Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.stecf.org/goods/spectroscopy/CDFS_Mastercat/vanderwel04.pdf Äàòà èçìåíåíèÿ: Thu Sep 16 11:25:45 2004 Äàòà èíäåêñèðîâàíèÿ: Tue Oct 2 01:16:20 2012 Êîäèðîâêà:

The Astrophysical Journal, 601:L5­L8, 2004 Januar y 20
2004. The American Astronomical Society. All rights reser ved. Printed in U.S.A.

THE FUNDAMENTAL PLANE OF FIELD EARLY-TYPE GALAXIES AT z p 1 A. van der Wel,2 M. Franx,2 P. G. van Dokkum,3 and H.-W. Rix
Received 2003 October 3; accepted 2003 December 8; published 2004 January 19
4

1

ABSTRACT We present deep VLT spectra of early-type galaxies at z 1 in the Chandra Deep Field­South, from which we derive velocity dispersions. Together with structural parameters from Hubble Space Telescope imaging, we can study the fundamental plane for field early-type galaxies at that epoch. We determine accurate mass-to-light ratios (M/L) and colors for four field early-type galaxies in the redshift range 0.96 ! z ! 1.14 , and two with 0.65 ! z ! 0.70. The galaxies were selected by color and morphology, and have generally red colors. Their velocity dispersions show, however, that they have a considerable spread in M/L (by a factor of 3). We find that the colors and directly measured M/L correlate well, demonstrating that the spread in M/L is real and reflects variations in stellar populations. The most massive galaxies have M/L comparable to massive cluster galaxies at similar redshift and therefore have stellar populations that formed at high redshift (z 1 2 ). The lower mass galaxies at z 1 have a lower average M/L, and one is a genuine "E A" galaxy. The M/L indicate that their luminosityweighted ages are a factor of 3 younger at the epoch of obser vation, because of either a low formation redshift or late bursts of star formation contributing 20%­30% of the mass. Subject headings: cosmology: obser vations -- galaxies: evolution -- galaxies: formation
1. INTRODUCTION

Good understanding of the formation and evolution of earlytype galaxies is one of the major challenges for current structure-formation models. Models of hierarchical structure generally predict that field early-type galaxies form relatively late (e.g., Diaferio et al. 2001). One of the prime diagnostics of the formation histor y of early-type galaxies is the evolution of the M/L as measured from the fundamental plane (FP; Franx 1993). Studies of the evolution of the luminosity function, together with the evolution of M/L, quantify the evolution of the mass function. Previous studies of the evolution of the M/L have produced consistent results for the evolution of massive cluster early-type galaxies: the evolution is slow, consistent with star formation redshifts z 2 (e.g., van Dokkum & Stanford 2003). On the other hand, studies of the evolution of field galaxies have yielded more contradictor y results: whereas early studies found slow evolution (e.g., van Dokkum et al. 2001; Treu et al. 2001; Kochanek et al. 2000), more recently, evidence for much faster evolution was found by Treu et al. (2002) and Gebhardt et al. (2003); and yet other authors found that the majority of field early-types evolve slowly, with a relatively small fraction of fast-evolving galaxies (e.g., Rusin et al. 2003; van Dokkum & Ellis 2003; van de Ven, van Dokkum, & Franx 2003; Bell et al. 2003). These previous measurements suffered from several uncertainties: the signal-to-noise ratios (S/Ns) of the spectra were generally quite low, much lower than usual for nearby studies of the FP (e.g., Faber et al. 1989; JÜrgensen et al. 1996). Those studies, based on lensing galaxies, used stellar-velocity dispersions derived from image separations. In this Letter, we present high-S/N spectra and accurate measurements of the M/L of four field elliptical galaxies with
Based on obser vations collected at the European Southern Obser vator y, Chile (169.A-0458). 2 Leiden Obser vator y, P.O. Box 9513, J. H. Oort Building, Niels Bohr weg 2, NL-2300 RA, Leiden, Netherlands. 3 Department of Astronomy, Yale University, P.O. Box 208101, New Haven, CT 06520-8101. 4 Max-Planck-Institut fur Astronomie, Konigstuhl 17, D-69117 Heidelberg, ¨ ¨ Germany.
1

z p 1­1.14 and two at z 0.7 in the Chandra Deep Field­ South (CDF-S). The S/Ns are comparable to those obtained for nearby galaxies. Together with accurate multiband photometr y available for the CDF-S, we have measured the accurate M/L and rest-frame optical colors at z 1. Throughout this Letter we use Vega magnitudes and assume a L-dominated cosmology ([QM , QL ] p [0.3, 0.7]), with a Hubble constant of H0 p 70 km s 1 Mpc 1.
2. SPECTROSCOPY

2.1. Sample Selection and Observations The galaxies were selected from the COMBO-17 catalog (see Wolf et al. 2003), and imaging was obtained by the Great Obser vatories Origin Deep Sur vey5 (GOODS; data release ver. 0.5) from the Advanced Camera for Sur veys (ACS) on the Hubble Space Telescope. We selected compact, regularly shaped galaxies with photometric redshifts higher than 0.8 and I z ( I z)Sbc , z ! 22. ( I z)Sbc denotes the color of the Sbc template of Coleman, Wu, & Weedman (1980) at the photometric redshift. This template has (U V )zp0 p 0.95. The typical uncertainty in the color is 0.1 mag. Lower priority galaxies were included with either lower redshifts or later types. The CDF-S was obser ved in MXU mode with FORS2 on VLT-UT4 during three runs from 2002 September through 2003 Februar y, for a total of 14 hr. The 600 z grism (central wave° ° length 9010 A, resolution 5.1 A or jinst r p 72 km s 1) was used. During the obser vations toward the CDF-S the seeing varied between 0 .7 and 1 .5 , with a median seeing of about 1 . The sky was clear all the time. 2.2. Velocity Dispersions It turned out that 10 out of the 11 high-priority objects at z phot 1 have the spectrum of a quiescent galaxy; the other one has a bright [O ii] emission line. The brightest four (z 21.0, independent of the color) had sufficient S/Ns to perform reliable dispersion measurements. The rest-frame spectra of these four galaxies are shown in Figure 1. They all show
5

See http://www.stsci.edu/science/goods.

L5


L6

VAN DER WEL ET AL.

Vol. 601

Fig. 1.--Left: Unsmoothed rest-frame spectra of the six objects with velocity dispersions. Regions with bright sky lines are interpolated. The wavelength scale ° is interrupted at l p 4500 A. Right: ACS images (F850LP) of the four galaxies at z 1, and the residual images from the r1/4 fit. From left to right: 20950, 19990, 19375, and 22432.

° a strong 4000 A break and Ca lines. Balmer lines (especially the Hd line) are also present, although var ying in strength from object to object (see Table 1). Object 19375 is an "E A" galaxy, according to the criteria used by Fisher et al. (1998). For two ellipticals with 0.65 ! z ! 0.70 we also have sufficient signal to determine a velocity dispersion. Dispersions were measured by convolving a template star spectrum to fit the galaxy spectrum as outlined by van Dokkum & Franx (1996). We tested this procedure extensively, using different template stars and masking various spectral regions. The final values for the velocity dispersions (see Table 1) were obtained by masking the Ca H and K and Balmer lines and using the best-fitting template spectrum, which was a highresolution solar model spectrum6 smoothed and rebinned to match the resolution of the galaxy spectra. The Ca lines were not included in the fit because this greatly reduced the dependence of the measured velocity dispersions on template type. The tests using different templates and different masking of the Ca lines indicate that the systematic uncertainty is 5%. In order for the results to be comparable to previous studies, an aperture correction as described by JÜrgensen, Franx, & KjÔrgaard (1995b) was applied to obtain velocity dispersions within a circular aperture with a radius of 1 .7 at the distance of the Coma Cluster. This correction is 7%. This is the first extensive sample of such objects at z 1 0.9 with high S/N (see van Dokkum & Ellis 2003; Treu et al. 2002; Gebhardt et al. 2003, for other spectroscopic studies).
6

3. PHOTOMETRY

See http://bass2000.obspm.fr.

Photometr y and structural parameters were determined from the GOODS ACS images (data release ver. 1.0). Images are available in four filters (F435W, F606W, F775W, and F850LP), which we refer to as b, v, i, and z, respectively. For each object, the effective radius (re) and the surface brightness at the effective radius (me ) were obtained by fitting an r 1/ 4 profile, convolved by the point-spread function (PSF; van Dokkum & Franx 1996); z-band images were used for the z 1 objects, and i-band images for the z 0.7 objects. Stars were used as the PSF. The resulting values for re and me var y by 10% when using different stars, but also correlate such that the error is almost parallel to the FP (van Dokkum & Franx 1996). Therefore, the errors in our results are dominated by the errors in the velocity dispersions. The results are listed in Table 1. The images of the z 1 objects and the residuals of the fits are shown in Figure 1. To determine the i z and v i colors, fluxes were calculated from the r 1/ 4 model within the measured effective radius. To this model flux we added the flux within the same radius in the residual images. We corrected for Galactic extinction based on the extinction maps from Schlegel, Finkbeiner, & Davis (1998). The correction is extremely small: E( B V ) p 0.007. Rest-frame B-band surface brightnesses and rest-frame U V colors were obtained by transforming obser ved flux densities in two filters to a rest-frame flux density exactly as outlined by van Dokkum & Franx (1996). The spectral energy distribution used to calculate the transformations is the earlytype spectrum from Coleman et al. (1980). We found the same

TABLE 1 Photometric and Spectroscopic Properties ID 19375 19990 20950 22432 21376 22239 ...... ...... ...... ...... ...... ...... a (arcsec) 0 32 73 83 129 236 d (arcsec) 50 35 6 41 10 35 zspec 1.089 0.964 0.964 1.135 0.685 0.660 i 21.72 21.32 21.18 22.38 21.46 20.67

vi
1.71 1.89 2.07 2.00 1.77 1.67

iz 1.04 1.00 1.07 1.42 0.58 0.50

log re (kpc) 0.410 0.388 0.0085 0.109 0.447 0.842 0.012 0.007 0.026 0.040 0.001 0.002 21.75 21.45 22.95 23.35 22.16 19.83

me 0.05 0.03 0.08 0.13 0.04 0.03

S/N ° (A 1) 26 49 39 21 27 40

j (km s 1) 198 159 261 217 156 177 25 14 23 20 24 19

(Hg

Hd)/2 ° (A) 4.1 2.5 !1 !1 ... ...

[O ii] ° (A) 4.6
11 11

4.4 ... ...

Note.--Coordinates are in arcseconds east and north of R.A. p 03h32m25s, decl. p 27 54 00 . Errors in the magnitudes and colors are, respectively, 0.03 and 0.05 mag. Effective radii and surface brightnesses (mag arcsec 2 at re) are measured in the z-band for objects 19375, 19990, 20950, and 22432, and in the i-band for objects 21376 and 22239. The listed errors in the velocity dispersions are fitting errors and do not include a 5% systematic error.


No. 1, 2004

FUNDAMENTAL PLANE AT z p 1 FOR FIELD GALAXIES

L7

Fig. 2.--(a) FP points of the field galaxies presented in this paper (errors include a 5% systematic effect in the velocity dispersions), the FP points of cluster galaxies at z p 1.27 (van Dokkum & Stanford 2003), and the FP of the Coma Cluster (JÜrgensen et al. 1996). (b) Offsets in M/LB from the Coma Cluster FP (square; derived from JÜrgensen et al. 1996). Besides the z p 1.27 cluster, this figure also contains the data from van Dokkum et al. (1998) on the MS 1054 cluster at z p 0.83. The full, dashed and dotted cur ves are the model predictions for a single burst of star formation with a Salpeter (1955) initial mass function for redshifts 3, 2, and 1.5, respectively. Filled symbols indicate galaxies with masses M 1 3 # 1011 M,; other symbols indicate galaxies less massive than that. Crosses and squares distinguish between galaxies at z ! 0.8 and z 1 0.8 , respectively. All galaxies occupying the region below the z p 1.5 model cur ve are "E A" galaxies, except 19990, and have masses less than 3 # 1011 M,. The more massive galaxies have significantly older stellar populations.

results for other template spectra. The results are listed in Table 1.

4. MASS-TO-LIGHT RATIOS FROM THE FUNDAMENTAL PLANE

Fig. 3.--Rest-frame U V color vs. M/LB in solar units. Filled symbols are objects more massive than M 1 3 # 1011 M,; open symbols represent the less massive ones. The cluster galaxies are the z p 1.27 galaxies from Fig. 2. The redshift zero data point is the average for galaxies with j 1 150 km s 1 in the clusters Abell 194 and DC 2345 28 (JÜrgensen, Franx, & KjÔrgaard 1995a). The lines are solar-metallicity Bruzual & Charlot (2003) models with constant star formation during the first 200 Myr (dotted line) and exponentially decaying star formation on the same timescale (dashed line).

Figure 2a shows the FP for the six field galaxies described above, and the FP for Coma derived by JÜrgensen et al. (1996). Additionally, we show the results from van Dokkum & Stanford (2003) on three cluster galaxies at z p 1.27 . The offsets of the high-redshift galaxies from the Coma FP are a measure of the evolution of M/L. We show the evolution of M/L in Figure 2b as a function of redshift. Obviously, the field galaxies at z 1 0.6 span a wide range in offsets, by a factor of approximately 3 in M/L. The error bars on the individual points are much smaller than the offsets. A model with a single-formation redshift can be ruled out at the 99% confidence level, as measured from the x 2 method. The rest-frame colors of the galaxies confirm the reality of the variations in the M/L. As shown in Figure 3, a ver y strong correlation exists between the colors and the M/L, in the direction predicted by population synthesis models. The good correlation demonstrates that colors can be used to estimate the M/L, as applied, for example, by Bell et al. (2003), to a large sample of field early-type galaxies. We note that galaxies in our study lie fairly close to the red sequence and were characterized by Bell et al. (2003) to have red colors. The overall spread in colors of field galaxies is much larger (1.5 mag) compared to the spread found here (0.3 mag).


L8
5. DISCUSSION

VAN DER WEL ET AL.

Vol. 601

On the basis of our high-S/N spectra we have found a rather wide range in M/L for early-type galaxies at z p 1, indicating a range in star formation histories. The M/L and colors are well correlated, as expected from stellar population models. Hence, the scatter in M/L is real. The results agree surprisingly well with earlier results based on lensing galaxies. Rusin et al. (2003) and van de Ven et al. (2003) found a range in M/L, and van de Ven et al. (2003) found a similar correlation between rest-frame colors and M/L. Other authors found either low M/L (e.g., Treu et al. 2002; Gebhardt et al. 2003) or high M/L (e.g., van Dokkum & Ellis 2003), and this is most likely due to (still unexplained) sample selection effects. The last authors found that galaxies with residuals from the r 1/ 4 profile had young ages. However, we find no such relation in our sample. Stellar population models indicate that the low M/L of the blue z 1 galaxies may be due to an age difference of a factor of 3. Alternatively, bursts involving 20%­30% of the mass can produce similar offsets. The current sample is too small to determine the fraction of young early-type galaxies at z 1 reliably. Large, mass-selected samples are needed for this, as current samples are generally optically selected and therefore biased toward galaxies with lower M/L. It is striking that the most massive galaxies have modest evolution in M/L, similar to what van Dokkum & Stanford (2003) found for massive cluster galaxies. The evolution of the galaxies with M p 6.07rej 2 3 # 10 11 M, (JÜrgensen et al. 1996) in our sample is D ln ( M/L B ) p ( 1.17 0.14)z. The implied formation redshift is above 2, ignoring the effects of progenitor bias (van Dokkum & Franx 2001). The mass limit is comparable

to the M mass of an early-type galaxy: if we take the j of early-type galaxies derived by Kochanek (1994) of 225 km s 2, we derive a typical mass of M p 3.1 # 10 11 M,, based on the sample measured by Faber et al. 1989. The sample as a whole evolves as D ln ( M/L B ) p ( 1.64 0.45)z, whereas the sample with masses smaller than 3 # 10 11 M, evolves as D ln ( M/L B ) p ( 1.95 0.29)z. The results are therefore consistent with little or no (recent) star formation in massive early-type galaxies out to z p 1, and younger populations in less massive galaxies, possibly caused by bursts involving up to 30% of the stellar mass. Since these less massive galaxies have much more regular stellar populations at z ! 0.5 without signs of recent star formation, these results are consistent with the downsizing seen in the field population (Cowie et al. 1996): at progressively higher redshifts, more and more massive galaxies are undergoing strong star formation. It remains to be seen how this trend continues out to even higher redshifts. The biases inherent in studies of galaxies at z p 2 and higher make it ver y hard to perform similar studies: the optical light has shifted to the near-IR, and spectroscopy is extremely hard at those wavelengths. More studies at redshift z 1 are needed to determine the distribution of colors and M/L of the progenitors of field earlytypes. Such a determination should be based on mass-selected samples. Further studies of spectral energy distributions extending to the rest-frame infrared will be ver y useful to better constrain the star formation histories of the bluer galaxies. We thank the ESO staff for their support during the observations. We thank C. Wolf for making available the COMBO17 catalog.

REFERENCES Bell, E. F., et al. 2003, ApJ, submitted (astro-ph/0303394) Bruzual A., G., Charlot, S. 2003, MNRAS, 344, 1000 Coleman, G. D., Wu, C.-C., Weedman, D. W. 1980, ApJS, 43, 393 Cowie, L. L., Songaila, A., Hu, E. M., & Cohen, J. G. 1996, AJ, 112, 839 Diaferio, A., Kauffmann, G., Balogh, M. L., White, S. D. M., Schade, D., & Ellingson, E. 2001, MNRAS, 323, 999 Faber, S. M., Wegner, G., Burstein, D., Davies, R. L., Dressler, A., LyndenBell, D., & Terlevich, R. L. 1989, ApJS, 69, 763 Fisher, D., Fabricant, D., Franx, M., & van Dokkum, P. 1998, ApJ, 498, 195 Franx, M. 1993, PASP, 105, 1058 Gebhardt, K., et al. 2003, ApJ, 597, 239 JÜrgensen, I., Franx, M., & KjÔrgaard, P. 1995a, MNRAS, 273, 1079 ------. 1995b, MNRAS, 276, 1341 ------. 1996, MNRAS, 280, 167 Kochanek, C. S. 1994, ApJ, 436, 56 Kochanek, C. S., et al. 2000, ApJ, 543, 131 Rusin, D., et al. 2003, ApJ, 587, 143 Salpeter, E. E. 1955, ApJ, 121, 161 Schlegel, D. J., Finkbeiner, D. P., & Davis, M. 1998, ApJ, 500, 525 Treu, T., Stiavelli, M., Bertin, G., Casertano, S., & MÜller, P. 2001, MNRAS, 326, 237 Treu, T., Stiavelli, M., Casertano, S., MÜller, P., & Bertin, G. 2002, ApJ, 564, L13 van de Ven, P. M., van Dokkum, P. G., & Franx, M. 2003, MNRAS, 344, 924 van Dokkum, P. G., & Ellis, R. S. 2003, ApJ, 592, L53 van Dokkum, P. G., & Franx, M. 1996, MNRAS, 281, 985 ------. 2001, ApJ, 553, 90 van Dokkum, P. G., Franx, M., Kelson, D. D., & Illingworth, G. D. 1998, ApJ, 504, L17 ------. 2001, ApJ, 553, L39 van Dokkum, P. G., & Stanford, S. A. 2003, ApJ, 585, 78 Wolf, C., Meisenheimer, K., Rix, H.-W., Borch, A., Dye, S., & Kleinheinrich, M. 2003 A&A, 401, 73