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

2: \documentclass{emulateapj}

3: \usepackage{natbib}

4: \slugcomment{{\sc Accepted to the ApJ:} August 16, 2008} 

5: \usepackage{lscape}

6: \parskip=5pt

7: 

8: %%%%%  PERSONAL MACROS ... DO NOT MODIFY BELOW %%%%%

9: 

10: \def\CIVdblt{{\rm C~}\kern 0.1em{\sc iv}~$\lambda\lambda 1548, 1550$}

11: \def\MgIIdblt{{\rm Mg~}\kern 0.1em{\sc ii}~$\lambda\lambda 2796, 2803$}

12: \def\NVdblt{{\rm N}\kern 0.1em{\sc v}~$\lambda\lambda 1238, 1242$}  

13: \def\OVIdblt{{\rm O}\kern 0.1em{\sc vi}~$\lambda\lambda 1031, 1037$}

14: \def\SiIVdblt{{\rm Si~}\kern 0.1em{\sc iv}~$\lambda\lambda1394, 1403$}

15: \def\AlIIIdblt{{\rm Al~}\kern 0.1em{\sc iii}~$\lambda\lambda1855,1863$}

16: \def\FeIIdblt{{\rm Fe~}\kern 0.1em{\sc ii}~$\lambda\lambda 2383, 2600$}

17: 

18: \def\AlII{\hbox{{\rm Al~}\kern 0.1em{\sc ii}}}

19: \def\AlIII{\hbox{{\rm Al~}\kern 0.1em{\sc iii}}}

20: \def\CaI{\hbox{{\rm Ca}\kern 0.1em{\sc i}}}

21: \def\CaII{\hbox{{\rm Ca}\kern 0.1em{\sc ii}}}

22: \def\CrII{\hbox{{\rm Cr}\kern 0.1em{\sc ii}}}

23: \def\CII{\hbox{{\rm C~}\kern 0.1em{\sc ii}}}

24: \def\CIII{\hbox{{\rm C~}\kern 0.1em{\sc iii}}}

25: \def\CIV{\hbox{{\rm C~}\kern 0.1em{\sc iv}}}

26: \def\CV{\hbox{{\rm C}\kern 0.1em{\sc v}}}

27: \def\H{\hbox{{\rm H}}}

28: \def\HI{\hbox{{\rm H~}\kern 0.1em{\sc i}}}

29: \def\HII{\hbox{{\rm H~}\kern 0.1em{\sc ii}}}

30: \def\Lya{\hbox{{\rm Ly}\kern 0.1em$\alpha$}}

31: \def\Lyb{\hbox{{\rm Ly}\kern 0.1em$\beta$}}

32: \def\Lyg{\hbox{{\rm Ly}\kern 0.1em$\gamma$}}

33: \def\Lyfive{\hbox{{\rm Ly}\kern 0.1em$5$}}

34: \def\Lysix{\hbox{{\rm Ly}\kern 0.1em$6$}}

35: \def\Lyseven{\hbox{{\rm Ly}\kern 0.1em$7$}}

36: \def\Lyeight{\hbox{{\rm Ly}\kern 0.1em$8$}}

37: \def\Lynine{\hbox{{\rm Ly}\kern 0.1em$9$}}

38: \def\Lyten{\hbox{{\rm Ly}\kern 0.1em$10$}}

39: \def\HeI{\hbox{{\rm He}\kern 0.1em{\sc i}}}

40: \def\HeII{\hbox{{\rm He}\kern 0.1em{\sc ii}}}

41: \def\FeI{\hbox{{\rm Fe~}\kern 0.1em{\sc i}}}

42: \def\FeII{\hbox{{\rm Fe~}\kern 0.1em{\sc ii}}}

43: \def\FeIII{\hbox{{\rm Fe~}\kern 0.1em{\sc iii}}}

44: \def\MnII{\hbox{{\rm Mn}\kern 0.1em{\sc ii}}}

45: \def\MgI{\hbox{{\rm Mg~}\kern 0.1em{\sc i}}}

46: \def\MgII{\hbox{{\rm Mg~}\kern 0.1em{\sc ii}}}

47: \def\MgIII{\hbox{{\rm Mg~}\kern 0.1em{\sc iii}}}

48: \def\MgIV{\hbox{{\rm Mg~}\kern 0.1em{\sc iv}}}

49: \def\NaI{\hbox{{\rm Na}\kern 0.1em{\sc i}}}

50: \def\NV{\hbox{{\rm N}\kern 0.1em{\sc v}}}

51: \def\NII{\hbox{{\rm N}\kern 0.1em{\sc ii}}}

52: \def\NIII{\hbox{{\rm N}\kern 0.1em{\sc iii}}}

53: \def\OVI{\hbox{{\rm O}\kern 0.1em{\sc vi}}}

54: \def\OII{\hbox{[{\rm O}\kern 0.1em{\sc ii}]}}

55: \def\SiII{\hbox{{\rm Si~}\kern 0.1em{\sc ii}}}

56: \def\SiIII{\hbox{{\rm Si~}\kern 0.1em{\sc iii}}}

57: \def\SiIV{\hbox{{\rm Si~}\kern 0.1em{\sc iv}}}

58: \def\SII{\hbox{{\rm S}\kern 0.1em{\sc ii}}}

59: \def\SIII{\hbox{{\rm S}\kern 0.1em{\sc iii}}}

60: \def\SIV{\hbox{{\rm S}\kern 0.1em{\sc iv}}}

61: \def\TiII{\hbox{{\rm Ti}\kern 0.1em{\sc ii}}}

62: \def\ZnII{\hbox{{\rm Zn}\kern 0.1em{\sc ii}}}

63: \newcommand{\kms}{\hbox{km~s$^{-1}$}}

64: \newcommand{\cmsq}{\hbox{cm$^{-2}$}}

65: \newcommand{\cc}{\hbox{cm$^{-3}$}}

66: \def\kms{\hbox{km~s$^{-1}$}}      

67: \def\cmsq{\hbox{cm$^{-2}$}}

68: \def\cc{\hbox{cm$^{-3}$}}

69: \def\etal{et~al.\ }

70: \def\minfit{\sc minfit}

71: \def\DR{\hbox{\sc dr}}

72: 

73: \begin{document}

74: 

75: \title{The Chemical and Ionization Conditions in Weak {\MgII} Absorbers}

76: \author{Anand~Narayanan\altaffilmark{1}, Jane~C.~Charlton\altaffilmark{1}, Toru~Misawa\altaffilmark{1}, Rebecca~E.~Green\altaffilmark{1}, \\ and Tae-Sun Kim\altaffilmark{2}}

77: 

78: \altaffiltext{1}{Department of Astronomy and Astrophysics, The Pennsylvania State University, University Park, PA 16802. Email: anand, charlton, misawa, reg5021@astro.psu.edu}

79: 

80: \altaffiltext{2}{Astrophysiakalisches Institut Potsdam, An der Sternwarte 16, 14482 Potsdam, Germany, {\it tkim@aip.de}}

81: 

82: \begin{abstract} 

83: We present an analysis of the chemical and ionization conditions in a sample of $100$ weak {\MgII} absorbers identified in the VLT/UVES archive of quasar spectra. In addition to {\MgII},

84: we present equivalent width and column density measurements of other low ionization species such as {\MgI}, {\FeII}, {\AlII}, {\CII}, {\SiII} and also {\AlIII}. We find that the column densities of {\CII}  and {\SiII} are strongly correlated with the column density of {\MgII}, with minimal scatter in the relationships. The column densities of {\FeII} exhibit an appreciable scatter when compared with the column density of {\MgII}, with some fraction of clouds having $N(\FeII) \sim N(\MgII)$, in which case the density is constrained to n$_{\H} > 0.05$~{\cc}. Other clouds in which $N(\FeII) << N(\MgII)$ have much lower densities. From ionization models, we infer that the metallicity in a significant fraction of weak {\MgII} clouds is constrained to values of solar or higher, if they are sub-Lyman limit systems. Based on the observed constraints, we hypothesize that weak {\MgII} absorbers are predominantly tracing two different astrophysical processes/structures.  A significant population of weak {\MgII} clouds, those in which $N(\FeII) << N(\MgII)$, identified at both low ($z \sim 1$) and high ($z \sim 2$) redshift, are likely to be tracing gas in the extended halos of galaxies, analogous to the Galactic high velocity clouds. These absorbers might correspond to $\alpha$-enhanced interstellar gas expelled from star-forming galaxies, in correlated supernova events. The $N(\MgII)$ and $N(\FeII)/N(\MgII)$ in such clouds are also closely comparable to those measured for the high velocity components in strong {\MgII} systems. An evolution is found in $N(\FeII)/N(\MgII)$ from $z = 2.4$ to $z = 0.4$, with an absence of weak {\MgII} clouds with $N(\FeII) \sim N(\MgII)$ at high-$z$. The $N(\FeII) \sim N(\MgII)$ clouds, which are prevalent at lower redshifts ($z < 1.5$), must be tracing Type Ia enriched gas in small, high metallicity pockets in dwarf galaxies, tidal debris, or other intergalactic structures.  

85: 

86: \end{abstract}

87: \keywords{galaxies: evolution --- halo --- intergalactic medium --- quasars: absorption lines.}

88: 

89: \section{INTRODUCTION}

90: \label{sec:1}

91: The {\HI} gas directly associated with galaxies that intercept the

92: line of sight to background quasars appears as optically thick Lyman

93: Limit systems in the quasar spectrum. The prominent metal lines 

94: associated with these intervening absorbers are typically observed to be

95: kinematically broad ($\Delta v \sim 100 - 400$~{\kms}), strong, and often saturated

96: \citep[e.g.,][]{ss92,cwcvogt01} . Studying the

97: properties of a large population of such strong {\MgII} absorbers is a

98: technique used for constraining the evolution of metals in the

99: interstellar media, gaseous halos and coronae of galaxies over a large

100: history of the universe \citep{lanzetta87,cwc96}. Apparently distinct from these strong {\MgII} absorbers are 

101: the population of quasar absorption line systems

102: in which the low ionization metal lines are weak.  These systems are

103: separated from the strong ones based on the standard definition of the

104: rest-frame equivalent width of $\MgII~\lambda 2796$~{\AA} line being

105: $W_r(2796) < 0.3$~{\AA}. This is not a firm criterion for division,

106: but has been followed as a convention on the following basis. The

107: survey of \citet{ss92}, which identified a large

108: population of strong {\MgII} absorbers used a sample of intermediate

109: resolution spectra ($\Delta \lambda \sim 5$~{\AA}) which had an

110: equivalent width threshold of $\sim 0.3$~{\AA}. Later surveys of

111: higher sensitivity and spectral resolution found that the equivalent

112: width distribution of {\MgII} systems at $z \sim 1$ increases steeply

113: for $W_r(2796) < 0.3$~{\AA} such that $\sim 67\%$ of all {\MgII}

114: absorbers (down to $0.02$~{\AA}) from that epoch are in fact weak

115: \citep{weak1, anand07}. It later became clear that such an empirical 

116: basis for the classification of {\MgII}

117: systems into {\it strong} and {\it weak} does bear some physical

118: significance in that the two classes might be tracing two or more different

119: populations of objects \citep{weak1, weak2, charlton03}.

120: 

121: The class of weak {\MgII} quasar absorption systems have several

122: remarkable properties that are unique. To begin with, unlike the

123: strong systems, the weak {\MgII} systems are optically thin in neutral

124: hydrogen and produce metal lines that are narrow [$b$(Mg) $\sim

125: 4$~{\kms}] and often unsaturated \citep{weak1}. If weak

126: {\MgII} absorbers are sub-Lyman limit systems with $10^{15.8} < N(\HI)

127: < 10^{16.8}$~{\cmsq}, they would account for a significant fraction

128: ($> 25\%$) of the high column density regime of the {\Lya} forest

129: \citep{weak2}. Surveys of quasar fields to identify host

130: galaxies have not often found weak {\MgII} systems at close impact

131: parameters (physical distance, D $< 30$~Kpc) of luminous star forming galaxies ($L >

132: 0.05$~$L_*$) \citep{cwc98, cwc05, milni06}. This is a surprising result particularly in

133: light of the fact that, in a substantial number of weak systems, the

134: metallicity of the low ionization gas where the {\MgII} absorption

135: arises is constrained to values greater than 0.1Z$_\odot$.  In some

136: cases the best constraints require metallicities that are between

137: Z$_\odot$ and 10Z$_\odot$ \citep{weak2, charlton03, misawa07}. 

138: Thus, even though they have {\HI} column

139: densities that are $\sim 4$ orders of magnitude smaller than DLAs,

140: weak {\MgII} absorbers are produced in gas clouds with metallicities

141: that are 0.5 - 2 dex higher than the average metallicity of DLA

142: absorbers.

143: 

144: The astrophysical systems associated with weak {\MgII} absorbers have

145: not been identified yet. Several possibilities exist, which partly

146: account for the observed statistical and physical properties of weak

147: {\MgII} systems. Examples include extragalactic high velocity clouds

148: \citep{anand07}, dwarf galaxies \citep{lynch06}, gas clouds expelled 

149: in  super winds from dwarfs \citep[e.g.][]{zonak04, stocke04, keeney06} 

150: and/or massive starburst galaxies and metal enriched gas in intergalactic star clusters

151: \citep{weak2}. 

152: Recently, a number of authors have reported the detection

153: of several high metallicity ($> 0.1$~Z$_\odot$) gas clouds that are

154: residing in the intergalactic medium, both in the local

155: universe \citep{aracil06, tripp06} and at high redshift [$z > 2$, \citet{simcoe04, 

156: schaye07}]. The observed column densities of {\MgII}, {\CII}, {\SiII} and {\FeII}

157: in these gas clouds, and the metallicities inferred for them are comparable to several weak {\MgII} absorbers studied so far \citep{weak2,lynch07,misawa07}. 

158: 

159: Photoionization models of specific weak {\MgII} systems have shown

160: that they possess a two phase structure. The low ionization gas,

161: traced by such species as {\MgII}, {\FeII}, {\CII}, {\SiII}, etc. has

162: a gas density of n$_{\H} > 0.1$~{\cc} that is roughly 2 - 3 orders of

163: magnitude higher than the density of the associated high ionization

164: gas traced by {\CIV} lines \citep{charlton03, lynch07, misawa07}. 

165: The column density measured for

166: weak {\MgII} lines is typically $N(\MgII) \sim 10^{12} - 10^{13}$

167: {\cmsq}. For such relatively small values, the high number density of

168: ions derived from the photoionization modeling constraints the

169: thickness of the low ionization gas to $\sim 10$~pc. The weak {\MgII}

170: population occupies a significant volume of the universe with a

171: cross-section similar to the absorption cross section of luminous

172: ($L_*$) galaxies \citep{weak1, anand07}. Thus, given their small thickness, 

173: if their gas clouds had a

174: spherical geometry then they would be a million times more numerous

175: than luminous galaxies at $z \sim 1$ in order to reproduce the

176: observed cross-section on the sky. However, an analysis comparing the

177: relative incidence of high and low ionization gas in a sample of weak

178: {\MgII} and {\CIV} systems at low redshift ($z < 0.3$) favors a

179: filamentary or sheet-like configuration for the absorber's physical

180: geometry \citep{milni06}, instead of millions of individual, {\it spherical} low 

181: ionization {\MgII} clouds of $\sim 10$~pc size, embedded in a $\sim $~kpc 

182: higher ionization halo, traced by {\CIV}.

183: 

184: In addition, \citet{lynch06} and \citet{anand07} discovered an evolution

185: in the redshift number density ($dN/dz$) of weak {\MgII}

186: absorbers over the interval $0.4 < z < 2.4$. The $dN/dz$ was found

187: to peak at $z = 1.2$, and subsequently decline (from a no-evolution trend,) towards 

188: higher redshift. The equivalent width distribution 

189: function was also found to be different between low and high redshift.

190: At $z \sim 1$, the equivalent width distribution of weak {\MgII} is

191: significantly higher than an extrapolation of the exponential

192: distribution for strong {\MgII} absorbers. In contrast, at $z \sim 2$, 

193: the equivalent width distribution of weak {\MgII} clouds

194: is only slightly in excess of an extrapolation of the strong {\MgII}

195: distribution. In the context of these observed changes in the 

196: absorber statistics between $z \sim 1$ and $z \sim 2$, it becomes

197: important to investigate if the changes are reflective of

198: an underlying change in the absorbers' physical or chemical

199: properties. Such an investigation may yield valuable clues

200: into the physical nature of the kind of astrophysical processes/structures

201: that produce weak {\MgII} absorption at low and high redshifts. The large

202: sample size considered here, provides the scope for such

203: an analysis.

204: 

205: In this paper we present constraints on the chemical abundances,

206: metallicity and

207: ionization conditions for a sample of 100 weak {\MgII} absorbers

208: identified in spectra that were extracted from the

209: VLT/UVES archive. We compare the observed line properties between the

210: various chemical elements in order to constrain the metallicity,

211: density and line-of-sight thickness of the low ionization

212: gas. Previous studies have focused on individual weak {\MgII} systems,

213: whereas here we have attempted to derive the range of properties for a

214: large ensemble of weak {\MgII} systems, which have only recently been

215: discovered (Narayanan {\etal} 2007). In \S~\ref{sec:2} and

216: \S~\ref{sec:3}, we explain the measurement of line parameters and the

217: comparison between the prominent metal lines in these systems. In

218: \S~\ref{sec:4}, we present the Cloudy photoionization constraints on

219: the densities and metallicities of these absorbers. An observed trend

220: in the {\FeII} to {\MgII} ratio with redshift is presented in \S~\ref{sec:5}, and a comparison

221: of weak {\MgII} clouds with the high velocity subsystems in a sample of

222: strong {\MgII} systems in \S~\ref{sec:6}. We conclude the paper with a summary

223: of the significant results (in \S~\ref{sec:7}) and a detailed discussion (\S~\ref{sec:8}) on the 

224: nature of the gaseous structures selected by weak {\MgII} absorption.

225: 

226: \section{THE SAMPLE OF WEAK {\MgII} ABSORBERS}

227: \label{sec:2}

228: 

229: The sample of weak {\MgII} systems presented in this study were

230: identified in $51$ quasar spectra extracted from the ESO archive. The

231: quasars were observed using the UVES high resolution echelle

232: spectrograph \citep{dekker00} on VLT at $R \sim 45,000$ (FWHM~$= 6.6$~{\kms}). The

233: detailed information on each quasar observation, such as the exposure time, emission

234: redshift of the quasar, wavelength coverage, program ID and PI of the observation

235: are listed in Table 1 of \cite{anand07}. The

236: reduction and wavelength calibration of the spectra were carried out

237: using the ESO-provided MIDAS pipeline. When multiple exposures of the

238: same target were available, they were co-added to enhance the $S/N$

239: ratio of the spectrum, after applying vacuum-heliocentric velocity

240: corrections. The final combined spectrum was continuum fitted using

241: IRAF\footnote{IRAF is distributed by the National Optical Astronomy Observatories (NOAO), which are operated by AURA, Inc., under cooperative agreement with NSF}, and subsequently normalized. The detailed reduction procedure can

242: be found in \S~2.1 of \citet{anand07}. The redshift path

243: length searched in each quasar spectrum for weak {\MgIIdblt} lines

244: excluded regions blueward of the {\Lya} emission to avoid

245: contamination from the Ly-$\alpha$ forest.

246: 

247: The 100 {\MgII} systems that we analyze here are taken from 

248: \citet{anand07} which described a survey for weak {\MgII} systems at

249: $0.4 < z < 2.4$. In addition to magnesium, we use the information from

250: ions of four other elements, viz. iron, carbon, silicon and aluminum,

251: in each system to estimate the chemical and ionization

252: conditions in the gas. Specifically, the lines that we consider

253: are the following; $\MgI$~$\lambda$~2853~{\AA}; {\MgIIdblt}~{\AA}; {\FeIIdblt}~{\AA};

254: {\AlIIIdblt}~{\AA}; $\AlII$~ $\lambda$~1671~{\AA}; $\CII$~$\lambda$~1335~{\AA}; 

255: and $\SiII$~$\lambda$~1260~{\AA}\footnote{The wavelengths are vacuum wavelengths rounded to the nearest natural number}. The coverage

256: of the individual lines vary depending on the redshift of the system

257: and the wavelength coverage of the spectrum in which the line is identified.

258: The system plots are presented in Figure~\ref{fig:1} ({\it Note: The full set of system plots will  be available in the online version of the journal. Here we provide only a few as examples}). 

259: 

260: Most weak {\MgII} absorbers also have associated high ionization gas in a 

261: separate phase, traced by {\CIV} lines \citep[e.g.][]{weak1}. 

262: The relative incidence of high and low ionization

263: phase is a useful constraint on the physical geometry of the absorber population

264: \citep{milni06}. The UVES spectra offer simultaneous coverage of {\CIV} and {\MgII} over the redshift interval $1 < z < 2.4$. Within this interval, in almost all cases {\CIV} is detected. However, the {\CIV} absorption profile is in many cases offset in velocity with {\MgII}, indicating the presence of a separate phase for the high ionization gas. The {\CIVdblt} profiles are shown in the various system plots of Figure~\ref{fig:1} ({\it Note: The full set of system plots will  be available in the online version of the journal. Here we provide only a few as examples}). In this paper, our focus is on determining the ionization conditions and metallicity in the low ionization  gas, and hence we defer the detailed analysis of the high ionization {\CIV} phase and its association with the low ionization gas to a forthcoming paper.

