astro-ph0511452/ms.tex
1: %\documentstyle[12pt,aaspp4]{article}
2: %\documentstyle [12pt,aasms4]{article}
3: %\documentstyle [emulateapj]{article}
4: %\documentstyle [aaspp4]{article}
5: \documentclass [12pt,preprint]{aastex}
6: \begin{document}
7: 
8: \title{Early Cosmic Chemical Evolution: Relating the Origin of a
9: Diffuse Intergalactic Medium and the First Long-Lived Stars}
10: 
11: \author{F.D.A. Hartwick}
12: 
13: \affil{Department of Physics and Astronomy, \linebreak University of 
14: Victoria,
15: Victoria, BC, Canada, V8W 3P6}
16: \begin {abstract}
17: 
18: Nucleosynthetic signatures in common between the gas responsible for the
19: high redshift Lyman $\alpha$ forest and a subsample of extremely metal
20: poor stars are found.  A simple mass loss model of chemical evolution with
21: physically motivated parameters provides a consistent picture in which the
22: gas is identified with that lost by supernova-driven winds during the first 
23: generation of
24: star formation.  Substantial mass loss occurs which can account for a
25: diffuse IGM with up to $80\%$ of the total baryon content and a peak
26: [C-O/H] abundance of $\sim-2.9$.  This mass loss component differs from one
27: produced later during galaxy formation and evolution that contributes to
28: a circum-galactic medium (CGM).  The CGM was shown earlier to 
29: have a mass of $\sim10\%$ of all baryons and peak [Fe/H]$\sim-1$. 
30: 
31: \end{abstract}
32: 
33: \keywords{intergalactic medium}
34: 
35: \section {Introduction}
36: 
37: It is well known that the stellar content of the universe is only about
38: $\sim10\%$ of the total baryonic content (e.g. Fukugita \& Peebles 2004).
39: It is also well established that vigorous star formation, which occurs
40: during galaxy formation and evolution, is accompanied by mass loss (e.g.
41: Veilleux et al. 2005).  According to some models this mass loss can account
42: for another $\sim10\%$ of all baryons (e.g. Hartwick 2004, hereafter H04).
43: What is the origin and composition of the remaining $\sim80\%$ of all
44: baryons? This question is addressed here using recent observational data
45: and simple chemical evolution arguments. 
46: 
47: Generally the abundance distribution exhibited by stellar spectra reflects
48: the material that formed the stars.  If one were to observe spectra of the
49: oldest stars (most metal poor?) in our Galaxy, then one expects to have
50: to look at high redshifts in order to find gas with the same
51: nucleosynthetic signature. 
52: 
53: At redshifts of order 3, the baryon budget is believed to be dominated
54: by the tenuous gas responsible for producing the Lyman $\alpha$ forest
55: observed in absorption in high redshift quasar spectra (e.g. Rauch 1998).
56: As discussed below, recent high quality data from the largest telescopes
57: combined with sophisticated reduction techniques have allowed the
58: detection of the heavy elements C, O, and Si in low column densities of 
59: hydrogen.  These new observational results have prompted a re-examination of a
60: possible connection between this gas and the earliest star formation
61: (Pop~III?) phase.  By modelling the observed distribution of carbon or
62: oxygen in this gas, we can predict the distribution of these elements and
63: characteristic abundance ratios expected in long-lived stars, if they had
64: been formed at the same early time.  Unfortunately both the gas and
65: stellar samples, are small and incomplete.  Yet, we find encouraging
66: consistency with the hypothesis when comparing these observations
67: with the results from a simple, physically motivated, mass loss model of
68: chemical evolution.  An important and surprising conclusion
69: is that long-lived stars with nucleosynthetic signatures in common 
70: with the Lyman $\alpha$ forest gas do exist. 
71: 
72: \section{The Observations}
73: 
74: \subsection{The Diffuse Gas}
75: 
76: Early work on the detection of carbon in low column density Lyman $\alpha$
77: clouds was carried out by Cowie, Songaila, Kim \& Hu (1995).  More recently, 
78: Simcoe, Sargent \& Rauch (2004) used survival analysis (a technique which makes
79: use of non-detections as well as detections) to construct the distribution
80: of [O/H] and [C/H] in low column density absorption lines along lines of
81: sight to seven quasars.  The data become incomplete at [O/H]$\lesssim-3.0$
82: (no floor was detected), but above that the distribution was log-normal
83: with mean [C-O/H]$\sim-2.85$ and dispersion $\pm0.75$, with these results
84: depending on the assumed shape of the UV background (UVB).  A second well
85: fit model used a softer UVB resulting in medians of [C/H]$\sim-3.1$ and
86: [O/H]$\sim-2.7$.  The ionizing UVB models are those of Haardt \& Madau
87: (1996) and are made up of contributions from a quasar continuum and a galaxy 
88: contribution.  A softer spectrum implies more contribution
89: from galaxies and less from quasars. 
