1: %\documentclass[preprint]{emulateapj}.
2: %\documentclass[12pt,preprint]{aastex}
3: \documentclass[apj]{emulateapj}
4: %\usepackage{apjfonts}
5: %%\received{}
6: %%\accepted{}
7: %%\journalid{}{}
8: %%\articleid{}{}
9:
10: \slugcomment{To appear in {\it{The Astrophysical Journal}}}
11: \shortauthors{Gonzalez et al.}
12: \shorttitle{Baryons in Clusters and Groups}
13:
14: \newcommand \sbu {mag arcsec$^{-2}$}
15: %\newcommand \ho {$h_{70}^{-1}$ }
16: \newcommand \ho {}
17: \newcommand{\kms}{~km~s$^{-1}$}
18: \newcommand \ddev {2-deV }
19: \newcommand \dev{$r^{1/4}$ }
20: \newcommand \ser {S\'ersic }
21: \newcommand\rfive{$r_{500}$}
22: \newcommand\rtwo{$r_{200}$}
23: %\newcommand\mfive{M$_{500}$}
24: \newcommand\mfive{{\fontfamily{pzc}\selectfont M}$_{500}$}
25: %\newcommand\mtwo{M$_{200}$}
26: \newcommand\mtwo{{\fontfamily{pzc}\selectfont M}$_{200}$}
27:
28:
29: \begin{document}
30: \title{A Census of Baryons in Galaxy Clusters and Groups}
31:
32: \author{Anthony H. Gonzalez\altaffilmark{1},
33: Dennis Zaritsky\altaffilmark{2,3}, \& Ann I. Zabludoff\altaffilmark{2,3}}
34: \altaffiltext{1}{Department of Astronomy, University of Florida, Gainesville, FL 32611-2055}
35: \altaffiltext{2}{Steward Observatory, University of Arizona, 933 North Cherry
36: Avenue, Tucson, AZ 85721}
37: \altaffiltext{3}{Center for Cosmology and Particle Physics, Dept. of
38: Physics, NYU, New York, NY, 10003}
39:
40: \begin{abstract}
41: We determine the contribution of stars in galaxies, intracluster stars, and
42: the intracluster medium to the total baryon budget in nearby galaxy clusters
43: and groups. We find that the baryon mass fraction
44: ($f_b\equiv\Omega_b/\Omega_m$) within \rfive\ is constant for systems with
45: \mfive\ between $6\times10^{13}\ \mathrm{M}_\odot$ and $1\times10^{15}\
46: \mathrm{M}_\odot$. Although $f_b$ is lower than the WMAP value, the shortfall
47: is on the order of both the observational systematic uncertainties and the
48: %both
49: depletion of baryons within \rfive\ that is predicted by simulations. The
50: data therefore provide no compelling evidence for undetected baryonic
51: components, particularly any that would be expected to vary in importance with
52: cluster mass. A unique feature of the current analysis is direct inclusion of
53: the contribution of intracluster light (ICL) in the baryon budget. With the
54: addition of the ICL to the stellar mass in galaxies, the increase in X-ray gas
55: mass fraction with increasing total mass is entirely accounted for by a
56: decrease in the total stellar mass fraction, supporting the argument that the
57: behavior of both the stellar and X-ray gas components is dominated by a
58: decrease in star formation efficiency in more massive environments. Within
59: just the stellar component, the fraction of the total stellar luminosity in
60: the central, giant brightest cluster galaxy (BCG) and ICL (hereafter the
61: BCG+ICL component) decreases as velocity dispersion ($\sigma$) increases for
62: systems with $145 {\rm \ km\ s}^{-1} \le \sigma \le 1026$ km s$^{-1}$,
63: suggesting that the BCG+ICL component, and in particular the dominant ICL
64: component, grows less efficiently in higher mass environments. The degree to
65: which this behavior arises from our sample selection, which favored systems
66: with central, giant elliptical galaxies, remains unclear. A more robust
67: result is the identification of low mass groups with large BCG+ICL components,
68: demonstrating that the creation of ``intracluster" stars does not require a
69: massive cluster environment. Within \rfive\ and \rtwo, the BCG+ICL
70: contributes on average 40\% and 33\% of the total stellar light, respectively,
71: for the clusters and groups in our sample. Because these fractions are
72: functions of both enclosed radius and system mass, care should be exercised
73: when comparing these values with other studies and simulations.
74: \end{abstract}
75:
76: \keywords{galaxies: clusters: general --- galaxies:cD, formation, evolution, fundamental parameters}
77:
78: %Introduction
79: \section{Introduction}
80: \label{sec:intro}
81:
82: An important step in understanding any astrophysical system is a full
83: accounting of its constituent components. Over scales ranging from groups to
84: clusters of galaxies, measuring how baryons are distributed among their
85: various gaseous and stellar phases can provide clues to the roles played by
86: AGN feedback, starburst winds, galaxy mergers, and tidal stripping
87: \citep{merritt1984,Roussel,valdarnini,delucia,kravtsov,cheng,ettori2006,monaco2006}
88: in driving galaxy evolution in these complex environments. Determining the
89: total contribution of these baryonic components to the cluster mass is
90: similarly valuable. While the baryon fraction in clusters may be expected to
91: reflect the universal value \citep[see][]{White1993,ettori2006}, baryon
92: accountings in clusters to date have fallen short of this expectation
93: \citep{ettori2003,mccarthy} and uncovered possible
94: intriguing trends with cluster mass \citep{lin2003}. With the precise
95: measurement of the universal baryon fraction from WMAP
96: \citep[$f_b\equiv\Omega_b/\Omega_m=0.176^{+0.008}_{-0.013}$;][]{spergel}, this
97: shortfall has gained in physical significance, leading to suggestions that
98: physical processes lower the cluster baryon fraction relative to the Universe
99: \citep[see, for example,][]{he}, that there may be significant, undetected
100: baryon components \citep[see][]{ettori2003,lin2004b}, or that WMAP
101: underestimates $\Omega_m$ \citep{mccarthy}. Any investigation along these
102: lines first requires measurement of {\it all} the baryon components, including
103: the hot, X-ray emitting gas and the stars in and out of galaxies.
104:
105: One baryonic component that past studies were unable to include directly is
106: the intracluster stars (hereafter the intracluster light or ICL). In recent
107: years, multiple studies have detected intracluster light via stacking of large
108: cluster samples \citep{zibetti2005}, deep observations of individual clusters
109: \citep{feldmeier2004,gonzalez2005,krick2006}, and even identification of {\it
110: individual} intracluster stars in very nearby clusters
111: \citep{durrell2002,feldmeier2004b,aguerri2005,ciardullo2005,gerhard2005}. The
112: general consensus is now that all clusters have an ICL component. The natural
113: extension of these studies is placing the ICL in the larger context of cluster
114: and group evolution, examining directly whether this component is important in
115: the chemical enrichment of the intragalactic medium
116: \citep{lin2004b,zaritsky2004}, the total stellar budget, or even in the total
117: baryon budget. While recent work suggests that the ICL does not contribute
118: much to the overall baryon budget of the most massive clusters (the ones most
119: often observed), it could become important in lower mass systems where the
120: plasma is less dominant and the apparent discrepancy between observations and
121: the WMAP baryon fraction is most severe \citep[e.g.][]{lin2003}. This issue
122: is mostly unexplored, as previous ICL measurements have come either from a few
123: systems with little dynamic range \citep{feldmeier2004b,krick2006} or from
124: stacked systems for which mass estimators have been crude \citep{zibetti2005}.
125:
126: In our program we include the ICL's contribution to the total baryon mass
127: budget directly, explore the dependence of total baryon fraction on cluster
128: mass, and consider the relative importance of the stars in the ICL to those in
129: galaxies and of the total stellar mass to the total gas mass in the
130: intracluster medium (ICM). We focus on a sample of 23 nearby galaxy clusters
131: and groups with ICL measurements from \citet{gonzalez2005} that span a wide
132: range in velocity dispersion, $145 {\rm \ km \ s}^{-1} \le \sigma \le 1026$
133: \kms. We determine the cluster radii and masses at an overdensity of 500
134: times critical density (\rfive\ and \mfive;\S3.1), revisit our previous ICL
135: measurements (\S3.2), measure the total luminosity in the cluster galaxies
136: (\S3.3), determine stellar masses (\S3.4), and obtain gas masses from the
137: literature (\S3.5). We then derive relationships for the stellar and total
138: baryon mass fraction with cluster mass (\S4.1), examine the detailed behavior
139: of the ICL relative to the galaxy luminosity (\S4.2), and speculate about the
140: origin of the observed trends (\S4.3). Throughout this paper we assume a
141: cosmology with $\Omega_M=0.27$, $\Omega_\Lambda=0.73$, and $H_0=70$ \kms.
142:
143: \section{Sample Definition and Data}
144:
145: In Paper I \citep{gonzalez2005}, we present a sample of 24 clusters and groups
146: for which we obtained drift scan imaging in Gunn $i$ using the Great Circle
147: Camera \citep{zaritsky1996} on the Las Campanas 1m Swope telescope. The sample
148: is comprised of systems at $0.03<z<0.13$ that contain a dominant brightest
149: cluster galaxy (BCG) with a major axis position that lies within 45$^\circ$ of
150: the direction of the drift scan, east-west, and span a range of velocity
151: dispersions and Bautz-Morgan types \citep{bautz1970}. We refer the reader to
152: Paper I for further details concerning the data and photometric reductions.
153:
154: In Paper II \citep{Zaritsky2006}, we present velocity dispersions for 23 of
155: these clusters and groups. Twelve of these $\sigma$'s are based on our own
156: observations with the Hydra spectrograph at the CTIO Blanco 4m telescope,
157: while the rest are derived from galaxy redshifts in the literature. For all
158: but four systems, the velocity dispersions are calculated using $>20$
159: confirmed members within 1.5 Mpc of the BCG. The system with the smallest
160: number of confirmed members has 8 galaxies and the lowest $\sigma$
161: ($144^{+78}_{-57}$ \kms). Details of the spectroscopic reductions and
162: velocity dispersion measurements are given in Paper II. The information from
163: both Papers I and II relevant to the current analysis is presented in Table
164: \ref{tab:data}. The magnitudes in this Table do not include extinction or
165: k-corrections, which are listed in a separate column. When converting to
166: stellar masses in \S \ref{sec:stellarmass} we do include these corrections.
167:
168: \input{tab1}
169: \section{Analysis}
170:
171: To determine the ratio of baryons to total mass for each group and cluster in
172: our sample, we 1) estimate the total masses and cluster radii, 2) calculate
173: the combined luminosity of the brightest cluster galaxy and intracluster
174: light, 3) determine the total luminosity associated with the other cluster
175: members, 4) convert the luminosities of these components into stellar masses,
176: and 5) obtain X-ray gas mass fractions from the literature. In \S4, we discuss
177: the relationships between total system mass and stellar mass, X-ray gas mass,
178: and total baryon mass.
179:
180: \subsection{Cluster Radii and Masses}
181:
182: To estimate the total baryon fraction from the stellar components and ICM, we
183: must first choose a fixed radius within which the baryons can be summed and
184: then determine the baryonic and total masses within this radius. We choose
185: \rfive, the radius at which the cluster mass density exceeds the critical
186: value by a factor of 500, and the largest radius for which the current X-ray
187: data require no model extrapolation \citep{vikhlinin}. Measurements of the
188: ICL typically do not extend even out to \rfive, but the light is sufficiently
189: concentrated within a few hundred kpc that the model extrapolation to \rfive\
190: introduces a minimal uncertainty.
191:
192: The systems in our ICL sample have measured velocity dispersions, but generally
193: lack X-ray data. To obtain \rfive\ and \mfive\ values for the clusters in
194: our ICL sample that are directly
195: comparable to those determined in X-ray studies, we therefore derive
196: calibrations of the $\sigma$-\rfive\ and $\sigma$-\mfive\ relations using the
197: clusters from \citet{vikhlinin} that also have published velocity dispersions
198: \citep{girardi,wu}. The fit to the $\sigma$-\rfive\ relation has a slope
199: consistent with the theoretical expectation ($1.07\pm 0.12$ vs 1) and
200: corresponds to a cluster with $\sigma=1000$ \kms\ having an \rfive\ of 1.41
201: Mpc. The best fit to the $\sigma$-\mfive\ relationship using the
202: \citet{vikhlinin} sample has a slope consistent with the theoretical
203: expectation ($3.37 \pm 0.53$ vs. 3) and corresponds to a cluster with $\sigma
204: = 1000$ \kms\ having an enclosed mass, \mfive, of $7.6 \times 10^{14}
205: M_\odot$. A similar analysis can be performed for \rtwo\ using the sample of
206: \citet{arnaud2005}, but with the limitation that only five clusters in this
207: sample have published velocity dispersions. Fixing the slope to the value
208: derived for the \citet{vikhlinin} sample, the resulting normalization yields
209: an \rtwo\ of 2.2 Mpc for a $\sigma=1000$ \kms. We utilize this \rtwo\ relation
210: in \S\ref{sec:vdrad} when quantifying the radial dependence of the ICL.
211:
212: For comparison, we also derive \rfive\ and \rtwo\ using the approach of
213: \citet{hansen2005}, directly calculating the radius within which the number
214: density of cluster galaxies implies that the mass density exceeds the critical
215: value by a factor of 200 or 500. If the bias between the galaxy distribution
216: and dark matter is not large, this approach is expected to yield a reasonable
217: estimate for these radii. As in \citet{hansen2005}, we use the
218: \citet{blanton2003} SDSS $i-$band luminosity function integrated over redshift
219: to compute the field density \citep[see][]{hansen2005}. When computing \rtwo\
220: and \rfive, we impose a faint magnitude limit of $m_I=18$; varying this limit
221: by $\pm0.5$ mag alters the derived \rtwo\ values by $\pm5$\%. We find that two
222: approaches yield consistent values except for clusters with low velocity
223: dispersions ($\la 400$ \kms), where the \citet{hansen2005} approach yields
224: larger radii (by as much as a factor of two). For all analyses below we use
225: the X-ray derived values; no results in the paper change qualitatively if we
226: instead use the values from the \citet{hansen2005} approach.
227:
228:
229: \subsection{Total Luminosity of BCG+ICL}
230:
231: A key question in previous discussions of these data
232: \citep{zaritsky2004,gonzalez2005} was whether the BCG and ICL are distinct
233: entities. A central result of Paper I is that the stellar surface brightness,
234: position angle, and ellipticity profiles of these clusters are best reproduced
235: when the BCG and ICL are fit as separate components. In this analysis we
236: choose to combine the luminosity of these two components and refer to the
237: combined entity as simply the BCG+ICL. There are several strong motivations
238: for this choice in the current context. Observationally, the separation of
239: the two components is difficult and the combined luminosity is therefore a
240: more robust quantity than the luminosity of either separately. Using the
241: combined quantity also avoids choosing among the various ICL definitions in
242: the literature \citep[e.g.,][]{feldmeier2004,zibetti2005}, enabling a more
243: straightforward comparison between our results and other observational studies
244: and simulations. Lastly, because we are interested here in a full baryon
245: accounting, the question of whether they reside in the central galaxy or in
246: intracluster space is not as critical as in other contexts. In later
247: discussions of the ICL relative to the galaxy luminosities (\S
248: \ref{sec:stellar}), where the distinction between BCG and ICL is relevant, we
249: reiterate from Paper I that the bulk ($> 80$\%) of the luminosity in the
250: cluster's BCG+ICL is contained in the ICL.
251:
252: We determine the luminosity of the BCG+ICL component as follows. In Paper I
253: we used the GALFIT package \citep{peng2002} to fit the surface brightness
254: distribution in each cluster out to a radius of 300 \ho kpc from the BCG. The
255: BCG and ICL were best fit with separate $r^{1/4}$ profiles. Here we use those
256: best-fit profiles to construct a 2-D model image from which we determine the
257: flux within circular apertures corresponding to \rfive\ (in our analysis of
258: the baryon mass fraction; \S4.1) and to fractions of \rtwo\ (in our
259: determination of the BCG+ICL's contribution to the total stellar light;
260: \S4.2). We estimate the systematic errors in the measured fluxes using the
261: $\pm1\sigma$ systematic uncertainties in the structural parameters of the
262: best-fit profiles (these exceed the statistical errors due to uncertainties in
263: the sky level; see Paper I). The advantages of using the models rather than
264: placing apertures on the original data are that we minimize contamination from
265: cluster galaxies, interpolate directly over masked regions, and provide a
266: robust means of quantifying the systematic uncertainties. We use circular
267: rather than elliptical apertures to enable direct comparison with other
268: studies and simulations.
269:
270: An implicit assumption in this work is that our models from Paper I, which are
271: determined for $r<300$ kpc, adequately describe the ICL at all radii. In our
272: models, 80\% of the total light of the BCG+ICL is contained in the central 300
273: kpc, making it unlikely that we are significantly overestimating the
274: luminosity of this component with our extrapolation. Moreover, the BCG+ICL
275: dominate the total cluster luminosity within this radius, and for our data the
276: potential bias from inclusion of faint, unresolved cluster galaxies in the ICL
277: is at most a few percent \citep{gonzalez2000}. Although the uniformity of the
278: profiles within 300 kpc argues that extrapolating is reasonable, we will
279: underestimate the ICL luminosity if there is any significant contribution at
280: larger radii that is not included in our model. We expect that the best method
281: of testing this assumption is via extension of current planetary nebulae
282: studies in nearby clusters like Virgo \citep{feldmeier2004b} to large
283: fractions of \rtwo. Because of the nature of this study, we are most concerned
284: with overestimating the ICL, and large overestimates ($>$ 30 \%) are not
285: possible.
286:
287:
288: \subsection{Total Luminosity of Cluster Galaxies}
289: \label{sec:luminosity}
290: We compute the total luminosity of non-BCG cluster galaxies within the same
291: apertures (\rfive\ and fractions of \rtwo) used for the BCG+ICL. Hereafter,
292: we refer to the non-BCG cluster galaxies as simply ``cluster galaxies". After
293: excluding stars using the SExtractor \citep{Bertin1996} stellarity index, we
294: sum the flux of all galaxies fainter than the BCG and brighter than $m_I=18$
295: lying within the aperture. We then perform a statistical background
296: subtraction, and include a completeness correction to account for the
297: contribution of fainter cluster members (see below). In cases where the
298: aperture size is larger than the width of the drift scan images, we also apply
299: a correction to account for the lost area. Specifically, we weight the
300: luminosity contribution of each galaxy by the inverse of the fractional area
301: lost at that radius (e.g., a galaxy would have a weighting factor of two if
302: half the area is lost).
303:
304:
305: Because the determination of the flux density from background galaxies is
306: critical in this type of measurement, we estimate the background using two
307: independent methods. First, we compute the background galaxy density with an
308: annular region located $30\arcmin$ to $60\arcmin$ from the BCG. This method,
309: which was also used in Paper I, has the advantage that the photometry for the
310: background and cluster galaxies is performed in the same fashion from the same
311: images. However, large scale structure may bias the observed background level
312: even at these large radii. To ascertain whether we are suffering from such a
313: bias, we also compute the background level using the Sloan Digital Sky Survey
314: luminosity function from \cite{blanton2003}, integrating over redshift and
315: applying mean evolutionary and k-corrections derived from \citet{bruzual2003}
316: models with the Padova isochrones \citep{padova1994}. A comparison of the
317: results from the two techniques shows that for the ensemble of systems the
318: total cluster galaxy luminosities after background subtraction are consistent
319: to within 10\%. In the subsequent discussion, we apply the first method of
320: direct background subtraction because this approach relies on fewer
321: assumptions.
322:
323: After background subtraction, we include a completeness correction to account
324: for the luminosity contribution from cluster members fainter than the limiting
325: magnitude. For this correction, we adopt the cluster luminosity function of
326: \citet{christlein2003}, $\alpha=-1.21$ and $M_{*,R}=-21.14$, using $R-I=0.82$
327: to convert from $R$ to $I$. Although the value of $M_*$ is identical to the
328: $i-$band value for the field from \cite{blanton2003}, the cluster luminosity
329: function has a slightly steeper faint end slope. Using a different value of
330: $\alpha$ systematically shifts the total luminosity due to cluster galaxies
331: and thus the total stellar light of the cluster. For $\alpha=-1$, the
332: computed total luminosity of the cluster galaxies would on average be 12\%
333: lower, increasing the relative importance of the BCG+ICL, but decreasing the
334: stellar contribution to the overall baryon budget. The total luminosities of
335: the BCG+ICL and of the cluster galaxy components are presented in Table
336: \ref{tab:data}.
337:
338: \begin{figure*}
339: %\epsscale{1.20}
340: \plotone{f1_color.eps}
341: \caption{Determination of cluster and group baryon fractions within \rfive\ as
342: a function of \mfive\ (bottom axis) and velocity dispersion (top axis).
343: Plotted are X-ray gas mass fractions from \citet{vikhlinin} (filled circles) and
344: \citet{gastaldello2006} (filled triangles), and the stellar mass fractions
345: (BCG+ICL+galaxies, open squares) for systems in our sample with masses that overlap
346: the range shown for the X-ray studies. All measurements are within \rfive.
347: Overplotted are the best fit relations
348: for the \citet{vikhlinin} sample gas mass fractions and for our stellar mass
349: fractions. The WMAP 1$\sigma$ confidence region for the universal baryon
350: fraction from \citet{spergel} is shown for comparison, and the right-hand axis
351: shows $Y$, the ratio of the baryon fraction for each component to the
352: universal value from WMAP.
353: {\it Lower panel:} The total baryon fraction derived for the \citet{vikhlinin}
354: and \citet{gastaldello2006} clusters if our best-fit stellar baryon relation
355: is used to estimate a stellar baryon contribution for each of these
356: systems. The weighted mean for the sample (dashed line) is $\Omega_b/\Omega_m=0.133\pm0.004$.
357: We observe no trend in baryon fraction with cluster mass. The error bars on the individual
358: data points and the weighted mean include only statistical uncertainties.
359: The unweighted mean for the combined \citet{zhang2007} and \citet{rasmussen2006}
360: samples (dotted line) is included to provide a sense of the systematic uncertainties.
361: \label{fig:baryons}
362: }
363: \epsscale{1.0}
364: \end{figure*}
365:
366:
367: \subsection{Total Stellar Masses}
368: \label{sec:stellarmass}
369: The conversion from luminosity to stellar mass for the cluster galaxies and
370: the BCG+ICL potentially incurs the most significant uncertainty in the total
371: baryon mass budget. A straightforward theoretical calculation, based on
372: PEGASE2.0 \citep{pegase} for a Salpeter IMF and a 10 Gyr old population,
373: yields $M/L_I = 4.7$, but recent observations indicate that real systems
374: exhibit lower $M/L$. To estimate a mean $M/L$, we use the SAURON results from
375: \citet{Cappellari2006}, which provide an empirical determination of the
376: central stellar $M/L$ using Schwarschild dynamical modeling of two-dimensional
377: kinematic data. Equation 9 in \citet{Cappellari2006} provides the luminosity
378: dependence of the mass-to-light ratio in the $I-$band. We compute a
379: luminosity-weighted M/L for $L>0.25 L_*$ (the range over which the SAURON
380: relation is determined) using the same \citet{christlein2003} luminosity
381: function as in \S\ref{sec:luminosity}. We adopt the resultant value,
382: $M/L=3.6$, for both the cluster galaxies and the BCG+ICL in our baryon
383: calculation, but note that the uncertainty is at least 10\% and that $M/L$
384: varies with both mass \citep{Cappellari2006,Zaritsky2006} and stellar
385: population. An improved accounting for the ICL component would require
386: knowledge of the source of the ICL stars if the variations in $M/L$ are driven
387: by population, rather than dynamical, differences. An improved accounting for
388: the cluster galaxy population would require determining the fraction of spiral
389: galaxies in these clusters and using a more appropriate $M/L$ for the spirals.
390:
391: \subsection{Total Mass in the Intracluster Medium}
392: Determination of total cluster baryon fractions also requires measurement of
393: the mass of hot gas in the intracluster medium. Because we lack these data for
394: clusters in our sample, in \S \ref{sec:results} we fit the behavior of the
395: stellar mass fraction with cluster mass and use this relation to derive total
396: baryon fractions for clusters with published X-ray gas fractions. For the gas
397: mass fractions we adopt the values of X-ray gas fraction inside of \rfive ,
398: $f_{g}$, and of enclosed mass, \mfive, from the tabulations by
399: \citet{vikhlinin} and \citet{gastaldello2006}. The latter adds only two new
400: systems, but helps constrain the baryon fraction at low mass. We note that
401: there are several systems for which the X-ray gas fraction was measured by
402: both groups. In these instances there is reasonably good agreement, and hence
403: we expect no significant relative systematic biases between the two samples.
404:
405: %\vskip 5.5cm
406: \section{Results and Discussion}
407: \label{sec:results}
408:
409: \subsection{Baryon Mass Fraction versus Cluster Mass}
410: \label{sec:baryonfraction}
411:
412: In this section, we obtain two principal results. First, summing the three
413: baryonic components --- the hot gas of the intracluster medium, the stars of
414: the BCG+ICL, and the stars of the other cluster galaxies --- gives a total
415: baryon mass fraction, $f_b$, that is constant as a function of total mass
416: (Figure 1). This constancy contradicts previous studies and arises in large
417: part from the inclusion of the ICL for systems with \mfive\ $<$ few $\times
418: 10^{14} M_\odot$. Second, the weighted mean value of the total baryon mass
419: fraction is $f_b=0.133$.\footnote{Defining $Y$ to be the ratio of the baryon
420: fraction to the cosmic value from the WMAP third year results, this value for
421: the total baryon fraction corresponds to $Y=0.76$.} For reasons discussed
422: below, we conclude that there is no compelling case for undetected baryons in
423: the groups and clusters in our sample.
424:
425:
426: We fit the behavior of the stellar and X-ray gas mass fractions with mass as
427: power laws (top panel of Figure \ref{fig:baryons}). Within \rfive, the stars
428: result in a relationship $\log(f_{*}) = (7.57 \pm 0.08) -
429: (0.64\pm0.13)\log$(\mfive) and the plasma in $\log(f_{g}) = (-3.87 \pm
430: 0.04) + (0.20 \pm 0.05)\log$(\mfive) for the sample of \citet{vikhlinin}.\footnote{Note
431: that while this fit applies only to the \citet{vikhlinin} data set, the \citet{gastaldello2006} points
432: in Figure \ref{fig:baryons} are consistent with this relation.}
433: fit the \citet{vikhlinin} sample alone The behavior of the two components
434: is clearly inverted. The plasma component, which dominates at cluster masses
435: $> 10^{14} M_\odot$, has a shallower dependence on mass than the stellar
436: component. We use our best fit relation for the stellar mass to convert the
437: X-ray gas fractions from \citet{vikhlinin} and \citet{gastaldello2006} to
438: total baryon fractions on a cluster-by-cluster basis. The resulting total
439: baryon fractions, shown in the lower panel of Figure \ref{fig:baryons}, are
440: independent of cluster mass over the range of \mfive\ from $6\times10^{13}$ to
441: $10^{15}$ M$_\odot$. If one considers only the statistical uncertainties
442: associated with our total cluster baryon fraction and the WMAP measurement,
443: then the two values are discrepant at the level of 3.2$\sigma$.
444:
445: There are three possible interpretations of this result in the context of a
446: full baryon accounting. First, systematic uncertainties may be significantly
447: larger than the random errors, and our measurement may therefore be consistent
448: with the universal WMAP value. Because the gas mass fraction is much larger
449: than the stellar mass fraction over the majority of the mass range we probe,
450: we focus this discussion on potential errors in the gas mass
451: fraction. Systematic errors, by their nature, are often difficult to calculate
452: and can best be illuminated by comparing independent studies.
453: \citet{zhang2007} and \citet{rasmussen2006} provide gas mass fractions for
454: high and low mass systems, respectively. \citet{zhang2007} find gas fractions
455: that are 15\% larger than those we adopted over the range probed by their
456: sample. \citet{rasmussen2006} also find gas fractions that are larger than
457: those we adopted for comparably low mass systems. Together, these two samples
458: cover the mass range of our adopted sample and result in an average
459: $f_b=0.149$, which is $2.1\sigma$ discrepant with the WMAP value and 12\%
460: larger than the $f_b$ we derive for the \citet{vikhlinin} data set. We
461: conclude that indeed the systematic uncertainties dominate, that the sense of
462: the uncertainty works to diminish the discrepancy between measurements and
463: expectations, and that the case for undetected baryon components rests on
464: removing these systematic uncertainties. Second, the baryon mass fraction
465: with \rfive\ in clusters may be different from the universal value. Cluster
466: simulations routinely predict baryon fractions that are lower than the
467: universal baryon fraction by $\sim$10\%. We have deliberately not adopted any
468: of those as the target because there are still a minority of simulations that
469: predict a baryon fraction larger than universal
470: \citep[e.g.][]{kravtsov}. Third, the shortfall may be resolved by yet
471: undetected baryon components. The constancy of the measured baryon fraction
472: with total system mass argues against a large undetected component, but one
473: could contribute at a modest level and bypass current detection. In the end,
474: all of these possibilities may contribute at some level to a final resolution
475: of the baryon accounting, but at this point we are left to conclude that there
476: is no compelling evidence for undetected baryons.
477:
478:
479: We close this section by comparing our study to that of \citet{lin2003}, who
480: obtained a dependence of the baryon fraction on cluster mass in conflict with
481: our result. The range of masses covered by the two studies is the same, and
482: we both use \rfive\ as our fiducial radius, so the discrepancy is real. The
483: most obvious differences are our inclusion of the ICL component and our
484: different normalization of the gaseous component. We model the effect of
485: including the ICL into their accounting by using our relationship for the
486: BCG+ICL mass vs. system mass and ``correcting'' their values of the baryon
487: fraction on a cluster-by-cluster basis.\footnote{This is a slight ($\sim$20\%)
488: overcorrection because it includes the BCG, which \cite{lin2003} had already
489: included.} The relationship between baryon fraction and mass decreases from a
490: $\sim$3$\sigma$ result in the original \citet{lin2003} study to a 1.8$\sigma$
491: result in the corrected case. Therefore, the inclusion of the ICL explains
492: part of the discrepancy between our result and that of \citet{lin2003},
493: although apparently not all of it. The lack of the ICL in the \citet{lin2003}
494: accounting was not an oversight --- they attempted to model the effect of the
495: ICL in \cite{lin2004b}, but simply lacked data of the sort we present here.
496: While the exclusion of the ICL causes a relative underprediction of the
497: contribution of stellar baryons for the lowest mass sytems, the Lin et
498: al. adopted X-ray gas fractions, which are $\sim 25$\% larger than ours and
499: hence consistent with the WMAP baryon mass fraction for the most massive
500: systems, cause a relative overprediction of baryons in the highest mass
501: systems. These two effects lead to the apparent mass-dependent baryon
502: fraction of Lin et al. As with the comparison among different studies
503: discussed previously, we have an unresolved discrepancy in the gas mass
504: fractions at the $\sim 20$\% level that highlights the difficulty in such an
505: analysis and supports our contention that there is yet no compelling evidence
506: for either additional baryonic components or a trend in baryon fraction with
507: system mass.
508:
509: \begin{figure}
510: \plotone{f2.eps}
511: \caption{Calibration of velocity dispersion to X-ray derived masses within
512: \rfive. We plot the seven systems from \cite{vikhlinin} for which we found
513: velocity dispersions in the literature \citep{girardi,wu}. The line is our fit
514: to the data, and hence our calibration for converting the velocity dispersions
515: measured for our optical clusters into \mfive 's. }
516: \label{fig:masscal}
517: \end{figure}
518:
519:
520: \subsection{Stellar Component of Clusters and Groups}
521: \label{sec:stellar}
522: \subsubsection{Behavior vs. \mfive}
523:
524: We now consider the behavior of the stellar component in more detail. We
525: compute the dependence of the optical luminosity on \mfive\ for the cluster
526: galaxies, the BCG+ICL, and the sum of these populations. There are multiple
527: studies of the dependence of the total stellar luminosity of cluster galaxies
528: on cluster mass. In the $K-$band, \citet{lin2004a} find that for the galaxies
529: alone (excluding the BCG) $L\propto M^\gamma$ with $\gamma=0.82\pm0.05$
530: ($\gamma=0.82\pm0.04$) within \rfive\ (\rtwo), and \citet{ramella2004} find
531: $\gamma=0.74\pm0.06$ within \rtwo. A least-squares fit to our $I-$band data
532: for the cluster galaxies yields $\gamma=0.71\pm0.07$ within \rfive, which is
533: consistent with these results and most other published values \citep[][but see
534: \citet{kochanek2003} for an exception]{girardi2000,rines2004}. We conclude
535: that the conflict between our result --- that the baryon mass fraction is
536: insensitive to system mass for groups and clusters --- and those of past
537: studies does not arise predominantly from our measurement of cluster galaxy
538: properties.
539:
540: \begin{figure}
541: \plotone{f3.eps}
542: \caption{The total absolute magnitude of the BCG+ICL component (circles) and
543: of cluster galaxies (triangles) within $r_{500}$ as a function of cluster
544: velocity dispersion. The filled circles denote systems where the BCG+ICL
545: contributes less than 40\% of the total luminosity within this radius, while
546: open-crossed circles correspond to systems with higher fractions of light in
547: the BCG+ICL. The BCG+ICL luminosity depends less strongly on the velocity
548: dispersion than does the total galaxy luminosity. We also note that there is
549: no systematic difference in the BCG+ICL absolute magnitudes for clusters with
550: high and low BCG+ICL fractions. The BCG+ICL fraction differences are instead
551: driven predominantly by variations in the total absolute magnitudes of the
552: non-BCG cluster galaxy populations.
553: \label{fig:magsig}}
554: \end{figure}
555:
556: In comparison with the cluster galaxy luminosity versus cluster mass relation,
557: the correlation between BCG+ICL luminosity and cluster mass is weaker,
558: increasing by only about a factor of two in total luminosity between poor
559: groups with $\sigma=400$\kms~and rich clusters with $\sigma=1000$\kms~(Figure
560: \ref{fig:magsig}). The existence of groups in which the fraction of stars in
561: the BCG+ICL exceeds that in galaxies out to \rfive\ (Figure \ref{fig:icl})
562: demonstrates that the rich cluster environment is not a requirement for the
563: production of a significant intracluster stellar population. Instead, it
564: appears that local density maxima in both the group and cluster environments
565: are capable of generating intracluster stars, presumably via tidal processes.
566: This conclusion is valid whether one is interested in the combination of the
567: BCG and ICL, or in the ICL alone, because the latter contains the bulk
568: (typically $>80$\%) of the total light in the two components
569: \citep{gonzalez2005}. However, our results should not be interpreted to mean
570: that all groups have a significant ICL component, because our original sample
571: selection focused on systems with dominant BCGs.
572:
573: The observed weak dependence in the BCG+ICL is given by
574: $\gamma=0.39\pm0.05$. Physically, this result implies that production of the
575: BCG+ICL --- presumably from the stripping of stars from galaxies --- is more
576: efficient in group (or subcluster) environments than in clusters. Again, we
577: caution that when selecting clusters and groups for this sample, we
578: specifically targeted systems with dominant BCGs. If there is a strong
579: correlation between the presence of a dominant BCG (i.e., our sample) and the
580: total BCG+ICL luminosity (which is not necessarily the case because the ICL
581: contributes much more light), then our selection criteria would artificially
582: produce a smaller $\gamma$ than that actually found in nature.
583:
584: \subsubsection{Total Cluster Luminosities and Mass-to-Light Ratios}
585:
586: The observed weak dependence of the BCG+ICL luminosity on cluster mass, if
587: real, implies that $\gamma$ for the total stellar light within \rfive\
588: ($L_{500}$) must be smaller than the value for the galaxy population alone.
589: The best fit value for the behavior of $L_{500}$ is $\gamma=0.47\pm0.05$, with
590: a scatter of only 0.26 mag about the best fit. Recasting in terms of a
591: mass-to-light ratio yields $M_{500}/L_{500} \propto L_{500}^{1.13\pm0.23}$.
592: For comparison, literature values that include only the cluster galaxy
593: contribution typically have $M/L\propto L^{0.3}$. Thus, if the slope we
594: measure for the BCG+ICL component is not significantly biased by our exclusion
595: of groups without a dominant BCG, then our data argue that galaxy cluster
596: mass-to-light ratios are much more strongly dependent upon system mass than is
597: commonly presumed.
598:
599: \begin{figure}
600: \plotone{f4.eps}
601: \caption{The fraction of the total stellar luminosity, measured within \rfive,
602: which is contained in the BCG+ICL component as a function of cluster velocity
603: dispersion. The dotted vertical line corresponds to $\sigma=750$ \kms,
604: roughly the median velocity dispersion for an Abell richness class $R\ge1$
605: cluster \citep{zabludoff1990}.
606: \label{fig:icl}}
607: \end{figure}
608:
609:
610: Conversely, if our selection against groups lacking a dominant BCG is the
611: reason that the BCG+ICL and cluster galaxy luminosities have different
612: observed cluster mass dependences, then there must exist a correlation between
613: mass-to-light ratio and the presence of a dominant BCG. Specifically, for a
614: given velocity dispersion, groups with a dominant BCG have higher BCG+ICL
615: luminosities and hence lower mass-to-light ratios.
616:
617: \subsubsection{BCG+ICL Luminosity Fraction: Dependence on Velocity Dispersion and Radius}
618: \label{sec:vdrad}
619:
620: We determine the fraction of total stellar luminosity contained in the BCG+ICL
621: component as a function of velocity dispersion and radius. In Figure
622: \ref{fig:icl}, we plot the BCG+ICL luminosity fraction, $L_{BCG+ICL}/
623: (L_{gal}+L_{BCG+ICL})$ as a function of velocity dispersion within \rfive,
624: which is slightly larger than the physical region within which the BCG and ICL
625: were modeled in Paper I. We find a trend of decreasing BCG+ICL luminosity
626: fraction with increasing velocity dispersion. The same trend is observed if we
627: use a fixed physical radius of 300 kpc, but the scatter is higher. The
628: presence of this trend is also independent of whether \rfive\ or a fixed
629: fraction of \rtwo\ is used for the abscissa.
630:
631: Within the clusters, the radial dependence of the BCG+ICL luminosity fraction
632: is shown in Figure \ref{fig:icl_rad}, where we plot the BCG+ICL luminosity
633: fraction enclosed within different radii. Combining the full ensemble of
634: clusters, the BCG+ICL luminosity fraction decreases smoothly from 65\% within
635: 0.1\rtwo\ to 33\% within \rtwo. This radial decline, which indicates that the
636: ICL is more concentrated than the galaxies, is also seen by
637: \citet{zibetti2005} and is predicted by the simulations of
638: \citet{murante2004}.
639:
640: \begin{figure}
641: \plotone{f5.eps}
642: \caption{The cumulative fraction of light contained in the BCG+ICL as a
643: function of radius. Subdividing the sample into clusters with $\sigma>750$
644: \kms~(open boxes) and $\sigma<750$\kms~(filled triangles) demonstrates that
645: the BCG+ICL fraction is systematically lower in the more massive clusters. The
646: horizontal dashes included in the plots denote the {\it rms} variations for
647: the cluster ensemble at each radius.
648: \label{fig:icl_rad}}
649: \end{figure}
650:
651: Lastly, we use a threshold of $\sigma=750$ \kms~--- roughly the median
652: velocity dispersion of an Abell richness class $R \ge 1$ cluster
653: \citep{zabludoff1990} --- to divide the data set into two subsamples (Figure
654: \ref{fig:icl}). The clusters with higher velocity dispersions exhibit lower
655: BCG+ICL luminosity fractions at all radii. To confirm that this result is not
656: driven purely by the BCG, we also compute the corresponding luminosity
657: fractions within circular annuli from 0 to $0.5$\rtwo. We find that at all
658: radii the fraction of the cluster luminosity contained in the ICL is lower for
659: the more massive clusters. Moreover, the rms scatter about the mean for the
660: BCG+ICL luminosity fractions is only 10\% for the high-$\sigma$ sample
661: compared with 20\% for the low-$\sigma$ sample within 0.5\rtwo. Thus, for
662: clusters and groups with a dominant BCG, the BCG+ICL luminosity fractions are
663: smaller and more uniform in the more massive systems.
664:
665: \subsection{Origin of the Baryon Trends}
666:
667: We have identified two independent trends with mass that shed light on the
668: environmental dependencies of physical processes that distribute baryons among
669: the various phases. First, we find that the fraction of stars in the ICL grows
670: (and that in galaxies diminishes) with decreasing system mass. This result is
671: predicated on the assumption that we are not strongly biased in the type of
672: system observed, as discussed above. Even so, we now know that in at least
673: some low mass systems the ICL forms efficiently. As a result, we can rule out
674: cluster-only processes, such as harassment \citep{moore1996,moore1998}, as the
675: dominant ICL formation mechanism, and find support for mechanisms that would
676: be more effective in lower mass systems, such as galaxy-galaxy interactions
677: \citep{merritt1984}. Such tidal interactions, which could strip stars from
678: galaxies and build up the ICL, are favored in lower mass groups, because the
679: internal velocity dispersions of group galaxies are similar to the velocity
680: dispersion of the group as a whole. If all low mass systems do have high ICL
681: luminosity fractions, and some are the building-blocks of clusters, then a
682: diluting mechanism, such as the infall of individual galaxies into clusters,
683: would be necessary to reduce the final ICL fraction of more massive systems.
684: Of course, it may be that not all low mass groups have high ICL fractions like
685: those in our sample. The low ICL fractions of high mass clusters may therefore
686: simply reflect their growth by the accretion of groups with lower ICL
687: fractions than we observe. A related possibility is that the clusters accrete
688: and mix progenitor groups before they have time to form substantial ICL via
689: galaxy-galaxy interactions, whereas the ICL continues to grow in more isolated
690: groups. Simulations of the formation of the ICL must reproduce these trends
691: and will lead to a better understanding of the details of the formation
692: mechanisms \citep{Willman2004,murante2004,SL2005,SL2006,Rudick2006}.
693:
694: Second, we find that the baryon mass fraction across the range of systems for
695: which we have both optical and X-ray data is roughly constant over the mass
696: range $6\times10^{13}\ \mathrm{M}_\odot <$ \mfive$\ < 1\times10^{15}\
697: \mathrm{M}_\odot$. This result requires a fairly well-tuned balance between
698: the X-ray gas mass fraction, which declines by a factor of two between systems
699: with \mfive\ of $10^{15} M_\odot$ and $10^{14} M_\odot$, and the stellar mass
700: fraction, which increases by a factor of four to compensate. The matching of
701: the two trends supports the suggestion \citep{bryan,lin2003} that the X-ray
702: gas mass fraction behavior is due to the decreased efficiency of turning gas
703: into stars in the more massive environments. The reduction in star formation
704: efficicency in higher mass systems may arise from the decreased efficiency of
705: tidal interactions among galaxies, the removal of the gaseous reservoirs of
706: galaxies by interactions with the intracluster medium, or increased feedback
707: processess that quench star formation. As usual, interpreting such trends is
708: complicated by the fact that today's lower mass systems are not necessarily
709: similar to the lower mass antecedents of today's higher mass systems.
710:
711: \section{Conclusions}
712: \label{sec:conclusions}
713:
714: We have conducted a study of 23 nearby galaxy clusters and groups with a
715: dominant, early-type brightest cluster galaxy. For this sample we determine
716: the luminosities of the different stellar constituents (including stars
717: associated with cluster galaxies and the intracluster stars) as a function of
718: cluster velocity dispersion ($\sigma$), cluster mass (\mfive), and
719: cluster-centric radius. Combining these results with current measurements of
720: the X-ray gas mass as a function of $\sigma$, we calculate the dependence of
721: the baryon budget on environment. We reach the following conclusions:
722:
723: 1. The observed total baryon mass fraction within \rfive\ is constant
724: for systems with \mfive\ between $6\times10^{13}$ and $10^{15}$ M$_\odot$.
725: Although the mean baryon mass fraction is formally $>3\sigma$ lower than the
726: WMAP measurement, both the observational systematic uncertainties --- which
727: dominate the formal random errors --- and the predicted deficit of baryons in
728: clusters relative to the universal value are on the order of the observed
729: baryon shortfall. Thus, we conclude that there is no compelling evidence for
730: undetected baryons at this time.
731:
732: 2. The inclusion of the combined light from the brightest cluster galaxy and
733: the intracluster stars, which we refer to as the BCG+ICL component, becomes
734: increasingly important in the baryon budget as the system mass decreases. For
735: systems with \mfive $= 10^{14} M_\odot$, the BCG+ICL component is nearly 35\%
736: of the total stellar luminosity within \rfive, and the mass of the stellar
737: component including the BCG+ICL and the cluster galaxies is comparable to that
738: of the X-ray gas component. Consequently, the increase in X-ray gas mass
739: fraction with increasing \mfive\ is directly reflected by a decrease in the
740: total stellar mass fraction. The matching of the falling stellar mass and
741: rising gas mass with total mass supports the suggestion made previously
742: \citep{bryan, lin2003} that the star formation efficiency decreases with
743: increasing system mass.
744:
745: 3. Considering only the stellar components, we observe that the fraction of
746: the total stellar luminosity in the BCG+ICL decreases with system velocity
747: dispersion. The origin of this trend is that the luminosity of the BCG+ICL
748: component changes more slowly with cluster mass than does the luminosity of
749: cluster galaxies, increasing by only a factor of two between 400 \kms~ and
750: 1000 \kms. The presence of large BCG+ICL luminosity fractions in the lower
751: mass systems demonstrates that the rich cluster environment is not required
752: for production of ``intracluster'' stars.
753:
754: The decrease in the importance of the BCG+ICL component, and in particular the
755: dominant ICL component, with increasing cluster mass is consistent with more
756: efficient stripping of stars from galaxies -- presumably via galaxy-galaxy
757: interactions -- in the lower mass systems. Although it remains unclear the
758: degree to which our sample selection, which favored systems with central,
759: giant elliptical galaxies, impacts this result, it is noteworthy that
760: \citet{purcell2007} predict precisely the type of relation that we observe
761: using a simple analytic model for BCG+ICL formation from satellite accretion
762: and disruption. A similar model is also able to reproduce the relative
763: luminosities that we observe for the BCG and ICL \citep{conroy2007}, implying
764: that this basic physical picture can largely explain two significant results
765: from our work without invoking significant bias due to sample selection.
766:
767:
768: 4. The mean BCG+ICL luminosity fraction for our sample decreases
769: monotonically from 65\% within 0.1\rtwo\ to 33\% within \rtwo. This trend is
770: qualitatively similar to that seen in the stacked sample of
771: \citet{zibetti2005}.
772:
773: 5. Interior to \rtwo, we find that the BCG+ICL optical luminosity fraction is
774: inversely correlated with cluster velocity dispersion. Our mean value of 33\%
775: is somewhat larger than one might expect based upon the work of
776: \citet{zibetti2005}, who found a similar luminosity fraction of 32\% within a
777: smaller, 500 kpc radius. This difference likely reflects a difference between
778: the systems included in the two studies. For clusters with $\sigma>750$
779: \kms~, we obtain a mean value of 23\% within \rtwo, which is consistent with
780: both the \citet{zibetti2005} results and simulated BCG+ICL luminosity
781: fractions within \rtwo\ for rich clusters \citep{Willman2004}.
782:
783: \acknowledgements
784: The authors thank the referee, Stefano Ettori, for excellent suggestions that
785: significantly improved the paper. AHG also thanks Andrey Kravtsov and
786: Yen-Ting Lin for insightful discussions regarding this project. AIZ is
787: supported by NSF grant AST-0206084 and NASA LTSA grant NAG5-11108. DZ
788: acknowledges financial support for this work from a Guggenheim fellowship and
789: NASA LTSA grant 04-0000-0041. AIZ and DZ thank the KITP for its hospitality
790: and financial support through the National Science Foundation grant
791: PHY99-07949, and generous support from the NYU Physics department and Center
792: for Cosmology and Particle Physics during their sabbatical there.
793:
794:
795:
796:
797:
798:
799:
800: %References
801: \bibliographystyle{apj}
802: \bibliography{ms}
803:
804:
805:
806:
807:
808: \end{document}
809: