Draftversion February2,2008 PreprinttypesetusingLATEXstyleemulateapjv.6/22/04 AN OVI BARYON CENSUS OF THE LOW-z WARM-HOT INTERGALACTIC MEDIUM Charles W. Danforth & J. Michael Shull1 CASA,DepartmentofAstrophysicalandPlanetarySciences,UniversityofColorado,389-UCB,Boulder,CO80309; [email protected], [email protected] Draft versionFebruary 2, 2008 ABSTRACT Intergalactic absorbers along lines of sight to distant quasars are a powerful diagnostic for the evolutionand content of the intergalactic medium (IGM). In this study, we use the FUSE satellite to search 129 known Lyα absorption systems at z <0.15 toward 31 AGN for corresponding absorption 5 from higher Lyman lines and the important metal ions OVI and CIII. We detect OVI in 40 systems 0 over a smaller range of column density (logNOVI = 13.0–14.35) than seen in H I (logNHI = 13.0– 0 16.0). The co-existence of O VI and H I suggests a multiphase IGM, with warm photoionized and 2 hot ionized components. With improved O VI detection statistics, we find a steep distribution in O VI column density, dN /dN ∝ N−2.2±0.1, suggesting that numerous weak O VI absorbers n OVI OVI OVI a contain baryonic mass comparable to the rare strong absorbers. Down to 30 m˚A equivalent width J (OVIλ1032)wefindanabsorberfrequencydN /dz ≈17±3. Thetotalcosmologicalmassfraction OVI 4 in this hot gas is at least ΩWHIM = (0.0022±0.0003)[h70(ZO/0.1Z⊙)(fOVI/0.2)]−1 where we have scaled to fiducial values of oxygen metallicity, OVI ionization fraction, and Hubble constant. Gas in 1 the WHIM at 105−6 K contributes at least 4.8±0.9% of the total baryonic mass at z < 0.15. We v thencombineempiricalscalingrelationsfortheobserved“multiphaseratio”,N /N ∝N 0.9±0.1, HI OVI HI 4 and for hydrogen overdensity in cosmological simulations, N ∝ δ0.7, with the H I photoionization HI H 5 correctiontoderivethemeanoxygenmetallicity,ZO ≈(0.09Z⊙)(fOVI/0.2)−1 inthelow-z multiphase 0 gas. Given the spread in the empirical relations and in f , the baryon content in the OVI WHIM OVI 1 couldbeaslargeas10%. OursurveyisbasedonalargeimprovementinthenumberofOVIabsorbers 0 (40 vs. 10) and total redshift pathlength (∆z ≈2.2 vs. ∆z ≈0.5) compared to earlier surveys. 5 0 Subject headings: cosmological parameters—cosmology: observations—intergalactic medium— / quasars: absorption lines h p - 1. INTRODUCTION pathlength ∆z ≈ 0.5 (Tripp, Savage, & Jenkins 2000; o Savage et al. 2002). For the portion of WHIM contain- r One of the great,unanticipated legaciesof the Far Ul- t ing O VI, the estimated baryon fractions were ∼5% s traviolet Spectroscopic Explorer(FUSE)missionissurely a the detection of O VI absorption lines along extragalac- for O/H metallicity equal to 10% of the former solar : value, 7.41×10−4 (Grevesse et al. 1996). Because the v tic sight lines from the warm-hot intergalactic medium Xi c(WalHsiImMu)laatitonTs (=Ce1n05&−7OKst,rikaesrp1r9e9d9ic;tDedavb´eyetcoasl.m2o0l0o1g)i-. saotla(rOo/xHy)g⊙en=ab4.u9n0d×an1c0e−h4as(ArlelceenndtelyPbrieeetno erte-aml.ea2s0u0r1e)d, we have re-scaled our baryon estimates to this lower r Hot gas in the intergalactic medium (IGM) is produced a value. These estimates have large uncertainties, includ- by shocks generated by gravitational instability during ing untested assumptions of uniform metallicity, O VI the formation of large-scale structure, and its detection ionization equilibrium, and multiphase IGM structure. in highly ionized oxygen is an indicator of widespread The initial O VI searches reported only a handful of metal transport into the IGM through feedback from galaxy formation. The O VI absorption probes gas at absorbers – four systems by Tripp, Savage, & Jenkins 105−6 K, somewhat cooler than the bulk of the WHIM, (2000) and six by Savage et al. (2002). At low redshift, which has been shock-heated to temperatures of several down to 50 m˚A equivalent width in O VI λ1032, the ×106 K. This latter, very hot gas is only detectable number of O VI absorbers per unit redshift is 10–15% through weak X-ray absorption lines from higher ions of that found in the H I Lyα surveys: dNOVI/dz ≈ (e.g., O VII, O VIII, Ne VIII, N VI, N VII) as described 14+−96 (Savage et al. 2002) versus dNHI/dz ≈ 112 ± recently by Nicastro et al. (2005). 9 (Penton, Stocke, & Shull 2004). In this study, we The search for the WHIM gas has now begun in dramatically increase the total surveyed pathlength to earnest, using sensitive UV resonance absorption lines, ∆z > 2 and the total number of O VI absorbers to primarilytheOVIdoublet(1031.926,1037.617˚A)which 40. We also study the relationship between absorbers is ∼100 times more sensitive than the X-ray transi- from the WHIM and the warm neutral medium (WNM; tions of O VII (21.602 ˚A) or O VIII (18.97 ˚A). Thus, 103.5−4.5 K). With the increased number of absorbers, the initial WHIM detections of O VI were made by we can begin to look at WHIM absorption in a statisti- FUSE at z ≤ 0.15 and by the Hubble Space Telescope calmanner,particularlythe distributioninOVIcolumn (HST) at z ≥ 0.12, but over a modest total redshift density. 1 alsoatJILA,UniversityofColoradoandNationalInstituteof Standards andTechnology 2. OBSERVATIONSANDDATAANALYSIS 2 Fig. 1.— Comparison of HI and OVI column densities in 124 known Lyα absorbers. The left panel shows a mild correlation, if any, between NHI and NOVI. While NHI varies over a factor of nearly 1000, NOVI varies by a factor of only 20. The right panel shows the “multiphase ratio”, NHI/NOVI, with a dispersion of less than one dex over nearly three orders of magnitude in NHI, suggesting that photoionized H I and warm-hot gas occupy different phases of the IGM. The multiphase ratio is well fitted by a power law, NHI/NOVI =2.5×[NHI/1014 cm−2]0.9±0.1 (dashedline). NHI canberelatedtotheoverdensity,δH ≡ρ/hρi,asdiscussedin§4andshownalongthe topaxis. We began our survey with published lists of 3. THE MULTIPHASEIGM intervening Lyα absorption systems toward low-z The O VI lines are ideal tracers (at 105.5±0.3 K) for AGN obtained from GHRS and STIS surveys by the warm-hot ionized medium (WHIM; 105 − 107 K), Penton, Shull, & Stocke(2000); Penton, Stocke, & Shull while HI traces the photoionized warm neutral medium (2004). Other sight lines were covered by literature (WNM; 103.5−4.5 K). The left panel of Figure 1 shows a sources or measured at the University of Colorado from weak correlation, if any, between N and N . While OVI HI STIS/E140MspectraasdiscussedintheupcomingDan- N variesovernearlythreeordersofmagnitude,N is HI OVI forth,Shull, &Rosenberg(2005,hereafterPaperII). We detected between 1013 cm−2 and a few times 1014 cm−2 disregarded any weak Lyα absorbers (Wλ < 80 m˚A, (afactorof∼20). Togaugetherelativeamountsofwarm logNHI .13.2) and searched the FUSE data for higher- photoionized gas (H I) and hot collisionally ionized gas orderLymanlines plus OVIand CIIIcounterparts. We (OVI), we use the “multiphase ratio”, N /N . This HI OVI have been conservative in our identification of OVI sys- ratio was defined previously (Shull 2003) as a means of tems,requiringunambiguous(≥4σ)features,andexam- assessing the range in contributions from photoionized ining multiple channels of FUSE data and both lines of gas(HI)andcollisionallyionizedgas(OVI),whichoften the OVI doublet when possible. In all, we analyzed 171 appear to be associated kinematically (Sembach et al. absorbersin31FUSEsightlines. Wemeasuredallavail- 2003; Collins, Shull, & Giroux 2004). able Lyman series lines to determine accurate NHI and AsshownintherightpanelofFigure1,theNHI/NOVI dopplerb values for HI via curve-of-growthconcordance ratio exhibits a strong correlationwith N , with a typ- HI curves (Shull et al. 2000). ical range N /N ≈ 0.5–50 and a dispersion of less HI OVI Of the Lyα absorbers, 129 were at z ≤ 0.15, where than one (dex). Higher H I column absorbers typically O VI absorption could potentially be observed. We de- display a higher multiphase ratio, while weak H I ab- tected O VI at 4σ or greater level in one or both lines sorbershave N ≈(0.1−2)N . We are well awareof OVI HI of the doublet for 40 absorbers, and we obtained 4σ or the factthatplots suchasFigure1aresubjectto the ef- greater upper limits in 84 other cases. The remaining fects of correlatederrorsin N . However,this ratiohas HI five absorbers fell on top of airglow or strong lines from physical utility in deriving a mean (O/H) metallicity in the interstellar medium (ISM), were blended, or were in multiphasegas(see§4),whencombinedwiththeempir- some other way inaccessible. ical correlation between N and hydrogen overdensity, HI We determined metal-ion column densities via Voigt δ , seen in cosmological simulations (Dav´e et al. 1999). H profile fits and/or the apparent column method The large range in N and the correlation in multi- HI (Savage & Sembach 1991). We assumed that any satu- phase ratio provide evidence that the IGM has at least ration in these lines was mild, and that profile fits accu- two phases (WHIM and WNM). If the H I and O VI rately determine the column density. In cases where we materials were well mixed, we would expect that the detected both O VI lines, we assigned a weighted mean multiphase ratio would be constant with N . We sug- HI ofthecolumndensitiestoNOVI. Wedescribethefullde- gest instead that the WHIM occupies a shell around a tails of absorber selection and data analysis in Paper II. warm neutral core. The outer WHIM shell is heated by 3 a combinationof externalionizing photons (fromAGN), shocks from infalling clouds, and possible shocks from cluster outflows such as superwinds and SNR feedback. ThenarrowrangeofN comparedtoN impliesthat OVI HI the WHIM shell may have a characteristic column den- sity of 1013−1014 cm−2, while the neutral core can be arbitrarily large. In this scenario, the very large values of the multi- phase ratio may arise in “quiescent gas”, possibly lo- cated in high-column density H I gas in halos. The absence of associated O VI suggests the lack of shocks at velocities greater than about 150 km s−1. Many strong H I (Lyα) absorbers have been seen in proximity (within 200h−1 kpc) of bright galaxies (Lanzetta et al. 70 1995; Chen et al. 1998; Penton, Stocke, & Shull 2003). In our current O VI sample, we found three absorbers with extremely high multiphase ratios: very high neu- tralhydrogencolumnswithnodetectableWHIM.Inthe first case, the absorber at cz = 1586 km s−1 toward 3C273 shows logN =15.67 and logN ≤13.17 and is71h−1kpcontheHskIyfromadwarf(M OV=I−13.9)post- Fig. 2.— Column density histogram for the observed OVI ab- 70 B sorbers. ApproximateequivalentwidthsforOVIλ1032areshown starburstgalaxy(Stocke et al.2004). The secondcaseis along the top axis. The decreasing number of absorbers at low the absorber at cz = 23,688 km s−1 toward PHL1811 columndensitiessuggeststhatoursurveyisincompleteforweaker with logN = 15.94 and logN ≤ 13.06. This sys- absorptionlines. HI OVI tem is beyond the L∗ survey depth, so we cannot com- ment on surrounding galaxies. Finally, the Lyman-limit system at cz = 24,215 km s−1 toward PHL1811 shows The total redshift pathlength of our survey is a func- logNHI = 18.11 and logNOVI ≤ 13.06. This absorber tion of equivalent width, and our survey is more com- lies 23′′ (34h−1 kpc) from an L∗ galaxy at z = 0.0808 plete for strong absorbers than for weak absorbers. The 70 (Jenkins et al. 2003). Tripp et al. (2005) note a sub- equivalent-width sensitivity, W (λ), is a function of min DLAsystematzabs =0.00632towardPG1216+069with spectrograph resolution, R = λ/∆λ, and the signal-to- logNHI = 19.32 and logNOVI ≤ 14.3 which lies 86 kpc noise ratio (S/N) of the data per resolution element. from a sub-L∗ galaxy. These four absorption systems, Thus, for a 4σ detection limit, we define W (λ) = min with multiphase ratios log[NHI/NOVI] ≥ 2.5, 2.9, 5.0, 4(λ/R)/(S/N). We assume R = 20,000 and calculate and 5.0 respectively, may be shielded from external ion- W profilesforeachdatasetbasedonS/Nmeasuredev- min izing flux by adjacent IGM clouds and have unshocked ery 10 ˚A. Strong instrumental features, ISM absorption, gas. andairglowlinesaremaskedout(S/Nsettozero). From Metallicity is one of the great unknowns in interpret- thisprocedure,wecalculateN (λ)forbothOVItran- min ing the baryon content of the WHIM absorbers. In pre- sitions, using curves of growth with b = 25 km s−1. OVI vious studies, (O/H) has been assumed to be constant By moving an absorber doublet along the profile as a at 10% solar. However, one might expect that metallic- function of z , we generate a profile N (z). By abs min ity variations could play a role in the multiphase ratio: addingupthetotalpathlengthforeachabsorberateach outflows of material from galaxies should be denser and N (z), we determine the relationship between N min OVI more enriched than primordial IGM clouds. This would and ∆z as shown in Figure 3. With a total high-N OVI contribute a negative slope in the multiphase ratio plot pathlength ∆z = 2.21, we are at least 80% complete in (sFhoigwurae 1rebl)a.tivIenlsytelaodw,erthreatdieonsoefrO(hiVghI/-HNHII,) aanbasloorgboeurss OinVthIed1e0te3c2ti˚Aonlidnoew).nHtoowleovgeNr,O∆VIz=an1d3t.4he(Wcoλm=ple3t0enmes˚As to lower metallicity. While metallicity variations almost fall off rapidly for weaker absorbers. Dividing Figure 2 certainly exist in the IGM, they are not the main cause byFigure3,weobtaintheprofileofdN /dz asafunc- of the positive slope in the multiphase ratio plot. Some OVI tion of N (Figure 4). To first order, this procedure of this effect could arise from changes in the mean OVI OVI should correct for incompleteness in our OVI survey. ionization fraction, typically chosen to be the maximum Ourincompleteness-correctedvalueoftheabsorberfre- value, f ≈ 0.2 at T = 105.45 K in collisional ion- OVI max quency, dN /dz, compares well with previous, un- ization equilibrium (Sutherland & Dopita 1993). This OVI corrected values (Table 1). For O VI absorbers with fraction is expected to vary, depending on the range of W ≥ 50 m˚A (in λ1032), we find dN /dz = 9±2, shock velocities that produce the O VI (Heckman et al. λ OVI somewhatlowerthanthevalue,dN /dz =14+9,found 2002; Rajan & Shull 2005). OVI −6 by Savage et al. (2002) using six OVI absorbers toward 4. THE HOTBARYONCONTENT OFTHE UNIVERSE PG0953+415. For the weaker O VI absorbers, we find Using O VI as a tracer of the 105−6 K portion of the dNOVI/dz = 17±3 (Wλ ≥ 30 m˚A) and dNOVI/dz = WHIM,wecanemploythe numberofabsorbersperunit 19± 3 (W ≥ 15 m˚A) from our total sample of 40+7 λ −6 redshift,dN /dz,todetermineΩ ,thefractionof measured O VI absorbers. Uncertainties are based on OVI WHIM the critical density contributed by this WHIM gas. Our single-sided 1σ confidence limits in Poisson statistics detection statistics are shown in Figure 2 for OVI. (Gehrels 1986). Tripp, Savage, & Jenkins (2000) found 4 Fig. 3.—CumulativeFUSEredshiftpathlength,∆z,asfunction of 4σ absorption-linedetection limitsinOVI columndensity and λ1032 equivalent width. Strong lines can be detected in poor- quality data. Weak linescan only bedetected inthe highest-S/N observations and thus contribute less total path length. We use the∆z profiletocorrectforincompleteness inourOVIsurvey. Fig. 5.—CumulativeequivalentwidthdistributionforOVIab- sorbers after Figure 10 of Chenetal. (2003). Our data points (filled diamonds) at (OVI λ1032) equivalent width detection thresholdsof10,15,30,50,and100m˚Aareslightlylowerthanval- dN /dz > 17 at 90% confidence for W ≥ 30 m˚A, ues from Savageetal. (2002) (square), Tripp,Savage, &Jenkins OVI λ (2000) (circle), and Trippetal. (2004) (triangle). Models from based on four absorbers toward H1821+643. Chenetal. (2003) show the simulated distribution at a range of A recent study (Tripp et al. 2004) using STIS/E140M metallicities: Z = 0.1Z⊙ (solid curve), Z = Z⊙ (dotted), and data finds dN /dz = 23±4 for W ≥ 30 m˚A based an overdensity-dependent metallicity Z = Z(δH) (dashed). The on 44 OVI abOsoVrIbers. This sample waλs taken at slightly Zda=she0d.5)Zre⊙prmesoednetlsuasemdobreysFimanpgl,ifiBerdyapnh,ys&icaClamniozdareel.s(2002)(dot- higher redshift (0.12 ≤ z ≤ 0.57) than our sample abs (z ≤ 0.15) and shows a higher value of dN /dz. abs OVI Thisisnotpredictedbysimulations,whichpredictanin- andthediscrepancyindN /dz maybearesultofcos- creasing WHIM fraction at recent epochs. However, the OVI mic variance. difference in redshift between the two samples is small, Recent cosmological simulations of the X-ray forest predict a distribution of O VI absorbers and provide a convenient way to check current simulations with ob- served data. Chen et al. (2003) model the X-ray forest assuming a ΛCDM model similar to Dav´e et al. (2001), both collisional ionization and photoionization from a UV background, radiative cooling, and a range of dif- ferent metallicities. The cumulative distribution of OVI absorbers(dN /dln[1+z])downtoaminimumequiv- OVI alent width is drawn from these models and provides a convenientcomparisonwith observedstatistics from our work and previous surveys (see Figure 5). The simulation with Z = 0.1Z⊙ shows a reasonably good match to the observed cumulative distribution of O VI absorbers in the local universe. Weaker absorbers (W <30m˚A)areoverpredictedcomparedwithobserva- λ tions, but this may be a matter of small-number statis- tics or differing detection thresholds between observa- tions and simulations. A simulationin which metallicity is a function of overdensity fits the data slightly better than the fixed-metallicity model. However, Chen et al. (2003) caution that there is a substantial scatter in the simulations, so that the lower-metallicity curves are es- unFiitgr.e4d.s—hiftNuafmtebrerouorfOfirVstI-oarbdseorrbceormsppleertecnoelsusmcnordreencstiitoyn.binEvpeenr sentially indistinguishable. The Z = 0.5Z⊙ model of Fang, Bryan, & Canizares (2002) also fits the observed afterthiscorrection,thenumberofOVIabsorbersappearstofall offrapidlybelowlogNOVI≤13.4. However,thestatisticsarepoor data reasonably well, but it is based on a less physical withonlyfiveabsorbersinthelowesttwobins. simulationwithnophotoionizationandnoradiativecool- 5 and TABLE 1 n IGM OVI AbsorberStatistics δ ≡ H H (1.90×10−7 cm−3)(1+z)3 Criteria Nabs dN/dz ΩWHIMa reference ≈20 N0.7 10−0.4z . (2) 14 W ≥10m˚A 40 21±3 0.0022±0.0003 thiswork TheaboverelationsallowustorelateN tothephysical W ≥15m˚A 38 19±3 0.0021±0.0003 thiswork HI gasdensity,n ,neededforthehydrogenphotoionization W ≥30m˚A 35 17±3 0.0021±0.0004 thiswork H W ≥50m˚A 19 9±2 0.0016±0.0005 thiswork correction. Equation (1) is derived by fitting the multi- W ≥100m˚A 6 3+2 0.0008±0.0004 thiswork phase correlation in Fig. 1b over the approximate range −1 10 ≤ δ ≤ 300. The scaling constant C = 2.5±0.2 is H 0 W ≥30m˚A 44 23±4 0.0027 Trippetal.(2004) themeanmultiphaseratioatlogN =14,andthebest- HI W ≥30m˚A 4 >17 >0.006b Tripp,Savage, &Jenfiktintsin(g20s0l0o)pe is α = 0.9±0.1. Equation (2) comes from W ≥50m˚A 6 14+−96 ≥0.003 Savageetal.(2002) cosmological simulations (Dav´e et al. 1999) and relates aAllΩWHIM valueshavebeenconvertedtoaconsistentsetofas- the hydrogenoverdensity,δH,to the HI columndensity, sumptions: fOVI =0.2, Z =0.1Z⊙, H0 =70h70 kms−1 Mpc−1, NHI ≡(1014 cm−2)N14. and(O/H)⊙ =4.9×10−4 (AllendePrietoetal.2001). From these relations, we can derive a statistical value bAll four OVI absorbers in Tripp,Savage, &Jenkins (2000) lie of the (O/H) metallicity from the formula, alongtheunusuallyrichH1821+643sightline. f HI hN /N i=hN /N i× . (3) O H OVI HI (cid:18)f (cid:19) OVI We employ the multiphase ratio, N /N , together HI OVI ing. The solar metallicity model of the IGM can clearly with appropriate ionization correction factors, fOVI and be ruled out by our observations. Simulations at lower fHI. The H I fraction is derived from photoionization metallicities (Z = 0.01Z⊙) would be helpful, given the equilibrium in the low-z IGM wealth of new observational results. n α(A) δ From dN/dz, we can calculate the contribution to Ωb f = e H =(1.80×10−5)(1+z)3 H T−0.726Γ−1 , fromWHIMgas,asdiscussedinSavage et al.(2002). We HI Γ (cid:18)20(cid:19) 4 −13 H assume a Hubble constant H = 70h km s−1 Mpc−1, (4) 0 70 rather than their value h , and we make standard as- for gas with n = 1.16n at temperature (104 K)T , 75 e H 4 sumptions regardingOVI ionizationfractionand(O/H) photoionized at rate ΓH = (10−13 s−1)Γ−13. Since metallicity(Tripp, Savage, & Jenkins2000;Savage et al. the mean-free path of a Lyman continuum photon is 2002). Weadoptanionizationfraction,f =0.2char- very large in the low-overdensity IGM, we use the OVI acteristic of its maximum value in collisional ionization case-A recombination rate coefficient α(A) = (4.09 × H equilibrium (CIE) and assume an O/H abundance 10% 10−13 cm3 s−1)T−0.726. Combiningthetwoempiricalre- of the solar value. Of course, CIE is almost certainly 4 lations with the relation of photoionization equilibrium, not a valid assumption for the low-density IGM. As the we find that the mean oxygen metallicity of the O VI infalling gas is shock-heated during structure formation, absorbers at hzi=0.06 is the ionization states O VI, O VII, and O VIII undergo transient spikes in abundance, followed by cooling and f −1 recombination. Recent time-dependent models of the ZO =(0.09Z⊙)N1−40.2T4−0.726Γ−−113(cid:18) 0O.V2I(cid:19) . (5) ionization, recombination, and cooling of shock-heated, low-density WHIM (Rajan & Shull 2005) find that the Given the uncertainties in these empirical relations, it is mean,time-averagedOVIionfractionishfOVIi=8–35%, remarkable that this formula arrivesat an oxygenabun- over a range of initial temperatures 5.4 ≤ logT ≤ 6.2. dance near the fiducial value of 10% solar. In fact, our These fractions are compatible (with a factor of two estimated value, ZO = 0.09Z⊙, is probably accurate to spread) with the fiducial value, fOVI = 0.2, based on only a factor of 2. the maximum fraction of O VI at T = 105.45 K, in max CIE (Sutherland & Dopita 1993). As shown above (Fig. 5), a metallicity Z = 0.1Z⊙ is reasonablyconsistentwiththeobservedequivalentwidth distribution of OVI absorbers at z ∼0. However, using our sample of O VI and H I absorbers, we can make a more direct estimate of the (O/H) metallicity. To do so,we must estimate the amountof hydrogenassociated with each component of the the multiphase system of photoionized H I and collisionally ionized O VI. Since the HI andOVIabsorbersare associatedkinematically, we assume they share the same metallicity. We then use theempiricalrelationsbetweenN andoverdensity,δ , HI H and between the multiphase ratio (N /N ) and N . HI OVI HI N HI =C Nα , (1) (cid:18)N (cid:19) 0 14 OVI 6 We now derive the fractional contribution to closure this analysis can be applied to WHIM species. We find density of WHIM baryons associated with hot OVI, that the differential number of OVI absorbers with col- umn density is a power law, dN /dN ∝N−2.2±0.1 H µm OVI OVI OVI ΩWHIM= 0 H for absorbers with logNOVI ≥ 13.4. This is some- (cid:18)cρcr(cid:19)(O/H)⊙ Z fOVI what steeper than the corresponding H I distribution, Nmax dN β ∼ 1.6 (Penton, Stocke, & Shull 2004) and means that × hNOVIidNOVI (6) low-columnabsorberscontributecomparableamountsto ZNmin (cid:18) dz (cid:19) theOVIbaryoncensusastherarehigh-columnsystems. dN Note that in eq. [6], Ω scales as N−0.2 for β =2.2. =(1.85×10−18) h−1 hN i . WHIM min 70 (cid:18) dz (cid:19) OVI i Eventhoughwecorrectforincompletenessintheweaker Xi i absorbers,thestatisticsarestillpooratlogN ≤13.4. OVI Weperformtheabovesumusingthevalue,(dN/dz)i,for It is unclear whether the turnover in dNOVI/dz at lower each column-density bin in Figure 4, with hN i cho- columns is real or a statistical fluctuation from small OVI i sen as the mean column density (cm−2) in the bin. We numbersofweakabsorbers. Nevertheless,theweakOVI find Ω = (0.0022±0.0003)h−1. Values of Ω absorbersappeartomakesignificantcontributionstothe WHIM 70 WHIM baryonmassdensity. Thesteeppowerlawalsoreinforces for other equivalent width thresholds are listed in Ta- ourconclusionsaboutthe natureofthe multiphase IGM ble 1 alongwith the resultsofother studies convertedto with a core-halo structure. uniform values of H0, fOVI, and (O/H)⊙. We have demonstrated that the contributions from ItshouldbestressedthattheseestimatesofΩ are basedonanassumeduniform10%solar(O/H)WmHeItMallic- weak OVI absorberscannot be neglected in an accurate WHIM baryon census. The rare strong O VI absorbers ity and that further uncertainty arises in our assumed O VI ionization fraction. Changes in f within the and numerous weak absorbers contribute nearly equally OVI to Ω . Further FUSE observations at high-S/N will rangeofmeanvalues(Rajan & Shull2005)couldchange WHIM allow us to probe weak O VI absorbers and refine the our result by a factor of two. Based on the standard statistics at the low-column end of the absorber distri- assumptions, our result is slightly lower than the result published by Savage et al. (2002), Ω ≥ 0.002h−1. bution. AnanalysisofOVIabsorbersathigherredshifts Savage et al. (2002) used an older sWolHaIrMoxygen abu75n- (z >0.5) would allow us to track changes in the WHIM dance, (O/H)⊙ =7.41×10−4, which is 50% larger than density,confirmingthedecreaseintheamountofshocked gasathigherredshifts predictedbycosmologicalsimula- the Allende Prieto et al. (2001) value. After correcting tions (Cen & Ostriker 1999; Dav´e et al. 1999, 2001). for this difference, their result for Ω is larger than WHIM Onlybyenlargingoursampleofmultiphase(HI,OVI) ours. Tripp et al. (2004) find Ω =0.0027 based on WHIM absorbers beyond the 40 discussed here will we be able 44 absorbers at z ≥0.12. torefineourstatisticalestimateforthemetallicityofthe Our estimate corresponds to Ω /Ω = 0.048 ± 0.007, so that WHIM gas in the rWanHgIMe 105b−6 K makes WHIM.WithamuchlargersampleofOVIabsorbers,we could use maximum-likelihood techniques to search for up roughly 5% of the baryonic mass in the local uni- the expected trends: (1) decreasing metallicity at lower verse. The contribution could exceed 10% if we ac- overdensityδ (Gnedin & Ostriker1997);(2)decreasing count for the likely spread in (O/H) metallicities and H O VI ionization fractions. Simulations predict that the ΩOVIathigherredshift(Dav´e et al.1999);and(3)trends current universe is composed of ∼ 30% WHIM gas, so of NHI/NOVI with other metal indicators such as C III, CIV, or SiIV. our value of ∼ 3−10% falls short of this mark. How- ever, O VI is only useful as a proxy for gas within the lowerportionoftheWHIMtemperaturerange. Account- ing for the hotter 106−7 K gas using similar method- ology will require high-sensitivity, high resolution X- We are grateful to Steve Penton, John Stocke, Jessica ray observations of O VII and O VIII with spectro- Rosenberg, Todd Tripp, Bill Blair, Blair Savage, Ken graphs aboard future missions such as Constellation-X Sembach,andJasonTumlinsonfor usefuldiscussionsre- or XEUS (Fang, Bryan, & Canizares 2002; Chen et al. garding this project. This work contains data obtained 2003). Limited observationalwork has been done in this for the Guaranteed Time Team by the NASA-CNES- area(Nicastro et al.2005),butthe currentgenerationof CSAFUSEmissionoperatedby the JohnsHopkins Uni- X-ray telescopes are not ideal for this kind of investiga- versity,as well as data from the Hubble Space Telescope. tion. Financial support to the University of Colorado has The distribution of H I absorbers with column been provided by NASA/FUSE contract NAS5-32985 density is often expressed as a power law with in- and grant NAG5-13004, by our HST Lyα survey pro- dex β: dN/dN ∝ N−β (Weymann et al. 1998; gram6593,andby theoreticalgrantsfromNASA/LTSA Penton, Stocke, & Shull 2000, 2004). For the first time, (NAG5-7262)and NSF (AST02-06042). REFERENCES AllendePrieto,C.,Lambert,D.L.,&Asplund,M.2001,ApJ,556, Danforth, C. W., Shull, J. M., & Rosenberg, J. L., 2005, ApJ, in L63 preparation(PaperII) Cen,R.,&Ostriker,J.P.1999,ApJ,519,L109 Dav´e,R.,etal.1999,ApJ,511,521 Chen, H-W.,Lanzetta, K. M.,Webb, J. K.,& Barcons, X.,1998, Dav´e,R.,etal.2001,ApJ,552,473 ApJ,498,77 Fang,T.,Bryan,G.L.,&Canizares,C.R.2002,ApJ,564,604 Chen, X.,Weinberg, D.H.,Katz, N.,& Dav´e,R.2003, ApJ, 594, Gehrels,N.1986,ApJ,303,336 42 Gnedin,N.Y.,&Ostriker,J.P.1997,ApJ,486,581 Collins,J.,Shull,J.M.,&Giroux,M.2004,ApJ,605,216 7 Grevesse, N., Noels, A., & Sauval, A. J. 1996, in ASP Conf. Ser. Savage, B.D.,&Sembach,K.R.1991,ApJ,379,245 99, Cosmic Abundances, ed. S. S. Holt & G. Sonneborn (San Sembach,K.R.,etal.2003,ApJS,146,165 Francisco:ASP),117 Shull,J.M.,Giroux,M.,etal.2000, ApJ,538,L13 Heckman, T. M., Norman, C. A., Strickland, D. K., & Sembach, Shull, J. M. 2003, in The IGM/Galaxy Connection: The K.R.2002, ApJ,577,691 Distribution of Baryons at z = 0, ed. J. L. Rosenberg & M. Jenkins, E. B., Bowen, D. V., Tripp, T. M., Sembach, K. R., E.Putman,(Dordrecht: Kluwer),1 Leighly, K. M., Halpern, J. P., & Lauroesch, J. T. 2003, AJ, Stocke, J.T.,etal.2004,ApJ,609,94 125,2824 Sutherland,R.S.,&Dopita,M.A.1993,ApJS,88,253 Lanzetta, K.M.,etal.1995,ApJ,442,538 Tripp,T.M.,Savage, B.D.,&Jenkins,E.B.2000,ApJ,534,L1 Nicastro,F.,etal.2005, Nature,inpress Tripp, T. M., et al. 2004, in Astrophysics in the Far Ultraviolet: Penton, S.V.,Shull,J.M.,&Stocke, J.T.2000,ApJ,544,150 FiveYearsofDiscoverywithFUSE,Victoria,BC. Penton, S.V.,Stocke, J.T.,&Shull,J.M.2000,ApJS,130,121 Tripp,T.M.,etal.2005,ApJ,inpress Penton, S.V.,Stocke, J.T.,&Shull,J.M.2003,ApJ,565,720 Weymann,R.,etal.1998, ApJ,506,1 Penton, S.V.,Stocke, J.T.,&Shull,J.M.2004,ApJS,152,29 Rajan,N.&Shull,J.M.,2005,ApJ,inpreparation Savage, B. D.,Sembach, K. R., Tripp,T. M., & Richter, P. 2002, ApJ,564,631