265: 

266: \subsection{Measurement of Equivalent Widths}

267: \label{sec:2.1}

268: 

269: For each system within the redshift interval of $0.4 < z < 2.4$,

270: besides {\MgIIdblt}~{\AA}, only {\MgI}~$\lambda 2853$~{\AA} and

271: {\FeIIdblt} lines have wavelength long enough to be in the regions of

272: the spectrum typically uncontaminated by {\HI} lines in the

273: forest. The other prominent metal lines that we have measured have

274: rest-frame wavelengths, $\lambda < 2000$~{\AA}. As a consequence they

275: are susceptible to blending with {\Lya} forest lines, particularly

276: since the redshift of the intervening absorber is often much less than

277: the emission redshift of the quasar. In instances where line blending

278: is apparent, we quote an upper limit on the measurement of the

279: rest-frame equivalent width. For doublet/multiplet lines such as {\FeII} and

280: {\AlIII} we have measured the equivalent width of the stronger member

281: of the doublet. We also quote a $3\sigma$ upper limit when a line is

282: not detected at the $3\sigma$ level. Table 1 lists the rest-frame

283: equivalent width measured for the various lines in each system.

284: 

285: \subsection{Measurement of Column Densities}

286: \label{sec:2.2}

287: 

288: Absorption lines were fit with a Voigt function to estimate the column

289: density. An initial model for the line profile was derived using the

290: automated profile fitter AUTOVP \citep{dave97}. The AUTOVP

291: routine generated its model profile by performing a Voigt profile

292: decomposition of the absorption feature and subsequently minimizing

293: the $\chi^2$ by adjusting the velocity ($v$), column density ($N$) and

294: Doppler parameter ($b$) for all the components in the model. The

295: output of AUTOVP was then refined using a maximum likelihood least

296: square fitter, MINFIT, which returns a best-fit model with a minimum

297: number of Voigt profile components based on an F-test \citep{cwc03}. 

298: To retain a component, requires an improvement in the model fit at an

299: 80\% significance.  MINFIT derives the

300: model absorption profile after convolving with a Gaussian kernel of

301: FWHM = 6.6~{\kms}, corresponding to the UVES spectral resolution 

302: of $R=45,000$. The column density and Doppler parameter with 

303: their $1 \sigma$ errors are obtained for this final model.

304: 

305: Voigt profile fits were applied to the following lines associated with each system : {\MgIIdblt};

306: {\MgI}~$\lambda 2853$~{\AA}; {\FeIIdblt}; {\AlIIIdblt};

307: {\AlII}~$\lambda 1671$~{\AA}; {\CII}~$\lambda 1335$~{\AA}; and {\SiII}

308: ~$\lambda 1260 $~{\AA}. In

309: the situation where a line is not detected at the $3\sigma$ level, we

310: quote an upper limit on the column density determined from the

311: $3\sigma$ limit on the equivalent width. Our sample consists of

312: only relatively high-$S/N$ spectra. The $3\sigma$ limits are hence low, so that we can

313: assume linear part of the curve-of-growth for estimating the

314: corresponding upper limit in column density. To get robust constraints

315: on the fit parameters, we use both members of the doublet while

316: fitting profiles for {\MgII} and {\AlIII}, and both of the

317: strong members of the multiplet in the case of {\FeII}, viz. {\FeIIdblt}. Weaker 

318: members of the {\FeII} multiplet were rarely detected at the 3-$\sigma$ level. 

319: By simultaneous fitting of members of the doublet/multiplet, it 

320: is possible to recover the true column density, even if the

321: stronger member of the doublet/multiplet is saturated \citep[see Sec 4.4.2][]{cwcthesis}. 

322: Thus for example, in the case of {\MgII}, by using both members of the doublet, it is possible to 

323: recover the true column density for values of $N(\MgII)$ up to 

324: $10^{14}$~{\cmsq} \citep[see Figure 4.3][]{cwcthesis}. For lines that are not doublets,

325: Voigt profile fits are unique only when the lines are unsaturated. In our

326: sample, this would a problem only for the strongest of {\CII}~$\lambda 1335$~{\AA}

327: and {\SiII}~$\lambda 1260 $~{\AA} lines. 

328: 

329: Table 2 lists the line parameters ($v$, $N$ and $b$) thus measured for the various lines in

330: each system. As mentioned earlier, the lines with rest-wavelength

331: $\lambda < 2000$~{\AA} are often found within the region of the

332: spectrum that is contaminated by the {\Lya} forest.  For 

333: {\AlIII}, we found that blending with {\HI} lines of the forest could

334: be identified by comparing the profile shapes of the individual

335: members of the doublets. For the rest of the transitions, their

336: profiles were compared to {\MgII} to rule out possible contamination.

337: Figure~\ref{fig:1} shows the line profiles of the various low ionization 

338: transitions and {\AlIII} associated with each system in our 

339: sample. For each line, the positions of the individual clouds, 

340: determined from Voigt profile fitting, are labeled. 

341: 

342: \section{RESULTS FROM MEASUREMENT OF METAL LINES}

343: \label{sec:3}

344: 

345: \subsection{The Population of Single and Multiple Clouds}

346: \label{sec:3.1_01}

347: 

348: From comparing the frequency distribution of the number of clouds per

349: system between strong and weak absorbers, \citet{weak2} discovered

350: that unlike strong absorbers, weak {\MgII} systems have a

351: non-Poissonian frequency distribution. Approximately two-thirds of the

352: weak systems in their sample of 30 at $z \sim 1$ had absorption in a

353: single cloud, isolated in redshift. The clouds were narrow ($ b \sim

354: 4$~{\kms}) indicating a small temperature and velocity dispersion in

355: the gas. These systems were consequently called {\it single cloud}

356: weak {\MgII} absorbers referring to the low ionization gas in a single

357: narrow component, unresolved at $R=45,000$ (FWHM$=

358: 6.6$~{\kms}). The other set of weak absorbers were called {\it

359: multiple cloud} weak {\MgII} systems as they had the low ionization

360: absorption in multiple clouds that are resolved at $R=45,000$

361: and kinematically broad ($\Delta v > 30$~{\kms}) compared to 

362: single clouds. 

363: 

364: The incidence of the number of low ionization clouds in any given weak

365: {\MgII} system is important for considering the physical geometry of

366: the absorbing structure \citep{ellison04, milni06}.

367: Figure~\ref{fig:2} shows the distribution of the number of Voigt profile

368: components per system in our sample. In nine systems

369: \footnote{$z=0.599512$ in Q2217-2818, $z=1.091866$ in Q0042-2930,

370: $z=1.153704$ in Q1151+068, $z=1.330502$ in Q1157+014, $z=1.395635$ in

371: Q0011+0055, $z=1.405367$ in Q2347-4342, $z=1.491972$ in Q0551-3637,

372: $z=1.755704$ in Q2243-6031 and $z=1.796233$ in Q2347-4342}, we found the

373: {\MgII} line profile to have a slight asymmetry, sometimes yielding

374: two components in the Voigt profile model. We have classified these as

375: single cloud systems, and plotted them in the $N_c = 1$ bin, since the

376: low ionization gas is predominantly still in a single gas cloud with

377: an internal velocity dispersion is less than 6.6~{\kms}.  A similar

378: asymmetry in single cloud line profiles was also noticed by \citet{weak1}

379: in HIRES/Keck high resolution spectra. However, their

380: formal fitting procedure, with the lower $S/N$ data, did not

381: statistically favor a two component fit.  The occasional asymmetry in

382: the line profile is likely due to a contribution to the low ionization

383: absorption from a slightly offset higher ionization gas

384: cloud. Photoionization models have succeeded in reproducing the

385: observed asymmetry in the line profile using a single low ionization

386: phase and separate high ionization phases (e.g., see the ionization

387: models for $z=1.405367$ and $z=1.796237$ systems in \citet{lynch07}

388: and the $z=1.755704$ system in \citet{misawa07}. We have therefore

389: chosen to classify the above nine systems as single cloud systems.

390: 

391: Taking this into account, in our larger sample of weak systems we find

392: that the single cloud absorbers account for 48\% of the total

393: population, which is much smaller than their observed fraction in the

394: HIRES sample described in \citet{weak1} and \citet{weak2}, 

395: but is consistent within 1$\sigma$. Within the redshift

396: interval of $0.4 \leq z \leq 1.4$, identical to the redshift path

397: length covered by the \citet{weak1} sample, we find that

398: only 45\% (34/76) of the weak absorbers are single cloud systems,

399: indicating that our results are not affected by any evolutionary

400: effect in which a larger fraction of $z > 1.4$ systems have multiple

401: clouds. This is further confirmed in Figure~\ref{fig:3} where we illustrate the

402: distribution of single and multiple cloud absorbers as a function of

403: redshift. We find weak absorbers showing absorption in both single and

404: multiple clouds at all redshifts within $0.4 < z < 2.4$. A preference is

405: not evident for a certain type of weak absorber (i.e. single or multiple

406: cloud) towards either low or high redshift.  

407: 

408:  

409: \subsection{Equivalent Width of {\MgII}}

410: \label{sec:3.1_02}

411: 

412: Figure~\ref{fig:4} shows the distribution of rest-frame equivalent width of

413: {\MgII}$\lambda 2796$ as a function of system redshift. The

414: strength of the low ionization phase as traced by {\MgII} demonstrates

415: considerable scatter within the interval $0.4 < z < 2.4$. A

416: Spearmann-Kendall test supports the null hypothesis that the

417: equivalent width is statistically uncorrelated (Spearman's $\rho =

418: 0.04$) with the redshift of the absorber. In strong {\MgII} absorbers, 

419: the low ionization absorption is never confined to a single cloud. The

420: line profiles are often kinematically complex with the absorption

421: spread in several clouds separated in velocity. From the statistical

422: analysis of 23 strong {\MgII} systems along 18 quasar lines of sight,

423: \citet{cwc03} found an average of $\sim 8$ clouds per

424: system, with the absorption profile of one system resolved into as

425: many as 19 different components. In addition, in that sample of strong

426: {\MgII} systems, a very strong correlation was found between the

427: number of clouds and the rest-frame equivalent width $W_r(2796)$. In

428: the bottom panel of Figure~\ref{fig:2}, we illustrate that such a strong

429: correlation ($>9\sigma$) also exists for weak {\MgII} systems, where

430: most of the weaker systems ($W_r(2796) < 0.1$~{\AA}) have absorption only in

431: one or two clouds. Both \citet{weak1} and \citet{anand07} found that the equivalent 

432: width distribution of weak systems at

433: $z \sim 1$ rises rapidly towards smaller equivalent widths. This

434: observation is also reflected in the bottom panel of Figure~\ref{fig:2} where we

435: find 67\% of our sample of weak absorbers to be at $W_r(2796) <

436: 0.15$~{\AA}.

437: 

438: \subsection{Comparison of Rest-Frame Equivalent Width}

439: \label{sec:3.2}

440: 

441: In Figure~\ref{fig:5}, we present the rest-frame equivalent width of the various

442: metal lines compared to the equivalent width of {\MgII}~$\lambda 2796$

443: line for both single and multiple clouds. The difference in the number

444: of data points in each plot is attributed to the spectral coverage for

445: the various transitions. A Spearman-Kendall nonparametric

446: correlation test shows that the rest-frame equivalent width of {\MgI},

447: {\FeII}, {\CII}, {\AlII}, and {\AlIII} are all correlated with

448: the rest equivalent width of {\MgII} at a greater than 98\% confidence level. Owing to fewer data

449: points, {\SiII} exhibits a correlation with {\MgII} of lesser

450: significance ($ > 3 \sigma$). The Spearman and Kendall 

451: correlation tests were carried out using the ASURV astrostatistics

452: package which takes into consideration measurements that are upper

453: limits \citep{edf85, lavalley92}. 

454: 

455: We note that the {\CII} equivalent width has a strong linear

456: relationship with {\MgII}, with little scatter.  The {\SiII} may have

457: a similar relationship, but it is hard to demonstrate with the smaller

458: number of data points.  All other transitions, though we have shown

459: their equivalent widths to be correlated with {\MgII}, show a much

460: larger scatter in this relationship. 

461: 

462: For {\FeII}, there is more than a factor of ten spread in the ratio

463: $W_r(2383)/W_r(2796)$ at $0.2 < W_r(2796) < 0.3$~{\AA}.  At small

464: $W_r(2796)$, many of the $W_r(2383)$ measurements are upper limits,

465: but a spread of more than a factor of two can still be demonstrated at

466: $W_r(2796) \sim 0.05$~{\AA}.  A similarly large spread is also found

467: for the relationships between {\MgI} and {\MgII}, {\AlII} and {\MgII}, and

468: {\AlIII} and {\MgII}.

469: 

470: The large scatter in the observed ratios between the various

471: transitions can be brought about by a number of factors, and can be

472: exploited to diagnose the physical conditions of the absorber.  For a

473: given strength of {\MgII}, the spread in the strength of the other

474: transitions can be due to variations in abundance patterns or to

475: differences in the density/ionization parameters of the gas clouds (addressed

476: in \S~\ref{sec:4.2},~\ref{sec:4.6} and ~\ref{sec:5}).

477: There is also the possibility that absorption from two different ions

478: arises in separate phases, in which case the scatter between their

479: equivalent widths could be quite large.  If we are to distinguish

480: between these different factors, it is important not to use the observed

481: equivalent widths since they average together the contributions from

482: different clouds.  The physical conditions are better probed through

483: comparison of cloud-by-cloud column densities.

484: 

485: 

486: \subsection{Comparison of Column Densities}

487: \label{sec:3.3}

488: 

489: In Figure~\ref{fig:6}, we compare the {\MgII} column density measured for

490: each cloud with the corresponding column densities in {\MgI}, {\FeII},

491: {\SiII}, {\CII}, {\AlII} and {\AlIII}. In multiple cloud systems, the

492: comparison is between each component of {\MgII} and the corresponding

493: component in the other transition. Non-detections at the $3\sigma$

494: level are given as upper limits. To test for likely dependence between

495: the measured quantities, we apply Spearman and Kendall's

496: non-parametric correlation tests. We find that the column densities of

497: all ionization species except for {\SiII} are correlated with the

498: column density of {\MgII} at greater than $7\sigma$

499: significance. The statistical measure of the correlation is smaller for {\SiII}

500: ($3\sigma$ significance) because of fewer data, 40\% of which are

501: censored points. The strongest correlation is observed between

502: $N(\CII)$ and $N(\MgII)$ (rank correlation coefficient, $\rho =

503: 0.762$), in spite of being limited by fewer data points. Such a strong

504: positive correlation in the equivalent width and column density of

505: {\CII} with {\MgII} justifies the use of {\CII} lines, in conjunction with

506: other low ionization lines such as {\SiII}, to select weak

507: absorbers at redshifts $z > 2.5$ where it becomes more difficult to

508: use {\MgIIdblt} lines because of the redshifting of these lines

509: into the near-infrared regime.

510: 

511: Among the various ions, the column densities of {\FeII}, {\CII} and {\SiII}

512: display the least scatter with the column density of {\MgII}. The

513: correlation between these ions and {\MgII} can be formalized as:

514: 

515: \begin{center}

516: log $N(\FeII) = (1.15 \pm 0.14)$ log $N(\MgII) - 2.36$, ($\sigma = 0.37$) 

517: 

518: log $N(\CII) = (0.82 \pm 0.10)$ log $N(\MgII) + 3.42$, ($\sigma = 0.19$)

519: 

520: log $N(\SiII) = (1.02 \pm 0.25)$ log $N(\MgII) - 0.26$, ($\sigma = 0.28$)

521: \end{center}

522: 

523: \noindent The best-fit slope, the y-intercept and the corresponding $1 \sigma$

524: uncertainties of these regression lines were calculated using the

525: survival analysis package ASURV Rev 1.2 \citep{edf85, lavalley92}

526: which implements the methods

527: presented in \citet{isobe86}. The $\sigma$ values

528: are the standard deviation of the respective fits.

529: 

530: In Figure~\ref{fig:6}, we also plot the Solar composition of Fe, C, Si and Al

531: with respect to Mg, for reference. The abundance of carbon

532: [log (C/Mg)$_\odot$ = 0.970] is taken from \citet{allende01, allende02}, of

533: silicon [log (Si/Mg)$_\odot$ = -0.030], iron [log (Fe/Mg)$_\odot$ = -0.069] and

534: magnesium from \citet{holweger01}, and for aluminum [log (Al/Mg)$_\odot$ = -1.110] from \citet{grevesse98}. The observed ratio of column densities between the various

535: ions and {\MgII} when compared with the respective solar abundance ratios, can indicate whether the ionization fractions of {\CII}, {\SiII}, {\AlII}, {\AlIII} and

536: {\MgI} are comparable with that of {\MgII} in the low 

537: ionization gas. We note that this is the case for the observed {\SiII} to {\MgII} and

538: {\AlII} to {\MgII} column density ratios, which closely follow the respective

539: solar abundance ratios. 

540: On the other hand, the observed {\CII} to {\MgII} ratios 

541: are above the solar abundance ratio, while the {\FeII} to {\MgII} ratios are below. 

542: A number of factors such as differences in ionization parameter, differences in the elemental abundances and/or contributions from different gas phases can combine to produce these observed trends, which are discussed in the next section. 

543: 

544: In addition, differential depletion of elements onto dust can lead to deviations from Solar composition. The presence of dust has not been directly measured in weak {\MgII} systems. However, it has been found that dust extinction is significant only in stronger {\MgII} absorbers \citep[$W_r(\MgII) > 1.5$~{\AA};][]{khare05, york06}. For low column density absorbers such as the weak {\MgII} systems, interstellar dust may not be a substantial component influencing metallicity estimates derived from gas phase abundances. In addition, CLOUDY models incorporating varying amounts of dust levels find dust having a negligible effect on the density of the absorbing gas as well \citep{weak2}.

545: 

546: 

547: \section{CHEMICAL AND IONIZATION PROPERTIES OF WEAK {\MgII} ABSORBERS}

548: \label{sec:4}

549: 

550: The physical conditions in the low ionization gas clouds are

551: constrained using the standard photoionization code Cloudy

552: \citep[ver.07.02.01,][]{ferland98}. The primary objective is to

553: derive limits on the metallicity, density, and line-of-sight thickness

554: for the gas phase where the bulk of the {\MgII} absorption arises. For

555: this purpose, the observed column densities of the other prominent low

556: and intermediate ionization transitions - namely {\MgI}, {\FeII},

557: {\SiII}, {\AlII}, {\CII}, and {\AlIII}, and their ratios to {\MgII} - 

558: are used.

559: 

560: The ionization fraction for a given element is controlled by the

561: density in the gas cloud as well as by the strength of the incident

562: ionizing radiation. Weak {\MgII} systems are not known to reside at

563: small impact parameters (d~$< 30$~kpc) from luminous star-forming

564: galaxies ($L > 0.05L_*$, \citet{weak1}). Hence the ionization balance in them is

565: likely dictated by the intensity of the extragalactic background

566: radiation (EBR). We choose the \citet{hm96} model for

567: the EBR which incorporates ionizing photons from quasars and

568: star-forming galaxies after propagation through a thick IGM.  A 10\%

569: escape fraction from galaxies is used for ionizing photons with

570: $\lambda \leq  912$~{\AA}.

571: 

572: To determine the overall properties for our sample of weak {\MgII}

573: absorbers, we generate a grid of Cloudy models for a range of

574: ionization parameters ($-8.0 <$ log $U < -1.0$) and neutral hydrogen

575: column densities ($14.0 <$ log $N(\HI) < 19.0$). The weak {\MgII}

576: systems in our sample span the redshift range $0.4 < z <

577: 2.4$. Therefore we consider two separate Cloudy grids modeled using

578: the integrated ionizing photon density ($h\nu \geq 1$~Ryd) at $z = 1$

579: and $z = 2$.  The difference of $\sim 0.5$~dex in the intensity

580: of the extragalactic background radiation field between these two redshifts

581: does not critically affect the output of the Cloudy models. Nonetheless, to have

582: a more tenable comparison between the data and the models, we plot 

583: the $z < 1.5$ and $z \geq 1.5$ systems on the $z=1$ and the $z=2$ grids 

584: respectively. We adopt a solar abundance pattern for the Cloudy models, but 

585: discuss the effects of abundance variations. In the following sections, we discuss the

586: constraints that the various ions provide towards the chemical and

587: ionization conditions in the absorbing gas.

588: 

589: \subsection {Constraints from {\MgII}}

590: \label{sec:4.1}

591: 

592: In our sample, the {\MgII} column densities of the individual clouds

593: in weak {\MgII} absorbers fall within the range $10^{11.0} < N(\MgII)

594: < 10^{13.3}$~{\cmsq}. Our search for weak {\MgII} systems along the 81

595: quasar lines of sight is 86\% complete down to the equivalent width

596: threshold of $W_r (2796) = 0.02$~{\AA}, corresponding to $N(\MgII)

597: \sim 10^{11.8}$~{\cmsq} \citep{anand07}. The $N(\MgII)$ is

598: useful to place limits on the metallicity of the low ionization gas

599: phase. Figure~\ref{fig:7} presents how the column densities of the various ions

600: change with respect to the ionization parameter, log $U$, for

601: different values of $N(\HI)$ and metallicity. For a given metallicity

602: and ionization parameter, i.e., a certain density, an increase in

603: $N(\HI)$ would correspond to an increase in the size of the

604: absorber. Also, with increasing ionization parameter, the neutral

605: fraction of hydrogen declines such that to converge on the same value

606: of $N(\HI)$, the size of the absorber has to further increase. It

607: is evident from Figure~\ref{fig:7} that for $N(\HI) = 10^{15}$~{\cmsq}, at

608: sub-solar metallicity (e.g. $0.1$Z$_\odot$), the model column density of {\MgII} is

609: inadequate to explain the observed $N(\MgII)$ even for the weakest

610: {\MgII} lines in our sample.  For a given log $U$, the column

611: densities of the ionization stages of various elements scale almost

612: linearly with both $N(\HI)$ and metallicity. Thus, higher $N(\MgII)$ can be

613: recovered by raising either $N(\HI)$ or metallicity. For example, at $N(\HI) =

614: 10^{15}$~{\cmsq}, by raising the metallicity by 1 dex (to Z$_\odot$),

615: we find that the ionization models reproduce the observed column

616: densities in the weaker {\MgII} systems ($N(\MgII) < 10^{12}$~{\cmsq})

617: in our sample. For the same $N(\HI)$ value, at 10Z$_\odot$, a

618: substantial fraction of the range of observed $N(\MgII)$ is covered by

619: the ionization models, except for those systems with $N(\MgII) >

620: 10^{13}$~{\cmsq}. Alternatively, with a 1 dex increase in $N(\HI)$,

621: the curves shift correspondingly such that systems with $N(\MgII) <

622: 10^{12}$~{\cmsq} can be produced in 0.1Z$_\odot$ gas. However, for

623: $N(\HI) > 10^{17}$~{\cmsq}, the low ionization gas cloud is an

624: optically thick, Lyman-limit absorber (i.e. able to produce a break in

625: the spectrum of the background quasar at $\lambda = 912$~{\AA} in the

626: rest-frame of the absorber).

627: 

628: It can be concluded that the column density of {\MgII} is a suitable

629: parameter for estimating limits on the metallicity of the

630: absorber. Our sample of weak {\MgII} systems spans a range of 2 dex in

631: {\MgII} column density. Assuming solar abundance pattern, the

632: metallicity in many of their low ionization gas clouds is constrained

633: to be at least 0.1 Z$_\odot$ if the gas is optically thin in neutral

634: hydrogen ($N(\HI) < 10^{17}$~{\cmsq}, see \S~\ref{sec:8}).  

635: Moreover, the strongest {\MgII} lines ($N(\MgII) > 10^{13}$~{\cmsq}) 

636: among the weak systems require supersolar metallicity.

637: 

638: \subsection {Constraints from {\FeII}}

639: \label{sec:4.2}

640: 

641: In our sample of weak absorbers, 32\% (66/205) of {\MgII} clouds have

642: {\FeII} detected at the $> 3\sigma$ level, out of which 81\% are firm

643: detections (i.e. detections unaffected by blending with other

644: absorption features).  The column density ratio, N(\FeII)/N(\MgII),

645: falls between 0.02 and 4.0. The range of values for the ratio remains

646: unchanged even when we exclude upper limit measurements. Among the

647: clouds with {\FeII} detected, 13 are single cloud systems and the

648: remaining 53 are part of multiple cloud systems.

649: 

650: Constraints for metallicity, similar to the ones derived using

651: observed $N(\MgII)$, can also be derived based on $N(\FeII)$.  

652: Figure~\ref{fig:7} illustrates how the column density of {\FeII} changes with

653: ionization parameter for different values of $N(\HI)$ and metallicity.

654: At $N(\HI) \leq 10^{16}$~{\cmsq} and Z$ = 0.1$Z$_{\odot}$, $N(\FeII) <

655: 10^{11.2}$~{\cmsq} which is inadequate to explain the observed column

656: density in systems with {\FeII} detected. By raising $N(\HI)$ by one

657: dex, we find a corresponding increase in the column density of {\FeII}

658: in the models, such that a column density of $N(\FeII) <

659: 10^{12.2}$~{\cmsq} is possible at sub-solar metallicity. This still

660: does not account for some fraction ($\sim 10$\%) of the observed {\FeII} lines. With

661: metallicity increased to solar and super-solar values the models begin

662: to produce enough {\FeII} to explain the full range of observed values. In

663: general, we can infer that for the systems in which {\FeII} is

664: detected in our sample, the metallicity is constrained to values of Z

665: $\geq$ Z$_\odot$ if $N(\HI) < 10^{17}$~{\cmsq}.

666: 

667: It is evident from the column density comparison in Figure~\ref{fig:6} that, for

668: a given $N(\MgII)$, the observed $N(\FeII)$ has a spread of $\sim

669: 1$~dex between the various systems. This spread is also evident in

670: Figure~\ref{fig:5} which compares the rest-frame equivalent widths.  For gas

671: that is optically thin, the ratio of column density between various

672: ionization stages does not depend on metallicity since all individual

673: column densities scale linearly. An exception to this can occur

674: (discussed in \S~\ref{sec:4.4}) for certain ions at supersolar

675: metallicities where cooling leads to much lower gas temperatures. In

676: the optically thin regime, for a given abundance pattern, the ratio of

677: column densities between different elements varies primarily with

678: ionization parameter. The relative strength of {\FeII} compared to

679: {\MgII} in a system is particularly sensitive to ionization parameter

680: for log $U > -4.0$. Thus we over-plot, in Figure~\ref{fig:8}, the observed column density

681: ratios of {\FeII} to {\MgII} on a Cloudy grid of photoionization

682: models. The Cloudy grid is for a range of log

683: $U$ and $N(\HI)$ at sub-solar, solar and super-solar metallicities.

684: The censored data points that occupy the left of Figure~\ref{fig:8} 

685: are systems in which {\MgII} is very weak. The {\FeII}, being even weaker, is not

686: detected at the $3 \sigma$ significance threshold. The $S/N$ of the

687: best of our sample of quasar spectra are comparable and therefore the

688: envelope of the ratio $N(\FeII)/N(\MgII)$ for censored data points is

689: seen as increasing with decreasing $N(\MgII)$.

690: 

691: We can place constraints of log $U$ assuming that the {\FeII} and

692: {\MgII} arise in the same phase.  To begin with, we notice that the

693: column density ratios of all clouds in our sample confine the

694: ionization parameter to log $U < -2.0$, corresponding to $n_H >

695: 0.002$~{\cc} (for log $n_\gamma = -4.70$ at $z = 2$).  

696: At 0.1Z$_{\odot}$, the systems with {\FeII} detected require {\HI}

697: column densities greater than $10^{16}$~{\cmsq}. By increasing the

698: metallicity, the grids shift to the right proportionately such that

699: the same {\FeII} to {\MgII} ratios can now be recovered from weaker

700: {\HI} lines ($N(\HI) \sim 10^{14} - 10^{15}$~{\cmsq}). Thus, if the

701: low ionization gas is thin in neutral hydrogen, the metallicity in

702: systems where {\FeII} is detected is constrained to solar or

703: super-solar values. 

704: 

705: In the sample of 17 single cloud weak absorbers studied by Rigby

706: {\etal} (2002), a subset of systems with log $[N(\FeII)/N(\MgII)] > -0.3$

707: were classified as {\it iron-rich}. These systems were found to

708: have high metallicity ($ > 0.1 Z_{\odot}$), particularly strong

709: constraints on density (log $U < -4.0$, $n_{H} > 0.09$~{\cc}), and small

710: sizes [$N(\HI) + N(\HII) < 10^{18}$~{\cmsq}, R $ < 10 $~pc]. The

711: relatively high {\FeII} to {\MgII} ratio indicated that the high

712: density, low ionization gas in the {\it iron-rich} systems is not

713: $\alpha$-enhanced. Following the definition of \citet{weak2}, we find that 30

714: clouds in our sample are {\it iron-rich}, excluding censored data

715: points.  Comparing the data to the Cloudy grid, we find that the

716: ionization parameter in these systems is constrained to an upper limit

717: on the ionizing parameter between -3.2 and -3.7 depending on the

718: difference in ionizing photon number density between $z = 2$ and $z

719: =1$ respectively. A limit of log $U < -3.7$ translates to a density of

720: $n_H > 0.05$~{\cc} in the low ionization gas for log $n_\gamma =

721: -5.04$~{\cc} (the number density of ionizing photons with $h\nu \geq

722: 13.6$~eV at $z = 1$).  The density constraint for the {\FeII} rich

723: systems translates into a small upper limit for the thickness ($R <

724: 10$~pc) of the absorber.  In systems where {\FeII} is weak

725: compared to {\MgII}, the constraint on density is much lower

726: (log $U < -2.0$).

727: 

728: In this analysis, we have assumed a Solar abundance pattern.  Changing

729: the abundance of any element from this pattern would result in a

730: corresponding change in all ionization stages of that

731: element. Thus, the {\it iron-rich} systems can have a lower constraint on

732: density if the abundance of iron in the low ionization cloud is

733: enhanced relative to the solar abundance pattern, since the Cloudy

734: grids would be shifted upwards. Such an abundance pattern is not

735: physically well motivated.  On the other hand, an $\alpha-$enhanced

736: abundance pattern is ruled out for these {\it iron-rich} systems as its

737: effect would be to shift the Cloudy grids further down such that the

738: ionization models will not be able to reproduce the observed {\FeII}

739: to {\MgII} ratio. The $\alpha-$enhancement is, however, conceivable 

740: for the clouds in which {\FeII} is low compared to {\MgII}. The ionization

741: model, in that case, would infer higher densities for the low ionization

742: gas.

743: 

744: Finally, \citet{weak2} found that the $N(\FeII)/N(\MgII)$ in

745: their HIRES sample had a bimodal distribution with an apparent gap of

746: $\sim 0.5$~dex at $-0.8 <$ log $[N(\FeII)/N(\MgII)] < -0.3$. It was

747: therefore used as basis for defining the {\it iron-rich} systems, and

748: to suggest that there may be a separate class where {\FeII} is weak

749: relative to {\MgII}. Figure~\ref{fig:9}, shows the histogram distribution of

750: the {\FeII} to {\MgII} ratio for our sample of weak {\MgII} single and

751: multiple clouds. The bin size is equivalent to the gap in the

752: distribution that \citet{weak2} found for their sample. The

753: distribution from our sample does not suggest a bimodality, either for

754: single or multiple clouds. This remains true for smaller bin sizes as well.

755: Hence the apparent gap that was suggested

756: by the \citet{weak2} data can be attributed to inadequate

757: sample size. Additionally, we also note that there is no difference in the

758: observed {\FeII} to {\MgII} ratio between single and multiple cloud systems.

759: The individual clouds in the multiple cloud systems have the similar

760: log $U$ constraints as single clouds, with {\it iron-rich} systems detected

761: in both category. 

762: 

763: \subsection{Constraints from {\MgI}}

764: \label{sec:4.3}

765: 

766: In this section, we explain the constraints that are available from

767: the observed {\MgI} to {\MgII} ratio in weak systems.  In the past,

768: single phase photoionization models (using CLOUDY 90; \citet{ferland98})

769: have failed to reproduce the observed {\MgI} to {\MgII} ratio in some

770: strong {\MgII} systems \citep{rauch02, cwc03, ding03}. 

771: The {\MgI}/{\MgII} ratio derived from the models was

772: lower than the observed neutral to singly-ionized ratio.  To

773: circumvent this, a separate phase was proposed in which the {\MgI}

774: ionization fraction is higher \citep{cwc03, ding03}. 

775: This separate phase would have a higher density ($n_H >

776: 1$~{\cc}) and lower temperature ($T < 600$~K) than the gas phase

777: associated with the {\MgII} absorption. The {\MgI} lines corresponding

778: to such low temperatures are very narrow ($b \sim 2$~{\kms}) and are

779: therefore unresolved at the $R=45,000$ of the earlier HIRES and UVES

780: observations. However, through superhigh spectral resolution

781: observations, at $R = 120,000$ ($\Delta v = 2.5$~{\kms}), it has been

782: demonstrated that the {\MgI} lines are not narrower than what is

783: derived for $R=45,000$ (Narayanan {\etal} 2007). 

784: 

785: Compared to {\FeII}, only a few weak {\MgII} systems in our sample

786: have {\MgI} detected at the $3\sigma$ level. Out of the 200 weak

787: {\MgII} clouds for which there is coverage, {\MgI} is detected in only

788: 7 single cloud systems and in 20 clouds in multiple cloud

789: systems. Both single and multiple clouds span roughly the same range

790: of values for the {\MgI} to {\MgII} ratio, between -2.2 and -0.5,

791: considering only firm detections. The neutral magnesium fraction (Mg$^0$/Mg$_{total}$) 

792: is thus small, compared to the {\MgII} fraction ({\MgII}/Mg$_{total}$), in these systems. 

793: This most likely explains the large

794: scatter in the range of limits, evident in Figures~\ref{fig:5} and ~\ref{fig:6} , 

795: since we are

796: sampling a large number of quasar spectra with differences in

797: sensitivity. The spectra with the highest

798: $S/N$ ratio in our sample, however, constrain the {\MgI} column

799: density to values as low as $10^{9.5}$~{\cmsq}, $\sim 2$~dex smaller

800: than $N(\MgII)$, indicating that the neutral fraction in the {\MgII}

801: phase is indeed not very high.

802: 

803: Figure~\ref{fig:10} shows the observations compared to the grid of Cloudy

804: (version 07.02.01) ionization models. To begin with, the ionization

805: models are able to recover the observed {\MgI} to {\MgII} ratio from a

806: single phase. Compared to the CLOUDY (version 90) grid of ionization

807: models presented in Churchill {\etal} (2003), the models displayed in

808: Figure~\ref{fig:10}, have the {\MgI}/{\MgII} fraction higher by $\sim 0.5$~dex for a

809: given log $U$. The difference in the ionization fraction of magnesium

810: is a result of improvements in the rate coefficients for charge

811: transfer reactions, incorporated into the more modern versions of

812: Cloudy \citep{kingdon96}. The relevance of the charge transfer reaction  

813: (H + Mg$^+$ $\rightarrow$ H$^+$ + Mg) in controlling the ionization

814: fractions of  {\MgI} and {\MgII} has also been noted by \citet{tappe04}. For our sample of

815: weak {\MgII} systems, the ionization models derived from the revised

816: version of the photionization code suggests that a single phase

817: solution is possible. In fact, the observed ratio of {\MgI} to {\MgII}

818: in strong {\MgII} absorbers can also now be explained without invoking

819: a separate cold phase for {\MgI}.

820: 

821: We find that, in a large majority of the systems for which information on

822: {\MgI} is available, the ionization parameter is confined to log $U < -2.5$, for

823: solar and super-solar metallicity. At Z $<$ Z$_\odot$, the constraint

824: on ionization parameter is higher by $\sim 1$~dex. The fraction

825: {\MgI}/{\MgII} is expected to decrease with an increase in the ionization

826: conditions in the gas. This is evident in Figure~\ref{fig:7}. Therefore, the

827: systems with higher {\MgI} to {\MgII} ratios will have lower

828: constraints on ionization parameter (log $U < -3.0$) and

829: correspondingly higher constraints on density ($n_H > 0.02$~{\cc}),

830: identical to iron-rich systems. Moreover, if the {\HI} lines are

831: weaker ($N(\HI) < 10^{16}$~{\cmsq}), solar or supersolar metallicities

832: will be necessary to reproduce the observed {\MgI} and {\MgII} column

833: densities (see Figure~\ref{fig:10}) .

834: 

835: 

836: \subsection{Constraints from {\CII}}

837: \label{sec:4.4}

838: 

839: In our sample, all the 15 weak {\MgII} systems for which there is

840: coverage of {\CII}~$\lambda 1335$~{\AA} show prominent {\CII}

841: lines. Among these, eight are single cloud and the remaining are

842: multiple cloud systems. The ratio of $N({\CII})$ to $N({\MgII})$ is

843: always significantly greater than 1, with a range of values between 4

844: and 60.   A larger oscillator strength for the {\MgII}$\lambda$ 2796

845: line, and a longer wavelength compared to {\CII}$\lambda$ 1335, leads to

846: them having comparable equivalent widths.

847: 

848: The Cloudy grid of single phase photoionization models, comparing the

849: column density of {\CII} to {\MgII}, is shown in Figure~\ref{fig:11}. We

850: find that for most of the systems, the ionization parameter would be

851: constrained to log $U > -2.5$, implying a gas phase density of $n_H <

852: 0.006$~{\cc}.  Such large values of log $U$ would be inconsistent

853: with that inferred for the low ionization phase for the clouds in

854: which {\FeII} is detected.  However, 

855: in Figure~\ref{fig:15}, we compare the observed {\CII}

856: to {\MgII} ratio against {\FeII} to {\MgII} in systems with

857: simultaneous coverage of both lines.  Only for two systems ($z =

858: 1.585464$, and $z = 1.988656$), do we have firm (i.e. measurements that

859: are not limits) detections for both {\CII} and {\FeII}. In these two

860: systems we find $N({\FeII})$ to be $\sim 1$~dex smaller than

861: $N(\MgII)$, and therefore a constraint of log $U

862: > -2.5$ (see the $z \geq 1.5$ iron grid).  This is consistent with the log

863: $U$ derived using the {\CII} to {\MgII} ratio in the corresponding

864: systems allowing for a single low ionization

865: phase solution.  The density of this low ionization phase would

866: be smaller than what is estimated for the {\it iron-rich}

867: systems.  Unfortunately, such a definite statement cannot be extended

868: for all the other systems plotted in Figure~\ref{fig:15}, as their {\FeII}

869: measurements are upper limits. Nonetheless, even these limits are consistent

870: with a low density.

871: 

872: Many of the detected clouds in Figure~\ref{fig:8} did not appear in

873: Figure~\ref{fig:15} because {\CII} could not be measured for them due

874: to contamination or lack of coverage.  For these clouds the inferred

875: ionization parameters range from log $U \sim -2.2$ to $\sim -3.5$.  If

876: the {\CII} and {\MgII} arise in same phase, we would expect the

877: {\CII} in the clouds in which {\FeII} is detected to have

878: $N({\CII})/N({\MgII}) > 10$, such that consistent log $U$ values

879: would be derived from {\CII}. Such a direct comparison between the

880: measured {\CII} and {\FeII} may not apply because {\CII} may arise

881: partly from a higher ionization phase.  In addition to the low

882: ionization phase, most weak {\MgII} systems also have an associated

883: high ionization phase where the density is low ($n_{\H} \leq

884: 10^{-3}$~{\cc}).  Although dominated by higher ionization states of

885: carbon ({\CIII} and {\CIV}), the {\CII} ionization fraction

886: (i.e. {\CII}/C$_{total}$) can be non-negligible in this phase. For

887: example, at log $U = -1.5$, $N({\HI}) = 10^{15}$~{\cmsq}, and 

888: Z = 0.3Z$_\odot$ (typical values derived from photoionization models,

889: e.g. see Table 5 of \citet{misawa07}), $N(\CII) =

890: 10^{13.3}$~{\cmsq}, which is comparable to the detected {\CII} in many

891: of our clouds.  The $N(\MgII)$ contribution from this high ionization

892: phase is negligible.  In summary, even though {\CII} is detected in

893: all weak {\MgII} systems, since it does not arise exclusively in the low

894: ionization phase, it may not provide as robust a constraint on the

895: ionization parameter as does {\FeII}.

896: 

897: We note, also, that the grid with a metallicity of 10Z$_\odot$ does

898: not cover many of the {\CII} data points.  At this metallicity,

899: temperatures fall to $T < 1000$~K because of metal cooling, even

900: at log $U > -1.5$.  The gas, including both magnesium and carbon,

901: is less heavily ionized at the low temperatures, but the

902: effect is stronger for {\MgII} so that the density of {\MgII} 

903: is larger relative to {\CII}.  Clouds that have supersolar metallicity constraints,

904: based on other transitions, would then need to have a large

905: contribution to the {\CII} from a separate phase.

906: 

907: \subsection {Constraints from {\SiII}}

908: \label{sec:4.5}

909: 

910: In our sample, we find that the {\SiII} column density

911: is comparable to the column density of {\MgII}, as shown in Figure~\ref{fig:6}.

912: The ratio of column densities has values between 0.2 and 3.2, with majority of them

913: at $\sim 1$.  Most of the clouds have {\SiII} to {\MgII} ratios that

914: fall on the grid of Cloudy models in Figure~\ref{fig:12}.  However, the grids do not provide

915: much leverage in determining log $U$ since {\SiII} and {\MgII} are

916: similar over most of the parameter space, particularly for solar and

917: higher metallicities.

918: 

919: The single phase ionization models suggest that systems in which

920: $N(\SiII) < N(\MgII)$ require log $U > -1.5$ and high metallicity,

921: assuming a Solar abundance pattern. This is evident in the Cloudy

922: grids where data points corresponding to low {\SiII} to {\MgII} column

923: density ratio are below the model expectations for Z $<$

924: Z$_\odot$. Only for supersolar metallicities, do any of the models

925: reproduce the low {\SiII} to {\MgII} ratio. This is also evident in

926: Figure~\ref{fig:7} where the {\SiII} and {\MgII} ionization curves cross only at

927: super-solar metallicity for log $U > -1.5$, corresponding to $n_H <

928: 10^{-3}$~{\cc}. The low {\SiII} to {\MgII} ratio can also be obtained

929: from single phase models by lowering the abundance of silicon compared

930: to other $\alpha-$process elements, in which case the metallicity could

931: be lower, and the density higher.

932: 

933: In general, the {\SiII} does not provide a robust constraint on the

934: ionization parameter for the low ionization gas.  In the small number

935: of clouds for which both {\SiII} and {\CII} are covered they usually

936: provide consistent constraints on log $U$, taking into account that

937: some fraction of the {\CII} can arise in a separate phase.

938: 

939:   

940: \subsection {Constraints from {\AlII}}

941: \label{sec:4.6}

942:   

943: Figure~\ref{fig:13} shows the Cloudy grid of photoionization models, with

944: measurements of {\AlII} to {\MgII} overplotted.  Within the sample of

945: systems, we find that $N({\AlII)}$ is always smaller than

946: $N({\MgII})$, sometimes by as much as 0.5 to 1.5 dex.  For a solar or

947: smaller metallicity, many systems (those with $N({\AlII})\geq 0.1

948: N({\MgII})$) are covered by the photoionization grids.  Particularly

949: at solar metallicity, the ratio of {\AlII} to {\MgII} is not very

950: sensitive to ionization parameter, so it cannot be used to effectively

951: measure this quantity.  Of special note are a number of systems, both

952: at high and low redshift, with $N({\AlII})\leq 0.1 N({\MgII})$.  These

953: points are not covered by the grid for all metallicities, based on a

954: solar abundance pattern.  The 10Z$_\odot$ grid does extend (for $\log

955: U > -1.5$) to somewhat lower values of {\AlII} to {\MgII}.  A

956: supersolar metallicity could help to explain some systems with a low

957: {\AlII} to {\MgII} ratio, such as the $z=1.68079$ system towards Q~$0429-4091$

958: described in Misawa {\etal} (2007).  However, for many of these systems, the most likely

959: explanation would be a reduction of the aluminum abundance relative to

960: magnesium by up to $\sim$0.7 dex.  A reduction would effectively

961: shift the grids down by the same amount.  Such an abundance pattern is

962: feasible, since it is consistent with $\alpha$-enhancement.  We note

963: that the same shift should apply for these systems in the {\AlIII}

964: grids as well.

965: 

966: \subsection {Constraints from {\AlIII}}

967: \label{sec:4.7}

968:  

969: Among the 74 systems with simultaneous coverage of {\MgIIdblt}~{\AA}

970: and {\AlIIIdblt}~{\AA} lines, {\AlIII} is detected in 12 single cloud

971: systems and in 44 clouds in 20 multiple cloud systems. The ratio of

972: {\AlIII} to {\MgII} column density falls within the range 0.04 to

973: 0.98, excluding upper limits. The Cloudy grids of single phase models

974: are shown in Figure~\ref{fig:14}.  Assuming that the {\AlIII} and {\MgII} are

975: produced in the same phase, for the {\AlIII} detections, the ionization

976: parameter ranges from $-3.5 <$ log $U < -1.0$.

977: 

978: In the previous section, in order to reconcile the {\AlII} to {\MgII}

979: grid with the points below that grid, we proposed a reduction of the

980: aluminum abundance relative to magnesium.  If this is applied to

981: the {\AlIII}, it shifts this grid downwards so that it does not

982: cover some of the data points.  If the systems with small {\AlII}

983: to {\MgII} also have small {\AlIII} to {\MgII} this is not a problem,

984: though it does require large log $U$ for even these systems.

985: This is found to be the case in Figure~\ref{fig:15}, where we have plotted

986: the ratio of {\AlIII} to {\MgII} versus {\AlII} to {\MgII}.  For all

987: of the systems below the {\AlII} grid (see Figure~\ref{fig:13}),

988: $N$({\AlIII})~$< 0.3 N({\MgII})$.  

989: 

990: Although we have not identified specific clouds for which {\AlIII} and

991: {\MgII} cannot arise in the same phase, we note that some of our {\AlIII}

992: detections imply large ionization parameters, log $U > -2.0$.  This is

993: even more the case if we rely on a decrease of the aluminum abundance.

994: So either there is a sub-population of weak {\MgII} clouds which are

995: of a higher ionization state, or some of the {\AlIII} is produced in a

996: higher ionization phase, such as the one giving rise to the bulk of the

997: {\CIV} absorption.

998: 

999: \subsection {{\AlIII} to {\AlII} ratio}

1000: \label{sec:4.8}

1001: 

1002: In damped Ly-$\alpha$ (DLA) systems, the chemical abundance

1003: estimations are often carried out under the assumption that the ionization

1004: corrections are not significant, since the gas is expected

1005: to be predominantly in the low ionization phase. However, the

1006: detection of {\AlIII} lines at the same velocity as the low ionization

1007: lines in several DLAs lead \citet{vladilo01} to investigate the

1008: relevance of ionization corrections for these systems.  In their

1009: analysis, \citet{vladilo01} observed that the

1010: $N(${\AlIII}$)/N(\AlII)$ ratio in DLA systems exhibits an

1011: anti-correlation with $N(\HI)$. This relationship was described as

1012: intrinsic to DLAs, and was used to suggest that the {\AlIII} to

1013: {\AlII} ratio in these systems could be a sensitive probe of the

1014: ionization conditions in the gas. Using a sample of sub-DLA systems,

1015: \citet{dz03} examined if this anti-correlation

1016: extends to lower $N(\HI)$. They found the {\AlIII} to {\AlII}

1017: ratio in sub-DLAs to be in the same range as for DLA systems. In other

1018: words, the anti-correlation trend did not seem to extend to sub-DLAs

1019: ($10^{19} < N(\HI) < 10^{20.3}$~{\cmsq}). However, more recently

1020: \citet{meiring07} found that the anti-correlation could apply

1021: even to sub-DLAs, based on a different sample of systems.

1022: 

1023: In our sample, 28 weak {\MgII} clouds have measurements of both

1024: {\AlIII} and {\AlII}, of which 15 are firm detections in both

1025: (i.e. measurements that are not limits). In Figure~\ref{fig:16}, we plot their

1026: ratio with respect to the corresponding $N(\MgII)$. We also plot, in

1027: an adjacent panel, the {\AlIII} to {\AlII} ratio in DLA and sub-DLA

1028: systems as a function of $N(\HI)$, based on information extracted from

1029: the literature \citep{vladilo01, dz03, meiring07}. 

1030: We find the {\AlIII} to {\AlII} ratio in weak

1031: {\MgII} systems to be considerably higher than in DLAs and

1032: sub-DLAs. On average, the ratio is $\sim 0.5 - 1$~dex higher than what

1033: has been measured for the other two classes of systems. This indicates

1034: that the ionization conditions are higher in weak {\MgII}

1035: systems than in DLA or sub-DLA systems. 

1036: 

1037: Based on photoionization modeling (see \S~\ref{sec:4.1}) we have

1038: concluded that a large fraction of weak {\MgII} clouds have a

1039: metallicity of solar or higher if the clouds are optically thin in

1040: neutral hydrogen.  The observed redshift number density of weak

1041: {\MgII} absorbers is too large for all of them to be Lyman limit

1042: systems (as explained in

1043: \S~\ref{sec:8}).  If $N(\HI) < 10^{17}$~{\cmsq} for the weak systems

1044: plotted in Figure~\ref{fig:16}, then we conclude that

1045: the anti-correlation trend discovered by Vladilo {\etal} (2001) for DLA

1046: systems, and supported by \citet{meiring07} for sub-DLA systems,

1047: continues to lower $N(\HI)$ values.

1048: 

1049: 

1050: \section {Evolution of the low ionization phase structure}

1051: \label{sec:5}

1052: 

1053: One of our objectives in carrying out the chemical and ionization

1054: analysis on a large sample of weak {\MgII} systems is to find out if

1055: there are any evolutionary trends observable in the absorber

1056: population.  Our VLT/UVES sample of weak {\MgII} systems span the

1057: redshift interval $0.4 < z < 2.4$. Photoionization constraints have

1058: already suggested that a range of ionization properties and

1059: metallicities can be expected for the low ionization phase. It would be

1060: unusual to assume that the entire population of weak {\MgII} systems

1061: are tracing some unique type of physical process/structure, given these

1062: variations and the large redshift interval surveyed.

1063: 

1064: To investigate, we compared the observed {\FeII} to {\MgII} ratio

1065: between the various systems, as it is a reliable constraint on

1066: density and chemical enrichment history. Figure~\ref{fig:17} shows the measured

1067: $N(\FeII)/N(\MgII)$ as a function of redshift.

1068: Because of the many non-restrictive limits at small $N({\MgII})$,

1069: particularly for low redshift clouds, it is hard to evaluate whether

1070: there is a significant relationship between $N(\FeII)/N(\MgII)$

1071: and $z$.  In order to consider this issue, we separated clouds

1072: with log $N({\MgII}) < 12.2$, those that were

1073: likely to have only limits on {\FeII} (based on inspection of

1074: Figure~\ref{fig:8}), and considered only the stronger

1075: of the weak {\MgII} clouds, plotted in {\it red} color in the top panel of

1076: Figure~\ref{fig:17}.  It appears

1077: that there is an anti-correlation between $N(\FeII)/N(\MgII)$ and

1078: $z$.  At high redshift, there is an absence of detections with

1079: larger $N(\FeII)/N(\MgII)$ values, while at low redshift there

1080: are many detections with large $N(\FeII)/N(\MgII)$ and few limits

1081: that could even be consistent with small $N(\FeII)/N(\MgII)$

1082: values.  We applied a Kolmogorov-Smirnov (K-S) test to compare the

1083: distributions of $N(\FeII)/N(\MgII)$ at $z \geq 1.5$ and $z < 1.5$

1084: for the clouds with log $N({\MgII}) > 12.2$.

1085: In the cases where only upper limits are available, we conservatively

1086: include these as values when performing the K-S test.  

1087: The distributions are shown in histogram form in the

1088: lower panel of Figure~\ref{fig:17}.  We find that

1089: there is a probability of only P(KS) = 0.006 (KS statistic D = 0.505) that

1090: the two samples are drawn from the same distribution.  The probability

1091: is likely to decrease if upper limits could be replaced with actual

1092: detections.  Thus we find that the

1093: observed anti-correlation of $N(\FeII)/N(\MgII)$ with $z$ is

1094: statistically significant for log $N({\MgII}) > 12.2$ clouds.

1095: 

1096: We have found an absence of log $N({\MgII}) > 12.2$ clouds at high

1097: redshift with large values of $N(\FeII)/N(\MgII)$ and an apparent absence of

1098: log $N({\MgII}) > 12.2$ clouds at low redshift with small values of

1099: $N(\FeII)/N(\MgII)$.  There are a number of low redshift clouds with

1100: limits that could be consistent with small values.  Furthermore,

1101: for weaker clouds (with log $N({\MgII}) < 12.2$), there are some examples of

1102: low $N(\FeII)/N(\MgII)$ values at low redshifts.  We conclude

1103: that large $N(\FeII)/N(\MgII)$ clouds are present only at $z<1.5$ and

1104: not at higher redshifts.  The other population with small

1105: $N(\FeII)/N(\MgII)$ exists both at low and high redshifts.

1106: 

1107: As demonstrated earlier, systems in which log $N(\FeII)/N(\MgII) >

1108: -0.3$ are constrained to have a high density (log $U < -3.7$, $n_H >

1109: 0.05$~{\cc}).  Those with a lower {\FeII} to {\MgII} ratio have lower

1110: densities, ranging down to log $U < -2.0$, which corresponds to $n_H <

1111: 0.001$~{\cc}.  The observed trend in the {\FeII} to {\MgII} ratio with

1112: redshift, therefore, could imply the absence of high density clouds in

1113: the low ionization phase in weak absorbers at high-$z$.  Such

1114: variations in the phase structure are plausible if the weak systems are probing a different

1115: combination of astrophysical systems/processes at $z \sim 2$ and $z

1116: \sim 1$. Furthermore, if the absorbers are optically thin {\HI}

1117: clouds, then we are also seeing a change in the thicknesses

1118: of the low ionization gas clouds, from kiloparsec-scale at $z \sim 2$

1119: to a range of values including both parsec-scale and kiloparsec-scale

1120: clouds at $z \sim 1$.

1121: 

1122: Alternatively, gas clouds that are enriched primarily by Type II SNe

1123: events will have [$\alpha$/Fe] $> 0$, in which case the observed

1124: {\FeII} to {\MgII} column density ratios will be low.  Thus the

1125: observed trend could also indicate that the weak {\MgII} clouds are

1126: predominantly $\alpha$-enhanced at high redshift, with an increasing

1127: contribution to the population at lower redshift from clouds with a

1128: higher iron-abundance.  Increasing the [$\alpha$/Fe] in the Cloudy

1129: models, would then lead to low {\FeII} to {\MgII} clouds having high

1130: densities (log $U < -3.0$, $n_H > 0.01$~{\cc}), similar to the {\it

1131: iron-rich} clouds.  The relevance of abundance pattern variations is

1132: discussed in detail in \S~\ref{sec:8}, where we speculate on the

1133: physical origin of these absorbers at the two redshift epochs.

1134: 

1135: \section{Weak {\MgII} Absorbers \& Satellites of Strong {\MgII} Systems}

1136: \label{sec:6}

1137: 

1138: Strong {\MgII} systems are understood to be absorption arising in the

1139: disk and extended halos of normal galaxies \citep{bergeron91, steidel94, steidel02}.  

1140: Their broad ($\Delta$v

1141: $\sim 150$~{\kms}) and kinematically complex {\MgII} line profiles are

1142: found to be consistent with this picture \citep{charlton98}. A characteristic 

1143: feature in many $z \sim 1$ strong {\MgII} systems is

1144: weak, kinematic subsystems separated in velocity from the dominant

1145: absorption component \citep{cwcvogt01}. Such kinematic subsystems are

1146: likely to be gas clouds in the extended halo of the absorber in an

1147: arrangement analogous to the Galactic high velocity cloud (HVC) and 

1148: intermediate velocity cloud (IVC) populations.  In

1149: earlier work, we hypothesized that a non-negligible fraction of weak

1150: {\MgII} systems could be the extragalactic analogs of Milky Way HVCs,

1151: in which a random line of sight intercepts the surrounding halo

1152: cloud(s), but misses the optically-thick absorber. The possibility of

1153: such an event is favored strongly by some recent observations which

1154: find a patchy distribution (less than unity covering factor) for the

1155: gas in the extended halos of galaxies \citep{tripp05, cwc07}. 

1156: For a patchy halo, a sight line that passes only

1157: through the clouds in the halo is more likely to produce a weak

1158: {\MgII} system than a strong one \citep{cwc05}.  Our

1159: hypothesis was primarily based on the observed evolution in the

1160: redshift number density ($dN/dz$) of weak {\MgII} systems and the

1161: evolution in the gas kinematics of strong {\MgII} absorbers over the

1162: same redshift interval of $0.4 < z < 2.4$ \citep{anand07, mshar07}.

1163: 

1164: To extend this postulate further, and also to test its validity, we

1165: compared the {\FeII} to {\MgII} ratio for the low ionization gas in weak

1166: {\MgII} systems to that for the satellite clouds of strong {\MgII} systems

1167: presented in \citet{cwcvogt01}. Based on an observed break in

1168: the velocity distribution of Voigt Profile components in their sample

1169: of strong {\MgII} systems, \citet{cwcvogt01} specified clouds

1170: at $|\Delta v| > 40$~{\kms} as intermediate velocity or high velocity

1171: subsystems (i.e. satellite clouds). The satellite clouds in that

1172: sample were separated in velocity by as much as $|\Delta v| \sim

1173: 350$~{\kms} from the system center, with a median value of $|\Delta v|

1174: = 165$~{\kms}. In Figure~\ref{fig:18}, we plot the {\FeII} to {\MgII} column

1175: density ratio of these subsystems and compare it to the same in our

1176: sample of weak {\MgII} clouds. We have included only those weak

1177: absorbers that are within $0.4 \leq z \leq 1.2$, equivalent to the

1178: redshift interval of the \citet{cwcvogt01} sample.

1179: 

1180: The comparison shows that the weak {\MgII} clouds closely resemble the

1181: satellite clouds of strong {\MgII} systems. To begin with, the column

1182: density of {\MgII} in the satellite clouds spans roughly the same

1183: range of values as that of weak {\MgII} systems. The {\FeII} to

1184: {\MgII} in the satellite clouds have a scatter which is also

1185: comparable to the scatter in weak absorbers at the same redshift. A

1186: Kolmogorov-Smirnov test estimates that the two

1187: samples are consistent with being drawn from the same distribution [P(KS) =

1188: 0.633, D=0.196]. Comparable to the subset of {\it iron-rich} weak absorbers are

1189: several satellite clouds with $N(\FeII) \sim N(\MgII)$, which

1190: consequently constrains their density to $n_{\H} > 0.05$~{\cc}. In

1191: addition, a significant subset of the satellite clouds also have much

1192: lower densities ($n_{\H} < 0.001$~{\cc}), which make them analogous

1193: to the $\alpha$-enhanced weak {\MgII} clouds.

1194: 

1195: \section {Summary}

1196: \label{sec:7}

1197: 

1198: Using a large sample of recently discovered weak {\MgII} systems

1199: \citep{anand07}, we have derived constraints on the chemical

1200: and ionization conditions in their low ionization gas. In addition to

1201: {\MgII}, we have measured the equivalent widths and column densities of

1202: a number of other prominent metal lines associated with these

1203: absorbers. The significant results reported in this paper can be

1204: summarized as follows:

1205: 

1206: \noindent 1. In our sample of 100 weak {\MgII} systems, we find that

1207: only 48\% are single cloud absorbers. This fraction is smaller than

1208: the past results of \citet{weak2} where the majority (67\%) of

1209: weak {\MgII} absorbers were found to be single cloud systems, but is

1210: consistent within errors. The VLT/UVES sample that we consider in this

1211: paper is a factor of $\sim 5$ larger than the Keck/HIRES sample used

1212: by \citet{weak2}. We find no evidence for an evolution in the

1213: ratio of single to multiple cloud absorbers over $0.4 < z < 2.4$.

1214: 

1215: \noindent 2. We find the equivalent widths and column densities of {\CII}

1216: and {\SiII} are well correlated with the equivalent widths of {\MgII}, with minimal

1217: scatter in the respective relationships. The column densities of

1218: {\CII} and {\SiII} yield the following relationships with

1219: {\MgII}; log $N(\CII) = (0.82 \pm 0.10)$ log $N(\MgII) + 3.42$, and

1220: log $N(\SiII) = (1.02 \pm 0.25)$ log $N(\MgII) - 0.26$. The presence

1221: of a significant correlation in the equivalent widths, extends the possibility of using {\CII}

1222: and {\SiII} as proxy doublets for detecting analogs of weak {\MgII}

1223: systems at $z > 2.5$ in the optical spectra of quasars.

1224: 

1225: \noindent 3. If a large fraction of weak {\MgII} clouds are sub-Lyman limit 

1226: systems (i.e. optically thin in {\HI} with $N(\HI) <

1227: 10^{17}$~{\cmsq}), then the observed column density of {\MgII}

1228: constrains the metallicity in the low ionization gas to Z $\geq 0.1$

1229: Z$_\odot$. We also find the neutral fraction of magnesium to be very

1230: low in almost all weak {\MgII} clouds, approximately $\sim 2$~dex

1231: smaller than the corresponding $N(\MgII)$.

1232: 

1233: \noindent 4. From assuming a solar abundance pattern, we find

1234: that the clouds for which $N(\FeII) \sim N(\MgII)$ have their

1235: ionization parameters constrained to log $U < -3.7$, corresponding to

1236: $n_H > 0.05$~{\cc}. If the low ionization gas is optically thin in

1237: neutral hydrogen, then this places an upper limit of $R < 10$~pc on

1238: the thickness of these gas clouds.  Similarly, clouds with

1239: $N(\FeII) << N(\MgII)$ are constrained to have higher ionization

1240: parameters (log $U \sim -2$ in some cases) and lower densities.

1241: If the weak {\MgII} clouds, in which {\FeII} is observed to be weak 

1242: relative to {\MgII}, are $\alpha$-enhanced, then that would yield 

1243: higher constraints on density similar to the $N(\FeII) \sim N(\MgII)$ 

1244: absorbers.

1245: 

1246: \noindent 5. In the past, ionization models using CLOUDY (version 90) have often 

1247: not succeeded in recovering the observed {\MgI} to {\MgII} ratio in

1248: both strong and weak {\MgII} systems. The ionization fraction of

1249: {\MgI}, compared to {\MgII}, predicted by the models was not

1250: sufficiently large to explain the observed

1251: $N(\MgI)/N(\MgII)$. Therefore, a separate, cold (T $\sim 500$~K), high

1252: density (n$_{\H} > 1$~{\cc}) phase, centered at the same velocity as

1253: the {\MgII} phase was proposed in order to recover the observed {\MgI} in

1254: the models. However, in the current version of Cloudy (ver 07.02.01),

1255: with improvements in the rate coefficients of charge transfer

1256: reactions, the model {\MgI} to {\MgII} fraction is higher by $\sim

1257: 0.5$~dex for a given ionization parameter log $U$. Such an increase

1258: makes it consistent for {\MgI} to be in the same low ionization phase

1259: as {\MgII}, in both weak as well as strong {\MgII} systems.

1260: 

1261: \noindent 6. Most of our {\CII} and {\SiII} measurements are for systems

1262: at $z > 1.5$. In single phase models, the constraints from {\CII} and 

1263: {\SiII} are typically high for the ionization parameter (log $U > -2.5$),

1264: which is inconsistent with the constraints derived for clouds in which

1265: $N(\FeII) \sim N(\MgII)$. However, we also find an evolution in the

1266: relative strength of {\FeII}, compared to {\MgII}, such that towards

1267: higher redshift ($z > 1.5$) there might be a paucity of {\it iron-rich}

1268: systems (see Figure~\ref{fig:17}) . The absorbers in our sample, for 

1269: which there is simultaneous coverage

1270: of {\CII}, {\MgII} and {\FeII}, suggest that the $N(\FeII)$ could be sufficiently

1271: small compared to $N(\MgII)$ in the high redshift clouds. Moreover, a non-negligible

1272: fraction of {\CII} can arise in the high ionization gas, traced by {\CIII} and {\CIV},

1273: such that {\CII},  in itself, cannot be used to determine the physical

1274: conditions in the low ionization gas in weak absorbers. 

1275: 

1276: \noindent 7. We find that deviations from a solar abundance pattern

1277: is required to explain the observed column density of {\AlII} in many

1278: weak {\MgII} clouds. In particular, systems in which $N(\AlII) < 0.1

1279: N(\MgII)$ require the abundance of aluminum in the low ionization gas

1280: to be lowered by $\sim 0.7$~dex, consistent with $\alpha$-enhacement.

1281: Models with super-solar metallicity generally produce less {\AlII}

1282: relative to {\MgII}, but some reduction of the aluminum abundance is

1283: still required for many clouds.  When the abundance of aluminum is

1284: reduced, models underpredict {\AlIII} absorption unless the ionization

1285: parameter is high, which is sometimes inconsistent with that derived from

1286: other ions.  This suggests that {\AlIII}, like {\CII}, sometimes arises

1287: partly in a separate, higher ionization phase.

1288: 

1289: \noindent 8. In our sample, we find a relative absence of weak {\MgII}

1290: clouds with $N(\FeII) \sim N(\MgII)$ at high redshift ($z \geq 1.5$)

1291: compared to many detections of $N(\FeII) \sim N(\MgII)$ towards

1292: low-$z$. This observed trend can be interpreted in two ways : (1) an

1293: absence of high density, low ionization gas at high-$z$ and/or (2) the presence

1294: of  [$\alpha$/Fe] $ > 0$ in weak {\MgII} clouds at high-$z$. The other

1295: population of weak absorbers, in which $N(\FeII) << N(\MgII)$, are 

1296: detected at all intervals within $0.4 < z < 2.4$.

1297: 

1298: \noindent 9. We find similarities between the observed column density

1299: of {\MgII} as well as the {\FeII} to {\MgII} column density ratio in weak {\MgII} clouds

1300: and the high velocity subsystems (i.e. satellite clouds) of strong {\MgII} absorbers. 

1301: The range of $N(\MgII)$ and $N(\FeII)/N(\MgII)$ for the two groups are comparable. 

1302: This could be suggestive of the fact that some fraction of weak absorbers could be

1303: probing a similar type of physical structure as the satellites of strong {\MgII} systems.

1304: 

1305: \section {Discussion}

1306: \label{sec:8}

1307: 

1308: Weak {\MgII} absorbers have been identified over a large redshift

1309: interval $0 < z < 2.4$ \citep{weak1, anand05, anand07}, corresponding 

1310: to a great majority 

1311: of the history of the universe. Within this interval, their redshift

1312: number density ($dN/dz$) is found to be evolving, with a peak value of

1313: $dN/dz = 1.76$ at $<z> = 1.2$ \citep{anand07}. Towards lower

1314: redshift, the decrease in number density follows the expected curve

1315: for a non-evolving population (for a $\Lambda$CDM concordance model)

1316: \citep{anand05}. At $z > 1.2$, the $dN/dz$ has been found to

1317: decrease rapidly, such that an extrapolation to $z > 3$ would yield a

1318: value of zero. In other words, the observed redshift number density

1319: does not suggest that a significant population of weak {\MgII} systems

1320: exists at $z > 3$. In contrast, the number density of Lyman limit

1321: systems (LLSs) has been found to increase towards high redshift. At $z

1322: \sim 0.7, 1.5$ and 3, the $dN/dz$ of LLSs is estimated to be 0.7, 1.1

1323: and 1.9, respectively \citep{strengler95, sargent89}. These values are 

1324: in turn closely matched by the redshift number

1325: density of strong {\MgII} systems (with $W_r(2796) > 0.3$~{\AA}) at

1326: those same redshifts \citep[e.g.][]{nestor05}. Therefore, a

1327: substantial fraction of the observed weak {\MgII} clouds at $z\sim0.7$

1328: and $z\sim 1.5$ ought to be gaseous structures that are optically thin

1329: in {\HI} (i.e. sub-Lyman limit systems with $N(\HI) <

1330: 10^{17}$~{\cmsq}). This, consequently, would constrain the metallicity

1331: in the low ionization gas of many weak {\MgII} absorbers to Z $\geq$

1332: Z$_\odot$, in order to reproduce the observed column density of

1333: {\MgII}. Detailed photoionization models, where information on the

1334: {\HI} column density has been available, further support this

1335: inference \citep{charlton03, masiero05, misawa07}.

1336: 

1337: Bearing in mind these observed number statistics and constraints on

1338: the chemical and ionization conditions described in this work, we now

1339: proceed to discuss the plausible hosts of these low ionization, high metallicity 

1340: weak absorbers.  Given the range of ionization properties 

1341: and chemical abundances, it would be unusual to assume that the entire 

1342: population of weak {\MgII} systems would

1343: correspond to one unique type of astrophysical process/structure at 

1344: all redshifts.

1345: 

1346: \citet{schaye07} have recently suggested that weak {\MgII} clouds

1347: are likely to arise in gas ejected from starburst supernova-driven

1348: winds during an intermediate stage in free expansion, before settling

1349: in pressure equilibrium with the surrounding IGM.  The galaxy

1350: populations detected at high redshift ($z \sim 2 - 3$) are found to be

1351: rapidly star-forming, with a star-formation rate of $10 - 100$~M$_\odot$ yr$^{-1}$

1352: \citep[e.g.][]{pettini01, choi06}. The starburst events

1353: associated with these could give rise to galactic scale outflows that can

1354: displace large amounts of chemically enriched gas from the ambient ISM

1355: into the extended halo \citep[e.g.][]{heckman01, pettini01}. The strong 

1356: clustering of {\CIV} systems with Lyman break

1357: galaxies, which dominate the star formation rate at high-$z$, is possibly

1358: a signature of such outflows \citep{adelberger03, adelberger05a}. Such 

1359: supernova driven winds are observed to have a multiphase

1360: structure, with a non-negligible fraction of the interstellar gas in a

1361: warm neutral phase ($T < 10,000$~K) traced by such lines as {\CII},

1362: {\SiII}, {\FeII} and {\AlII} \citep{schwartz06} and a cold neutral 

1363: component ($T \sim 100$~K)

1364: detected in {\NaI} \citep{heckman00, rupke02}.  A

1365: sight line that directly intercepts the outflow close to the

1366: starburst region is likely to produce a very strong, saturated, and

1367: kinematically broad absorption feature \citep{bond01}. However,

1368: as described in \citet{schaye07}, as the wind material moves

1369: farther into the outskirts of the extended halo of the galaxy, the

1370: column densities would decrease in response to a decreasing density in

1371: the ambient medium.  At this stage, fragments in the wind, generated

1372: through hydrodynamical instabilities, would manifest as weak {\MgII}

1373: clouds, and later as weaker {\CIV} absorption associated with {\HI}

1374: lines in the Ly-$\alpha$ forest \citep{zonak04, schaye07}.

1375: 

1376: The interstellar clouds, ejected from correlated supernova events, are

1377: likely to be highly chemically enriched because of the close

1378: association with the feedback from star formation. \citet{simcoe06}

1379: discovered evidence for such chemically enriched gas (Z $>

1380: 0.1$Z$_\odot$) at $z \sim 2.3$, at distances of $\sim 100 - 200$~kpc

1381: from luminous star-forming galaxies, which they interpret as feedback

1382: from supernova winds or perhaps tidally stripped gas. The low

1383: ionization lines such as {\MgII}, {\FeII}, {\AlII}, {\CII}, and {\SiII} in the absorbers

1384: presented in that study have column densities similar to 

1385: those of weak {\MgII} clouds in our sample. Material that is directly related

1386: to star-forming events is likely to have [$\alpha$/Fe] $> 0$.  Weak

1387: {\MgII} clouds associated with such events would therefore have

1388: $N(\FeII) << N(\MgII)$.  This is consistent with the dominant

1389: population of high redshift ($z \sim 2$) weak {\MgII} clouds, and with

1390: some fraction of the clouds towards low ($z \sim 1$) redshift.  The

1391: high metallicity weak {\CIV} absorption clouds presented in \citet{schaye07} were estimated to have sizes that are small (R $\sim100$~pc),

1392: less than the Jean's length for self-gravitating clouds, implying that

1393: they are likely to be short lived.  Such a transient physical nature 

1394: is also a feature of weak {\MgII} clouds (Narayanan {\etal} 2005), and

1395: is anticipated for relics of winds.

1396: 

1397: The metal enriched interstellar gas expelled from the disk would

1398: resemble the high velocity gaseous structures surrounding the Milky Way, 

1399: as they move through the galaxy's halo. 

1400: \citet{ellison04}, from estimating the coherence scales of low and high ionization gas

1401: associated with weak absorbers, have suggested a scenario in which weak

1402: {\MgII} absorption could arise in the outskirts of ordinary galaxies,

1403: where the filling factor of the low ionization clouds (i.e. number of

1404: clouds per cubic parsec) is small compared to that in the center.  For

1405: low ionization gas, the coherence scale is $\sim 2$~kpc, i.e.,

1406: there is a high probability of seeing weak {\MgII} absorption along

1407: two lines of sight separated by this distance \citep{ellison04}.

1408: However, the separate low ionization phases must not fully cover this $\sim 2$~kpc

1409: region, since individual absorbing clouds are not seen along both

1410: lines of sight separated by tens to hundreds of parsecs \citep{rauch99}.  

1411: Also, photoionization models have shown that cloud

1412: line-of-sight thicknesses are often $< 100$~pc.  This suggests a

1413: clustering of separate clouds on a $\sim 2$~kpc scale, as well

1414: as implying a flattened geometry for the coherent structure, consistent

1415: with the findings of \citet{milni06}.  This scale could

1416: be consistent with dwarf galaxies \citep{ellison04} or with 

1417: tidal streams. Sight lines through gas stripped in tidal interactions 

1418: of galaxies can also produce sub-Lyman limit systems, and related weak {\MgII}

1419: absorbers. Gas that is tidally stripped in merger or accretion events

1420: could also form stars and provide a source of enriched gas clouds

1421: to the halos of high redshift galaxies.  These would be analogous to

1422: the Milky Way circumgalactic gaseous streams, related to accretion of

1423: interstellar gas from satellite galaxies.

1424: 

1425: In this context, we emphasize that the Milky Way analogs of weak {\MgII} absorbers are not likely to be the HVC complexes 

1426: detected in 21cm and/or H$\alpha$ emission, since those have $N(\HI) > 10^{18}$~{\cmsq} \citep{wakker97, putman03}. The weak absorbers must instead correspond to a population of halo clouds with  lower {\HI} column densities. Spectroscopic observations in the ultraviolet along various sight lines through the Milky Way halo have detected such high velocity gas in which the {\HI} column densities are sub-Lyman limit \citep[$N(\HI) \sim 10^{16.5}$~{\cmsq}; ][]{collins04, fox05, ganguly05}. These clouds, which exhibit multiple gaseous phases, have the column densities of {\CII}, {\SiII} and {\FeII} constrained to values similar to what we find for these ions in our weak {\MgII} sample. Using the HST/STIS archival spectra of quasars, \citet{richter08} have identified a population of high velocity clouds in the Milky Way halo in which $N(\HI) < 10^{18}$~{\cmsq}, with a few having $N(\HI) \lesssim 10^{17}$~{\cmsq}. The low ionization metal lines associated with these halo clouds are kinematically narrow and weak, identical to the high-$z$ weak {\MgII} systems. Two of the sight lines that cover {\MgII} measure $W_r(2796) < 0.2$~{\AA}. These observations lend further support to the proposition that at least some fraction of the weak {\MgII} absorption systems are likely to have their physical origin in gas clouds residing in the halos of high-$z$ galaxies.

1427: 

1428: The star formation per co-moving volume is known to be roughly

1429: constant between $z \sim 1$ and $z \sim 3$ \citep[e.g.][]{bouwens03, wang06}. 

1430: In this same interval however, the number

1431: density of weak {\MgII} clouds has been found to decline from a value

1432: of $dN/dz = 1.76$ at $z = 1.2$ to $dN/dz = 0.65$ at $z=2.2$ \citep{anand07}. 

1433: If a significant fraction of weak {\MgII} clouds,

1434: especially those in which $N(\FeII) << N(\MgII)$, form in supernova

1435: driven outflows from star-forming galaxies, then it would seem

1436: perplexing that a declining trend is observed for $dN/dz$ from $z \sim

1437: 1$ to $z \sim 2.4$.  We would expect $dN/dz$ of weak {\MgII} absorbers to be 

1438: not decreasing so drastically if they were {\it all}

1439: directly connected to interactions and outflows.  However, we find a

1440: spread in the physical properties of weak {\MgII} absorbers, and an

1441: evolution in these properties from $z \sim 1$ to $z \sim 2$.

1442: We would expect that those weak {\MgII} absorbers with

1443: $N({\FeII}) << N({\MgII})$ could be consistent with an origin

1444: in superwind condensations, since that process would lead to

1445: $\alpha$-enhancement and could produce high metallicities.  We

1446: have found such absorbers both at $z \sim 1$ and $z \sim 2$, and

1447: their numbers are roughly consistent with a constant $dN/dz$ for

1448: the sub-population, and thus with a constant star formation rate over the 

1449: same interval.

1450: 

1451: We then must also consider the sub-population of weak {\MgII}

1452: absorbers with $N({\FeII}) \sim N({\MgII})$.  These objects have a

1453: $dN/dz$ that peaks at $z \sim 1$, and they are apparently rare at $z

1454: \sim 2$.  When we combine the two populations, the $N({\FeII}) \sim

1455: N({\MgII})$ and the $N({\FeII}) << N({\MgII})$ clouds, the result is a

1456: $dN/dz$ that declines from $z \sim 1$ to $z \sim 2.4$, as is observed.

1457: The evolution and the physical properties of the $N({\FeII}) \sim

1458: N({\MgII})$ absorbers provide clues to their origins.  It has been shown that, to reproduce the observed solar composition of iron,  

1459: nucleosynthetic yields from Type Ia events have to included \citep{timmes95}. 

1460: Thus, the weak {\MgII} clouds in which the {\FeII} to {\MgII} column density ratio approximately reflects Solar abundance (i.e. $N(\FeII) \sim N(\MgII)$, see Figure~\ref{fig:6}) should also be Type Ia-enriched. Enrichment by Type Ia supernovae requires that there is a $\sim 1$ billion year delay from

1461: the onset of star formation until the elements produced in the

1462: Type Ia enter the interstellar medium.  What is needed to explain

1463: these clouds is a process that peaks at $z \sim 1.5$ ($\sim 1$~Gyr

1464: before the peak of $dN/dz$ at $z =1.2$) to produce

1465: the stars that subsequently give rise to the iron enrichment.

1466: It has been noted in \citet{lynch07} and \citet{misawa07} that this peak 

1467: closely compares to the peak in the global star

1468: formation rate in dwarf galaxies \citep{kauffmann04, bauer05}.

1469: More generally, for Type Ia SNe to contribute to enrichment, the

1470: absorbing structure must persist for more than $1$ billion

1471: years.  This may exclude superwinds from starbursts and tidal

1472: debris, that are likely to be relatively short-lived.  However,

1473: it would be consistent with Type Ia enriched gas trapped in

1474: the potential wells of dwarf galaxies, or with intergalactic

1475: star-forming structures in the cosmic web \citep{weak2, milni06}.  

1476: Either of these sites could

1477: also be consistent with the coherence lengths of low ionization gas

1478: as estimated in \citet{ellison04}. 

1479: 

1480: In the local universe, \citet{stocke04} and \citet{keeney06} have 

1481: associated weak {\MgII} absorbers with unbound winds from dwarf starburst 

1482: galaxies. In their respective

1483: examples, a post-starburst galaxy is identified at impact parameters of $71h^{-1}$~kpc

1484: and $33h^{-1}$~kpc from the line of sight. Compared to

1485: massive star-forming galaxies, the halo escape velocities are smaller

1486: for dwarfs and hence, they are more efficient in transporting metal

1487: enriched gas clouds into intergalactic environments. Yet, outflows

1488: associated with starbursts from dwarfs are also likely to be $\alpha$-enhanced,

1489: such that the $N(\FeII) \sim N(\MgII)$ weak absorbers would remain largely

1490: unexplained. Thus the clouds that have large $N({\FeII})$ require 

1491: a site that has been previously enriched by local star formation, such as

1492: the dwarf galaxies themselves.

1493: 

1494: We conclude, based on the range of physical properties derived for the weak

1495: {\MgII} absorbers, that they arise from at least two types of astrophysical

1496: processes.  The $N(\FeII) << N(\MgII)$ clouds, some of which are $\alpha$--enhanced,

1497: are produced in processes related to the concerted action of massive stars, such as superwinds.

1498: They thus evolve in number along with the global star formation rate in massive

1499: galaxies, so we would expect roughly a constant $dN/dz$ from $z\sim 1$ to $z\sim 5$

1500: from this population.  The $N(\FeII) \sim N(\MgII)$ clouds, which would be enriched

1501: by local Type Ia supernova, could be housed in relatively dense pockets within

1502: the potential wells of dwarf galaxies or intergalactic structures.  Because the

1503: star formation rate in dwarfs peaks at a later epoch ($z \sim 1.5$), we expect

1504: these clouds to emerge from $z\sim 2$ to $z\sim 1$.  The two populations combined

1505: would lead to a gradual increase in the total weak {\MgII} absorber $dN/dz$

1506: from $z\sim 2.4$ to $z\sim 1$.  This picture would predict that the $dN/dz$

1507: of weak {\MgII} absorbers would remain constant from $z\sim 2.4$ up to higher

1508: redshifts, because the star formation rate in massive galaxies was constant.

1509: 

1510: However, the redshift interval $1 < z < 3$ also corresponds to the

1511: epoch over which  hierarchical structure growth and mergers are most

1512: active. The fraction of interacting galaxies, and proto-galaxies

1513: with irregular luminosity profiles,

1514: is observed to be higher towards high redshift \citep[$z > 1$, $\sim 40$\%, ][]{abraham96, vandenbergh96, conselice03, elmegreen05}. Indirect evidence of this

1515: dynamical evolution of galaxies was also noticed by \citet{mshar07} in the evolution of the kinematic profiles of strong {\MgII} systems. Towards high redshift ($z \sim 2$), the

1516: kinematic profiles in a number of strong {\MgII} lines were found to

1517: be particularly complex, with absorption in multiple clouds

1518: that were linked with each other continuously in velocity.  This

1519: is suggestive of ongoing accretion events.  In contrast, $z \sim 0.5$

1520: strong {\MgII} systems typically have a distinct region of strong

1521: absorption due to several blended clouds, surrounded by one or more

1522: weaker high velocity components (i.e. satellite clouds).  This

1523: is the expected absorption signature for quiescent disk/halo galaxies.

1524: It was proposed by Narayanan {\etal} (2007) that the superpositions

1525: of the numerous halo clouds present in high redshift proto-galactic

1526: structures lead to a rarity of isolated weak {\MgII} absorbers during

1527: that epoch.  Instead, the weak {\MgII} absorbers would be consolidated

1528: into a stronger {\MgII} absorber.  The result would be a deficit of

1529: weak {\MgII} absorbers at $z>2.5$.  Near-IR high resolution spectroscopic observations 

1530: are needed to determine the $dN/dz$ of weak {\MgII} absorbers at $z > 2.5$

1531: and to determine their metallicities, ionization conditions, and

1532: chemical abundances.

1533: 

1534: \acknowledgements

1535: We express our gratitude to the ESO for the public data archive and for providing the MIDAS UVES pipeline. The authors are grateful to Gary Ferland for answering questions related to the charge transfer reaction rates and also for making the Cloudy photoionization code openly available. We also wish to thank an anonymous referee for providing valuable suggestions which improved the scope of the paper significantly. AN is thankful to Eric Feigelson for several useful discussions on the application of survival analysis in statistical calculations. This research was funded by NASA under grants NAG5-6399 and NNG04GE73G and by the National Science Foundation (NSF) under grant AST-04-07138.

1536: 

1537: \begin{thebibliography}{XXX}

1538: 

1539: \bibitem[Abraham {\etal}(1996)]{abraham96}

1540: Abraham, R. G., van den Bergh, S., Glazebrook, K., Ellis, R. S., Santiago, B. X., Surma, P., \& Griffiths, R. E. 1996, ApJS, 107, 1.

1541: 

1542: \bibitem[Adelberger {\etal}(2005a)]{adelberger05a}

1543: Adelberger, K. L., Shapley, A. E., Steidel, C. C., Pettini, M., Erb, D. K., \& Reddy, N. A. 2005, ApJ, 629, 636. 

1544: 

1545: \bibitem[Adelberger {\etal}(2003)]{adelberger03}

1546: Adelberger, K. L., Steidel, C. C., Shapley, A. E., \& Pettini, M. 2003, ApJ, 584, 45.

1547: 

1548: \bibitem[Allende Prieto {\etal}(2001)]{allende01}

1549: Allende Prieto, C., Lambert, D. L., \& Asplund, M. 2001, \apj, 556, L63.

1550: 

1551: \bibitem[Allende Prieto {\etal}(2002)]{allende02}

1552: Allende Prieto, C., Lambert, D. L., \& Asplund, M. 2002, \apj, 573, L137. 

1553: 

1554: \bibitem[Aracil {\etal}(2006)]{aracil06}

1555: Aracil, B., Tripp, T. M., Bowen, D. V., Prochaska, J. X., Chen, H. -W., \& Frye, B. L. 2006, MNRAS, 367. 139.

1556: 

1557: \bibitem[Bauer {\etal}(2005)]{bauer05}

1558: Bauer, A. E., Drory, N., Hill, G. J., \& Feulner, G. 2005, ApJ, 621, 89.

1559: 

1560: \bibitem[Bergeron {\etal}(1991)]{bergeron91}

1561: Bergeron, J., \& Boiss\'e, P. 1991, A\&A, 243, 344.

1562: 

1563: \bibitem[Bond {\etal}(2001)]{bond01}

1564: Bond, N. A., Churchill, C. W. C., Charlton, J. C., \& Vogt, S. S. 2001, ApJ, 562, 641.

1565: 

1566: \bibitem[Bouwens {\etal}(2003)]{bouwens03}

1567: Bouwens, R. J. {\etal} 2003, ApJ, 595, 589.

1568: 

1569: \bibitem[Charlton {\etal}(2003)]{charlton03}

1570: Charlton, J. C., Ding, J., Zonak, S. G., Churchill, C. W., Bond, N. A., \& Rigby, J. R. 2003, ApJ, 589, 111.

1571: 

1572: \bibitem[Charlton \& Churchill (1998)]{charlton98}

1573: Charlton, J. C., \& Churchill, C. W. C. 1998, ApJ, 499, 181.

1574: 

1575: \bibitem[Choi {\etal}(2006)]{choi06}

1576: Choi, P. I. {\etal}2006, ApJ, 637, 227.

1577: 

1578: \bibitem[Churchill(1997)]{cwcthesis} 

1579: Churchill, C.~W.\ 1997, Ph.D.~Thesis

1580: 

1581: \bibitem[Churchill {\etal}(2007)]{cwc07}

1582: Churchill, C. W. C., Kacprzak, G. G., Steidel, C. C., \& Evans, J. L. 2007, ApJ, 661, 714.

1583: 

1584: \bibitem[Churchill {\etal}(2005)]{cwc05}

1585: Churchill, C. W. C., Kacprzak, G. G., Steidel, C. C. 2005,  IAU Colloquium Proceedings, Cambridge University Press, pp 24-41.

1586: 

1587: \bibitem[Churchill \& Le Brun(1998)]{cwc98}

1588: Churchill, C. W., \& Le Brun, V. 1998, ApJ, 499, 677.

1589: 

1590: \bibitem[Churchill {\etal}(1999)]{weak1}

1591: Churchill, C. W., Rigby, J. R., Charlton, J. C., \& Vogt, S. S. 1999, \apjs, 120, 51.

1592: 

1593: \bibitem[Churchill {\etal}(1996)]{cwc96}

1594: Churchill, C. W., Steidel, C. C., \& Vogt, S. S. 1996, \apj, 471, 164. 

1595: 

1596: \bibitem[Churchill \& Vogt(2001)]{cwcvogt01}

1597: Churchill, C. W., \& Vogt, S. S. 2001, \aj, 122, 679.

1598: 

1599: \bibitem[Churchill {\etal}(2003)]{cwc03}

1600: Churchill, C. W. C., Vogt, S. S., \& Charlton, J. C. 2003, ApJ, 125. 98

1601: 

1602: \bibitem[Collins et al.(2004)]{collins04} Collins, J.~A., Shull, 

1603: J.~M., \& Giroux, M.~L.\ 2004, \apj, 605, 216 

1604: 

1605: \bibitem[Conselice {\etal}(2003)]{conselice03}

1606: Conselice, C. J., Bershady, M. A., Dickinson, M., \& Papovich, C. 2003, AJ, 126, 1183.

1607: 

1608: \bibitem[Dav\'e {\etal}(1997)]{dave97}

1609: Dav\'e, R., Hernquist, L., Weinberg, D. H., \& Katz, N. 1997, ApJ, 477, 21.

1610: 

1611: \bibitem[Dekker {\etal}(2000)]{dekker00}

1612: Dekker, H., D'Odorico, S., Kaufer, A., Delabre, B., \& Kotzlowski, H. 2000, SPIE, 4008, 534.

1613: 

1614: \bibitem[Dessauges-Zavadsky {\etal}(2003)]{dz03}

1615: Dessauges-Zavadsky, M., P\'eroux, C., Kim, T. S., D'Odorico, S., \& McMahon, R. G. 2003, MNRAS, 345, 447.

1616: 

1617: \bibitem[Ding {\etal}(2003)]{ding03}

1618: Ding, J., Charlton, J. C., Bond, N. A., Zonak, S. G., \& Churchill, C. W. 2003, ApJ, 587, 551.

1619: 

1620: \bibitem[Ellison {\etal}(2004)]{ellison04}

1621: Ellison, S. L., Ibata, R., Pettini, M., Lewis, G. F., Aracil, B., Petitjean, P., \& Srianand, R. 2004, A\&A, 414, 79. 

1622: 

1623: \bibitem[Elmegreen {\etal}(2005)]{elmegreen05}

1624: Elmegreen, D. M., Elmegreen, B. G., Rubin, D. S., \& Schaffer, M. A. 2005, \apj, 631, 85.

1625: 

1626: \bibitem[Feigelson, \& Nelson(1985)]{edf85}

1627: Feigelson, E. D., \& Nelson, P. I. 1985, ApJ, 293, 192.

1628: 

1629: \bibitem[Ferland {\etal}(1998)]{ferland98}

1630: Ferland, G., Korista, K. T., Verner, D. A., Ferguson, J. W., Kingdon, J. B., \& Verner, E. M. 1998, PASP, 110, 76.

1631: 

1632: \bibitem[Fox et al.(2005)]{fox05} Fox, A.~J., Wakker, B.~P., 

1633: Savage, B.~D., Tripp, T.~M., Sembach, K.~R., 

1634: \& Bland-Hawthorn, J.\ 2005, \apj, 630, 332 

1635: 

1636: \bibitem[Ganguly et al.(2005)]{ganguly05} Ganguly, R., Sembach, 

1637: K.~R., Tripp, T.~M., \& Savage, B.~D.\ 2005, \apjs, 157, 251 

1638: 

1639: \bibitem[Grevesse \& Sauval (1998)]{grevesse98}

1640: Grevesse, N., \& Sauval, A. J. 1998, Space Science Reviews, 85, 161-174.

1641: 

1642: \bibitem[Haardt \& Madau(1996)]{hm96}

1643: Haardt, F., \& Madau, P. 1996, ApJ, 461, 20.

1644: 

1645: \bibitem[Heckman {\etal}(2000)]{heckman00}

1646: Heckman, T. M., Lehnert, M. D., Strickland, D. K., \& Armus, L. 2000, ApJS, 129, 493.

1647: 

1648: \bibitem[Heckman {\etal}(2001)]{heckman01}

1649: Heckman, T. M., Sembach, K. R., Meurer, G. R., Strickland, D. K., Martin, C. L., Clazetti, D., \& Leitherer, C. 2001, ApJ, 554, 1021.

1650: 

1651: \bibitem[Holweger {\etal}(2001)]{holweger01}

1652: Holweger, H. 2001, {\it Solar and Galactic Composition} Workshop, Ed. by Robert F. Wimmer-Schweingruber. Pub. American Institute of Physics Conference Proceedings. Vol 598.

1653: 

1654: \bibitem[Isobe {\etal}(1986)]{isobe86}

1655: Isobe, T., Feigelson, E. D., \& Nelson, P. I. 1986, ApJ, 306, 490.

1656: 

1657: \bibitem[Kauffmann {\etal}(2004)]{kauffmann04}

1658: Kauffmann, G., White, S. D. M., Heckman, T. M., M\'enard, B., Brinchmann, J., Charlot, S., Tremonti, C., \& Brinkmann, J. 2004, MNRAS, 353, 713.

1659: 

1660: \bibitem[Keeney {\etal}(2006)]{keeney06}

1661: Keeney, B. A., Stocke, J. T., Rosenberg, J. L., Tumlinson, J., \& York, D. G. 2006, AJ, 132. 2496.

1662: 

1663: \bibitem[Khare et al.(2005)]{khare05} Khare, P., et al.\ 2005, 

1664: IAU Colloq.~199: Probing Galaxies through Quasar Absorption Lines, 427 

1665: 

1666: \bibitem[Kingdon {\etal}(1996)]{kingdon96}

1667: Kingdon, J. B., \& Ferland, G. F. 1996, \apjs, 106, 205.

1668: 

1669: \bibitem[Lanzetta {\etal}(1987)]{lanzetta87}

1670: Lanzetta, K. M., Wolfe, A. M., \& Turnshek, D. A. 1987, \apj, 322, 739.

1671: 

1672: \bibitem[Lavalley {\etal}(1992)]{lavalley92}

1673: Lavalley, M., Isobe, T., \& Feigelson, E. 1992, 25, 245.

1674: 

1675: \bibitem[Lynch {\etal}(2006)]{lynch06}

1676: Lynch, R. S., Charlton, J. C., \& Kim, T. -S. 2006, ApJ, 640, 81. 

1677: 

1678: \bibitem[Lynch \& Charlton(2007)]{lynch07}

1679: Lynch, R. S., \& Charlton, J. C. 2007, ApJ, 666, 64.

1680: 

1681: \bibitem[Masiero {\etal}(2005)]{masiero05}

1682: Masiero, J. R., Charlton, J. C., Ding, J., Churchill, C. W., \& Kacprzak, G. 2005, \apj, 623, 57.

1683: 

1684: \bibitem[Meiring {\etal}(2007)]{meiring07}

1685: Meiring, J. D., Lauroesch, J. T., Kulkarni, V. P., P\'eroux, C., Khare, P., York, D. G., \& Crotts, A. P. S. 2007, MNRAS, 376, 557.

1686: 

1687: \bibitem[Milutinovi\'c {\etal}(2006)]{milni06}

1688: Milutinovi\'c, N., Rigby, J. R., Masiero, J. R., Lynch, R. S., Palma, C., \& Charlton, J. C. 2006, ApJ, 641, 190.

1689: 

1690: \bibitem[Misawa {\etal}(2007)]{misawa07}

1691: Misawa, T., Charlton, J. C., \& Narayanan, A. 2007, ApJ submitted.

1692: 

1693: \bibitem[Mshar {\etal}(2007)]{mshar07}

1694: Mshar, A. C., Charlton, J. C., Lynch, R. S., Churchill, C. W. C., \& Kim, T. S. 2007, ApJ in press, arXiv0706.0515.

1695: 

1696: \bibitem[Narayanan {\etal}(2005)]{anand05}

1697: Narayanan, A., Charlton, J. C., Masiero, J. R., \& Lynch, R. 2005, \apj, 632, 92.

1698: 

1699: \bibitem[Narayanan {\etal}(2007)]{anand07}

1700: Narayanan, A., Misawa, T., Charlton, J. C., \& Kim, T. S. 2007, ApJ, 660. 1093.

1701: 

1702: \bibitem[Nestor {\etal}(2005)]{nestor05}

1703: Nestor, D. B., Turnshek, D. A., \& Rao, S. M. 2005, ApJ, 628, 637.

1704: 

1705: \bibitem[Pettini {\etal}(2001)]{pettini01}

1706: Pettini, M., Shapley, A. E., Steidel, C. C., Cuby, J. -G., Dickinson, M., Moorwood, A. F. M., Adelberger, K. L., \& Giavalisco, M. 2001, ApJ, 554, 981.

1707: 

1708: \bibitem[Putman et al.(2003)]{putman03} Putman, M.~E., 

1709: Bland-Hawthorn, J., Veilleux, S., Gibson, B.~K., Freeman, K.~C., 

1710: \& Maloney, P.~R.\ 2003, \apj, 597, 948 

1711: 

1712: \bibitem[Rauch {\etal}(1999)]{rauch99}

1713: Rauch, M., Sargent, W. L. W., \& Barlow, T. A. 1999, ApJ, 515, 500.

1714: 

1715: \bibitem[Rauch {\etal}(2002)]{rauch02}

1716: Rauch, M., Sargent, W. L. W., Barlow, T. A., \& Simcoe, R. A. 2002, ApJ, 576, 45.

1717: 

1718: \bibitem[Richter {\etal}(2008)]{richter08}

1719: Richter, P., Charlton, J. C., Fangano, A. P. M., Bekhti, N. B., \& Masiero, J. R. 2008, ApJ submitted.

1720: 

1721: \bibitem[Rigby {\etal}(2002)]{weak2}

1722: Rigby, J. R., Charlton, J. C., \& Churchill, C. W. 2002, ApJ, 565, 743.

1723: 

1724: \bibitem[Rupke {\etal}(2002)]{rupke02}

1725: Rupke, D. S., Veilleux, S., \& Sanders, D. B. 2002, ApJ, 570, 588.

1726: 

1727: \bibitem[Sargent {\etal}(1989)]{sargent89}

1728: Sargent, W. L. W., Steidel, C. C., \& Boksenberg, A. 1989, ApJS, 69, 703.

1729: 

1730: \bibitem[Schaye {\etal}(2007)]{schaye07}

1731: Schaye, J., Carswell, R. F., \& Kim, T. S. 2007, MNRAS, 379, 1169.

1732: 

1733: \bibitem[Schwartz {\etal}(2006)]{schwartz06}

1734: Schwartz, C. M., Martin, C. L., Chandar, R., Leitherer, C., Heckman, T. M., \& Oey, M. S. 2006, ApJ, 646, 858.

1735: 

1736: \bibitem[Simcoe {\etal}(2004)]{simcoe04}

1737: Simcoe, R. A., Sargent, W. L. W., \& Rauch, M. 2004, ApJ, 606, 92.

1738: 

1739: \bibitem[Simcoe {\etal}(2006)]{simcoe06}

1740: Simcoe, R. A., Sargent, W. L. W., Rauch, M., \& Becker, G. 2006, \apj, 637, 648.

1741: 

1742: \bibitem[Steidel {\etal}(1994)]{steidel94}

1743: Steidel, C. C., Dickinson, M., \& Persson, S. E. 1994, ApJ, 437, 75.

1744: 

1745: \bibitem[Steidel {\etal}(2002)]{steidel02}

1746: Steidel, C. C., Kollmeier, J. A., Shapley, A. E., Churchill, C. W., Dickinson, M., \& Pettini, M. 2002, ApJ, 570, 526.

1747: 

1748: \bibitem[Steidel \& Sargent(1992)]{ss92}

1749: Steidel, C. C., \& Sargent, W. L. W. 1992, \apjs, 80, 1.

1750: 

1751: \bibitem[Stengler-Larrea {\etal}(1995)]{strengler95}

1752: Stengler-Larrea {\etal}. 1995, ApJ, 444, 64.

1753: 

1754: \bibitem[Stocke {\etal}(2004)]{stocke04}

1755: Stocke, J. T., Keeney, B. A., McLin, K. M., Rosenberg, J. L., Weymann, R. J. \& Giroux, M. L. 2004, ApJ, 609, 94.

1756: 

1757: \bibitem[Tappe \& Black(2004)]{tappe04}

1758: Tappe, A. \& Black, J. H. 2004, A\&A, 423, 943.

1759: 

1760: \bibitem[Timmes {\etal}(1995)]{timmes95}

1761: Timmes, F. X., Woosley, S. E., \& Weaver, T. A. 1995, \apjs, 98, 617.

1762: 

1763: \bibitem[Tripp {\etal}(2006)]{tripp06}

1764: Tripp, T. M., Aracil, B., Bowen, D. V., \& Jenkins, E. B. 2006, ApJ, 643, 77.

1765: 

1766: \bibitem[Tripp \& Bowen (2005)]{tripp05}

1767: Tripp, T. M., \& Bowen, D. V. 2005, IAU Colloquium Proceedings, Cambridge University Press, pp 5-23.

1768: 

1769: \bibitem[van den Bergh {\etal}(1996)]{vandenbergh96}

1770: van den Bergh, S., Abraham, R. G., Ellis, R. S., Tanvir, N. R., Santiago, B. X., \& Glazebrook, K. G. 1996, AJ, 112, 359.

1771: 

1772: \bibitem[Vladilo {\etal}(2001)]{vladilo01}

1773: Vladilo, G., Centuri\'on, M., Bonifacio, P., \& Howk, J. C. 2001, \apj, 557, 1007.

1774: 

1775: \bibitem[Wakker 

1776: \& van Woerden(1997)]{wakker97} Wakker, B.~P., \& van Woerden, H.\ 1997, \araa, 35, 217 

1777: 

1778: \bibitem[Wang {\etal}(2006)]{wang06}

1779: Wang, W.-H., Cowie, L. L., \& Barger, A. J. 2006, \apj, 647, 74.

1780: 

1781: \bibitem[York et al.(2006)]{york06} York, D.~G., et al.\ 2006, 

1782: \mnras, 367, 945 

1783: 

1784: \bibitem[Zonak {\etal}(2004)]{zonak04}

1785: Zonak, S. G., Charlton, J. C., Ding, J., \& Churchill, C. W. C. 2004, ApJ, 606, 196.

1786: 

1787: \end{thebibliography}

1788: 

1789: 

1790: %%%%%%%%%%%%%%%%%% TABLES %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1791: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

1792: 

1793: \clearpage

1794: \LongTables

1795: \begin{landscape}

1796: \begin{deluxetable}{lllcrrrrrrr}

1797: \tabletypesize{\scriptsize} 

1798: \tablewidth{0pt}

1799: \tablecaption{EQUIVALENT WIDTH OF METAL LINES ASSOCIATED WITH WEAK {\MgII} ABSORBERS}

1800: \setlength{\tabcolsep}{0.06in}

1801: \tablehead{

1802: \colhead{QSO}&

1803: \colhead{$\lambda~{\AA}$}&

1804: \colhead{z$_{abs}$}&

1805: \colhead{type}&

1806: \colhead{{\MgII} 2796}&

1807: \colhead{{\MgI} 2853}&

1808: \colhead{{\FeII} 2383}&

1809: \colhead{{\AlII} 1671}&

1810: \colhead{{\CII} 1335}&

1811: \colhead{{\SiII} 1260}&

1812: \colhead{{\AlIII} 1855}

1813: }

1814: \startdata

1815: 3c336 & 3530 - 6650 & 0.702901 & s & $	0.028 {\pm} 0.004	$ & $	< 0.010	$ & $	< 0.009 	$ & $	...	$ & $	... 	$ & $	...	$ & $	...	$	\\

1816: 

1817: CTQ 0298 & 3520 - 8550 & 1.256069 & s & $	0.057 {\pm} 0.004	$ & $	< 0.004	$ & $	...	$ & $	...	$ & $	... 	$ & $	...	$ & $	...	$	\\

1818: 

1819: Q 0001-2340 & 3060 - 10070 & 0.452394 & m & $	0.138 {\pm} 0.003	$ & $	< 0.019 ^*	$ & $	0.019 {\pm} 0.001	$ & $	...	$ & $	...	$ & $	...	$ & $	...	$	\\

1820: 

1821:  &  & 0.685957 & s & $	0.035 {\pm} 0.001	$ & $	< 0.002 	$ & $	...	$ & $	...	$ & $	...	$ & $	...	$ & $	< 0.008 	$	\\

1822: 	

1823:  & & 1.651484 & s & $	0.077 {\pm} 0.001	$ & $	0.010 {\pm} 0.001	$ & $	< 0.004 ^* 	$ & $	< 0.003	$ & $	0.075 {\pm} 0.001	$ & $	< 0.081 ^* 	$ & $	0.004 {\pm} 0.001	$	\\

1824: 

1825: Q 0002-4220 & 3160 - 10070 & 1.446496 & s & $	0.042 {\pm} 0.000	$ & $	0.002 {\pm} 0.000	$ & $	...	$ & $	...	$ & $	...	$ & $	...	$ & $	< 0.001	$	\\

1826: 

1827:  &  & 1.988656 & m & $	0.276 {\pm} 0.001	$ & $	... 	$ & $	0.016 {\pm} 0.003	$ & $	0.081 {\pm} 0.001	$ & $	0.171 {\pm} 0.001	$ & $	0.180 {\pm} 0.001	$ & $	0.045 {\pm} 0.002	$	\\

1828: 	

1829: Q 0011+0055 & 3770 - 10000 & 0.487264 & s & $	0.244 {\pm} 0.019	$ & $	< 0.040 	$ & $	   ...	$ & $	...	$ & $	...	$ & $	...	$ & $	...	$	\\

1830: 	

1831: &  & 1.395635 & s & $	0.186 {\pm} 0.004	$ & $	< 0.013	$ & $	...	$ & $	...	$ & $	...	$ & $	...	$ & $	0.043 {\pm} 0.008 	$	\\

1832: 	

1833:  &  & 1.777926 & s & $	0.127 {\pm} 0.003	$ & $	< 0.051	$ & $	...	$ & $	0.059 {\pm} 0.006	$ & $	...	$ & $	...	$ & $	...	$	\\

1834: 	

1835: Q 0013-0029 & 3060 - 9890 & 0.635069 & m & $	0.205 {\pm} 0.014	$ & $	< 0.005 	$ & $	< 0.053 ^*	$ & $	...	$ & $	...	$ & $	...	$ & $	...	$	\\

1836: 	

1837: &   & 0.857469 & m & $	0.150 {\pm} 0.005	$ & $	< 0.005 	$ & $	0.014 {\pm} 0.004	$ & $	...	$ & $	...	$ & $	...	$ & $	0.023 {\pm} 0.003	$	\\

1838: 	

1839:  &   & 1.146770 & m & $	0.047 {\pm} 0.001	$ & $	< 0.009	$ & $	0.004 {\pm} 0.001	$ & $	...	$ & $	...	$ & $	...	$ & $	< 0.034	$	\\

1840: 	

1841: Q 0042-2930 & 3530 - 6800 & 0.798665 & m & $	0.239 {\pm} 0.007	$ & $	< 0.007	$ & $	0.063 {\pm} 0.004	$ & $	...	$ & $	...	$ & $	...	$ & $	...	$	\\

1842: 	

1843: &   & 1.091866 & s & $	0.162 {\pm} 0.005	$ & $	< 0.012	$ & $ 0.69 {\pm} 0.001	$ & $	...	$ & $	...	$ & $	...	$ & $< 0.011	$	\\

1844: 	

1845: Q 0100+130 & 3520 - 10000 &1.758694 & m & $	0.108 {\pm}  0.008	$ & $	< 0.013	$ & $	...	$ & $	...	$ & $	0.125 {\pm} 0.004	$ & $	...	$ & $	...	$	\\

1846: 

1847:  &		 & 2.298494 & m & $	0.232 {\pm}  0.004	$ & $	 ...	$ & $	< 0.003	$ & $	...	$ & $	< 0.291 $ & $	0.081 {\pm} 0.002	$ & $	...	$	\\

1848: 	

1849: Q 0109-3518 & 3060 - 10070 & 0.769646 & s & $	0.044 {\pm} 0.001	$ & $	0.002 {\pm} 0.000	$ & $	< 0.001	$ & $	...	$ & $	...	$ & $	...	$ &	$	< 0.005 ^*	$	\\

1850: 	

1851:  & 	& 0.896004 & m & $	0.020 {\pm} 0.001	$ & $	< 0.001 	$ & $	< 0.003 	$ & $	< 0.009 	$ & $	...	$ & $	...	$ & $	...	$	\\

1852: 	

1853:  &	& 1.182696 & m & $	0.135 {\pm} 0.001	$ & $	0.009 {\pm} 0.001	$ & $	0.016 {\pm} 0.001	$ & $	< 0.020 ^*	$ & $	...	$ & $	... 	$ &	$	< 0.030 ^*	$	\\

1854: 	

1855: Q 0122-380 & 3060 - 10190 &0.822606 & m & $	0.269 {\pm} 0.003	$ & $	0.020 {\pm} 0.002	$ & $	0.034 {\pm} 0.003	$ & $	...	$ & $	...	$ & $	...	$ & $	< 0.020 ^*	$	\\

1856: 	

1857:  & 	& 0.910117 & m & $	0.060 {\pm} 0.004	$ & $	< 0.006 	$ & $	< 0.003 	$ & $	...	$ & $	...	$ & $	...	$ & $	...	$	\\

1858: 	

1859:  & 	& 1.174224 & s & $	0.020 {\pm} 0.001	$ & $	< 0.007	$ & $	< 0.006	$ & $	< 0.005	$ & $	... 	$ & $	...	$ & $	< 0.005	$	\\

1860: 	

1861:  & 	& 1.450076 & m & $	0.061 {\pm} 0.003	$ & $	< 0.009	$ & $	...	$ & $	< 0.005	$ & $	...	$ & $	...	$ & $	< 0.004	$	\\

1862: 	

1863:  & 	& 1.911015 & m & $	0.199 {\pm} 0.006	$ & $	< 0.015	$ & $	0.020 {\pm} 0.002	$ & $	0.037 {\pm} 0.001	$ & $	0.104 {\pm} 0.001	$ & $	0.084 {\pm} 0.001	$ & $	0.015 {\pm} 0.001	$	\\

1864: 	

1865:  &	& 1.974182 & m & $	0.271 {\pm}  0.020	$ & $	< 0.019	$ & $	< 0.008	$ & $	0.020 {\pm} 0.001	$ & $	0.251 {\pm} 0.006	$ & $	< 0.323 ^*	$ & $	0.057 {\pm} 0.007	$	\\

1866: 	

1867: Q 0128-2150 & 3050 - 6800 &1.398315 & s & $	0.018 {\pm} 0.001	$ & $	...	$ & $	< 0.016	$ & $	...	$ & $	< 0.050	$ & $	...	$ & $	...	$	\\

1868: 	

1869:  & 	& 1.422086 & s & $	0.042 {\pm} 0.001	$ & $	...	$ & $	...	$ & $	...	$ & $	< 0.060	$ & $	...	$ & $	...	$	\\

1870: 	

1871: Q 0130-4021 & 3550 - 6800 & 0.962487 & m & $	0.089 {\pm} 0.004	$ & $	< 0.011 	$ & $	...	$ & $	...	$ & $	...	$ & $	...	$ & $	...	$	\\

1872: 	

1873: Q 0136-231 & 3500 - 6640 & 1.261761 & s & $	0.102 {\pm} 0.003	$ & $	< 0.013	$  & $	< 0.009	$ & $	0.026 {\pm} 0.003	$ & $	...	$ & $	...	$ & $	0.021 {\pm} 0.003	$	\\

1874: 	

1875:  & 	 & 1.285796 & s & $	0.021 {\pm} 0.003	$ & $	< 0.013	$ & $	 < 0.009	$ & $	< 0.011	$ & $	...	$ & $	...	$ & $	< 0.010	$	\\

1876: 	

1877:  & 	& 1.353662 & m & $	0.170 {\pm} 0.004	$ & $	... 	$ & $	< 0.019	$ & $	...	$ & $	...	$ & $	...	$ & $	0.029 {\pm} 0.003	$	\\

1878: 	

1879: Q 0141-3932 & 3060 - 10000 & 1.781686 & s & $	0.039 {\pm} 0.001	$ & $	< 0.011	$ & $	< 0.008	$ & $	0.006 {\pm} 0.001	$ & $	0.025 {\pm} 0.001	$ & $	0.020 {\pm} 0.001	$ & $	0.004 {\pm} 0.000	$	\\

1880: 	

1881: Q 0151-4326 & 3060 - 10070 & 0.737248 & s & $	0.022 {\pm} 0.001	$ & $	< 0.001	$ & $	< 0.002	$ & $	...	$ & $	...	$ & $	...	$ & $	< 0.003	$	\\

1882: 	

1883:  & 	& 1.708492 & s & $	0.026 {\pm} 0.001	$ & $	< 0.002	$ & $	0.005 {\pm} 0.001	$ & $	...	$ & $	...	$ & $	...	$ & $	0.003 {\pm} 0.001	$	\\

1884: 	

1885: Q 0237-23 & 3060 - 10070 & 1.184624 & m & $	0.140 {\pm} 0.002	$ & $	0.004 {\pm} 0.000	$ & $	0.009 {\pm} 0.001	$ & $	0.033 {\pm} 0.002	$ & $	...	$ & $	...	$ & $	< 0.016	$	\\

1886: 	

1887: Q 0328-272 & 3500 - 6630 & 0.570827 & s & $	0.168 {\pm} 0.008	$ & $	< 0.013	$ & $	< 0.044 ^* 	$ & $	...	$ & $	...	$ & $	...	$ & $	...	$	\\

1888: 	

1889:  &	& 1.269042 & m & $	0.105 {\pm} 0.007	$ & $	< 0.016	$ & $	< 0.015	$ & $	< 0.031	$ & $	...	$ & $	...	$ & $	< 0.017	$	\\

1890: 	

1891: Q 0329-2550 & 3060 - 10070 &0.992899 & m & $	0.279 {\pm} 0.002	$ & $	< 0.001	$ & $	0.027 {\pm} 0.002	$ & $	...	$ & $	...	$ & $	...	$ & $	0.005 {\pm} 0.001	$	\\

1892: 	

1893: &	 & 1.398230 & m & $	0.025 {\pm} 0.001	$ & $	< 0.006	$ & $	< 0.003	$ & $	...	$ & $	...	$ & $	...	$ & $	...	$	\\

1894: 	

1895: Q 0329-3850 & 3070 - 8500 &0.929608 & s & $	0.073 {\pm} 0.002	$ & $	< 0.003	$ & $	< 0.003	$ & $	< 0.007 	$ & $	...	$ & $	...	$ & $	< 0.004 	$	\\

1896: 	

1897:  &	& 0.970957 & s & $	0.055 {\pm} 0.001	$ & $	< 0.003 	$ & $	< 0.004	$ & $	< 0.011 	$ & $	...	$ & $	...	$ & $	< 0.031 ^*	$	\\

1898: 	

1899: Q 0429-4901 & 3050 - 10080 & 0.584249 & s & $	0.016 {\pm} 0.002	$ & $	< 0.004 	$ & $	< 0.004 	$ & $	...	$ & $	...	$ & $	...	$ & $	...	$	\\

1900: 	

1901:  &	& 1.680766 & s & $	0.015 {\pm} 0.001	$ & $	...	$ & $	< 0.012	$ & $	< 0.009	$ & $	0.012 {\pm} 0.001	$ & $	< 0.024	$ & $	0.005 {\pm} 0.001	$	\\

1902: 	

1903: Q 0453-4230 & 3060 - 10070 &0.895865 & s & $	0.034 {\pm} 0.001	$ & $	< 0.001	$ & $	< 0.001	$ & $	< 0.003 ^*	$ & $	...	$ & $	...	$ & $	...	$	\\

1904: 	

1905:  & 	& 1.039517 & m & $	0.189 {\pm} 0.003	$ & $	...	$ & $	0.105 {\pm} 0.001	$ & $	...	$ & $	... 	$ & $	... 	$ & $	...	$	\\

1906: 	

1907: Q 0549-213 & 3500 - 6640 & 1.343495 & s & $	0.181 {\pm} 0.010	$ & $	...	$ & $	< 0.012	$ & $	< 0.12	$ & $	... 	$ & $	...	$ & $	< 0.013	$	\\

1908: 	

1909: Q 0551-3637 & 3060 - 9370 &0.505437 & m & $	0.052 {\pm} 0.014	$ & $	< 0.013	$ & $	...	$ & $	...	$ & $	...	$ & $	...	$ & $	...	$	\\

1910: 	

1911:  & 	& 1.491972 & s & $	0.176 {\pm} 0.004 	$ & $	< 0.018	$ & $	< 0.010	$ & $	0.043 {\pm} 0.005	$ & $	0.108 {\pm} 0.010	$ & $	...	$ & $	< 0.019	$	\\

1912: 	

1913: Q 0810+2554 & 3050 - 6640 & 0.821727 & m & $	0.252 {\pm} 0.001	$ & $	0.019 {\pm} 0.001	$ & $	...	$ & $	...	$ & $	...	$ & $	...	$ & $	< 0.003	$	\\

1914: 	

1915:  &	& 0.831511 & m & $	0.158 {\pm} 0.002	$ & $	0.004 {\pm} 0.001	$ & $	...	$ & $	< 0.008 	$ & $	...	$ & $	...	$ & $	< 0.031 ^*	$	\\

1916: 	

1917: Q 0926-0201 & 3060 - 10000 &1.096336 & s & $	0.020 {\pm} 0.001	$ & $	< 0.001	$ & $	< 0.006	$ & $	< 0.013	$ & $	...	$ & $	...	$ & $	< 0.023	$	\\

1918: 	

1919:  &	 & 1.232206 & m & $	0.069 {\pm} 0.004	$ & $	< 0.003	$ & $	0.010 {\pm} 0.001	$ & $	< 0.002	$ & $	...	$ & $	...	$ & $	< 0.005	$	\\

1920: 	

1921: Q 0940-1050 & 3110 - 10070 & 2.174535 & s & $	0.035 {\pm} 0.001	$ & $	< 0.008	$ & $	< 0.005	$ & $	0.004 {\pm} 0.000	$ & $	< 0.029 ^*	$ & $	< 0.033 ^*	$ & $	0.011 {\pm} 0.000	$	\\

1922: 	

1923: Q 1122-1648 & 3060 - 10070 & 0.806215 & m & $	0.249 {\pm} 0.001	$ & $	< 0.003	$ & $	0.032 {\pm} 0.001	$ & $	...	$ & $	...	$ & $	...	$ & $	0.045 {\pm} 0.001	$	\\

1924: 	

1925:  &	 & 1.234140 & m & $	0.200 {\pm} 0.000	$ & $	0.010 {\pm} 0.000 	$ & $	0.017 {\pm} 0.000	$ & $	< 0.133	$ & $	...	$ & $	...	$ & $	0.028 {\pm} 0.000	$	\\

1926: 	

1927: Q 1140+2711 & 3775 - 10000 & 2.196632 & m & $	0.193 {\pm} 0.002	$ & $	< 0.061 ^* 	$ & $	...	$ & $	...	$ & $	0.173 {\pm} 0.001	$ & $	< 0.295 ^*	$ & $	...	$	\\

1928: 	

1929: Q 1151+068 & 3705 - 10000 & 1.153704 & s & $	0.108 {\pm} 0.003	$ & $	< 0.008	$ & $	< 0.004	$ & $	...	$ & $	...	$ & $	...	$ & $	< 0.016	$	\\

1930: 	

1931: Q 1157+014 & 3520 - 7400 & 1.330502 & s & $	0.120 {\pm} 0.002	$ & $	< 0.007	$ & $	0.020 {\pm} 0.004	$ & $	0.024 {\pm} 0.006	$ & $	...	$ & $	...	$ & $	0.013 {\pm} 0.002	$	\\

1932: 	

1933: Q 1158-1843 & 3070 - 10070 & 0.506041 & s & $	0.022 {\pm} 0.001	$ & $	< 0.003	$ & $	< 0.021 ^*	$ & $	...	$ & $	...	$ & $	...	$ & $	...	$	\\

1934: 	

1935:  &	& 0.818119 & m & $	0.063 {\pm} 0.001	$ & $	< 0.001	$ & $	0.009 {\pm} 0.001	$ & $	...	$ & $	...	$ & $	...	$ & $	< 0.004 	$	\\

1936: 	

1937: Q 1209+0919 & 3520 - 7770 & 1.264983 & s & $	0.083 {\pm} 0.007	$ & $	< 0.019	$ & $	< 0.015	$ & $	...	$ & $	... 	$ & $	...	$ & $	...	$	\\

1938: 	

1939: Q 1229-021 & 3530 - 6650 & 0.700377 & s  & $	0.010 {\pm} 0.001	$ & $	< 0.004	$ & $	< 0.006	$ & $	...	$ & $	...	$ & $	...	$ & $	...	$	\\

1940: 	

1941:  &	& 0.768862 & s & $	0.033 {\pm} 0.002	$ & $	< 0.005 	$ & $	< 0.005	$ & $	...	$ & $	...	$ & $	...	$ & $	...	$	\\

1942: 	

1943:  &	& 0.830858 & m & $	0.134 {\pm} 0.000	$ & $	0.011 {\pm} 0.001	$ & $	0.032 {\pm} 0.002	$ & $	...	$ & $	... 	$ & $	...	$ & $	...	$	\\

1944: 	

1945: Q 1418-064 & 3765 - 9945 & 1.516673 & s & $	0.075 {\pm} 0.003	$ & $	< 0.019 ^*	$ & $	...	$ & $	...	$ & $	...	$ & $	...	$ & $	< 0.047 ^*	$	\\

1946: 	

1947:  &	& 2.174224 & s & $	0.178 {\pm} 0.004	$ & $	< 0.023 ^*	$ & $	0.018 {\pm} 0.003	$ & $	...	$ & $	...	$ & $	...	$ & $	...	$	\\

1948: 	

1949: Q 1444+014 & 3520 - 5830 & 0.509719 & m & $	0.193 {\pm} 0.007	$ & $	0.022 {\pm} 0.003	$ & $	< 0.129 ^*	$ & $	...	$ & $	...	$ & $	...	$ & $	...	$	\\

1950: 	

1951:  &	& 1.101989 & m & $	0.033 {\pm} 0.005	$ & $	< 0.007	$ & $	0.005 {\pm} 0.000	$ & $	...	$ & $	...	$ & $	...	$ & $	0.032 {\pm} 0.005	$	\\

1952: 	

1953: Q 1448-232 & 3060 - 10070 & 1.019191 & m & $	0.033 {\pm} 0.005	$ & $	...	$ & $	< 0.002	$ & $	< 0.070	$ & $	... 	$ & $	...	$ & $	< 0.002	$	\\

1954: 	

1955:  &	& 1.473201 & m & $	0.269 {\pm} 0.009	$ & $	0.011 {\pm} 0.000	$ & $	0.007 {\pm} 0.001	$ & $	0.024 {\pm} 0.001	$ & $	...	$ & $	< 0.116 ^* 	$ & $	< 0.051	$	\\

1956: 	

1957:  & 	& 1.585464 & m & $	0.088 {\pm} 0.002	$ & $	0.008 {\pm} 0.001	$ & $	0.005 {\pm} 0.000	$ & $	0.005 {\pm} 0.001	$ & $	0.069 {\pm} 0.001	$ & $	0.044 {\pm} 0.001	$ & $	0.022 {\pm} 0.002	$	\\

1958: 	

1959: Q 1621-0042 & 3530 - 6800 & 1.174521 & m & $	0.237 {\pm} 0.012	$ & $	0.004 {\pm} 0.001	$ & $	0.003 {\pm} 0.001	$ & $	...	$ & $	...	$ & $	...	$ & $	...	$	\\

1960: 	

1961: Q 1629+120 & 3050 - 6800 & 1.379330 & m & $	0.142 {\pm} 0.007	$ & $	< 0.018	$ & $	< 0.017	$ & $	...	$ & $	...	$ & $	...	$ & $	...	$	\\

1962: 	

1963: Q 2000-330 & 3495 - 9945 & 1.249864 & s & $	0.032 {\pm} 0.001	$ & $	< 0.003	$ & $	< 0.004	$ & $	...	$ & $	... 	$ & $	...	$ & $	...	$	\\

1964: 	

1965: Q 2044-168 & 3520 - 9900 & 1.342525 & m & $	0.057 {\pm} 0.004	$ & $	< 0.013	$ & $	...	$ & $	< 0.014	$ & $	...	$ & $	...	$ & $	< 0.013	$	\\

1966: 	

1967: Q 2059-360 & 3750 - 9280 & 1.242973 & s & $	0.015 {\pm} 0.001	$ & $	< 0.006	$ & $	...	$ & $	...	$ & $	...	$ & $	...	$ & $	< 0.009	$	\\

1968: 	

1969:  &	& 1.399947 & m & $	0.109 {\pm} 0.002	$ & $	< 0.006	$ & $	0.006 {\pm} 0.002	$ & $	...	$ & $	...	$ & $	...	$ & $	< 0.011	$	\\

1970: 	

1971: Q 2116-358 & 3530 - 6640 & 0.539154 & s & $	0.115 {\pm} 0.004	$ & $	< 0.018	$ & $	...	$ & $	...	$ & $	...	$ & $	...	$ & $	...	$	\\

1972:  &	 & 0.775358 & m & $	0.183 {\pm} 0.010	$ & $	... 	$ & $	< 0.013	$ & $	...	$ & $	...	$ & $	...	$ & $	...	$	\\

1973: 	

1974: Q 2132-433 & 3500 - 6640 & 0.793600 & m & $	0.184 {\pm} 0.007	$ & $	0.012 {\pm} 0.005	$ & $	0.030 {\pm} 0.007	$ & $	...	$ & $	...	$ & $	...	$ & $	...	$	\\

1975: 	

1976: Q 2204-408 & 3520 - 6800 & 1.335279 & m & $	0.052 {\pm} 0.004	$ & $	< 0.009	$ & $	< 0.008	$ & $	...	$ & $	...	$ & $	...	$ & $	...	$	\\

1977: 	

1978: Q 2206-199 & 3420 - 6640 & 0.948363 & m & $	0.265 {\pm} 0.007	$ & $	< 0.005	$ & $	0.016 {\pm} 0.002	$ & $	...	$ & $	...	$ & $	...	$ & $	< 0.007	$	\\

1979: 	

1980:  & 	& 1.297044 & s & $	0.148 {\pm} 0.001	$ & $	0.008 {\pm} 0.001	$ & $	0.005 {\pm} 0.001	$ & $	< 0.064 ^*	$ & $	... 	$ & $	...	$ & $	< 0.005	$	\\

1981: 	

1982: Q 2217-2818 & 3060 - 9890 & 0.599512 & s & $	0.115 {\pm} 0.003	$ & $	 0.005 {\pm} 0.001 	$ & $	< 0.016 ^*	$ & $	...	$ & $	...	$ & $	...	$ & $	...	$	\\

1983: 	

1984:  & 	& 0.786515 & m & $	0.204 {\pm} 0.001	$ & $	< 0.001 	$ & $	0.025 {\pm} 0.005	$ & $	...	$ & $	...	$ & $	...	$ & $	< 0.028 ^*	$	\\

1985: 	

1986:  &	& 1.054310 & m & $	0.046 {\pm} 0.002	$ & $	< 0.002	$ & $	< 0.002	$ & $	< 0.021	$ & $	...	$ & $	...	$ & $	...	$	\\

1987: 	

1988:  &	& 1.082780 & m & $	0.125 {\pm} 0.001	$ & $	< 0.001	$ & $	< 0.001	$ & $	< 0.003	$ & $	...	$ & $	...	$ & $	< 0.003 ^*	$	\\

1989: 	

1990:  &	& 1.200162 & m & $ 	0.099 {\pm} 0.002	$ & $	< 0.001 ^*	$ & $	< 0.002 ^* 	$ & $	...	$ & $	...	$ & $	...	$ & $	< 0.022 ^*	$	\\

1991: 	

1992:  &	& 1.555849 & m & $	0.268 {\pm} 0.001	$ & $	< 0.076	$ & $	0.043 {\pm} 0.000	$ & $	0.036 {\pm} 0.000	$ & $	< 0.271 ^*	$ & $	0.125 {\pm} 0.002	$ & $	< 0.006  ^*	$	\\

1993: 	

1994: Q 2222-3939 & 3530 - 6640 & 1.227553 & s & $	0.114 {\pm} 0.005	$ & $	< 0.019	$ & $	< 0.013	$ & $	< 0.013	$ & $	...	$ & $	...	$ & $	< 0.012	$	\\

1995: 	

1996: Q 2225-2258 & 3050 - 10000 & 0.831374 & m & $	0.033 {\pm} 0.002	$ & $	< 0.003	$ & $	< 0.005 	$ & $	< 0.03 	$ & $	...	$ & $	...	$ & $	< 0.005 	$	\\

1997: 	

1998:  &	 & 1.433018 & m & $	0.167 {\pm} 0.002	$ & $	< 0.011	$ & $	...	$ & $	0.036 {\pm} 0.003	$ & $	0.162 {\pm} 0.004	$ & $	0.010 {\pm} 0.010	$ & $	0.009 {\pm} 0.002	$	\\

1999: 	

2000: Q 2243-6031 & 3140 - 10000 & 0.828081 & m & $	0.263 {\pm} 0.005	$ & $	< 0.003 	$ & $	< 0.007 	$ & $	...	$ & $	...	$ & $	...	$ & $	< 0.009	$	\\

2001: 	

2002:  &	& 1.389707 & m & $	0.106 {\pm} 0.022	$ & $	< 0.003	$ & $	...	$ & $	< 0.007	$ & $	...	$ & $	...	$ & $	...	$	\\

2003: 	

2004:  &	& 1.755704 & s & $	0.108 {\pm} 0.001	$ & $	0.007 {\pm} 0.001	$ & $	< 0.004 	$ & $	< 0.008	$ & $	< 0.106 ^*	$ & $	< 0.058 ^*	$ & $	...	$	\\

2005: 	

2006: Q 2314-409 & 3520 - 6640 & 0.843114 & s & $	0.043 {\pm} 0.003	$ & $	< 0.008	$ & $	< 0.010 	$ & $	...	$ & $	...	$ & $	...	$ & $	...	$	\\

2007: 	

2008: Q 2347-4342 & 3100 - 10070 & 1.109640 & s & $	0.040 {\pm} 0.004	$ & $	< 0.003	$ & $	< 0.003	$ & $	...	$ & $	...	$ & $	...	$ & $	...	$	\\

2009: 	

2010:  &	& 1.405367 & s & $	0.074 {\pm} 0.001	$ & $	< 0.003	$ & $	< 0.004	$ & $	...	$ & $	< 0.095 ^*	$ & $	...	$ & $	< 0.030 ^*	$	\\

2011: 	

2012:  &	& 1.796233 & s & $	0.160 {\pm} 0.001	$ & $	< 0.009	$ & $	0.010 {\pm} 0.001	$ & $	...	$ & $	< 0.163 ^*	$ & $	0.066 {\pm} 0.001	$ & $	0.004 {\pm} 0.000	$

2013: 

2014: \enddata

2015: \tablecomments{$*$ indicates lines that were contaminated by absorption features at other redshifts. In most cases, the contamination was from {\HI} lines of the Ly-$\alpha$ forest. The second column gives the wavelength coverage of the UVES spectrum of each quasar, the third column gives the redshift of each absorber, and the fourth column indicates whether the weak {\MgII} absorption was in a single cloud (s) or in multiple clouds (m). Columns 5 - 11 are the total rest-frame equivalent widths of the respective lines.}

2016: \end{deluxetable}

2017: \clearpage

2018: \end{landscape}

2019: \clearpage

2020: 

2021: \medskip

2022: \medskip

2023: \begin{center}

2024: \vspace{2 in}

2025: \textsf{\large NOTE: Table 2 lists the velocity, column density and Doppler parameter for {\MgI}, {\MgII}, {\FeII}, {\CII}, {\SiII}, {\AlII} and {\AlIII} associated with the 100 weak {\MgII} systems in our sample. The table (18 pages) can be downloaded from \bf{http://www.astro.wisc.edu/$\sim$anand/vptable.pdf}}

2026: \end{center}

2027: \clearpage

2028: 

2029: %%%%%%%%%%%%%%%%%% FIGURES %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2030: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

2031: 

2032: \clearpage

2033: 

2034: \begin{landscape}

2035: \begin{figure*}

2036: %\figurenum{1}

2037: \begin{center}

2038: \epsscale{0.4}

2039: \rotatebox{270}{\plotone{f1a.eps}}

2040: \hspace{8 mm}

2041: \epsscale{0.4}

2042: \rotatebox{270}{\plotone{f1cg.eps}} \\

2043: \epsscale{0.4}

2044: \rotatebox{270}{\plotone{f1w.eps}}

2045: \hspace{1 mm}

2046: \epsscale{0.4}

2047: \rotatebox{270}{\plotone{f1cn.eps}}

2048: \end{center}

2049: \protect

2050: \caption{A few example systems plots from the $100$ weak {\MgII} absorbers in our sample represented on a velocity scale with the system redshift corresponding to $0$~{\kms}. The vertical axis is continuum normalized scale for the spectra. The vertical tick marks correspond to the centroid of the absorption components in each spectral feature, derived from Voigt profile fitting. Regions that are blueward of the Ly-$\alpha$ emission of the quasar are labeled in the respective panels as {\it in  Ly-$\alpha$ forest}. The quasar name and redshift of the system are also labeled in each system plot. ({\it Note: The full set of system plots will be presented in the online version of the journal.})}

2051: \label{fig:1}

2052: \end{figure*}

2053: \clearpage

2054: \end{landscape}

2055: 

2056: 

2057: \begin{figure*}

2058: %\figurenum{2}

2059: \begin{center}

2060: \epsscale{1.0}

2061: \vspace{1.5 in}

2062: \rotatebox{90}{\plotone{f2.eps}}

2063: \end{center}

2064: \protect

2065: \caption{The {\it top panel} shows the frequency distribution of the number of weak {\MgII} 

2066: clouds (i.e. number of Voigt profile components) in our sample. Single cloud systems account 

2067: for 48\% of the total number of systems.The {\it bottom panel} compares the rest-frame integrated equivalent width of $\MgII \lambda 2796$ line as a function of the number of clouds. Most of the weaker systems [$W_r(2796) < 0.1$~{\AA}] have absorption only in one or two clouds.}

2068: \label{fig:2}

2069: \end{figure*}

2070: \clearpage

2071: 

2072: \begin{figure*}

2073: %\figurenum{3}

2074: \epsscale{1.0}

2075: \begin{center}

2076: \vspace{2 in}

2077: \plotone{f3.eps}

2078: \end{center}

2079: \protect

2080: \caption{Distribution of single and multiple cloud weak {\MgII} systems 

2081: as a function of redshift. The redshift distribution is comparable between the two types of 

2082: systems (N$_c = 1$ and N$_c \geq 2$; i.e., single and multiple cloud) and they do not exhibit preference for either 

2083: low ($z \sim 1$) or high ($z \sim 2$) redshift. }

2084: \label{fig:3}

2085: \end{figure*}

2086: \clearpage

2087: 

2088: \begin{figure*}

2089: %\figurenum{4}

2090: \epsscale{1.0}

2091: \begin{center}

2092: \vspace{2 in}

2093: \plotone{f4.eps}

2094: \end{center}

2095: \protect

2096: \caption{The total (i.e. integrated) rest-frame equivalent width of $\MgII \lambda 2796$~{\AA} in each absorber, as a function of the absorber's redshift. The {\it red} and {\it black} data points correspond to single and multiple cloud systems respectively. It is evident that there is no significant correlation between the strength of weak absorber and its redshift.}

2097: \label{fig:4}

2098: \end{figure*}

2099: \clearpage

2100: 

2101: \begin{landscape}

2102: \begin{figure*}

2103: %\figurenum{5}

2104: \epsscale{0.6}

2105: \begin{center}

2106: \vspace{0.8 in}

2107: \rotatebox{90}{\plotone{f5.eps}}

2108: \end{center}

2109: \protect

2110: \caption{The comparison of the total rest-frame equivalent widths of {\MgII} ($W_r(2796)$) in each absorber, with other corresponding low ionization transitions and {\AlIII}. The large solid dots are 3$\sigma$ detections, the large open circles are detections that are affected by blending with an absorption feature at some other redshift, in which case the measurement is considered as an upper limit, indicated by a downward pointing arrow from the open circle. Non-detections at the 3$\sigma$ level are plotted using just the downward pointing arrow. In most cases, the blending was from {\HI} lines of the Ly-$\alpha$ forest. The equivalent width corresponding to non-detections at the 3$\sigma$ level are plotted using downward pointing arrows. The {\it dash-dot} line in the {\CII} and {\SiII} panels are the $y = x$ line for comparison with {\MgII}. The correlation coefficients (Spearman's $\rho$) between $W_r(\MgII)$ and $W_r(\FeII)$, $W_r(\MgI)$, $W_r(\CII)$, $W_r(\SiII)$, $W_r(\AlII)$, $W_r(\AlIII)$ are  0.64(0.00), 0.50(0.00), 0.75(0.00), 0.38(0.20), 0.67(0.00), and 0.38(0.00) respectively. The value in parentheses represents the significance level, i.e., the probability that the observed value of $\rho$ would be greater than or equal to the actual value by chance.}

2111: \label{fig:5}

2112: \end{figure*}

2113: \clearpage

2114: \end{landscape}

2115: 

2116: \begin{landscape}

2117: \begin{figure*}

2118: %\figurenum{6}

2119: \epsscale{0.7}

2120: \begin{center}

2121: \rotatebox{90}{\plotone{f6.eps}}

2122: \end{center}

2123: \protect

2124: \vspace{0.2in}

2125: \caption{The comparison of {\MgII} column densities with the column density of other low 

2126: ionization transitions and {\AlIII}. The large solid dots are 3$\sigma$ detections, the large open circles are detections that are affected by blending with an absorption feature at some other redshift, 

2127: in which case the measurement is considered as an upper limit, indicated by a downward 

2128: pointing arrow from the open circle. The column density of non-detections, estimated from the $3\sigma$ 

2129: equivalent width limit, are plotted using just the downward pointing arrows. The solid line is a linear 

2130: regression fit formalizing the relationship between the two ions. The standard deviation of the 

2131: corresponding fits are indicated by the {\it dashed-dot} line. The regression lines are drawn for the three ions ({\FeII}, {\CII} and {\SiII}) that display the least scatter with N(\MgII). The {\it red dashed} line in each panel indicates the solar abundance pattern based on values given in \citet{grevesse98,allende01,allende02} and \citet{holweger01}. In the lower left panel, the {\it red dashed} line therefore corresponds to $ y = x$.} 

2132: \label{fig:6}

2133: \end{figure*}

2134: \clearpage

2135: \end{landscape}

2136: 

2137: 

2138: \begin{landscape}

2139: \begin{figure*}

2140: %\figurenum{7}

2141: \epsscale{0.9}

2142: \plotone{f7.eps}

2143: \protect

2144: \vspace{0.3in}

2145: \caption{Cloudy photoionization curves indicating how the column density of {\SiII}, {\CII}, {\MgII}, {\FeII}, {\AlII} and {\MgI} change with ionization parameter (log $U$), metallicity ($Z$)

2146: and neutral hydrogen column density [$N(\HI)$]. The density ($n_{\H}$) was calculated using the expression log $n_{\H}$ = log $n_\gamma$ - log $U$, for log $n_\gamma$ = -4.70, corresponding to the number density of ionizing photons ($h\nu \geq 13.6$~eV) at $z = 2$.}

2147: \label{fig:7}

2148: \end{figure*}

2149: \clearpage

2150: \end{landscape}

2151: 

2152: 

2153: \begin{landscape}

2154: \begin{figure*}

2155: %\figurenum{8}

2156: \epsscale{0.7}

2157: \begin{center}

2158: \rotatebox{270}{\plotone{f8.eps}}

2159: \end{center}

2160: \protect

2161: \caption{Cloudy grid of photoionization models for 0.1Z$_\odot$, Z$_\odot$ and 

2162: 10Z$_\odot$ metallicity with measurements of {\FeII} and {\MgII} column density over-plotted. 

2163: Vertical curves correspond to lines of constant $N(\HI)$ and the horizontal curves lines of 

2164: constant ionization parameter (log $U$). The Cloudy models were computed for the intensity of 

2165: the extragalactic ionizing background radiation (EBR) at $z = 2$ and $z=1$. Weak {\MgII} 

2166: clouds at $z \geq 1.5$ are plotted in the EBR at $z = 2$ panel and clouds at $z < 1.5$ are 

2167: plotted in the EBR at $z =1$ panel. The large solid dots are 3$\sigma$ detections, the large open circles 

2168: are detections that are affected by blending with an absorption feature at some other redshift, 

2169: in which case the measurement is considered as an upper limit, indicated by a downward 

2170: pointing arrow from the open circle. The column density of non-detections, estimated from the $3\sigma$ 

2171: equivalent width limit, are plotted using just the downward pointing arrows.}

2172: \label{fig:8}

2173: \end{figure*}

2174: \clearpage

2175: \end{landscape}

2176: 

2177: \begin{figure*}

2178: %\figurenum{9}

2179: \epsscale{0.55}

2180: \begin{center}

2181: \vspace{1.5 in}

2182: \plotone{f9.eps}

2183: \end{center}

2184: \protect

2185: \caption{The distribution of {\FeII} to {\MgII} column density ratio for single and multiple cloud 

2186: systems. The histogram filled with the light color includes measurements that are upper limits in {\FeII}, whereas the 

2187: histogram filled with the dark color excludes those. For multiple clouds, we have included the {\FeII} to {\MgII} ratio in individual clouds. It is evident from the figure that both single and 

2188: multiple cloud systems span roughly the same range of {\FeII} to {\MgII} column density ratios.}

2189: \label{fig:9}

2190: \end{figure*}

2191: \clearpage

2192: 

2193: \begin{landscape}

2194: \begin{figure*}

2195: %\figurenum{10}

2196: \epsscale{0.75}

2197: \rotatebox{270}{\plotone{f10.eps}}

2198: \protect

2199: \caption{Cloudy grid of photoionization models with measurements of {\MgI} and {\MgII} 

2200: column density over-plotted. The description of the Cloudy curves and the data points are the 

2201: same as in Figure~\ref{fig:8}.}

2202: \label{fig:10}

2203: \end{figure*}

2204: \clearpage

2205: \end{landscape}

2206: 

2207: 

2208: \begin{landscape}

2209: \begin{figure*}

2210: %\figurenum{11}

2211: \epsscale{0.3}

2212: \begin{center}

2213: \rotatebox{90}{\plotone{f11.eps}}

2214: \end{center}

2215: \protect

2216: \vspace{0.1in}

2217: \caption{Cloudy grid of photoionization models with measurements of {\CII} and {\MgII} column 

2218: density over-plotted. The description of the Cloudy curves and the data points are the same as 

2219: in Figure~\ref{fig:8}. Almost all weak {\MgII} absorbers with coverage of $\CII \lambda 1335$ are at $z \geq 1.5$ and therefore we plot all of them on the $z = 2$ EBR plot.}

2220: \label{fig:11}

2221: \vspace{0.2in}

2222: %\begin{figure*}

2223: %\figurenum{12}

2224: \epsscale{0.30}

2225: \begin{center}

2226: \rotatebox{90}{\plotone{f12.eps}}

2227: \end{center}

2228: \protect

2229: \vspace{0.1in}

2230: \caption{Cloudy grid of photoionization models with measurements of {\SiII} and {\MgII} column 

2231: density over-plotted. The description of the Cloudy curves and the data points are the same as 

2232: in Figure~\ref{fig:8}. Almost all weak {\MgII} clouds with coverage of $\SiII \lambda 1260$ are at $z \geq 1.5$ and therefore we plot all of them on the $z = 2$ EBR plot.}

2233: \label{fig:12}

2234: \end{figure*}

2235: \clearpage

2236: \end{landscape}

2237: 

2238: \begin{landscape}

2239: \begin{figure*}

2240: %\figurenum{13}

2241: \epsscale{0.65}

2242: \rotatebox{270}{\plotone{f13.eps}}

2243: \protect

2244: \vspace{0.3in}

2245: \caption{Cloudy grid of photoionization models with measurements of {\AlII} and {\MgII} column 

2246: density over-plotted. The description of the Cloudy curves and the data points are the same as 

2247: in Figure~\ref{fig:8}.}

2248: \label{fig:13}

2249: \end{figure*}

2250: \clearpage

2251: \end{landscape}

2252: 

2253: \begin{landscape}

2254: \begin{figure*}

2255: %\figurenum{14}

2256: \epsscale{1.0}

2257: \vspace{0.6 in}

2258: \plotone{f14.eps}

2259: \protect

2260: \caption{Cloudy grid of photoionization models with measurements of {\AlIII} and {\MgII} 

2261: column density over-plotted. The description of the Cloudy curves and the data points are the 

2262: same as in Figure~\ref{fig:8}.}

2263: \label{fig:14}

2264: \end{figure*}

2265: \clearpage

2266: \end{landscape}

2267: 

2268: \begin{landscape}

2269: \begin{figure*}

2270: %\figurenum{15}

2271: \epsscale{0.40}

2272: \begin{center}

2273: \vspace{2 in}

2274: \rotatebox{90}{\plotone{f15.eps}}

2275: \end{center}

2276: \protect

2277: \vspace{0.2in}

2278: \caption{{\it Left Panel :} The figure compares the {\CII} to {\MgII} ratio with {\FeII} to {\MgII} in weak {\MgII} clouds that have measurements for both {\CII} and {\FeII}. The filled circles correspond to weak {\MgII} clouds that had firm measurements for both {\CII} and {\FeII}. The open circles correspond to data points in which either {\CII} or {\FeII} or both have measurements that are upper limits. {\it Middle Panel :} compares the {\AlIII} to {\MgII} ratio with {\AlII} to {\MgII} in weak {\MgII} clouds that have measurements for both {\AlIII} and {\AlII}. {\it Right Panel :} compares the {\CII} to {\MgII} ratio with {\SiII} to {\MgII} in weak {\MgII} clouds that have simultaneous measurements for both {\CII} and {\SiII}.}

2279: \label{fig:15}

2280: \end{figure*}

2281: \clearpage

2282: \end{landscape}

2283: 

2284: \begin{landscape}

2285: \begin{figure*}

2286: %\figurenum{16}

2287: \begin{center}

2288: \epsscale{0.8}

2289: \vspace{1.3 in}

2290: \plotone{f16.eps}

2291: \end{center}

2292: \protect

2293: \caption{On the {\it left panel} is the {\AlIII} to {\AlII} ratio in weak {\MgII} clouds discussed in this paper. Measurements that are upper limits in {\AlIII} and {\AlII} are indicated using downward and upward pointing arrows respectively. The {\it right panel} shows the {\AlIII} to {\AlII} ratio in sub-DLA and DLA systems taken from the literature. If weak {\MgII} clouds are optically thin in {\HI} (i.e. $N(\HI) < 10^{17}$~{\cmsq}), then the measurements indicate that the anti-correlation between $N(\HI)$ and {\AlIII} to {\AlII} ratio that is observed for DLA and sub-DLA systems, also extends to lower $N(\HI)$ values.}

2294: \label{fig:16}

2295: \end{figure*}

2296: \clearpage

2297: \end{landscape}

2298: 

2299: \begin{figure*}

2300: %\figurenum{17}

2301: \epsscale{0.62}

2302: \vspace{2 in}

2303: \plotone{f17.eps}

2304: \protect

2305: \caption{{\it Top panel : } shows the {\FeII} to {\MgII} ratio in all the weak {\MgII} clouds as a function of redshift of the system. The filled circles are firm detections (i.e. detections that are not limits), the large open circles with arrow pointing downward 

2306: corresponds to {\FeII} lines that are affected by blending with an absorption feature at some 

2307: other redshift, in which case the measurement is considered as an upper limit. The non-

2308: detections at the $3\sigma$ level are plotted using downward pointing arrows. Those weak {\MgII} clouds in which $N(\MgII) > 10^{12.2}$~{\cmsq} are plotted in {\it red}. {\it Bottom 

2309: panel :} shows the distribution of {\FeII} to {\MgII} in absorbers at $z \geq 1.5$ and $z < 1.5$. The width of each bin in the distribution is 0.3. The data used to create the frequency distribution includes only those weak {\MgII} clouds in which $N(\MgII) > 10^{12.2}$~{\cmsq}, below which a large fraction of systems only have upper limits for {\FeII}.}

2310: \label{fig:17}

2311: \end{figure*}

2312: \clearpage

2313: 

2314: \begin{figure*}

2315: %\figurenum{18}

2316: \epsscale{0.7}

2317: \vspace{2 in}

2318: \rotatebox{90}{\plotone{f18.eps}}

2319: \protect

2320: \vspace{0.3in}

2321: \caption{The figure compares the column density ratio between {\FeII} and {\MgII} for weak {\MgII} clouds and the high velocity components (i.e. satellite clouds) of strong {\MgII} systems. In the {\it right panel} the downward pointing arrows indicate measurements that are upper limits in $N(\FeII)$. The {\FeII} and {\MgII} column densities for the satellite clouds of strong {\MgII} systems is taken from \citet{cwcvogt01}, and are represented as {\it large blue} points. The {\it small red} points correspond to weak {\MgII} clouds presented in this paper. The {\it left panel} compares the distribution of the column density ratio between {\FeII} and {\MgII} for weak {\MgII} clouds and the satellite clouds of strong {\MgII} systems. The {\it blue} histogram is for the satellite clouds that are offset in velocity from the main absorbing component by $|\Delta v| > 40$~{\kms}. The histogram shaded with lines is for the satellite clouds that are offset in velocity by $|\Delta v| > 90$~{\kms}. The {\it red} histogram is the distribution for the sample of weak {\MgII} clouds presented in this paper.}

2322: \label{fig:18}

2323: \end{figure*}

2324: \clearpage

2325: 

2326: \end{document}