1: \documentclass[useAMS,usenatbib]{mn2e}
2: \usepackage{graphics}
3: \usepackage{epsfig}
4: \usepackage{natbib}
5: \usepackage{float}
6: \bibliographystyle{mn2e}
7:
8:
9: \newcommand{\hMpc}{{\ifmmode{h^{-1}{\rm Mpc}}\else{$h^{-1}$Mpc }\fi}}
10: \newcommand{\hkpc}{{\ifmmode{h^{-1}{\rm kpc}}\else{$h^{-1}$kpc }\fi}}
11: \newcommand{\hMsun}{{\ifmmode{h^{-1}{\rm {M_{\odot}}}}\else{$h^{-1}{\rm{M_{\odot}}}$}\fi}}
12: \newcommand{\bimp}{\tilde{\rm b}} %\frac{\rm b_{imp}}{r_{\rm vir}}}
13: \newcommand{\vgap}{{\rm v_{12}}}
14: \newcommand{\mgap}{\ensuremath{\Delta m_{12}}}
15:
16:
17: \title[Fossil Groups]{The Fossil Phase in the Life of a Galaxy Group}
18: \author[von Benda-Beckmann et al.]{ Alexander M. von
19: Benda-Beckmann$^{1}$, Elena D'Onghia$^{2}$\thanks{Marie Curie
20: Fellow}, Stefan Gottl{\" o}ber$^{1}$, \newauthor Matthias
21: Hoeft$^{3}$, Arman Khalatyan$^{1}$, Anatoly Klypin$^{4}$,
22: and Volker M{\" u}ller$^{1}$
23: \\
24: $^{1}$Astrophysical Institute Potsdam, An der Sternwarte 16, Germany \\
25: $^{2}$Institute for Theoretical Physics - University of Zurich,
26: Winterthurerstrasse 190, Switzerland\\
27: $^{3}$Jacobs University Bremen, Campus Ring 12, Germany \\
28: $^{4}$ Astronomy Department, New Mexico State University,
29: MSC 4500, P.O.Box 30001, Las Cruces, NM, 880003-8001, USA}
30:
31:
32:
33: \renewcommand{\topfraction}{0.85}
34: \renewcommand{\textfraction}{0.1}
35:
36: \begin{document}
37:
38: \date{Submitted Version 2007 October}
39:
40: \pagerange{\pageref{firstpage}--\pageref{lastpage}} \pubyear{2007}
41:
42: \maketitle
43:
44: \label{firstpage}
45:
46: \begin{abstract}
47:
48: We investigate the origin and evolution of fossil groups in a concordance
49: LCDM cosmological simulation. We consider haloes with masses between
50: $(1-5)\times10^{13} \hMsun$ and study the physical mechanisms that lead to
51: the formation of the large gap in magnitude between the brightest and the
52: second most bright group member, which is typical for these fossil systems.
53: Fossil groups are found to have high dark matter concentrations, which we
54: can relate to their early formation time. The large magnitude-gaps arise
55: after the groups have build up half of their final mass, due to merging of
56: massive group members. We show that the existence of fossil systems is
57: primarily driven by the relatively early infall of massive satellites, and
58: that we do not find a strong environmental dependence for these systems. In
59: addition, we find tentative evidence for fossil group satellites falling in
60: on orbits with typically lower angular momentum, which might lead to a more
61: efficient merger onto the host. We find a population of groups at higher
62: redshifts that go through a ``fossil phase'': a stage where they show a
63: large magnitude-gap, which is terminated by renewed infall from their
64: environment.
65:
66:
67: \end{abstract}
68:
69: \begin{keywords}
70: galaxies: formation - galaxies : clusters - cosmology : dark matter -
71: galaxies : evolution - methods : N-body simulations
72: \end{keywords}
73:
74: \section{INTRODUCTION}
75:
76: Observations in the last decade revealed the existence of groups of galaxies
77: containing extended X-ray-emitting hot gas with properties expected for poor
78: clusters such as the Virgo cluster, but in the optical light completely
79: dominated by a single luminous, giant elliptical galaxy \cite[][]{Ponman1994a,
80: Vikhlinin1999a}. The second brightest galaxy in these systems is more than a
81: factor five less luminous than the dominant elliptical.
82:
83: To be specific, these systems are defined as spatially extended X-ray sources
84: with luminosities $L_{\rm X,bol} \ge 10^{42} h_{50}\;{\rm erg}\; {\rm s}^{-1}$.
85: The optical counterparts are galaxy groups with $\Delta m_{12} \ge 2 $ mag,
86: where $\Delta m_{12}$ is the absolute R-band magnitude-gap between the
87: brightest and second-brightest galaxies.
88:
89: These systems are extremely interesting for several reasons: Although they
90: have X-ray temperatures comparable to the Virgo cluster these systems show a
91: galaxy luminosity function with a deficit of bright galaxies beyond the
92: characteristic magnitude of the Schechter function M* , or of visible galaxies
93: as compared to the predictions of cosmological simulations
94: \cite[][]{DOnghia2004a}, whereas Virgo contains six M* galaxies
95: \cite[][]{Jones2000a}. Therefore they have been interpreted as the final
96: outcome of galaxy-galaxy mergers. Numerical simulations suggest that the
97: luminous galaxies in a group will eventually merge to form a single giant
98: elliptical galaxy \cite[e.g.][]{Barnes1989a}. The merging timescales for the
99: brightest group members (with magnitudes M $\sim$ M* or brighter) in compact
100: groups are typically a few tenths of a Hubble time. Therefore, by the present
101: day several group galaxies have likely merged into the giant elliptical.
102: Outside of the high-density core, the cooling time for the intra-group medium
103: is larger than a Hubble time; thus, while the luminous galaxies in some groups
104: have had enough time to merge into a single object, the large-scale X-ray halo
105: of the original groups should remain intact. This means that a merged group
106: might appear today as an isolated elliptical galaxy with a group-like X-ray
107: halo \cite[][]{Ponman1993a}. Hence these systems have been termed ``fossil''
108: groups.
109:
110: Using the ROSAT All-Sky Survey data, \citet[][]{Ponman1994a} found the first
111: ``fossil'' group candidate. The RXJ1340.6+4018 system has an X-ray luminosity
112: comparable to a group, but 70\% of the optical light comes from a single
113: elliptical galaxy \cite[][]{Jones2000a}. The galaxy luminosity function of
114: RXJ1340.6+4018 indicates a deficit of galaxies at the characteristic magnitude
115: M*. \citet[][]{Jones2000a} have studied the central galaxy in detail and
116: found no evidence for spectral features implying recent star formation, which
117: indicates the last major merger occurred at least several Gyrs ago.
118:
119: These systems are not rare. With a number density of $(5 \times 10^{-7}$ -- $2
120: \times 10^{-6}) \; h^{3} {\rm{Mpc}^{-3}}$, they constitute 10 to 20 per cent
121: of all clusters and groups with an X-ray luminosity greater than $2.5 \times
122: 10^{42} h \;{\rm ergs\; s}^{-1}$ \cite[][]{ Vikhlinin1999a, Romer2000a,
123: Jones2003a}. The number density of fossil groups is comparable to that of
124: brightest cluster galaxies \cite[][]{Jones2003a}. Thus they may be of
125: considerable importance as the place of formation of a significant fraction of
126: all giant ellipticals.
127:
128: So far different approaches to model and measure the number density of fossil
129: systems have been undertaken. \citet[][]{Milosavljevic2006a} adopted an
130: extended Press-Schechter approach to estimate (5-40)\% for the expected
131: fraction of fossil groups in a mass range of $10^{13}-10^{14} \hMsun$ and
132: decreasing to (1-3)\% for massive clusters. Recently
133: \citet[][]{vandenBosch2007a} used 2dF data to measure a fossil fraction of 6.5
134: \% for group with masses $10^{13}-10^{14}\hMsun$ . Using N-body simulations
135: \citet[][]{dOnghia2007a} estimated a fraction of 18\% and
136: \citet[][]{Sales2007a} (8-10)\% for $M_{\rm group} > 10^{13}\hMsun $ based on
137: an analyses of the Millennium simulation \cite[][]{Springel2005a}. A
138: compilation of measurements of the fossil fraction and a discussion of
139: differences due to selection effects is given in \citet[][]{Dariush2007a}.
140: Note, however, that some of these systems seem to be fossil clusters rather
141: than fossil groups \cite[][Zibetti et 2007, in preparation]{Gastaldello2007a,
142: Cypriano2006a}, i.e. galaxy clusters with the typical magnitude-gap of 2
143: between the brightest and the second brightest cluster galaxy. It seems at
144: least that a significant fraction of groups is fossil and it is a strong
145: function of group mass \cite[][]{Milosavljevic2006a, Sales2007a,
146: vandenBosch2007a} .
147:
148: Most of the theoretical work on fossil groups focused on the predictions of
149: the statistics of the magnitude-gap in the luminosity function. The physical
150: processes that lead to the formation of a mass or magnitude-gap in these
151: systems are still poorly understood. Early work suggested that fossil groups
152: result from mergers of the largest galaxies within compact groups
153: \cite[][]{Barnes1989a} and are due to early formation time
154: \cite[][]{dOnghia2005a,Dariush2007a}. However, it is not yet understood under
155: which conditions mergers are so efficient that they produce such an extreme
156: gap in magnitude. When do fossil groups typically form their magnitude-gap?
157: Are fossil systems isolated systems or can they also populate high density
158: regions like galaxy clusters? Are fossil systems early formed systems, are
159: they more concentrated than other systems? Are fossil groups long lasting
160: systems, or does the group environment regulate its lifetime by infall of new
161: massive structures?
162:
163: These are the open questions which we address here using a high-resolution
164: N-body simulation. The answers should guide the interpretation of
165: observational datasets especially by surveys like PANSTARS combined with
166: COSMOz that can search for fossil systems at higher redshift and provide a
167: framework for understanding the formation of giant ellipticals within the
168: current cosmology.
169:
170: This paper is organized as follows. We describe the numerical simulation in
171: $\S2.1$ and the selection criterion of the sample of fossil groups in $\S2.2$.