90: 
91: The information in Fig.\ 14 of Simcoe, Sargent \& Rauch (2004) is the
92: starting point for the present work.  A second important observational
93: constraint comes from Schaye et al.\ (2003) who determined the
94: silicon to carbon ratio [Si/C] in the Lyman $\alpha$ forest down to very
95: low column densities.  They found this ratio to be $\sim0.77$, again
96: depending on the shape of the UVB that is assumed, with a softer UVB (more
97: galaxy and less quasar contribution) resulting in a lower [Si/C] ratio. 
98: The authors concluded that this [Si/C] ratio is unlikely to be lower than
99: $\sim0.5$. 
100: 
101: \subsection{The Stellar Sample}
102: 
103: When the chemical evolution model is constrained by observations of the
104: diffuse gas, one can predict the [C-O/H] distribution in stars, assuming
105: that luminous long-lived remnants were produced.  Further, the stars should
106: also show appropriate [Si/C] signatures.  If such stars were formed, we can
107: reasonably expect to find them among the various samples of extremely
108: metal poor stars which have been spectroscopically analyzed. The data 
109: discussed below follows from 
110: earlier work by Gratton \& Sneden (1988, 1991), McWilliam et al.\ (1995), 
111: Ryan, Norris \& Beers (1996), and others. 
112: 
113: Here we are concerned primarily with the abundances of C, O, Mg, Si, and
114: Fe.  Unfortunately, not all of these abundances are available from every
115: study, and care must be exercised when interpreting and combining these
116: results, since the samples are subject to strong observational selection
117: effects.  In addition, most of the work has been done on giant stars whose
118: atmospheres are subject to non-LTE effects, possible mixing effects, 1-D
119: versus 3-D effects, and incompleteness due to the weakness of lines
120: analyzed.  Carbon and oxygen are particularly vulnerable.  Our main
121: sources of data are the work of Cayrel et al.\ (2004) and a further
122: discussion of the carbon and oxygen abundances of the same stars by Spite
123: et al.\ (2005).  For carbon [C/Fe], the sources are the unmixed stars of
124: Spite et al.\ (2005), Akerman et al.\ (2004), as well as those stars from
125: Honda et al.\ (2004) and Barklem et al.\ (2005) with log
126: L/L$_{\odot}\lesssim2.3$ (calculated from log T$_{eff}$, log g, and
127: assumed mass of 0.8~M$_{\odot}$).  This selection criterion was used in order
128: to reduce the possibilty of including mixed-CNO atmospheres.  Note that all
129: sources, except Barklem et al., avoided including carbon-rich stars.  Data
130: for oxygen [O/Fe] are from Akerman et al.\ (2004) and Spite et al.\
131: (2005).  Following Spite et al., their [O/Fe] data were lowered by 0.25 dex to
132: agree
133: with the Akerman et al.\ data.  Magnesium abundances [Mg/Fe] are available
134: from all sources except Akerman et al., silicon [Si/Fe] from all sources
135: except Akerman et al.\ and Barklem et al., while values of [Fe/H] are
136: available for all stars. 
137: 
138: Generally, the above samples contain some of the most metal poor stars known. 
139: Most of these stars are more metal poor than the metal poorest globular 
140: clusters ([Fe/H]$\sim-2.4$, [O/H]$\sim-1.8$). As such they are assumed to have 
141: comparable or older ages ($\sim13-14 \times10^{9}$ yrs) and hence masses 
142: $\sim0.8$M$_{\odot}$. 
143: 
144: 
145: The data are presented as plots of [O/H] and [C/H] versus [Fe/H], [Si/H],
146: and [Mg/H] in Figs.\ 1--3.  The solid line in each figure shows the
147: Population II sequence.  It is defined by the apparent clumping of many of the
148: stars (especially at the metal rich end) and from the figures, by the following
149: implied abundance ratios (i.e. [C/Fe]$\sim0.25$,
150: [O/Fe]$\sim0.65$, [Mg/Fe]$\sim0.30$, and [Si/Fe]$\sim0.35$). These ratios are 
151: similar to those found in earlier studies of the less metal poor Pop~II stars 
152: (e.g. the works cited in the first paragraph of this section).
153: The dashed lines delineate second
154: sequences where stars with nucleosynthetic signatures in common with the
155: Lyman $\alpha$ forest gas lie. In
156: every figure, the Cayrel et al.\ stars labelled 5, 30, 32, and 34 are
157: offset from the Pop~II sequence and straddle the dashed line.  Other stars
158: may also belong to this group, but either one or both of carbon, oxygen or
159: silicon abundances are not available to confirm this possibility or they are 
160: masked by observational uncertainty especially near the observational limit. 
161: The [C-O/H] abundances of these 4 stars are plotted as a histogram in the
162: right hand panel of Fig.\ 4.  Note that compared to Pop II stars, these stars 
163: have reduced carbon and oxygen but enhanced silicon with respect to iron. In 
164: particular, we note that the C and O independent ratio [Si/Fe] 
165: differs by 0.2 dex between the two groups.  
166: From Fig.2 the enhanced [Si/C] ratio is
167: $\sim0.6$.  Note also that the [O/C] ratio for these stars is $\sim0.4$,
168: similar to that of the Pop~II star sequence.  Based on the similarity of
169: the observed [C-O/H] distribution and [Si/C] ratio with that of the
170: diffuse gas, it is suggested that these 4 stars, and possibly several
171: others, are first generation stars. Note that because the model distributions 
172: in Figure 4 begin with an assumed oxygen abundance of zero and end when the 
173: Pop~II stars start to dominate, the distributions and the appropriate 
174: abundance ratios define what is referred to as first generation.
175: 
176: From Figs. 1--3 the first stellar
177: generation nuclear yields [C/Mg], [O/Mg], [Si/Mg], and [Fe/Mg] are $-0.4$, 
178: 0.0, 0.2, and $-0.35$ respectively, with an estimated uncertainty in each of 
179: $\pm0.1$. 
180: These yields can be compared to those calculated by Heger \& Woosley (2002) 
181: for a variety of Population III enrichment scenarios that are shown 
182: graphically in 
183: their Figs. 3--5. The closest fit to the first three observed ratios above 
184: is the model represented by the solid line in their Fig. 5 (i.e. high energy 
185: explosions of 12--40 M$_{\odot}$ 
186: stars including exploding very massive stars ($\sim140-260$ M$_{\odot}$) 
187: modelled with a Salpeter-like IMF. The predicted ratios [C,O,Si/Mg] are 
188: $-0.31$, 0.03, and 0.48. Both carbon and especially silicon are over produced, 
189: but a very massive star contribution appears to be required.
190: 
191: \section { The Chemical Evolution Model}
192: 
193: The model employed here is similar to the one used to describe the global
194: star formation history (H04), and it is based on the simple mass loss
195: model (e.g. Hartwick 1976, also see Pagel 1997, and Binney \& Merrifield
196: 1998).  Its novelty is a premature halting of the chemical evolution in a
197: distributed but physically motivated way.  Consider the standard one zone
198: mass loss model of chemical evolution.  Gas is slowly turned into stars
199: while being gradually enriched in heavy elements according to a specified
200: chemical yield, p.  Similtaneously, as a result of supernova driven winds,
201: gas with the same abundance as the gas currently forming stars is also being 
202: lost to further star formation at a rate which is proportional to the star 
203: formation rate. The `effective' yield (p$_{eff}$) determines the constant of 
204: proportionality, c=(p/p$_{eff}$)$-1$. 
205: 
206: If chemical evolution is suddenly halted before all of the gas is
207: exhausted, a discontinuity occurs in the metallicity distribution of stars 
208: formed as well as in the lost gas, leaving a reservoir of gas with
209: uniform composition available for later star formation.  The stellar
210: metallicity distribution can be described mathematically as the
211: undisturbed distribution multiplied by a complementary Heaviside unit step
212: function.
213:  
214: If the step function (or gas starvation function, f) is replaced with one
215: distributed in metallicity (i.e. we round the corners of the step function), 
216: then the one zone model appears as though it
217: contains many individually evolving boxes.  In H04, for example, the
218: complementary error function replaces the Heaviside function to be
219: consistent with the Gaussian distribution of stoppages assumed to occur as
220: a result of collisions among the numerous individual clumps. To provide more 
221: context for the model we give a brief summary of its application in H04.
222: 
223: There, by confining attention to a representative volume of the universe, 
224: the chemical evolution and star formation history associated with galaxy 
225: formation and evolution is derived from observations of stars and clusters 
226: in the Galaxy and in M31. The picture considers the 
227: anisotropic collapse of a number of star forming clumps. The gas in these 
228: clumps is assumed to have been enriched by a previous generation of star 
229: formation. 
230: The first collapse, perpendicular to the eventual rotation axis, results in 
231: collisions between low angular momentum clumps (M$_{t,blue}$) which terminates 
232: the star formation and is assumed to create the metal poor (blue) globular 
233: clusters. The stars already formed (M$_{s,blue}$) then constitute the extended 
234: metal poor 
235: halo and the gas released in the collisions (M$_{ml,blue}$) falls to the center
236: to form the bulge. The higher angular momentum clumps (M$_{t,red}$) continue 
237: to form stars and become more metal rich until they too begin to collide as 
238: they fall along the rotation axis. These collisions again terminate star 
239: formation and give rise to the metal rich (red) globular clusters. The stars 
240: already formed (M$_{s,red}$) are released to form a metal rich spheroid 
241: population, and the 
242: gas (M$_{ml,red}$) dissipates to form the disk.
243: Interestingly, in this picture only $\sim20$\% of the available baryons are 
244: required to provide an acceptable fit to the observed cosmic star formation 
245: rate density. Further, before 
246: collisions terminate star formation within the clumps, 
247: approximately half of these baryons are returned to the IGM by supernova 
248: driven winds (designated (M$_{WHIM,blue}$+M$_{WHIM,red}$) in H04 but 
249: renamed here M$_{CGM}$, the 
250: circum-galactic medium (CGM) component). Accounting for the `leftover' 80\% of
251: the baryons provided one of the main motivations for the present work. 
252: 
253: 
254: Here we assume there are a large number of star forming clumps approaching
255: the transition to Pop~II as the carbon and oxygen abundances approach
256: [C-O/H]$\sim-3$.  At this point the gas can cool more efficiently which in
257: turn allows low-mass Pop~II stars to form more readily (e.g. Bromm \& Loeb
258: 2003).  To characterize this transition we define f to be a complementary
259: extreme value or Fisher Tippett distribution (see equation (6) below). 
260: Because the present context of the model is different from that of H04, we
261: redefine the variables and rewrite the equations governing the chemical
262: evolution.  The conserved quantity is the total baryon mass, M$_{t}$.  It
263: is made up of 4 components: the gas contributing to star formation M$_{g}$
264: which initially is the same as M$_{t}$, the mass in long-lived stars
265: and/or remnants M$_{s}$, the gas lost due to supernova driven winds
266: identified with the diffuse intergalactic medium M$_{dIGM}$, and the gas
267: left to form later generations M$_{LG}$.  From H04, with the oxygen
268: abundance O replacing Z, M$_{dIGM}$ replacing M$_{WHIM}$, and M$_{LG}$   
269: replacing M$_{ml}$ the equations become: 
270: 
271: \begin{equation}
272: M_{t}=M_{g}+M_{s}+M_{dIGM}+M_{LG}
273: \end{equation}
274: 
275: \begin{equation}
276: M_{g}={M_{t}}\times{exp(-(\rm{O-O}_{0})/p_{eff})}\times{f}
277: \end{equation}
278: 
279: \begin{equation}
280: \frac{dM_{s}}{dlog \rm O}={ln10}\times {\rm O}\times {M_{g}}/p
281: \end{equation}
282: 
283: \begin{equation}
284: M_{dIGM}={({{p}/{p_{eff}}-1})}\times{M_{s}}
285: \end{equation}
286: 
287:  and
288: 
289: \begin{equation}
290: \frac{dM_{LG}}{dlog \rm O}={-M_{t}}\times{exp(-(\rm{O-O}_{0})/p_{eff})}\times{
291: \frac{df}{dlog \rm O}}
292: \end{equation}
293: 
294: where the complementary extreme value function is written
295: 
296: \begin{equation}
297: f=1-exp(-exp(-log(\rm{O/O_{f}}))/W)
298: \end{equation}
299: 
300: For this particular problem the initial conditions are M$_{g}$=M$_{t}$,
301: M$_{s}$=0, M$_{dIGM}$=0, M$_{LG}$=0, and O$_{\rm 0}$=0. The parameters O$_{f}$ 
302: and W in equation (6) represent the point of inflection and the `width' of the 
303: transition from one to zero of the extreme value distribution \linebreak (see 
304: e.g. http://mathworld.wolfram.com/ExtremeValueDistribution.html).
305: 
306: 
307: \subsection{Model Results}
308: 
309: As shown in Table 1, 4 input parameters are required to construct a
310: model.  The constraints considered are a given true yield p for oxygen, a
311: fit of the distribution of M$_{dIGM}$ to the high [O/H] end of the Simcoe
312: et al.\ histogram (Fig.\ 4), and that M$_{LG}$ be 19\% of the total baryon
313: content as required by the successful model of cosmic chemical evolution
314: in H04.  The oxygen yield was determined by combining the [O/Fe] value of
315: 0.35 from Fig.\ 1 with the yield assumed in H04 of [p$_{\rm Z}$/H]$=-0.11$ 
316: (representing the iron yield) 
317: to give [p$_{\rm O}$/H]$=0.24$. Note that while Z was used as the model 
318: variable in H04 its observational counterpart there was always assumed to be 
319: [Fe/H]. 
320: The remaining 3 input parameters were then
321: varied in order to satisfy the last two constraints resulting in the
322: `nominal' values in Table 1.  In order to gauge the robustness of the
323: results, each input parameter was varied separately by the amount
324: tabulated, and its effects on each of the output values are also shown.
325: The top row for each output variable shows the result of a positive
326: variation, and the bottom row shows the negative variation.  For example, 
327: the nominal value of M$_{dIGM}$/M$_{t}$ is 0.81. If [p$_{\rm O}$/H] is 
328: increased by 0.3 M$_{dIGM}$/M$_{t}$ remains unchanged at 0.81. If 
329: [p$_{\rm O}$/H] is 
330: decreased by 0.3, M$_{dIGM}$/M$_{t}$ also remains unchanged at 0.81. However 
331: [O$_{eff}$/H] is increased by 0.3, M$_{dIGM}$/M$_{t}$ becomes 0.69 and it is 
332: increased to 0.9 for a decrease in [O$_{eff}$/H] of 0.3 etc. In order to
333: compare with the observations, the calculated distribution in the left
334: hand plot of Fig.\ 4 was smoothed with a Gaussian kernel of width $\pm0.5$
335: and the stellar distribution with a $\pm0.2$ kernel to allow for
336: observational uncertainties.  The two calculated distributions appear to be
337: quite consistent with the (limited) observations. The [O/H] distribution of 
338: gas from which later stellar generations (Pop~II and beyond) will form is 
339: given by equation (5).
340: 
341: One interesting output parameter is the value of M$_{s}$/M$_{t}$ (i.e. the
342: first generation fraction of long-lived stars and remnants, fg).  Assuming 
343: that this
344: number is representative of the universe as a whole, then
345: $\Omega_{s,fg}/\Omega_{b}=8.3 \times10^{-4}$.  If we further assume that a
346: Magorrian-like relation holds between black hole mass and stellar mass
347: (i.e. M$_{\bullet}$/M$_{\star}\sim0.002$), then from the WMAP value of the 
348: baryon density ($\rho_{b}h^{2}=0.0224$, Spergel et al.\ 2003), one obtains 
349: $\rho_{\bullet,fg}\sim1.0\times10^{4}$ M$_{\odot}$/Mpc$^{3}$.  It is
350: interesting to compare this number to the Yu \& Tremaine (2002)
351: determination of the same statistic at redshift zero:
352: $2.5\pm0.4\times10^{5}h^{2}_{0.65}$ M$_{\odot}$/Mpc$^{3}$.  If the black
353: hole seeds were formed during this first generation of star formation,
354: then they had to have grown by $\sim25$ times between then and the
355: present time under our assumed cosmogony and implied `normal' IMF. However,
356: the nuclear yields discussed earlier suggest that the first generation IMF 
357: extends to very massive stars. This should result in additional massive 
358: black holes being formed and make the above growth factor an upper limit.
359: 
360: The value of the output parameter, M$_{LG}$/M$_{t}$, was constrained to be 
361: the baryon fraction that was determined in H04 to successfully describe the
362: cosmic star formation history through galaxy formation and evolution. (From 
363: Table 2 of H04 $\Omega_{b}=\Omega_{\star}+\Omega_{WHIM}+\Omega_{diffuse}$. 
364: Hence the constrained value of 
365: M$_{LG}$/M$_{t}$ is ($\Omega_{\star}+\Omega_{WHIM})/\Omega_{b}=0.19$ with an 
366: estimated uncertainty of $\pm0.04$) This 
367: number includes the mass that is lost during this process, and is the 
368: component that we 
369: now refer to as the circum-galactic medium (CGM but called WHIM in H04)
370: in order to distinquish it from the dIGM discussed in this paper. In H04 it 
371: was found that 
372: M$_{CGM}$/M$_{t}\sim0.09$ with a heavy element content
373: [Fe/H]$\sim-1$.  Thus, we expect there to be two gaseous components. As shown 
374: in Table 1, the dIGM is much more massive (M$_{dIGM}$/M$_{t}\sim0.81$) and 
375: much more metal poor ([O/H]$\sim-2.9$).
376: 
377: If long-lived stars are formed during the Pop~III phase, one may ask 
378: where are the stars with [Fe/H]$<-6$. The answer is that such stars would be 
379: exceedingly rare. In a 
380: sample of 10,000 first generation stars only 15 are predicted to have 
381: [Fe/H]$<-6$ ([O/H]$<-5.65$) based on the model distribution in Fig.4. 
382: Currently 2 stars are known with [Fe/H]$<-5$: HE 0107-5240 and HE 1327-2326 
383: (e.g. Aoki et al.\ 2005). However, both of these stars show very large over 
384: abundances of carbon and so are apparently different from the stars discussed 
385: above.
386: 
387: \section{Discussion}
388: 
389: By identifying nucleosynthetic signatures in common, a case has been made 
390: for a connection between the diffuse IGM, as identified at high
391: redshift with the Lyman $\alpha$ forest, and the first stellar generation as 
392: represented by a subsample of extremely metal poor stars.
393: This is clearly a first step as both sets of data are small and
394: incomplete. Some theoretical arguments suggest that only high mass
395: stars with no long-lived luminous stellar remnants are made during the 
396: Population~III phase (e.g. Bromm \& Larson 2004) while others (e.g. Nakamura 
397: \& Umemura 2001) have  
398: argued that the first generation IMF is bimodal with peaks at $\sim100$ and 
399: $\sim1$ M$_{\odot}$. Our result suggests that 
400: nucleosynthesis from intermediate mass stars as well as very massive stars is 
401: required to satisfy the yields found in \S2.2. These nuclear yields, as well as
402: others 
403: available but not discussed here, should allow much tighter constraints to be
404: placed on the Pop~III IMF and the accompanying ratio M$_{\bullet}$/M$_{\star}$.
405:  
406: 
407: Other extensions to this work are obvious.  Every effort should be made to
408: measure both oxygen and silicon in the burgeoning samples of extremely
409: metal poor stars.  Work should be concentrated on the highest gravity stars
410: to circumvent problems with CNO mixing which occurs in the most luminous
411: stars.  In the case of the Lyman $\alpha$ forest observations, it would be
412: useful to investigate the ionizing UV background to see if it could be
413: adjusted consistently to similtaneously allow [O/C]$\sim0.4$ and 
414: [Si/C]$\sim0.6$ in the
415: gas as is observed in the stellar sample. 
416: 
417: Finally, our simple cosmogony requires two gaseous components.  A
418: circum-galactic one (CGM) arising from mass loss associated with galaxy
419: formation and evolution, whose mass is $\sim10\%$ of all baryons, and a
420: diffuse IGM which contains $\sim80\%$ of the total baryon content but
421: which is also much more metal poor than the CGM.  However, the model makes
422: no predictions about the thermodynamic state of the gas or how it evolves.
423: Rather, sophisticated cosmological simulations follow the evolution of
424: these baryons which are presently considered to be in the form of a
425: filamentary web of very low density gas (e.g. Cen \& Ostriker 1999).  One
426: aspect of the model, which may be relevant to the search for local
427: `missing' baryons (e.g. Nicastro et al.\ 2005) and that is independent of
428: the thermodynamic state, is that the $composition$ of most of the gas may
429: not have evolved and may still be peaked at [C-O/H]$\sim-2.9$.  This could
430: make it very difficult to observe with present instrumentation. 
431: 
432: \acknowledgments
433: 
434: The author wishes to acknowledge financial support from an NSERC (Canada)
435: discovery grant. 
436: 
437: %\clearpage
438: 
439: \begin{references}
440: 
441: \reference{}
442: Akerman, C.J., Carigi, L., Nissen, P.E., Pettini, M., \& Asplund, M. 
443: 2004,
444: \aap, 414, 931
445: 
446: \reference{}
447: Aoki, W., Frebel, A., Christlieb, N., Norris, J.E., Beers, T.C. et al. 2005, 
448: astro-ph/0509206
449: 
450: \reference{}
451: Barklem, P.S., Christlieb, N., Beers, T.C., Hill, V., Bessell, M.S.,
452: Holmberg, J., Marsteller, B., Rossi, S., Zickgraf, F.-J., \& Reimers, D.
453: 2005, \aap, 439, 129
454: 
455: \reference{}
456: Binney, J., \& Merrifield, M. 1998, Galactic Astronomy, (Princeton: Princeton
457: University Press)
458: 
459: \reference{}
460: Bromm, V., \& Larson, R.B. 2004, \araa, 42, 79
461: 
462: \reference{}
463: Bromm, V., \& Loeb, A. 2003, Nature, 425, 812
464: 
465: \reference{}
466: Cayrel, R., Depagne, E., Spite, M., Hill, V., Spite, F., Francois, P.,
467: Plez, B., Beers, T., Primas, F., Andersen, J., et al. 2004, \aap, 416, 1117
468: 
469: \reference{}
470: Cen, R., \& Ostriker, J. 1999, \apj, 514, 1
471: 
472: \reference{}
473: Cowie, L.L., Songaila, A., Kim, T-S., \& Hu, E.M. 1995, \aj, 109, 1522
474: 
475: \reference{}
476: Fukugita, M., \& Peebles, P.J.E. 2004, \apj, 616, 643
477: 
478: \reference{}
479: Gratton, R.G., \& Sneden, C.S. 1988, \aap, 204, 193
480: 
481: \reference{}
482: Gratton, R.G., \& Sneden, C.S. 1991, \aap, 241, 501
483: 
484: \reference{}
485: Haardt, F., \& Madau, P. 1996, \apj, 461, 20
486: 
487: \reference{}
488: Hartwick, F.D.A. 2004, \apj, 603, 108, H04
489: 
490: \reference{}
491: Hartwick, F.D.A. 1976, \apj, 209, 418
492: 
493: \reference{}
494: Heger, A. \& Woosley, S.E. 2002, \apj, 567, 532
495: 
496: \reference{}
497: Honda, S., Aoki, W., Kajino, T., Ando, H., Beers, T.C., Izumiura, H.,
498: Sadakane, K., \& Takada-Hidai, M. 2004, \apj, 607, 474
499: 
500: \reference{}
501: McWilliam, A., Preston, G.W., Sneden, C., \& Searle, L. 1995, \aj, 109, 2757
502: 
503: \reference{}
504: Nakamura, F., \& Umemura, M. 2001, \apj, 548, 19
505: 
506: \reference{}
507: Nicastro, F., Mathur, S., Elvis, M., Drake, J., Fang, T., Fruscione, A.,
508: Krongold, Y., Marshall, H., Williams, R., \& Zezas, A. 2005, Nature, 433, 495
509: 
510: \reference{}
511: Pagel, B.E.J. 1997, Nucleosynthesis and Chemical Evolution of Galaxies, 
512: (Cambridge: Cambridge University Press)
513: 
514: \reference{}
515: Rauch, M. 1998, \araa, 36, 267
516: 
517: \reference{}
518: Ryan, S.G., Norris, J.E., \& Beers, T.C. 1996, \apj, 471, 254
519: 
520: \reference{}
521: Simcoe, R.A., Sargent, W.L.W., \& Rauch, M. 2004, \apj, 606, 92
522: 
523: \reference{}
524: Spergel, D.N., et al. 2003, \apjs, 148, 175
525: 
526: \reference{}
527: Spite, M., Cayrel, R., Plez, B., Hill, V., Spite, F., Depagne, E., Franois, P.
528: , Bonifacio, P., Barbuy, B., Beers, T., et al. 2005, \aap, 430, 655
529: 
530: \reference{}
531: Veilleux, S., Cecil, G., \& Bland-Hawthorn, J. 2005, astro-ph/0504435
532: 
533: \reference{}
534: Yu, Q., \& Tremaine, S. 2002, \mnras, 335, 965
535: 
536: \end{references}
537: 
538: \clearpage
539: 
540: \begin{figure}
541: \plotone{f1.eps}
542: \caption{ (left) [C/H] vs [Fe/H], (right) [O/H] vs [Fe/H] for extremely
543: metal poor stars.  Red numbers - data from Spite et al.\ (2005).  Blue
544: numbers correspond to stars in order of their appearance in the data
545: tables of Honda et al.\ (2004), black dots are data from Akerman et al.
546: (2004), and yellow dots are data from Barklem et al.\ (2005).  The solid line 
547: is the Population II sequence and the dashed line is the locus of stars with
548: abundances considered to be in common with the Lyman $\alpha$ forest. Typical 
549: error bars are shown.} 
550: \end{figure}
551: 
552: \clearpage
553: 
554: \begin{figure}
555: \plotone{f2.eps}
556: \caption{The same as Fig.\ 1 except for [Si/H] vs [C/H] and [Si/H] vs [O/H]}
557: \end{figure}
558: 
559: \clearpage
560: 
561: \begin{figure}
562: \plotone{f3.eps}
563: \caption{The same as Fig.\ 1 except for [Mg/H] vs [C/H] and [Mg/H] 
564: vs [O/H]}
565: \end{figure}
566: 
567: \clearpage
568: 
569: \begin{figure}
570: \plotone{f4.eps}
571: \caption{(left) A comparison of the observed abundance distribution of
572: [O/H] in the low column density Lyman $\alpha$ forest from Simcoe et al.
573: (2004), with the mass loss component from the model (solid line).  (right) 
574: The solid line 
575: shows the predicted distribution of [O/H] for the stellar component of the 
576: model. The histogram shows the distribution of [C/H] $and$ 
577: [O/H] for the 4 stars considered prime first
578: generation candidates.  The normalization of the ordinate in both figures
579: is arbitrary.} 
580: \end{figure}
581: 
582: \clearpage
583: 
584: \begin{deluxetable}{cccccc}
585: \tablenum{1}
586: \tablecolumns{6}
587: \tablecaption{Model Parameters and Changes due to Variations in Input
588: Parameters\tablenotemark{a}}
589: \tablehead{Parameter & Input &
590: \colhead{[p$_{\rm O}$/H]} &
591: \colhead{[O$_{eff}$/H]} &
592: \colhead{[O$_{f}$/H]} &
593: \colhead{1/W} \\
594: \colhead{Output} &
595: \colhead{Nominal value} &
596: \colhead{$0.24\pm0.3$} &
597: \colhead{$-2.75\pm0.3$} &
598: \colhead{$-2.50\pm0.3$} &
599: \colhead{$1.5\pm0.5$}
600: }
601: \startdata
602: & & $4.16\times10^{-4}$ & $1.42\times10^{-3}$ & $9.24\times10^{-4}$ &
603: $8.51\times10^{-4}$  \nl
604:    M$_{s}$/M$_{t}$ & $8.30\times10^{-4}$ & \nl
605: & & $1.66\times10^{-3}$ & $4.63\times10^{-4}$ & $7.10\times10^{-4}$ &
606: $7.95\times10^{-4}$  \nl
607: \nl
608: & & $0.81$ & $0.69$ & $0.90$ & $0.83$ \nl
609:    M$_{dIGM}$/M$_{t}$ & $0.81$ & \nl
610: & & $0.81$ & $0.90$ & $0.69$ & $0.78$ \nl
611: \nl
612: & & $0.19$ & $0.31$ & $0.10$ & $0.17$ \nl
613:    M$_{LG}$/M$_{t}$ & $0.19$ & \nl
614: & & $0.19$ & $0.10$ & $0.31$ & $0.22$ \nl
615: \nl
616: & & $-2.90$ & $-2.70$ & $-2.85$ & $-2.90$  \nl
617:    {[O/H]$_{max}$} & $-2.90$ & \nl
618: & & $-2.90$ & $-3.15$ & $-2.95$ & $-2.90$ \nl
619: \nl
620: \enddata
621: \tablenotetext{a}{For each output parameter, the upper row gives the
622: value due to the indicated positive variations in each input parameter
623: separately.  The lower row are results from a negative variation. See text.}
624: \end{deluxetable}
625: 
626: \end{document}
627: