1: %\newif\ifsubmode
2: %\submodetrue
3: %\submodefalse
4:
5: %%%%%%%%%%%%%%%
6: % Preamble
7: %%%%%%%%%%%%%%%
8:
9: %\ifsubmode
10: % \documentclass[12pt,preprint]{aastex} %% For preprint style with 12pt type
11: % \received{}
12: % \revised{}
13: % \accepted{}
14: % \journalid{ }
15: % \articleid{ }{ }
16:
17: %\else
18: % \documentclass{emulateapj} %% To emulate ApJ style
19: % \slugcomment{Accepted to \apjl}
20: %\fi
21: \documentclass{emulateapj}
22: %\documentclass[12pt,preprint2]{aastex}
23: \newcommand{\lre}{$\log r_{\rm e}$}
24: \newcommand{\re}{$r_{\rm e}$}
25: \newcommand{\mie}{$<\! \mu\! >_{\rm e}$}
26: \newcommand{\ls}{$\log \sigma_0$}
27: \newcommand{\dt}{$\delta(\log t)$}
28: \newcommand{\dz}{$\delta(\log Z)$}
29: \newcommand{\ct}{$c_{\rm t}$}
30: \newcommand{\cz}{$c_{\rm Z}$}
31: \newcommand{\ml}{$ M/L$}
32: \newcommand{\mls}{$M_{\ast}/L$}
33:
34: \shorttitle{SDSS-UKIDSS FPs}
35: \shortauthors{La Barbera et al.}
36:
37: \begin{document}
38: \title{The SDSS-UKIDSS Fundamental Plane of Early-type Galaxies}
39:
40: \author{La Barbera, F. \altaffilmark{1},
41: Busarello, G. \altaffilmark{1},
42: Merluzzi, P. \altaffilmark{1},
43: de la Rosa, I. \altaffilmark{2},
44: Coppola, G. \altaffilmark{3},
45: Haines, C.P. \altaffilmark{4}
46: }
47:
48: \altaffiltext{1}{INAF -- Osservatorio Astronomico di Capodimonte, Napoli, Italy, }
49: \altaffiltext{2}{Instituto de Astrofisica de Canarias, Tenerife, Spain,}
50: \altaffiltext{3}{University of Naples Federico II, Department of Physics, Napoli, Italy,}
51: \altaffiltext{4}{School of Physics and Astronomy, University of Birmingham,
52: Birmingham, UK}
53:
54: %
55: \begin{abstract}
56: We derive the Fundamental Plane (FP) relation for a sample of 1430
57: early-type galaxies in the optical (r band) and the near-infrared (K
58: band), by combining SDSS and UKIDSS data. With such a large,
59: homogeneous dataset, we are able to assess the dependence of the FP on
60: the waveband. Our analysis indicates that the FP of luminous
61: early-type galaxies is essentially waveband independent, with its
62: coefficients increasing at most by $8\%$ from the optical to the NIR.
63: This finding fits well into a consistent picture where the tilt of the
64: FP is not driven by stellar populations, but results from other
65: effects, such as non-homology. In this framework, the optical and NIR
66: FPs require more massive galaxies to be slightly more metal rich than
67: less massive ones, and to have highly synchronized ages, with an age
68: variation per decade in mass smaller than a few percent.
69: \end{abstract}
70: \keywords{Galaxies: fundamental parameters -- Galaxies: evolution}
71:
72: \section{Introduction}
73: \label{sec:INTR}
74:
75: Early-type galaxies (ETGs) populate a two-dimensional surface in the
76: space of parameters that reflect size (effective radius), density
77: (mean surface brightness), and kinetic temperature (velocity
78: dispersion) \citep{George92}. A key feature of the Fundamental Plane
79: (FP; \citealt{Dressler87}; \citealt{George87}) is its deviation
80: (`tilt') from the virial theorem, which may be interpreted as a
81: variation of the M/L ratio along the sequence of ETGs and/or the
82: breaking of homology assumption, i.e. the fact that, for all
83: galaxies, the observed parameters have the same power-law dependence
84: on the corresponding physical quantities (namely, the central velocity
85: dispersion on kinetic energy, the effective radius on `gravitational'
86: radius, and the effective surface brightness on the overall light
87: profile, see e.g.~\citealt{DjS93}). Despite all the observational
88: efforts, the origin of the tilt is still under debate. The change of
89: the M/L ratio can be explained by a change in either the stellar
90: population (e.g.~\citealt{PrS96}) or dark matter content with galaxy
91: mass (\citealt{CLR96}). Both structural and dynamical non-homology
92: have also been invoked as physical explanations of the observed tilt
93: (see e.g.~\citealt{HjM95},~\citealt{CdC95},~\citealt{GrC97},
94: and~\citealt{BCC97}). Recently, ~\citet[from now TBB04]{Tru04} showed
95: that the tilt is mostly driven by dynamical and structural
96: non-homology, while stellar populations account for only a small
97: fraction of it.~\citet{BBT07} argued that the tilt is more likely
98: because of a variation of the dark matter content with mass, still
99: favoring a picture where stellar populations play a minor role. Since
100: the contribution of different stellar populations to galaxy luminosity
101: is expected to be wavelength-dependent, while other effects (e.g.
102: non-homology) are not, the dependence of the FP on wavelength directly
103: informs on how properties of the stellar populations change with mass,
104: which is a crucial point to understand galaxy formation and evolution.
105:
106: Previous studies of the wavelength dependence led to contradictory
107: results.~\citet{PDdC98b} and~\citet{SCO98} found the tilt to
108: significantly decrease from optical to NIR wavelengths, interpreting
109: this result as an increase of age and metallicity with mass.
110: ~\citet{MGA99} and~\citet{ZGS02} found only a small decrease of the
111: tilt with wavelength, with the FP still being significantly tilted in
112: the NIR. However, ~\citet{BER03b}, deriving the FP for ETGs observed
113: in the Sloan Digital Sky Survey (SDSS), found evidence for the FP to
114: be wavelength independent from the $g$ to the $z$ bands. Several
115: different effects can produce this puzzling picture. The FP in
116: different wavebands has often been derived for small samples, with
117: inhomogeneous measurements of galaxy parameters, different selection
118: criteria (e.g. galaxy samples spanning different ranges in magnitude
119: and/or velocity dispersion), and with different fitting methods. The
120: FP by~\citet{BER03b} avoided all these problems by analyzing the same
121: sample of galaxies at different wavebands, but it was limited to the
122: short wavelength baseline provided by the SDSS. In the present work,
123: for the first time, we derive the FP by using the same, large,
124: homogeneous sample of ETGs over the wide wavelength baseline provided
125: by the r- and K-band data of the SDSS and the UKIRT Infrared Deep Sky
126: Survey (UKIDSS).
127:
128: The layout of the paper is the following. In Sec.~2, we describe the
129: selection of the sample, while Sec.~3 details how we obtain the r- and
130: K-band structural parameters, and the central velocity
131: dispersions. Sec.~4 deals with the comparison of the r- and K-band
132: FPs. In Sec.~5, we show how the optical and NIR FPs constrain the
133: variation of stellar population parameters along the galaxy sequence.
134: The discussion follows in Sec.6. Throughout the paper, we adopt the
135: cosmology $\rm H_0 \! = \! 75 \, km \, s^{-1} \, Mpc^{-1}$,
136: $\Omega_{\rm m} \! = \! 0.3$, and $\Omega_{\Lambda} \! = \! 0.7$.
137:
138: \section{Sample selection}
139: \label{sec:DATA}
140: We select a sample of ETGs, with available K-band photometry from the
141: second data release of UKIDSS, and r-band photometry and central
142: velocity dispersions from the fifth data release (DR5) of SDSS.
143: First, a complete volume-limited catalog of galaxies is defined,
144: consisting of all the 105036 objects in DR5 with an r-band absolute
145: magnitude M$_{r}{<}-20$, and a spectroscopic redshift in the range of
146: 0.05 to 0.095. Absolute magnitudes are obtained from the SDSS r-band
147: Petrosian magnitudes, k-corrected to redshift $0.1$ by using the $
148: kcorrectv4\_1\_4 $ software~\citep{BL03}. The lower redshift limit is
149: chosen to minimize the aperture bias~\citep{GOMEZ03}, while the upper
150: redshift limit guarantees a high level of completeness
151: (see~\citealt{SAR06}). ETGs are defined according to the SDSS
152: spectroscopic parameter $eclass$, that classifies the spectral type on
153: the basis of the principal component analysis technique, and the
154: photometric parameter $fracDev_r$, which measures the fraction of
155: galaxy light that is fitted by a de Vaucouleurs law. Following
156: ~\citet{BER03a}, we define as ETGs those objects with $eclass \! < \!
157: 0$ and $fracDev_r \! > \! 0.8$, resulting in a list of 47061
158: galaxies. Out of them, we retain only those 33628 galaxies with
159: available central velocity dispersion, $\sigma_0$, between $70$ and
160: $420$ km\,s$^{-1}$. These cuts are required to obtain reliable
161: $\sigma_0$'s from the SDSS-DR5 database. All the selected galaxies
162: have spectra with median per-pixel $S/N$ larger than $10$, which is a
163: further requirement to retrieve reliable SDSS velocity
164: dispersions~\footnote{\footnotesize See the list of requirements in
165: the Algorithms section of the SDSS-DR5 website, at
166: http://www.sdss.org/dr5/algorithms/veldisp.html }. The SDSS catalog
167: is cross-matched with UKIDSS, resulting in 1570 galaxies. We notice
168: that all the SDSS galaxies with reliable velocity dispersion, that are
169: covered by the UKIDSS survey, are then included in this sample. In
170: other words, the matching with UKIDSS does not change the magnitude
171: limit of the present sample. We select only those galaxies observed in
172: K-band images with good seeing ($FWHM \leq 1''$). This selection
173: reduces the sample size by only $10 \%$, and excludes cases where
174: structural parameters might be affected by large uncertainties. Since
175: all the r-band images have FWHM smaller than~$1.4''$ and the galaxies
176: in our sample have on average effective radii larger in the optical
177: than in the K-band (see below), we do not apply any seeing selection
178: to the SDSS photometry. The above procedure leads to a final sample
179: of 1430 galaxies.
180:
181: \begin{deluxetable*}{c|c|c|c|c|c|c}
182: \tablewidth{0pc}
183: \tabletypesize{\scriptsize}
184: \tablecaption{FP parameters in the $r$ and K bands for the sample of 1430 gala
185: xies. }
186: \tablehead{RA & DEC & $\log R_{\rm e,r}$ & $< \!\mu \!>_{\rm e,r}$ & $\log R_{\rm e,K}$ & $< \!\mu \!>_{\rm e,K}$ & $\log \sigma_0$ }
187: \startdata
188: 145.34432 & -0.01692 & 0.860 & 21.602 & 0.527 & 17.046 & 2.282 \\
189: 147.24805 & -0.03572 & 0.834 & 21.346 & 0.680 & 17.437 & 2.228 \\
190: 146.81199 & -0.19005 & 0.459 & 19.968 & 0.349 & 16.319 & 2.217 \\
191: 146.09369 & -0.79309 & 0.985 & 21.521 & 0.475 & 16.216 & 2.259 \\
192: 146.46892 & -0.09284 & 1.622 & 23.236 & 1.090 & 18.009 & 2.317 \\
193: 146.19333 & -0.03887 & 0.625 & 20.555 & 0.445 & 16.439 & 2.226 \\
194: 145.68114 & -0.86722 & 1.001 & 21.576 & 0.721 & 17.350 & 2.196 \\
195: 145.70894 & -0.74768 & 0.134 & 18.936 & 0.088 & 15.653 & 2.112 \\
196: 145.48725 & -0.80693 & 0.965 & 21.993 & 1.065 & 19.167 & 2.180 \\
197: 145.42694 & 0.04954 & 0.470 & 19.852 & 0.018 & 14.473 & 2.367 \\
198: 145.44549 & -0.12268 & 0.381 & 19.576 & 0.166 & 15.351 & 2.353 \\
199: 145.34160 & -0.57727 & 0.087 & 18.124 & -0.152 & 13.574 & 2.355 \\
200: 145.19382 & 0.16887 & 0.192 & 18.862 & 0.137 & 15.669 & 1.930 \\
201: 146.28017 & -0.40695 & 0.033 & 17.760 & -0.129 & 13.996 & 2.288 \\
202: 147.30829 & 0.15116 & 0.830 & 21.046 & 0.454 & 16.295 & 2.287 \\
203: 146.72794 & -0.55688 & 0.262 & 19.223 & 0.178 & 15.409 & 2.316 \\
204: 148.85664 & -0.05916 & 0.978 & 21.643 & 1.160 & 19.055 & 2.226 \\
205: 147.79347 & 0.12326 & 0.614 & 20.337 & 0.227 & 15.566 & 2.187 \\
206: 147.74868 & 0.11584 & 0.480 & 20.362 & 0.125 & 15.682 & 2.233 \\
207: 148.58499 & -0.94207 & 1.180 & 22.575 & 0.703 & 17.637 & 2.109 \\
208: 149.12382 & -0.39828 & 0.365 & 19.682 & -0.037 & 14.726 & 2.324 \\
209: 149.11298 & -0.34883 & 0.685 & 20.638 & 0.593 & 16.954 & 2.270 \\
210: 149.18631 & -0.31181 & 1.291 & 23.217 & 0.823 & 18.064 & 2.101 \\
211: 148.84251 & -0.04411 & 0.469 & 20.031 & 0.387 & 16.276 & 2.202 \\
212: 149.11264 & -0.47563 & 0.572 & 20.065 & 0.471 & 16.069 & 2.398 \\
213: 149.17153 & -0.41298 & 0.359 & 19.539 & 0.266 & 15.894 & 2.260 \\
214: \enddata
215: \end{deluxetable*}
216:
217:
218: \section{FP parameters}
219: \label{sec:FP}
220: The photometric parameters entering the FP, namely, the effective
221: radius, $r_{\rm e}$, and the mean surface brightness within that
222: radius, \mie, are derived using 2DPHOT~\citep{LBdC08}. The r- and
223: K-band images are processed by adopting the same 2DPHOT options,
224: allowing homogeneous structural parameters to be derived between both
225: bands. For each galaxy, a local PSF model is computed by fitting
226: simultaneously the four closest stars to that galaxy on the image.
227: Structural parameters, i.e. the effective parameters and the Sersic
228: index $n$ (shape parameter), are then derived by modeling the galaxy
229: images with two-dimensional seeing-convolved Sersic models. Effective
230: radii are converted to physical units by using the angular diameter
231: distance corresponding to the DR5 spectroscopic redshift, $z$, of each
232: galaxy. Mean surface brightnesses are de-reddened using the
233: extinction maps of~\citet{SFD98}, are corrected for cosmological
234: dimming, by subtracting the term $\rm 7.5 \log (1+z)$, and are
235: k-corrected to redshift $0.1$ with the $ kcorrectv4\_1\_4 $
236: software~\citep{BL03}. In order to estimate the accuracy on
237: structural parameters, we use 160, out of the 1430, galaxies having
238: repeated observations in different UKIDSS frames. We find that the
239: averaged differences between repeated measurements of structural
240: parameters are fully consistent with zero, amounting to $-0.01 \! \pm
241: \! 0.01$, $-0.037 \! \pm \! 0.04$, and $0.005 \! \pm \! 0.01$ for
242: $\log r_{\rm e}$, \mie \, and $\log n$, respectively. The rms values
243: of these differences amount to $32 \%$ in \re, $0.6 \rm \,
244: mag/arcsec^2$ in \mie, and $25 \%$ in $n$. Notice that the scatter in
245: $\log r_{\rm e}$ is fully consistent with the typical accuracy of the
246: measurement of the half-light radii~\citep{Kelson00}. The quantity
247: $\log r_{\rm e} \! - \! 0.3 \! < \! \mu \! >_{\rm e}$, that is
248: the relevant photometric parameter entering the FP, has an uncertainty
249: of only $7 \%$, as expected due to the correlation of measurement
250: errors of the effective parameters. The comparison of r- and K-band
251: structural parameters for the present sample is fully consistent with
252: what is found in our previous studies (e.g.~\citealt{LMB04}). In
253: particular, the mean ratio between r- and K-band radii is $-0.11 \!
254: \pm \! 0.01$dex, i.e. on average effective radii decrease by $\sim
255: \! 25 \%$ from the optical to the NIR. This value is consistent with
256: that of $\sim 20 \%$ found by~\citet{PdCD98a}, and is in agreement
257: with the existence of negative color gradients in early-type galaxies.
258: Sersic indices are fully consistent between optical and NIR. The
259: average ratio of r- to K-band $n$ values amounts to $-0.007 \! \pm \!
260: 0.009$dex.
261: %Comparison of re in the different wavebands...e..
262: Central velocity dispersions are retrieved from DR5 and are corrected
263: as in \citet{BER03b} to a relative aperture of $r_{\rm e}/8$,
264: following~\citet{JFK95}. As shown by~\citet{BERN07}, for $\sigma_0 \!
265: < \! 150$ km s$^{-1}$ the DR5 velocity dispersions are slightly
266: overestimated. This small bias increases up to $12 \%$ at $\sigma_0
267: \sim 100$ km s$^{-1}$. We remove this effect by applying the
268: correction curve shown in fig.~4 of~\citet{BERN07} (see the grey line
269: in the upper-left panel)\footnote{\footnotesize The SDSS-DR6 velocity
270: dispersions are not affected by this bias, but they are available
271: only for $85\%$ of our sample. However, we verified that
272: restricting the analysis to the sample with DR6 velocity dispersions
273: changes the FP coefficients by less than $2\%$.}. The r- and K-band
274: effective parameters, as well as the corrected DR5 velocity
275: dispersions are given in Tab.~1 (fully available in electronic form)
276: for all the 1430 galaxies analyzed in the present study. Columns in
277: the table provide the following quantities. Columns~1 and~2 report
278: right ascension (RA) and declination (DEC) in units of degrees.
279: Columns~3 and~4 provide the logarithm of the effective radius (in unis
280: of $kpc$) and the effective mean surface brightness (in unis of
281: $mag/arcsec^2$) in the r-band. Columns~5 and~6 list the same
282: quantities as columns~3 and~4 but for the K-band. Column~7 reports
283: the corrected DR5 velocity dispersions.
284:
285:
286:
287: %\setcounter{table}{1}
288:
289:
290: \begin{figure*}[t!]
291: %\includegraphics[width=8.5cm]{FP_edge_short_rK3.eps}
292: \epsscale{0.9}
293: \plotone{f1.eps}
294: \caption{Short edge-on view of the r- and K-band FPs, with the
295: photometric quantity, $\log R_{\rm e} \! - \! b < \! \mu \!>_{\rm
296: e}$, being plotted as a function of \ls. Both quantities have been
297: normalized by subtracting their median values, enabling a direct
298: comparison of the r- and K-band projections. The median \ls \, value
299: amounts to $2.215 \rm dex$ . The FPs are represented by two lines,
300: whose slopes are equal to the coefficients $a$ of the \ls \,
301: fits. Black and grey colors correspond to the r and K bands,
302: respectively. The average value of $b=0.308$ was adopted for both
303: bands. The inset compares the r- and K-band slopes of the FP, with
304: the concentric ellipses denoting $1$ and $2 \sigma$ confidence
305: contours. Results of the orthogonal and \ls \, fits are represented
306: by solid and dashed ellipses, respectively.}
307: \label{fig:FP_EDGE}
308: \end{figure*}
309:
310:
311: \section{The SDSS and UKIDSS Fundamental Planes}
312: \label{sec:RES}
313:
314: We adopt the usual representation of the FP, $\log r_{\rm e} =
315: a \log \sigma_{\rm 0} + b < \!\mu\! >_{\rm e} + c \ $, where $a$ and
316: $b$ are the ``slopes'', and $c$ is the offset. These coefficients
317: were derived by minimizing the sum of the absolute residuals around
318: the plane. This method has the advantage of being less affected by
319: outliers (e.g.~\citealt{JFK96}). We adopted two fitting procedures,
320: minimizing the residuals either in \ls\, or in the orthogonal
321: direction to the plane. The first method is virtually independent of
322: selection effects in the plane of photometric parameters, while the
323: orthogonal fit, adopted in most previous works, treats all the
324: variables symmetrically (see~\citealt{LAB00}).
325:
326: The FP coefficients were corrected for selection effects through
327: Monte-Carlo simulations. First, we generated galaxy magnitudes
328: according to the r-band luminosity function of ETGs~\citep{BER03a}.
329: For each magnitude, we derived~\lre \, and \mie \, from the
330: luminosity--size relation of~\citet{Shen03}. Values of \ls \, were
331: assigned by using the equation of the FP, assuming given values of
332: $a$, $b$, and $c$, and a given scatter in \ls. All these quantities
333: were chosen to match the observed FP. Notice that when deriving the
334: simulated FP we applied the same cuts in magnitude and~\ls \, as we
335: did for the real sample. The corrections for selection effects on
336: $a$, $b$, and $c$, were estimated by not applying the cuts in
337: magnitude and~\ls \, to the simulated FP. These corrections amount to
338: $+0.01\%$, $+8\%$ and $+5\%$ for the \ls \, fit, and $+35 \%$, $+7\%$
339: and $+17\%$ for the orthogonal fit. As expected
340: (e.g.~\citealt{LAB00}), the magnitude cut underestimates the
341: coefficient $a$ of the orthogonal fit, while for the \ls \, fit the
342: effect is negligible. The above corrections depend mainly on the
343: scatter around the FP, and, because of the very similar dispersions of
344: the r- and K-band FPs (see below), were applied to both the r- and the
345: K-band coefficients. We notice that the above procedure assumes that
346: our sample of early-type galaxies is magnitude complete. However,
347: because of the matching of the initial volume-complete SDSS catalog
348: with the UKIDSS database, that reduces the sample size from 33628 to
349: 1430 ETGs, the above assumption might not necessarly hold. To address
350: this point, we retrieved effective parameters and velocity dispersions
351: for the whole sample of 33628 ETGs in the SDSS catalog, and estimated
352: how the r-band FP coefficients change between the whole and the
353: UKIDSS-matched samples. We found the variation to be completely
354: negligible, amounting to $1 \%$ and to $ 2\%$ for the values of $a$
355: and $b$ obtained by the orthogonal fit.
356:
357:
358: \begin{deluxetable}{c|c|c|c|c}
359: \tablecolumns{5}
360: \tablewidth{0pc}
361: %\small
362: \tablecaption{FP coefficients.~\label{tab:FPCOF}}
363: \tablehead{& a & b & c & $\rm rms$ }
364: \startdata
365: $\log \sigma_o $ fit &&&&\\
366: $\rm r \! - \! band$ & $1.55 \pm 0.07$ & $0.312 \pm 0.003$ & $-9.1 \pm 0.1$ & $0.081 $ \\
367: $\rm K \! - \! band$ & $1.51 \pm 0.04$ & $0.307 \pm 0.003$ & $-8.6 \pm 0.1$ & $0.073 $ \\
368: \hline
369: orthogonal fit &&&&\\
370: $\rm r \! - \! band$ & $1.42 \pm 0.05$ & $0.305 \pm 0.003$ & $-8.8 \pm 0.1$ & $0.064 $ \\
371: $\rm K \! - \! band$ & $1.53 \pm 0.04$ & $0.308 \pm 0.003$ & $-8.6 \pm 0.1$ & $0.062 $
372: \enddata
373: \end{deluxetable}
374:
375:
376: The coefficients of the FP are reported in Tab.~\ref{tab:FPCOF}. In
377: Fig.~\ref{fig:FP_EDGE}, we compare the edge-on views of the r- and
378: K-band planes, and in the inset we show the values of $a$ and $b$.
379: The $a$ coefficients differ by only $2\sigma$ for the orthogonal fit,
380: and they almost coincide for the \ls \, fit, which is virtually
381: unaffected by selection effects. The value of $b$ ($\sim 0.308$) is
382: independent of the waveband, as well as the scatter of the FP, which
383: presents a tiny difference ($<2 \%$) for the \ls \, fit (see also
384: ~\citealt{PDdC98b}). The r-band value of $a$ is fully consistent with
385: that of $a=1.49 \pm 0.05$ found by~\citet{BER03b}, while it is larger
386: than that of $a=1.24 \pm 0.07$ found by~\citet{JFK96}. For the K-band,
387: the \ls \, coefficient is consistent with the value of $a=1.53 \pm
388: 0.08$ found by~\citet{PDdC98b}, while it is larger than that of
389: $a=1.38 \pm 0.1$ found by~\citet{ZGS02}. We notice that, because of
390: the large sample size, the accuracy of our K-band FP coefficients is
391: significantly higher (by $50\%$) than in previous studies. The
392: invariance of the FP with waveband is in agreement with~\citet{Cap06},
393: who found for 25 ETGs from the SAURON project the M/L versus L
394: relation to have the same slope in both the I- and K-bands.
395:
396: \begin{figure*}[t!]
397: \epsscale{0.9}
398: \plotone{f2.eps}
399: \caption{Differences of age, \dt, and metallicty, \dz, per decade
400: in mass between more and less massive galaxies, as derived from the FP
401: slopes in the r and K bands, by adopting the mean values of $a$ of the
402: orthogonal and \ls \, fits (see the text). Red crosses and blue
403: circles correspond to the cases of homology assumption with $f=0$ and
404: $f=1$, respectively, while the green squares correspond to the case
405: where non-homology is taken into account (see Sec.~5). For each symbol
406: type, different points mark the values of \dt \, and \dz \, obtained
407: by shifting $a_{\rm r}$ and $a_{\rm K}$ according to the corresponding
408: uncertainties. For each case, the median values, $\delta \rm t$ and
409: $\delta \rm Z$, of \dt \, and \dz \, are plotted as black symbols and
410: are reported in the lower-left part of the plot. The effect of
411: changing $f$ on the median values of \dt \, and \dz \, is shown by the
412: black solid and dashed lines, for the BrC03 and the Vaz96 models,
413: respectively. The horizontal dotted lines mark the values of
414: $\delta \rm Z$ obtained from previous studies of the color-magnitude
415: relation (KOD98) and absorption line indices (THO05).}
416: \label{fig:DLOGT_DLOGZ}
417: %\end{center}
418: \end{figure*}
419:
420:
421: \section{Constraints on the stellar populations}
422: \label{sec:ANALYSIS}
423:
424: The tilt of the FP can be parameterized as a power-law relation
425: between \ml \, and $M$. We assume that the stellar mass-to-light ratio
426: of galaxies, \mls, is a power-law: \mls$\propto M^{\beta^\ast}$. This
427: agrees with what was found in previous studies for bright ETGs (see
428: TBB04). The \ml \, vs. $M$ relation can then be written as $M/L\!
429: \propto \! M^{\beta +\beta^\ast}$ (see also~\citealt{PrS96}). The
430: quantity $\beta$ is related to the ratio of stellar to total mass as
431: $M_{\ast}/M\! \propto \! M^{-\beta}$, describing the variation of
432: wavelength independent properties with mass (e.g. dark matter
433: content). The quantity $\beta^\ast$ depends on how stellar population
434: parameters vary with mass. Considering only the effects of age and
435: metallicity, for each waveband, we can write:
436: \begin{eqnarray}
437: \frac{\delta(\log \! M_{\ast}/L)}{\delta(\log \! M_{\ast})} =\frac{\beta^{\ast}}{1-\beta}=c_{\rm t} \cdot \frac{ \delta(\log \! t) }{\delta(\log \! M_{\ast})} + c_{\rm Z} \cdot \frac{\delta(\log \! Z)}{\delta(\log \! M_{\ast})},
438: \label{eq:mlchange}
439: \end{eqnarray}
440: where the quantities \dt \, and \dz \, are the logarithmic differences
441: of age and metallicity defined between more and less massive galaxies
442: (per decade in stellar mass), while $c_{\rm t}=\frac{\partial \log \!
443: M_{\ast}/L}{\partial \log t}$ and $c_{\rm Z}=\frac{\partial \log \!
444: M_{\ast}/L}{\partial \log Z}$ are the partial derivatives of $\log
445: \! M_{\ast}/L$ (in a given band) with respect to $t$ and $Z$.
446: Writing Eq.~\ref{eq:mlchange} for r and K bands, we obtain two
447: independent constraints on \dt \, and \dz. Then, expressing the
448: values of $\beta^\ast$ in the r and K bands ($\beta^\ast_{\rm r}$ and
449: $\beta^\ast_{\rm K}$) as a function of the corresponding coefficients
450: of the FP, we can estimate \dt \, and \dz. We introduce the parameter
451: $f=\beta^\ast_{\rm K} / (\beta + \beta^\ast_{\rm K})$ which defines
452: the fraction of the K-band slope of the \ml \, vs. $M$ relation
453: (i.e. the fraction of the K-band tilt) due to stellar population
454: effects.We note that $f$ can vary between $0$ and $1$. For $f=0$, the
455: K-band tilt is independent of stellar populations ($\beta^{\ast}_{\rm
456: K}=0$), while for $f=1$ the tilt is entirely explained by stellar
457: population effects ($\beta=0$). Under the assumption of homology, the
458: slope of the \ml \, vs. $M$ \, relation can be written as
459: $(2-a)/(2+a)$. With the above notation, the following relations hold:
460: \begin{eqnarray}
461: \beta+\beta^\ast_{\rm r} = (1-f)/f \! \cdot \! \beta^\ast_{\rm K}+\beta^\ast_{\rm r} = (2-a_{\rm r})/(2+a_{\rm r}), \nonumber & (2a) \\
462: \beta+\beta^\ast_{\rm K} = \beta^\ast_{\rm K}/f = (2-a_{\rm K})/(2+a_{\rm K}). \nonumber & (2b)
463: \label{eq:mlslopes}
464: \end{eqnarray}
465: For different $f$'s and using the values of $a$ in
466: Tab.~\ref{tab:FPCOF}, we computed $\beta^\ast_{\rm r}$ and
467: $\beta^\ast_{\rm K}$ from Eqs.~\ref{eq:mlslopes}, and then, inverting
468: Eq.~\ref{eq:mlchange} for both bands, we derived \dt \, and \dz. The
469: values of \ct \, and \cz \ were estimated using simple stellar
470: population models, with solar metallicity and an age of
471: $12$~Gyr~\footnote{Computing \ct \, and \cz \, by varying $t$ between
472: $5$ and $12$Gyr, and $Z$ from $0.5 Z_{\odot}$ to $2.0 Z_{\odot}$, we
473: found that the change of the median values of \dt \, and \dz \, are
474: negligible, amounting to less than $0.01$dex and $0.005$dex,
475: respectively.}, using both the \citet{BrC03} (from now BrC03) and the
476: updated \citet{Vazdekis:96} (from now Vaz96) models. We adopted a
477: Scalo IMF and a Salpeter IMF for the BrC03 and Vaz96 models,
478: respectively. Fig.~\ref{fig:DLOGT_DLOGZ} shows \dz \, versus \dt \,
479: obtained for $f=0$ and $f=1$, as well as the mean values of \dt \,
480: and \dz \, as a function of $f$. The scatter seen in the plot
481: reflects the uncertainties listed in Tab.~\ref{tab:FPCOF} for the
482: $a_{\rm r}$ and $a_{\rm K}$ coefficients. The figure shows that if the
483: NIR tilt of the FP is not caused by stellar population effects
484: ($f=0$), more massive galaxies have to be more metal rich than less
485: massive ones (\dz$>0$), with galaxy ages being remarkably homogeneous
486: ($\delta t/t \sim 1 \%$). As $f$ increases, we see that
487: \dz \, decreases, while \dt \, becomes larger. Specifically, for
488: $f \sim 1$, more massive galaxies are much older ($\delta
489: t/t \sim50 \%$) and less metal rich than low mass systems.
490:
491: \section{Discussion}
492: \label{sec:DISC}
493:
494: This work presents the waveband dependence of the FP by comparing the
495: optical and NIR FPs for a large sample of galaxies, with homogeneous
496: measurements of structural parameters and velocity dispersions. This
497: is allowed, for the first time, thanks to the availability of both
498: SDSS r-band photometry and spectroscopy, and UKIDSS K-band photometry
499: for the same sample of ETGs. Such a dataset, together with the use of
500: the same fitting procedure in both bands, makes our study virtually
501: free from any methodological effect on the waveband dependence of the
502: FP. Our analysis shows that the FP does not change significantly from
503: the optical to the NIR, bringing interesting questions about the
504: nature of the sequence of ETGs.
505:
506: In Sec.~5, we have shown how the r- and K-band FPs constrain the
507: variation of the stellar population properties (age and metallicity)
508: with stellar mass, and how such a constraint is strongly dependent on
509: the fraction, $f$, of the FP tilt resulting from stellar population
510: effects. Previous studies of the color-magnitude (CM) relation and of
511: line-strength indices of ETGs might help us to solve this dependency,
512: by deriving the proper value for $f$. \citet{Kod98} showed that the
513: little redshift evolution of the CM relation implies (i) all the
514: (luminous) ETGs to be equally old and (ii) more massive galaxies to be
515: more metal rich, with a metallicity change of $\delta (\log Z) \sim
516: 0.22$dex per decade in stellar mass. This finding is qualitatively
517: consistent with that of~\citet{THO05}, who found absorption-line
518: indices consistent with a metallicity change of $\delta (\log Z) \sim
519: 0.12$dex per mass decade. They also found evidence for an age
520: gradient along the sequence of ETGs, with $\delta (\log t) = +0.05 \pm
521: 0.07$ (see their eq.3). Fig.~\ref{fig:DLOGT_DLOGZ} compares these
522: values of $\delta (\log Z)$ with those derived by the optical--NIR FP.
523: For the BrC03 (Vaz96) model, the maximum value of $\delta (\log Z)$,
524: which is consistent with the FP, amounts to $0.06 $($0.09$) $\pm 0.04
525: $($0.04$)dex for $f=0$. This value is $4 \sigma$ ($3.7 \sigma$) lower
526: than that derived by the CM relation, while it is only $1.5 \sigma$
527: ($0.8 \sigma$) lower than that found by~\citet{THO05}. As $f$
528: increases, the FP requires the value of $\delta (\log Z)$ to decrease,
529: making the above differences even larger. Therefore, reconciling
530: previous estimates of $\delta (\log Z)$ with our results leads to a
531: scenario where $f=0$, which means that the FP tilt is not driven by
532: stellar populations. We have to remark, however, that this
533: interpretation is troublesome, since galaxy colors and line indices
534: are always measured within a given fixed aperture, and the presence of
535: internal population gradients in galaxies can significantly affect the
536: inferred values of $\delta (\log Z)$ and $\delta (\log t)$
537: (e.g.~\citealt{Scodeggio:01}).
538:
539: Further constraints come from previous works addressing the origin of
540: the FP tilt itself. Performing a detailed dynamical analysis of
541: twenty-five galaxies, \citet{Cap06} derived \ml \, ratios consistent
542: with those obtained from the virial theorem in the assumption of
543: homology, concluding that structural and orbital non-homology have a
544: negligible role in the tilt of the FP (see also~\citealt{ZZG08}). In
545: support to this view, they also showed that the variation of the
546: dynamical \ml \, is correlated with the H$_\beta$ line-strength, thus
547: ascribing most of the tilt to stellar population (age)
548: effects. However, as the authors notice, this result strictly applies
549: to their measurement of the velocity dispersion as the average over an
550: aperture of radius equal to $r_e$, a fact that alone might compensate
551: part of the dynamical non-homology. Moreover, most of the galaxies in
552: their sample (68\%) are fast rotators, while five of them (20\%) have
553: low velocity dispersion ($\sigma$=60-85 km s$^{-1}$). As found
554: by~\citet{ZGZ06} and~\citet{DOF08}, bright and faint spheroids have
555: different FPs, with the tilt becoming larger for galaxies having low
556: velocity dispersion. Hence, the different selection of our sample and
557: that of~\citet{Cap06} prevents a straightforward comparison.
558: \citet{BBT07} showed that by replacing mean surface brightness with
559: mass density, the FP relation closely approaches the virial theorem
560: expectation, implying that most of the tilt is caused by a variation
561: of dark matter content with galaxy mass. However, the uncertainties
562: on their FP coefficients and the possible biases introduced by the
563: gravitational-lens selection do not definitively exclude the
564: contribution of non-homology to the FP tilt. Our result agrees with
565: \citet{BBT07} regarding the minor role played by stellar populations
566: on the tilt. TBB04, agreeing with~\cite{BCC97} and~\citet{GrC97},
567: found that structural and dynamical non-homology can account for more
568: than two-thirds of the FP tilt, with the remaining part being
569: explained by stellar population effects. In particular, restricting
570: the analysis to the magnitude-complete subsample, they found that the
571: contribution of stellar populations to the tilt becomes negligible.
572: Notice also that their FPs are derived from different sources, with
573: significantly different coefficients in the optical and NIR wavebands.
574:
575: To understand how our results may be affected by the assumption of
576: homology, we followed an approach similar to that of TBB04, using
577: spherical, isotropic, non rotating, one-component models of ETGs
578: following the Sersic law (see~\citealt{LBC05}). For each galaxy, we
579: considered the model with the corresponding Sersic index in the
580: r-band, and used that model to correct the central velocity dispersion
581: to the quantity $\sigma_{\rm t}$ (defined as the square-root of the
582: total specific kinetic energy), the effective radius to the
583: gravitational radius, $r_{\rm g}$, and to calculate the mean surface
584: brightness within the gravitational radius, $ <\! \mu \! >_{\rm g}$.
585: Applying the orthogonal fit, we obtain the following equation of the
586: FP in the K band: $\log r_{\rm g} \propto (2.3 \pm 0.2) \log \sigma_t
587: + (0.4 \pm 0.02) <\! \mu \! >_{\rm g}$, which is remarkably close to
588: the virial theorem expectation, implying that non-homology may account
589: for the entire tilt. To explore how this result would affect stellar
590: population properties, we normalized the r- and K-band FP coefficients
591: in such a way to match the virial theorem expectation ($a=2.0$ and
592: $b=0.4$) in the K band. Then, we derived the corresponding values of
593: \dz \, and \dt \, (see Sec.~\ref{sec:ANALYSIS}). As we can see in
594: Fig.~\ref{fig:DLOGT_DLOGZ}, accounting for non-homology leads to the
595: same \dz \, and \dt \, values as those derived under the assumption of
596: homology if the tilt of the FP is not due to stellar populations (i.e.
597: $f=0$). As discussed above, this is consistent with what expected
598: from the color-magnitude relation and absorption-line indices of ETGs.
599:
600: %~\citet{Ger01} find that the M/L ratios derived from a
601: %line-profile analysis of a sample of elliptical galaxies show no systematic
602: %trend with respect to the \ml derived for the virial theorem. They
603: %conclude that the \ml vs $L$ relation does not originate from dynamical
604: %non-homology, although photometric non-homology could affect its
605: %slope. Since they also rule out a possible role of dark matter
606: %content, they conclude that stellar populations must cause the
607: %dependence of \ml on $L$, the main driver being metallicity, with some
608: %secondary effect explaining the K-band tilt.
609: %A comparison with our result is however not feasible, due to the
610: %fact that their sample is composed of 21 ellipticals, mostly luminous,
611: %nearly round and slowly rotating, that admittely constitute a very
612: %uniform group for what concerns the dynamical structure, that is
613: %certainly not the case for our sample due to the completely different
614: %selection.
615:
616: In summary, our analysis suggests a consistent picture where (i) the
617: tilt of the FP does not originate from stellar population effects but
618: is due to other effects, such as non-homology; (ii) the SDSS-UKIDSS
619: FPs require more massive galaxies to be mildly more metal rich than
620: less massive systems, and to have extremely synchronized ages, with
621: the age variation per mass decade being smaller than few percent.
622:
623: %The
624: %similarity of the optical and NIR FPs should be taken into account by
625: %future investigations on the origin of the sequence of ETGs.
626:
627: %Aside from this interpretation
628: %, althought limited to certain
629: %luminosity and velocity dispersion ranges,
630: %
631: %account when the origin of the tilt is investigated.
632:
633: \acknowledgements
634:
635: We thank R.R. de Carvalho, S.G. Djorgovski, M.Capaccioli, and
636: A.Mercurio for the helpful comments and suggestions. We also
637: acknowledge the referee for his/her helpful suggestions. We also thank
638: A. Vazdekis for providing us with the most recent version of his
639: stellar population code. We have used data from the 2nd data release
640: of the UKIDSS survey (\citet{Law07}), which is described in detail in
641: \citet{War07}. Funding for the SDSS and SDSS-II has been provided by
642: the Alfred P. Sloan Foundation, the Participating Institutions, the
643: National Science Foundation, the U.S. Department of Energy, the
644: National Aeronautics and Space Administration, the Japanese
645: Monbukagakusho, the Max Planck Society, and the Higher Education
646: Funding Council for England.
647:
648: \begin{thebibliography}
649:
650: \bibitem[Bernardi et al.(2003a)]{BER03a} Bernardi, M., et al. 2003a, AJ, 125, 1849
651: \bibitem[Bernardi et al.(2003b)]{BER03b} Bernardi, M., et al. 2003b, AJ, 125, 1866
652: \bibitem[Bernardi(2007)]{BERN07} Bernardi, M. 2007, AJ, 133, 1954
653: \bibitem[Blanton et al.(2003)]{BL03} Blanton, M.R., Lin, H., Lupton, R.H., et al., 2003, AJ, 125, 2276
654: \bibitem[Bolton et al.(2007)]{BBT07} Bolton, A.S., et al. 2007, ApJ 665, 105
655: \bibitem[Bruzual \& Charlot(2003)]{BrC03} Bruzual, G., \& Charlot, S. 2003, MNRAS 344, 1000 (BrC03)
656: \bibitem[Busarello et al.(1997)]{BCC97} Busarello, G., et al. 1997, A\&A 320, 415
657: \bibitem[Capelato et al.(1995)]{CdC95} Capelato, H.V., de Carvalho, R.R., Carlberg, R.G. 1995, ApJ, 451, 525
658: \bibitem[Cappellari et al.(2006)]{Cap06} Cappellari, M., et al. 2006, MNRAS 366, 1126
659: \bibitem[Ciotti et al.(1996)]{CLR96} Ciotti, L., Lanzoni, B., \& Renzini, A. 1996, \mnras, 282, 1
660: \bibitem[Djorgovski \& Davis(1987)]{George87} Djorgovski, S.G., \& Davis, M. 1987, ApJ, 313, 59
661: \bibitem[Djorgovski(1992)]{George92} Djorgovski, S. 1992, ASPC, 24, 19
662: \bibitem[Djorgovski \& Santiago(1993)]{DjS93} Djorgovski, S.G., \& Santiago, B.X. 1993, {\it ESO/EIPD Workshop on structure, dynamics, and chemical evolution of early-type galaxies}, Danziger, I.J, Zelinger, W.W., Kj\"ar, eds., (ESO: Garching) p. 59
663: \bibitem[D'Onofrio et al.(2008)]{DOF08} D'Onofrio, M., et al. 2008, astro-ph/0804.1892
664: \bibitem[Dressler et al.(1987)]{Dressler87} Dressler, A., et al. 1987, ApJ, 313, 42
665: \bibitem[G\'{o}mez et al.(2003)]{GOMEZ03} G\'{o}mez, P.L., et al. 2003, ApJ, 584, 210
666: \bibitem[Graham \& Colless(1997)]{GrC97} Graham, A., \& Colless, M. 1997, MNRAS 287, 221
667: \bibitem[Hjorth \& Madsen(1995)]{HjM95} Hjorth, J., \& Madsen, J. 1995, ApJ, 445, 55
668: \bibitem[J\o rgensen et al.(1995)]{JFK95} J\o rgensen, I., Franx, M., \& Kjaergaard, P. 1995, MNRAS, 276, 1341
669: \bibitem[J\o rgensen et al.(1996)]{JFK96} J\o rgensen, I., Franx, M., \& Kjaergaard, P. 1996, MNRAS, 280, 167
670: \bibitem[Kelson et al.(2000)]{Kelson00} Kelson, D.D., Illingworth, G.D., van Dokkum, P.G., \& Franx, M. 2000, ApJ, 531, 137
671: \bibitem[Kodama et al.(1998)]{Kod98} Kodama, T., et al. 1998, A\&A 334, 99
672: \bibitem[La Barbera et al.(2000)]{LAB00} La Barbera, F., Busarello, G., \& Capaccioli, M. 2000, A\&A, 362, 851
673: \bibitem[La Barbera et al.(2004)]{LMB04} La Barbera, F., Merluzzi, P., Busarello, G., Massarotti, M., \& Mercurio, A. 2004, A\&A, 425, 797
674: \bibitem[La Barbera et al.(2005)]{LBC05} La Barbera, F., et al. 2005, MNRAS 358, 1116
675: \bibitem[La Barbera et al.(2008)]{LBdC08} La Barbera, F., et al. 2008,
676: PASP, 120, 681
677: \bibitem[Lawrence et al.(2007)]{Law07} Lawrence, A., et al. 2007, MNRAS, 379, 1599
678: \bibitem[Mobasher et al.(1999)]{MGA99} Mobasher, B. et al. 1999, \mnras, 304, 225
679: \bibitem[Pahre et al.(1998a)]{PdCD98a} Pahre, M.A., de Carvalho, R.R., \& Djorgovski, R.R. 1998a, AJ 116, 1606
680: \bibitem[Pahre et al.(1998b)]{PDdC98b} Pahre, M.A., Djorgovski, S.G., \& de Carvalho, R.R. 1998b, AJ 116, 1591
681: \bibitem[Prugniel \& Simien(1996)]{PrS96} Prugniel, P., \& Simien, F. 1996, A\&A 309, 749
682: \bibitem[Schlegel, Finkbeiner and Davis(1998)]{SFD98} Schlegel, D., Finkbeiner, D.P., \& Davis, M. 1998, ApJ, 500, 525
683: \bibitem[Scodeggio et al.(1998)]{SCO98} Scodeggio, MM., et al. 1998, \mnras, 301, 1001
684: \bibitem[Scodeggio(2001)]{Scodeggio:01} Scodeggio, M. 2001, AJ, 121, 2413
685: \bibitem[Shen et al.(2003)]{Shen03} Shen, S., et al. 2000, 343, 978
686: \bibitem[Sorrentino et al.(2006)]{SAR06} Sorrentino, G., Antonuccio-Delogu, \& V., Rifatto, A. 2006, A\&A, 460, 673
687: \bibitem[Thomas et al.(2005)]{THO05} Thomas, D., et al. 2005, ApJ 621, 673
688: \bibitem[Trujillo et al.(2004)]{Tru04} Trujillo, I., Burkert, A., \& Bell, E. 2004, ApJ 600, 39 (TBB04)
689: \bibitem[Vazdekis(1996)]{Vazdekis:96} Vazdekis, et al. 1996, ApJS, 106, 307 (Vaz96)
690: \bibitem[Warren et al.(2007)]{War07} Warren, S.J., et al. 2007, astro-ph/0703037
691: \bibitem[Zaritsky et al.(2006)]{ZGZ06} Zaritsky, D., Gonzalez, A.H., \& Zabludoff, A.I. 2006, \apj, 638, 725
692: \bibitem[Zaritsky et al.(2008)]{ZZG08} Zaritsky, D., Zabludoff, A.I., \& Gonzalez, A.H. 2008, astro-ph/0711.2071v1
693: \bibitem[Zibetti et al.(2002)]{ZGS02} Zibetti, S., et al. 2002, ApJ 579, 261
694:
695: \end{thebibliography}
696:
697:
698:
699:
700:
701:
702: \end{document}
703: