1: %\documentclass[preprint]{aastex}

2: \documentclass[useAMS,usenatbib]{mn2e}

3: %\documentclass[preprint,useAMS,usenatbib]{mn2e}

4: \bibliographystyle{mn2e}

5: \usepackage{epsfig}

6: \usepackage{amsmath}

7: %\usepackage{mn}

8: %\usepackage{mn-nat}

9: \newcommand{\be}{\begin{equation}}

10: \newcommand{\beq}{\begin{equation}}

11: \newcommand{\ba}{\begin{eqnarray}}

12: \newcommand{\ee}{\end{equation}}

13: \newcommand{\eeq}{\end{equation}}

14: \newcommand{\ea}{\end{eqnarray}}

15: \newcommand{\msun}{$M_{\odot}\hspace{1mm}$}

16: \newcommand{\wmap}{{\it WMAP }}

17: \newcommand{\kel}{$^{\circ} K\hspace{1mm}$}

18: \newcommand{\hs}{\hspace{1mm}}

19: \newcommand{\vs}{\vspace{2mm}}

20: \newcommand{\lya}{Ly$\alpha \hspace{1mm}$}

21: \newcommand{\ha}{H$\alpha \hspace{1mm}$}

22: \newcommand{\kms}{km s$^{-1}\hspace{1mm}$}

23: \newcommand{\cm}{cm$^{2}\hspace{1mm}$}

24: \newcommand{\GHz}{{\rm GHz}}

25: \newcommand{\Mpc}{{\rm Mpc}}

26: \newcommand{\kpc}{{\rm kpc}}

27: \newcommand{\dPsidL}{$\partial \log \Psi/\partial \log L_B$}

28: \newcommand{\apj}{ApJ}

29: \newcommand{\aap}{A\&A}

30: \newcommand{\apjl}{ApJL}

31: \newcommand{\mnras}{MNRAS}

32: \newcommand{\aj}{AJ}

33: \newcommand{\apjs}{ApJS}

34: \newcommand{\procspie}{Proc. SPIE}

35: \newcommand{\nat}{{\it Nature}}

36: \newcommand{\pasj}{PASJ}

37: \newcommand{\araa}{ARA\&A}

38: % definition to produce a "less than or similar to" symbol

39: \def\lsim{~\rlap{$<$}{\lower 1.0ex\hbox{$\sim$}}}

40: % definition to produce a "greater than or similar to" symbol

41: \def\gsim{~\rlap{$>$}{\lower 1.0ex\hbox{$\sim$}}}

42: %

43: \begin{document}

44: 

45: \title[CMB Anisotropies From LBGs]{CMB Anisotropies from Outflows in Lyman Break Galaxies}

46: 

47: \author[Daniel Babich and Abraham Loeb]

48: {Daniel Babich$^{1,2}$\thanks{E-mail:dbabich@cfa.harvard.edu} and Abraham

49: Loeb$^{1}$\thanks{E-mail:aloeb@cfa.harvard.edu}\\ $^1$

50: Astronomy Department, Harvard University, 60 Garden Street, Cambridge, MA

51: 02138, USA\\

52: $^2$ Current address: California Institute of Technology, Theoretical Astrophysics, MC 130-33

53: Pasadena, CA 91125, USA}

54: 

55: %

56: \date{\today}

57: \pagerange{\pageref{firstpage}--\pageref{lastpage}} \pubyear{2006}

58: %

59: \maketitle

60: %

61: \label{firstpage}

62: %

63: \begin{abstract}

64: Thomson scattering of the Cosmic Microwave Background (CMB) on moving

65: electrons in the outflows of Lyman Break Galaxies (LBGs) at redshifts 2--8

66: contributes to the small-scale CMB anisotropies. The net effect produced by

67: each outflow depends on its level of deviation from spherical symmetry,

68: caused either by an anisotropic energy injection from the nuclear starburst

69: or quasar activity, or by an inhomogeneous intergalactic environment.  We

70: find that for plausible outflow parameters consistent with spectroscopic

71: observations of LBGs, the induced CMB anisotropies on arcminute scales

72: reach up to $\sim 1 \mu$K, comparable to the level produced during the

73: epoch of reionization.

74: \end{abstract}

75: 

76: \begin{keywords}

77: cosmology -- cosmic microwave background -- galaxies:high-redshift

78: \end{keywords}

79: 

80: \section{Introduction}

81: Several experiments to observe the Cosmic Microwave Background (CMB)

82: anisotropies on arcminute scales are currently, or will soon be, underway

83: \citep{Kosowsky03,Ruhl04, Lo05}.  These experiments plan to measure the CMB

84: power-spectrum for a spherical harmonic multipole index of $10^3 \le \ell

85: \le 10^4$ at several frequencies centred around the thermal

86: Sunyaev-Zel'dovich null of 217 \GHz. Photon diffusion damps the CMB

87: anisotropies on these small scales during cosmological recombination at

88: redshift $z\sim 10^3$ \citep{Silk68}, and so any observed signal must

89: originate at much lower redshifts.  Indeed, the above experiments plan to

90: constrain the epoch of reionization and the growth of structure in the low

91: redshift universe \citep{Zahn05}.

92: 

93: The primary physical mechanism which is responsible for the small scale CMB

94: anisotropies is Thomson scattering of CMB photons off moving electrons.

95: Any peculiar velocity induces a Doppler anisotropy of the scattered

96: radiation along the direction of motion.  This accounts for the CMB

97: anisotropies produced by the peculiar velocities of clusters (the so-called

98: {\it kinetic Sunyaev Zel'dovich effect}) \citep{Zeldovich80}, by peculiar

99: velocities of linear overdensities in the intergalactic medium (the

100: so-called {\it Ostriker-Vishniac effect}) \citep{Ostriker86, Vishniac87}

101: and by the peculiar velocities of the fluctuations in the ionization

102: fraction during patchy reionization \citep{Gruzinov98}.

103: 

104: In this {\it Letter} we examine the contribution of outflows in Lyman Break

105: Galaxies (LBGs) to the small-scale CMB anisotropies. LBGs are believed to

106: be the ancestors of present-day luminous elliptical galaxies. They are

107: observed to produce gas outflows with velocities of several hundred \kms

108: (see \cite{Giavalisco02} for a comprehensive review). In contrast with the

109: traditional kinetic Sunyaev-Zel'dovich effect where the bulk velocity of

110: the virialized gas is responsible for the induced CMB anisotropy, we focus

111: here on the Doppler effect of the outflowing gas and ignore any bulk motion

112: of the LBG as a whole (which produced a smaller effect at the redshifts of

113: interest). This bulk effect is included in standard calculations of the

114: non-linear generalization of the Ostiker-Vishniac effect \citep{Hu00}.

115: 

116: The contribution from a single LBG to the fractional temperature

117: fluctuation of the CMB can be expressed as \be \label{los_int} \frac{\Delta

118: T}{T} = - \int dl \sigma_T n_e\frac{\vec{n} \cdot \vec{v}}{c}, \ee where

119: $\sigma_T = 6.65 \times 10^{-25}$ \cm is the Thomson cross section, $n_e$

120: is the electron number denisty, $\vec{v}$ is the electron peculiar

121: velocity, $c$ is the speed of light and $\hat{n}$ is the observer's

122: line-of-sight toward the LBG \citep{Zeldovich80}. The integration traces

123: the photon's path through the LBG outflow.

124: 

125: The radial extent of the outflow is found by solving the corresponding

126: hydrodynamics equations. These coupled non-linear partial differential

127: equations can be reduced to a single ordinary differential equation

128: \citep{Tegmark93, Furlanetto03} under the assumption that the gas swept-up

129: by the outgoing blast wave lies in a thin shell behind the propagating

130: shock front (the so-called {\it thin shell approximation}). The validity of

131: this approximation is illustrated by the self-similar Sedov-Taylor-von

132: Neumann solution for a point explosion in which $90\%$ of the swept-up mass

133: resides in a shell of thickness $10\%$ of the outflow's radius (see

134: Ostriker \& McKee 1988 and Ikeuchi et al. 1983 for additional discussion on

135: this approximation).  The thin shell approximation allows us to treat the

136: radiative transfer of CMB photons through the shock front in a plane

137: parallel geometry. When the thickness of the shock front is small compared

138: with the shock front's radius of curvature, the path length through the

139: shock front can be expressed as \be \delta l \approx \frac{\delta R}{|

140: \hat{n}\cdot \hat{v} |}, \ee where $\delta R$ is the thickness of the shock

141: front. Using conservation of mass and the density compression ratio for a

142: strong adiabatic shock, one gets $\delta R/R = (\gamma -1)/(3\gamma + 3)$,

143: where $R$ is the radius of the outflow and $\gamma$ is the adiabatic index

144: of the gas \citep{Ostriker88}.

145: 

146: Within the thin shell approximation the line-of-sight integration in

147: Eq. (\ref{los_int}) is simplified to \be \left(\frac{\Delta T}{T}

148: \right)_i = - \frac{\sigma_T~\delta R}{c}~[n_e(\vec{r}_1) v(\vec{r}_1) -

149: n_e(\vec{r}_2) v(\vec{r}_2)], \ee where $\vec{r}_1$ and $\vec{r}_2$ are the

150: location, in a coordinate system centred on the LBG, on the shock front 

151: where the line-of-sight respectively enters and exits the shock front. 

152: The subscript $i$ labels the contribution from a particular LBG.

153: 

154: If the outflow is spherically symmetric then there is an exact cancellation

155: of the anisotropies produced at the entry and exit points of the

156: line-of-sight through the LBG shock front\footnote{Note that even if the

157: outflow had a perfect spherical shape, the finite light crossing time

158: through the outflow would produce a net non-zero signal because the flow

159: parameters are time-dependent.  This effect would produce a signal

160: $\sigma_T n_e R (\Delta v)/c \sim \sigma_T n_e (dv/dt) (R/c)^2$ that is

161: extremely small and is ignored here.}. However, observations of

162: low-redshift starburst galaxies, which serve as analogs of the higher

163: redshift LBGs, show evidence for highly non-spherical outflow geometries

164: \citep{Martin99}. It is therefore reasonable to expect that the early

165: stages of LBG outflows, whether they are driven by starburst or quasar

166: activity\footnote{In this {\it Letter} we will only include the feedback

167: driven by supernovae. Energy input from quasars will lead to enhancements

168: in the bubble size and outflow velocity [see \cite{Furlanetto01} for a

169: description of quasar outflows].}, would produce CMB anisotropies.

170: 

171: For simplicity, we will assume that the outflow is axisymmetric about some

172: axis $\hat{z}$ (possibly the rotation axis of the galactic disc or possibly

173: the jet axis of a central quasar) and expand both the outflow velocity and

174: shock front density in Legrendre polynomials, $ \vec{v}(\vec{r}) =

175: \hat{r}~\sum_{\ell} v_{\ell}(r) P_{\ell}(\hat{z} \cdot \hat{r})$ and

176: $n_e(\vec{r}) = \sum_{\ell} n_{\ell}(r) P_{\ell}(\hat{z} \cdot \hat{r})$,

177: where we have assumed that the velocity field is radial ($\hat{v} \equiv

178: \hat{r}$).  Even if the outflow began highly collimated, it will eventually

179: isotropize as it propagates into the intergalactic medium (IGM; an

180: analogous tendency exists in relativistic flows, see Ayal \& Piran

181: 2001). On longer timescales, IGM inhomogeneities, such as filaments and

182: voids, will once again make the outflows non-spherical. Numerical

183: simulations are needed in order to properly model these effects.  

184: Here we parameterize the level of asphericity in the outflow by a

185: coefficient $\epsilon$ (see Eq. \ref{eq:dT2} below).

186: 

187: The outline of this {\it Letter} is as follows.  In \S 2 we analyze the

188: CMB anisotropy induced by a single LBG with an arbitrarily non-spherical

189: outflow. In \S 3 we calculate the resulting CMB power spectrum. In \S 4

190: we present numerical results and in \S 5 we summarize our conclusions.

191: Throughout our discussion, we will assume the WMAP3 cosmological model

192: \citep{Spergel06}.

193: 

194: \section{Single LBG Signal}

195: The signal along a given line-of-sight includes contributions from LBGs

196: with different masses, outflow ages, orientations of the outflow symmetry

197: axis and line-of-sight impact parameters. For simplicity, we begin by

198: analyzing the expectation value for LBGs formed at a certain redshift and

199: of certain mass and age. Since the net effect from a single LBG can be

200: either positive or negative we will find that the mean of the signal along

201: a given direction is always zero, but a non-zero variance will be produced

202: due to Poisson fluctuations in a manner equivalent to a random walk. The

203: analysis will be done separately for the distinct cases where either one or

204: two lines of sight intersect the same LBG outflow.

205: 

206: \subsection{One Sightline}

207: First we consider the case where one line-of-sight intersects a single LBG

208: outflow.  Averaging over symmetry axis orientation and impact parameter,

209: the expectation value of the fractional temperature perturbation is \be

210: \label{eq:sing_los} \left \langle \left( \frac{\Delta T}{T} \right)_i

211: \right \rangle = \int d^2\hat{z} P(\hat{z}) \int d^2 \vec{b}~P(b)

212: \left(\frac{\Delta T}{T} \right)_i.  \ee Here $\hat{z}$ is the symmetry

213: axis of the outflow and $\vec{b}$ is the impact parameter at which the

214: line-of-sight enters the LBG (see Fig. \ref{fig:outflow} for definitions

215: of the variables we use in describing the LBG outflow).

216: \begin{figure}

217: \includegraphics[width=8cm]{diag_tmp.ps}

218: \caption[]{Schematic illustration of the geometry for a single LBG

219: outflow. The black circle represents the shock front, the orange arrow

220: labelled $\hat{n}$ is the line-of-sight from the observer, the red arrows

221: labelled $\vec{r}_1, \vec{r}_2$ are the entry and exit points of the

222: line-of-sight on the outflow, the blue arrow labelled $\hat{z}$ is the

223: symmetry axis of the outflow and the green arrow labelled $\vec{b}$ is the

224: impact parameter of the line-of-sight.}

225: \label{fig:outflow} 

226: \end{figure}

227: We assume uniform probability distributions for the symmetry axis

228: orientation, $P(\hat{z}) = 1/4\pi$, and for the impact parameter,

229: $P(\vec{b}) = 1/\pi R^2.$

230: 

231: Now we change the integration variable from the impact parameter $\vec{b}$

232: to the location of entrance point of the line-of-sight into the LBG 

233: outflow $\hat{r}$. Thus, the expectation value defined in Eq. (\ref{eq:sing_los}) 

234: can be written as 

235: \be \left \langle \left( \frac{\Delta T}{T} \right)_i \right \rangle =

236: \int_{4\pi} \frac{d^2\hat{z}}{4\pi} \int_{2\pi} \frac{d^2\hat{r}}{\pi}

237: \left(\frac{\Delta T}{T} \right)_i, \ee 

238: where the integration over the

239: location of the impact parameter is restricted to a single hemisphere.

240: Performing this integration we find that the signal vanishes on average, as

241: expected from the fact that the net signal for a given LBG is just as

242: likely to be negative as it is to be positive.

243: 

244: There will be a non-zero contribution from a given LBG since the induced

245: anisotropies will be different at the entry and exit points of the

246: line-of-sight through the shock front. Because the LBG symmetry axes is

247: randomly oriented with respect to the direction of the observer, the signal

248: vanishes once the average over this direction is done. This implies that

249: the mean signal and therefore the one-point function vanishes. We are

250: ultimately interested in the two-point correlation function and the related

251: power spectrum. There are two contributions to the power spectrum

252: \citep{Cooray02}. The first is a clustering (two halo) term, originating

253: from the correlated perturbations in the cold dark matter density produced

254: during inflation.  The second is a Poisson (one halo) term, originating

255: from Poisson fluctuations in the number density of halos. The one-point

256: function vanishes because net temperature anisotropy produced by a given

257: LBG is uncorrelated with the signal from other LBGs along the

258: line-of-sight. The clustering term only implies that the number density of

259: LBGs nearby another LBG is greater than average, not that the symmetry axes

260: are somehow correlated\footnote{On small scales, the asphericity of the

261: outflows may be correlated because they propagate into the same

262: inhomgeneous IGM, or because tidal gravitational forces produced

263: correlations in the shapes of nearby galaxies \citep{Mackey02}.}. Since

264: there is no correlation in the signals between the two distinct

265: lines-of-sight, there will be no contribution to the resulting CMB power

266: spectrum from a clustering term.

267: 

268: \subsection{Two Sightlines}

269: For a single sightline through each outflow we found that the two-point

270: correlation function vanishes, as the symmetry axes of LBGs are randomly

271: oriented. When both lines-of-sight intersect the same LBG this cancellation

272: does not take place. The average value of the two-point temperature

273: anisotropy when both lines of sight intersect the same LBG is \be \left

274: \langle \left( \frac{\Delta T}{T} \right)_i^2 \right \rangle = \int

275: d^2\hat{z} P(\hat{z}) \int d^2 \vec{b}_1~P(\vec{b}_1) \int

276: d^2\vec{b}_2~P(\vec{b}_2) \left(\frac{\Delta T}{T} \right)^2_i \ee

277: Performing the relevant integrations we find a non-zero answer when the

278: product $n_e(\vec{r})~v(\vec{r})$ has odd parity. The induced fluctuations

279: produced when two lines of sight intersect the same LBG are due to Poisson

280: fluctuations. As mentioned above, we parameterize the deviation from

281: sphericity with a fudge factor $\epsilon$. Then the expectation value for

282: two lines-of-sight intersecting the same LBG outflow region is \be

283: \label{eq:dT2} \left \langle \left( \frac{\Delta T}{T} \right)_i^2 \right

284: \rangle = \sigma^2_T n^2_e \delta R^2 \frac{\epsilon^2 v^2}{c^2}.  \ee

285: 

286: \section{Poisson Fluctuations}

287: We have seen that there is an exact cancellation when the two

288: lines-of-sight intersect two different LBG outflows and that a non-zero

289: signal arises when the two lines-of-sight intersect a single LBG

290: outflow. Correlations in the one LBG terms 

291: are produced by Poisson fluctuations in the number of intercepted LBGs. We

292: will analyze this effect in two stages: first, we will consider the effect

293: along a given line-of-sight (pencil-beam survey) and then generalize to the

294: case of a finite beam size.

295: 

296: \subsection{Pencil Beam Survey}

297: The temperature anisotropies induced by LBGs of halo mass between $M$ and

298: $M+dM$, formed between redshifts $z_f$ and $z_f + dz_f$ and scattering the

299: CMB between redshifts $z$ to $z+dz$ is \be \frac{\Delta T}{T} =

300: \sum_{n=1}^{\infty} P_{dN}(n_{\rm LBG}) \sum_{i=1}^{n_{\rm LBG}}

301: \left(\frac{\Delta T}{T}\right)_i, \ee where $P_N(n_{\rm LBG})$ is the

302: Poisson probability that $n_{\rm LBG}$ LBGs are observed and $dN$ is the

303: mean number of LBGs in the redshift interval between $z$ and $z+dz$, \be dN

304: = \pi R^2 \frac{-c~dz}{(1+z)H(z)} \frac{d^2n}{dM dz_f} dM dz_f, \ee where

305: $R$ is the radius of an outflow at redshift $z$ produced by an LBG of mass $M$

306: formed at redshift $z_f$. The total signal is found by integrating over

307: $dM$, $dz_f$ and $dz$.  The expectation value toward a given line of sight

308: vanishes since the expectation value for the temperature anisotropy

309: produced by a single LBG vanishes. Nevertheless the variance of the signal

310: does not vanish due to Poisson fluctuations. For a single sightline through

311: an LBG outflow region the contribution to the temperature anisotropies can

312: be either positive or negative with equal probability, however for two

313: sightlines the contribution is always non-negative.  The resultant

314: variance, which is the case of two sightlines at a separation less than the

315: characteristic angular size of LBG outflows, is \be \left\langle

316: \left(\frac{\Delta T}{T}\right)^2\right\rangle = \int dN \left\langle

317: \left(\frac{\Delta T}{T}\right)_i^2\right\rangle.  \ee The {\it rms} value

318: of the anisotropy can be evaluated in terms of the unknowns and the time

319: changing Legendre coefficients $n_{\ell}$ and $v_{\ell}$. For simplicity,

320: we adopt the spherically symmetric solution for the shock radius and

321: velocity, and parameterize the degree of asymmetry in the outflow by the

322: fudge factor $\epsilon$.

323: 

324: \subsection{Window Function Effects}

325: The average angular size of an LBG outflow $\bar{\theta}_{\rm LBG} \approx

326: 10\arcsec$ is below the resolution of the upcoming generation of

327: experiments, and so we must properly account for the beam's window

328: function. The observed temperature anisotropy will be an average over a

329: window function $W(\hat{n})$ \be \frac{\Delta \tilde{T}}{T}(\hat{n}) = \int

330: d^2 \hat{n}'~W(\hat{n}-\hat{n}')~\frac{\Delta T}{T}(\hat{n}').  \ee For

331: simplicity, we will take the window function shape to be a top hat of

332: angular size $\theta$, namely $W(\hat{n}) = 1/\theta^2 \mbox{ if }

333: |\hat{n}| \le \theta$, and $W(\hat{n})= 0 \mbox{ if } |\hat{n}| > \theta$.

334: 

335: The variance in an angular aperature defined by the window function, \be \label{eq:cmb_var}

336: \left \langle \left( \frac{\Delta \tilde{T}}{T}\right)^2

337: \right\rangle_{\theta} = \int d^2\hat{n}'_1

338: d^2\hat{n}'_2~W(\hat{n}'_1)~W(\hat{n}'_2)~\left\langle\frac{\Delta

339: T}{T}(\hat{n}'_1)~\frac{\Delta T}{T}(\hat{n}'_2) \right\rangle, \ee is

340: related to the power spectrum as \be \frac{\ell(\ell+1)C_{\ell}}{2\pi}

341: \approx \left \langle\left( \frac{\Delta \tilde{T}}{T}\right)^2

342: \right\rangle_{\theta = 2\pi/\ell}.  \ee

343: 

344: In order for Poisson fluctuations to give a nonzero result, the two lines

345: of sight $\hat{n}'_1$ and $\hat{n}'_2$ must intersect the same

346: LBG. Therefore they must be separated by less than $\bar{\theta}_{\rm

347: LBG}$. This requirement allows us to evaluate Eq. (\ref{eq:cmb_var}) as \be

348: \left \langle \left( \frac{\Delta \tilde{T}}{T}\right)^2

349: \right\rangle_{\theta} = \frac1{\theta^2} \int dN \left( \frac{R}{D_{\rm

350: A}(z)} \right)^2 \left \langle \left( \frac{\Delta T}{T}\right)_i^2

351: \right\rangle.  \ee Note that the Fourier multipole corresponding to

352: $\bar{\theta}_{\rm LBG}$ is $\ell = 2\pi/\bar{\theta}_{\rm LBG}$.  Here

353: $D_{\rm A}(z)$ is the angular diameter distance to redshift $z$.

354: 

355: A broad window function allows for a larger number of LBGs within the beam.

356: The window function is normalized such that it integrates to unity, and so

357: one is observing the fractional fluctuations in the signal as an average

358: over $\theta^2/\theta^2_{\rm LBG}$ independent coherence patches in the

359: beam. This is equivalent to a fractional fluctuation of $1/\sqrt{N}$ as

360: expected from Poisson fluctuations.

361: 

362: \section{Results}

363: We numerically solve for the evolution of the shock front in the thin shell

364: approximation \citep{Tegmark93, Furlanetto03}, taking into account the

365: effects of the halo gravity and the self gravity of the mass shell, the

366: internal pressure of shocked IGM and the acceleration due to the

367: cosmological constant. When calculating the internal pressure we include

368: Compton cooling, adiabatic cooling, and the addition of shock heated

369: gas. Initially we assume that a fraction $f_{*} = 0.1$ of the baryons

370: assembled into the central LBG form stars with a Scalo initial mass

371: function. In this case there is one supernova per $126$ \msun of star

372: formation \citep{Furlanetto03}. Supernovae characteristically produce $10^{51}$ 

373: ergs of energy, but only a

374: small fraction of this energy, $f_{\rm SN}$, couples to the outflow, with

375: the rest being radiated away. We adopt a value of $f_{\rm SN}=0.01$; see

376: \cite{Furlanetto03} and reference within for a more detailed description

377: of our outflow model.

378: 

379: Assuming a Sheth-Tormen mass function \citep{Sheth99} we include the

380: effects from all possible halos above the minimum galaxy mass. The lowest

381: galaxy mass is determined by the maximum between the Jeans filtering mass

382: and the cooling mass (dictated by the halo's ability to cool through atomic

383: hydrogen line emission)\footnote{Note that we include haloes with masses

384: below the observational sensitivity for LBGs, as there is no fundamental

385: reason to exclude these haloes.}. We allow the LBGs to form

386: between\footnote{We include this lower redshift limit because the

387: phenomenon of downsizing, as well as the evolution of AGN luminosity

388: function, implies that massive galaxies finished forming stars around that

389: redshift.} $z=2$ and $z=8$. We continue to allow the outflow to evolve

390: until its velocity equals to Hubble flow at its radius from the LBG or

391: until $z=0$. Increasing the upper redshift has little effect on our results

392: because of the higher minimum LBG mass, as well as, the higher velocity of

393: the Hubble flow which causes the LBG outflows to merge with the IGM at an

394: relatively earlier time. Decreasing the minimum LBG mass could have a

395: significant effect on our results because the ratio of the initial outflow

396: velocity to the halo escape velocity at the initial radius scales as

397: $v_{\rm init}/v_{esc} \propto M^{-2/9}$.  However, in low mass haloes the

398: star formation timescale increases and supernova feedback becomes capable

399: of decreasing $f_{*}$.  Some starburst activity continues to lower

400: redshifts, albeit with a reduced intensity compared to high redshift star

401: formation, changing the lower redshift of from $z_F = 2$ to $z_F = 1$ would

402: change our results by a factor of $2$. Since the observed star formation

403: efficiency decreases with decreasing $z_F$, our model, which assumes that

404: star formation and the subsequent supernova feedback only depends on the

405: LBG mass, overestimates this change.

406: 

407: In Fig. \ref{fig:power_spec} we show the CMB power spectrum produced by

408: several secondary mechanisms. The three blue dot-dashed curves denote the LBG outflow

409: signal calculated in this {\it Letter} for the values of $\epsilon = 1,

410: 0.5, 0.25$ from top to bottom. The green dotted curve delineates the Ostriker-Vishniac 

411: effect and the dashed cyan curve describes the patchy reionization effect as 

412: calculated by \cite{Mcquinn06}\footnote{Note that \cite{Mcquinn06} used a 

413: different cosmological model with a higher value of $\sigma_8 = 0.9$. This 

414: artificially raises the amplitude of the results compared to value of $\sigma_8 = 0.73$

415: adopted in this work in line with WMAP3.}  

416: Also, for comparison, we show the primary CMB anisotropies (solid, black curve) and

417: the small scale Cosmic Background Imager (CBI) data points (red triangles) 

418: \citep{Readhead04}.

419: 

420: \begin{figure}

421: \includegraphics[width=8cm]{cl_3yr.ps}

422: \caption[]{Power spectra produced by several secondary mechanisms. The

423: contribution from the LBG outflow is shown as the three blue dot-dashed curves for

424: $\epsilon = 1, 0.5, 0.25$ from top to bottom. The primary CMB anisotropies are shown in the

425: solid-black curve; the Ostriker-Vishniac and patchy reionization

426: contributions as calculated by \cite{Mcquinn06} are dotted-green and

427: dashed-light-blue respectively. The observed CBI data points are plotted as

428: red triangles.}

429: \label{fig:power_spec} 

430: \end{figure}

431: 

432: To characterize our results, let us mention some typical quantities for a

433: common LBG halo of mass $5 \times10^{9} M_{\odot}$ formed at $z_F = 2$. In

434: this case the outflow reaches a maximum comoving radius of 150 \kpc ~ at

435: $z= 0.9$ before it merges with the Hubble flow. At late times the outflow

436: velocity with respect to the local Hubble flow is $(v-HR)/c \sim 7 \times

437: 10^{-4}$ and the Thomson scattering optical depth through the shock front

438: is $\tau \approx 10^{-5}$.

439: 

440: These small characteristic values for the Thomson scattering optical depth

441: and outflow velocity imply that the signal should not be notably

442: polarized. Thomson scattering of a radiation field containing a quadrupole

443: moment will produce polarized radiation.  There are two standard ways in

444: which scattering by a halo can produce polarization -- (i) photons will

445: double scatter in the halo; (ii) the peculiar velocity of the scatterer

446: will induce a quadrupole moment in the radiation field

447: \citep{Zeldovich80}. In the first case, the radiation can scatter in the

448: halo producing an anisotropic radiation field; a second scattering of that

449: radiation field can produce polarization at the level $\mathcal{O}(\tau^2

450: v/c)$.  In the second case, the peculiar velocity of the scatterer

451: perpendicular to its line-of-sight with respect to the observer will induce

452: a quadrupole in the incident radiation field at the order

453: $\mathcal{O}(v^2/c^2)$.  A fraction $\tau$ of the radiation field will

454: scatter and become polarized at the level $\mathcal{O}(\tau v^2/c^2)$.

455: Since values of $\tau$ and $v/c$ are so small, we conclude that the

456: polarization power spectrum (which is sixth order in the small parameters

457: of $\tau$ and $v/c$) is negligible.

458: 

459: \section{Discussion}

460: 

461: The contribution of outflows from Lyman-break galaxies to the CMB

462: power-spectrum on arcminute scales is proportional to the square of their

463: characteristic level of deviation from sphericity, $\epsilon^2$. Future CMB

464: experiments could therefore calibrate the intricate feedback process of

465: galactic outflows on the IGM. Most tools used to study these feedback

466: processes focus on the inner few \kpc ~of the LBGs, even though the shock

467: front is typically located at several of tens or hundreds of \kpc. The

468: secondary CMB anistropies calculated in this work provide a unique probe of

469: these extended perturbed regions around starburst galaxies at high

470: redshifts.

471: 

472: Even though the amplitude of the power spectrum produced by this effect is

473: small, the signal has distinctive spectral and spatial characteristics. The

474: power spectrum has the same scaling with Fourier multipole ($\propto

475: \ell^2$) as radio or IR point sources. However, the frequency dependence of

476: the anisotropies produced by our effect has the standard blackbody

477: spectrum, whereas the radio point sources have a power law frequency

478: spectrum. This distinct feature of our effect may allow future small scale

479: experiments, many of which have excellent frequency coverage, to separate

480: the anisotropies produced by LBGs from radio point sources.

481: 

482: \bigskip

483: 

484: {\bf Acknowledgements}

485: 

486: This work was supported in part by Harvard university grants. D.B. thanks

487: the Harvard Institute for Theory and Computation for its hospitality when

488: this work was completed, and acknowledges helpful conversations with Re'em

489: Sari and Niayesh Afshordi. We also thank Matt McQuinn for making available

490: the numerical results from his work.

491: 

492: \begin{thebibliography}{19}

493: 

494: \bibitem[Ayal \& Piran(2001)]{Ayal01} 

495: Ayal, S., \& Piran, T.\ 2001, \apj, 555, 23 

496: 

497: \bibitem[Cooray \& Sheth(2002)]{Cooray02} 

498: Cooray, A., \& Sheth, R.\ 2002, Phys. Repts., 372, 1 

499: 

500: \bibitem[Furlanetto \& Loeb(2001)]{Furlanetto01} 

501: Furlanetto, S.~R., \& Loeb, A.\ 2001, \apj, 556, 619 

502: 

503: \bibitem[Furlanetto \& Loeb(2003)]{Furlanetto03} 

504: Furlanetto, S.~R., \& Loeb, A.\ 2003, \apj, 588, 18 

505: 

506: \bibitem[Giavalisco(2002)]{Giavalisco02} 

507: Giavalisco, M.\ 2002, \araa, 40, 579 

508: 

509: \bibitem[Gruzinov \& Hu(1998)]{Gruzinov98} 

510: Gruzinov, A., \& Hu, W.\ 1998, \apj, 508, 435 

511: 

512: \bibitem[Hu(2000)]{Hu00} 

513: Hu, W.\ 2000, \apj, 529, 12 

514: 

515: \bibitem[Ikeuchi et al.(1983)]{Ikeuchi83} 

516: Ikeuchi, S., Tomisaka, K., \& Ostriker, J.~P.\ 1983, \apj, 265, 583 

517: 

518: \bibitem[Kosowsky(2003)]{Kosowsky03} Kosowsky, A.\ 2003, New 

519: Astronomy Review, 47, 939

520: 

521: \bibitem[Lo et al.(2005)]{Lo05} 

522: Lo, K.~Y., et al.\ 2005, IAU Symposium, 201, 306 

523: 

524: \bibitem[Mackey et al.(2002)]{Mackey02} 

525: Mackey, J., White, M., \& Kamionkowski, M.\ 2002, \mnras, 332, 788

526: 

527: \bibitem[Martin(1999)]{Martin99} Martin, C.~L.\ 1999, \apj, 513, 156 

528: 

529: \bibitem[McQuinn et al.(2006)]{Mcquinn06} 

530: McQuinn, M., et al.\ 2006, New Astronomy Review, 50, 84 

531: 

532: \bibitem[Ostriker \& Vishniac(1986)]{Ostriker86} 

533: Ostriker, J.~P., \& Vishniac, E.~T.\ 1986, \apjl, 306, L51 

534: 

535: \bibitem[Ostriker \& McKee(1988)]{Ostriker88} 

536: Ostriker, J.~P., \& McKee, C.~F.\ 1988, Reviews of Modern Physics, 60, 1 

537: 

538: \bibitem[Readhead et al.(2004)]{Readhead04}

539: Readhead, A.~C.~S., et al. \ 2004, \apj, 609, 498

540: 

541: \bibitem[Ruhl et al.(2004)]{Ruhl04} Ruhl, J., et al.\ 2004, 

542: \procspie, 5498, 11 

543: 

544: \bibitem[Sheth \& Tormen(1999)]{Sheth99} 

545: Sheth, R.~K., \& Tormen, G.\ 1999, \mnras, 308, 119 

546: 

547: \bibitem[Silk(1968)]{Silk68} Silk, J.\ 1968, \apj, 151, 459 

548: 

549: \bibitem[Spergel et al.(2006)]{Spergel06} 

550: Spergel, D.N. et al.\ 2006, submitted to \apj, astro-ph 0603449

551: 

552: \bibitem[Tegmark et al.(1993)]{Tegmark93} 

553: Tegmark, M., Silk, J., \& Evrard, A.\ 1993, \apj, 417, 54 

554: 

555: \bibitem[Vishniac(1987)]{Vishniac87} 

556: Vishniac, E.~T.\ 1987, \apj, 322, 597 

557: 

558: \bibitem[Zahn et al.(2005)]{Zahn05} 

559: Zahn, O., et al. \ 2005, \apj, 630, 657 

560: 

561: \bibitem[Zel'dovich \& Sunyaev(1980)]{Zeldovich80} 

562: Zel'dovich, Y.~B., \& Sunyaev, R.~A.\ 1980, Pis ma Astronomicheskii Zhurnal, 6, 545 

563: 

564: \end{thebibliography}

565: 

566:  \end{document}

567: 

568: 

569: 

570: 

571: 

572: 

573: 

574: 

575: 

576: 

577: 

578: 

579: 

580: