Table Of Content

February2,2008 PreprinttypesetusingLATEXstyleemulateapjv.11/12/01 SUPER-SOLAR METALLICITY IN WEAK MGII ABSORPTION SYSTEMS AT Z ∼ 1.71 Toru Misawa2, Jane C. Charlton2, and Anand Narayanan2 misawa,charlton,[email protected] February 2, 2008 ABSTRACT Through photoionization modeling, constraints on the physical conditions of three z ∼ 1.7 single- cloud weak Mg II systems (Wr(2796) ≤ 0.3˚A) are derived. Constraints are provided by high resolution R = 45,000, high signal-to-noise spectra of the three quasars HE 0141−3932, HE 0429−4091, and 8 HE2243−6031whichwehaveobtainedfromtheEuropeanSouthernObservatory(ESO)archiveofVery 0 Large Telescope (VLT), Ultraviolet and Visual Echelle Spectrograph (UVES). Results are as follows. 0 2 (1) The single-cloud weak Mg II absorption in the three z ∼ 1.7 systems is produced by clouds with n ionization parameters of −3.8 < logU ∼< −2.0 and sizes of 1 – 100 pc. a J (2) In addition to the low-ionization phase Mg II clouds, all systems need an additional 1 – 3 high- 6 ionizationphaseCIVcloudswithin100kms−1oftheMgIIcomponent. Theionizationparameters 1 ofthe CIVphasesrangefrom−1.9< logU <−1.0,withsizesoftens ofparsecstokiloparsecs. In allcases at least one CIV cloud is centered at the same velocity (within 5 km s−1 in our systems) ] h as the MgII clouds. p - (3) Two of the three single-cloud weak Mg II absorbers have near-solar or super-solar metallicities, if o we assume a solar abundance pattern. Although such large metallicities have been found for z < tr 1 weak Mg II absorbers, these are the first high metallicities derived for such systems at higher s redshifts. Thesestrongconstraintswerepossible becauseofthe specific shapes ofthe Lyαprofiles a [ inthesecases. Allsingle-cloudweakMgIIabsorbersmayhavehighmetallicities,butinsomecases the kinematic spread of the CIV cloud contributions to Lyα do not allow such a determination. 1 v (4) Two of the three weak Mg II systems also need additional low-metallicity, broad Lyα absorption 2 lines, offset in velocity from the metal-line absorption, in order to reproduce the full Lyα profile. 5 4 (5) Metallicity in single-cloud weak Mg II systems are more than an order of magnitude larger than 2 those in Damped Lyα systems at z ∼ 1.7. In fact, there appears to be a gradual decrease in . metallicity with increasing N , from these, the most metal-rich Lyα forest clouds, to Lyman 1 HI 0 limit systems, to sub-DLAs, and finally to the DLAs. Weak Mg II absorbers could be near local 8 pocketsinwhichstar-formationhasoccurred,butwherethereislittlegastodilutethemetalsthat 0 are dispersed into the region, resulting in their very high metallicities. We speculate that DLAs : may be subject to the opposite effect, where a large dilution of metals produced in the vicinity v will occur, leading to a small metallicity. i X Subject headings: quasars: intergalactic medium – galaxies: abundances — absorption lines – quasars: r individual (HE 0141−3932,HE 0429−4901,and HE 2243−6031) a 1. introduction local star-formationactivities in the full range of environments over cosmic time. Many single-cloud weak Mg II absorbers (Wr(2796) ≤ Narayananet al.(2007)presentedtheresultsofasearch 0.3˚A) at 0.4 < z < 1.0 have been found to have so- for weak Mg II absorption in VLT/UVES quasar spec- lar or even super-solar metallicities (Rigby et al. 2002; tra over a cumulative redshift path of 77.3 at 0.4 < z Charlton et al. 2003). This is quite puzzling in view of < 2.4. A total of 116 Mg II absorbers with 0.02 < the fact that they tend to be found at distances of 40 – W (2796) < 0.3 ˚A were detected, with ∼ 60% of these 100 kpc from star-forming galaxies (Churchill et al. 2005; r having just a single-cloud, narrow component in Mg II. Milutinović et al. 2006), and not near detected sites of It has been argued that a large fraction of the popu- star formation. These objects have a cross-section sim- lation of multiple-cloud weak Mg II absorbers (i.e., ab- ilar to the absorption cross section of luminous galax- sorbers with multiple-clouds, which are resolved by high- ies, thus they occupy a significant volume of the universe resolutional (R ∼ 40,000) spectroscopies) have a different (Rigby et al.2002). BecauseweakMgIIsystemsonlycon- physicaloriginfromthesingle-cloudweakMgIIabsorbers tain a small mass, they do not trace the cosmic metal (Rigby et al. 2002; Masiero et al. 2005; Ding et al. 2005). abundance, however, it is still important to understand Narayananet al. (2007) found that the redshift path den- 1 Based on archivedata of observations made withthe ESO Telescopes at the La Sillaor Paranal Observatories under programs, 65.O-0411, 66.A-0221,67.A-0280 2 DepartmentofAstronomy&Astrophysics,ThePennsylvaniaStateUniversity,UniversityPark,PA16802 1 2 Misawa et al. sity, dN/dz, of weak Mg II absorbers increases from z = km s−1Mpc−1, Ωm=0.3, and ΩΛ=0.7. 2.4 to z = 1.2, where it peaks, and subsequently declines until the present. 2. data The paper of Narayananet al. (2007) followed a weak All three weak MgII systems in our sample were found Mg II survey of Lynch et al. (2006), which used a small in the survey of Narayananet al. (2007) who studied the subset of high quality VLT/UVES quasar spectra. From statistical properties of weak Mg II systems, using 81 that survey, Lynch & Charlton (2007) selected the nine z VLT/UVES spectra. Data were retrieved from the ESO > 1.4 weak Mg II absorbers and applied photo-ionization archive, and reduced using the ESO MIDAS pipeline. In models. Of these, four were single-cloud weak Mg II ab- the caseofmultiple epochobservations,the individualex- sorberswithcoverageofthecorrespondingLyαtransitions posures were combined to enhance the S/N ratio. The so that metallicities could be constrained. detailed description of the original data analysis is pre- In general, Lynch & Charlton (2007) found that the sented in Narayananet al. (2007). physical conditions of the z > 1.4 weak Mg II absorbers The main purpose of this paper is to place strict con- were similar to those of the same class of absorbers at straints on physical properties (such as metallicity, ion- lowerredshift,suggestingthatthe sameproductionmech- ization parameter4, gas temperature, and size) of single anisms are responsible over this time period. However, in cloud, weak Mg II absorbers at high redshift. Among the contrasttothesituationatlowredshift(Rigby et al.2002; 116 weak Mg II systems in Narayananet al. (2007), 16 Charlton et al.2003),Lynch & Charlton(2007)wereonly systems at relatively high redshift (z > 1.5)have Lyα ab- able to derivelowerlimits to the metallicity,rangingfrom sorption lines covered with the observed spectra. Lyα is 1/100th – 1/10th of the solar value. Single-cloud weak quite important for determining metallicities. Of the 16 Mg II absorbers, in addition to the cloud that produced systems, 7 are classified as single-cloud systems. Four of the Mg II absorption, have separate, lower density C IV theseseven,thez=1.65146systemtowardHE0001−2340, clouds that are spread over tens to hundreds of km s−1 the z = 1.70849 system toward HE0151−4326, the z = (Charlton et al. 2003; Lynch & Charlton 2007). The rea- 1.79624systemtowardHE2347−4342,andthez=2.17455 son that only lower limits could be derived for the metal- system toward HE0940−1050 had already been mod- licityinthesecaseswasbecauseofthecontributionstothe eled as described in Lynch & Charlton (2007), since they LyαprofilefromthekinematicallyspreadCIVclouds. On were first identified in the earlier survey of Lynch et al. the other hand, Lynch & Charlton (2007) also suggested (2006). The other three systems, the z = 1.78169 sys- that there couldbe a build-upofmetals in the population tem toward HE 0141−3932, the z = 1.68079 system to- of single-cloud weak Mg II absorbers from z ∼ 2 to z ∼ ward HE 0429−4091,and the z = 1.75570 system toward 0.4. HE 2243−6031,are modeled in the present paper. In view of the large spread of metallicity constraints In Table 1, we give an observation log for the three at any redshift, a sample of only four z > 1.4 single-cloud quasars toward which the single-cloud weak Mg II ab- weakMgIIabsorberswasnotsufficienttoresolvetheissue sorbers were detected. Column (1) is the quasar name, of whether metallicity evolution occurs in the population. columns (2) and (3) are the coordinates of the quasars, Wethereforeidentifiedthreeadditionalz >1.4candidates columns (4) and (5) are the emission redshift and optical for photo-ionization modeling from the larger survey of magnitude of the quasars, column (6) is observed wave- Narayananet al. (2007). These systems, the z = 1.78169 length range, column (7) is the central wavelengthfor the absorbertowardHE0141−3932,thez =1.68079absorber blue and red CCDs of VLT/UVES, columns (8) and (9) toward HE 0429−4091,and the z = 1.75570 absorber to- are total exposure time and observing date, and column ward HE 2243−6031, have high S/N-ratio VLT/UVES (10) is the proposal ID. By chance, all three quasars were coverageofkeyconstrainingmetal-linetransitions,aswell observedaspartofprogramswiththesameP.I.,Sebastian as Lyα. We apply Cloudy (Ferland et al. 1998) photoion- Lopez. ization models to constrainthe ionizationparametersand The absorption profiles for key transitions used to con- metallicities of the phases3 of gas that are required to fit strainthe three single-cloudweakMgIIabsorbersaredis- the data. Our focus is on the question of the metallicity playedinFigures1,2,and3. Rest-frameequivalentwidths evolutionofthesingle-cloudweakMgIIabsorbers,andon and 5σ rest-frame equivalent width limits for these same comparing the result to the metallicity evolution of other transitions are listed in Table 2. classes of quasar absorption line systems. We will begin in § 2 with a summary of the data that 3. photoionization models we have used to constrain the properties of the three z Inthissectionwebrieflysummarizethestrategyforpho- ∼ 1.7 single-cloud weak MgII absorbers. Next, in § 3, we toionization modeling, which is similar to the procedure describeourprocedureforphoto-ionizationmodelingwith practiced in previous studies (e.g., Churchill & Charlton Cloudy,whichisbasedoncomparingsyntheticmodelpro- 1999; Charlton et al. 2003; Lynch & Charlton 2007). At files to the data. § 4 presents the detailed results for the first, we determine the line parameters (column den- three absorbers. Finally, in § 5, we summarize those results, compare the metallicities of single-cloudweak MgII sity, Doppler parameter, and redshift) for the Mg II doublets, using a Voigt profile fitting code (MINFIT; absorbersto DLAs, to sub-DLAs, andto Lymanlimit ab- Churchill et al. 2003). For each of the three absorbers sorbers, and discuss the implications of this comparison. modeled here, a single absorptioncomponent provided an Throughout this paper, we use a cosmology with H =72 0 adequate fit to the Mg II doublet. The best fit parame- 3 Phasesaredefinedasregionsofgaswithsimilarmetallicity,temperature, andvolumedensity. 4 Theionizationparameterisdefinedtheratioofionizingphotons (nγ)tothenumberdensityofhydrogenintheabsorbinggas(nH). Super-Solar Metallicity in Weak MgII Systems 3 ters are listed in Table 3: column (1) is the name of the dance pattern for the system. Our method is illustrated transition,columns(2),(3),and(4)arethefitparameters for a specific example, the v = 1 km s−1 cloud for the of velocity shift fromthe systemcentralredshift5, column z = 1.78169 system toward HE0141-3932 (System 1) in density, and Doppler parameter with their 1σ errors. For Figure 4. We show, for this component, that only values completeness, we also applied formal Voigt profile fits to within the range −0.7 < logZ < 0.1 and within 0.1 dex otherdetectedtransitions,andlistedtheresultsinTable3. of logU = −2.3 produce acceptable fits to Lyα, SiII, and We emphasize, however, that we did not use these values SiIV, which are the main constraints for this case. (except for the optimized transitions) to determine model Usually, low-ionization phase clouds that produce the constraints. Instead we compared directly to the shapes Mg II absorption lines also produce other low-ionization of the observed absorption profiles as described below. transitions such as Fe II, Si II, C II, and Al II. How- Assuming that the absorbingclouds arein photoioniza- ever, the higher-ionization transition, C IV, and some- tion equilibrium, their ionization conditions are derived times intermediate-ionization transitions such as SiIV are using the photoionization code Cloudy, version 07.02.00 not fully produced by the Mg II clouds. Therefore, once (Ferland et al. 1998), optimizing on the observed column we find the best model for the Mg II clouds, we repeat densityofMgIIandcomparingmodelpredictionstoother a similar Cloudy analysis for the C IV clouds. For the observed low ionization transitions. The clouds in each high-ionizationphase,multiplecomponentsaresometimes phase aremodeledasplane-parallelstructuresofconstant required to fit C IV absorption lines detected at differ- density. The ionization parameter (logU = log[n /n ]) ent velocities. The results of Voigt profile fits to these γ H andmetallicity(Z/Z⊙)ofthecloudsserveasfreeparame- C IV components are also listed in Table 3. Finally, the ters in the modeling. The elemental abundance pattern is combined models, with contributions from both low- and initially assumed to be solar, and variations are explored high-ionizationphases,arecomparedtotheobservedspec- when they are suggested by the data. trum. Since Lyα absorptionlines havecontributions from Following Haardt & Madau (1996, 2001), we take the both the low- and high-ionization phases, determination combined flux from quasars and star forming galaxies at ofmetallicitiesforthelow-andhigh-ionizationphasescan z ∼1.7(with aphotonescapefractionof0.1)asanextra- be degenerate. However, in the three single-cloud weak galacticbackgroundradiation(i.e.,incidentfluxontheab- Mg II absorbers that we are considering, it is possible to sorbers). We will consider the possible effects of a nearby placelimitsonmetallicityofbothphases,soasnottoover- stellar contribution to the ionizing radiationfield in § 4.4, produce the Lyα absorption. Also, in some cases, models althoughitshouldbesafetoignorethem,sinceweakMgII withhighmetallicity(leadingtolowtemperatures)canbe absorbers are not often within 40 kpc of bright galaxies favored when comparing model profiles to some observed (Rigby et al. 2002; Churchill et al. 2005). low ionization transitions. The metallicity and ionization parameter are initially varied in steps of 0.5 dex, and then fine-tuned to select 4. individual systems the model that corresponds best with the observed data For photoionization models, our procedure is adjusted in steps of 0.1 dex. The column densities for various ion- for each system based on the available constraints in var- izationstagesofeachelementandtheequilibriumgastem- ious transitions. Some transitions are blended with other perature are providedby the Cloudy model. The Doppler lines, and others may be saturated. These issues are con- parameter for each element can then be calculated from sideredcarefully during photoionizationmodeling. In this the expression for total line width, b2 = b2 + b2 , tot ther turb section, we present the spectra of our three weak Mg II where bther (= p2kT/m) is the thermal contribution to systems and then describe the results of photoionization the line width corresponding to a gas temperature T and models. We can place strong constraints on metallicity bturb, the contribution from internal gas turbulence. This andionizationparameterbecausemultipletransitions(in- latter quantity, which is uniform across elements, is esti- cludingLyα)havebeendetectedinhighS/N-ratioregions mated using the gas temperature from the Cloudy model ofthe spectra(seeFigures1 –3). We will, hereafter,refer and the observed Doppler parameter of the element on to the three weak Mg II systems as Systems 1, 2, and 3. which the model is optimized. From the derived column After presenting the constraints, we discuss in § 4.4 the densities andDopplerparameters,asynthetic spectrumis effect of varying model parameters and assumptions such generated and compared to the observed spectrum after as photoionization equilibrium, the shape of the incident convolving with a Gaussian instrumental spread function radiation field, and the abundance of dust. (R = 45,000 in the case of our VLT/UVES spectra). Although the observed and the synthesized spectra are 4.1. System 1 (HE 0141−3932; z=1.78169) compared “by eye”, the method generally gives good re- This system is detected in Lyα, Mg II, Si II, Al II, C II, sults inthe sense thatother models whosemetallicity and Al III, Si III, Si IV, and C IV, as shown in Figure 1. The ionization parameter are (sometimes only slightly) differ- observed spectrum covers Mg I, O I, and N V, but these ent than the best values would make the model spectrum transitions are not detected to a 5σ detection limit. Al- deviate greatly from the observed spectrum. It is not thoughFeIIλ2383isformallydetectedatonly3σandmay practical to apply formal procedures to assess the qual- be contaminated by spurious features, Fe II λ2600 is also ity of fit because of the effect of blends and data de- detected at 2.5σ and both are precisely aligned with the fects on a χ2 statistic. If all transitions of a certain el- Mg II. We therefore are confident in applying the Fe II as ementareover/under-producedcomparedtothe observed a constraint. Rest-frame equivalent widths and 5σ equiv- spectrum in the best model, we consider a specific abun- alentwidthlimits arelistedinTable 2. Althoughthe Lyα 5 Thesystemcentralredshiftisdefinedastheflux-weighted centeroftheMgIIλ2796absorptionline. 4 Misawa et al. profile is complex, it provides a very strong constraint on transitions except for Lyα. There are absorption features metallicity because the flux at the position of the compo- at both sides of the Lyα profile, at ∆v < −50 and at ∆v nent centered on Mg II begins to recover on its blue side, > 70 km s−1 (see Figure 1). Neither of them are Lyβ which enables us to fit this component of Lyα effectively. profiles from higher redshift systems, because the corre- Alltransitionsexceptforhigh-ionizationlinessuchasCIV sponding Lyα absorption lines would be located redward and SiIV have only single components detected. of the quasar Lyα emission line. The C IV component at We begin by fitting a Voigt profile to the Mg II dou- ∆v = −77 km s−1 cannot itself give rise to all of the ob- blet. The best fit parameters are listed in Table 3. A served Lyα around that velocity because it is too narrow. lowerlimittotheionizationparameterisprovidedbyFeII, Thus, we fit the broad Lyα profile (b(H) = 42 km s−1 as whichisover-producedatlogU <−3.8. MgIis alsoover- listed in Table 3), and placed upper limits on its metal- produced at logU < −6.0. (Mg I was not detected in licity: (logZ < 0.0 for logU ∼ −3 and logZ < −2.0 for the observed spectrum, and Fe II is only detected in the logU ∼ −1.5) so as not to produce other transitions at λ2383transitionatalevelof∼3 σ.) Thereis alsoastrict this velocity. Ionization parameter is not constrained for upper limit on the ionization parameter of logU < −2.8 this broad Lyα cloud, although high values of logU are above whichthe high-ionizationtransitions CIVand SiIV ruled out unless the metallicity is extremely small. are over-produced. Based onthese considerations,models We conclude that, for a solar abundance pattern, the with logU = −3.8 – −2.8 seem acceptable, however Si II metallicity of the z = 1.78169 weak Mg II system in the andAlIIprovideadditionalconstraints. Anionizationpa- HE0141-3932spectrumisconstrainedtobe−0.7 ∼< logZ rameter, −3.8 < logU < −3.7, is consistent with all con- ∼< −0.5. AstrictlowerlimitonmetallicityoflogZ >−0.7 straints. At lower values of logU Fe II is over-produced, appliesinorderthatLyαabsorptionisnotover-produced. and at higher values AlII and SiII are over-produced. In- Two of the three C IV clouds related to this system are dependent of ionization parameter, the metallicity must constrainedto havesimilar orhighermetallicities, andfor be logZ > −0.7 in order that Lyα is not over-produced. the other no significant metallicity constraint is available. An upper limit of logZ < −0.5 would also apply, for so- The MgII cloud has a sub-parsec size, much smaller than lar abundancepattern, if logU = −3.7,in orderthat FeII the three C IV phase absorbers, with sizes of >10 pc. All absorptionisnotover-produced. Thelowerlimitonmetal- components are optically thin, the MgII cloud giving rise licity oflogZ > −0.7applies for the solarabundance pat- to logN ∼ 15.9. HI tern,andwouldbeadjusteddownwardsforanα-enhanced Asummaryofhowspecifictransitionswereusedtocon- pattern. Similarly,logU couldbelowerintheα-enhanced strain logU and logZ for this system is given in Table 4. case. If dust depletion was important, the lower limit on Ranges of acceptable model parameters for each model metallicity would be increased. cloud are listed in Table 5, in which column (1) is the ab- In addition to the Mg II cloud, three more higher- sorption redshift, column (2) is the optimizing transition ionization clouds are needed to reproduce the three de- used in Cloudy, column (3) is the velocity shift from the tected C IV components, the two Si IV components, and system center, and column (4) is the acceptable range of the right wing of the Lyα profile. There are no N V lines ionization parameter. Column (5) of the table presents detected,whichisusefultoplaceanupperlimitontheion- the best constraint on metallicity, considering both com- ization parameter. The results of Voigt profile fits to the parisontotheobservedLyαandthefittometal-linetran- C IV doublet are given in Table 3. Beginning with these sitions, assuming a solar abundance pattern. Column (6) CIVfits,weadjusttheionizationparameterandmetallic- is a more conservative constrainton the metallicity of the ity to satisfy constraints from other transitions. In order cloud, using only the requirement that Lyα absorption is toavoidover-productionofSiIVandlowionizationtransi- not over-produced. Column (7) is the depth of absorber tions(atlowvalues)orNV(athighvalues),theionization along the line of sight, column (8) is the Doppler parame- parametersfortheadditional3cloudsat∆v=−77,1,and teroftheoptimizingtransition(listedincolumn(2)), and 31 kms−1 fromthe systemcenter shouldbe −1.5< logU column (9) is the column density of neutral hydrogen gas <−1.0,−2.35<logU <−2.25,and−1.9<logU <−1.8, in the absorbers. An example of a best fit model and its respectively. For these ionizationparameters, lower limits specific parameters is presented in Figure 1 and Table 6. on metallicity of logZ > −1.8, logZ > −0.7, and logZ > −0.5 will apply for the three CIV components in order 4.2. System 2 (HE 0429−4091; z=1.68079) that Lyα is not over-produced. For the first two components, at ∆v = −77 and 1 km s−1 upper limits on logZ ThisisaveryweakMgIIsystemwhoserest-frameequiv- of 0.5 and 0.1 in order that low ionization transitions are alent width (Wr(2796) = 0.015˚A) is just above the lower not over-produced. The third C IV component, at ∆v = limit of the survey presented in Narayananet al. (2007). 31 km s−1, has a strong constraint on its metallicity, as- In this system, Lyα, Si II, Al II, C II, Al III, Si III, Si IV, sumingthatitisresponsiblefortheLyαabsorptioninthe and C IV are clearly detected, but O I, Fe II, and N V are red wing of the profile (since the Mg II cloud cannot ac- not detected at 5σ, as shown in Figure 2. The rest-frame count for that absorption without over-producing Lyα in equivalent widths and 5σ limits are tabulated in Table 2. the blue wing). For that CIV we find that −0.5 ∼< logZ Again, we begin by fitting Voigt profiles to the Mg II ∼< −0.4 in order to match the Lyα profile. We empha- doublet (see Table 3). Low ionizationparameters,smaller size that the three components have different ionization than logU = −3.5, are not acceptable because of over- parameters from each other, although these values are all production of Fe II. At logU > −2.0, C IV is over- higher than that of the MgII cloud. produced. For this system, the Lyα absorption is strong, The one Mg II and three C IV clouds reproduced all and it probably arises from offset CIV components rather than fromthe MgII cloud. Because of the strong Lyα ab- Super-Solar Metallicity in Weak MgII Systems 5 sorption we cannot place a meaningful constraint on the high-ionizationphasecloudsat∆v of−47,−31,−12,and metallicity of the system; the blue wing of Lyα only re- 3 km s−1 from the system center. As with the maximal quires logZ > −3.0. However, at −3.5 < logU < −2.0, CIVmodel, the metallicity ofthe ∆v = −47kms−1 com- AlIIisover-producedatlowmetallicity(suchthatlogZ > ponent is constrained to be logZ = −1.0 – −0.9. We −1.0forlogU ∼−3.0andlogZ >−0.3forlogU ∼−2.0). found a range of acceptable ionization parameters for the Because the acceptable range of ionization parameter is first three C IV components, with logU = −1.9 – −1.7. large for this system, we consider two extreme cases: (i) However, there are no acceptable solutions for the fourth the Mg II cloud has the highest acceptable ionization pa- component(thatoverlapswiththeMgIIcomponent). The rameter (logU = −2.0) that produces the maximum pos- best model, with logZ = −0.5 and logU = −1.9, repro- sible C IV and Si IV absorption, and (ii) the Mg II cloud duces observations for most all transitions, but it gives has the lowest acceptable ionization parameter (logU = rise to additional Mg II absorption. These double contri- −3.5) so that it does not significantly contribute to the butions from low- and high-ionization phase clouds over- CIV andSiIVprofiles. We willrefer to these below as the produce the observed Mg II profile. This result suggests maximal and minimal CIV models. that the Mg II absorption is partially produced by the InthemaximalCIVmodel,weneedanadditionalthree same cloud as the C IV, and supports the maximal C IV high-ionization components to fit C IV profile at ∆v = model above. Thus, we conclude that the maximal C IV −47, −28, and 5 km s−1 from the system center. The fit- model with three CIV components is more appropriateto ting parameters, after removing the contributions to CIV describe this system. from the low-ionization clouds, are listed in Table 3. Fig- As with System 1, the right wing of the Lyα profile is ure 2 alsoshows that the MgIIcloudat 0 km s−1 is offset not produced either by the Mg II cloud or by the three and is not broad enough to explain all of the C IV ab- C IV clouds. Therefore, we placed an additional Lyα line sorption at a similar velocity. Only the C IV cloud at in this region of the profile. The unexplained absorption ∆v = −47 km s−1 has its metallicity constrained to be cannotbeLyβ becausethisregionisnotintheLyβ forest, logZ = −1.1 – −1.0, assuming that it accounts for the aswasthecasewithSystem1. Wefittheregionusingone blueward edge of the saturated Lyα profile. The metal- additional Lyα component with logN ∼ 14.6 and b ∼ HI licity cannot be lower than this value or Lyα would be 34 km s−1. This component is not well-constrained, but overproduced. However,aconstraintontheionizationpa- for logU > −2.0 the C IV and NV are over-producedun- rameters of the three C IV clouds can be obtained by the less logZ < −2.0. For logU < −2.5, the metallicity can requirement that they also fit the Si III and Si IV dou- beaslargeaslogZ =0.0beforelowionizationtransitions blet profile. For the three C IV clouds, we find logU in are over-produced. the rangebetween −1.8and−1.6. The only disagreement We conclude that the z = 1.68079 system in the between this maximal C IV model and the observed spec- HE 0429-4091 spectrum has three C IV components as trum is an over-production of N V. Adjusting the abun- well as one Mg II component. The transitions that pro- dance pattern so that nitrogen is reduced by at least 0.5 vided constraints onlogU and logZ based onour Cloudy dex compared to the other elements gives adequate re- models are listed in Table 4. All components have simi- sults (see Figure 2). In fact, the production mechanisms lar ionization parameters, logU ∼ −1.9. It is difficult to for nitrogen are poorly understood (e.g., Russel & Dopita constrain the metallicity of the low ionization phase for 1992, and references therein). Namely, it is not well un- this system since there are kinematically distributed CIV derstood yet whether massive stars or intermediate-mass components that contribute to the Lyα absorption, how- stars(orboth)contributetotheprimarynitrogenproduc- ever metal line absorption is best reproduced with about tion before the secondary production starts through the solar metallicity for the Mg II component. The blueward, CNO cycle (Spite et al. 2005). This nitrogen “problem” offset CIV cloud is constrainedby the Lyα profile to have has also been reported frequently in DLA systems (e.g., logZ > −1.1. Because of its higher ionization state, the Pettini et al. 2002) and for the Magellanic Bridge (e.g., size of the Mg II cloud of this system (∼ 100 pc) is two Lehner et al. 2001), as well as for multiple-cloud weak orders of magnitude larger than that of the Mg II cloud Mg II systems (Zonak et al. 2004). Once the best model in System 1. All components are optically thin. Ranges parameters for the three C IV components are found, we of acceptable model parameters are listed in Table 5. An adjust the parameters for MgII again to compensate, and example of the best models and its model parameters are get logU = −2.0 and logZ = −0.2 – 0.1. The lower limit presented in Figure 2 and Table 6. For the model curves on metallicity was placed in order not to over-produce showninFigure2,thenitrogenabundanceisdecreasedby Al II, and the upper limit in order not to under-produce 0.5 dex from the solar abundance. SiIVandCIV.Acceptable rangesofmodelparametersfor one Mg II and three C IV components (the maximal C IV 4.3. System 3 (HE 2243−6031; z=1.75570) model) are summarized in Table 5. An example of the best fit model is overplotted on the observed spectrum in This weak Mg II system has simple, single component Figure 2. Here, we emphasize that the Mg II component profiles detected in Lyα, the low-ionization lines, Mg II has a relativelyhigh ionizationparametersimilar to those and Si II, and in the higher-ionization lines, Al III, Si III, of the CIV components. Si IV and C IV. Fe II and Al II lines are not detected to We also consider a minimal C IV model, for which we a 5σ equivalent width limit. See Table 2 for the rest- need four more components to reproduce C IV and Si IV frame equivalent widths and 5σ limit values. The regions because there is no contribution to the C IV absorption of spectrum where O I, C II, and N V would appear are from the Mg II phase. We fit the C IV profile with four contaminated by blends, so that they only serve to provide weak upper limits. It is immediately apparent that 6 Misawa et al. the equivalent width of MgII is very large for this system thesolarpatternandthe siliconabundancewasdecreased compared to its relatively weak Lyα profile. This implies by 0.2 dex, logZ = 0.0 would be consistent with the data that the system has a high metallicity. Velocity plots of forlogU =−2.0. Bothsuper-solarmetallicitymodelsand various transitions are displayed in Figure 3. small changes in abundance patterns are reasonable ways As usual, we first optimize on the column density and to reconcile models with the data. However, we empha- Doppler parameter of Mg II obtained from Voigt profile sizethatthemetallicityconstraintoflogZ >0.0isdirectly fitting (see Table 3). A lower limit on the ionization pa- determined based on comparison with the Lyα profile. rameter of logU > −3.5 can be determined, since Fe II Next, we added a high-ionization phase because the is over-produced at smaller values. On the other hand, broadlineprofileofCIVcannotbereproducedbythelow- higher ionization conditions with logU > −2.0 are ruled ionizationMgIIcomponent(seethedottedlineinFigure4 out because of overproductionof SiIV. The observedCIV on the C IV profile). Si IV is also under-produced by an absorption is also over-producedif logU > −1.5. MgIIcomponentwithlogU =−2.0andlogZ =1.0. Inor- More stringent constraints on the ionization parameter dertofindconstraintsontheCIVphase,weoptimizedon canbefoundifweconsidertheobservedabsorptioninAlII the CIV column density from our Voigt profile fit. In this and SiII. For metallicities logZ ∼< 0.0, the constraintson case,thecontributiontotheCIVabsorptionfromthelow- ionization parameter are independent of metallicity. For ionization phase was minimal, so that removing it made larger metallicities, cooling leads to different constraints. no significant difference to the fit, listed in Table 3. From WewillfirstconsidertheconstraintsforlogZ <0.0. Con- Cloudy photoionization models, we derive a constraint of sidering Al II and Si II, we found that an ionization pa- logU =−1.5–−1.4onthe ionizationparameter,inorder rameter at the higher end of the range −3.5 < logU < to simultaneously fit the C IV and Si IV profiles. If the −2.0 produces the best agreement,but these ions are still metallicity of the MgII cloud is lower so that it gives rise overproduced by the model. To reconcile a logU = −2.0 to more Si IV absorption, then the ionization parameter model with the data for logZ < 0.0, we can reduce the of the CIV cloud would need to be higher to compensate. abundance of aluminum by 0.5 dex, and that of siliconby Similarly, if the ionizationparameter of the MgII cloud is 0.2 dex. Further variations in abundance pattern could lower, that of the C IV cloud would also need to be lower lead to different preferred values of logU. in order to produce more SiIV absorption. Independent of the constraint on the ionization param- Since the Lyαprofile does not havea contributionfrom eter, the Lyα profile can be used to place a strong lower kinematicallyseparateCIVcomponents,itcanbe usedto limit on the metallicity of the Mg II cloud. The Mg II placealowerlimitonthemetallicityinthehigh-ionization component is centered exactly on the Lyα profile, so that phase. InorderthatLyαisnotover-producedbytheCIV both the blue and the red side of the Lyα profile provide cloud, a metallicity of logZ > −0.2 is required. No firm the same lowerlimit, logZ > 0.0,inorderthatLyαis not upper limit canbe placedsince the Lyα absorptionwould overproduced. However, there are still some uncertainties be fully accounted for by the Mg II cloud if logZ = 0.0. on this metallicity constraint. On one hand, lower values However,iftheMgIIcloudhaslogZ =1.0inordertopro- would be permitted if the system is in α-enhanced con- ducetheobservedAlIIIabsorption,itdoesnotaccountfor dition because this constraint does assume a solar abun- all of the observed Lyα absorption. In this case, logZ = dancepattern. Ontheotherhand,therequiredmetallicity −0.1 – 0.0 for the C IV cloud would account for the ma- wouldbehigherifmagnesiumisdepletedontodustgrains, jority of the Lyα absorption. A cloud with logU = − 1.4 althoughwedo notexpectalargeamountofdustinweak and logZ = 0.0 has a size of ∼ 5 kpc. Mg II systems as discussed below. A cloud with logU = We conclude that the system can be reproduced with −2.0andlogZ =0.0hasatemperatureofT =9870Kand only two components: one low-ionization and one high- asizeof1.5kpc. AsmallerlogU,whichwouldbepossible ionization phase. The constraints on logU and logZ and with further reductions in the abundances of aluminum the transitions they were derived from are presented in and silicon, would lead to a smaller cloud size. Table 4 for two models. Model 1 assumes a solar abun- Metallicities greater than the solar value are also possi- dance pattern and relies on a super-solar metallicity for ble forthe MgIIcloud. Infact, asolarabundancepattern the low-ionization phase to fit the data. Model 2 allows canbeconsistentwiththedatabecausethecoolingathigh anadjustment ofthe abundances of aluminumand silicon metallicity leads to less Al III and SiIV production. If we relative to the solar pattern, but still requires at least a accept the constraint on ionization parameter of logU ∼ solar metallicity. Our favored two-phase model has a rel- −2.0, as discussed above, and assume a solar abundance atively high ionization parameter of logU = −2.0 for the pattern, we find that Si IV will be slightly over-produced MgIIcloud,andlogU =−1.4fortheCIVcloud. However, unless logZ > 0.4. It is over-producedon the blue side of with a different abundance pattern, the ionizationparam- the Si IV λ1403 profile. For logZ > 0.4, Si IV is under- eterforthe MgIIcloudcouldbe lower. Allcomponentsof produced, but in this case SiIV absorption can arise from acceptable models are optically thin in H i. Ranges of the a second, higher ionizationphase. Similarly,AlIIIis over- acceptablemodel parametersarelisted inTable 5, includ- produced unless logZ > 0.9. (The λ1863 transition is ingbothmodelswithlogZ =0.0andwithlogZ >0.9for overproduced; the λ1855 transition is contaminated by a the Mg II cloud. An example of best fit models and their blend.) Due to cooling at extreme metallicities, a logZ model parameters are presented in Figure 3 (only for the = 1.0 cloud would have a temperature of ∼ 500 K and model with logZ > 0.9) and Table 6 (for both). a size of ∼25 pc. These constraints are quite dependent on abundance pattern. As we noted above, if the abundance of aluminum was decreased by 0.5 dex relative to 4.4. Effect of Photoionization Modeling Assumptions Super-Solar Metallicity in Weak MgII Systems 7 In using the Cloudy code to derive constraints for the 5. discussion threesystems,wehaveassumedthatphotoionizationequi- Inthisstudy,weappliedphotoionizationmodelstocon- librium applies. This is almost certain to be the case for strain the physical properties of three single-cloud weak hydrogen since the photoionization timescale is short, on theorderof104years. Formetals,photoionizationequilib- Mg II absorption systems at z ∼ 1.7. Along with results presented in Lynch & Charlton (2007), these complete a rium might not apply if the gas has cooled from a higher sample of the seven z > 1.5 single-cloud weak Mg II ab- temperature, in which case the gas would still be more sorbers found in the VLT archive (Narayananet al. 2007) ionized than our calculations would suggest. This would for which metallicity constraintscould be derivedbecause clearly affect our model constraints. However, our most of simultaneous coverage of various metal lines and Lyα. important conclusion of high metallicities for weak Mg II We start this final section by summarizing the model absorber would only be stronger if a larger fraction of the constraintsderivedforthe threez ∼1.7single-cloudweak magnesium was in higher ionization states. MgIIabsorbersstudiedinthispaper(§5.1). Wethenpro- Another important assumption we applied in our mod- ceed to compare our results to the other four single-cloud eling was that of a solar abundance pattern, unless a de- weak Mg II systems at z ∼ 1.7 from Lynch & Charlton viationwasrequiredbythedata. We notethatdeviations (2007) in § 5.2. In this section, we also compare to lower from the solar pattern are to be expected, and may apply redshift single-cloud weak Mg II systems from the litera- whether the data require them or not. In other words, turetoinvestigateevolutionaltrends. Weaddressthehigh there is a degeneracy between the parameters logU and metallicityofthesesystemsinthecontextofDLAs(whose logZ and the abundance pattern. One common type of metallicities are usually smaller than the solar value by a abundancepatterndeviationwouldbeanα-enhancedcon- factor of 10 – 30) in § 5.3, sub-DLAs in § 5.4, and strong dition. In this case, we tend to over-estimate metallicity and multiple-cloud weak Mg II systems (the strong ones because our metallicity estimations are based on magne- are usually associated with bright (L > 0.05L∗) galaxies sium,anα-element. Anothersourceofthedeviationisde- within 40h−1 kpc; e.g., Bergeron& Boissé 1991) in § 5.5, pletion onto dust grains. We do not really expect a large respectively. We finally discuss a possible correlation be- amount of dust in weak MgII absorbers because they are tween metallicity and total hydrogen column density in not typically close to sites with large N or with current HI § 5.6. star formation. However, it is important to note that if magnesiumisdepletedbysomefactor,ourinferredmetal- 5.1. Summary of Results licity would increase by roughly that same factor. Finally, we consider the effect of changing the shape of System 1 can be fit with a single low-ionization Mg II the ionizing radiation field. For the previous calculations cloud and three higher ionization clouds (two of them off- wehadassumedaHaardtandMadaumodelforthe ioniz- setinvelocityfromtheMgII)thatgiverisetoCIVabsorp- ingradiationfromquasarsandstarforminggalaxies,with tion. ThelowionizationcloudhaslogZ >−0.8andtwoof an escape fractionof 0.1of ionizing photons fromgalaxies the high ionization clouds must have metallicities at least (Haardt & Madau 1996, 2001). We have also considered this high as well. Two offset low-metallicity clouds, pro- the opposite extreme, in which only quasars contribute ducing only Lyα absorption appear to be clustered with to the ionizing radiation field. For a given ionization pa- this system. rameter, the absence of a stellar contribution leads to an System2isalsofitwithoneMgIIandthreeCIVclouds, increase in the relative number of high energy photons. though all four clouds have similar, relatively high ioniza- With that change of spectral shape, we found negligible tion parameters (logU ∼ −1.9). The metallicity of the difference in the properties derived for the low ionization Mg II cannot be well-constrained directly from compari- phase. For the high ionization phase, we found that a sontothe Lyαbecausethe kinematicalspreadofthe CIV lower ionization parameter (e.g., by about 0.5 dex) would lines give rise to a large Lyα equivalent width, however beneededtofitthedata. Duringthephotoionizationmod- near solar metallicities are possible, and even favored for eling presentedabove,weneglectedpossible contributions a solarabundance pattern. Also, the bluewardCIV cloud from nearby stellar sources. However, we see only small will over-produce absorption in Lyα unless its metallicity differences in the column densities of low/intermediate- is logZ > −1.1. A separate,offset low-metallicitycompo- ionization phase gas (e.g., SiII, SiIII, and SiIV) by a fac- nent was again needed to fit the red portion of the Lyα tor of 2 or 3, even after adding stellar radiation from an profile. O7 star with an effective temperature of T = 38000 K System 3 has the best metallicity constraint among the eff (model# C3 of Schaerer & de Koter 1997). We adopt sevensingle-cloudweakMgIIabsorbersatz>1.5because an ionizing photon number density 10 times greater than of the absence of offset C IV components, thereby keep- that ofthe extragalacticbackgroundradiation,asa maxi- ing the Lyα profile very narrow. Two phases are still re- mum flux model. Thus, contributions from nearby stellar quired,with clouds of differentionizationparameterscen- sources, even if they exist, have no significant qualitative teredatthesamevelocitythatproduceboththenarrower effect on our conclusions. low-ionizationtransitions,andthebroaderhigh-ionization We conclude that none of the assumption behind our transitions. Both the low- and the high-ionization phases photoionization models have a qualitative impact on our are constrained to have solar or super-solar metallicities. conclusions. In particular, there is no effect that works The most striking thing about these model results, is against our conclusion of a very high metallicity for Sys- that in all three z ∼ 1.7 systems, the metallicity of at tem 3. least one phase of gas is constrained to be greater than one tenth the solar value. The three systems all have at least two phases of gas: the low ionization phase that 8 Misawa et al. arises in a layer of gas ∼1 – 100 pc thick with a density rived for the z = 1.7557 absorber in this paper is high abs of 0.001 – 0.1 cm−3, and the high ionization phase that becausethissystemisararecasethatdoesnothaveoffset comes from a larger region (∼ 0.1 – 10 kpc) with a lower C IV absorption that contributes to the Lyα absorption. density. For these systems, even metallicities of the larger Inotherwaysthereisnoreasontothinkitisunusual. The high-ionization phase regions range from one tenth solar sameappliedforthe lowredshiftweakMgIIsystemswith uptosolar. Wewillreturntoadiscussionofthesurprising the highest metallicity constraints (Charlton et al. 2003). issue ofhowregionswithlogNHI < 15(i.e., overa million We thereforeknowthatsingle-cloudweakMgIIabsorbers times less than the threshold for star formation) can be at both low and high redshift definitely have metallicities enriched in metals to the solar value. ofatleast1/10thsolar,butthatsomeandprobablymany have solar or even supersolar metallicities. Thus even before z = 1.5 star formation must have 5.2. Comparison to Other Single-Cloud Weak MgII polluted certain environments with metal-rich gas. Since Systems Lynch & Charlton(2007) could not constrainanyz > 1.5 The properties of the three z ∼ 1.7 single-cloud weak single-cloud weak Mg II absorbers to have close to solar Mg II systems, derived from the observed profiles, are metallicity, and since severallower redshift solar metallic- comparedto those of23other single-cloudweakMgIIab- itycaseswereknown,theytentativelysuggestedthatthere sorbersinFigure5,bothatsimilarandatlowerredshifts. might be a gradual build-up of metals in the single-cloud Figure5showsthatthereisnosignificantevolutioninthe weak Mg II absorber population. In this paper, we have observed Mg II λ2796 profiles. Figure 6 presents quan- shownthere is atleast one strong counter-example,i.e. of tities derived from photoionization models, for our three z >1.5solarmetallicitycases,andthereforethereappears absorbers and for others taken from the literature, using to be no metallicity evolution in the population. This is thesamemethodsaswehaveusedhere(Rigby et al.2002; nottosaythatthepopulationhasanarrowrangeofphys- Charlton et al. 2003; Ding et al. 2005; Lynch & Charlton ical properties: there is a large spread of cloud densities 2007). The specific constraints that are displayed in Fig- and metallicities at all redshifts (see Figure 6). However, ures 5 and 6 are also listed in Table 7, with the relevant it does suggestthat commonprocessesareat workto cre- references. ate the metals, and that the metals are mixed into fairly Lynch & Charlton(2007)derivedsimilarconstraintsfor similar surrounding environments. the phase structure of four different single-cloud weak The existence of high metallicity (greater than solar) Mg II absorbers at z ∼ 1.7. They also placed constraints compact intergalactic clouds (∼ 100 pc) has also been on the metallicities of those absorbers, of logZ > −1.5, demonstratedby Schaye et al.(2007), who derivedrobust > −1.5, > −1.0, and > −2.0 for the four different low- lowerlimitsonmetallicitiesofCIVabsorbersusingsimilar ionization phases. They note that the metallicities could photoionizationmodelingtechniques. Schaye et al.(2007) be significantly higher than these lower limits, however, find that the number density of this population of high these absorbers (by chance) tended to have larger contri- metallicity C IV absorbers is of the same order as that butions to the LyαabsorptionfromoffsetCIVcloudsand of single-cloud weak Mg II absorbers, and suggests that from Lyα-only clouds. the former might evolve from the latter as material ex- Results from modeling lower redshift (0.4 < z < 1.4) pands. More observations allowing direct connections be- single-cloudweakMgIIsystemsalsotypicallyyieldatwo- tween these populations will be of great interest. In fact, phase structure, with similar densities. Metallicity con- two of our three z > 1.5 weak Mg II systems have C IV straints are again often limited because of the difficulty phases with metallicities comparable to those derived by in separating the contributions of the two phases. How- Schaye et al. (2007) for the high metallicity C IV cloud ever, of the 11 cases of single-cloud weak Mg II absorbers population. for which some metallicity constraint could be obtained, 2 cases require the metallicity of the low-ionization phase 5.3. Comparison to Damped Lyα Systems to be be solar or even super-solar (Charlton et al. 2003). Furthermore, a total of 7/11 of the low redshift cases re- Damped Lyα systems (DLAs), whose neutral hydro- quiredametallicityofatleastonetenthofthesolarvalue. gen column densities are greater than logN = 20.3, HI We note that the metallicities are often quite likely to be dominate the neutral gas in the Universe at high red- higher than these strict lower limits because they are de- shift(e.g.,Lanzetta et al.1995;Storrie-Lombardi& Wolfe rived assuming that none of the Lyα absorption comes 2000;Péroux et al.2003). Byidentifyingmetalabsorption from the high-ionization clouds. lines that correspondto high redshift DLAs, it is possible We conclude that some fraction of weak Mg II systems to constrain the global metal abundances and trace the havebeendemonstratedtohavesolarorsuper-solarmetal- evolution of metallicity in neutral gas from z = 5 to 0 licity, at low redshift (2 systems at z = 0.8181 and (e.g., Pettini et al. 1994; Prochaska & Wolfe 2000). abs 0.9056; Charlton et al. 2003) and even at z ∼ 1.7 (1 sys- The metallicity evolution, measured from DLAs, how- tematz =1.7557;thispaper),whichcorrespondstoan ever, has two big problems: (i) the global mean metal- abs age of the Universe of 3.7 Gyr. Values constrained to be licity is less than 1/6th of the solar abundance at z > greaterthan1/10ththesolarvalueareevenmorecommon. 1 (e.g., Pettini et al. 1999; Prochaska & Wolfe 2000), and Furthermore,itislikelythatinmanycasesforwhichlower is still less than the solar by a factor of > 4 if the limits on metallicity have been derived the actual value is metallicity redshift relationship is extrapolated down to much higher. The metallicity constraint that we have de- z ∼ 0 using a least-squares linear fit (Prochaska et al. 6 At z < 1, there are only a few cases of metallicity measurements in DLAs using high quality spectra, and in these cases the metallicity is evaluated tobe∼1/10thofthesolarvalue(delaVargaetal.2000;Pettinietal.2000) Super-Solar Metallicity in Weak MgII Systems 9 2003; Kulkarni et al. 2005; Péroux et al. 2006b) 6, and are not the same type of regions that would produce high (ii) the metallicity is also much smaller than the theo- metallicity DLA absorption. retically expected value from the star forming activity Relating to the idea that DLAs sample low metallic- at higher redshift (e.g., Madau et al. 1998; Pettini et al. ity parts of galaxies, the weak Mg II absorbers, despite 1999; Wolfe et al. 2003). This discrepancy (i.e., “missing- their low column densities, must somehow sample higher metals problem”) remains an unresolved problem. metallicity regions which are not closely related to lumi- There are at least two possible classes of ideas pro- nous galaxies. This providesclear evidence for metallicity posed for the origin of this discrepancy: (i) a large inhomogeneities, which because of the size constraints for amount of dust in metal-rich DLAs obscures background single-cloudweakMgIIabsorbers,areonverysmallscales quasars, which prevent us from detecting DLAs with (parsecs to hundreds of parsecs). higher metallicities (e.g., Boisse et al. 1998; Fall & Pei 1993; Vladilo & Péroux 2005), and (ii) the DLA region 5.4. Comparison to sub-DLA Absorbers is sampling a lower metallicity than is found in the star-forming parts of the galaxy (Ellison et al. 2005a; Sub-DLA absorbers, with logNHI = 19.0 – 20.3, have Wolfe et al. 2003). beenfoundtohavesystematicallyhighermetallicitiesthan Thedustobscurationscenarioissupportedbythesmall DLAs with the NHI-weighted value larger by 0.5 – 0.8 abundances of depleted elements such as Cr and Fe, rela- dex, at 0.6 < z < 3.2 (Kulkarni et al. 2007). Several sub- tive toundepleted elementslike Zn,inDLAsystems(e.g., DLA systems (super-LLSs) at 0.5 < z < 2.0 have been Pettini et al. 1997; Khare et al. 2004). Vladilo & Péroux found with metallicities that aresolaror super-solar(e.g., (2005) estimated that ∼30 % to 50% DLAs are missed as Péroux et al. 2006a; Prochaska et al. 2006). Even small a consequence of obscuration. In absorbers expected to numbers of such high metallicity systems contribute sig- have a lower dust abundance, sub-DLA systems (super- nificantly to the total metal mass density at high red- LLSs) withlogN = 19.0–20.3,a highermetallicity has shift (Prochaska et al. 2006; Kulkarni et al. 2007). The HI beenmeasured(e.g.,Péroux et al.2006a;Prochaska et al. Kulkarni et al. (2007) mean values of metallicity for sub- 2006; Pettini et al. 2000; Kulkarni et al. 2007). Although DLAs,determinedfrom[Zn/H],arereproducedinourFig- Ellison et al. (2001) did not see any significant differences ure 7 from their Table 1. Although there is a range of inDLAstowardradio-selectedquasars,comparedtothose metallicities at each redshift, there is a clear increase in toward optically selected quasars at z > 2, the dust- the metallicity of sub-DLAs from hzi = 1.5 to hzi = 0.9. depletion effectis expectedto be moreimportantatlower This is consistent with the global star formation history redshift. inthe universeoverthis period,andthemetallicity values Alternatively, Ellison et al. (2005a) and Chen et al. aremuchmoreinlinewithexpectationsfromchemicalevo- (2005) suggest that there could be an metallicity gradi- lution models than those for DLAs (Kulkarni et al. 2007; ent as a function of a distance from the galactic cen- Somerville et al. 2001). ter, with DLAs produced at a larger impact parame- Somesingle-cloudweakMgIIabsorbers,withlogNHI ∼ ter than the stellar emission. DLAs that are produced 15 – 16, also have solar or super-solar metallicities, both in the host galaxies of gamma-ray bursts tend to have at hzi = 0.9 and at hzi = 1.5. There is considerable over- higher metallicities (e.g., Djorgovski et al. 2004), which lap between the metallicity constraints for sub-DLAs and could be more representative of typical star-forming re- single-cloudweakMgIIabsorbersatlowandintermediate gions. Also, small regions with DLA column densities redshifts. However, it is important to note that many of couldbeejectedfromgalaxiesbysuperwindsorasapartof the weak Mg II metallicities are only lower limits, and it AGN outflows (Mac Low & Ferrara 1999; Hamann et al. is quite likely that the values are higher at least in some 1997; Gabel et al. 2006; Bouché et al. 2006). cases. The reason that we cannot derive higher metallic- Figure 7 summarizes the metallicities derived for DLAs ities for those systems could simply be the separate C IV at z = 0.4 to 4.8 from Prochaska et al. (2003) and cloudsthathappentooverlapinvelocity. Itisalsoimpor- Kulkarni et al. (2005), and compares to the values for tantto note that for z ∼ 1.7,the redshiftof the absorbers single-cloud weak Mg II absorbers. It is striking that we have studied in this paper, there are not many sub- weak Mg II absorbers, despite having drastically smaller DLA systems with super-solar metallicity, but all single- NHI values, have considerably higher metallicities than cloud weak Mg II absorbers potentially could have solar- the DLAs, both at low redshift and at z ∼ 1.7. or super-solarmetallicity because we can place only lower Becausethey arenotnear regionswith currentstar for- limits on metallicity. Although two super-solar metallic- mation, the dust-to-gas ratio in single-cloud weak Mg II itysub-DLAswereinvestigatedbyProchaska et al.(2006), absorbers is expected to be very small. Evenfor DLAs, it theywereselectedforstudyfromamuchlargersamplebe- has been demonstrated that there is no bias against ob- causeoftheirextremelystrongZnIIabsorption. However, serving high metallicities due to dust obscuration of their muchlargerdatasamplesofsub-DLAsandweakMgIIsys- background quasars Ellison et al. (2001, 2005b). Even if tems will be requiredbefore concluding whether there are thedust-to-gasratioissimilartothoseinDLAs,theirtotal anydifferencesofmetallicitybetweenthesetwocategories. dustcolumndensityshouldbemuchsmallerthanthosein Another apparent difference between the metallicities DLAs, because of their lower gas column densities. Thus, of single-cloud weak Mg II absorbers and sub-DLAs, is they wouldnotbe subjectto the metallicity bias thatwas that there is no clear increase in the metallicity of the more likely to have affected DLAs. Thus the weak Mg II single-cloud weak Mg II absorbers over the range 0.4 < z absorbers provide an opportunity to see some high metal- < 1.7. This type of absorption is apparently produced by licity regionsof the universe at high redshift, though they the same types of regions presentover this redshift range, thoughtheyareknowntobelesscommonatlowredshifts 10 Misawa et al. and at z > 2 (Narayanan et al. 2005, 2007; Lynch et al. muchlowervaluesfortheLyαforestoverlapwiththeDLA 2006; Lynch & Charlton 2007). metallicities at z >2.5, but are smaller than those for low The sample size of weak Mg II systems is now much redshift DLAs. smaller than those for DLAs and Sub-DLAs, and only a The fact that the high metallicity population of single- few cases have been confirmed to have super-solar metal- cloud weak Mg II absorbers represents only a small frac- licities. It is quite important to increase the sample size tion of all Lyα forest clouds does not diminish their sig- of weakMgIIabsorbers,particularlyto include other sys- nificance. The single-cloud weak Mg II absorbers at z ∼ tems with only one C IV component such that a strong 1 would account for 25 – 100% of the Lyα forest clouds constraint on metallicity can be derived. At z < 1.5 this with 15.8 < logN < 16.8 (Rigby et al. 2002). More im- HI will involve high resolution observations with HST/COS portantly, the cross-sectionof single-cloud weak MgII ab- or HST/STIS of quasars for which high-resolutionoptical sorbers in the plane of the sky is comparable to that of spectra are also available. galaxiesatz ∼ 1, consideringthe regionsof galaxies ∼<30 –40kpcthatproduceLymanlimitabsorption. Thusthere 5.5. Comparison to Strong MgII and Multiple-Cloud aresignificantregionsoftheuniversecoveredbythesemys- Weak MgII Absorbers terious near-solar or super-solar metallicity objects with small logN . IntermediateinlogNHI betweensingle-cloudweakMgII As descriHbIed above, one explanation proposed for the absorption and sub-DLA’s are Lyman limit and partial larger metallicities for sub-DLAs than for DLAs is the Lyman limit systems, with logN = 16 – 19. These HI dust obscuration of quasars that have the highest metal- populations loosely correspond to the bulk of the strong licity DLAs in their foregrounds. Although there may be Mg II absorber population, with Wr(2796) > 0.3 ˚A, and somedebateaboutwhetherthis is aplausibleexplanation to multiple-cloud weak Mg II absorbers. Fewer of these of that trend, it is not likely to explain the continued in- systems have been studied in detail to date, but it is still possible to derive good constraints on metallicities creaseinmetallicity towardthe lowestlogNHI weakMgII absorbers. There is not likely to be a largeselectioneffect in many cases,using photoionizationmodeling techniques due to dust obscurationfor eventhe sub-DLAsystems. It the same as those employed here. The ten cases of strong seems more plausible that different types of absorbers are and multiple-cloud weak MgII absorbers at 0.7 < z < 1.9 samplingdifferenttypesofregionsinandaroundgalaxies, (Masiero et al.2005;Zonak et al.2004;Lynch & Charlton which can have large variations in metallicities even on 2007; Prochaska & Burles 1999; Ding et al. 2003), shown small scales. Different types of galaxies, e.g. star-forming on Figure 7, have metallicities ranging from −1.5 < logZ versus quiescent, will have different fractions of area cov- < 0.6. These are consistent with the ranges for both sub- eredbyregionsofthedifferenttypes. Bycomparingtothe DLAs and single-cloud weak MgII absorbers, but tend to mass-metallicity relation seen in star-forming galaxies at be higher than those for DLAs. Without larger samples, local universe (Tremonti et al. 2004) and higher redshift we cannot distinguish differences or evolutionary trends, (Savaglio et al. 2005; Erb et al. 2006), York et al. (2006) but again we note that many of these are measurements and Khare et al. (2007) proposed that the DLAs are as- aobfstohrebemrsettahlelicloitwy,mwehtiallelicfoitrytvhaeluseisngalree-calolluldowweeralkimMitgs.II sociated with low-mass (< 109 M⊙) galaxies, while sub- DLAsandweakerLLSsareprobablythesystemsthatarise in massive spiral/elliptical galaxies. 5.6. Metallicity vs. N HI The question remains: how do the single-cloud weak In summary, Figure 7 shows that it is the lowest NHI Mg II absorbers develop such high metallicities even at systems, that appear to have the highest metallicities, redshifts as high as z ∼ 1.7? We know that the lines of bothatz <1andatz ∼1.7. Thedataareevenconsistent sight that pass through these objects have N five or HI withagradualincreaseinmetallicitywithdecreasingNHI, six orders of magnitude below the star formation thresh- though there is a huge spread at any given value. Such a old. We also know that most single-cloud weak Mg II trendhasalreadybeenpointedoutusingcompiledsamples absorbers are not located very close to luminous galaxies, ofDLAsandsub-DLAs(Boisse et al.1998;Akerman et al. but that they tend to be in the vicinity (Churchill et al. 2005; Meiring et al. 2006; Khare et al. 2007), extending 2005; Milutinović et al. 2006). Outflowing gas from dwarf downtosub-DLAswithlogNHI∼19(Péroux et al.2003). galaxies in the local universe (e.g., Stocke et al. 2004) York et al.(2006)also suggestedthat a similartrend con- or from Lyman break galaxies at higher redshift (e.g., tinued to strong MgII systems with smaller hydrogencol- Adelberger et al. 2005) could partially contribute to the umn densities. high metallicities in single-cloud weak Mg II systems. The individual systems plotted on Figure 7, however, Schaye et al. (2007) have argued that high metallicity in- are only the systems with detected Mg II absorption. In tergalactic CIV clouds may be ejected as small clumps in particular, for single-cloud weak Mg II absorbers we are superwinds from star-forming galaxies. However, at least sure to be missing objects with metallicities significantly some of the very high metallicity single-cloud weak Mg II less than those of the systems that we do detect. In fact, absorbersshow signs of FeII line detections (e.g. our Sys- those lowermetallicity objects arepartofthe muchlarger tem 1 and severalsystems in Rigby et al. 2002), although Lyα forest population. Figure 7 also shows the typical sometimesitisdetectedwithonlyafewσ level. Forthose metallicities of Lyαforestclouds, both those with logNHI systems, it has been stated that they are not α-enhanced, >14.5,whichhavelogZ ∼< −2.0(Songaila & Cowie1996; and that ”in situ” star formation, including the less ener- Cowie et al.1995;Tytler et al.1995),andthosewithlower geticTypeIaSNethatincreasetheirontomagnesiumra- column densities, which have a mean metallicity of logZ tio,isresponsibleforenrichingthegas(Rigby et al.2002). ∼< −3.2 (Cowie & Songaila 1998; Lu et al. 1998). These