1: \def\km{{\rm\,km}}

2: \def\cms{{\rm\,cm\,s^{-1}}}

3: \def\kms{\ifmmode{\,\hbox{km}\,s^{-1}}\else {\rm\,km\,s$^{-1}$}\fi}

4: \def\kpc{{\rm\,kpc}}

5: \def\mpc{{\rm\,Mpc}}

6: \def\msun{{\rm\,M_\odot}}

7: \def\lsun{{\rm\,L_\odot}}

8: \def\rsun{{\rm\,R_\odot}}

9: \def\pc{{\rm\,pc}}

10: \def\cm{{\rm\,cm}}

11: \def\yr{{\rm\,yr}}

12: \def\au{{\rm\,AU}}

13: \def\AU{{\rm\,AU}}

14: \def\gm{{\rm\,g}}

15: \def\kmsm{{\rm\,km\,s^{-1}\,Mpc^{-1}}}

16: \def\kmps{{\rm\,km\,s^{-1}}}

17: \def\hmpc{\ifmmode{h^{-1}\,\hbox{Mpc}}\else{$h^{-1}$\thinspace Mpc}\fi}

18: \def\hkpc{$h^{-1}$\thinspace kpc}

19: \def\eg{{\it e.g.}~}

20: \def\etal{{\it et~al.}~}

21: \def\et{{\it et~al.}~}

22: \def\cf{{\it cf.}~}

23: \def\ie{{\it i.e.}~}

24: 

25: \documentclass[12pt,preprint]{aastex}

26: \begin{document}

27: \relax

28: \tolerance 9000

29: %\slugcomment{DRAFT: \today}

30: 

31: \title

32: {Globular Clusters at High Redshift}

33: 

34: \author{R.~G.~Carlberg}

35: 

36: \affil{Department of Astronomy \& Astrophysics}

37: \affil{University of Toronto, Toronto ON, M5S~3H8 Canada}

38: \email{carlberg@astro.utoronto.ca}

39: 

40: 

41: %\clearpage

42: 

43: \begin{abstract}

44: Globular clusters will be present at high redshifts, near the very

45: beginning of the galaxy formation process. Stellar evolution ensures

46: that they will be much more luminous than today. We show that the

47: redshift distribution at nano-Jansky levels should be very broad,

48: extending up to the redshift of formation.  A bracketing range of

49: choices for the redshift of formation, spectral energy evolution

50: models and population density evolution, leads to the conclusion that

51: the sky densities should be around $10^7$ per square degree at 1~nJy

52: ($m_{AB}=31.4$ mag) in bands around 4 microns. Such high sky densities

53: begin to present a confusion problem at these wavelengths to

54: diffraction limited 6m class telescopes.  These star-like, low

55: metallicity, clusters will be a significant foreground population for

56: ``first light'' object searches. On the other hand they are an

57: exceptionally interesting ``second light'' population in their own

58: right. Depending on the details of galaxy assembly, the clusters will

59: have a noticeable cross-correlation with galaxies on scales of about

60: 20 arcsec, or less, depending on the details of the buildup of galaxy

61: assembly after globular cluster formation. High redshift globular

62: clusters will be an accessible, direct, probe of the earliest stages

63: of the formation of galaxies and the buildup of metals in the

64: universe.

65: \end{abstract}

66: 

67: \keywords{galaxies: clusters: general, galaxies: interactions, galaxies: star clusters, stars: formation}

68: 

69: \clearpage

70: \section{Introduction}

71: 

72: Globular clusters contain some of the oldest stars in the universe and

73: have long been vital clues to the earliest phases of the star

74: formation in galaxies \citep{searle+zinn,harris_ext,cmw}. In our own

75: galaxy the known globular clusters are very old \citep{donv} but there

76: is evidence that they can form at lower redshift in suitably extreme

77: conditions, generally associated with merging galaxies

78: \citep{zepf,zf,ashman+zepf,cen,larsenetal}.  The great age and

79: low metallicity of globular cluster systems indicates that they should

80: be present at very high redshifts and predate the bulk of their

81: eventual host galaxies' stars.

82: 

83: The exciting prospect is that direct studies of globular cluster

84: formation and evolution will soon become possible.  The next

85: generation of optical-infrared telescopes on the ground and in space

86: will have the capability to detect objects at the nano-Jansky

87: level. An estimate of the faint number counts in the optical was

88: undertaken for HST \citep{vdb} but we concentrate on the IR where the

89: redshifts and rise in numbers is much more dramatic.  In the 2 to 5

90: micron bands, the combination of large k-corrections and substantial

91: stellar brightening raises the fluxes from high redshift clusters into

92: the range of 29-32 AB mag. These nanojansky flux levels are within the

93: capabilities expected of future telescopes.

94: 

95: Today's globular globular clusters are likely the survivors of a

96: larger population present at the various times of formation

97: \citep{fall+rees,fall+zhang}. If their co-moving density increases by

98: an order of magnitude over those at low redshift then the globular

99: clusters are likely to appear with numbers at a given flux level that

100: are comparable to sub-galactic mass dark matter halos which are the

101: sites of the ``first stars'' \citep{hugh+martin,haiman+loeb,HAM}.

102: 

103: This paper calculates the magnitude limited distribution of the

104: expected numbers, $n(m)$, the redshift distribution, $n(z|m)$, and

105: estimates the angular clustering properties of the globular cluster

106: population relative to their host galaxies.  The predictions are made

107: for filter pass bands sufficiently red that Lyman $\alpha$ trough

108: absorption will not normally be an issue. In the next section we

109: describe the calculation of the co-moving number density of globular

110: clusters (GC) as a function of redshift for different evolutionary

111: assumptions. In Section 3 we present the results of the number

112: calculations. Section 4 considers the apparent sky clustering of the

113: distant GC. We conclude with a discussion of the opportunities and

114: complications that this population presents. The calculations are

115: presented in a cosmology for which $H_0=70

116: \kmsm$, $\Omega_M=0.3$, $\Omega_\Lambda=0.7$.

117: 

118: \section{An Evolving Globular Cluster Luminosity Function}

119: 

120: The expected sky density of GCs at magnitude $m$ and redshift $z$

121: depends on the product of the cosmological volume element and their

122: luminosity function, $\phi_{GC}(L,z)$, integrated with the volume

123: element along the line of sight.  The luminosity function has three

124: sources of evolution. First, it is generally accepted that galactic

125: tidal fields and stellar dynamical ``shocks'' erode a more numerous

126: high redshift GC population into the remnant population we see

127: today. We use the results of a relatively secure theoretical analysis

128: of the evolution of the population, but also show results for a

129: non-evolving distribution. Second, as the stellar population becomes

130: younger with increasing redshift its spectral energy distribution

131: changes. Third, the GCs form at some high, but as yet poorly

132: determined, redshift. Our approach to each of these evolutionary terms

133: along with the normalization to the present day globular cluster

134: population is discussed in the following section.

135: 

136: \subsection{An Evolving Mass Distribution}

137: 

138: Recently Fall \& Zhang (2001, hereafter referred to as FZ) have

139: discussed a generalized dynamical model for the evolution of the mass

140: distribution of GCs. They find that within 1-2 Gyr of origin, a wide

141: range of initial GC mass distribution assumes a characteristic form

142: which then evolves in a nearly self-similar way. At small mass, all

143: clusters (in the same tidal field) go to zero mass at the same rate

144: due to two-body relaxation driven evaporation. At high mass,

145: gravitational shocks impose a characteristic maximum mass above which

146: there is a rapid cutoff of numbers. The continual depletion of GCs

147: implies that over a Hubble time about 10\% of the initial cluster

148: population survives, under the assumptions that the system is not

149: replenished and that the galactic potential does not change. FZ have

150: kindly made their results for the evolution of the globular cluster

151: mass distribution available for use in this paper. Specifically we use

152: the differential number of GCs at mass $M$ at time $t$, $n_{GC}(M,t)\,

153: dM$, which FZ present in their Figure 3.

154: 

155: It should be noted that the FZ model {\it predicts} the mass of the

156: peak.  To test this aspect of the models FZ have put the Milky Way

157: globular clusters on a mass scale using $M/L_V=3$, as is appropriate

158: for an old, metal poor stellar population. The agreement between model

159: and observation is impressively good.  The extensive testing of FZ

160: shows that the results should not change much with galaxy mass or

161: galaxy type.  

162: 

163: The FZ model results are specified at times of 0.01, 1.5, 3, 6 and 12

164: Gyr.  We use a double spline function in the variable $M$ and

165: $\log{t}$ to interpolate to other times. We extrapolate slightly

166: beyond their 12 Gyr model to the age of 13.4 Gyr age of our

167: cosmology. The minimum age of their models is a 0.01 Gyr, where the

168: ``formation distribution'' is close to a power law in mass. We will

169: show the sensitivity of our results to the GC number evolution.

170: 

171: \subsection{The Redshift Dependent Luminosity Function}

172: 

173: We require the redshift dependent globular cluster luminosity function,

174: $\phi_{GC}(L_{\bar{\lambda}},z)$, where $L_{\bar{\lambda}}$ is the

175: observed luminosity in some filter band centered around

176: $\bar{\lambda}$.  The conversion from $n_{GC}(M)$ to $\phi_{GC}(L)$ is made

177: using a spectral synthesis model which gives the entire spectral

178: energy distribution, $F_\lambda$, as a function of model age for given

179: metallicity and star formation history.  We use the PEGASE.2 code

180: \citep{pegase} to calculate $\ell_{\bar{\lambda}}$, 

181: the observed frame luminosity per unit mass in the filter band

182: $\bar{\lambda}$,

183: 

184: \begin{equation}

185: \ell_{\bar{\lambda}} = {{\int_0^\infty T(\lambda)F_\lambda((1+z)\lambda) 

186: 	\,d\lambda}

187: 	\over{(1+z)\int_0^\infty T(\lambda) \,d\lambda}},

188: \label{eq:ml}

189: \end{equation}

190: where $F_\lambda[(1+z)\lambda]/(1+z)$ is the model's mass normalized

191: absolute flux in the observed frame and $T(\lambda)$ is the filter

192: transmission function.  Noting that the photon redshift and the time

193: dilation are included in Eq.~\ref{eq:ml}, 

194: an object of mass $M$ gives an observed flux in the

195: $\bar{\lambda}$ filter of

196: \begin{equation}

197: f_{\bar{\lambda}} =  {{M\ell_{\bar{\lambda}}}\over{4\pi r^2(z)}},

198: \label{eq:flux}

199: \end{equation}

200: where $r(z)$ is the co-moving distance in the adopted cosmology.  The

201: observed flux is converted to magnitudes using the definition

202: $m_{\bar{\lambda}}\equiv -2.5\log_{10}(f_{\bar{\lambda}}) + C$, where

203: $C$ is 31.4 AB magnitudes at 1 nano-Jansky. 

204: 

205: \subsection{Normalizing the Luminosity Function}

206: 

207: The luminosity functions of the GC systems of the Milky Way and more

208: than 50 nearby galaxies have been studied

209: \citep{harris_ext,harris_mw}.  A single galaxy's GC luminosity

210: function is conventionally described as a Gaussian (in absolute

211: magnitude, hence a lognormal distribution in luminosity) centered at

212: $\langle M_V

213: \rangle= -7.27+5\log_{10}{(H_0/75)}$ mag with a dispersion of about

214: 1.2 magnitudes.  Although a more complex function, the FZ mass model

215: appears to describe the data at least as well as a Gaussian. Moreover,

216: it is based on a dynamical theory that allows its history to be

217: predicted. To use the FZ function in our calculation we need to fix

218: the volume normalization and we will also introduce a small shift in

219: the $M/L$ value.

220: 

221: \subsubsection{Mass-Luminosity Normalization}

222: 

223: The mass normalization of the FZ models is determined by the dynamics

224: of the GCs within the model galaxy. Although fairly insensitive to

225: variations in the potential, the mass function does shift slightly

226: depending on the specific galactic potential.  Here we need the

227: luminosity function typical of a mix of galaxies. We adopt the

228: functional form of the FZ mass function and could adopt their

229: $M/L_V=3$ value, however we prefer to make a small adjustment to

230: provide an alternate match to the observational data.  The FZ mass

231: function is a power law on the low mass side and much steeper than a

232: Gaussian on the high mass side. Here we chose an $M/L_V$ value that

233: brings the mean luminosities of the FZ distribution to the mean of the

234: Gaussian fits. A numerical integration finds that $M_V=-7.27$ mag

235: should be identified as $\log{M/\msun}= 5.36$, which implies an

236: $M/L_V=3.3\msun/\lsun$.  This small $M/L$ change is well within the

237: uncertainty of stellar population modeling.  In particular, $M/L_V$ is

238: $2.1\msun/\lsun$ for the $Z=0.1$ solar PEGASE models we compute at an

239: age of 13.4 Gyr. Our normalization effectively raises the $M/L$ values

240: of the PEGASE models by a multiplicative factor of 1.57.

241: 

242: The V band luminosity is widely used to describe low redshift

243: clusters. However, it is beneficial for the accuracy of our

244: application to high redshift galaxies main application to use K band

245: luminosities. Furthermore, the GC population is most closely connected

246: to the old stellar population which is most accurately measured at low

247: redshift by K band luminosities.  The conversion from V to K must use

248: the IR colors of a population with a metal abundance of about

249: one-tenth solar, $V-K=2.93 +0.5 Z/Z_\odot$ mag \citep{vk_colors}.  The

250: $Z=0.1Z_\odot$ PEGASE models find $V-K=2.4$ mag at 13.4 Gyr which is

251: in essentially exact agreement with the observational relation.  Using

252: this color we find that the mean peak K band luminosity for GCs is

253: $\langle M_K \rangle =-9.70+5\log_{10}{(H_0/75)}$. This value is

254: converted to the flux based AB magnitude system with the addition of

255: $AB(K)=+1.88$ mag.

256: 

257: \subsubsection{Number Density Normalization}

258: 

259: The mean co-moving density of GCs for a single galaxy is modeled as

260: being directly proportional to its luminosity,

261: \citep{S,harris_ext},

262: \begin{equation}

263: S_N= N_t 10^{-0.4(M_V+15)},

264: \label{eq:SV}

265: \end{equation}

266: where $M_V$ is the galaxy's absolute magnitude in the V band and $N_t$

267: is the total number in a Gaussian luminosity function. The $S_N$

268: relation has significant variations with Hubble type and possibly

269: environment but appears to be accurate in the mean \citep{harris_ext}.

270: Since GC are most clearly associated with old stellar light it is

271: natural to use a K band luminosity function, in which case,

272: \begin{equation}

273: S_N= N_t 10^{-0.4(M_K+17.9)},

274: \label{eq:SK}

275: \end{equation}

276: where a galaxy with solar metallicity has $V-K\simeq 2.9$ mag

277: \citep{vk_colors}. 

278: Adopting the Gardner \et\ (1993) luminosity

279: function (for our purposes, similar to the recent Cole et al. 2001

280: results), for which $M_\ast(K) = -23.1$ mag, we find that in

281: the K band the number of globular clusters around a galaxy rises

282: linearly with luminosity,

283: \begin{equation}

284: N_t(L_K) =120S_N {L_K\over {L_\ast(K)}}. 

285: \label{eq:nk}

286: \end{equation}

287: 

288: To convert the normalization from globular cluster per galaxy to 

289: a volume normalization we use the luminosity density of galaxy light 

290: in the K band, $j(K)$. The normalizing constant for the GC luminosity

291: function is defined such that the integral over all GC luminosities

292: must be equal to the mean number of GC expected for the mean amount of

293: galaxy light in that volume. That is,

294: \begin{equation}

295: \int_0^\infty \phi_{GC}(L_{\bar{\lambda}},z=0)\, dL

296: 	 = 120 S_N {j(K)\over L_\ast(K)}.

297: \label{eq:norm}

298: \end{equation}

299: For Gardner's (1993) $\alpha=-1$ Schechter luminosity function fit

300: $j(K)=\phi_\ast(K) L_\ast(K)$, where

301: $\phi_\ast(K)=0.0166h^{-3}$~Mpc$^{-3}$. Note that the dependence on

302: $L_\ast(K)$ cancels in Eq.~\ref{eq:norm}.  Then the co-moving volume

303: density is GCs is $n_{GC}(0)=2.0h^{-3}S_N$~Mpc$^{-3}$, where

304: $h=H_0/100$.  We adopt $S_N=2$ as a reasonable and somewhat

305: conservative value, given that the bulk of the K light emerges from

306: relatively luminous early type galaxies. Figure~4 and Table~3 of

307: Harris (1991) might suggest a value of about 3 for the early type

308: galaxies, with evidence that strongly clustered early type galaxies

309: have higher $S_N$. Of course the origin of these effects may well be

310: directly visible in the future.  The outcomes is that our complete GC

311: luminosity function is,

312: \begin{equation}

313: \phi_{GC}(L_{\bar{\lambda}},z) \,dL_{\bar{\lambda}} = 

314: 	n_{GC}(0)

315: 	n_{GC}[L_{\bar{\lambda}}/\ell_{\bar{\lambda}}(t),t(z)] 

316: 	\,dL_{\bar{\lambda}}.

317: \end{equation}

318: 

319: The redshift distribution per unit sky area of GC at a given flux

320: level, is simply

321: \begin{equation}

322: n(z|f_{\bar{\lambda}}) \, d\ln{f_{\bar{\lambda}}} = \int_0^\infty \phi_{GC}(4\pi

323: 	r^2(z)f_{\bar{\lambda}},z) {{dV}\over {dz}}\, dz\, d\ln{f_{\bar{\lambda}}}.

324: \label{eq:sky}

325: \end{equation}

326: where $dV/dz$ is the volume element within the model

327: cosmology. Integrating over the redshift distribution gives the

328: number-magnitude relation In practice, these calculations are done

329: using magnitudes rather than fluxes.

330: 

331: \section{Counts and Distributions}

332: 

333: With the modeling apparatus in hand we first pause to show the low

334: redshift $n(z)$ at $m_R(AB)=25, 26, 27$ and $28$ mag in

335: Figure~\ref{fig:nzR}.  The total sky densities of the FZ model at

336: these depths are 87, 321, 1,260 and 4,840 per square degree per

337: magnitude, respectively.  These GCs will be clearly associated with

338: relatively bright galaxies, typically about $m_R(AB)=13-17$ mag, with

339: the redshift distributions shown.

340: 

341: \subsection{Number Redshift Distributions}

342: 

343: Precisely how globular clusters form is, at this time, unknown.  Part

344: of the point of this paper is that the plausible formation redshifts

345: for the bulk of GCs will shortly come within reach of telescopes. To

346: try to bracket the situation, we examine a number of somewhat extreme

347: alternative models and look at the effects of co-ordinated bursts in a

348: galaxy. To examine the importance of ``luminosity spikes'' at the time

349: of formation, we use (arbitrarily, for the purpose of illustration) 10

350: bursts of star formation of duration 10~Myr spread over the 0 to 1 Gyr

351: time interval.  The formation age of all of the GC is put at 0.5

352: Gyr. The results are shown in Figure~\ref{fig:nzL}. Such bursts would

353: only effect the counts in an area small enough that only a few dozen

354: actively GC forming galaxies were present.

355: 

356: In Figure~\ref{fig:nzL_alt} we use the same set of ten star formation

357: bursts but shifted in time to the 1 to 2 Gyr time interval with a

358: uniform formation age of 1 Gyr. Clearly the bursts of star formation

359: produce spikes in the redshift distribution but those effects quickly

360: die away. It could be that the earliest phases of globular cluster

361: formation are cloaked in dust which later disperses, following an

362: age-extinction relation \citep{shapley}.  In that case the high

363: luminosity peaks will be a briefly obscured phase in the life of

364: GCs. However the Figures show that if those short-lived bright spikes

365: are removed, neither the counts or redshift distribution will be

366: greatly altered.

367: 

368: The difference between the no-evolution and FZ density evolution

369: models are small at redshifts below about three, under the assumption

370: that most globular clusters were formed at redshifts greater than

371: three. The differences would be much larger if significant globular

372: cluster formation continued to much lower redshift.  The numbers

373: predicted with a non-evolving mass model are nearly a full decade

374: below the evolving model beyond redshifts of five. Since density and

375: luminosity evolution are independent in these models the same

376: difference applies to all formation histories.

377: 

378: \subsection{Number Magnitude Relations}

379: 

380: The number-magnitude relation is shown in Figure~\ref{fig:nm} in the

381: V, R, J, K, L and M bands (spanning roughly 0.5 to 5 microns) for our

382: model with ten, bursts of 10\% of the mass, spread between 0 and

383: 1~Gyr. There are two effects. At low redshift the counts rise with

384: increasing magnitude at a rate governed by the volume element,

385: enhanced by k-corrections in an old, low metallicity, population.

386: Although not shown, at 1~nJy in the V band the $n(z)$ peaks at about

387: redshift 0.3. At 1~nJy in the R band $n(z)$ peaks at about redshift

388: 0.5, with a few of the actively star-forming $z=5$ clusters in

389: formation being visible.  Clearly deep optical band observations are

390: not the ideal way to probe the formation epoch. The character of the

391: redshift distribution changes in the infrared bands as the peak of the

392: spectrum is redshifted into them. The combination of k-correction and

393: luminosity evolution causes the counts in the redder bands to rapidly

394: climb to several million per square degree.  As Figure~\ref{fig:nm}

395: shows, the counts are steeper than Euclidean near 30 AB mag in the IR

396: bands.

397: 

398: The predicted counts for a wide range of model star formation

399: histories are shown in Figure~\ref{fig:nmE}.  We display the L band

400: counts for models having ten bursts of 10\% of the final mass star

401: formation extending over 10~Myr in the 0-1 Gyr interval (triangles),

402: the 1-2 Gyr interval (diamonds), and exponential models with $\tau=1$

403: pentagons), 2 (heptagons) and 4 Gyr (hexagons) for both evolving and

404: non-evolving mass function. The number-magnitude relation shows

405: substantial model dependencies beginning at about 1~nJy ($m_{AB}=31.4$

406: mag). However, the result that the L band counts will be around $10^8$

407: per mag per square degree at 0.2~nJy is reasonably robust. It does not

408: depend a lot on star formation, internal internal dust shrouding in

409: the early phases, and is not unduly sensitive to the exact amount of

410: GC density evolution. The biggest potential over-prediction of numbers

411: is if GC initially form in dusty disk environments which makes them

412: hard to detect. As long as the clusters become visible within about 2

413: Gyr of formation the numbers predicted here should be fairly

414: accurate. The basic prediction that the sky density becomes about

415: $10^7$ per square degree around 1 nJy is difficult to escape, given

416: the assumptions about globular cluster origins, evolution and

417: visibility made in this paper.

418: 

419: \section{Angular Clustering}

420: 

421: Globular clusters are strongly concentrated around their host

422: galaxies.  At the very low flux levels we are investigating here it is

423: natural to ask to what degree this clustering will remain evident and

424: whether the nano-Jansky sky will effectively be covered with a nearly

425: uniform distribution of GC. A prediction of clustering uses the

426: results above but requires additional information about the degree to

427: which galaxies and their GC systems merge into larger and larger

428: units. Furthermore, the host galaxies may not always be visible at the

429: highest redshifts considered here, since galaxies are generally

430: younger and much lower surface brightness than GCs. Hence, the

431: following estimates of galaxy-GC cross-correlations will be upper

432: limits, although we do incorporate a model for galaxy merging into our

433: calculations.

434: 

435: The real space cross-correlation of galaxies and GCs can be derived

436: from the average radial profile of GCs in their host galaxies. As

437: shown below the auto-correlation of galaxies makes no significant

438: contribution at the angles of interest.  The FZ calculations find that

439: after approximately 1 to 2 Gyr the radial distribution converges to a

440: stable, nearly power-law form.  An approximate power law fit to the

441: Milky-Way data of Harris (1996) is,

442: \begin{equation}

443: n(r)=3\times 10^2 \left({r\over{10 {\rm kpc}}}\right)^{-3.5} {\rm kpc}^{-3}.

444: \label{eq:gcr}

445: \end{equation}

446: If there is a core in the radial distribution it appears at a radius

447: of order a few kpc where the GCs become superimposed on significant

448: galaxy light and hard to find. The Milky Way is probably somewhat less

449: than $L_\ast$ in luminosity and its GC system, with a total of

450: $160\pm20$ clusters \citep{harris_ext}, has numbers about 2/3 half of

451: the 240 expected at $L_\ast$.  To calculate the cross-correlation

452: function with galaxies we need $\delta(r)= (n(r) -n_0)/n_0$, where

453: $n_0$ is the mean density.  We normalize these numbers to the volume

454: average for $L_\ast$ galaxies.  The mean GC density of

455: $4.0h^{-3}$Mpc$^{-3}$ we derived above becomes a physical density of

456: $1.2\times 10^{-8}$~kpc$^{-3}$ for $H_0=70$. Consequently we can

457: re-express Eq.~\ref{eq:gcr}, as the over-density,

458: \begin{equation}

459: \delta(r)=2.5\times 10^{10} \left({r\over {10 {\rm kpc}}}\right)^{-3.5}.

460: \label{eq:delta}

461: \end{equation}

462: Converting this to the standard correlation length form and using

463: co-moving co-ordinates ($H_0=70$),

464: \begin{equation}

465: \xi_{gGC}(x) = \left({{9.4 {\rm Mpc}}\over x}\right)^{3.5}

466: 		{L_h\over L_\ast},

467: \label{eq:xi}

468: \end{equation}

469: where we have included the luminosity dependence, with $L_h$ being the

470: luminosity of the host galaxy.  Note that an alternate description of

471: this correlation length is $6.6\hmpc$. In this calculation we have

472: assumed that the low redshift $S$ relationship holds in the earliest

473: phases of the life of a galaxy which needs to be tested.  An overall

474: density normalization change has no effect on the correlations since

475: the mean field density changes at the same rate, leaving $\delta$

476: invariant. 

477: 

478: GC systems appear to always be associated with more or less virialized

479: galaxies. They are not part of a clustering hierarchy that extends

480: into the linear regime. Therefore we describe the over-density

481: distribution as being fixed in physical co-ordinates. We therefore

482: multiply Eq.~\ref{eq:xi} by the correlation function

483: evolution term $(1+z)^\epsilon$. The quantity $\epsilon$ is equal to

484: $\gamma-3$ for fixed over-density in physical co-ordinates, as is

485: appropriate here.

486: 

487: Galaxies are assembled over time through the merger process.  A simple

488: model for the increase of mass $M$ is $dM/dt = {\cal R} (1+z)^{\cal

489: M}$.  Approximating $1+z=t_0/t$ (as in an empty universe) this

490: integrates to 

491: \begin{eqnarray}

492:          & M_0 - {\cal R}({\cal M}t_0)^{-1} (1+z)^{{\cal M}-1}, 

493: 		&  {\cal M} > 1, \\

494: M(z) =\Bigg\{  &   &  \\

495: 	& M_0-{\cal R}t_0 \log{(1+z)},~~~~~~~ & {\cal M}=1.  \\

496: \end{eqnarray}

497: Note that $M(z)$ goes to zero for finite $z$ for ${\cal

498: M}>1$.  Reasonable values are ${\cal M}\simeq 1-3$ and ${\cal

499: R}t_0\simeq 0.2-0.5$ \citep{mergers,cfrs_mg}. The resulting redshift

500: dependent co-moving correlation function, $\xi_{gGC}(x|z)$, is,

501: \begin{equation}

502: \left({{r_0(z)}\over x}\right)^\gamma 

503: = (1+z)^\epsilon \left( {{r_{00}}\over x(z)}\right)^\gamma

504: 	{M(z)\over M_0}.

505: \label{eq:xiz}

506: \end{equation}

507: 

508: The angular correlation function is simply related to the volume

509: correlation through a projection over redshift,

510: \begin{equation}

511: \omega(\theta) = A(\gamma)

512: \theta^{1-\gamma} N^{-2}

513: \int n^2(z) \left({{r_0(z)}\over x}\right)^\gamma x  \, {{H(z)}\over{c}} dz,

514: \end{equation}

515: where $N=\int n(z)\, dz$,

516: $A(\gamma)=\Gamma(\onehalf)\Gamma((\gamma-1)/2)/ {\Gamma(\gamma/2)}$,

517: and $H(z) = H_0

518: [\Omega_M(1+z)^3+\Omega_R(1+z)^2+\Omega_\Lambda]^{1/2}$, with

519: $\Omega_M+\Omega_R+\Omega_\Lambda=1$. 

520: 

521: We express the results as an angular correlation $\omega(\theta) =

522: (\theta_0/\theta)^{\gamma-1}$. We evaluate the integral using the L

523: band $n(z)$ at 1nJy. For our $r_{00}=6.6\hmpc$ Mpc, $\gamma=3.5$,

524: $\epsilon=0.5$, we find $\theta_0 = 22$ and $18\arcsec$, with ${\cal

525: R}t_0=0.5$ for ${\cal M}=1$, and $2$, respectively and 21 and

526: $17\arcsec$ for ${\cal R}t_0=0.3$ for the same ${\cal M}$.

527: Correlation angles of $20\arcsec$ correspond to physical distances of

528: about 100~kpc around redshift three. Therefore the bulk of the GCs

529: will be clearly associated with their host galaxies. The galaxies, if

530: they exist and are not obscured, will be some $\sim10-12$ mag brighter

531: than the GCs, depending on the relative roles of merging and

532: luminosity evolution. At $K_{AB}\simeq 20-22$ mag galaxies have mean

533: sky separations of $\sim 20-40\arcsec$, so the sky will be effectively

534: covered, albeit with a concentration toward galaxies, or, the

535: still-dark halos that will become the sites of galaxies.

536: 

537: The GC-galaxy cross-correlation calculation ignores the contribution

538: due to galaxy-galaxy clustering. The same of calculation shows that

539: the much shallower $\gamma=1.8$ of galaxy clustering the

540: auto-correlation angle is about 2 arcsec, for $r_{00}=5\hmpc$. For the

541: GCs their steep cross-correlation with galaxies allows them to rapidly

542: climb out of the projected distribution, which does not occur for the

543: galaxy-galaxy correlation. The galaxy-galaxy contribution will only be

544: visible at about an arc-minute, where the projected clustering

545: amplitude is only $\sim 0.03$. At angles less than 20\arcsec\ the

546: contribution is less than 10\%, given our modelling for clustering.

547: Of course globular cluster formation during merging is a special case.

548: 

549: 

550: \section{Discussion and Conclusions}

551: 

552: Globular clusters are, in the main, very old objects, likely formed in

553: the first quartile of the age of the universe, implying strong

554: luminosity evolution at high redshift. The combination of

555: k-corrections and luminosity evolution put the bulk of their energy in

556: the 3 to 5 micron bands. For a fairly wide range of density and

557: luminosity evolution models there should be approximately $10^7$ GC

558: per square degree per magnitude, with a continuing steep rise in

559: counts.  In the optical bands the counts rise slowly with few GC

560: appearing beyond redshift one.

561: 

562: Source confusion noise in flux and position measurements increases in

563: proportion to the density of sources relative to the beam density,

564: $(\ln{2}/\pi) (D/\lambda)^2$

565: \citep{scheuer,condon}.  The problems associated with confusion

566: begin to arise when the source density is about $\sim5$\% of the beam

567: density. Moreover the strong clustering of GC toward galaxies will

568: create enhanced confusion in the neighborhood of galaxies.  A

569: diffraction limited 6m telescope operating at $4\mu$ will have one

570: source per beam (severe confusion) at a sky density $1.5\times 10^8$

571: per square degree. For the relatively steep source counts found here

572: and the high confidence detections that would be of interest to

573: photometric redshift estimation techniques, the source density below

574: about 1~nJy presents an issue to be carefully approached. In detail

575: this problem could be more quantitatively addressed with simulated

576: observations using the predicted counts.  In a future paper we will

577: also consider a more detailed model that incorporates dust and

578: emission line nebula effects and a number of potential astrophysical

579: complications.

580: 

581: The large density of high redshift GCs is both an opportunity and a

582: challenge. In as much as GC are key indicators of how the extended low

583: metallicity stellar halos of galaxies came into being, observations at

584: nJy flux levels will directly probe their origins. On the other hand,

585: the sky densities and flux levels are similar to those predicted for

586: zero metallicity, ``first light'' objects. It will require some care

587: to distinguish a young cluster of fairly normal stars with strong

588: ionizing radiation from the unusual zero metallicity stars that are

589: the first luminous objects. It will be fascinating to understand the

590: relationship between these two ``early light'' populations.

591: 

592: \acknowledgements

593: 

594: I thank Mike Fall \& Qing Zhang for providing the results of their

595: mass evolution models.  Mike's comments also improved the presentation

596: of the results. Chris Pritchet and Sidney van den Bergh provided

597: inspiration and comments on this subject.  Research support from NSERC

598: and CIAR are gratefully acknowledged.

599: 

600: \begin{thebibliography}{}

601: 

602: 

603: \bibitem[Aaronson Cohen Mould \& Malkan 1978]{vk_colors} 

604: 	Aaronson, M., Cohen, J.~G., Mould, J., \& Malkan, M.\ 1978, 

605: 	\apj, 223, 824

606: \bibitem[Ashman \& Zepf 2001]{ashman+zepf} 

607: 	Ashman, K.~M.~\& Zepf, S.~E.\ 2001, \aj, 122, 1888

608: \bibitem[Carlberg,~et al. 2000]{mergers} 

609: 	Carlberg, R. G., and 13 colleagues 2000, \apjl, 532, L1

610: \bibitem[Cen 2001]{cen}

611: 	Cen, R.\ 2001, \apj, 560, 592

612: \bibitem[Cole et al. 2001]{cole_etal}

613: 	Cole, S.~et al.\ 2001, \mnras, 326, 255

614: \bibitem[Condon 1974]{condon}

615: 	Condon, J.~J.\ 1974, \apj, 188, 279

616: \bibitem[Cote Marzke \& West 1998]{cmw} 

617: 	Cote, P., Marzke, R.~O., \& West, M.~J.\ 1998, \apj, 501, 554

618: \bibitem[Couchman \& Rees 1986]{hugh+martin}

619: 	Couchman, H.~M.~P.~\& Rees, M.~J.\ 1986, \mnras, 221, 53

620: \bibitem[Fall \& Rees 1977]{fall+rees}

621: 	Fall, S. M. \& Rees, M. J. 1977, \mnras, 181, 73P

622: \bibitem[Fall \& Zhang 2001]{fall+zhang}

623: 	Fall, S. M. \& Zhang, Q. 2001, \apj, 561, 751

624: \bibitem[Fioc \& Rocca-Volmerange 1997]{pegase}

625:         Fioc, M.\ \& Rocca-Volmerange, B.\ 1997, \aap, 326, 950

626: \bibitem[Gardner, Sharples, Frenk, \& Carrasco 1997]{gardner}

627: 	Gardner, J.~P., Sharples, R.~M., Frenk, C.~S., \& 

628: 	Carrasco, B.~E.\ 1997, \apjl, 480, L99

629: \bibitem[Haiman \& Loeb 1997]{haiman+loeb}

630: 	Haiman, Z. \& Loeb, A. 1997, \apj, 483, 21

631: \bibitem[Haiman Abel \& Rees 2000]{HAR} 

632: 	Haiman, Z., Abel, T., \& Rees, M.~J.\ 2000, \apj, 534, 11

633: \bibitem[Haiman Abel \& Madau 2001]{HAM}

634: 	Haiman, Z., Abel, T., \& Madau, P.\ 2001, \apj, 551, 599

635: \bibitem[Harris 1991]{harris_ext}

636: 	Harris, W. E. 1991, \araa, 29, 543-79

637: \bibitem[Harris 1996]{harris_mw} 

638: 	Harris, W.~E.\ 1996, \aj, 112, 1487

639: \bibitem[Harris \& van den Bergh 1981]{S} 

640: 	Harris, W.~E.~\& van den Bergh, S.\ 1981, \aj, 86, 1627

641: \bibitem[Larsen et al. 2001]{larsenetal} 

642: 	Larsen, S., Brodie, J.~P., Elmegreen, B.~G., 

643: 	Efremov, Y.~N., Hodge, P.~W., \& Richtler, T.\ 2001, \apj, 556, 801

644: \bibitem[Le F{\` e}vre et al. 2000]{cfrs_mg}

645:         Le F{\` e}vre, O.\ et al.\ 2000, \mnras, 311, 565

646: \bibitem[Searle \& Zinn 1978]{searle+zinn}

647: 	Searle, L.~\& Zinn, R.\ 1978, \apj, 225, 357

648: \bibitem[Scheuer 1974]{scheuer}

649: 	Scheuer, P. A. G., 1974, \mnras, 167, 329

650: \bibitem[Shapley et al. 2001]{shapley}

651: 	Shapley, A.~E., Steidel, C.~C., Adelberger, K.~L., 

652: 	Dickinson, M., Giavalisco, M., \& Pettini, M.\ 2001, 

653: 	\apj, 562, 95

654: \bibitem[VandenBerg 2000]{donv} 

655: 	VandenBerg, D.~A.\ 2000, \apjs, 129, 315

656: \bibitem[van den Bergh 1979]{vdb}

657:         van den Bergh, S. 1979, {\it in Scientific Research with the

658:         Space Telescope}, IAU Colloquium 54, ed. M. S. Longair and J. W.

659:         Warner, pp 151-161.

660: \bibitem[Zepf et al. 1999]{zepf}

661: 	Zepf, S.~E., Ashman, K.~M., English, J., Freeman, K.~C., \& 

662: 	Sharples, R.~M.\ 1999, \aj, 118, 752

663: \bibitem[Zhang \& Fall 1999]{zf}

664: 	Zhang, Q.~\& Fall, S.~M.\ 1999, \apjl, 527, L81

665: 

666: 

667: \end{thebibliography}

668: 

669: 

670: \newpage

671: \newcounter{figi}

672: \newcommand{\nfig}{\addtocounter{figi}{1}\thefigi}

673: 

674: \figcaption[fig1.eps]

675: {The redshift distribution for $m_R(AB)= 25, 26, 27$ and 28 mag,

676: comparing our evolving luminosity function (solid line) with its

677: no-evolution form (dotted). 

678: \label{fig:nzR}}

679: 

680: \figcaption[fig2.eps]

681: {The redshift distribution in the L band (around 3.5 microns) where

682: the GCs form over the time range of zero to one Gigayear in 

683: 10 bursts of 10Myr. The solid line is for the density evolution model the

684: dotted line is for no density evolution. Curves are presented

685: for 0.2, 0.5, 1, 2, 4, 10 and 20 nJy, equivalent to 33.15, 32.15,

686: 31.4, 30.65, 29.9, 28.9 and 28.15 AB mag, respectively.

687: \label{fig:nzL}}

688: 

689: \figcaption[fig3.eps]

690: {The redshift distribution in the L band for 10 bursts of 10Myr of GC

691: formation over the 1 to 2 Gyr time interval. Line types are as in

692: Fig.~\ref{fig:nzL}. The same magnitude limits as in Fig~\ref{fig:nzL}

693: are used.

694: \label{fig:nzL_alt}}

695: 

696: \figcaption[fig4.eps]

697: {The number per magnitude per square degree as a function of limiting

698: AB magnitude for the V (triangles, orange), R (diamonds, light green),

699: J (pentagons, green), K (hexagons, blue), L (heptagons, purple) and M

700: (octagons, red) bands. The solid lines are evolving density models and

701: the dotted for fixed co-moving density models.

702: \label{fig:nm}}

703: 

704: \figcaption[fig5.eps]

705: {The L band counts for cluster formation in the 0-1 Gyr interval

706: (triangles, orange), the 1-2 Gyr interval (diamonds, light green), and

707: exponential models with $\tau=1$~Gyr (pentagons, green), 2~Gyr

708: (hexagons, blue) and 4~Gyr (heptagons, purple) for both evolving and

709: non-evolving mass functions.

710: \label{fig:nmE}}

711: 

712: \begin{figure} \figurenum{\nfig}\includegraphics{fig1.eps}\caption{}\end{figure}  

713: \begin{figure} \figurenum{\nfig}\includegraphics{fig2.eps}\caption{}\end{figure}  

714: \begin{figure} \figurenum{\nfig}\includegraphics{fig3.eps}\caption{}\end{figure}  

715: \begin{figure} \figurenum{\nfig}\includegraphics{fig4.eps}\caption{}\end{figure}  

716: \begin{figure} \figurenum{\nfig}\includegraphics{fig5.eps}\caption{}\end{figure}  

717: 

718: \end{document}

719: