arXiv:astro-ph/0202241v1 12 Feb 2002 ot hwta iia eainmgthl loa high at also hold might relation (0 similar a Star that the show North deep to from related obtained directly that Chandra results suggested is Preliminary is Rate. emission It Formation X-ray ones. and hard re- far luminosity the linear and X-ray radio tight keV the a 2–10 both that the argue between we holds sample lation galaxies, well-defined forming a Bep- star of of of observations use ASCA Making and relation. linear poSAX tight a follow galaxies nrrd aais–msin:BpoA,ASCA, – BeppoSAX, galaxies missions: continuum: – radio galaxies infrared: – galaxies X-rays: words: Key S P48 uut20 d.F asn&TBD & Jansen F. eds. 2002 August SP-488, ESA Netherlands The ESTEC, F = SFR ra- as: and written SFR be can between luminosities For- relation dio/FIR (1992) Star the Condon (1998), global Following galaxy. Kennicutt the a and of in indicators (SFR) be Rate mation to assumed are ties msin(awt&Pcn 95 eo ta.18) Since 1985). al. et Helou ( 1975; massive Pacini & synchrotron (Harwit observed ac- the emission may reradiated producing explosions their electrons and the of while grains celerate band, fraction dust infrared by a the mas- absorbed in dust: of is in presence radiation embedded the UV Condon to stars 1985; due young al. as et sive, interpreted Jong is lumi- De This in 1973; 1992). a magnitude Kruit der of show (van orders to nosity four known of spanning are luminosities relationship galaxies (FIR) infrared forming far star and continuum Radio rc Symposium Proc. ihteFRflxdfie fe eo ta.(95 as: (1985) al. et 1 Helou = after FIR defined flux FIR the with = SFR where otXryeiso,oiiae nhtpamsassociated plasmas hot in originated emission, X-ray soft Jy. in fluxes frared . 2 ai n a nrrdlmnste fsa forming star of luminosities infrared far and Radio trfrigglxe r nw oso luminous show to known are galaxies forming Star . L z 2 4 . L 1 . 26 . . M n ai bevtoso h ubeDe Field Deep Hubble the of observations radio and . 2 0 L 4GHz 1 . 1 · · FIR 4GHz · & . 10 10 H –0KVLMNST SASA OMTO AEINDICATOR RATE FORMATION STAR A AS LUMINOSITY KEV 2–10 THE 10 ). 2) M 5 43 28 − si r s erg in is 1 11 iatmnod srnma nvri` iBlga i Ra via Bologna, Universit`a di Astronomia, di Dipartimento M M NwVsoso h -a nvrei h M-etnadChan and XMM-Newton the in Universe X-ray the of Visions ‘New (2 ⊙ 2 .Introduction 1. NF–Osraoi srnmc iBlga i azn ,I 1, Ranzani via Bologna, di Astronomico Osservatorio – INAF ⊙ ⊙ tr r hr-ie,teeluminosi- these short-lived, are stars ) . 58 y (2) (1) /yr /yr Abstract S 60 − 1 µ Hz + S − 100 1 , µ L r s erg ) .Ranalli P. FIR nerg in − 1 cm ih linear tight − 1 1 .Comastri A. , Chandra − 2 n in- and (3) rgtGlxe ySnae&Tman(91 (RSA) (1981) of Tammann Catalog & declination Shapley-Ames Sandage with Revised by (1997) the Galaxies al. in Bright et galaxies Ho of by sur- vey spectroscopic spectra complete a nuclear represents HFS97) optical (hereafter of atlas The band. X-ray keV) (2–10 hard the study into the extend relations to these possible of make galaxies forming star of 1990). Padovani & (Griffiths e)rlto a on ewe I n ot(0.5–3.0 soft and the FIR by measured between luminosities found X-ray keV) was relation dex) 2 osa omn ein n aatcwns o linear non A winds. galactic and ( regions forming star to htteFCol oestesywt aatclatitude galactic with sky the notice covers FSC; only | 1989, FSC al. the et that (Moshir Catalogue Source forming Faint star of “H typical as labeled ratios are line in- systems (1987); nuclear line Osterbrock of with & basis Veilleux galaxies the to on according HFS97 ratios in tensity classified are spectra cal 0kV -a ula pcrmi oiae yastrong a by dominated (0.5– is broad-band spectrum its nuclear since X-ray sample keV) the 10 (M33) in object included and One not only. spectra was absorption best-fit Galactic from in bands for calculated keV corrected 2–10 were and Fluxes 0.5–2.0 confusions. look the Survey) to source order Sky in possible images for (Digital cm) (20 radio optical available, im- where against and, The checked files. were were event spectra ages pipeline-screened and the Images re- from data. we extracted archival available, ASCA not were the data duced published where cases few catalogues. 1996) from (1990, obtained al. were et fluxes any Condon GHz) add (1.4 them the to Radio include sample. not loose the did too we in are thus limits and information, detected. upper significant not flux were keV observations 2–10 ASCA of The in view objects of additional field Four instruments. the GIS with band or keV MECS 2–10 the the in archives. detected were ASCA then galaxies and Eighteen was BeppoSAX which the sample”) with cross-correlated “parent the (hereafter galaxies 60 at eeu apeo 9 ery( nearby 193 of sample geneous b L | X > epSXadAC bevtoso ial sample sizable a of observations ASCA and BeppoSAX ecoscreae h F9 apewt h IRAS the with sample HFS97 the cross-correlated We oto h aahv led enpbihd na in published; been already have data the of Most 2 n .Setti G. and , ∝ 10 µ ◦ n . ya 100 at Jy 1.0 and L n scmlt ont iiigflxso . Jy 0.2 of fluxes limiting to down complete is and FIR 0 . 6 zn ,I417Blga Italy Bologna, I–40127 1, nzani n uhsatrd(iprino about of (dispersion scattered much and ) .Telclsample local The 2. > δ 417Blga Italy Bologna, –40127 1 r Era’ dra 0 ◦ n magnitude and µ ii bann opeehomo- complete a obtaining ) nuclei”. 63 oebr2001, November 26–30 < z 0 . B 1 trforming star 01) Einstein T < 12 . .Opti- 5. satellite 1 2

Figure 1. The 0.5–2.0 keV luminosity of local star forming is related to radio and FIR ones. Straight line: best-fit relation for the local sample. Dotted line: best-fit for the local+supplementary samples. Squares: local sample. Crosses: supplementary sample. variable source (M33 X-8) identified as a black hole candi- Following Helou et al. (1985) we also calculate the date (Parmar et al. 2001). Although M33 is identified by mean ratio q between the logarithms of FIR and radio HFS97 as an H ii nucleus, it has a very low SFR (∼ 0.009 fluxes, obtaining q ≃ 2.05 with a variance σ ≃ 0.23. These M⊙/yr) so that the spectral signatures related to star for- values can be compared with the mean q =2.34 ± 0.01 for mation can be easily hidden by a single powerful source. the 2809 galaxies in the IRAS 2 Jy sample by Yun et al. Our sample (“local sample”) consists of the 17 galax- (2001). Although our sample shows a lower q, this result ies listed in Tab. 1. It contains 11 galaxies with SFR > may not be statistically significant. 1 M⊙/yr out of 77 in the parent sample (14%), and 5 out As a further test we checked the soft X-ray/FIR rela- of 27 with SFR > 3 M⊙/yr (19%). Since the parent sam- tion (Fig. 1), finding ple is complete within the statistical errors, these numbers represent an estimate of our sample completeness. All sta- Log(L.5−2) = (0.90 ± 0.10) Log(LFIR) − 3.2 ± 0.4 (6) tistical tests presented here are performed on this sample Log(L − ) = (0.90 ± 0.10) Log(L )+14.1 ± 2.9 (7) unless otherwise stated. .5 2 1.4 In the figures we also plot data for 6 other well-known starburst galaxies which were not in the HFS97 survey with mean δ’s of 58% and 46%, respectively. 0.95±0.06 because they are in the southern emisphere. On the basis Our result is consistent with the L0.5−4.5keV ∝ LFIR of their line intensity ratios they should be classified as relation found by David et al. (1992) for normal and star- H ii nuclei. In Tab. 1 we label them as “supplementary sample”. Table 1. Galaxies in the local sample. All galaxies were observed by ASCA, except those marked with * observed by BeppoSAX. 3. X-rays and the Rate Local Sample As a preliminary test, we perform a least-squares analysis M82* NGC 2276 NGC 4449 for the well-known radio/FIR correlation, which yields M101 NGC 2403 NGC 4631 M108 NGC 2903 NGC 4654 Log(LFIR)=(0.94 ± 0.09) Log(L1.4)+16.4 ± 2.5 (4) NGG 891 NGC 3310 NGC 6946 The dispersion around the best-fit relation is estimated NGC 1569 NGC 3367 IC 342 as: NGC 2146 NGC 3690 δ =1/N · X |(Lobs − Lpred)/Lpred| (5) Supplementary Sample where Lpred is the luminosity expected from the best fit relation and Lobs the observed one. For the radio/FIR NGC 55 NGC 1672 NGC 3256 correlation (eq. 4) one has δ ≃ 43%. NGC 253* NGC 1808 Antennae 3

Figure 2. The 2–10 keV luminosity of local star forming galaxies is linearly related to radio and FIR ones. Symbols as in fig. (1). burst galaxies from the IRAS Bright Galaxy Sample, but is inconsistent with the much flatter relationship obtained by Griffiths & Padovani (1990) for a sample of IRAS se- 0.62±0.14 lected galaxies (L0.5−3keV ∝ L60µ ) and for a sample 0.70±0.12 of starburst/interacting galaxies (L0.5−3keV ∝ L60µ ). The inclusion of the supplementary objects (Tab. 1) in our analysis would lead to a slight flattening of the 0.84±0.08 slope, i.e. L0.5−2.0keV ∝ LFIR ; likewise, if we use the 60µ luminosity instead of FIR, we obtain L0.5−2.0keV ∝ 0.83±0.10 L60µ with δ = 64%. However these results are still consistent with slopes derived in eq. (6) and (7) within 1σ. On the other hand the radio/FIR/hard X-ray relation is definitively linear.In Fig. (2) we plot 2–10 keV luminosi- ties versus FIR and radio ones. Least-squares fits yield:

Log(L2−10) = (1.08 ± 0.09) Log(LFIR) − 3.8 ± 0.3 (8)

Log(L2−10) = (1.04 ± 0.11) Log(L1.4)+10.1 ± 3.4 (9) with mean δ’s of 51% and 69%, respectively. The linearity Figure 3. The radio/X-ray luminosity relation holds in the and the dispersion are not significantly changed neither Hubble Deep Field too. Open symbols as in fig. (1). Filled cir- by the inclusion of the supplementary sample (L2−10 ∝ 0.98±0.07 0.98±0.07 cles: deep sample. Line: best fit for local galaxies (eq. 11). The L and L2−10 ∝ L , δ = 50% and 57% re- FIR 1.4 two galaxies with highest luminosity are both at z ≃ 1.2 spectively), nor by the use of the 60µ luminosity (L2−10 ∝ 1.00±0.11 L60µ ). It is also worth noticing that while the soft X-ray re- lations involve some further uncertainties related to the ever this seems unlikely for our sample because of the se- possible presence of intrinsic absorption, this is negligible lection of the objects by optical spectroscopy and of their 42 in the 2–10 keV band for column densities usually found moderate X-ray luminosity (< 10 erg/s). 22 in normal galaxies. A high column density (NH & 10 ) The existence of a tight linear relation implies that could partially obscure a possible AGN contribution; how- the three considered bands all carry the same information. 4

Since the radio and far infrared luminosity are indicators and found that there is a tight linear correlation between of the SFR, the 2–10 keV luminosity should also be an the X-ray luminosities in the 2–10 keV interval with both indicator of SFR. the radio and the FIR luminosities, normally assumed as By repeating the best fit procedure under the assump- the indicators of the star formation rate. We conclude that tion of strict linear relationships, consistent within 1σ with the origin of the hard X-ray emission must be closely re- the result of eq. (8) and (9), we find: lated to star formation. By selecting candidate starburst galaxies from the HDF Log(L2−10) = Log(LFIR) − 3.51 (10) North, whose redshift range extends up to z ∼ 1.2, we Log(L2−10) = Log(L1.4)+11.12 . (11) find that their X-ray (L2−10) and radio (L1.4) luminosi- From eq. (1) and (2) we propose: ties closely follow the same linear relationship derived for −40 the local sample. This linear correlation encompasses five SFR = 1.7 · 10 L2−10keV M⊙/yr (12) orders of magnitude in luminosity, up to L2−10 ∼ sev- 42 −1 with L2−10 keV in erg/s. eral 10 erg s and a corresponding star formation rate ∼ 1000 M⊙ yr−1. We also notice that for the highest lu- 4. Star-forming galaxies in the Hubble Deep Field minosity galaxies (which have z ≃ 1.2) the observed 2–10 keV band corresponds to a rest-frame 4.5–22 keV, where The 1 Ms Chandra (Brandt et al. 2001) and the radio 8.4 the emission is likely to be non-thermal. A plasma hot GHz (Richards et al. 1998) catalogs of the Hubble Deep enough to dominate the emission in this band would pose Field North (HDF) reach a limiting flux which is suffi- many problems in heating and bounding the gas. ciently deep to detect star-forming galaxies at up The way to understand the physics involved in hard X- to z ∼ 1.2, and can be used to check wether the radio/X- ray emission must go through a careful analysis of Chandra ray relation holds also for distant galaxies. and XMM-Newton observation, which have the necessary Radio and X-ray positions were matched with an en- spatial resolution and spectral sensitivity. Moreover, since circling radius of 1.1′′. All objects classified by Richards the radio/FIR relation holds also locally in galaxies down et al. (1998) as AGN, or with an elliptical or undefined to scales of about 100pc, it is worth remarking that the morphology were excluded; all other objects are accepted two star-bursting nuclei of NGC 3256 closely follow the as “candidate starbursts”. Fluxes at 1.4GHz were calcu- hard X-ray/radio relation, as suggested by our preliminary lated making use of 8.4GHz data and spectral indexes work on the Chandra observation for this galaxy. The ex- measured for each object and reported in Richards et al. planation of the radio/FIR correlation is still a matter of (1998). Redshifts were taken from Cohen et al. (2000). discussion; we hope that these results may help in clarify- Fluxes in the 2–10 keV were derived from a power-law fit ing this issue. We are currently analysing possibilities for to the observed 0.5–8 keV counts by Brandt et al. (2001). a theoretical interpretation which, along with further con- Radio and X-ray k-corrections were calculated for a range straints on this observational results, will be the subject of spectral indexes and redshifts; we chose to use mean k- of a forthcoming paper. corrections since the induced error would not significantly affect the rest-frame fluxes. References We find that radio and X-ray luminosities of candidate starburst galaxies at larger redshifts follow the same rela- Brandt W.N. et al. 2001, AJ 122, 2810 tion we observe in the very local universe (Fig. 3), which Cohen J. et al. 2000, ApJ 538, 29 Condon J.J. 1992, ARA&A 30, 575 now spans 5 orders of magnitude. Condon J.J. et al. 1990, ApJS 73, 359 Condon J.J. et al. 1996, ApJS 103, 81 5. Conclusions David L.P., Jones C. & Forman W. 1992, ApJ 388, 82 De Jong T. et al. 1985, A&A 147, L6 We have analyzed a small but well defined sample of 17 Griffiths R.E. & Padovani P. 1990, ApJ 360, 483 star forming galaxies for which there is a homogeneous Harwit M. & Pacini F. 1975, ApJ 200, L127 information on optical, FIR, radio and X-ray bands (lo- Helou G., Soifer B.T. & Rowan-Robinson M. 1985, ApJ 298, L7 cal sample). The BeppoSAX and ASCA X-ray data have Ho L.C., Filippenko A.V. & Sargent W.L.W. 1997, ApJS 112, 315 been corrected for Galactic absorpion only. In agreement Kennicutt R.C. Jr. 1998, ApJ 498, 541 b > with a previous work (David et al. 1992) we find that the Moshir M. et al. 1989, “IRAS Faint Source Catalog, | | 10 Degrees”, Infrared Processing and Analysis Center logarithms of the soft (0.5–2 keV) X-ray luminosities are Parmar A.N. et al. 2001, A&A 368, 420 linearly correlated with the logarithms of either radio (1.4 Richards E.A. et al. 1998, AJ 116, 1039 GHz) and FIR luminosities, and that within the statistical Sandage A.R. & Tammann G.A. 1981, A Revised Shapley- errors these relationships are consistent with linearity be- Ames Catalog of Bright Galaxies tween the corresponding luminosities. We have extended van der Kruit P.C. 1973, A&A 29, 263 our analysis to the harder X-ray band, essentially free from Veilleux S. & Osterbrock D. E. 1987, ApJS 63, 295 internal absorption which may affect the soft X-ray fluxes, Yun M.S., Reddy N.A. & Condon J.J. 2001, ApJ 554, 803