains a mystery what happens to the cool gas. The combination of Chandra imaging and
radio data (McNamara et al. 2000; Blanton et al. 2001) have shown that in many of these
objects there exist holes in the X-ray surface brightness which are filled in by radio emitting
plasma, thus lending credence to ideas that feedback from AGN in brightest cluster galaxies
strongly modifies the cooling and thus reduces the net amount of material available for star
Despite past evidence of star formation in these systems (e.g., McNamara & O’Connell
1989; Cardiel, Gorgas, & Aragon-Salamanca 1998; Crawford et al. 1999) and the apparent
preference for emission-line systems to inhabit the cores of high-central density, short cooling
time clusters (Hu et al. 1985), the fact that star formation rate (SFR) estimates often differed
by factors of ∼ 10 or more from inferred X-ray cooling rates led to doubt that the two
1Based on observations made with the NASA Galaxy Evolution Explorer. GALEX is operated for NASA
by the California Institute of Technology under NASA contract NAS5-98034.
– 3 –
phenomena were related. However, recent UV investigations (Mittaz et al. 2001; Hicks &
Mushotzky 2005), Spitzer data (Quillen et al. 2008; O’Dea et al. 2008), and precision optical
photometry (Bildfell et al. 2008) have definitively shown that CC clusters are indeed the sites
of star formation, and that there is an indisputable relationship between X-ray properties
Previous studies have shown a connection between activity, such as star formation and
radio AGN, in brightest cluster galaxies (BCGs) and the thermodynamic state (traced by
entropy, cooling time, etc.) of the intracluster medium in the cluster core (e.g., Cavagnolo
2008). The physical explanation for a connection between the state of the hot gas in the core
inside 100 kpc and star formation in the inner 10 kpc in the brightest cluster galaxy is not at
all obvious. However the situation in nearby BCGs might be similar to what models predict
is happening in massive galaxies at high redshift. Models such as Ostricker & Ciotti (2005)
suggest that almost all star formation in the high-redshift universe derives from accretion of
gas that fell into potential wells, shocked, heated, then cooled (White & Frenk 1991; Fabian
et al. 1986). More recent simulations (e.g., Kereˇ s et al. 2009) show that this “hot mode”
of accretion might be the dominant mode of star formation for massive galaxies, but this
verdict is far from final owing to the unknown effects of feedback.
In either steady state or bursty star formation scenarios the dominant contribution to
the UV flux comes from short-lived main sequence stars, therefore the UV band constitutes
one of the prime routes to understanding star formation. Despite this fact, while cool core
clusters have been well observed in the radio (e.g., McNamara et al. 2000; Blanton et al. 2001),
IR (e.g., Quillen et al. 2008; O’Dea et al. 2008) and with emission line studies (e.g., Heckman
et al. 1989), there have been comparatively few published studies of UV observations of cool
core clusters (O’Dea et al. 2004; Hicks & Mushotzky 2005; Sparks et al. 2009; Donahue et al.
2010). Here we attempt to address and quantify the connection between X-ray properties
and star formation, using recent GALEX observations of a sample of 16 CC clusters. Unless
otherwise noted, this paper assumes a cosmology of H0= 70 km s−1Mpc−1, ΩM= 0.3, and
The Galaxy Evolution Explorer (GALEX) is an orbiting space telescope possessing both
imaging and spectroscopic capabilities in two ultraviolet wavebands, Far UV (FUV) 1350-
1780˚ A and Near UV (NUV) 1770-2730˚ A (Martin et al. 2005). GALEX has a very low
sky background, and very high sensitivity (∼ 24 apparent AB magnitudes in each filter for
2 ksec exposures). Pixels are 1.5??× 1.5??, and GALEX’s on-axis spatial resolution is 4.2??
– 4 –
and 4.9??for the FUV and NUV respectively (GALEX Technical Documentation2).
Our GALEX targets consist of 17 clusters of galaxies that exhibit evidence of central
cooling based on the indicators discussed in the introduction. These objects were chosen to
sample a wide range of redshifts (0.02 < z < 0.45) and central cooling times (0.5 < tcool< 4.6
Gyr at R = 20 kpc), be safely observable by GALEX (i.e., no bright nearby UV sources)
and have low attenuations (Av< 0.5); therefore they do not constitute a complete sample.
One of our targets (Abell 644) was observed after the loss of the GALEX FUV detector and
therefore only has NUV data. Figure 1 shows the GALEX NUV images of a representative
sample of our targets with Chandra X-ray contours overlayed. Table 1 lists the objects in
our sample, their redshifts, and GALEX exposure times. These exposure times were based
on our previous work with XMM-Newton Optical Monitor UV data (Hicks & Mushotzky
All photometry was performed on pipeline-processed GALEX intensity maps. FUV
data were convolved with a gaussian to match the NUV PSF, and the sizes of point sources
in the resulting images were checked against those in the NUV data with the IRAF tool
imexam. Photon counts were corrected for background using large (∼ 40??radius) source-
free regions taken from nearby areas of the camera in the same observation. Our photometric
measurements were compared to those obtained with the background-subtracted intensity
images provided in the GALEX pipeline, as an added check.
Final fluxes were determined by employing GALEX counts-to-flux conversions, and
correcting for average Galactic extinction in the line of sight to each cluster (Cardelli, Clay-
ton, & Mathis 1989). Errors were assessed by adding Poisson (root N) photon statistics in
quadrature to a conservative 5% fixed systematic error (Morrissey et al. 2007). All of our
targets were easily detected in both GALEX wavebands, with an average SNR of 40 (21) in
the NUV (FUV), and minimum SNRs of ∼ 6 in each band (for a fixed 7??radius aperture).
Photometric results are presented in Table 2.
– 5 –
4. Spatial Analysis
To investigate the spatial distribution of UV emission in our targets, radial flux profiles
were produced for each band from point source subtracted intensity maps (convolved to
produce matching PSFs, as above). Profiles were constructed using concentric annuli of at
least 5??width and binned to achieve S/N > 3. These profiles are shown in Figure 2. With
GALEX we are able to detect UV emission out to large radii in many of the CC clusters.
The surface brightness profiles were then background subtracted and combined to create
radial color profiles for each cluster (Figure 3). Overall the central colors of most of the BCGs
indicate the presence of a very young stellar population, and imply active star-formation.
We also see positive color gradients in nearly all of our targets, in keeping with the results
of Rafferty, McNamara & Nulsen (2008) and Wang et al. (2010) for cool core clusters.
Greater than 82% of elliptical galaxies have FUV-NUV colors of > 0.9 (Gil de Paz
et al. 2007), much redder than the central regions of all of our objects (Figure 3). However,
presumably due to variations in the UV upturn (thought to be caused by horizontal branch
stars) there is still a broad distribution of UV color among passive ellipticals, therefore we
have not attempted to determine a definitive extent of star formation in individual targets.
5. Fixed Aperture Analysis
To determine the amount of “excess” UV light present in our targets, we first need an
estimate of their “expected” UV emission. We obtain this empirically by examining the UV
emission of non-star forming ellipticals and BCGs, using 2MASS J band flux as a proxy for
the old stellar population. We avail ourselves of existing 2MASS photometric measurements
by adopting their fixed 7??radius aperture in this portion of our analysis. We note that this
aperture contains the majority of excess UV emission (Figures 2 and 3).
5.1. Control Sample
Our non-star forming control sample is composed of 17 cluster ellipticals and 22 BCGs
in non-CC clusters, all drawn from archival GALEX observations. The clusters used in our
calibration analysis are listed in Table 3 along with their redshifts and GALEX exposure
Elliptical galaxies were gathered from 4 clusters spanning a redshift range of 0.08 < z <
0.15. We used FUV-K colors as a proxy for galaxy type (Gil de Paz et al. 2007), adopting
– 6 –
a liberal cutoff of FUV-K = 7.5.
BCGs were selected for inclusion if they met one of the following criteria: 1) central
cooling time > 7 Gyr; 2) spectrally determined X-ray˙M consistent with zero; 3) X-ray
underluminous (Popesso et al. 2007) and 1.4 GHz luminosity < 1×1024W Hz−1(Sun 2009).
BCGs without ancillary X-ray or radio data were included when FUV-K > 7.5.
Our total (elliptical + BCG) calibration sample spans ranges in redshift and absolute J
magnitude that are well-matched to our target sample, with the exceptions of our highest-
redshift cluster (RXJ1347.5-1145 at z=0.45) and most luminous BCG (MKW4). We note
that none of our conclusions are based on individual objects in our sample.
Photometry was performed in 7??radius apertures centered on each galaxy (as described
in Section 3); measurements are given in Table 4. We see no relationship between NUV-J
color and redshift in our calibration sample, confirming that our aperture choice is sufficient
for meeting the goals of this study; e.g., our aperture choice is large enough that we capture
most of the “excess” UV light even at low redshifts and is not so large as to dilute the signal
below detection thresholds at high redshifts.
Least squares fits were executed between the properties of our calibration sample using
the wls regress algorithm of Akritas & Bershady (1996). This routine was chosen because
the scatter in UV luminosity (ostensibly stemming from variations in the UV upturn) vastly
dominates over J band luminosity uncertainties. Relationships are fitted with the form
log10Y = C1+ C2log10X (1)
Correlations between luminosities and flux ratios are given in Table 5 and are shown
in Figures 4 and 5. Clearly the FUV shows more scatter than the NUV. This scatter is
expected because the FUV filter covers a spectral region which is very sensitive to variations
in the magnitude of the UV upturn from object to object. This high sensitivity to the UV
upturn makes the NUV filter a more straightforward choice for investigating star formation
in early-type galaxies, so our discussion will focus on the NUV results.
5.2. UV Excesses
The UV luminosity excesses of our 17 cool core clusters were calculated by subtracting
the expected UV luminosity from the old stellar population (based on the fit obtained in
Section 5.1) from the measured value. The majority of our sample exhibits clear UV excesses,
indicating recent star formation. Figure 5 shows the J band luminosity of each galaxy in
– 7 –
our sample plotted against its UV/J flux ratios (UV-J colors). We have not attempted to
estimate and correct for internal dust absorption, and thus these excesses provide a lower
limit on the UV emission in these clusters. Our measured UV excesses are given in Table 6.
Starburst99 (Leitherer 1999) redshifted models corresponding to continuous Saltpeter
star formation over a 20 Myr period were used to estimate star formation rates for our
sample. The 20 Myr continuous model was chosen to grossly approximate episodic cooling
timescales, during which the system undergoes feedback processes with alternating heating
and cooling cycles; it is this model that we use in the figures to follow. We emphasize
that there are too many unknowns (e.g., internal reddening, IMF, continuous vs. burst star
formation, age of star formation, etc.) to predict accurate star formation rates, and that we
estimate SFRs with this model purely to facilitate comparisons with previous work and with
other wavebands. Resulting values are shown in Table 6.
Star formation rates estimated from the NUV and FUV bands show general agreement
(Figure 6), though FUV derived SFRs tend to be slightly lower, ostensibly due to the larger
scatter in the FUV calibration relationship. Because of the overall agreement between SFR
estimates in the two bands, and the tendency for NUV data to be less plagued by variations
in the UV upturn, we focus primarily on the NUV results in the following sections.
5.3. UV Colors
Some of our targets have very blue FUV-NUV colors when compared to our control
sample, but none of them are known to harbor a central AGN. It is possible that their dust
is of the Milky Way variety, which preferentially absorbs NUV emission and therefore results
in bluer observed colors (e.g., Witt & Gordon 2000).
In Figure 7, we show that the FUV-NUV colors of the central (7??) of our sample are
inversely correlated with excess UV luminosity (correlation coefficient = −0.74), such that
the reddest UV colors are associated with the weakest UV excesses. Typical FUV-NUV colors
for inactive BCGs are shaded grey on this plot, consistent with the colors of the BCGs with
the smallest excesses. A couple of the BCGs’ colors are undoubtedly affected by Lyman-α
emission contributing to the FUV band (ZwCl 3146) or NUV band (RXJ1347.5-1145).
If extinction by Milky Way-type dust is the explanation for the color trend, then the
conclusion from this plot is that the BCGs with the largest UV excesses have the largest
intrinsic dust extinction. If the color differences are intrinsic to the stellar population, then
the BCGs with the largest excesses have more hot main sequence stars and therefore may
host more recent bursts than galaxies with lower UV excesses.
– 8 –
Using our measured UV excesses, we can estimate the color of the young stellar popu-
lation in our targets (ranging from -0.7 to 0.5). All but two (MKW4 and MS1358.4+6245,
which are bluer than -0.3) can be explained using either continuous or burst models for
the star formation. Interestingly, objects with colors redder than ∼ 0.1 (about half of the
sample) can only be explained by a burst of star formation occurring 30-200 Myr ago.
6. Multiwavelength Comparisons
Hα measurements for our sample were taken from the literature (specific references
are given in Table 7), and are shown vs. NUV inferred SFR in Figure 8. We note that
Hα measurements are usually based on long-slit estimates, and may miss emission line flux
outside of the slit. Overlaid on the plot is the Kennicutt (1998) Hα-SFR relationship. This
relationship is based on an assumption of constant SFR at ages < 2 × 107years. The fact
that there is general agreement with UV SFRs estimated using similar assumptions suggests
that a recent, constant SFR model provides an adequate description of our targets.
Infrared fluxes were gathered from the literature for 12 of our targets, eight from Spitzer
data and four from IRAS (Table 7). Spitzer fluxes were used to determine total IR luminos-
ity following the method of Quillen et al. (2008), who interpolate a 15µm flux from 8µm and
24µm Spitzer data and then employ the relationship LIR= (11.1+5.5
baz et al. 2002). We note that this relationship will over-estimate the IR luminosity
from star formation if the 24µm point is contaminated with AGN emission. For the four
objects without Spitzer data, IRAS fluxes were converted to total IR luminosity using
LIR∼ 1.7L60 µm(Rowan-Robinson et al. 1997).
Total IR luminosities were then converted to SFRs using equation (5) of Bell (2003)
(as in O’Dea et al. 2008)
ψ(M?yr−1) = A
where A = 1.17 × 10−10for LIR < 1011L? and A = 1.57 × 10−10at higher luminosities.
These relationships were constructed using a sample of galaxies that did not include any
– 9 –
early-types and therefore may not be well suited to our target population. In addition there
is significant expected scatter (at least 50% at 109L?and 25% at 1011L?Bell 2003).
Overall, IR inferred SFRs are a factor of ∼ 10 higher than UV inferred (constant) SFRs,
though our results are somewhat more consistent with recent Herschel estimates (Abell
2597 and ZwCl 3146; Edge et al. 2010). Because the IR luminosities are so large and are
presumably due to dust absorbing UV radiation, the larger inferred IR SFRs potentially
indicate high UV extinction. To further explore the attenuation in our targets, we calculate
their IR excesses (e.g., Gordon et al. 2000):
IRX = log10(Ldust/LUV) (3)
where LUV= νLνand ν = c/1390˚ A.
We compare the IRX and FUV-NUV colors of our sample with Johnson et al. (2007),
who investigate these properties in SDSS galaxies (Figure 9). Overall the colors of the BCGs
in our sample fall into a region that is heavily populated with the colors of galaxies exhibiting
recent star formation. Many of our targets have high values of IRX, consistent with heavily
dust-enshrouded star formation which is very common in rapidly star forming galaxies.
Cluster entropy can be used as a tool for investigating the energy budget of baryons in
clusters (e.g., Ponman, Cannon & Navarro 1999). Thermodynamic entropy is proportional to
the log of the measurable quantity K ≡ TX/n2/3
gas to radiatively cool is also proportional to its density and temperature: tcool∝ T1/2n−1.
In general, if the cooling time is shorter than ∼ 109years, clusters tend to exhibit cool core
characteristics (Hudson et al. 2009). Our sample was chosen to span a range of cooling times
so that we could examine the relationship between these parameters and UV emission.
e . Likewise, the time it takes for intracluster
The X-ray properties of our sample are taken from the ACCEPT database3(Cavagnolo
et al. 2009). As well as using central estimates based on a fit, both entropy (K) and cooling
time profiles were interpolated to obtain values at R=20 kpc from the center of each cluster.
Interpolation enabled a more uniform comparison between objects with different redshifts
and/or data quality.
NUV inferred SFR vs. entropy is shown in Figure 10, and SFR is plotted vs. cluster
– 10 –
cooling time in Figure 11. The plots constructed with central (R→ 0) entropies and cooling
times indicate thresholds comparable to those reported by Cavagnolo et al. (2008), Voit et
al. (2008), and Rafferty, McNamara & Nulsen (2008). Comparisons with gas properties at
R = 20 kpc from the cluster center, however, yield smoother trends in entropy and cooling
time. These plots show a clear tendency for lower entropy, shorter cooling time objects
to exhibit more star formation, providing convincing evidence that the star formation in
these objects is directly related to cooling gas in the cluster cores. A BCES regress fit
between NUV inferred SFR and the R = 20 kpc cooling time data yields the relationship
log10SFRNUV= a log10tcool,20+ b, where a = −3.9 ± 0.7 and b = −0.6 ± 0.1.
7.Summary and Discussion
In our UV study of 17 cool core clusters we find that GALEX easily detects star forma-
tion in cluster BCGs out to z ≥ 0.45 and to unprecedented radii. The BCGs are found to
be bluest in the center, with colors that become increasingly redder with radius, suggesting
that star formation is most easily detected in the central regions.
We construct UV/J band calibration relationships from 17 cluster ellipticals and 22
quiescent BCGs that enable us to subtract the expected UV light from older populations. In
most of the CC clusters studied, we find significant UV luminosity excesses and colors that
strongly suggest recent and/or current star formation.
Star formation rates are estimated using Starburst99 templates for both continuous
and burst models, for easy comparison to results in the literature from other wavebands.
Our findings are corroborated by Hα observations, showing good agreement with Kennicutt
(1998) models of recent, continuous star formation.
To investigate attenuation in CC BCGs, IRX values are calculated using our GALEX
data and IR fluxes from the literature. Comparisons with the SDSS sample of Johnson et al.
(2007) indicate that our sample has moderate-to-high extinction and has NUV-FUV colors
consistent with or bluer than other star-forming galaxies. These results emphasize a need
for additional observations and detailed studies of cluster BCGs, as currently there are no
adequately large samples of quiescent BCGs to provide a sufficient context for our findings.
We also compare our UV results to properties of the intragalactic medium using X-ray
observations. We find clear correlations between UV excess, cluster entropy, and central
cooling time, demonstrating that the star formation is directly and incontrovertibly related
to the cooling gas in these objects.
– 11 –
Support for this work was provided by NASA through GALEX award NNX07AJ38G,
Chandra Archive award AR7-8012X, and LTSA grant NNG05GD82G. This research has
made use of the NASA/IPAC Extragalactic Database (NED) which is operated by the Jet
Propulsion Laboratory, California Institute of Technology, under contract with the National
Aeronautics and Space Administration.
Akritas, M. G., & Bershady, M. A. 1996, ApJ, 470, 706
Bauer, F.E. et al. 2005, MNRAS, 359, 1481
Bell, E. F. 2003, ApJ, 586, 794
Bildfell et al. 2008, MNRAS, 389, 1637
Blanton, E.L. et al. 2001, ApJ, 558, L15
Burns, J. O., Gregory, S. A., & Holman, G. D. 1981, ApJ, 250, 450
Canizares, C.R. et al. 1979, ApJ, 234, 33
Cardelli, J.A., Clayton, G.C., & Mathis, J.S. 1989, ApJ, 345, 245
Cardiel, N., Gorgas, J., & Aragon-Salamanca 1998, MNRAS, 298, 977
Cavagnolo, K. 2008, Ph.D. Thesis
Cavagnolo, K. W., Donahue, M., Voit, G. M. & Sun, M., ApJ, 683, L107
Cavagnolo, K. W. et al. , APJS 182, 12
Cowie, L. L., & Binney, J. 1977, ApJ, 215, 723
Cowie et al. 1893, ApJ, 272, 29
Crawford, C.S. et al. 1999, MNRAS, 306, 857
Donahue, M., Stocke, J.T., & Gioia, I.M. MNRAS 385, 49
Donahue, M. et al. , ApJ670, 231
Donahue, M. et al. , 2010, ApJ, accepted
– 12 –
Edge, A.C. MNRAS 328,762
Edge, A. C. et al. , 2010, astro-ph/1005.1211
Egami, E. et al. , ApJ647, 922
Elbaz, D., Cesarsky, C. J., Chanial, P., Aussel, H., Franceschini, A., Fadda, D.,& Chary, R.
R. 2002, A&A, 384, 848
Fabian, A. C., & Nulsen, P. E. J. 1977, MNRAS, 180, 479
Fabian, A.C. et al. 1986, ApJ, 305, 9
Ge, J. & Owen F.N. 1994, AJ, 108, 1523
Gil de Paz, A. et al. , 2007, ApJS, 173, 185
Golombek, D., Miley, G. K. & Neugebauer, G. 1988, AJ, 95, 26
Gordon, K. D., Clayton, G. C., Witt, A. N., & Misselt, K. A. 2000, ApJ, 533, 236
Heckman, T.M., Baum, S.A., van Breugel, W.J.M., & McCarthy, P. 1989, ApJ, 338, 48
Hicks, A.K. & Mushotzky, R. 2005, ApJ, 635, L9
Hu, E. M., Cowie, L. L., & Wang, Z. 1985, ApJS, 59, 447
Hudson, D. S., Mittal, R., Reiprich, T. H., Nulsen, P. E. J., Andernach, H., & Sarazin, C. L.
Jaffe, W., Bremer, M.N., & Baker, K. 2005, MNRAS, 360, 748
Johnson, B. D., et al. 2006, ApJ, 644, L109
Johnson, B. D., et al. 2007, ApJS, 173, 392
Kaastra, J.S. et al. 2004, A&A, 413, 415
Kennicutt, R.C. 1998, ARA&A, 36, 189
Kereˇ s, D., Katz, N., Fardal, M., Dav´ e, R., & Weinberg, D. H. 2009, MNRAS, 395, 160
Knapp, G. R., Guhathakurta, P., Kim, D. & Jura, M. A. 1989, ApJS, 70, 329
Lamareille, F., Contini, T., Le Borgne, J.-F., Brinchmann, J., Charlot, S., & Richard, J.
2006, A&A, 448, 893
– 13 –
Leitherer, C. et al. 1999, ApJS, 123, 3
Martin, D.C. et al. 2005, ApJ, 619, 1
Mathews, W. G., & Bregman, J. N. 1978, ApJ, 224, 308
McNamara, B.R. & O’Connell, R.W. 1989, AJ, 98, 201
McNamara, B.R. et al. 2000, ApJ, 534, 135
Morrissey et al. 2007, ApJS, 173, 682
Mittaz, J.P.D. et al. 2001, A&A, 365, L93
Mushotzky, R. et al. 1981, ApJ, 244, 47
O’Dea, C. P., Baum, S. A., Mack, J., Koekemoer, A. M., & Laor, A. 2004, ApJ, 612, 131
O’Dea, C.P. et al. 2008, ApJ, 681, 1035
Ostricker, J. & Ciotti, L. 2005, RSPTA, 363, 667
Peres, C.B. et al. 1998, MNRAS, 298, 416
Peterson, J.R., Kahn, S.M., Paerels, F.B.S., Kaastra, J.S., Tamura, T, Bleeker, J.A.M.,
Ferrigno, C., & Jernigan, J.G. 2003, ApJ, 590, 207
Piffaretti, R. & Kaastra, J.S. 2006, A&A, 453, 423
Ponman, T. J., Cannon, D. B., & Navarro, J. F. 1999, Nature, 397, 135
Popesso, P., Biviano, A., B¨ ohringer, H., & Romaniello, M. 2007, A&A, 461, 397
Quillen, A.C. et al. 2008, ApJS, 176, 39
Rafferty, D. A., McNamara, B. R., & Nulsen, P. E. J. 2008, ApJ, 687, 899
Rowan-Robinson, M. et al. 1997, MNRAS, 289, 490
Salome, P. & Combes, F. 2003, A&A, 412, 657
Sparks, W. B., Pringle, J. E., Donahue, M., Carswell, R., Voit, M., Cracraft, M., & Martin,
R. G. 2009, ApJ, 704, L20
Sun, M. 2009, astro-ph/0904.2006
– 14 –
Voit, G. M., Cavagnolo, K. W., Donahue, M., Rafferty, D. A., McNamara, B. R., & Nulsen,
P. E. J. 2008, ApJ, 681, L5
Wang, J., Overzier, R., Kauffmann, G., von der Linden, A., & Kong, X. 2010, MNRAS, 401,
White, S. & Frenk, C. 1991, ApJ, 379, 52
Witt, A. N., & Gordon, K. D. 2000, ApJ, 528, 799
This preprint was prepared with the AAS LATEX macros v5.2.
– 15 –
Fig. 1.— GALEX NUV images with X-ray contours overlaid, shown for a representative
subset of our sample. Note that the UV emission is generally well aligned with the densest
X-ray emitting gas.