172:
173: In $\S3$ we describe the fossil group properties we find, like number density,
174: formation time, concentration and time of last major merger. We investigate
175: the formation mechanisms leading to the large magnitude-gap in $\S4$. Our main
176: results are summarized in $\S5$.
177:
178:
179:
180:
181:
182:
183: \section{Methods}
184:
185: We have selected our sample of fossil groups from a 80 $h^{-1}{\rm Mpc}$ dark
186: matter only N-body simulation which is large enough to lead to a statistically
187: meaningful sample. Since we are mainly interested in the dynamical properties
188: of the massive group members, we focus on a dark matter simulation in this
189: work.
190:
191: \subsection{Simulation}
192:
193: The initial conditions were generated for a WMAP3 cosmology with matter
194: density $\Omega_{\rm m} = 0.24$, an linear mass variance on 8
195: $h^{-1}$Mpc-scale $\sigma_{8} = 0.76$, a dimensionless Hubble parameter $h =
196: 0.73$ and a spectral index of primordial density perturbations $n = 0.96$
197: \cite[][]{Spergel2007a}. We used $N =512^{3}$ dark matter particles, i.e. a
198: particle mass of $4 \times 10^{8} h^{-1} {\rm M}_{\odot}$. Starting at
199: redshift $z = 40$ we evolved the simulation until the presence with the MPI
200: version of the Adaptive Refinement Tree (ART) code \cite[][]{Kravtsov1997a}.
201: The ART code enabled us to reach a force-resolution of $~1-3$ kpc in the most
202: refined regions, making sure our massive sub-haloes do not suffer from
203: over-merging \cite[][]{Klypin1999a} inside of our group sized haloes.
204:
205: We identified groups with a friend-of-friends (FOF) algorithm with a linking
206: length of $l = 0.17\; d$ (corresponding to a mean over-density of roughly
207: 330), with $d$ the mean inter-particle distance. The advantage is that it
208: identifies groups of any shape. In a second step we identified the bound
209: (sub-)structures in the groups. To this end we used the Bound Density Maximum
210: (BDM) halo finder \cite[][]{Klypin1999a}. This algorithm removes unbound
211: particles from the haloes and is therefore particularly suitable to identify
212: sub-haloes and their properties, like their circular velocity. Since the
213: determination of sub-halo masses in groups is uncertain we characterized them
214: by their maximum circular velocity. The BDM halo with the highest circular
215: velocity within the FOF group is considered the host group halo. We found 116
216: groups in the mass range of $(1\times 10^{13} - 5 \times 10^{13}) h^{-1}{\rm
217: M}_{\odot}$, corresponding
218: to the massive end of galaxy groups. %We repeated the FOF analysis for
219: %redshifts $z=0.3, \; 0.65, \;{\rm and} \; 0.93$ to study the evolution of the
220: %fraction of fossil groups.
221:
222: We calculated the mass accretion histories of the haloes using 130 time steps
223: of equal distance in the expansion parameter, $\Delta a = 0.006$. To this end
224: we selected the 20 per cent of the most bound particles of haloes and compare
225: them in 8 consecutive time steps. We uniquely associated a halo to its
226: progenitor by considering halo pairs, which have the largest number of
227: particles in common and do not differ by more than a factor 5 in mass. The
228: last criterion was included to avoid spurious misidentification by sub-haloes
229: with their host halo.
230:
231:
232:
233: \subsection{Selection of Fossil Groups}
234:
235: Fossil groups are defined on the basis of a measured magnitude-gap between the
236: brightest and the second brightest group member. In our simulation we can only
237: identify the dark matter haloes which host the group galaxies. Currently
238: methods are being developed in order to relate galaxy luminosities to dark
239: matter haloes in a statistical way \cite[e.~g.~][]{Yang2004a, Vale2004a,
240: Cooray2005a, Conroy2006a}. Here we follow this idea and associate
241: luminosities to the (sub-)haloes in the group. We assume simply that the most
242: luminous galaxy is the central galaxy of the group host halo with circular
243: velocity ${\rm v_{\rm circ,1}}$. Consequently, the halo with the second
244: highest circular velocity ${\rm v_{\rm circ,2}}$ will host the second
245: brightest group galaxy. Here the circular velocity ${\rm v_{circ}}$ is
246: always taken at the maximum of the rotation curve.
