1: %\documentstyle[epsfig]{aastex}
2: %\received{}
3: %\revised{}
4: %\accepted{}
5:
6: \documentstyle[epsfig, emulateapj5]{aastex}
7: \topmargin=0.5in
8:
9: \def\msun{{\rm\,M_\odot}}
10: \def\lsun{{\rm\,L_\odot}}
11: \def\zsun{{\rm\,Z_\odot}}
12: \newcommand{\etal}{et al.\ }
13: \newcommand{\kms}{\, {\rm km\, s}^{-1}}
14: \newcommand{\ikms}{(\kms)^{-1}}
15: \newcommand{\mpc}{\, {\rm Mpc}}
16: \newcommand{\kpc}{\, {\rm kpc}}
17: \newcommand{\hmpc}{\, h^{-1} \mpc}
18: \newcommand{\ihmpc}{(\hmpc)^{-1}}
19: \newcommand{\hkpc}{\, h^{-1} \kpc}
20: \newcommand{\lya}{Ly$\alpha$ }
21: \newcommand{\lyaf}{Ly$\alpha$ forest}
22: \newcommand{\ch}{\bf change}
23: \newcommand{\gmo}{{\gamma-1}}
24: \newcommand{\bF}{\bar{F}}
25: \newcommand{\hi}{\mbox{H\,{\scriptsize I}\ }}
26: \newcommand{\heii}{\mbox{He\,{\scriptsize II}\ }}
27: \newcommand{\kpa}{k_\parallel}
28: \newcommand{\vk}{{\mathbf k}}
29: \newcommand{\df}{\delta_F}
30: \newcommand{\sF}{{F_s}}
31: \newcommand{\sdelta}{{\delta_s}}
32: \newcommand{\seta}{{\eta_s}}
33: \newcommand{\dt}{\Delta \theta}
34: \newcommand{\dv}{\Delta v}
35: \newcommand{\pa}{\parallel}
36: \newcommand{\pe}{\perp}
37: \newcommand{\dz}{\Delta z}
38: \def\h2{${\rm\,H_2}$}
39: \def\muG{\rm \mu{G}}
40:
41:
42: %%%%% apj5 emulate stuff %%%%%
43: \makeatletter \newenvironment{tablehere} {\def\@captype{table}} {}
44: \newenvironment{figurehere} {\def\@captype{figure}} {}
45: %%%%% apj5 emulate stuff %%%%%
46:
47: \begin{document}
48: \title{Gamma-Ray Bursts May Be Biased Tracers of Star Formation}
49:
50: \author{Renyue Cen\altaffilmark{1} and Taotao Fang\altaffilmark{2}}
51:
52: \begin{abstract}
53:
54: Based on a simulation of galaxy formation in the standard
55: cosmological model, we suggest that a consistent picture
56: for Gamma-Ray Bursts and star formation may be found
57: that is in broad agreement with observations:
58: {\it GRBs preferentially form in low metallicity environments
59: and in galaxies substantially less luminous that $L_*$.}
60: We find that the computed formation rate
61: of stars with metallicity less than $0.1\zsun$
62: agrees remarkably well with the rate evolution
63: of Gamma-Ray Bursts observed by Swift from $z=0$ to $z=4$,
64: whereas the evolution of total star formation rate
65: is weaker by a factor of about $4$.
66: Given this finding, we caution
67: that any inference of star formation rate based
68: on observed GRB rate may require a more involved exercise than
69: a simple proportionality.
70: \end{abstract}
71:
72:
73:
74: \keywords{stars: abundances --- supernovae: general --- galaxies:
75: formation --- cosmology: theory}
76:
77: \altaffiltext{1}{Department of Astrophysical Sciences, Princeton
78: University, Peyton Hall, Ivy Lane, Princeton, NJ 08544}
79:
80: \altaffiltext{1}{Department of Astronomy, University of California,
81: Irvine, CA 92697; {\sl Chandra} Fellow}
82:
83:
84: \section{Introduction}
85:
86: The intriguing observational linkage between long duration Gamma-Ray Bursts (GRBs)
87: and core-collapse supernovae
88: (e.g., Stanek \etal 2003; Hjorth \etal 2003) suggests
89: that the progenitors of GRBs may be very massive stars.
90: This possible connection was predated by a proposed unified picture
91: (Cen 1998).
92: As such, it has been hoped that GRBs may be a good tracer of cosmic
93: star formation (Wijers \etal 1998;
94: Totani 1999;
95: Lamb \& Reichart 2000;
96: Blain \& Natarajan 2000;
97: Porciani \& Madau 2001; Daigne \etal 2006; Coward 2006;
98: Le \& Dermer 2007; Li 2007).
99: However, recent observations indicate that typical GRBs may prefer
100: relatively low metallicity environments ($\sim 0.1\zsun$)
101: (Fynbo \etal 2003;
102: Le Floch \etal 2003;
103: Christensen \etal 2004;
104: Fruchter \etal 2006;
105: Stanek \etal 2006)
106: and host galaxies significantly less luminous than $L_*$
107: (Fruchter \etal 1999,2006),
108: although there is
109: evidence that the actual spread in metallicity may be
110: wide (Berger \etal 2006; Prochaska 2006; Wolf \& Podsiadlowski 2007).
111:
112:
113: The aim of this {\it Letter} is to first address
114: the issue of consistency of GRB environment with respect to
115: metallicity and galaxy luminosity,
116: i.e., the galaxy luminosity-metallicity relation,
117: in the context of detailed simulation of galaxy formation in
118: the standard cosmological model.
119: Then, we make predictions on the evolution of GRB rate with
120: redshift and highlight a possible dramatic difference between overall star formation
121: history and GRB rate history, if GRBs are not an unbiased tracer of star formation.
122: In particular, if GRBs are predominantly produced by stars with
123: metallicity $\le 0.1\zsun$, the GRB rate is expected in our model to
124: rise obstinately from $z=0$ to $z\sim 5$ by a factor of
125: $\sim 100$, when it flattens out towards higher redshift,
126: whereas the overall star formation rate rises rapidly only from $z=0$ to $z\sim 3$
127: and is roughly flat from $z\ge 3$ until $z\sim 7$.
128: The evolution of GRB rate with redshift is thus expected to be
129: stronger than that of star formation.
130:
131:
132: \section{Evolution of GRB and Star Formation Rates}
133:
134: Observations seem to indicate that GRB galaxy hosts are preferentially
135: dwarf galaxies of about $0.1L_*$
136: at low redshift (Fruchter \etal 1999,2006).
137: This would be surprising, if GRB rate
138: is directly proportional to total star formation rate,
139: because the latter is known to peak in significantly larger galaxies
140: (see, for example, Figure 1 below).
141: The implication is that GRB rate is not exactly proportional to
142: the overall star formation rate.
143: On the other hand, analysis of the metallicity of GRB progenitors
144: suggests that GRB progenitors
145: tend to have relatively lower metallicity of $\sim 0.1\zsun$ than
146: that of typical forming stars at low redshift
147: (Fynbo \etal 2003;
148: Le Floch \etal 2003;
149: Christensen \etal 2004;
150: Fruchter \etal 2006;
151: Stanek \etal 2006).
152: We would like to ask the following question:
153: are these two observational facts consistent
154: in the galaxy formation model
155: in the standard cosmological model?
156:
157: Figure 1 shows the distribution of SDSS U-band light
158: as a function of the stellar mass of galaxies,
159: with the galaxies being divided into two subgroups
160: according to the mean metallicity of galaxies.
161: The results are based on a cold dark matter cosmological simulation
162: with galaxy formation
163: in a cold dark matter universe
164: (Nagamine \etal 2006;
165: Cen \& Ostriker 2006; Cen \& Fang 2006)
166: with the following essential parameters:
167: $\Omega_M = 0.31$,
168: $\Omega_b = 0.048$,
169: $\Omega_\Lambda=0.69$, $\sigma_8 = 0.89$,
170: $H_0 = 100 h$km s$^{-1}$ Mpc$^{-1}=69$ km s$^{-1}$ Mpc$^{-1}$
171: and $n_s = 0.97$.
172: The simulation box size is $85 h^{-1}$Mpc
173: comoving with a number of cells of
174: $1024^3$, giving a cell size of 83~h$^{-1}$ kpc comoving
175: and dark matter particle mass equal to
176: $3.9 \times 10^8 h^{-1}\msun$.
177: Given a lower bound of the temperature for almost all the gas in the simulation of
178: $T\sim 10^4$ K, the Jeans mass is
179: $\sim 10^{10}\msun$ for mean-density gas,
180: which is comfortably larger than our mass resolution.
181: Galaxies are produced using a grouping scheme HOP (Eisenstein \& Hut1998)
182: (see Nagamine \etal 2001 for details),
183: which gives a catalog of galaxies each with stellar mass,
184: mean stellar metallicity, luminosities in all SDSS bands, etc.
185:
186:
187: We see a general trend in Figure 1
188: that larger galaxies tend to have higher metallicity,
189: albeit a significant dispersion exists (not shown here).
190: Specifically, it is seen that, if the stellar metallicity
191: is, say, typically lower than $0.1\zsun$ for GRBs progenitors,
192: galaxies of stellar mass $10^{10}\msun$ is expected to make the dominant contribution,
193: whereas the overall star formation rate is peaked in galaxies
194: of stellar mass $10^{11}-10^{12}\msun$.
195: Apparently, this is in broad agreement with observations of GRB
196: being concentrated in dwarf galaxies of metallicity $<0.1\zsun$.
197: This is reassuring in that our simulation that has been shown to produce
198: consistent results compared to the real universe in many other aspects
199: appears to be in broad agreement with observations with regard to
200: the general luminosity-metallicity trend
201: (e.g., Kobulnicky \& Kewley 2004)
202: and with the observed preference of GRB hosts
203: of typical metallicity $\sim 0.1\zsun$ and typical luminosity
204: $\le 0.1L_*$.
205:
206: \begin{figurehere}
207: \begin{center}
208: \resizebox{3.5in}{!}{\includegraphics[angle=90]{lum.ps}}
209: \end{center}
210: \caption{
211: shows the distribution of U band light
212: as a function of the stellar mass of galaxies,
213: for two groups of galaxies,
214: one with metallicity greater than 10\% of solar metallicity (dashed histogram)
215: and the other with metallicity lower than 10\% of solar metallicity (solid histogram).
216: }
217: \label{fig:sfr}
218: \vskip7pt
219: \end{figurehere}
220:
221: Having verified that our simulation
222: is able to reproduced the general luminosity-metallicity relation for galaxies,
223: we will take a step further to infer the rate evolution for GRBs,
224: given the computed star formation rate and stellar metallicity as a function of redshift.
225: Since our simulated galaxies are composed of
226: thousands to millions of ``stellar" particles,
227: which typically resemble globular clusters,
228: we will use the ``resolved" metallicity of individual ``stellar" particles.
229: Usually, there is a wide dispersion in stellar metallicity among
230: the ``stellar" particles within an individual galaxy,
231: reflecting complicated star formation history of each galaxy.
232:
233: Figure 2 shows histories of three rates: total star formation (dotted curve),
234: star formation with metallicity less than $0.3\zsun$ (dashed curve)
235: and less than $0.1\zsun$ (solid curve), respectively.
236: We see that, if GRBs are preferentially hosted by stellar progenitors
237: with metallicity lower than $0.1\zsun$,
238: one should expect to see their rate
239: to rise approximately exponentially as a function of redshift
240: from $z=0$ until $z\sim 5$.
241: This is contrasted with a flattening of the overall star formation rate
242: at a much lower redshift $z\sim 2$ accompanied by only a modest rise (a factor of $2$)
243: from $z\sim 2$ to $z\sim 5$.
244:
245:
246: \begin{figurehere}
247: \begin{center}
248: \resizebox{3.5in}{!}{\includegraphics[angle=90]{sfr.ps}}
249: \end{center}
250: \caption{
251: shows histories of three rates: total star formation (dotted curve),
252: star formation with metallicity less than $0.3\zsun$ (dashed curve)
253: and less than $0.1\zsun$ (solid curve), respectively.
254: }
255: \label{fig:sfr}
256: \vskip7pt
257: \end{figurehere}
258:
259: We note that, while the overall star formation seen in Figure 2
260: is higher by a factor of
261: $\sim 30$ at $z=4$ than at $z=0$,
262: the ratio of GRB rate at $z=4$ to that at $z=0$ is $\sim 120$,
263: if GRB progenitors
264: predominantly have metallicity less than $0.1\zsun$,
265: roughly a factor of four larger.
266: This result is, curiously, in remarkable agreement
267: with recent observations of Kistler \etal (2007) where
268: they suggest that the GRBs observed by Swift
269: has an enhanced evolution by a factor of $\sim 4$ from $z=0$ to $z=4$
270: compared to the overall star formation rate.
271:
272: \begin{figurehere}
273: \begin{center}
274: \resizebox{3.5in}{!}{\includegraphics[angle=90]{modcum.ps}}
275: \end{center}
276: \caption{
277: shows the cumulative distribution of the observed (steps) and predicted (lines) GRB rates, normalized to the total number of bursts between z=0 and z=4. See Fig. 2 for line styles. Data shows the distribution of the 36 $\sl Swift$ GRBs taken from Butler et al.~(2007). The star formation rate of all-galaxy sample was normalized to the
278: analytic values of Hopkins \& Beacom (2006), and the star formation rates for the other two samples was normalized by the same factor.
279: }
280: \label{fig:sfr}
281: \vskip7pt
282: \end{figurehere}
283:
284: In the spirit of making as a direct comparison as possible
285: with observations,
286: we shows the cumulative distribution of the observed (steps) and predicted (curves) GRB rates, normalized to the total number of bursts between $z=0$ and $z=4$ in Figure 3, following Kistler \etal (2007).
287: Here, in order to make a direct comparison with Kistler \etal (2007)
288: and to remove uncertainties in the overall star formation
289: rate in our simulation,
290: we normalize
291: the star formation rate of the all-galaxy sample in our simulation
292: to the analytic values of Hopkins \& Beacom (2006),
293: as did Kistler \etal (2007).
294: Then, the star formation rates for the other two (lower metallicity)
295: samples are adjusted multiplicatively by the same factor.
296: It is seen that,
297: if one simply assumes that
298: the total GRB rate is proportional to
299: the star formation rate of metallicity less than $0.1\zsun$,
300: the observed evolution of GRB rate from $z=0$ to $z=4$ is reproduced
301: to a high degree,
302: clearly visible in the good agreement between the solid curve
303: and the step curve in Figure 3.
304:
305: \section{Discussion}
306:
307: Taken all the observational facts together along with
308: our theoretical results,
309: a broadly consistent picture appears to emerge:
310: {\it GRBs preferentially form in low metallicity environments
311: and (because of the luminosity-metallicity relation)
312: in galaxies substantially less luminous that $L_*$,
313: and GRB rate evolves more strongly than the overall star formation rate.}
314: Evidently,
315: the observed evolution of GRB rate from $z=0$ to $z=4$
316: can be explained, {\it if
317: GRBs are primarily produced by massive
318: stars with metallicity less than $0.1\zsun$}.
319: Including higher metallicity stars
320: would produce GRB rate evolution from
321: $z=0$ to $z=4$ that is inconsistent with observations.
322: Nevertheless, this overall picture would also be consistent
323: with a theoretical preference or possibly requirement
324: of low metallicity for GRB progenitors
325: in the context of ``collapsar" models (MacFayden \& Woosley 1999;
326: Woosley \& Heger 2006).
327:
328: What is implicitly assumed is that the stellar
329: initial mass function (IMF)
330: has remained the same over the redshift range considered.
331: In other words, whatever metallicity dependence
332: of GRB rate may have,
333: this dependence is assumed not to evolve with redshift.
334: One should note that it is not fully known observationally or
335: understood theoretically how the IMF evolves with time.
336: Therefore, additional possible evolutionary effect of IMF
337: would add another layer of complexity to this issue.
338: It is often thought
339: that lower metallicity environment might
340: favor formation of more massive stars.
341: If GRB progenitors are massive stars,
342: this would then translate to the expectation
343: that additional effect due to an evolving IMF
344: may further steepen the evolution of the GRB rate with redshift.
345: This, however, is not required or borne out
346: in our analysis.
347: Our results thus imply that
348: the evolution of IMF from $z=4$ to $z=0$, if any,
349: appears to be weak.
350:
351: If we place this result in a larger context of
352: star formation over the entire cosmic history,
353: one might come to the conclusion that,
354: while there may be a
355: dramatic transition of IMF from
356: Population III metal-free stars
357: (Nakamura\& Umemura, M. 2002;
358: Abel, Bryan \& Norman 2002;
359: Bromm, Coppi, \& Larson 2002)
360: to Population II stars at some high redshift
361: (e.g., Cen 2003; Trac \& Cen 2007),
362: further evolution of IMF at lower redshift may be modest,
363: in the sense that the mass fraction of high mass stars
364: that are presumably GRB progenitors of the total
365: stellar mass remains relatively constant.
366:
367:
368: \section{Conclusions}
369:
370: We utilize a simulation of galaxy formation in the standard
371: cosmological model that has been shown to produce results
372: consistent with extant observations of galaxy formation (e.g., Nagamine \etal 2006)
373: to shed light on the relation between GRB rate and star formation rate.
374: We find that a consistent picture
375: for Gamma-Ray Bursts and star formation
376: that is in broad agreement with observations
377: would emerge,
378: {\it if GRBs preferentially form in low metallicity environments
379: and in galaxies substantially less luminous that $L_*$.}
380: Because of the increase of metallicity with cosmic time,
381: GRB rate consequently evolves more strongly
382: with redshift than the overall star formation rate.
383: We find that
384: the observed evolution of GRB rate from $z=0$ to $z=4$
385: can be explained, {\it if
386: GRBs are primarily produced by massive
387: stars with metallicity less than $0.1\zsun$},
388: whereas an inclusion of stars with metallicity as high as $0.3\zsun$
389: yields GRB rate evolution from
390: $z=0$ to $z=4$ inconsistent with observations.
391:
392: Therefore, we reach the conclusion
393: that GRBs may not be a good tracer of cosmic star formation,
394: especially over a long timeline.
395: As a result, a simple inference of star formation rate
396: or its derived quantities such as the ionizing photon production rate at high redshifts,
397: based on the observed GRB rate,
398: should be done with caution and may require careful calibrations.
399:
400:
401: \smallskip
402: \acknowledgements
403: {We thank Ken Nagamine for providing simulated galaxy catalogs.
404: We gratefully acknowledge financial support by
405: grants AST-0407176 and NNG06GI09G.}
406:
407:
408: \begin{thebibliography}{99}
409:
410: \bibitem[Abel \etal (2002)]{a02}Abel, T., Bryan, G.L., \& Norman, M.L., 2002, Science, 295, 93
411:
412:
413: \bibitem[Berger \etal (2006)]{b06} Berger, E. \etal 2006, ApJ, 642, 979
414:
415: \bibitem[Blain \& Natarajan (2000)]{bn00} Blain, A.W., \& Natarajan, P. 2000, MNRAS, 312, L35
416:
417: \bibitem[Bromm \etal (2000)]{bn00}Bromm, V., Coppi, P.S., \& Larson, R.B. 2002, ApJ, 564, 23
418:
419: \bibitem[Butler \& Natarajan (2007)]{bn00} Butler, N.R., Kocevski, D., Bloom, J.S. \& Curtis, J.L., arXiv:0706.1275, ApJ in press
420:
421: \bibitem[Cen (1998)]{c98} Cen, R., 1998, ApJ, 507, L131
422:
423: \bibitem[Cen (2003)]{c03} Cen, R., 2003, ApJ, 591, 12
424:
425: \bibitem[Cen \& Fang (2006)]{cf06} Cen, R., \& Fang, T. 2006, ApJ, 650, 573
426:
427: \bibitem[Cen \& Ostriker (2006)]{co06} Cen, R., \& Ostriker, J.P. 2006, ApJ, 650, 560
428:
429: \bibitem[Christensen \etal (2004)]{c04} Christensen, \etal 2004;
430:
431: \bibitem[Coward(2007)]{2007NewAR..51..539C} Coward, D.\ 2007, New Astronomy Review, 51, 539
432:
433: \bibitem[Daigne et al.(2006)]{2006MNRAS.372.1034D} Daigne, F., Rossi, E.~M., \& Mochkovitch, R.\ 2006, \mnras, 372, 1034
434:
435: \bibitem[Eisenstein and Hut(1998)]{Eisenstein98} Eisenstein, D. J.~and~Hut, P. 1998, \apj, 498, 137
436:
437: \bibitem[Fruchter \etal (1999)]{f99} Fruchter, A.S. \etal 1999,
438:
439: \bibitem[Fruchter \etal (2006)]{f06} Fruchter, A.S. \etal 2006, Nature, 441, 463
440:
441: \bibitem[Fynbo \etal (2003)]{f03} Fynbo, J.P.U., \etal 2003, A\&A, 406, L63
442:
443: \bibitem[Hjorth \etal (2003)]{h03} Hjorth, J., \etal 2003, Nature, 423, 847
444:
445: \bibitem[Jakobsson \etal (2006)]{j06} Jakobsson, P., \etal 2006, A\&A, 443, 897
446:
447: \bibitem[Kobulnicky \& Kewley (2004)]{kk04} Kobulnicky, H.A., \& Kewley, L.J. 2004, ApJ, 617, 240
448:
449: \bibitem[Lamb \& Reichart (2000)]{lr00} Lamb, D.Q., \& Reichart, D.E. 2000, ApJ, 536, 1
450:
451: \bibitem[Le \& Dermer(2007)]{2007ApJ...661..394L} Le, T., \& Dermer, C.~D.\ 2007, \apj, 661, 394
452:
453: \bibitem[Le Floch \etal (2003)]{le03} Le Floch, E., 2003, A\&A, 400, 499
454:
455: \bibitem[Li (2003)]{l07} Li, L.-X. 2007, arXiv:0710.3587
456:
457: \bibitem[MacFayden \& Woosley (1999)]{mw99} MacFayden, A.I., \& Woosley, S.E. 1999, ApJ, 524, 262
458:
459: \bibitem[Nagamine \etal (2006)]{n06} Nagamine, K., Fukugita, M., Cen, R., \& Ostriker, J.P. 2001, ApJ, 588, 497
460:
461: \bibitem[Nagamine \etal (2006)]{n06} Nagamine, K., Ostriker, J.P., Fukugita, M., \& Cen, R. 2006, ApJ, 633, 881
462:
463: \bibitem[Nakamura \& Umemura (2002)]{n06}Nakamura, F., \& Umemura, M. 2002, ApJ, 569, 549
464:
465: \bibitem[Porciani \& Madau (2001)]{pm01} Porciani, C., \& Madau, P. 2001, ApJ, 548, 522
466:
467: \bibitem[Prochaska (2006)]{p06} Prochaska, J.X. 2006, ApJ, 650, 272
468:
469: \bibitem[Savaglio, Glazebrook, \& Le Borgne (2006)]{sgl06} Savaglio, S., Glazebrook, K., \& Le Borgne, D. 2006, in AIP Conf. Proc. Vol 836, Gamma-Ray Bursts in the Swift Era, eds. S. Holt, N. Gehrels, J. Nousek, Am. Inst. Phys., Melville, NY, p540
470:
471: \bibitem[Stanek \etal (2003)]{s03} Stanek, K.Z., \etal 2003, ApJ, 591, L17
472:
473: \bibitem[Stanek \etal (2006)]{s06} Stanek, K.Z., \etal 2006, Acta Astron. 56, 333
474:
475: \bibitem[Totani (1999)]{t99} Totani, T. 1999, ApJ, 511, 41
476:
477: \bibitem[Trac \& Cen (2007)]{tc07} Trac, H., \& Cen, R. 2007, astro-ph/0612406, ApJ, in press
478:
479: \bibitem[Wijers \etal (1998)]{w98} Wijers, R.A., \etal 1998, MNRAS, 294, L13
480:
481: \bibitem[Woosley \& Heger (2006)]{wh06} Woosley, S.E., \& Heger, A. 2006, ApJ, 637, 914
482:
483: \end{thebibliography}
484:
485: \end{document}
486:
487:
488:
489: