The evolution of star forming galaxies with the Wide Field X-ray Telescope
aa r X i v : . [ a s t r o - ph . I M ] A p r Mem. S.A.It. Vol. 1, 1 c (cid:13) SAIt 2008
Memorie della
The evolution of star forming galaxies with theWide Field X-ray Telescope
P. Ranalli Universit`a di Bologna – Dipartimento di Astronomia, via Ranzani 1, 40127 Bologna, Italye-mail: [email protected]
Abstract.
Star forming galaxies represent a small yet sizable fraction of the X-ray sky(1%–20%, depending on the flux). X-ray surveys allow to derive their luminosity functionand evolution, free from uncertainties due to absorption. However, much care must be putin the selection criteria to build samples clean from contamination by AGN. Here we re-view the possibilities o ff ered by the proposed WFXT mission for their study. We analyzethe expected luminosity and redshift distributions of star forming galaxies in the proposedWFXT surveys. We discuss the impact of such a mission on the knowledge of the cosmicstar formation history, and provide a few suggestions. Key words.
X-rays: galaxies – galaxies: luminosity function – galaxies: evolution – galax-ies: high-redshift – galaxies: spiral
1. Introduction
The X-ray luminosity of star forming galax-ies (SFG; they usually are spiral galaxieswithout AGN activity) appears to be a reli-able, absorption-free estimator of star forma-tion (Ranalli et al. 2003). This is justified onthe basis that the X-ray luminosities are lin-early and tightly correlated with the radio andFIR ones, which in turn are commonly usedas star formation rate (SFR) indicators. Thus,the X-ray emission of SFG may be consideredas a tool to investigate the cosmic star forma-tion history. To this end, the study of the X-ray luminosity function (XLF) of galaxies andof its evolution represents a necessary step.Ranalli et al. (2005, hereafter RCS05) built alocal ( z =
0) XLF of SFG and investigatedthe possibilities for evolution. In this paper,
Send o ff print requests to : P. Ranalli we build on the RCS05 XLF and methods toexplore the possible contribution of the WideField X-ray Telescope (WFXT) mission to ourunderstanding of the SFG content of the uni-verse, by analyzing the expected luminosityand redshift distributions.The WFXT is a proposed mission whichaims to perform very wide and moderatelydeep X-ray surveys. By taking a di ff erent ap-proach to mirror design than the classicalWolter type-1 (Burrows et al. 1992), it couldachieve a very large field of view ( ∼ )while maintaining a good angular resolution( ∼ ′′ ) and a large e ff ective area ( ∼ ) inthe 0.1-7 keV band (Conconi et al. 2010).Such a telescope would be able to ob-serve a number of X-ray sources far exceed-ing all those known today. While X-ray sur-veys mainly detect AGN, star-forming galax-ies (SFG) are also present, comprising a frac- Ranalli: star forming galaxies with the WFXT
Fig. 1.
Observed X-ray number counts in to-day’s surveys, and planned WFXT limitingfluxes. The thick upper line and the hornshow the observed Log N –Log S for allX-ray sources in the Chandra
Deep Fields(Moretti et al. 2003) and the limits from thefluctuation analysis (Miyaji & Gri ffi ths 2002).The bundle of histograms and data pointsshows several determinations of the star-forming galaxies Log N –Log S (see text). Thevertical lines illustrate the limiting fluxes of theplanned surveys. Table 1.
Covered area (deg ) and limitingfluxes (erg s − cm − in the 0.5–2.0 keV band)of the proposed WFXT surveys. wide medium deeparea 20000 3000 100flux (req.) 5 · − − − flux (goal) 3 · − · − · − tion in the range 1%–20% (depending on theflux) of all the sources detected in the 0.5–2.0keV band. Three major surveys are envisagedwith the WFXT, covering di ff erent amounts ofthe sky at di ff erent limiting fluxes and named wide , medium and deep (Fig 1). Their limitingsoft X-ray fluxes correspond broadly to thoseprobed in ROSAT (Tajer et al. 2005), XMM- Newton (Georgakakis et al. 2004) and deep
Chandra (Bauer et al. 2004; Norman et al.2004, RCS05) surveys of SFG. In Fig. 1 we show the total Log N –Log S from X-ray sur-veys, and di ff erent estimates of the SFG num-ber counts.Depending on technological developments,both a requirement and a goal value for the lim-iting fluxes can be quoted. Reaching the goalscould extend the number of detected objects bya factor ∼
5. However, given the early stage ofthe mission, here we will consider only the re-quirements, and regard the goal flux limit forthe deep survey only.We assume H =
70 km s − Mpc − , Ω M = . Ω Λ = .
2. The LF and evolution ofstar-forming galaxies
The local di ff erential luminosity function ϕ (Log L ) dLog L (1)is defined as the comoving number densityof sources per logarithmic luminosity interval.The evolution can be described as pure lumi-nosity with the form (Schmidt 1972) L ( z ) ∝ (1 + z ) η l . (2)Infrared surveys provide a powerfulmethod to select SFG, since the bulk of thefar and near IR emission is due to reprocessedlight from star formation, with AGN repre-senting only a minor population (de Jong et al.1984; Franceschini et al. 2001; Elbaz et al.2002). The FIR LFs may be assumed to beessentially una ff ected by a contribution fromSeyfert galaxies, as the fraction of Seyfertsis about ∼ µ LF derived from the IRASPoint Source Catalog Redshift (Saunders et al.2000) (PSC z ). It includes 15,411 galaxieswith z . .
07, covering 84% of the sky witha flux limit of 0.6 mJy at 60 µ . While T03reports pure-density evolution for their LF,pure-luminosity may provide an equally goodfit to the data (T. Takeuchi, priv. comm.).Other determinations of the SFG LF havebeen derived by the cross correlation of ra-dio surveys with optical ones (see references analli: star forming galaxies with the WFXT 3 Fig. 2.
Left:
IRAS, ISO, and radio local luminosity functions of SFG converted to the X-rays(see RCS05 for references to the individual LFs). All the LFs converge to the same location. Thelarge data points with error bars show an observational determination of the local XLF, basedon XMM-
Newton data by Georgantopoulos et al. (2005).
Right:
Comparison of the XLF derivedfrom IRAS data (solid curve; the grey area shows the uncertainty on the evolution) with the XLFderived by Norman et al. (2004) in the
Chandra
Deep Field (data points with error bars).in RCS05). The redshifts covered in these sur-veys are similar to those of the T03 galaxies,but the number of objects is smaller due to asmaller sky coverage, so reliable estimates ofthe evolution may not be derived.The local IR or radio LFs may be con-verted to X-ray ones by using the ap-proach first developed in Avni & Tananbaum(1986) (see also: Georgantopoulos et al. 1999;Norman et al. 2004), which may be sum-marised as follows. Given a galaxy with IR orradio luminosity L , let P ( L X | L ) be the proba-bility distribution of the possible values of thegalaxy’s X-ray luminosity L X , as given by theoptical / IR / radio vs. X-ray correlations. Thus,the X-ray LF may be otained by the convolu-tion of an optical / IR / radio LF with P ( L X | L ). InRanalli et al. (2003) it was reported that the X-ray luminosity is tightly correlated with radioand FIR luminosities. By assuming a Gaussianprobability distribution for these correlations,one has for example P (Log L . − | Log L µ ) == √ πσ e − Log L µ + . − Log L . − σ (3) with σ ∼ . z = –10 erg s − , encompassing theknee region after which all XLFs steepen to-ward higher luminosities; although departuresat lower and higher luminosities are present,the average local X-ray luminosity density, ∼ (3 · ± − Mpc − , appears to bewell defined.Norman et al. (2004) derived an XLF athigher redshifts (two bins: ¯ z ∼ .
27 and¯ z ∼ .
79; Fig. 2, right panel) than thoseprobed by the IR and radio surveys discussedabove. Other strong constraints at high redshiftcome from the COMBO-17 survey (Wolf et al.2003), and from the comparison of the ob-served X-ray Log N –Log S with that derivedby integrating the XLF. This work has beendone in detail in RCS05, and here we just quotethe results: the evolution is well described aspure-luminosity with an exponent η l ∼ . z ∼ Ranalli: star forming galaxies with the WFXT
3. Expected luminosity and redshiftdistributions with the WFXT
The XLF derived in the previous section canbe integrated in the volume of space probed bythe surveys to obtain luminosity distributionsd N dLog L = Z z max ϕ (Log L , z ′ ) min[ V ( z ) , V (Log L , F lim )] d z (4)and redshift distributionsd N d z = Z Log L max ( z ′ )Log L min ϕ (Log L , z ′ ) min[ V ( z ) , V (Log L , F lim )] dLog L (5)where z ′ = min( z , z stop ); F lim is the limiting fluxof the survey; V ( z ) is the comoving volume atredshift z ; and V ( L , F ) is the comoving volumeat the redshift at which a source with luminos-ity L is observed with flux F . All fluxes areconsidered in the 0.5–2.0 keV band.For the following calculations, we take z max = z stop = L min = max[10 ,4 π D lum ( z ) F lim ] erg s − and L max = (1 + z ′ ) η l erg s − . In words, this means that we inte-grate on the luminosity range (at z =
0) 10 –10 erg s − , that we allow the maximum lu-minosity to evolve with redshift, that we ex-clude luminosities lower than what could visi-ble given the redshift and limiting flux, and thatthe integration is done up to z = z stop =
1. The evolution ispure-luminosity as in Eq. (2) with η l = . · – 4 · objects per sur-vey). Reaching the development goal wouldenhance the number of SFG by a factor of ∼ · objects in the deep survey.It is important to check that the SFG XLFwill be well sampled at all luminosities. FromFig. 3 it is evident that at least 10 SFG with L < · erg s − should be detected in the medium and deep surveys, and that the “knee”region of the XLF (the range 10 –10 erg s − at z =
0, compare with Fig. 2) will be very well sampled with around 1 . · objects in eachof the medium and deep surveys. Similarly, thehigh luminosity tail ( L > erg s − ) willalso be well sampled with around 7 · ob-jects in the medium survey. This part of theXLF is especially important because objects inthis luminosity range are quite rare, and gen-erally suspected of having a substantial part oftheir emission due to an AGN. Refined clas-sification criteria, and the possibility of doingspectral analysis will clearly be essential.Reaching the development goal will en-large the sample of SFG with L < erg s − by a factor of ∼
5, while it should not makemuch di ff erence for brighter objects.The expected redshift distribution is shownin Fig. 4 (left panel). The wide and medium sur-veys should have redshift peaks around 0 . ∼ . < z < . wide survey, and ∼
900 in the range 0 . < z < . medium .The deep survey will probe much higher red-shifts: ∼
900 objects with 1 . < z < .
3, andother ∼
900 with 1 . < z < .
0; these num-bers would also be larger by a factor of ∼ η l = . up to whichredshift will the knee of the XLF be probed . Thelocal XLF exhibits its knee in the range 10 –10 erg s − (Fig. 2), thus we repeated the inte-gration in Eq. (5) taking L max = · (1 + z ′ ) η l erg s − . The result is shown in Fig. 4 (rightpanel). The wide and medium surveys will notprobe the knee of the XLF at redshift largerthan z ∼ . z ∼ .
2, respectively. The deep survey will extend the probed redshiftrange up to z ∼ .
5, while if the developmentgoal is reached, redshifts as large as ∼ . analli: star forming galaxies with the WFXT 5 Fig. 3.
Left:
Expected cumulative luminosity distributions for SFG in the WFXT surveys.
Right: Di ff erential luminosity distributions. Since the knee of the SFG XLF is comprised (at z =
0) inthe range 10 –10 erg s − , it appears that the XLF will be well sampled by the WFXT. Fig. 4.
Left:
Expected di ff erential redshift distributions for SFG in the WFXT surveys. The greyarea illustrates how the uncertainties about the XLF evolution could a ff ect the deep survey. Right:
Same as left, but only considering galaxies with luminosity L ≤ · (1 + z ′ ) η l erg s − : the kneeregion of the XLF will be probed up to z ∼ . Both:
Line styles as in Fig. 3.
4. Discussion
From the expected luminosity and redshift dis-tributions, it is evident that the WFXT will beable to determine the SFG XLF with an ac-curacy comparable to that of IRAS or opticalsurveys. Thus there will be many new possi-bilities to study how the X-ray emission de- pends on other parameters, such as morphol-ogy, colours, redshift, etc. However, such awork could only be made if multiwavelengthinformation is available. In fact, the first andmost important task will be the selection of theSFG, which are a minor fraction of the totalof X-ray surveys. Several di ff erent combina- Ranalli: star forming galaxies with the WFXT tions of the same basic parameters (X-ray lu-minosity, X-ray / optical flux ratio, hardness ra-tio, amount of absorption, presence of broadlines in optical spectra, etc.) have been ex-plored by di ff erent authors in deep Chandra surveys (RCS05, and references therein). Allthe determinations di ff er by up to a factor of ∼
2; this scatter can be reduced only witha better understanding of how these parame-ters are linked to each other, and how theya ff ect the selection (and the completeness ofsamples) of SFG. This only gets more di ffi -cult for wide-and-shallow surveys (respect todeep pencil-beam ones) because the SFG / AGNfraction in X-ray surveys depends on the limit-ing flux (Fig. 1). An attempt to investigate thisproblem for a sample of SFG in the
Chandra -COSMOS survey (Elvis et al. 2009) may befound in Ranalli et al. (2010, to be submitted).One of its main results is that no rigid bound-aries on the selection parameters can be put;a sensible approach should build on statisticalmethods for object classification.The need for the most complete multiwave-length coverage also requires that the choice ofthe sky areas covered by the WFXT surveys becoordinated with (or follow, if not possible oth-erwise) other present and future survey facili-ties (Pan-Starrs, the Large Synoptical SurveyTelescope, ALMA, LOFAR, E-VLA, etc.).The planned WFXT surveys will be able toderive the SFG XLF and determine its evolu-tion with unprecedented accuracy up to z ∼ . z ∼ . z . with a limiting flux of 10 − erg s − cm − would extend the coverage of the knee of the XLF up to z ∼ .
7, and of the high-luminosity tail up to redshifts well beyond thepeak of the cosmic star formation history.
Acknowledgements.
We thank Roberto Gilli andAndrea Comastri for stimulating discussions.
References
Avni, Y. & Tananbaum, H. 1986, ApJ, 305, 83Bauer, F. E., Alexander, D. M., Brandt, W. N.,et al. 2004, AJ, 128, 2048Burrows, C. J., Burg, R., & Giacconi, R. 1992,ApJ, 392, 760Conconi, P., Campana, S., Tagliaferri, G., et al.2010, MNRAS, 509de Jong, T., Clegg, P. E., Rowan-Robinson, M.,et al. 1984, ApJ, 278, L67Elbaz, D., Cesarsky, C. J., Chanial, P., et al.2002, A&A, 384, 848Elvis, M., Civano, F., Vignali, C., et al. 2009,ApJS, 184, 158Franceschini, A., Aussel, H., Cesarsky, C. J.,Elbaz, D., & Fadda, D. 2001, A&A, 378, 1Georgakakis, A. E., Georgantopoulos, I.,Basilakos, S., Plionis, M., & Kolokotronis,V. 2004, MNRAS, 354, 123Georgantopoulos, I., Basilakos, S., & Plionis,M. 1999, MNRAS, 305, L31Georgantopoulos, I., Georgakakis, A., &Koulouridis, E. 2005, MNRAS, 360, 782Miyaji, T. & Gri ffiffi