247:
248: To model the magnitude-gap we adopt a similar approach as
249: \citet[][]{Milosavljevic2006a} where we relate the the halo circular velocity
250: to the luminosity of the central galaxy using an empirically measured mean
251: R-band mass-to-light ratio \cite[][]{Cooray2005a}. Assuming a Sheth-Thormen
252: \cite[][]{Sheth1999a} mass distribution function for the dark matter haloes
253: and a functional form as in Equation 1, that expresses the halo mass in
254: luminosity for the central galaxies, \citet[][]{Cooray2005a} fit the measured
255: R-band luminosity function of \citet[][]{Seljak2005a}. We convert our circular
256: velocities to luminosities by the relation
257:
258:
259: \begin{equation}
260: L\left(M\right) = L_{0} \left(\frac{M}{M_{0}} \right)^{a} \left[ b + \left( \frac{M}{M_{0}}\right)^{cd} \right]^{-1/d}
261: \end{equation}
262:
263: \noindent with $L_{0} = 5.7\times10^{9} L_{\odot}$, $M_{0} = 2\times10^{11}
264: {\rm M_{\odot}}$, $a = 4$, $b = 0.57$, $c = 3.78$, $d = 0.23$, where we
265: substitute circular velocities for masses using the relation found by
266: \citet[][]{Bullock2001a}: $M/\left(h^{-1}\;{\rm M_{\odot}}\right) =
267: 10^{\alpha}\cdot \left[{\rm v_{circ}}\right/\left({\rm km
268: \;s^{-1}}\right)]^{\beta}$, with $\alpha = 4.3$ and $\beta = 3.4$. We
269: then define fossil groups as having masses in the range of $(1\times 10^{13} -
270: 5 \times 10^{13}) h^{-1}{\rm M}_{\odot}$ and a magnitude-gap $\mgap \ge 2$ mag
271: in the R-band .
272:
273: Mass accretion onto haloes stops at the time when they become sub-halos of a
274: more massive object like a group. After infall they start to lose matter due
275: to tidal interactions. Since baryons tend to lie deeper in the potential well,
276: they will be less prone to get tidally stripped. Therefore, the total
277: luminosity is more likely to be related to the mass at infall \cite[see
278: e.~g.~][]{Kravtsov2004a}. Following this idea we characterize the sub-haloes
279: of the groups by their masses and circular velocities {\it at infall times}
280: onto the group. The choice of relating luminosities to the circular velocities
281: of halos at infall time has been motivated by recent successes in matching the
282: data by modeling the two- and three-point correlation functions
283: \cite[][]{Conroy2006a, Berrier2006a, Marin2007a}.
284:
285:
286: \section{Distribution and properties of fossil groups}
287:
288: \subsection{Abundance}
289:
290: In this section our main aim is to characterize the properties of the fossil
291: groups of our group sample in order to guide the interpretation of future
292: observational constraints. We begin by computing the abundance of fossil
293: systems in our catalog.
294:
295: Assuming a magnitude-gap of $\mgap \ge 2$ (see dashed line in Figure 1) 24 per
296: cent of the groups of our catalog are classified as fossil, corresponding to
297: a number density of $5.5 \times 10^{-5} h^{3}{\rm Mpc}^{-3}$. This rate is
298: higher than previous estimates based on $N$-body simulations
299: \cite[][]{dOnghia2007a}, semi-analytic models \cite[][]{Sales2007a,
300: Dariush2007a}, and observational estimates \cite[][]{ Vikhlinin1999a,
301: Romer2000a, Jones2003a, vandenBosch2007a}, which all get a fraction of
302: around 10 per cent for groups in the mass range considered here.
303: However, only 15 well studied fossil groups are known at present with X-ray
304: data. Therefore these abundances present large uncertainties and might be
305: well underestimated. We believe that the over-estimate comes from our adopted
306: scheme for relating circular velocities to luminosities of the central
307: galaxies in groups, where we followed \citet[][]{Milosavljevic2006a} and
308: \citet[][]{Bullock2001a}. We are interested mainly in the formation process
309: of systems with a large magnitude-gap, which clearly corresponds to systems
310: with a large gap in circular velocities even if the related magnitudes are
311: uncertain. We therefore stick to our adopted method and study how our selected
312: fossil population differs from the normal group population.
313:
314: \begin{figure}
315: \hspace{-1.5cm} \epsfig{file=Fig/n_cum_v12_fin_highz.eps,width=10cm}
316: \caption{The fraction of galaxy-group sized haloes with a magnitude-gap
317: parameter larger than $\mgap$. The dashed line indicates our defining
318: magnitude-gap for fossil groups $\mgap \ge 2$.}
319: \label{vratio}
320: \end{figure}
321:
322:
323:
324: \subsection{Environment Density Dependence}
325:
326: Fossil groups are systems with many properties typical for galaxy clusters.
327: Hence, a further interesting test concerns the question whether fossil groups
328: are isolated systems that populate the low density regions or tend to reside
329: in higher density regions of the Universe like galaxy clusters. A good test
330: would be the cross-correlate the X-ray emitting fossil groups with galaxies in
331: the nearby universe, e.g with SDSS data. However the limited number of fossil
332: groups actually known makes an estimate of such correlations extremely
333: difficult. Some observational indications, though still uncertain, would
334: suggest that fossil groups could be fairly isolated systems
335: \cite[e.g.][]{Jones2000a, Jones2003a, Adami2007a}.
336:
337: We check in our simulated sample of groups whether fossil systems populate
338: preferentially low density regions in the universe. We estimate the
339: environmental density on a scale of 4 \hMpc. To this end we determine the
340: environmental over-density $\Delta_{4} = \rho_{4}/\rho_{\rm bg} - 1$, where
341: $\rho_4$ is the dark matter density within 4 \hMpc from the group center of
342: mass, with the inner one virial radius is subtracted, and $\rho_{\rm bg}$ is
343: the background matter density. Figure ~\ref{density} shows the distribution of
344: the over-density $\Delta_{4}$ for fossil and normal groups. Most of the
345: groups, in the range of mass considered here, independent of being fossil or
346: not, populate preferentially the intermediate over-density region. A two-sided
347: Kolmogorov-Smirnov test for the two cumulative distributions shown gives
348: $D=0.31$, corresponding to a probability of 0.03 that the the two samples are
349: drawn from the same distribution. We checked that adopting a slightly larger
350: or smaller volume (scales of 2-5 \hMpc) does not change our results
351: significantly. We do not find a strong tendency for fossil systems to be
352: preferentially located in low density environments. We suggest therefore that
353: observations might be biased to find fossil groups preferentially in low
354: density regions, which could be due to group selection effects.
355:
356: \begin{figure}
357:
358: \hspace{-1cm}\epsfig{file=Fig/v12_dens_4mpc_histo_cut.eps,width=10cm}
359: \caption{The group environment over-density $\Delta_{4}$ computed in a sphere of $ 4
360: \hMpc$ for fossil (black solid line) and non-fossil groups (grey dashed
361: line). To compute the over-density, the inner virial radius was
362: subtracted. For comparison, cluster typically populate regions of
363: $\Delta_{4} > 10$. There does not seem to be a strong tendency for fossil
364: groups to be in low density environments. Error bars indicate 1$\sigma$
365: Poissonian uncertainties.}
366: \label{density}
367: \end{figure}
368:
369:
370: \subsection{Formation Time}
371:
372: \begin{figure}
373: \hspace{-1cm}\epsfig{file=Fig/zform_dmag.eps,width=10cm}
374: \caption{ The correlation between the formation redshift of the
375: group host halo and its magnitude-gap parameter for fossil groups
376: (triangles) and normal groups (circles). Over-plotted are the mean
377: values (solid line)and lower and upper quartiles (dotted lines).}
378: \label{lastmm}
379: \end{figure}
380:
381: \begin{figure}
382: \hspace{-1cm} \epsfig{file=Fig/zform_conc_0.2.eps,width=10cm}
383: \caption{
384: The formation redshift of the host halo as a function of the group
385: concentration. The fossil group sample is marked with triangles and the
386: normal groups are drawn with circles. The dashed line is a linear fit to
387: all points. Crosses correspond to the mean values for the concentrations of
388: fossil: $c_{1/5} = 6.4$ (bold cross) and non-fossil groups $c_{1/5} =5.5$
389: (light cross), with the widths indicating the 1$\sigma$ standard deviations.}
390: \label{lastmm}
391: \end{figure}
392:
393: The halo formation redshift is defined as the epoch at which the system
394: assembled 50\% of its final mass \cite[e.g.][]{Lacey1993a}. Figure 2 shows a
395: correlation between the magnitude-gap parameter and the formation redshift of
396: the host halo for all the fossil systems (triangles) and the normal groups
397: (grey circles). As already pointed out in \citet[][]{dOnghia2005a,
398: dOnghia2007a} this correlation shows that fossil groups tend to form earlier
399: than normal groups, albeit with large scatter. In order to assess this
400: correlation we draw the mean and upper and lower quartiles (the solid and
401: dotted lines in Figure 2). The visual impression is quantified by statistical
402: measures as the Pearson's linear correlation coefficient $r = 0.39$, implying
403: a weak linear correlation between the magnitude-gap and the formation time.
404:
405:
406: \begin{figure*}
407: \begin{center}
408: \hspace{-2cm}
409: \begin{minipage}[t]{1.0\textwidth}
410: \begin{minipage}[t]{0.5\textwidth}
411: \epsfig{file=Fig/v_12_z0_mm.eps,width=10cm}
412: \end{minipage}
413: \hfill
414: \begin{minipage}[t]{0.5\textwidth}
415: \epsfig{file=Fig/v_12_z0.3_mm.eps,width=10cm}
416: \end{minipage}
417: \end{minipage}
418:
419: \vfill \hspace{-2cm}
420: \begin{minipage}[t]{1.0\textwidth}
421: \begin{minipage}[t]{0.5\textwidth}
422: \epsfig{file=Fig/v_12_z0.65_mm.eps,width=10cm}
423: \end{minipage}
424: \hfill
425: \begin{minipage}[t]{0.5\textwidth}
426: \epsfig{file=Fig/v_12_z0.93_mm.eps,width=10cm}
427: \end{minipage}
428: \end{minipage}
429: \end{center}
430: \caption{The evolution of the magnitude-gap parameter $\mgap$ for
431: fossil groups selected at $z$=0 (for redshifts $z$=0, 0.3, 0.65,
432: and 0.93). The non-fossil groups, which constitute the majority of
433: the groups, are left out from this plot for clarity. The plot shows that the
434: majority of the fossil systems had one or more massive satellites at higher
435: redshift. Note that the formation of the magnitude-gap happens typically
436: later (between $z=0-0.7$) than the formation of the groups, which occurred
437: around $z\ge0.8$ (see Figure 2).}
438:
439: \label{vratiotime}
440:
441:
442:
443: \end{figure*}
444: \begin{figure*}
445: \begin{center}
446: \hspace{-2cm}
447: \begin{minipage}[t]{1.0\textwidth}
448: \begin{minipage}[t]{0.5\textwidth}
449: \epsfig{file=Fig/v_12_z0.93_mm_phase.eps,width=10cm}
450: \end{minipage}
451: \hfill
452: \begin{minipage}[t]{0.5\textwidth}
453: \epsfig{file=Fig/v_12_z0.65_mm_phase.eps,width=10cm}
454: \end{minipage}
455: \end{minipage}
456:
457: \vfill \hspace{-2cm}
458: \begin{minipage}[t]{1.0\textwidth}
459: \begin{minipage}[t]{0.5\textwidth}
460: \epsfig{file=Fig/v_12_z0.3_mm_phase.eps,width=10cm}
461: \end{minipage}
462: \hfill
463: \begin{minipage}[t]{0.5\textwidth}
464: \epsfig{file=Fig/v_12_z0_mm_phase.eps,width=10cm}
465: \end{minipage}
466: \end{minipage}
467: \end{center}
468: \caption{The evolution of the magnitude-gap parameter $\mgap$ for
469: fossil groups selected at $z$=0.93 that end up in the $z$=0 group sample,
470: given at different epochs ($z$=0.93, 0.65, 0.3, and 0). Again
471: for clarity, only the selected sample at $z$=0.93 is shown. The open
472: circles indicate fossil groups that are selected in the same mass range as
473: the group sample at $z$=0. Most of the high redshift fossil systems, when
474: traced forward in time, experience renewed infall of massive satellites and
475: become normal groups at $z$=0. These systems are undergoing a `fossil
476: phase'. }
477:
478: \label{vratiotime}
479: \end{figure*}
480:
481:
482: \subsubsection{Concentration Parameter}
483:
484: The early formation redshift is also reflected in a higher concentration
485: parameter \cite[][]{Navarro1997a}.
486:
487: We define the concentration of our haloes by the ratio of the virial radius of
488: the host halo to the radius of a sphere enclosing one fifth of its virial
489: mass: $c_{1/5} = r_{\rm vir}/r_{1/5}$. This definition of the halo
490: concentration allows for a robust concentration determination when the haloes
491: are merger remnants and un-relaxed \cite[][]{AvilaReese2005a}. The correlation
492: shown in Figure 3 between formation redshift and concentration is well fitted
493: by a linear relation $z_{\rm form} = -0.79 + 0.27 \times c_{1/5}$ (marked with
494: the dashed bold line).
495:
496: The fossil groups clearly populate the early formed, more concentrated part of
497: the plot, and have a mean concentration of $c_{1/5} = 6.4$, which is about 16
498: per cent higher than the concentrations found for normal groups $c_{1/5} =
499: 5.5$. Our findings are consistent with fossil groups being systems with higher
500: dark matter concentrations than usual groups, which supports such a trend
501: found by \citet[][]{Khosroshahi2007a}. At present there are yet few
502: observational constraints on this issue \cite[e.g.][]{Gastaldello2007a,
503: Khosroshahi2007a}. However with the upcoming X-ray observations of fossil
504: groups with Chandra and XMM will provide soon better constraints.
505:
506: \subsection{Last Major Merger}
507:
508: Recent studies of the giant elliptical at the center of fossil groups report
509: no signs of a recent major merger activity, indicating that any major merger
510: should have happened at least more than approximately 3 Gyrs ago
511: \cite[][]{Jones2000a, Khosroshahi2006a}.
512:
513: For each halo we estimate the time of the last major merger of the group
514: haloes of our sample by studying the detailed mass assembly history. To identify
515: the time of the last major merger, we denote a halo as a major merger remnant
516: if its major progenitors were classified as a single group at one time but two
517: separate groups with a mass ratio less than 4:1 at the preceeding time. Note
518: that when the mass ratio defining the major merger event is restricted to
519: almost 1:1 progenitors, the time of the last major merger should in general
520: coincide with the formation time (as defined above). We find that only 15\% of
521: the fossil groups experienced the last major merger less than 2 Gyrs ago, and
522: at least 50\% had the last major merger longer than 6 Gyrs in qualitative
523: agreement with the observations.
524:
525:
526:
527: \section{Formation of Fossil Groups}
528:
529: In previous sections we investigated some properties of our galaxy-group sized
530: haloes that can be compared to observations. In particular, the fossil groups
531: appear to be more concentrated systems, they formed earlier, with the last
532: major merger happened a long time ago, in qualitative agreement with current
533: observational constraints. We now turn to the question why these systems are
534: devoid of a significant substructure. Obviously, no large satellite has fallen
535: in lately. The question then is, how long is such a major infall ago -- if it
536: ever occurred? And if they fell in, when do they merge to create the magnitude-gap?
537:
538: Furthermore, we look for signs of efficient merging in fossil systems. We
539: focus on two main conditions that lead to the formation of these systems: i)
540: the low infall rate of massive satellites into the host halo; and ii)
541: distribution of angular momentum of the in-falling massive satellites.
542:
543: \subsection{Forming the Magnitude-Gap}
544:
545: To quantify when the magnitude-gap was formed in fossil systems, we plot the
546: distribution of the magnitude-gap of fossil groups selected at $z=0$ in Figure
547: 5. This sample is traced backward in time to $z=0.3$ (the right top panel);
548: $z=0.65$ (the bottom panel on the left) and $z=0.93$ (the bottom panel on the
549: right). We find that the groups selected as fossil at present, show a lower
550: magnitude-gap once traced backwards in time. They therefore do not qualify
551: anymore as fossil systems at higher redshifts. The magnitude-gap is typically
552: formed over a wide range of redshifts between $z=0-0.7$. It worth noting that
553: this happens typically after the group halo has gained half of its mass, which
554: occurred earlier around redshift $z \ge 0.8$ (see Fig.~ 2).
555:
556: \subsection{The Fossil Phase}
557:
558: Is the `fossil stage' a final stage or will the systems fill their
559: magnitude-gap with the time? To assess this question, we select a sample of
560: fossil systems at high redshift $z=0.93$, and track these forward in time (as
561: shown in Figure 6). The open circles indicate massive fossil groups in the
562: same mass regime as the sample selected at $z=0$. Following all the fossil
563: systems forward in time we note that they leave the range where they would be
564: identified as fossil systems due to new in-falling satellites. Only three of
565: these systems did not experience further infall of a massive satellites from
566: their surrounding environment so that they end up as a fossil system today.
567:
568: Our simulation suggest that the existence of a gap in the galaxy luminosity
569: function in fossil systems is only a transition phase in the evolution of
570: groups, the duration of which is related to the merging of group members with
571: the central object and infall of new haloes.
572:
573: \subsection{Properties of In-falling Satellites}
574:
575: We showed in previous sections that fossil groups tend to assemble earlier
576: their mass than normal groups, and that the unusual magnitude-gap is a
577: transient phase in the evolution of the group in the hierarchical universe. A
578: group identified as fossil today did not experience any late infall of massive
579: satellites to fill the gap in the magnitude distribution function of the group
580: members. In order to assess this question quantitatively we compute the
581: average infall rate of haloes onto fossil systems as compared to normal
582: groups. This quantity is expressed in our analysis by computing the
583: cumulative number of massive (within 2 magnitudes of the host) satellites
584: falling into the host halo after redshift z divided by the total number of
585: fossil (normal) groups: $<N_{\rm infall}\left(<z\right)> = N_{\rm
586: infall}(<z)/N_{\rm groups}$. Figure~\ref{n_infall} shows that fossil groups
587: accrete the larger satellites earlier in time as compared to the normal
588: groups. An early infall of massive companions ensures enough time for the
589: massive satellite to merge into the host halo. Note that for $z>0.8$ fossil
590: and non-fossil groups have similar infall history of massive satellites. In
591: fact the slopes of the evolution of the infall rate is the same for both
592: distribution for $z>0.8$, showing that the infall rate of satellites is only
593: different at low redshifts.
594:
595: We checked whether the difference in infall rate is determined by environment.
596: This is done by splitting the sample up in $\Delta_{4} < 5$ and $\Delta_{4}
597: \ge 5$ and evaluating the cumulative number of in-falling satellites (see
598: dashed lines in figure 7). Although the denser regions do experience a bit
599: more infall, the difference is only of the order of 10-30 per cent, not as big
600: as we observe for fossil and non-fossil systems (in agreement with
601: \citet[][]{Maulbetsch2007a}, who find little environmental dependence of the
602: mass accretion history in this mass range). This supports lack of a strong
603: environmental preference for fossil groups found in section 3.2.
604:
605:
606: \begin{figure}
607: \hspace{-1cm}\epsfig{file=Fig/n_infall_cum_dens.eps,width=10cm}
608: \caption{The mean cumulative number of massive satellites with
609: $\mgap < 2$ mag in-falling into the host halo at redshifts $z_{\rm infall}
610: < z$. The grey solid line corresponds to normal groups and the black solid
611: line to fossil groups. The dashed curves indicate the mean cumulative
612: number of massive satellites when the group samples are split up by low
613: dense regions $\Delta_{4} \le 5$ and higher dense regions $\Delta_{4} > 5$.
614: }
615: \label{n_infall}
616: \end{figure}
617:
618:
619: We address the question whether the conditions under which the massive
620: satellites enter the group enable an efficient merger which could lead to a
621: gap in the magnitude distribution of the fossil group galaxies. We checked the
622: angular momenta values of the satellites at time of infall.
623: Figure \ref{angmom} shows the angular momentum of the satellites at infall
624: time. The angular momentum is calculated by taking the cross product of the
625: distance at infall and the velocity at infall $L_{\rm sat} = \left({\bf r_{\rm
626: inf}}\times{\bf v_{\rm inf}}\right)/\left(r_{\rm max}\cdot{\rm v_{\rm
627: circ}}\right)$, with $r_{\rm max}$ and ${\rm v_{circ}}$ both measured at
628: the maximum of the rotation curve. Both distributions peak at $L_{\rm sat}
629: \approx r_{\rm max}{\rm v_{\rm circ}}$. Fossil groups seems to be lacking
630: satellites in the high end tail of angular momentum distribution, which may
631: cause a faster merging of the satellites. However, since the distribution is
632: rather poorly sampled, better number statistics are needed to confirm this
633: result.
634:
635:
636: \begin{figure}
637: \hspace{-1cm}\epsfig{file=Fig/angmom_dens.eps,width=10cm}
638: \caption{ The distribution of the in-falling satellite normalized angular
639: momentum for fossil (black solid line) and non-fossil groups (grey dashed
640: line). The distribution seems more narrowly distributed around $L_{\rm sat}
641: \approx r_{\rm max}{\rm v_{\rm circ}}$ for fossil groups, i.e. its high
642: angular momentum tail seems to be less pronounced. Error bars indicate
643: $1\sigma$ Poissonian uncertainties.}
644: \label{angmom}
645: \end{figure}
646:
647:
648: \section {SUMMARY AND CONCLUSIONS}
649:
650: From an N-body simulation, we select and analyze a large sample of galaxy
651: group-sized haloes, that allows us to study in detail the mechanisms that lead
652: to the formation of fossil systems in the hierarchical universe.
653:
654: Our criterion to select fossil systems is based on identifying galaxy-group
655: sized haloes showing a gap in the magnitude between the two most massive
656: members. This selection criterion assumes that the circular velocity of a halo
657: traces its luminosity until it becomes a substructure of another more massive
658: system. We relate the magnitude of our haloes to the circular velocities by
659: using the empirical mean relation between dark matter halo mass and central
660: galaxy R-band luminosity found by \citet[][]{Cooray2005a}.
661:
662: Our results may be summarized as follows:
663:
664: \begin{itemize}
665: \item In the mass range $ 1\times 10^{13} - 5\times 10^{13} h^{-1}{\rm
666: M}_{\odot}$ 28 of the 116 groups sized haloes, i.~e.~ 24 per cent qualify
667: as fossil groups according to the definition of a magnitude-gap of 2
668: magnitudes between the brightest and second most bright galaxy. This
669: fraction is higher than the measured values. The largest uncertainty here
670: is how to relate the mass or circular velocity of the haloes to the
671: luminosities of their central galaxies. Because our adopted method of
672: relating mass to galaxy luminosity is uncertain due to the broad scatter in
673: this relation \cite[][]{Cooray2006a}, and the rate is sensitive to the group
674: definition, as well as selection effects, a robust comparison to observed
675: number densities obtained by other authors is difficult at the moment. We
676: are selecting systems with large circular velocity- or respectively
677: mass-gaps, so that our sample can be used to study the formation of the
678: extreme magnitude-gaps observed in fossil systems.
679:
680: \item The fossil groups identified in our sample tend to form earlier than the
681: other groups. The fossil systems have assembled half of their final mass
682: around $z\ge 0.8$ in agreement with the previous works. They form their
683: magnitude-gap typically between redshift $z=0-0.7$, much later than the
684: formation time of the groups. The early formation time is also expressed in
685: a slightly higher dark matter concentration. The average concentration for
686: the fossil groups is $c=6.4$ compared to $c=5.5$ of normal groups. Further,
687: we find that the majority of the fossil group seem to have experienced the
688: last major merger longer than 3 Gyr ago.
689:
690: \item We do not find a strong correlation between the environment and the
691: formation of fossil systems. Observations that indicate that fossil systems
692: are found preferentially in low density environments might be biased by
693: selection effects.
694:
695: \item The primary driver for the large magnitude-gap is the early infall of
696: massive satellites that is related to the early formation time of these
697: systems. The difference in infall rate for different group environments is
698: only of order 10-30 per cent, far less than the observed difference between
699: fossil and non-fossil groups, and hence the current environment does not
700: seem the primary driver for the lack of massive satellites. This is in
701: agreement also with the lack of strong correlation between fossil systems
702: and environment.
703:
704: \item We suggest that efficient mergers of massive members within the groups
705: can create the magnitude-gap typical of fossil groups at any redshift. The
706: efficiency of the merger process seems to be linked to the lower angular
707: momentum of the massive satellites in-falling into the host halo of fossil
708: groups when compared to normal groups. However, due to the limited number
709: statistics we need more data to substantiate this.
710:
711: \item By selecting samples of fossil groups at higher redshift ($z \approx 1$)
712: we find that many of them do not exhibit the magnitude-gap anymore once they
713: are traced forwards until present time. The majority of them fill the
714: magnitude-gap with time by infall of new massive satellites. We conclude
715: that the stage for a group to be ``fossil'' is a transient phase.
716: \end{itemize}
717:
718:
719:
720:
721:
722: \section*{Acknowledgments}
723:
724: The computer simulation was done at Columbia supercomputer at NASA Ames
725: Research Center. A.v.B.B. acknowledges support from Deutsche
726: Forschungsgemeinschaft (DFG) under the project MU 1020/6-3. E.D. is supported
727: by a EU Marie Curie Intra-European Fellowship under contract MEIF-041569. A.K.
728: acknowledges support of NSF grant AST-04070702. M.H. acknowledges support
729: from the DFG under the project Vo 855/2.
730:
731: \bibliography{articles_foss} \bsp
732:
733: \label{lastpage}
734:
735:
736: \end{document}
737: