Mon.Not.R.Astron.Soc.000,1–9(2003) Printed2February2008 (MNLaTEXstylefilev2.2) High-redshift X-ray properties of the haloes of simulated disc galaxies J. Rasmussen1⋆, J. Sommer-Larsen2, S. Toft1, K. Pedersen1 1 Astronomical Observatory, Universityof Copenhagen, Juliane MariesVej 30, DK-2100 Copenhagen Ø, Denmark 2 Theoretical Astrophysics Center, Juliane MariesVej 30, DK-2100 Copenhagen Ø, Denmark 4 0 0 2 n a ABSTRACT J X-ray luminosities and surface brightness profiles of the hot gas haloes of simulated 7 discgalaxiesatredshiftsz =0−2arepresented.Thegalaxiesareextractedfromfully cosmologicalsimulations and correspond in mass to the Milky Way. We find that the 2 bolometric X-ray luminosities of the haloes decrease by a factor 4−10 from z ∼1 to v z ∼ 0, reflecting the decrease in the rate at which hot halo gas cools out on to the 5 disc. At all redshifts, most of the emission is found to originate within 10–15 kpc of 3 the disc.Whencombinedwithmodels inwhichthe evolutionofdisc X-rayluminosity 6 is dominated by X-ray binaries, the predicted halo luminosities at z ∼ 1 show good 0 agreementwith constraints fromspiralgalaxies in Chandra Deep Field data. There is 1 an indication that haloes with a metal abundance of 0.3Z overpredict observed X- 3 ⊙ 0 ray luminosities at z ∼1, suggesting that halo metallicities are lower than this value. / Prospects for direct detection of the haloes of Milky Way–sized galaxies with current h andfutureX-rayinstrumentationarediscussed.ItisfoundthatXEUSshouldbe able p to single out the halo emission of highly inclined Milky Way–sized disc galaxies out - o to z ≈0.3. For such galaxies in this redshift interval, we estimate a lower limit to the r surface density of detectable haloes on the sky of ∼ 10 deg−2. More generally, owing t s to their luminosity evolution, the optimum redshifts at which to observe such haloes a could be 0.5<z <1, depending on their assembly history. : v Key words: methods: N-body simulations – cooling flows – galaxies: formation – i X galaxies: haloes – galaxies: spiral – X-rays: galaxies. r a 1 INTRODUCTION Katzet al.2003),whichsuggeststhatmostgalaxiesaccrete the majority of their gas at temperatures much lower than Standard models of disc galaxy formation require spiral thevirial temperatureof theirhalo. This would be particu- galaxies to be surrounded by large reservoirs of hot gas larly pronounced at higher redshifts, and would imply that which should be emitting at X-ray wavelengths and from most of the halo radiation is emitted as Lyα emission close which gas should stillbeaccreting on tothedisc at present to the disc rather than as soft X-rays from an extended, (e.g.White & Frenk1991).Thedetailed propertiesofthese quasi-spherical region. haloes and their role in the formation and evolution of Inapreviouspaper(Toft et al.2002,hereafterPaperI) galacticdiscsremainlargelyunknownfromanobservational weextracteddisc galaxies from fully cosmological hydrody- viewpoint, as the haloes have so far escaped direct detec- namical simulations and determined their present-day halo tion. Possible exceptions are NGC891 (Bregman & Houck X-ray luminosities, finding consistency with observational 1997; Strickland et al. 2003a) and the Milky Way itself upperlimitsonhaloemission from nearbyspirals.Itshould (e.g.Sidheret al.1996;Pietz et al.1998),ifneglectingcases be mentioned that we have subsequently detected an error where(i)thegalacticdiscsshowevidenceofbeingdisturbed intheX-raycalculationsemployedinthatstudy.Correcting by tidal interactions, or (ii) the X-ray emission at off-disc this increases the luminosity of hot haloes with gas metal- distances of a few kpc can be attributed to processes orig- inating in the disc such as feedback from star formation licity Z = 0.0 by a factor ∼ 2 and that of Z ≃ 0.3Z⊙ haloes by a factor ∼ 1.25. This correction applies equally (Dahlem et al. 2003; Strickland et al. 2003a,b). A possible to all our simulated galaxies, so the overall conclusions of explanation for this lack of halo detections may be pro- Paper I remain unchanged. Accounting for this error still vided by recent simulation work (Birnboim & Dekel 2003; producesresultsinconsistencywithobservations,astheup- perlimitsonobservedluminositiescaneasily accommodate ⋆ E-mail:[email protected] thecorrection. Onemain result of thestudyof PaperI was 2 J. Rasmussen et al. the conjecture that spiral haloes were possibly one order of the approach of Paper I, to which we refer for more de- magnitude brighter in soft X-rays at z = 1 than today, a tails: All SPH gas particles outside a box sized (1000 kpc)3 result which was based solely on the predicted mass accre- centred on the galaxy are cut away, as is ’cold’ gas (with tion rates of the discs. Analytical models of halo emission T < 3×104 K). Spatially smoothed temperature and den- overpredictemission atz =0byatleast anorderofmagni- sity fields are constructed from the values associated with tude (see Benson et al. 2000 and Paper I) and thus cannot individual particles, and X-ray volume emissivities are cal- beassumedtoprovideareliabledescriptionoftheevolution culated from these using the meka plasma emissivity code inX-rayemissionwithredshift.Toamendthissituation,we (Mewe, Lemen & van den Oord 1986). hereextendourpreviousworkbystudyingthehigh-redshift Atanygiven redshift, somevariation intotal X-raylu- propertiesofafewdiscgalaxiesextractedfromcosmological minosity is seen on time-scales of a few tens of Myr, dueto cold dark matter simulations. galaxy-galaxyinteractionsandmergingofsmallersatellites. We found in Paper I that the halo luminosity approx- Sincewearemainlyinterestedinthelong-termevolutionof imately scales as LX ∝ Vc5, where Vc is the characteristic LX, on the order of 10 frames separated by time intervals circularspeedinthediscatR2.2 =2.2Rd andRd isthedisc ∆t ≈ 100 Myr have been extracted around each redshift. scalelength. Although subject to considerable scatter, such This time interval is large enough that individual frames a trend is consistent with expectations from simple cooling canbeconsideredreasonablyindependent,giventhetypical flowmodelsandsuggeststhatmassivegalaxies(Vc &300km time-scale involved, yet small enough for overlaps between s−1)wouldbeoptimumtargetsforobservinghaloemission. adjacent redshift bins to be insignificant. X-ray properties Such galaxies are rare, however, and hence do not compare ofeachframehavebeencalculatedasprescribedabove,and welltothetypicalspiralseenindeepX-raysurveys.Theaim frameshavethenbeenomittedforwhichthetotalbolomet- of this work is therefore to study the predicted properties ric (0.012–12.4 keV) luminosity deviates by more than 2σ of, and the detection prospects for, haloes of more typical from the mean at that redshift (discarding on average less disc galaxies of size similar to theMilky Way(MW). thanoneframeperredshift,correspondingto∼7percentof The simulations andX-raycomputations aredescribed all frames). Mean valuesand dispersions of all quantitiesof in §2. Results are presented in §3 and compared to obser- interestwerederivedatagivenredshift from theremaining vations in §4. We discuss the possibility for detecting the frames and used in thesubsequent analysis. haloes in §5 and summarize ourfindings in §6. For the Z = 0.0 galaxies we follow the halo properties backto z=2.9 (gal15) andz =2.3 (gal18), beyondwhich the disc itself is not well-defined. The Z = 0.3Z⊙ galaxies show a more complex gas distribution at all redshifts and 2 SIMULATIONS AND X-RAY are followed only back to z≈2. CALCULATIONS The primary goal is to study the redshift evolution in halo 3 RESULTS X-ray properties of simulated disc galaxies. To this end, we have extracted two disc galaxies from fully cosmologi- cal TreeSPH simulations of galaxy formation in a ΛCDM 3.1 LX versus redshift cosmology with ΩM = 0.3, ΩΛ = 0.7, and H0 = 65 km Ofprimeinterestistheevolutionwithredshift oftheX-ray s−1 Mpc−1. These values are adopted throughout this pa- luminosity, in part because this is expected to reflect the per.Thesimulationsincludestarformation,stellarfeedback gas accretion history of the stellar disc as will be discussed processes, and ameta-galactic UVradiation field.Forade- below. scriptionofinitialconditionsandsimulationdetails,werefer Fig.1showstheredshiftevolutionoftherest-frame0.2– to Sommer-Larsen, G¨otz & Portinari (2003). 2keVandbolometricluminositiesLX,bolofthehothaloesof From a simulation with an adopted universal baryon the two galaxies. In general LX,bol is seen to increase with fraction of fb =0.10, we study in detail two galaxies (here- redshift, rising fairly steeply out to z ∼ 1, beyond which afterdenotedgal15andgal18),eachrunbothwithacool- the increase levels off. The increase from z = 0− 1 is a ingfunctionbasedonaprimordialgascomposition([Fe/H]= factor ∼4−10, thus verifying our prediction from Paper I −∞, i.e. Z = 0.0) and with one based on the typical in- that disc galaxy haloes could be up to an order of magni- tracluster gas metallicity ([Fe/H]= −0.5, i.e. Z ≃ 0.3Z⊙). tude brighter at z = 1 than present-day haloes. As can be The latter abundance can probably a priori be considered seen, gal15 shows a steady evolution in both X-ray bands, a reasonable upper limit to the metallicity of hot halo gas. whereas gal18 exhibits a slightly more complex behaviour. The two galaxies were selected from the requirement that The reason is that gal18 experiences a period of enhanced they should show disc circular speeds Vc at z = 0 com- accretion, as will be demonstrated below. parable to that of the present-day MW. With values at Itisseenthatathighz the0.2–2keVemissionissome- z = 0 of Vc = 225 (250) km s−1 for gal15 with Z = 0.0 what higher for the Z = 0.3Z⊙ hot haloes than for the (Z = 0.3Z⊙) and Vc = 202 (213) km s−1 for gal18 with Z = 0.0 haloes, falling short of the bolometric luminosity Z =0.0(Z =0.3Z⊙),theyhaveMW-likemasses andhence at all z only by a factor of ∼ 2−3. This is because the correspond to L ∼ L∗ galaxies in the local Universe. The Z = 0.3Z⊙ haloes display higher emission-weighted mean z = 0 version of both galaxies were included in Paper I. temperatures (∼ 0.20–0.25 keV, with little redshift depen- Sincethegalaxieswereformedinthesamecosmologicalsim- dence) than the Z = 0.0 haloes (∼ 0.10–0.15 keV), thus ulation,differencesbetweenthemonlyreflecttheirdifferent radiating a relatively larger fraction of their total emission assembly histories. in this band. Given the near-constancy of the hot gas tem- For the calculations of halo X-ray properties we follow perature,thedecreaseinLX withtimeispredominantlydue High-redshift X-ray properties of the haloes of simulated disc galaxies 3 Gal15 Gal15 10.0 10.0 ) ) s s g/ g/ r r e e 400 1.0 400 1.0 1 1 ( ( L X L X 0.1 0.1 0.0 0.5 1.0 1.5 2.0 2.5 3.0 0.0 0.5 1.0 1.5 2.0 Redshift Redshift Gal18 Gal18 10.0 10.0 ) ) s s g/ g/ r r e e 400 1.0 400 1.0 1 1 ( ( L X L X 0.1 0.1 0.0 0.5 1.0 1.5 2.0 0.0 0.5 1.0 1.5 2.0 Redshift Redshift Figure1.Redshiftevolutionoftherest-framebolometricand0.2–2keVX-rayluminositiesofthesimulatedgalaxies.Toppanelgal15, bottom panelgal18, withmetallicitiesaslabelled.Errorbarssignifythe1σ dispersionbetween framesatagivenredshift. to a decline in the available amount of hot halo gas along andvolume-weightedmeandensityofhotgasrelativetothe withadecreaseinthevolume-weightedmeandensityofthis Z =0.0 case. gas. Themassof hothalogasintheX-rayemittingvolume declinesonlybyafactor∼2(Z =0.0)and∼4(Z =0.3Z⊙) from z ∼ 1 to z ∼ 0, however. An implication of this com- 3.2 Accretion rates bined with Fig. 1 is that haloes radiate relatively more per The halo luminosity can be linked to the rate at which hot unitmass of hotgas at high z thanat z=0 (typicallybya halogas coolsoutandaccretesontothedisc,asillustrated factor ∼2 at z=1 relative to z=0). bythe following considerations. Note that for both galaxies, increasing the cooling effi- The cooling time is the gas energy density divided by ciencybyraisingthegasmetallicitytoZ =0.3Z⊙ leadstoa thecooling rate, decrease in bolometric LX at z =0, consistent with the re- 3NkT sultsfrom ourmuchlarger samplein Paper I.Thedecrease tcool≈ 2Λn2 , (1) is less significant for gal18, which, as will beargued below, H isprobablysomewhatatypicalofthegalaxiesstudiedinthat whereΛisthecoolingfunction,nH isthehydrogennumber paper.Also notethat thisdecrease persists back toat least density, and N =(ne+ ini) is the total number density z ≈ 2. This may at first seem counter-intuitive, given the of electrons and all ionPspecies i. The mass cool–out rate higherX-ray emissivity of a Z =0.3Z⊙ plasma. The reason dM˙ of gas with mass density ρ in a small volume element isthatalreadyatredshiftsz&2alargerfractionofhothalo dV becomes gas has cooled out to form stars in theZ =0.3Z⊙ galaxies, dM˙ . ρdV = 2µmpΛn2 dV, (2) leading at z ≈2 to a factor ∼2 deficit in the total amount tcool 3kT H 4 J. Rasmussen et al. whereµisthemeanmolecularweight,mp theprotonmass, 4 COMPARISON WITH OBSERVATIONS and equality corresponds to the case of absence of exter- nalheatsourcesandpdV work.Sincetheemission-weighted Although only a few relevant observational constraints on mean inversetemperature is halo emission are available even in the nearby Universe, we showedinPaperIthatLX ofoursimulatedgalaxiesatz=0 h1/Tiew = R ΛΛnn2H2T1ddVV (3) asigvreeespwiriatlhsaonbdsewrviethdeusptipmeratleimsoitfsthfoerahmaolouenstoafnndelaurmbyinomsaitsy- H R of hot halo gas in the Milky Way (these conclusions all re- and the bolometric luminosity is LX = Λn2HdV, we find main valid after correction for the error mentioned in §1). R Some recent observational progress in this context include M˙ . 2µmp Λn2HdV = 2µmpLXh1/Tiew (4) theresults of Kuntzet al. (2003) on emission in thenearby 3k Z T 3k spiral M101. These authors find that our halo prediction V from PaperI for a galaxy like M101 (Vc ≈170 kms−1) can easily beaccommodated within theamountof diffuseemis- ≈0.066(cid:18)1040LerXg s−1(cid:19)(cid:18)hk1e/VT−ie1w(cid:19) M⊙ yr−1. sdiiosnplaoybstaainseudffifocirenthtliys gsmalaaxllyv,eprrtoicvaidlesdcatlheehheaiglohtg.aOstdheenrsritey- cent results comprise those of Dahlem et al. (2003), Strick- Inthissimplifiedpicture,wethereforeexpectM˙ to(roughly) landetal.(2003a,b),andWang, Chaves& Irwin(2003),but scale with LXh1/Tiew. in these cases the X-ray emission is either affected by tidal As was shown in Paper I, the gas in the simulations interactionwithanothergalaxyorcanbeattributedtopro- is essentially two-phased, so M˙ is here estimated at any cesses in the disc, and they cannot bedirectly compared to givenredshiftbyconsideringtherateatwhichhotgas(T & ourmodels. 2×106 K)cools outto thecold phase(T .3×104 K),see Although at high redshifts there is currently no possi- Sommer-Larsen et al. (2003). bility of comparing the predicted levels of halo emission to In Fig. 2 we have plotted M˙ derived for the Z = 0.0 direct observations, useful constraints can nevertheless be galaxies along with the expectation from Eq. (4). There is extracted from theChandra Deep Field observations (these a remarkably good agreement in shape, particularly for the are superior in sensitivity to any other pointed X-ray ob- steadily accreting galaxy gal15. Although theagreement is servation). Spiral galaxies, classified morphologically from slightly less convincing for gal18, it is, however, clear that Hubble Deep Field data, have been detected in the 1 Ms this galaxy undergoes a period of strongly enhanced accre- Chandra Deep Field-North (CDF-N) data down to a nomi- tion around z ≈ 1.3 which affects its X-ray luminosity ac- nal limiting 0.5–2 keV point source flux of ≈3×10−17 erg cordingly.Inbothcases theaccretion rateat z≈1isfound cm−2s−1,andusingastackingtechniquedetectionhasbeen to be ∼ 5 times larger than at present, decreasing e.g. for gal15 from 2.5 to 0.5 M⊙ yr−1 from z =1 to z =0. Based pTehrefolramtteedratoutehvoernslcoownesridfleurexdesth(HeoinrtnesgcrhaetmedeiXer-reatyael.m2i0ss0i2o)n. onFig.2(left),anapproximaterelationdescribingthisevo- from L ≈ L∗ Sa–Sc spirals at z = 0.4−1.5 which did not lution in the redshift range z=0−1 is harbour a bright AGN or display strong starburst activity. logM˙(z)≈0.6z−0.25, (5) Resulting rest-frame 0.5–2 keV luminosities in several red- shift ranges were derivedand can betakenhereasobserva- with M˙ in M⊙ yr−1. The observed difference in normaliza- tionalupperlimitstoourpredictedhaloemission.Account- tion between this relation and Eq. (4) is due to the simpli- ingfordiscemission,notincludedinoursimulatedgalaxies, fying assumptions underlying the latter. We speculate that strengthens the CDF constraints on halo emission further. themainreasonsforthediscrepancyareinparttheneglect To this end, we consider the MW disc, for which Warwick of pdV work and in part a geometrical effect related to the (2002) lists a total 0.5–2 keV luminosity of ∼ 5×1039 erg fact that cooling proceeds in a cooling flow, so considering s−1, including all stellar and diffuse disc sources. This pro- a fixed volume for the estimation of M˙ is a clear simpli- vides a measure of the contribution to be added to that of fication. The problem is highly complex, however, and we our simulated haloes at z = 0, in order to obtain an es- cannot excludethat other effects play a role too. timate of the integrated X-ray luminosity of the simulated Giventhepredictedevolutionofdiscaccretionratesand galaxies. However, the MW disc is dominated by X-ray bi- halo bolometric X-ray luminosities, it is interesting to note naries (XRBs) at a level of ∼ 3×1039 erg s−1 (Warwick that Barger et al. (2001) found that accretion on to super- 2002). This contribution to the disc emission is unlikely to massive black holes, as measured by the bolometric X-ray remain constant with redshift, as it is expected to roughly luminosity of active galactic nuclei (AGN), follows a very scale with thedisc star formation rate (SFR).Globally, the similar trend to that seen in Figs. 1 and 2. The logarithm latterisknowntoriseafactorof∼10betweenz∼0–1(e.g. of the black hole mass accretion rate was found to grow Madau, Pozzetti & Dickinson1998),aresultwhichalsoap- roughlylinearlywithredshiftouttoz ∼1–1.5,abovewhich plies to the CDF-N in particular (Cohen 2003). Based on theincrease levels off.It seemsreasonable toassumethat a an assumed suchcoupling between discSFRand LX of the positive correlation should exist between the disc accretion XRB population, Ghosh & White (2001) present a model rate and that of any central black hole (given the observed for the redshift evolution of LX from XRBs. Assuming a correlation between black hole mass and host galaxy bulge Madau-type SFR profile, these authors predict a factor of mass, e.g. McLure & Dunlop 2002), in which case the ob- ∼5(3)increase in LX from XRBsat z=1(z=2) relative servationsofBarger et al.(2001)could betakentosupport to z =0. We have adopted the expected XRB contribution theoverall trendspredicted in Figs. 1 and 2. to the integrated disc emission from Warwick (2002) and High-redshift X-ray properties of the haloes of simulated disc galaxies 5 Gal15 Gal18 Figure 2.TotaldiscaccretionrateM˙ (solidline)ofgas,andthequantity LX,bolh1/Ti,normalizedaccordingtoEq.(4). evolved it with redshift according to the Ghosh & White (2001) model to the predicted halo luminosities, the agree- (2001) model. An addition from otherdiscand bulgeX-ray mentwiththeCDF-Nresultsisfairly good,ourpredictions sources of 2×1039 erg s−1, assumed constant for simplic- fortheintegratedemissionthenbeingconsistentwithresults ity, is included in order to match the observed total MW from thephotometric CDF-N sample. disc luminosity. The resulting disc luminosity of an MW- There is an indication, however, that the Z = 0.3Z⊙ likegalaxyiscomparedinFig.3(left)tothenominallimits haloes, in particular gal18, could be slightly too X-ray lu- in different redshift bins from the CDF-N sub-samples of minous at the relevant redshifts. Whilethis is certainly not galaxieswithspectroscopically confirmedandpurelyphoto- a large excess, we see at least two possible explanations for metric redshifts, respectively. Typical uncertainties on the this. Possibly gal18 is peculiar. It certainly displays a high observed X-ray luminosities are displayed for comparison; LX at z = 0 for its circular velocity and metallicity when the1σstatisticalerrorsare∼20percent(A.Hornschemeier, compared to othergalaxies of thestudy presentedin Paper priv.comm.), verysimilar at therelevant redshifts tothose I, supporting this idea. Alternatively, Fig. 3 may suggest of our simulated galaxies. Adopting the mean LX in the that assuming halo metallicities of Z = 0.3Z⊙ at all red- sampleofShapley,Fabbiano & Eskridge(2001)asrepresen- shifts probed here is unrealistic, overpredicting the result- tative of the typical spiral LX at z =0 (comparable to the ing X-ray emission at the CDF-N redshifts. The fact that MW value adopted here), Hornschemeier et al. (2002) note gal15,whichatz=0isverytypicalofthegalaxies studied that the observed increase in LX from z = 0 to z = 0.6 is in Paper I, is also close to the CDF constraints at z > 1 consistentwiththepredictionoftheGhosh & White(2001) would seem to substantiate this possibility. While only in- model.WhentakingtheCDFvaluesasupperlimits, Fig.3 dicative,sucharesultwouldnotbesurprising;asarguedby suggests that this consistency can be extended to redshifts e.g.Renzini(2003),theglobalmetallicityoftheUniverseat z ≈ 1.5, provided that LX of the other disc X-ray emit- z ≈3 is likely to be ∼ 0.1Z⊙ rather than ∼0.3Z⊙. On the ting components, such as hot diffuse gas in the disc and other hand, for haloes at z = 1, the predicted unabsorbed bulge, does not increase dramatically between z ≈ 0.5 and 0.5–2 keV flux originating inside a region 1×1 arcsec2 on ≈ 1.5. Some increase in diffuse disc luminosity is probably the sky (within the central ∼ 10×10 kpc of the halo) is expected, however, due to supernova driven feedback asso- ∼2×10−19 and∼6×10−19 ergcm−2s−1fortheZ =0.3Z⊙ ciated with theincrease in star formation with redshift. haloes of gal15 and gal18, respectively. This is well below Cohen (2003) has recently shown that the CDF-N spi- the CDF-N point source flux for normal spirals at this red- rals out to z = 1.4 follow an LX–SFR relationship com- shift,∼(3−6)×10−18 ergcm−2 s−1 (Hornschemeier et al. parable to that seen for local galaxies, albeit with larger 2002).Note,however,thatwe havenotattempted toquan- scatter. This result supports the approach employed by tifyanysystematicuncertaintiesonthepredictedintegrated Ghosh & White (2001) in their XRB model. Motivated by emission (related to the choice of SFR model, the coupling this, and by Fig. 3 (left), we compare in Fig. 3 (right) the between LX and SFR, the assumed XRB fraction of disc CDF constraints to our halo predictions, having added the LX etc.). For example, it is not obviously appropriate to scaled Ghosh & White (2001) model (plotted as upwards use a Madau-type SFR profile to extrapolate the SFR of arrows) to the upper limit of the highest predicted halo lu- individual galaxies to high redshifts. Taking again the MW minosity at any given redshift. Uncertainties on the CDF as an example and neglecting variations caused by short- values are here omitted for clarity. As can be seen, the term starbursts, evidence suggests that the MW disc has Z =0.0 galaxies are easily consistent with theCDF-N con- experienced a roughly constant SFR back to at least z ∼1 straints, whereas the metal–rich galaxies are much closer (Rocha-Pintoet al. 2000). Assuming a constant SFR, the to the observed values. When adding the Ghosh & White Ghosh & White (2001) model predicts a steady decline in 6 J. Rasmussen et al. 10.00 10.00 Gal15, Gal18, ) 1.00 ) 1.00 s s g/ g/ r r e e 400 400 Gal15, 1 1 ( ( L X 0.10 L X 0.10 Gal18, 0.01 0.01 0.0 0.5 1.0 1.5 2.0 0.0 0.5 1.0 1.5 2.0 Redshift Redshift Figure 3. Left: Chandra Deep Field-North results (0.5–2 keV) on L ≈ L∗ galaxies, along with the predicted evolution in disc X-ray luminosityofaMW-likegalaxy,assumingthatX-raybinariesdrivetheevolution(Ghosh&White2001;Warwick2002).Solidlineisthe CDF-N subsample of galaxies with spectroscopic redshifts, dashed line the subsample with photometric redshifts only, with typical 1σ errorsoverplotted. Right:CDF-Nresultsandourpredictedintrinsic0.5-2keVhaloluminosities.Dotted lineisgal15, whiledot-dashed lineis gal18, with metallicities as labelled. Upwardarrows show the shiftinLX resulting fromadding the discluminosities of the left figure. discLX from z=0toz=2of∼30percent,inwhichcase telescopeeffectiveareasandbackgroundlevels.Calculations ourpredictionsfortheZ =0.3Z⊙galaxiesdonotviolatethe were repeated for the 0.1–3 keV band, which, in this con- CDF constraints, but, in fact, show even better agreement text, is likely to be relevant to the next-generation XEUS with these values. mission, being well-calibrated down to energies at least as We finally note that when adding the Ghosh & White lowas0.1keVandathighenergiesprovidingamuchhigher (2001)modeltotheZ =0.0haloes,theintegratedemission S/N than e.g. XMM-Newton. iswellbelowthatoftheCDFvaluesandsoiseasily consis- The0.3–2 keVprofilesareshowninFig.4forthemost tentwiththesewhenviewed asupperlimits. Predicted and observed luminosities can in this case be reconciled if us- optimistic case of Z =0.3Z⊙ edge-on galaxies. An immedi- ate conclusion, apart from the apparent fact that resulting ingtheGhosh & White(2001)modelalongwith adifferent fluxesareexceedinglylow,isthatatallredshiftsthemajor- SFRprescription,e.g.the’hierarchical’modelofBlain et al. ity of emission originates within the innermost 10–15 kpc, (1999). aswasalsofoundtobethecaseatz=0(PaperI).Atleast twoadditionalinferencesfromtheseplotsareworthnoting: (1) Due to the strong luminosity evolution of the Z = 5 PROSPECTS FOR HALO DETECTION 0.3Z⊙ haloes (Fig. 1), the optimum redshift for observing such haloes in terms of their surface brightness is not nec- In order to assess the possibility of detecting the X-ray haloes of MW-sized discs with current and futureX-ray in- essarily z =0 but is not well-defined either. If LX depends as strongly on redshift as for gal15, the optimum redshifts strumentation, we have computed X-ray brightness profiles could be z &0.5, depending on the assembly history of the perpendicular to the discs of the simulated galaxies. The galaxy.FortheZ =0.0galaxies(notshown)thereisalarger discs have been oriented edge-on, and observer-frame sur- span between the flux at different redshifts, with z = 0 al- face brightness (energy flux per unit solid angle) and flux ways being the optimum redshift. The fluxes for haloes at profiles as a function of off-disc distance have been calcu- z > 1 are in all cases lower than those shown (this is also lated in 40 kpcwide and 5–8 kpchigh slices parallel to the thecase in the0.1–3 keV band). disc. Spectra were calculated for each SPH particle in the rest-frame band matching the observer-frame band of in- (2) To unambiguouslydetect thehaloes in imaging ob- terest, and the resulting total surface brightness has been servations, we estimate that a firm detection at least ∼ 10 diminished by a factor (1+z)4 to account for cosmological kpc above the disc is required in order to avoid confusion dimming.Galacticabsorptionhasbeenincorporatedassum- with disc emission. Taking gal15 as representative, it can ingan absorbing column densityat z =0of NH =3×1020 be inferred from the bottom panel of Fig. 4 that detection cm−2, using the photoelectric absorption cross sections of of haloes of MW-like galaxies at cosmological distances is Morrison & McCammon (1983). Profiles have been calcu- clearly beyond the capabilities of current instrumentation, lated in the 0.3–2 keV band, expected to provide the op- despitetheexpectedredshiftevolutionofLX.Evenwiththe timum signal-to-noise (S/N) ratio for Chandra and XMM- sub-arcsec spatial resolution of Chandra and point source Newton, given the predicted halo temperatures, as well as sensitivities exceeding those of the Chandra Deep Fields by High-redshift X-ray properties of the haloes of simulated disc galaxies 7 -14 -14 10 10 Gal15 Gal18 -2m] 10-15 -2m] 10-15 c c -2n 10-16 -2n 10-16 mi mi arc 10-17 arc 10-17 -1s -1s g 10-18 g 10-18 r r e e ) [ -19 ) [ -19 z 10 z 10 ( ( S S -20 -20 10 10 0 10 20 30 40 50 0 10 20 30 40 50 |z| [kpc] |z| [kpc] -16 -16 10 10 Gal15 Gal18 -17 -17 10 10 -2m] -2m] c -18 c -18 -1s 10 -1s 10 g g r -19 r -19 e 10 e 10 x [ x [ u -20 u -20 al fl 10 al fl 10 ot -21 ot -21 T 10 T 10 -22 -22 10 10 1 10 1 10 |z| [arcsec] |z| [arcsec] Figure 4. Top panel: 0.3–2 keV surface brightness of the Z = 0.3Z⊙ galaxies inside 40 kpc wide slabs oriented along the disc, as a function ofvertical distance |z|to the disc.The galaxies areviewededge-on; foreach vertical bin,the values above andbelow the disc havebeenadded. Bottom panel: Correspondingfluxprofiles(surfacebrightness integrated overthe skyareacoveredbytheslabs),asa functionofobservedangularseparationfromthedisc. a factor of a few, a direct detection of halo emission would adopted response assumes a four–spacecraft configuration not be possible. with a total effective area of 15,000 cm2 at E = 1.25 keV. With an expected spatial resolution comparable to The source was modelled as a combination of a T = 0.3 XMM-Newton,thenext-generationmission Constellation-X keVthermalmekalplasmawithZ =Z⊙ (representingdisc (e.g. White& Tananbaum 1999) will not be able to image diffuse emission), a Γ = 1.9 power-law (representing point the haloes directly either. However, normal spiral discs in sources), and a mekal plasma representing the hot halo, the nearby Universe typically exhibit two distinct spectral with each component normalized according to Figs. 1 and features,aT ∼0.2–0.3keVthermalcomponentfromdiffuse 3. We find that a 1 Ms observation will gather around 300 disc emission and a power-law from the point source popu- source counts over the full detector band (0.25-10 keV), of lation (e.g. Kuntzet al.2003;Swartz et al. 2003).Onemay which only ≈ 10 will originate in the halo. These numbers therefore ratherseek totest theexistenceof thehaloes and will be down by an order of magnitude at z = 1. It thus theirpredicted luminosity evolution byfindingevidence for seemsimpossiblethatConstellation-Xshouldbeabletode- anadditionallow-temperaturethermalcomponentinthein- tect the haloes within reasonable exposuretimes. tegratedspectraofdiscgalaxies. Toexplorethispossibility, Fromthe0.1–3keVprofiles(notshown),onefindsthat we generated artificial source spectra for a Constellation-X observation of gal15 (z = 0.3), using xspec v11.0 along extendedenergycoveragedoesresultinlargerobservedflux, withtheresponsematrixforthecalorimeter detector1.The but the increase in emitted flux when including lower ener- gies is nearly balanced by Galactic absorption. Depending on redshift, the increase in observed flux from 0.3–2 keV 1 Availablefromhttp://constellation.gsfc.nasa.gov/docs/ to 0.1–3 keV is at most a factor of 1.5–2 for the Z = 0.0 8 J. Rasmussen et al. galaxies and even less for Z = 0.3Z⊙, even for a relatively (2000), onearrives at adensity of ≈10 haloes deg−2 in the low column density NH as the one adopted. However, the adopted cosmology. Given our assumptions this would be a Wide-FieldImageronboardthefinalconfigurationofXEUS, conservative lower limit, since some haloes beyond z = 0.3 withaspatialresolutioncomparabletothatofChandraand would also be detectable (those with Vc well beyond 220 an improved sensitivity particularly at low energies (an ex- km s−1), as would some at z < 0.3 (where our inclination pected limiting point-sourcefluxof4×10−18 ergcm−2 s−1 requirement,i&80◦,is severely relaxed). for a 100 ks observation; Bleeker & M´endez 2002), should observe ∼6000 source counts in 1 Ms for gal15 at z =0.3, ofwhich∼300wouldbefromthehalo.Whilethisisnotsuf- 6 SUMMARY AND CONCLUSIONS ficient towarrant theintroduction of an additional thermal componentinaspectralfit,thehaloshouldbedetectableas FromtheX-raypropertiesofthehotgashaloesoftwoMilky extendedemissionsurroundingthediscinsoft-bandimages, Way–like disc galaxies extracted from a cosmological simu- afterremovalofdiscandobviouspointsourceemission. We lation, run both with primordial and intracluster chemical thus predict that XEUS should be able to single out halo composition, we find that halo X-ray luminosities increase emission of highly inclined disc galaxies out to z ∼0.3, but roughly an order of magnitude from z =0 to z =1, evolv- it will require exposure times comparable to those of the ingmoregentlyathigherredshifts.Thereisgoodagreement CDF.Atz≈1,about120halocountsoutof300intotalare between the redshift evolution of LXh1/Ti and that result- expected.Althoughtheratio of haloto discemission is sig- ing from rough estimates of the disc accretion rate, as sug- nificantly higher at this redshift, the lower fluxand smaller gestedbysimpleanalyticalconsiderations.Thelogarithmof spatial extent will makehalo detection more difficult. thediscaccretionrateincreasesapproximatelylinearlywith The use of a different prescription for the disc SFR redshift out to at least z ≈1, from a typical value of ∼0.5 should not impact on the halo detection prospects, but the M⊙ yr−1 at z =0 to ∼3 M⊙ yr−1 at z =1. inclinationofthegalaxywill.Theaboveresultsallapplyto Whenaddedtoaconstantcontributionfromdiffusegas edge-on galaxies, i.e. a galaxies in which the polar axis is inthediscandbulgealongwithamodelpredictingtheevo- inclined by i = 90◦ with respect to the line of sight. For a lution of LX from X-raybinaries, we find that the halo LX galaxy with smaller inclination, halodetection will bemore of the galaxies is consistent with values derived for spirals difficult,theeffectingeneraldependingonthe3-Ddistribu- in the CDF-N data. The luminosities of the Z = 0.3Z⊙ tionofhothalogas.Onemayestimatetheminimumgalaxy haloes suggest, however, that a constant halo metallicity of inclination at which a halo is still detectable. As a first ap- 0.3Z⊙ across the z ∼ 0–2 redshift range studied here over- proximation, we can simply evaluate the sky area covered predicts the level of X-ray emission. Although this result by the disc and assume that this area will effectively block is tentative rather than conclusive and depends on the as- allhaloemission.Thisseemsareasonableassumption,since sumed evolution in disc star formation rate, we speculate allhaloemissionoriginatinginfrontofthediscwillbehard that halo metal abundances would be somewhat lower in to distinguish from disc emission, while halo emission be- the z = 0−2 range than the typical metallicity of the in- hind the disc will be subject to the same effect as well as tergalactic/intracluster medium (IGM) at z . 1. Possible sufferfrom severeabsorption byneutralgas inthedisc.Al- explanationscouldincludeenrichedsupernovaejectafalling thoughthemethodonlytakesintoaccounttheareablocked back on to the disc (galactic chimneys), enriched gas being bythedisc,notthefactthatthehaloitselfwillshowasim- expelled out of the galactic gravitational potential (possi- ilar inclination, the error implied should be negligible for bly through supernova-drivensuperwinds during periods of largeinclinations. Assumingacirculardiscofradius15kpc strong star formation), or enriched gas being stripped from and employing the Z = 0.3Z⊙ halo of gal15 at different the haloes during galaxy-galaxy and galaxy-IGM interac- redshifts, we estimate that at an inclination of 80◦ around tions. half of thehalo emission will still bedetectable, whereas at We have assessed the possibility for detecting haloes i = 75◦ this number is down to ∼ 30 per cent. Adopting a of MW-like galaxies and shown that observations of such tolerancelevelof50percent,wethusrequiretheinclination haloes at cosmological distances will haveto await a future tobei&80◦,i.e.thediscshoulddeviatelessthan10◦ from generationofX-rayinstrumentation.WefindthatXEUSin an edge-on appearance for the halo to remain detectable. itsfinalconfigurationshouldbeabletodetecthaloemission This implies that ∼ 10 per cent of all MW-like discs at a out to at least z ≈ 0.3 in a 1 Ms observation, and that typical distance of z =0.3 should display a detectable halo thesurfacedensityofhaloesdetectableinsuchobservations in a 1 Ms XEUS observation. should be & 10 deg−2. In terms of surface brightness, the The haloes of massive (Vc &300 km s−1) disc galaxies optimalredshiftsfordetectinghaloescould infact be0.5< are predicted to have X-ray luminosity and surface bright- z<1.0, owing to theirluminosity evolution. ness at z ≃0 an order of magnitude larger than the haloes Itisworthemphasizingthattheseconclusionsarebased studied here (cf. Paper I). If the halo luminosities of such on a particular numerical simulation with a certain set of galaxies follow a similar behaviour with redshift to that input values; the aim was to assess the amount of asso- showninFig.1,andthegalaxyshowsaninclinationi&80◦, ciated X-ray emission in this specific case rather than to these haloes should in fact be detectable to redshifts z ∼1 test the effects of varying the physical parameters entering and less inclined haloes to well beyond z ≈0.3. We can es- in the simulation (some effort was devoted to this in Pa- timate the total surface density of detectable haloes on the per I). Given the observational difficulty in directly detect- sky by assuming that 10 percent of all haloes around discs ing the halo emission, the best way to test and improve on with Vc > 220 km s−1 can be detected to z = 0.3. Using thepredictionspresentedhereintheimmediatefuturemay the velocity function of spirals derived by Gonzalez et al. indeed be through additional simulations, perhaps coupled High-redshift X-ray properties of the haloes of simulated disc galaxies 9 with advances in our still incomplete understanding of the Mewe R., Lemen J.R., van den Oord G.H.J., 1986, A&A, formation of spiral discs. Higher numerical resolution and 65 511 yetmorerealisticinputphysicswithimprovedestimatesfor Morrison R., McCammon D., 1983, ApJ,270, 119 the important physical parameters such as baryon fraction Pietz J., Kerp J., Kalberla P.M.W., Burton W.B., Hart- and gas metallicity should refine the predictions presented mann D., Mebold U., 1998, A&A,332, 55 here. In particular, simulations of disc galaxy formation in Renzini A., 2003, in Mulchaey J.S., Dressler A., Oem- whichthegaschemicalevolutionisself-consistentlyincluded ler A., eds, Clusters of Galaxies: Probes of Cosmological mayshednewlightontheamountandphysicalstateofhot Structure and Galaxy Evolution, Cambridge Univ. Press halogasaroundnormalspiraldiscs(Sommer-Larsen,Porti- (astro-ph/0307146) nari&Romeo2003,inprep.).Withtheobservedintegrated Rocha-Pinto H.J., Scalo J., Maciel W.J., Flynn C., 2000, X-rayemissionfromspiralsactingasabasis,improvedmod- A&A,358, 869 elsfortheredshiftevolution ofdiscX-rayluminositywould Shapley A., Fabbiano G., Eskridge P.B., 2001, ApJS, 137, alternativelyallowtighterconstraintstobeimposedonhalo 139 emission. SidherS.D.,SumnerT.J.,QuenbyJ.J.,GambhirM.,1996, A&A,305, 308 Sommer-Larsen J., G¨otz M., Portinari L., 2003, ApJ, 596, 47 ACKNOWLEDGMENTS Strickland D.K., Heckman T.M., Colbert E.J.M., Hoopes C.G., Weaver K.A., 2003a, ApJS, submitted We are grateful to the referee for prompt and constructive (astro-ph/0306592) comments.ThisworkwassupportedbytheDanishNatural Strickland D.K., Heckman T.M., Colbert E.J.M., ScienceResearchCouncil(SNF),theDanishGround-Based Hoopes C.G., Weaver K.A., 2003b, ApJ, submitted Astronomical Instrument Centre (IJAF), the Carlsberg (astro-ph/0306598) foundation,andbyDanmarksGrundforskningsfondthrough Swartz D.A., Ghosh K.K., McCollough M.L., Pannuti its support for the establishment of the Theoretical Astro- T.G., Tennant A.F., Wu K., 2003, ApJS, 144, 213 physicsCenter. Toft S., Rasmussen J., Sommer-Larsen J., Pedersen K., 2002, MNRAS,335, 799 (Paper I) Wang Q.D., Chaves T., Irwin J.A., 2003, ApJ, in press (astro-ph/0308150) Warwick R.S., 2002, in Jansen F., ed., Proc. Symp., New REFERENCES Visions of the X-ray Universe in the XMM-Newton and Barger A.J., Cowie L.L., Bautz M.W., Brandt W.N., ChandraEra, ESA SP-488 (astro-ph/0203333) Garmire G.P., Hornschemeier A.E., Ivison R.J., Owen White S.D.M., Frenk C.S., 1991, ApJ, 379, 52 F.N., 2001, AJ, 122, 2177 White, N. E. & Tananbaum H., 1999, Astronomische Benson A.J,BowerR.G.,FrenkC.S.,WhiteS.D.M.,2000, Nachrichten,320, 280 MNRAS,314, 557 Birnboim Y., DekelA., 2003, MNRAS,345, 349 Blain A.W., Jameson A., Smail I., Longair M.S., Kneib J.-P., Ivison R.J., 1999, MNRAS,309, 715 Bleeker J., M´endez M., 2002, in Jansen F., ed., Proc. Symp., New Visions of the X-ray Universe in the XMM-Newton and Chandra Era, ESA SP-488 (astro-ph/0207283) Bregman J.N., HouckJ.C., 1997, ApJ, 485, 159 Cohen J.G., 2003, ApJ, in press (astro-ph/0307537) Dahlem M., Ehle M., Jansen F., Heckman T.M., Weaver K.A., Strickland D.K., 2003, A&A,403, 547 Ghosh P, White N.E., 2001, ApJ, 559, L97 Gonzalez A.H., Williams K.A., Bullock J.S., Kolatt T.S., Primack J.R., 2000, ApJ, 528, 145 HornschemeierA.E.,BrandtW.N.,AlexanderD.M.,Bauer F.E.,GarmireG.P.,SchneiderD.P.,BautzM.W.,Chartas G., 2002, ApJ,568, 82 Katz N., Keres D., Dav´e R., Weinberg D.H., 2003, in Rosenberg J.L., Putman M.E., eds, ASSL Conf. Proc. Vol.281,TheIGM/GalaxyConnection:TheDistribution of Baryons at z = 0, Kluwer Academic Publishers, Dor- drecht,p. 185 Kuntz K.D., Snowden S.L., Pence W.D., Mukai K., 2003, ApJ,588, 264 McLure R.J., Dunlop J.S., 2002, MNRAS,331, 795 Madau P., Pozzetti L., Dickinson M., 1998, ApJ, 498, 106