1: \documentstyle[procps]{kapproc} %For PostScript fonts
2: \begin{document}
3:
4: \articletitle{What fraction of stars formed in infrared galaxies
5: at high redshift?}
6:
7: \author{Neil Trentham}
8: \affil{Institute of Astronomy, University of Cambridge\\
9: Cambridge CB3 0HA, United Kingdom}
10: \email{trentham@ast.cam.ac.uk}
11:
12: \begin{abstract}
13: Star formation happens in two types of environment:
14: ultraviolet-bright starbursts (like 30 Doradus and HII
15: galaxies at low redshift
16: and Lyman-break galaxies at high redshift) and
17: infrared-bright dust-enshrouded regions (which may be moderately
18: star-forming
19: like Orion in the Galaxy or extreme like the core of Arp 220).
20: In this work I will estimate how many of the stars in the local
21: Universe formed in each type of environment, using observations
22: of star-forming galaxies at all redshifts
23: at different wavelengths
24: and of the evolution of the field galaxy population.
25: \end{abstract}
26:
27: \begin{keywords}
28: Galaxies, Cosmology
29: \end{keywords}
30:
31: \section{Introduction}
32:
33: It is now possible to estimate
34: the star-formation history of the Universe.
35: This is performed most directly by summing the contributions from
36: star-forming field galaxies in optical (corresponding to rest-frame
37: ultraviolet at high redshift) surveys.
38: The direct contribution from ultraviolet-bright star-forming
39: galaxies to the
40: comoving star-formation rate density is about
41: $3 \times 10^{-3} \, {\rm M}_{\odot}\,{\rm yr}^{-1}$ ${\rm Mpc}^{-3}$
42: at redshift $z=0$, rising to
43: $4 \times 10^{-2} \, {\rm M}_{\odot}\,{\rm yr}^{-1} \, {\rm Mpc}^{-3}$
44: at $z=1$, before slowly declining to
45: $1.5 \times 10^{-2} \, {\rm M}_{\odot}\,{\rm yr}^{-1} \, {\rm Mpc}^{-3}$
46: at $z=6$ ($h=0.7$, $\Omega_{\Lambda}=0.7$,
47: $\Omega_{\rm m}=0.3$; Salpeter IMF; ref.~1).
48: This is normally presented in the
49: uncorrected form of the ``Madau'' or ``Madau-Lilly'' plot.
50:
51: But galaxies that are forming stars also experience significant dust
52: extinction -- we know this because
53: we see that local star-forming regions like Orion are dusty and
54: because local spiral galaxies have spectral energy distributions
55: (SEDs) which peak in the far-infrared.
56: Correcting for this, Giavalisco et al.~[1] find that the
57: the total contribution from
58: optically selected galaxies to the comoving star-formation rate
59: density is about
60: $1.3 \times 10^{-2} \, {\rm M}_{\odot}\,{\rm yr}^{-1} \, {\rm Mpc}^{-3}$
61: at $z=0$, rising to
62: $0.13 \, {\rm M}_{\odot}\,{\rm yr}^{-1} \, {\rm Mpc}^{-3}$
63: at $z=1$, where it stays roughly constant out to at least $z=6$.
64: The corrections used come from the analysis of Adelberger \&
65: Steidel (ref.~2), which is based on multi-wavelength studies of
66: a large sample of star-forming
67: galaxies.
68:
69: An additional contribution may comes from extremely dusty galaxies
70: where the dust is optically thick -- these may be missing altogether
71: from optical surveys. Local ultraluminous infrared
72: galaxies (ULIGs) are examples of this
73: kind of galaxy; the $V$ extinction
74: to the core of Arp 220 is $>10$ mag [3].
75: Another example is the host of GRB 010222 [4], which is an
76: optical sub-$L^*$ galaxy but has a submillimetre star-formation rate of
77: $\sim 600$ M$_{\odot}$ yr$^{-1}$. This kind of galaxy may be similar to
78: the SCBUA galaxies [5] seen in submillimetre surveys
79: which might have a redshift distribution quite
80: different from galaxies in optically selected samples.
81:
82: In this work I assess the contributions from all three of these
83: modes of star formation in
84: generating the current cosmological
85: density in stars $\Omega_*$.
86: I provide estimates given current observations and outline how
87: future observations may provide stronger constraints.
88:
89: \section{Definitions}
90:
91: The total density in stars in critical units is
92: $\Omega_* = \Omega_*^{\rm UV} + \Omega_*^{\rm IR/Opt} + \Omega_*^{\rm IR}$
93:
94: The density of stars seen forming directly in optical galaxies $\Omega_*^{\rm UV}$
95: equals the integral of the uncorrected ultraviolet star formation rate density
96: over redshift.
97: The density of stars that formed in dusty regions within those
98: same galaxies $\Omega_*^{\rm IR/Opt}$
99: equals he integral of the ultraviolet star formation rate density
100: multiplied by an extinction correction (which may depend on $z$) over redshift.
101: Finally, the density of stars $\Omega_*^{\rm IR}$
102: that formed in highly obscured regions whose presence
103: cannot be inferred from optical observations equals to the infrared star
104: formation rate density (as measured by SCUBA or in the future ALMA)
105: minus the contributions to the previous integral,
106: integrated over redshift.
107:
108: The partition between $\Omega_*^{\rm IR/Opt}$ and $\Omega_*^{\rm IR}$ is somewhat
109: arbitrary.
110: Here we put in $\Omega_*^{\rm IR}$
111: all star formation in galaxies whose bolometric luminosity
112: is greater than some threshold luminosity corresponding to that at which the
113: optical luminosity no longer tracks the bolometric (mainly far-infrared) luminosity;
114: locally this happens at a 60-$\mu$m luminosity of 6.3 $\times 10^{10}
115: \, {\rm L}_{\odot}$ (ref.~6).
116: The extinction in these luminous galaxies surely comes from optically thick dust.
117: Optically thin extinction would tend to happen in galaxies
118: contributing to $\Omega_*^{\rm IR/Opt}$.
119:
120: Star formation that happened in ULIGs would be included in $\Omega_*^{\rm IR}$.
121:
122: The fraction of star formation that is obscured in any particular galaxy
123: varies from 0\% to about 99\% [2].
124: Obscured star formation at the low end of this range mostly
125: ontributes to $\Omega_*^{\rm IR/Opt}$
126: and at the high end to $\Omega_*^{\rm IR}$.
127: The host of GRB 010222 would be at the very high end of
128: this range.
129:
130: \section{The current cosmological stellar content: what
131: needs to be produced}
132:
133: The time integral of the cosmic star formation rate must equal the luminosity
134: integral of the galaxy luminosity function:
135: $$\Sigma_i \int_L \, L \, \phi_i(L) \, \Gamma_i \, {\rm d}L =
136: \int_{t(z)} \dot{\rho_*} \, {\rm d}t,
137: $$
138: where $\Gamma_i$ is the mass-to-light ratio of stellar population $i$ (this is derived from
139: stellar population synthesis models).
140: From the combination of the SDSS survey measurements at the bright end [7] and CCD mosaic
141: surveys [e.~g.~ref.~8] at the faint end, the galaxy luminosity function appears to be
142: well-described
143: by a Schechter function with $M_R^* = -22.0$ and
144: $\alpha^*=-1.28$ brightward of $M_R=-19$ and a power law with
145: $\alpha=-1.24$ faintward of $M_R=-19$.
146: Performing the sum of integrals on the LHS of this equation,
147: $$\Omega_* = 0.0036 \pm 0.0020$$
148: in units of the critical density.
149: About 3/4 of this is in spheroids and 1/4 in disks.
150: If a Salpeter, not KTG [9] IMF is used,
151: $\Omega_*$
152: is a factor of two higher.
153:
154: \section{Observational Constraints}
155:
156: \subsection{Field Galaxy Evolution}
157:
158: Multi-colour photometry of a near-infrared selected sample of galaxies has permitted
159: Dickinson, Papovich and colleagues [10] to measure the evolution of $\Omega_*$ with redshift.
160: Between $z=0$ and $z=1$ they found that about 40\% of the present-day stars in the Universe
161: formed. Between $z=1$ and $z=2$, their best-fitting models suggest that a further fraction
162: $>50$\% formed and only a few percent of the stars that we currently see had formed by
163: $z=2$. However, their is considerable uncertainty in the star-formation histories
164: used in modelling the SEDs of the sample galaxies and the fraction
165: of stars in place by $z=2$ may be as high as 25\%. An integral of the extinction-corrected
166: Madau Plot [1] over redshift from $z=\infty$ up to $z=2$ gives a fraction of about 25\%.
167:
168: Near-infrared surveys [11] have shown that about 30\% of the massive early-type galaxies seen today
169: were already in place by $z=2$ -- these are the distant red galaxies (DRGs). Therefore most of the
170: stars that we believe formed at $z>2$ are not only in massive galaxies today {\it but were
171: already in massive galaxies by} $z=2$. If these formed within the large galaxies, they
172: would have had to form in very extreme bursts.
173:
174: \subsection{GRB host galaxies}
175:
176: Long-duration gamma-ray bursts (GRBs)
177: are thought to be linked to the deaths of
178: massive stars, as suggested by the
179: coincident between SN 2003dh and GRB 030329 [12]. Since massive stars do not live long,
180: this opens up the possibility of using GRB host galaxies as a SFR-selected
181: sample of galaxies.
182:
183: There are, however, complications.
184: In the context of a collapsar [13] model, GRBs originate preferentially
185: from stars of high mass and low metallicity. This means that there will be
186: more GRBs per unit SFR in galaxies which have a high-mass biassed IMF
187: or a low metallicity. Perhaps these two effects
188: explain the preponderance of GRBs in ULIGs at $z \sim 1$ [14] and
189: Ly$\alpha$-emitters [ref.~15; see also ref.~16
190: about the very
191: low metallicity and high gas column
192: density of the host of GRB030323],
193: neither of which significantly contribute to $\Omega_*$.
194:
195: \subsection{Infrared and submillimetre backgrounds and counts}
196:
197: The extragalactic background light (EBL) is high at infrared wavelength:
198: $COBE$/DIRBE measured it as $32 \pm 13$ nW m$^{-2}$ sr$^{-1}$ at 140 $\mu$m
199: and $17 \pm 4$ nW m$^{-2}$ sr$^{-1}$ at 240 $\mu$m [17].
200: The submillimetre background is somewhat lower:
201: $0.55 \pm 0.15$ nW m$^{-2}$ sr$^{-1}$ at 850 $\mu$m [18].
202: The optical extragalactic background light also appears to be
203: high: 12/15/18 nW m$^{-2}$ sr$^{-1}$ at 300/550/800 nm,
204: with an uncertainty of 50\% [19].
205:
206: Madau \& Pozzetti [20] estimated the total EBL as
207: $55 \pm 20$ nW m$^{-2}$ sr$^{-1}$. This is lower than the
208: value of $100 \pm 20$ nW m$^{-2}$ sr$^{-1}$ quoted by Bernstein
209: et al.~[21], the main difference being an additional component
210: from the optical background that was previously undetected (however see
211: ref.~22).
212:
213: Most of this background is generated by stars, not AGN, else the local
214: density of supermassive black holes would be overproduced [20].
215: Assuming a recycling fraction of 0.4 (much of the material in stars is
216: returned to the ISM via winds and supernovae; ref.~23), the EBL
217: implies a stellar density of $\Omega_* = 0.003 - 0.006$, consistent
218: with the number in Section 3.
219:
220: One reason the range here is quite large is that the redshift distribution
221: of infrared star-forming galaxies is unknowm, and the contribution of
222: each galaxy to the EBL $\propto (1+z)^{-1}$.
223:
224: The submillimetre background has been resolved and redshifts determined for a
225: number of bright sources [24]. However, there are indications [25] that the galaxies which
226: dominate the infrared background are a different population to the galaxies
227: which dominate the submillimetre background.
228: The models described by Chary et al.~[25] suggest that these infrared galaxies
229: have lower bolometric luminosities and lower redshifts than the ULIGs observed by
230: Chapman et al.~[24].
231:
232: \subsection{Optical/Infrared observations of star-forming galaxies}
233:
234: The instantaneous
235: cosmic SFR at any redshift equals the infrared and optical contributions.
236: Making extinction corrections to
237: star formation rates for high-redshift galaxies is a substitute for direct
238: measurement at infrared wavelengths. Only at low redshift $z<1$ can the two be measured
239: directly.
240: The CFRS+$ISO$ survey [26] showed that about (i) 30\% of star formation at $z<1$,
241: was visible at optical wavelengths, (ii) most of the remaining 70\% that was in infrared galaxies
242: was in disturbed systems with red $I-K$ colours, and (iii) about 18\% of the
243: star formation happened in ULIGs with SFRs in excess of 100 M$_{\odot}$ yr$^{-1}$.
244: An implication of these results
245: is that most of the star formation at $z<1$ happened in infrared galaxies that are neither optical
246: galaxies with high internal extinction nor ULIGs of the kind seen
247: by SCUBA.
248:
249: The first $Spitzer$ results [25] seem to point towards a similar situation at higher redshifts.
250: The main difference between low and high redshift is that while the redshift distributions of
251: optical and infrared galaxies are similar {\it within} the $0<z<1$ range, they are very different
252: at high redshift.
253:
254: Another implication of these results is that in optically-selected star-forming
255: galaxies, most of the star formation is being observed directly and these are
256: at the low end of the obscuration range described by Adelberger and
257: Steidel [2].
258: Many of these may be blue compact emission-line galaxies of the type described at this
259: conference by Lowenthal and Bershady.
260:
261: \section{The IMF and density in infrared galaxies}
262:
263: The local star-forming region 30-Doradus has a stellar IMF that
264: is Salpeter ($\propto m^{-2.35}$) above
265: 3 M$_{\odot}$ [27].
266: Lyman-break galaxies at high redshift also have
267: Salpeter IMFs at high masses, an inference based on optical
268: spectroscopy of cB58 at
269: $z=2.7$ [28].
270: This IMF is attractive in that
271: it evolves into the KTG [9] IMF seen locally [29].
272: Unfortunately, no equivalent analysis can be made for infrared
273: galaxies and it remains a possibility that they have a different
274: IMF -- if it is high-mass biassed then they will generate
275: more energy per unit mass of stars formed than will optical
276: galaxies.
277:
278: Stars that form in infrared galaxies probably form in
279: dense star clusters which need to dissipate in order to
280: produce local galaxies.
281: Dissipation timescales are long for dense clusters but
282: are shorter if the cluster is
283: embedded in a gaseous medium.
284: Simulations of this physical process [30], along with
285: inferences about the star-formation history of the Universe from
286: stellar populations of nearby galaxies [31]
287: will provide additional constraints on the total amount
288: of cosmic star formation that occurred in infrared galaxies
289: and when it happened.
290:
291: \section{Concluding thoughts}
292:
293: My current thinking is that
294: (i) roughly equal amounts of stars formed in each of the following four types
295: of environments: optically
296: visible regions, dust-enshrouded regions within optical
297: galaxies, heavily obscured galaxies with $L_{\rm IR} < 10^{12}$ L$_{\odot}$
298: and ULIGs with $L_{\rm IR} < 10^{12}$ L$_{\odot}$, and (ii)
299: most star formation in optically visible regions was in small galaxies
300: at all redshifts [32, 33] while most star formation in ULIGs
301: happened at high redshift, perhaps producing the DRGs.
302: But these are not strongly held convictions
303: and I share the optimism felt at this conference that the
304: puzzle of star formation in galaxies
305: will be solved over the next few years.
306:
307:
308: \begin{chapthebibliography}{1}
309:
310: \bibitem[\protect\citename{bl}%
311: ]{giav04}
312: [1] Giavalisco M.~et al., 2004, ApJ, 600, L103
313:
314: \bibitem[\protect\citename{bl}%
315: ]{as00}
316: [2] Adelberger K.~L., Steidel C.~C., 2000, ApJ, 544, 218
317:
318: \bibitem[\protect\citename{bl}%
319: ]{scov98}
320: [3] Scoville N.~Z.~et al., 1998, ApJ, 492, L107
321:
322: \bibitem[\protect\citename{bl}%
323: ]{frail02}
324: [4] Frail D.~A.~2002, ApJ, 565, 829
325:
326: \bibitem[\protect\citename{bl}%
327: ]{blainrev}
328: [5] Blain A.~W., Smail I., Ivison R.~J., Kneib J.-P., Frayer D.~T.,
329: 2002, Physics Reports, 369, 111
330:
331: \bibitem[\protect\citename{bl}%
332: ]{rl96}
333: [6] Rieke G.~H., Lebofsky M.~J., 1986, ApJ, 304, 326
334:
335: \bibitem[\protect\citename{bl}%
336: ]{b07}
337: [7] Blanton M.~R.~et al., 2001, AJ, 121, 2358
338:
339: \bibitem[\protect\citename{bl}%
340: ]{tt}
341: [8] Trentham N., Tully R.~B., 2002, MNRAS, 335, 7
342:
343: \bibitem[\protect\citename{bl}%
344: ]{ktg}
345: [9] Kroupa P., Tout, C.~A., Gilmore, G., 1993, MNRAS, 262, 545
346:
347: \bibitem[\protect\citename{bl}%
348: ]{dp03}
349: [10] Dickinson M., Papovich C., Ferguson H.~C., Budavari T., 2003,
350: ApJ, 587, 25
351:
352: \bibitem[\protect\citename{bl}%
353: ]{dp03}
354: [11] Glazebrook K.~et al., 2004, Nature, 403, 181
355:
356: \bibitem[\protect\citename{bl}%
357: ]{hjorth030329}
358: [12] Hjorth J.~et al., 2003, Nature, 423, 847
359:
360: \bibitem[\protect\citename{bl}%
361: ]{mw99}
362: [13] MacFadyen A.~I., Woosley S.~E., 1999, ApJ, 524, 262
363:
364: \bibitem[\protect\citename{bl}%
365: ]{berg03}
366: [14] Berger E.~et al.,
367: 2003, ApJ, 588, 99
368:
369: \bibitem[\protect\citename{bl}%
370: ]{vreeslya}
371: [15] Fynbo J.~P.~U.~et al., 2003, A\&A, 406, L63
372:
373: \bibitem[\protect\citename{bl}%
374: ]{dylagrb}
375: [16] Vreeswijk P.~M.~et al., 2004, A\&A, 419, 927
376:
377: \bibitem[\protect\citename{bl}%
378: ]{Sch}
379: [17] Schlegel D.\,J., Finkbeiner D.\,P., Davis M., 1998, ApJ, 500, 525
380:
381: \bibitem[\protect\citename{bl}%
382: ]{Fix}
383: [18] Fixsen D.\,J., Dwek E., Mather J.\,C., Bennett C.\,L., Shafter R.\,A.,
384: 1998, ApJ, 508, 123
385:
386: \bibitem[\protect\citename{bl}%
387: ]{bfm1}
388: [19] Bernstein R.~A., Freedman W.~L., Madore B.~F., 2002, ApJ, 571, 56
389:
390: \bibitem[\protect\citename{bl}%
391: ]{mp00}
392: [20] Madau P., Pozzetti L., 2000, 312, L9
393:
394: \bibitem[\protect\citename{bl}%
395: ]{bfm1}
396: [21] Bernstein R.~A., Freedman W.~L., Madore B.~F., 2002, ApJ, 571, 107
397:
398: \bibitem[\protect\citename{bl}%
399: ]{matilla}
400: [22] Matilla K., 2003, ApJ, 591, 119
401:
402: \bibitem[\protect\citename{bl}%
403: ]{cole00}
404: [23] Cole S., Lacey C.~G., Baugh C.~M., Frenk C.~S., 2000, MNRAS, 319, 168
405:
406: \bibitem[\protect\citename{bl}%
407: ]{chscuba}
408: [24] Chapman S.~C., Blain A.~W., Ivison R.~J., Smail I.~R.,
409: 2003, Nature, 422, 695
410:
411: \bibitem[\protect\citename{bl}%
412: ]{chary04}
413: [25] Chary R.~et al., 2004, ApJS, 154, 80
414:
415: \bibitem[\protect\citename{bl}%
416: ]{floresiso}
417: [26] Flores H.~et al., 1999, ApJ, 517, 148
418:
419: \bibitem[\protect\citename{bl}%
420: ]{s99}
421: [27] Selman F., Melnick J., Bosch G., Terlevich R., 1999, A\&A, 347, 532
422:
423: \bibitem[\protect\citename{bl}%
424: ]{cb58}
425: [28] Pettini M.~et al.,
426: 2000, ApJ, 528, 96
427:
428: \bibitem[\protect\citename{bl}%
429: ]{kw03}
430: [29] Kroupa P., Weidner C., 2003, ApJ, 598, 1076
431:
432: \bibitem[\protect\citename{bl}%
433: ]{lamers}
434: [30] Lamers H.~J.~G.~L.~M., Gieles M., Portegeis Zwart S.~F., 2004,
435: astro-ph/0408235
436:
437: \bibitem[\protect\citename{bl}%
438: ]{h04a}
439: [31] Heavens A., Panter B., Jimenez R., Dunlop J., 2004, Nature, 428, 625
440:
441: \bibitem[\protect\citename{bl}%
442: ]{cbelg}
443: [32] Guzm{\'{a}}n R.~et al., 2003, ApJ, 586, L45
444:
445: \bibitem[\protect\citename{bl}%
446: ]{warren}
447: [33] Weatherley S.~J., Warren S.~J., 2003, MNRAS, 345, L29
448:
449:
450: \end{chapthebibliography}
451:
452:
453: \end{document}
454: