1: %\documentclass[12pt,preprint1]{aastex}
2: \documentclass[apj]{emulateapj}
3: \def\lap{\lower.5ex\hbox{$\; \buildrel < \over \sim \;$}}
4: \def\gap{\lower.5ex\hbox{$\; \buildrel > \over \sim \;$}}
5: \def\Msun{\hbox{M$_{\odot}$}}
6: \def\kms{km s$^{-1}$}
7: \def\m8{m$_{8}$}
8: \def\m45{m$_{4.5}$}
9: \usepackage{graphicx}
10: \usepackage{natbib}
11:
12: \begin{document}
13:
14: \shorttitle{Star Formation in the Bullet Cluster}
15: \shortauthors{Chung et al.}
16: \title{\footnote{This paper includes data gathered with the 6.5 meter Magellan Telescopes located at Las Campanas Observatory, Chile.}Impacts of a Supersonic Shock Front on Star Formation in the Bullet Cluster}
17: \author{Sun Mi Chung\altaffilmark{1}, Anthony H. Gonzalez\altaffilmark{1}, Douglas Clowe\altaffilmark{2}, Dennis Zaritsky\altaffilmark{3}, Maxim Markevitch\altaffilmark{4}, Christine Jones\altaffilmark{4}}
18:
19: \altaffiltext{1}{Department of Astronomy, University of Florida, Gainesville, FL 32611-2055; schung@astro.ufl.edu}
20: \altaffiltext{2}{Department of Physics and Astronomy, Ohio University, 251B Clippinger Lab, Athens, OH 45701}
21: \altaffiltext{3}{Steward Observatory, University of Arizona, 933 North Cherry Avenue, Tucson, AZ 85721}
22: \altaffiltext{4}{Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138}
23:
24: \begin{abstract}
25: We use the Bullet Cluster (1E0657-56) to investigate the extent to which star
26: formation in cluster galaxies is influenced by ram pressure from supersonic
27: gas (Mach 3) during a cluster merger. While the effects of ram pressure have
28: been studied for individual galaxies infalling into galaxy clusters, this
29: system provides a unique opportunity to investigate the impact of dramatic
30: merger events on the cluster galaxy population. In this analysis we use {\it
31: Spitzer} IRAC data to study star formation. At the redshift of the cluster the 6.2 $\mu$m PAH feature is redshifted into the 8 $\mu$m band, enabling use of the m$_{4.5}$-m$_{8}$ color as a proxy for specific star formation rate. We find that the color distribution on the two sides of the shock differ by less than 2$\sigma$, and conclude that ram pressure from the shock front has no dramatic, immediate impact on the star formation of cluster galaxies in the Bullet Cluster.
32: \end{abstract}
33:
34: \keywords{galaxies: clusters: individual (1E0657-56, The Bullet Cluster) -- galaxies: evolution}
35:
36: \section{Introduction}
37:
38: It has long been observed that there is a correlation between galaxy
39: morphology and local density \citep{dressler1980}, known as the
40: density-morphology relation. Although there are various physical mechanisms
41: that can transform star-forming late-type galaxies into quiescent early-type
42: galaxies, it is unknown whether this transformation occurs mostly in the
43: galaxy group or galaxy cluster environment.
44:
45: Preprocessing of galaxies in the group environment has been argued by some
46: \citep{zabludoff1998,kodama2001} to be a principal mechanism driving the
47: predominance of the passive galaxy populations observed in galaxy clusters.
48: In galaxy groups, the velocity dispersions are sufficiently low that processes
49: such as galaxy-galaxy interactions can quench star formation and
50: morphologically alter galaxies. In addition to galaxy-galaxy interactions,
51: \citet{kawata2008} show that ``strangulation'' is an important mechanism that
52: leads to quenched star formation in the group environment. Using a
53: cosmological chemodynamical simulation, \citet{kawata2008} show that most of
54: the hot gas in a disk galaxy is stripped away, thus cutting off a source of
55: new cold gas (some of the hot gas cools to form cold gas), which is necessary
56: to maintain star formation.
57:
58: Others have argued that cluster-specific processes are responsible for the
59: high early-type fraction. The simulations of \citet{berrier2008} argue that
60: the transformation of late-type to early-type galaxies is mostly attributed to
61: processes internal to the galaxy cluster environment, rather than the group
62: environment. \citet{berrier2008} find that most of the galaxies in a galaxy
63: cluster are accreted directly from the field, rather than from infalling
64: galaxy groups. In this case, mechanisms such as galaxy harassment
65: \citep{moore1996} and ram pressure stripping \citep{gunn1972} that are most
66: efficient in massive clusters must drive this transformation.
67:
68:
69: Simulations have shown that ram pressure stripping can remove some or all of
70: the interstellar medium (ISM) from late-type galaxies as they enter the
71: cluster potential \citep{roediger2005,mayer2006,bruggen2008}.
72: Studies of galaxies in clusters with observed HI deficiencies or truncated
73: disks support the idea that ram pressure stripping can transform a gas-rich
74: spiral galaxy into an anemic spiral galaxy
75: \cite[e.g.][]{vollmer2001,vollmer2008}, which may eventually evolve into a
76: lenticular \citep[e.g.][but see \citet{boselli2006}]{bekki2002}.
77: Ram pressure has also been shown to trigger bursts of
78: star formation in cluster galaxies \citep{gavazzi2001,gavazzi2003}, and
79: recent simulations \citep{kronberger2008} predict enhanced star
80: formation rates in galaxies undergoing a ram pressure event.
81:
82: While previous studies have examined the impact of ram
83: pressure on individual galaxies falling into the cluster potential \citep{cortese2007,chung2007} the Bullet Cluster provides a unique opportunity to examine the effect of ram pressure induced by cluster mergers. Although galaxies in any cluster environment undergo ram pressure as they travel through the ICM, galaxies in a major merger event such as those we observe in the Bullet Cluster will experience a dramatic enhancement in the pressure since $P\propto V^2$. It is thus possible that
84: this brief transient phase may have a significant impact on the properties of
85: the cluster galaxy population. The Bullet Cluster is an ideal site in which to
86: quantify the importance of such merger-induced ram pressure.
87:
88: In this paper we conduct an initial exploration of the impact of the shock
89: front upon polycyclic aromatic hydrocarbon (PAH) emission, and hence star
90: formation, in cluster galaxies. We analyze Spitzer data taken with the
91: Infrared Array Camera \citep[IRAC;][]{fazio2004}, in conjunction with optical
92: spectra observed with the Inamori Magellan Areal Camera and Spectrograph
93: \citep[IMACS;][]{bigelow1998}.
94:
95:
96: \section{Geometry and Cluster Properties}
97: The Bullet Cluster, at a redshift of $z=0.296$, is a head-on collision between
98: two clusters of unequal mass, which we call the main cluster and subcluster.
99: \citet{markevitch2002} [M02] observed a sharply defined bow shock front,
100: driven by the subcluster and propagating in the gas of the main cluster with a
101: velocity of 4740$^{+710}_{-550}$ \kms~(Mach number $M=3.0^{+0.45}_{-0.35}$)
102: (Markevitch et al. 2008, in preparation). Hydrodynamic simulations showed that
103: the subcluster itself has a lower velocity relative to the main cluster,
104: $\sim2700$ \kms\ (Springel \& Farrar 2007; see also Milosavljevi{\'c} et
105: al. 2007). While the subcluster and the shock front might be expected to move
106: together, the gravity of the subcluster causes the main cluster's gas in front
107: of the shock to flow onto the subcluster. As a result, the shock front has a
108: higher velocity in the reference frame of that gas inflow, which is what is
109: measured in X-rays. \citet{markevitch2004} constrain the inclination angle of
110: the collision, with respect to the plane of the sky, to $i<8^{o}$, based on
111: the Mach number of the shock and subcluster relative line-of-sight velocity
112: ($\sim$600 \kms) \citep{barrena2002}. From the velocity and geometry of the
113: system, it is computed that the subcluster exited the main cluster core
114: $\sim$0.15 Gyr ago.
115:
116: The simple geometry of the Bullet Cluster makes it an ideal system in which to
117: study how galaxy properties such as star formation are affected by ram
118: pressure. Because the merger is nearly in the plane of the sky, the velocity
119: of galaxies with respect to the intracluster medium (ICM) is well constrained,
120: providing a good quantitative measure of the ram pressure exerted by the gas
121: on the galaxies. In addition, the recentness of the collision (0.15 Gyr)
122: allows us to use sensitive indicators of recent star formation, direct and
123: indirect measures of the ionizing flux from O/B stars with lifetimes of 0.1
124: Gyr.
125:
126:
127: \section{Physical Assumptions}
128:
129: The analysis in this paper requires several assumptions related to the star
130: formation rate and the three-space galaxy distributions for the main cluster
131: and subcluster.
132:
133:
134: \subsection{Star Formation Rate}
135: To probe star formation rates, we use the PAH emission at rest wavelength
136: $\lambda=6.2\micron$, which at the cluster redshift is observed in the 8$\micron$
137: IRAC band. Ultra-violet (UV) photons that are produced by young hot stars are known to excite PAH molecules which then emit in the $\lambda=6.2\micron$ and other infrared bands \citep{leger1984,allamandola1985}. More recent studies have shown that the $6.2\micron$ luminosity is correlated with total infrared luminosity \citep{peeters2004,brandl2006}, which is a tracer of star formation rate \citep{kennicutt1998}. However, PAH emission is not a unique signature of star formation, which can also be excited by visible photons \citep{uchida1998,li2002}, as well as by active galactic nuclei (AGN) \citep{freudling2003}.
138: It has also been shown that PAH emission is dependent on metallicity
139: \citep{engelbracht2005,smith2007} and thus not a robust star formation rate indicator for galaxy samples that span a large range in metallicity \citep{boselli2004}.
140:
141:
142: For the Bullet Cluster, both effects are expected to be second order.
143: Our data are negligibly affected by AGN contamination (see \S5), and our
144: sample is dominated by massive galaxies, for which the metallicity
145: variations are not dramatic. Moreover, the metallicity distribution
146: should be the same for galaxies on both sides of the shock. The PAH flux
147: should nevertheless be considered a rough proxy for star formation rate
148: rather than a precision tracer.
149:
150:
151:
152: To normalize the 8$\micron$ flux, we use the 4.5$\micron$ flux as a proxy for mass. The 4.5$\micron$ flux originates
153: mostly from the old red giant stars, and probes the stellar mass of a galaxy
154: at the cluster redshift. By using the ratio of 8$\micron$ to 4.5$\micron$
155: flux or m$_{4.5}$-m$_{8}$ color, we can roughly trace star formation rate per
156: stellar mass. Specifically, $8\mu$m emission in excess of that expected from
157: the Rayleigh-Jeans tail of the cold stellar component is assumed to arise from
158: star-formation induced PAH emission. We estimate the level of impact that the
159: M$\sim$3 shock has had on star formation activity in individual galaxies by
160: looking at the m$_{4.5}$-m$_{8}$ color as a function of distance from the
161: shock front.
162:
163:
164:
165: \subsection{Galaxy Distribution and Projection}
166: We proceed with our analysis under the simplifying assumption that galaxies
167: that lie ahead of the shock front are main cluster ``pre-shock'' galaxies, and
168: those behind the shock front are main cluster ``post-shock'' galaxies
169: (Figure~\ref{fig:fig1}). The post-shock galaxies have already experienced the
170: ram pressure exerted by the gas behind the shock, whereas the pre-shock
171: galaxies have not yet been affected. The true physical situation is clearly
172: more complex, with contamination from the subcluster galaxies as well as
173: background/foreground sources not associated with the Bullet Cluster.
174: Although there is some contamination from the subcluster galaxies, we assume
175: that the galaxy populations on both sides of the shock front are dominated by
176: main cluster galaxies based on estimates of the total stellar mass. The
177: stellar mass of the main cluster is approximately an order of magnitude larger
178: than the stellar mass of the subcluster.
179:
180: We also note that any effect of contamination from the subcluster galaxies
181: will be strongest at the shock front, which bisects the subcluster brightest
182: group galaxy. Subcluster galaxies just ahead of the shock front experience a
183: strong ram pressure as they collide with the main cluster ICM at $\sim$4700
184: \kms, whereas subcluster galaxies just behind the shock are somewhat shielded
185: by the shock front and feel a much weaker ram pressure with a slower gas
186: inflow velocity of $\sim$1600 \kms. Therefore any effect of ram pressure on
187: star formation rate has the opposite sign for subcluster galaxies near the
188: shock front, compared to the main cluster galaxies.
189:
190:
191: We assume that there is a sharp division between post-shock and pre-shock galaxies, where all galaxies ``behind'' the shock front (leftward of the curve in Figure \ref{fig:fig1})
192: have experienced the effects of the shock. However, because of the projection of galaxies and curvature of the shock front surface along the line of sight, there is a gradual transition from post-shock to pre-shock galaxies.
193: This projection effect softens any step-like increase/decrease in color (or star formation rate) across the shock boundary.
194: By generating galaxies in a random distribution according to the Navarro, Frenk, and White (NFW) \citep{NFW1997} profile, and assuming the shock front is an axisymmetric cone projected in the plane of the sky and spanning $\sim$550 kpc from the axis of symmetry, we estimate that the transition region during which the observed galaxies go from being fully post-shock to fully pre-shock galaxies is $\sim$0.5 Mpc.
195:
196:
197:
198:
199: \section{Observations}
200: \subsection{IRAC}
201:
202: Observations of 1E0657-56 were taken on December 17, 2004, with Spitzer/IRAC.
203: Data were collected from all four IRAC channels - 3.6$\micron$, 4.5$\micron$,
204: 5.8$\micron$, and 8$\micron$, in full array readout mode. A medium scale,
205: cycling dither pattern was used, with 20 pointings. At 2 frames per pointing
206: and 100 s exposure time per frame, the total integration time per IRAC band
207: was 4000 s. The native IRAC pixel scale is 1.22 arcsec, but the final reduced
208: images were set to have a pixel scale of 0.86 arcsec pixel$^{-1}$, with an
209: effective field of view (FOV) of 3.7 $\times$ 3.7 arcmin covered by the four
210: channels. The frames were processed using the Spitzer Science Center (SSC)
211: IRAC Pipeline, and mosaics created from the basic calibrated data (BCD) frames
212: using \emph{MOPEX}.
213:
214: \subsection{IMACS}
215: We analyzed optical spectroscopic data taken with Magellan/IMACS in 2005 and
216: 2006. In 2005, 430 objects were observed with four masks, with exposure times
217: ranging from 4.8 to 5.4 ks. During this run, red sequence galaxies were given
218: highest priority as targets. The following year, two additional masks were
219: obtained, with exposure times of 10.8 and 12.0 ks. Target priority was given
220: to objects with relatively red m$_{4.5}$-m$_{8}$ colors. The wavelength range
221: of all spectra are 4000 \AA$-$9000 \AA, with a dispersion of 2 \AA\ per
222: pixel. We note that spectroscopic selection criteria for galaxies on both
223: sides of the shock front are identical.
224:
225:
226: \section{Data Analysis}
227: \subsection{IRAC Photometry}
228:
229: Source detection and photometry of the IRAC images are performed with Source
230: Extractor (Bertin \& Arnouts 1996). Sources are identified in the 4.5$\mu$m
231: image with a 3.5$\sigma$ detection threshold. Magnitudes are measured within a
232: 5$\arcsec$ aperture. We choose to apply no aperture corrections to the
233: photometry. In this analysis we are primarily interested in colors rather
234: than total magnitudes, which for m$_{8}$-m$_{5.8}$ and m$_{4.5}$-m$_{3.6}$
235: would change by only $\sim$0.05 magnitudes with the corrections applied.
236: Since the corrections themselves are uncertain by up to 10\%
237: (http://ssc.spitzer.caltech.edu/irac/calib/extcal), excluding systematics, we
238: use raw aperture magnitudes for all analyses. In addition, aperture
239: corrections are expected to be minor when examining colors rather than
240: absolute magnitudes.
241:
242: \subsection{IRAC Sample Selection}
243:
244: The curvature of the shock contour indicates that the direction of propagation
245: lies within $\sim$5 degrees of the east-west axis (M02). The shock front was
246: defined as the X-ray brightness contour whose most westward point lies at
247: 104.6 degrees in right ascension (Figure~\ref{fig:fig1}), with a maximum and
248: minimum declination of -55.908 and -55.985 degrees. Objects located above or
249: below (north or south) the shock front were not included in our analyses,
250: since these sources are not in the direct path of the shock front and thus
251: should be minimally affected by the propagating shock. We limit our analyses
252: to the region of spatial overlap between the four IRAC bands, detecting a
253: total of 758 sources within this ``analysis region''.
254:
255: We eliminate unsaturated stars from the IRAC catalog using Magellan optical photometry \citep{clowe2006}. In the inner region of the R-band image
256: where the point spread function (PSF) is small ($\sim$0.45 arcsec FWHM),
257: objects with a half-light radius of 2.6 to 3.2 pixels (0.111 arcsec/pixel) and
258: an R-band aperture magnitude of 18 to 24.1 (with a 7.8 pixel aperture), are
259: flagged as stars. Due to a gradient of the PSF in the image, half-light
260: radius and aperture magnitude cuts are scaled accordingly to provide a catalog
261: of unsaturated stars. Saturated stars are identified manually by inspecting
262: the Magellan R-band image to look for saturated objects.
263:
264:
265: To minimize contamination from bright foreground galaxies, we impose a
266: magnitude limit of m$_{4.5} > 14.20$ -- the magnitude of the brightest cluster
267: galaxy (BCG) in the main cluster. We also apply a magnitude cut of m$_{4.5} <
268: 17$, which is sufficiently faint to detect most of the 8$\micron$ sources.
269: While going deeper to m$_{4.5} < 18$ increases the number of cluster
270: candidates, we opted for a more conservative m$_{4.5} < 17$ magnitude limit so
271: as to minimize the background contamination. Most importantly, the analyses
272: described in this paper produce the same result within error bars, whether we
273: apply a magnitude limit of m$_{4.5} < 17$ or m$_{4.5} < 18$.
274:
275: Of the 758 sources, 18 are brighter than the main cluster BCG, 176 are stars,
276: leaving us with 564 sources. Eliminating sources fainter than m$_{4.5}=17.0$,
277: we are left with 169 sources in the direct path of the shock front and 31 that
278: are above/below the shock front. The distance from likely star-forming
279: galaxies to the shock front is illustrated by arrows in Figure~\ref{fig:fig1}.
280:
281:
282: \subsection{IMACS Spectroscopy}
283: To reduce contamination from background and foreground sources, we reduce a
284: total of 458 IMACS spectra with IRAF and IDL, then cross-correlate them with
285: template spectra, using the IRAF task \emph{xcsao} to obtain redshifts. We
286: use the same templates as \citet{tran2005}, each spectrum representing a
287: different type of galaxy -- giant elliptical, E+A, Sb spiral, and an emission
288: line galaxy (ELG). Of the 458 spectra, we obtain reliable redshifts for 326
289: sources. We augment our spectroscopic data set with 133 redshifts
290: \citep[][priv. comm]{barrena2002}, which more densely sample the cluster core,
291: yielding a total of 459 sources with known redshifts.
292:
293: We next associate the spectroscopic redshifts with IRAC sources for the 114
294: spectra that lie within the IRAC field-of-view. Of these 114 sources, 67 are
295: cluster members with redshifts that lie within $\pm 2000$ \kms~of the mean
296: cluster redshift ($z=0.296$), and 47 are interlopers. Sixty-three out of the
297: 67 cluster members, and 42 out of the 47 interlopers, meet the m$_{4.5} < 17$
298: criteria. This leaves us with 134 ``cluster candidates'', which include all
299: IRAC sources that meet our magnitude criteria, are located within the direct
300: path of the shock front, and exclude all known interlopers.
301:
302:
303: \section{Results and Discussion}
304:
305: \subsection{Impact of the Shock}
306:
307: We use the m$_{4.5}$-m$_{8}$ color to probe relative star formation rates per
308: unit stellar mass. At the redshift of the cluster ($z=0.296$), the prominent
309: PAH emission band at rest wavelength 6.2$\mu$m falls into the 8.0$\mu$m IRAC
310: band. PAH emission, which is excited primarily by UV photons, most often
311: originates from photodissociation regions of star formation sites
312: \cite[e.g.][]{Howell2007}. Although PAH emission is sometimes also associated
313: with AGN \citep{freudling2003}, only 2 out of 134 cluster candidates in the
314: direct path of the shock front have IRAC colors within the AGN wedge
315: \citep{lacy2004,stern2005}. Both AGN candidates have m$_{4.5}$-m$_{8} < 0.6$,
316: and thus do not contribute spurious star formation signatures.
317:
318: Figure~\ref{fig:fig2} shows color as a function of distance from the shock
319: front for 134 cluster candidates. The dotted horizontal lines show the
320: expected color of an elliptical and Sbc type galaxy at z=0.3
321: \citep{assef2007}. These predicted colors come from a set of low-resolution
322: spectral templates, derived by \citet{assef2007}, using optical and
323: near-infrared photometry of over 16,000 galaxies in the NOAO Deep Wide-Field
324: Survey Bo\"otes region, and redshifts from the AGN and Galaxy Evolution Survey
325: (AGES).
326: For elliptical galaxies the 8$\mu$m emission comes from the Rayleigh-Jeans
327: tail of the stellar component, for which the m$_{4.5}$-m$_{8}\sim-0.1$ at z=0.3
328: \citep{assef2007}. Colors redder than the elliptical locus are indicative of
329: PAH emission in the 8$\mu$m band.
330:
331: A majority of the IRAC sources have colors consistent with red sequence
332: galaxies in the cluster -- only 17\% of the cluster candidates have a color
333: m$_{4.5}$-m$_{8}>0.5$, and 6\% of the sources have m$_{4.5}$-m$_{8}>1$.
334: Although there is a selection effect in favor of red sequence galaxies with
335: the spectroscopic data, Figure~\ref{fig:fig2} uses the entire IRAC sample, for
336: which there is no bias towards quiescent ellipticals.
337:
338: The spectroscopic data
339: is useful for eliminating interlopers, but the sample of confirmed cluster
340: members (solid points in Figure \ref{fig:fig2} \& Figure \ref{fig:fig3})
341: is of limited value in the
342: current analysis because the spectroscopic program preferentially targeted red
343: sequence galaxies. Indeed, only 2 of the 23 cluster candidates with
344: m$_{4.5}$-m$_{8}>0.5$, presently have spectra confirming their membership.
345:
346:
347: Instead, we focus in the current analysis on the full sample of 134 candidate
348: members and assess whether there is evidence for a correlation between the
349: luminosity-weighted color of cluster members and their distance from the shock
350: front. For each galaxy we compute the distance from the shock (Figure
351: \ref{fig:fig1}), and then compute the integrated color of all galaxies in 275
352: kpc wide bins (Figure \ref{fig:fig3}). For each bin we use the bootstrap
353: technique with 10,000 realizations to calculate the uncertainties. The dotted
354: lines correspond to the total integrated color of all objects on each side of
355: the shock front, with the width of the shaded grey region corresponding to the
356: 1$\sigma$ confidence interval calculated via the bootstrap method.
357:
358: For the individual data points on Figure~\ref{fig:fig3}, the size of the
359: bootstrap error bars indicates that the mean color is strongly sensitive to
360: the small subset of strongly star-forming galaxies. For example, in the bin
361: directly ahead of the shock front, there are 2 (out of 22) galaxies that are
362: responsible for the relatively high integrated color. One of these two
363: galaxies is a spectroscopically confirmed, face-on spiral; the other is a
364: large disk galaxy whose optical color is bluer than the red sequence.
365:
366: The horizontal dotted-dashed line in Figure~\ref{fig:fig3} shows the predicted color of an elliptical galaxy at $z=0.3$ \citep{assef2007}. The data shown in Figure~\ref{fig:fig3} are meant to show an excess of color in the post-shock and pre-shock galaxies, in comparison to what one would expect from a purely quiescent elliptical galaxy.
367: Overall, Figure~\ref{fig:fig3} shows that there is no drastic change in color across the shock boundary. The integrated color of the post-shock versus pre-shock galaxies is consistent to within 2$\sigma$.
368:
369:
370: The above technique is a rather blunt means of assessing the impact of the
371: shock front, given that it includes the red sequence galaxies, where we expect
372: minimal change, as well as the star-forming galaxies. An alternate approach
373: is to use the Kolmogorov-Smirnoff (KS) test to understand how statistically
374: different the populations are on either side of the shock.
375: We first compare the full distributions on the two sides of the shock front.
376: Using all 134 candidate members, a KS-test indicates that the colors on the two sides of the shock differ at only the 1.3$\sigma$ level; consequently, the data do not exclude the hypothesis that the galaxies are all drawn from the same population.
377: We note that the full sample includes elliptical galaxies, whose m$_{4.5}$-m$_{8}$ color simply reflects the cold stellar component of the Rayleigh-Jeans tail rather than star formation. All galaxies are included in our initial KS-test because we are looking for changes in the global color distribution of galaxy colors across the shock front.
378:
379: To avoid the gas-poor elliptical galaxies whose m$_{4.5}$-m$_{8}$ color is not a good proxy for specific star formation rate, we can exclusively use galaxies whose m$_{4.5}$-m$_{8}$ color is redder than some value. Comparing the color distribution across the shock front for galaxies with m$_{4.5}$-m$_{8} > 0.2$, m$_{4.5}$-m$_{8} > 0.5$, and m$_{4.5}$-m$_{8} > 1$, the KS-test indicates that the colors on the two sides of the shock front differ at the 0.8$\sigma$, 1.5$\sigma$, and 1.2$\sigma$ level, respectively. Although we avoid the elliptical galaxies by imposing color cuts, we also introduce an uncertainty since objects may move in and out of the sample due to the impact of the shock front, without leaving a known trace on the final color distribution. However, we emphasize that the KS-test for all of our color cuts shows no evidence for significant change in the color distribution from pre-shock to post-shock galaxies.
380:
381: For comparison, we also investigate the optical properties of our IRAC
382: sample. Figure~\ref{fig:optical} presents the color-magnitude relation (CMR)
383: for the 88 objects out of the 134 IRAC cluster candidates with $R<20.5$, using our Magellan data and again excluding spectroscopically confirmed interlopers.
384: As indicated by the IRAC colors, Figure~\ref{fig:optical} confirms that most of the IRAC sources are red sequence galaxies.
385: However, to determine exactly which objects can be classified as part of the red sequence, we need three parameters of the CMR -- the slope, zero-point, and dispersion.
386:
387:
388: We adopt a CMR slope of -0.076, derived from 57 X-ray clusters examined by \citet{lopez-cruz2004}, with redshifts ranging from 0.02$\leq$ z $\leq$0.18.
389: To obtain the zero-point of the CMR relation, we fit the data with this fixed slope. This fit yields a zero-point of 3.927, which is $\sim$0.1 mag brighter than the expected zero-point calculated from the zero-point-redshift relation of \citet{lopez-cruz2004}. Finally, we calculate the dispersion by fitting a Gaussian function to the CMR
390: residuals, excluding outliers. We derive $\sigma=0.136$ mag, indicating
391: that our dispersion is indeed dominated by photometric uncertainty.
392:
393:
394: We classify all objects within 2$\sigma$ of the CMR as red sequence galaxies, indicated by the dotted lines in Figure~\ref{fig:optical}. Out of the 7 objects with m$_{4.5}$-m$_{8}>0.5$ and $R<20.5$, 5 of them lie more than 2$\sigma$ below the CMR (with one object just barely within 2$\sigma$). These red IRAC sources consitute 5/12 or 42\% of the objects below the 2$\sigma$ line. The optical data thus confirm that objects with red m$_{4.5}$-m$_{8}$ colors correspond to optically blue and likely star-forming galaxies.
395:
396:
397: There are a few caveats in interpreting the results discussed above. Foremost, there remains contamination by foreground and background sources for the galaxies that lack spectroscopic confirmation. Foreground interlopers among the star-forming galaxies may depress the observed significance of any color change across the shock front. It is unlikely however that removing these foregrounds will create a significant difference in the mean m$_{4.5}$-m$_{8}$ color across the shock boundary. A total of 55\% of IRAC sources with color
398: m$_{4.5}$-m$_{8}>0.5$ are known interlopers, while the fraction of known
399: interlopers is only 22\% for sources with m$_{4.5}$-m$_{8}<0.5$. Because the
400: fraction of interlopers is higher for late-type galaxies than early-type
401: galaxies, eliminating these objects will likely cause the mean color on both
402: sides of the shock front (Figure~\ref{fig:fig3}) to approach the quiescent
403: value.
404:
405:
406: \subsection{Physical Interpretation}
407:
408: The IRAC data show no significant change in color across the shock boundary
409: (Figure~\ref{fig:fig3}), indicating that the gas behind the shock front does
410: not have a large impact on the current or recent star formation activity in
411: the individual galaxies.
412: One possible explanation is that ram pressure
413: stripping may deplete existing gas reservoirs in the outskirts of galaxies
414: without disturbing current/recent star formation sites that may be more
415: centrally located in the disk. Following \citet{gunn1972}, a galaxy moving
416: face-on through an ICM experiences a ram pressure of $P=\rho V^{2}$. If ram
417: pressure exceeds the galaxy's internal gravitational pressure, ram pressure
418: stripping occurs. The gravitational pressure can be expressed as $P=2\pi
419: Gx\frac{M_{star}^2}{r_{g}^4}$, where $G$ is the gravitational constant, $x$ is
420: the interstellar gas to stellar mass ratio, $M_{star}$ is the stellar mass,
421: and $r_{g}$ is the galaxy radius. A galaxy with stellar mass
422: $M_{\star}=10^{10} M_{\odot}$ and $x=0.6$, should experience ram pressure
423: stripping at a distance of $\sim$4 kpc from its center, assuming that it moves
424: face-on through an ICM of density $4\times 10^{-3}$ cm$^{-3}$, at a velocity
425: of $\sim$2700 \kms. This is somewhat beyond the expected disk scale length for
426: a galaxy with stellar mass $M_{\star}=10^{10} M_{\odot}$ \citep{pizagno2005}.
427: Therefore, we might not expect to see evidence of disturbed star formation due
428: to ram pressure, since much of the current star formation occurs within the
429: disk scale length. However, the gas density and galaxy velocity (with respect
430: to the ICM) are highly uncertain at any significant distance from the shock
431: front, since both quanities change with time and position in the cluster.
432:
433:
434: If ram pressure stripping destroys gas reservoirs in galaxies that host
435: significant star formation, we might expect to see an excess of post-starburst
436: galaxies in $\sim$10 Myr (after a galaxy crosses the shock front), once the O
437: stars no longer significantly contribute to the stellar population. A
438: timescale of 10 Myr corresponds roughly to 75 kpc in Figure~\ref{fig:fig3}.
439: On such short timescales, it is difficult to detect an excess of any
440: particular galaxy type because of small number statistics (i.e.--there are not
441: enough galaxies within a single 75 kpc bin). Also, post-starburst galaxies are difficult to distinguish from quiescent elliptical galaxies with photometry, since both populations lie near the optical CMR \citep{tran2007}.
442:
443:
444: However, the lifetime of a post-starburst galaxy is $\sim$1 Gyr
445: \cite[e.g.][]{quintero2004}, providing us with a longer timescale (or distance
446: from the shock front) in which we can search for an excess of these galaxy
447: types. In future work, we will spectroscopically identify the E+A galaxies in
448: our sample. A higher fraction of E+A galaxies behind the shock front versus
449: ahead of the shock front, coupled with no change in current star formation
450: rates across the shock boundary, would favor ram pressure stripping of gas
451: reservoirs, rather than some effect on current star formation.
452:
453:
454: Although Figure~\ref{fig:fig3} shows that there is no significant change in
455: color in the pre-shock versus post-shock galaxies, the geometric effects and
456: contamination from non-cluster sources may dilute the signal. Future work
457: includes analysis of IRAC and MIPS 24$\mu$m data observed over a wider area,
458: which coupled with spectroscopic redshifts will be less
459: sensitive to geometric effects and contamination. However, even with our
460: current level of contamination, it is clear that the ram pressure exerted from
461: the supersonic gas does not have a dramatic impact on current star formation
462: activity in the main cluster galaxies.
463:
464: Another source of signal dilution is the large fraction of early-type galaxies
465: in our sample, for which there are two possible explanations. First, galaxy
466: preprocessing in the group environment may have already done its work on both
467: the cluster and subcluster galaxies, explaining the observed lack of late-type
468: galaxies \citep{zabludoff1998, cortese2006}. Second, the galaxies we observe
469: may have been affected by the cluster environment in such a manner that they
470: have already converted their gas reservoir into stars. \citet{marcillac2007}
471: and \citet{bai2007} have investigated star formation of galaxies in merging
472: clusters at $z\sim$0.83. They find evidence of triggered star formation in
473: infalling galaxies and galaxy groups, which coupled with mechanisms such as
474: ram pressure stripping or galaxy harassment, can act to eventually quench star
475: formation. In this scenario, many of the initially gas-rich, star forming
476: galaxies in the main cluster would have consumed much of their available gas
477: in an earlier epoch of triggered star formation, explaining the lack of
478: late-type galaxies in our current sample. However, the relative lack of
479: late-type galaxies cannot explain why we do not see a signal across the shock
480: front, especially considering that even among the late-type population, we do
481: not detect a significant difference in color in the post-shock versus
482: pre-shock galaxies.
483:
484: \section{Summary}
485: To trace specific star formation rate, we use IRAC color m$_{4.5}$-m$_{8}$.
486: The 8$\micron$ flux traces star formation via the PAH emission feature at rest
487: wavelength 6.2$\micron$, which is redshifted into the 8$\micron$ band at the
488: cluster redshift. The 4.5$\micron$ flux can be taken as a proxy for stellar
489: mass.
490:
491: Using spectroscopic redshifts, we confirmed 63 out of 200 IRAC sources (169 in
492: direct path of shock front + 31 above/below the shock front) as cluster
493: members, and found 42 IRAC sources that are interlopers. Applying a magnitude
494: cut at m$_{4.5}=17$, we are left with 158 cluster candidates, of which 134 are
495: in the direct path of the shock front.
496:
497: We observe no significant trend in m$_{4.5}$-m$_{8}$ as a function of distance
498: from the shock front. The KS-test reveals that the sample of galaxies behind the shock versus ahead of the shock front only differs in color by less than
499: 1.5$\sigma$, for galaxies with various cuts in m$_{4.5}$-m$_{8}$ color. This
500: indicates that even in a dramatic merger event like the Bullet Cluster, the
501: ram pressure induced from a supersonic collision does not drastically trigger
502: or quench current/recent star formation. A possible explanation is that the
503: ram pressure can remove the gas reservoirs in the outer disk, while preserving
504: star formation in the central disk of a galaxy. While geometric effects,
505: contamination from interlopers, and the lack of late-type galaxies in the
506: Bullet Cluster could contribute to diluting a signal, they are unlikely to
507: explain the complete lack of signal we observe. Even among our small sample
508: of late-type galaxies, we detect no change in color between the pre-shock and
509: post-shock galaxies to within $1.5\sigma$.
510:
511: \acknowledgements The authors would like to thank Rafael Barrena for sharing
512: his redshift data with us. We also acknowledge support for this work from
513: NASA/Spitzer grant 1319141. MM was supported by NASA contract NAS8-39073 and
514: Chandra grant GO8-9128X. Christine Jones acknowledges support from Spitzer
515: Contract 1265584.
516:
517: {\it Facilities:} \facility{Spitzer (IRAC)}, \facility{CXO (ACIS-I)}, \facility{Magellan:Baade (IMACS)}
518:
519:
520:
521:
522: \clearpage
523: \newpage
524: \begin{thebibliography}{1}
525:
526: \bibitem[Allamandola et al.(1985)]{allamandola1985} Allamandola, L.~J.,
527: Tielens, A.~G.~G.~M., \& Barker, J.~R.\ 1985, \apjl, 290, L25
528:
529: \bibitem[Assef et al.(2008)]{assef2007} Assef, R.~J., et al.\
530: 2008, \apj, 676, 286
531:
532: \bibitem[Bai et al.(2007)]{bai2007} Bai, L., et al.\ 2007,
533: \apj, 664, 181
534:
535:
536: \bibitem[Barrena et al.(2002)]{barrena2002} Barrena, R., Biviano,
537: A., Ramella, M., Falco, E.~E., \& Seitz, S.\ 2002, \aap, 386, 816
538:
539: \bibitem[Bekki et al.(2002)]{bekki2002} Bekki, K., Couch, W.~J.,
540: \& Shioya, Y.\ 2002, \apj, 577, 651
541:
542: \bibitem[Berrier et al.(2008)]{berrier2008} Berrier, J.~C.,
543: Stewart, K.~R., Bullock, J.~S., Purcell, C.~W., Barton, E.~J.,
544: \& Wechsler, R.~H.\ 2008, ArXiv e-prints, 804, arXiv:0804.0426
545:
546: \bibitem[Boselli et
547: al.(2004)]{boselli2004} Boselli, A., Lequeux, J., \& Gavazzi, G.\ 2004, \aap, 428, 409
548:
549: \bibitem[Boselli
550: \& Gavazzi(2006)]{boselli2006} Boselli, A., \& Gavazzi, G.\ 2006, \pasp, 118, 517
551:
552: \bibitem[Bertin \& Arnouts(1996)]{bertin1996} Bertin, E., \&
553: Arnouts, S.\ 1996, \aaps, 117, 393
554:
555:
556: \bibitem[Bigelow et al.(1998)]{bigelow1998} Bigelow, B.~C.,
557: Dressler, A.~M., Shectman, S.~A., \& Epps, H.~W.\ 1998, \procspie, 3355,
558: 225
559:
560: \bibitem[Brandl et al.(2006)]{brandl2006} Brandl, B.~R., et al.\
561: 2006, \apj, 653, 1129
562:
563: \bibitem[Br{\"u}ggen
564: \& De Lucia(2008)]{bruggen2008} Br{\"u}ggen, M., \& De Lucia, G.\ 2008, \mnras, 383, 1336
565:
566:
567:
568: \bibitem[Chung et al.(2007)]{chung2007} Chung, A., van Gorkom,
569: J.~H., Kenney, J.~D.~P., \& Vollmer, B.\ 2007, \apjl, 659, L115
570:
571: \bibitem[Clowe et al.(2006)]{clowe2006} Clowe, D., Brada{\v c},
572: M., Gonzalez, A.~H., Markevitch, M., Randall, S.~W., Jones, C.,
573: \& Zaritsky, D.\ 2006, \apjl, 648, L109
574:
575: \bibitem[Cortese et al.(2006)]{cortese2006} Cortese, L., Gavazzi, G., Boselli, A., Franzetti, P., Kennicutt, R.~C., O'Neil, K., \& Sakai, S.\ 2006, \aap, 453, 847
576:
577: \bibitem[Cortese et al.(2007)]{cortese2007} Cortese, L., et al.\
578: 2007, \mnras, 376, 157
579:
580: \bibitem[Dressler(1980)]{dressler1980} Dressler, A.\ 1980, \apj,
581: 236, 351
582:
583: \bibitem[Engelbracht et al.(2005)]{engelbracht2005} Engelbracht, C.~W.,
584: Gordon, K.~D., Rieke, G.~H., Werner, M.~W., Dale, D.~A.,
585: \& Latter, W.~B.\ 2005, \apjl, 628, L29
586:
587: \bibitem[Freudling et al.(2003)]{freudling2003} Freudling, W.,
588: Siebenmorgen, R., \& Haas, M.\ 2003, \apjl, 599, L13
589:
590: \bibitem[Gavazzi et al.(2001)]{gavazzi2001} Gavazzi, G., Boselli,
591: A., Mayer, L., Iglesias-Paramo, J., V{\'{\i}}lchez, J.~M.,
592: \& Carrasco, L.\ 2001, \apjl, 563, L23
593:
594: \bibitem[Gavazzi et al.(2003)]{gavazzi2003} Gavazzi, G., Cortese,
595: L., Boselli, A., Iglesias-Paramo, J., V{\'{\i}}lchez, J.~M., \& Carrasco,
596: L.\ 2003, \apj, 597, 210
597:
598:
599: \bibitem[Gunn
600: \& Gott(1972)]{gunn1972} Gunn, J.~E., \& Gott, J.~R.~I.\ 1972, \apj, 176, 1
601:
602: \bibitem[Howell et al.(2007)]{Howell2007} Howell, J.~H., et al.\
603: 2007, \aj, 134, 2086
604:
605:
606: \bibitem[Fazio et al.(2004)]{fazio2004} Fazio, G.~G., et al.\
607: 2004, \apjs, 154, 10
608:
609: \bibitem[Kawata
610: \& Mulchaey(2008)]{kawata2008} Kawata, D., \& Mulchaey, J.~S.\ 2008, \apjl, 672, L103
611:
612: \bibitem[Kennicutt(1998)]{kennicutt1998} Kennicutt, R.~C., Jr.\ 1998, \araa, 36, 189
613:
614: \bibitem[Kodama et al.(2001)]{kodama2001} Kodama, T., Smail, I.,
615: Nakata, F., Okamura, S., \& Bower, R.~G.\ 2001, \apjl, 562, L9
616:
617: \bibitem[Kronberger et
618: al.(2008)]{kronberger2008} Kronberger, T., Kapferer, W., Ferrari, C., Unterguggenberger, S., \& Schindler, S.\ 2008, \aap, 481, 337
619:
620: \bibitem[Lacy et al.(2004)]{lacy2004} Lacy, M., et al.\ 2004,
621: \apjs, 154, 166
622:
623: \bibitem[Leger
624: \& Puget(1984)]{leger1984} Leger, A., \& Puget, J.~L.\ 1984, \aap, 137, L5
625:
626: \bibitem[Li
627: \& Draine(2002)]{li2002} Li, A., \& Draine, B.~T.\ 2002, \apj, 572, 232
628:
629: \bibitem[L{\'o}pez-Cruz et al.(2004)]{lopez-cruz2004} L{\'o}pez-Cruz,
630: O., Barkhouse, W.~A., \& Yee, H.~K.~C.\ 2004, \apj, 614, 679
631:
632: \bibitem[Marcillac et al.(2007)]{marcillac2007} Marcillac, D., Rigby,
633: J.~R., Rieke, G.~H., \& Kelly, D.~M.\ 2007, \apj, 654, 825
634:
635:
636: \bibitem[Markevitch et al.(2002)]{markevitch2002} Markevitch, M.,
637: Gonzalez, A.~H., David, L., Vikhlinin, A., Murray, S., Forman, W., Jones,
638: C., \& Tucker, W.\ 2002, \apjl, 567, L27
639:
640: \bibitem[Markevitch et al.(2004)]{markevitch2004} Markevitch, M.,
641: Gonzalez, A.~H., Clowe, D., Vikhlinin, A., Forman, W., Jones, C., Murray,
642: S., \& Tucker, W.\ 2004, \apj, 606, 819
643:
644: \bibitem[Mayer et al.(2006)]{mayer2006} Mayer, L., Mastropietro,
645: C., Wadsley, J., Stadel, J., \& Moore, B.\ 2006, \mnras, 369, 1021
646:
647:
648: \bibitem[Milosavljevi{\'c} et al.(2007)]{milo2007}
649: Milosavljevi{\'c}, M., Koda, J., Nagai, D., Nakar, E.,
650: \& Shapiro, P.~R.\ 2007, \apjl, 661, L131
651:
652: \bibitem[Moore et al.(1996)]{moore1996} Moore, B., Katz, N.,
653: Lake, G., Dressler, A., \& Oemler, A.\ 1996, \nat, 379, 613
654:
655: \bibitem[Navarro et al.(1997)]{NFW1997} Navarro, J.~F., Frenk,
656: C.~S., \& White, S.~D.~M.\ 1997, \apj, 490, 493
657:
658: \bibitem[Peeters et al.(2004)]{peeters2004} Peeters, E., Spoon,
659: H.~W.~W., \& Tielens, A.~G.~G.~M.\ 2004, \apj, 613, 986
660:
661:
662: \bibitem[Pizagno et al.(2005)]{pizagno2005} Pizagno, J., et al.\
663: 2005, \apj, 633, 844
664:
665:
666: \bibitem[Quintero et al.(2004)]{quintero2004} Quintero, A.~D., et
667: al.\ 2004, \apj, 602, 190
668:
669:
670: \bibitem[Roediger
671: \& Hensler(2005)]{roediger2005} Roediger, E., \& Hensler, G.\ 2005, \aap, 433, 875
672:
673:
674: \bibitem[Smith et al.(2007)]{smith2007} Smith, J.~D.~T., et al.\
675: 2007, \apj, 656, 770
676:
677: \bibitem[Springel \& Farrar(2007)]{springel2007} Springel, V., \&
678: Farrar, G.~R.\ 2007, \mnras, 380, 911
679:
680: \bibitem[Stern et al.(2005)]{stern2005} Stern, D., et al.\ 2005,
681: \apj, 631, 163
682:
683: \bibitem[Tran et al.(2005)]{tran2005} Tran, K.-V.~H., van
684: Dokkum, P., Illingworth, G.~D., Kelson, D., Gonzalez, A., \& Franx, M.\
685: 2005, \apj, 619, 134
686:
687: \bibitem[Tran et al.(2007)]{tran2007} Tran, K.-V.~H., Franx, M.,
688: Illingworth, G.~D., van Dokkum, P., Kelson, D.~D., Blakeslee, J.~P.,
689: \& Postman, M.\ 2007, \apj, 661, 750
690:
691:
692: \bibitem[Uchida et al.(1998)]{uchida1998} Uchida, K.~I., Sellgren,
693: K., \& Werner, M.\ 1998, \apjl, 493, L109
694:
695:
696:
697: \bibitem[Vollmer et al.(2001)]{vollmer2001} Vollmer, B., Cayatte,
698: V., Balkowski, C., \& Duschl, W.~J.\ 2001, \apj, 561, 708
699:
700:
701: \bibitem[Vollmer et
702: al.(2008)]{vollmer2008} Vollmer, B., Soida, M., Chung, A., van Gorkom, J.~H., Otmianowska-Mazur, K., Beck, R., Urbanik, M., \& Kenney, J.~D.~P.\ 2008, \aap, 483, 89
703:
704: \bibitem[Zabludoff
705: \& Mulchaey(1998)]{zabludoff1998} Zabludoff, A.~I., \& Mulchaey, J.~S.\ 1998, \apj, 496, 39
706:
707:
708: \end{thebibliography}
709:
710: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
711:
712: \begin{figure}[h]
713: \epsscale{1}
714: \plotone{fig1_color.eps}
715: \caption{Bullet Cluster candidates (filled circles) shown with the shock front (a corresponding contour of constant X-ray surface brightness). Arrows indicate the distance between the object and the shock front for sources with IRAC color m$_{4.5}$-m$_{8} > 0.5$.}
716: \label{fig:fig1}
717: \end{figure}
718:
719:
720: \begin{figure}
721: \epsscale{1}
722: \plotone{fig2.eps}
723: \caption{The m$_{4.5}$-m$_{8}$ color as a function of distance from the shock front for 134 cluster candidates. The lower and upper axes indicate projected distance in degrees and Mpc, respectively. Spectroscopically confirmed cluster members are filled in boxes. The vertical dotted line shows the location of the shock front. Sources to the left of this line are considered post-shock galaxies, and those to the right are considered pre-shock galaxies. The horizontal dotted-dashed lines at m$_{4.5}$-m$_{8}$=-0.13 and m$_{4.5}$-m$_{8}$=2.43 represent the expected color of an Elliptical and Sbc galaxy at z=0.3, respectively \citep{assef2007}. The error bar in the upper left corner shows the mean color error of the 134 candidates.}\label{fig:fig2}
724: \end{figure}
725:
726: \begin{figure}
727: \epsscale{1}
728: \plotone{fig3.eps}
729: \caption{The m$_{4.5}$-m$_{8}$ color as a function of distance from shock front for cluster candidates in bins of 275 kpc, with the number of sources per bin denoted. Individual error bars are derived from bootstrapping method. Grey shaded regions show the integrated color for all objects behind and ahead of the shock front, $\pm$1$\sigma$ derived from bootstrapping.}\label{fig:fig3}
730: \end{figure}
731:
732: \begin{figure}
733: \epsscale{1}
734: \plotone{fig4_color.eps}
735: \caption{The optical color-magnitude diagram shown for 88 IRAC selected cluster candidates. The solid line is the color-magnitude relation whose slope and zero-point were adopted from \citet{lopez-cruz2004}. The dotted lines signify 2$\sigma$ from the fit, where the dispersion is obtained by fitting a Gaussian to the residuals. Objects overplotted with a cross symbol have IRAC color m$_{4.5}$-m$_{8}>0.5$ and are thus likely to be star-forming galaxies.}\label{fig:optical}
736: \end{figure}
737:
738:
739: \end{document}
740: