A&A 409, 479-484 (2003)
DOI: 10.1051/0004-6361:20031136

Near-infrared color evolution of LMC clusters[*]

J.-M. Kyeong1 - M.-J. Tseng2 - Y.-I. Byun3

1 - Korea Astronomy Observatory, Taejon, 305-348, Korea
2 - Institute of Astronomy, National Central University, Chung-Li, 32054, Taiwan, ROC
3 - Yonsei University Observatory and Department of Astronomy, Yonsei University, Seoul, 120-749, Korea

Received 5 May 2003 / Accepted 7 July 2003

We present here the digital aperture photometry for 28 LMC clusters whose ages are between 5 Myr and 12 Gyr. This photometry is based on our imaging observations in JHK and contains integrated magnitudes and colors as a function of aperture radius. In contrast to optical colors, our near-infrared colors do not show any strong dependence on cluster ages.

Key words: galaxies: photometry - galaxies: Magellanic Clouds - galaxies: star clusters - infrared: stars

1 Introduction

Star clusters of the Large Magellanic Cloud (LMC) provide excellent templates for studies of stellar populations in external galaxies. The LMC is close enough that star clusters within the galaxy can be resolved into individual stars by high spatial resolution imaging from both ground and space based telescopes. Their CMDs can be used for detailed age calibration. These star clusters are also populous enough that their integrated colors are less sensitive to stochastic effects when compared to galactic open clusters. Most importantly, unlike globular cluster in our Galaxy, clusters in the LMC are known to cover wide range of ages, making them ideal objects to investigate the integrated spectral behavior of simple stellar population as functions of both age and chemical composition.

Integrated photometry for LMC clusters had been studied since van den Bergh & Hagen (1968), who performed UBV photoelectric aperture photometry. In 1980, Searle, Wilkinson and Bagnuolo carried out an extensive program of uvgr aperture photometry, which resulted in a classification scheme of age-calibration with the reddening-free parameter Q(vgr) and Q(ugr). This scheme is now called the SWB classification. Their work showed that the integrated colors appear to correlate with cluster ages, and can be used as a primary age indicator. Elson & Fall (1985) translated the SWB classification into the UBV two-color diagram and also introduced a reddening independent parameter, s, which agrees with the SWB sequence for 90% of clusters. Girardi et al. (1995) revised the relation between the parameter s and cluster ages to consider the dispersion of observed colors due to stochastic effects and metallicity variation.

Table 1: Basic properties of LMC clusters in our sample.

Previous effort in near-infrared includes Persson et al. (1983, hereafter Persson83) and Ferraro et al. (1995). Persson83 studied the integrated light of 84 clusters in the Large and Small Magellanic Clouds using a single cell photoelectric aperture photometry. In contrast to the UV and optical clusters colors which vary smoothly with age, their infrared integrated colors display a large scatter among a given SWB classification. They defined a group of IR enhanced clusters that belong to SWB groups IV-VI, i.e., in the intermediate-age with the very red J-K and H-K colors. This color excess is due to the existence of luminous carbon stars on the asymptotic giant branch, which are responsible for as much as half of the bolometric luminosity in the near-infrared.

Ferraro et al. (1995) was the first who used a small, but two-dimensional InSb array for near-infrared observation of LMC clusters. With the classical and overshooting stellar evolution theories, they investigated how red stars dominate the bolometric luminosity of a simple stellar population after its early evolutionary stage. These red stars are considered to be in phase transitions; asymptotic giant branch (AGB) and red giant branch (RGB) branch, in particular. Their overall result is consistent with the earlier work by Persson83. The age dependent variation for V-K (substantially controlled by AGB and RGB stars) shows a large scatter, which makes it almost useless for the purpose of age calibration as well as for the evolutionary study of simple stellar populations.

Using a large format detector which enables more careful background estimates and covers a larger cluster area, we aim to collect a homogeneous data set for LMC clusters in the near-infrared and re-examine the age dependence of integrated infrared color.

This paper is a part of a long term study to understand the broad band color evolution of stellar systems with both simple and more complex stellar populations.

2 Sample and observation

2.1 Samples

A total of 28 LMC clusters have been selected for our infrared photometry. These clusters were chosen because all of them have been already studied with color-magnitude diagrams (CMD) and cover a wide range of ages from 5 Myr to 12 Gyr as determined by isochrone fitting to the CMDs. Table 1 gives the following basic information for each of our sample clusters.

2.2 Observations

Near-infrared observations were carried out during the nights of Dec. 20-22 in 1996 using the Australian National University 2.3-m telescope and the CASPIR (Cryogenic Array Spectrometer/Imager) instrument at the Siding Spring Observatory. The CASPIR system has InSb $256\times256$ chip with pixel size of 0.5 arcsec at the Cassegrain focus. We used the double sampling readout mode, whose characteristics can be found in McGregor (1995).

In order to handle possible detector instability and temporal changes of the bright IR background, cluster exposures were taken in a loop sequence. Each loop takes images in the following order ; bias - sky east - cluster - sky west - sky north - cluster - sky south - bias - dark. In most cases, the sky exposures were made in regions about 5 arcmin away from the center of the target cluster in the directions of east, west, north and south. When there was a bright contaminating source in the sky field, we moved the sky into another place nearby. The exposure time was usually 2 s long, except when the cluster has very bright center or stars.

By observing the standard stars in the beginning of the night, at midnight, and again at the end of the night, we attempted to get good sample of calibration data. A set of flatfield frames were obtained using artificial illumination of the inner surface of the dome to correct the pixel-to-pixel sensitivity variations across the array. The domeflat exposures contain a few pairs of lamp-on and lamp-off exposures for each filter, in order to achieve good signal statistics.

Table 2 lists our observation log, including date, cluster name, airmass, exposure time, and comments on the weather condition or centering problem if suspected.

3 Data reduction and calibration

We used the IRAF package including MSSSO CASPIR specific routines for basic image processing, and developed a set of software for final digital aperture photometry.

The IR data reduction method is more complex than optical CCD data mainly because of high sky background and its rapid variability. The CASPIR detector array also shows a quadratic non-linearity, which should be linearized to recover the low and high intensity information accurately (McGregor 1995). Each data frame was bias and dark subtracted and, was then median-combined to create a sky frame for that sequence. After subtracting the corresponding sky frames, the data images were subjected to flatfielding.

For each filter, the target clusters are exposed twice. Due to the excellent tracking and pointing capability of the ANU 2.3-m telescope, the two images turned out to be almost exactly aligned without needing an extra step of image relocation. These two images are stacked together by simple averaging. The resulting images are then ready for photometric measurement and analysis.

The instrumental magnitudes of 13 SAAO standard stars (Carter & Meadows 1995) are obtained by digital aperture photometry. The optimal size of aperture was chosen to be 20 pixels in radius after experiments with growth curves. We derived the transformation between our instrument magnitudes and the standard values J, H, K as follows

\begin{eqnarray*}j &=& J - 19.086(\pm0.004) + 0.112(\pm0.004) * X_j \\
& & + ~...
...4) * X_h \\
k &=& K - 18.022(\pm0.024) + 0.096(\pm0.019) * X_k

where j, h and k are instrumental magnitudes, Xj, Xh and Xk are airmass. We could not find any significant color term in the H and K bands. Residuals to the standard system are displayed in Fig. 1 and show the accuracy of our calibration.

\end{figure} Figure 1: Residuals between calibrated and standard magnitudes in J, H and K.
Open with DEXTER

4 Photometry

4.1 Concentric aperture photometry

Conventional aperture photometry based on photoelectric photometers requires exact centering of the aperture to the cluster center. This can be sometimes a rather difficult task due to contamination of bright stars. It is also difficult to maintain the same position for all the filters if the telescope has to be moved back and forth for sky measurements. The array observations such as ours do not have this problem.

\end{figure} Figure 3: Comparison with Persson83's photoelectric photometry in J,H,K ( a), $J-K,\ H-K$ ( b). Open circles refer to small aperture sizes of 12, 24, 29, 30 and 35 arcsec and filled circles are for bigger apertures of 56, 59, 60, 64, 75 arcsec.
Open with DEXTER

Using fully reduced images, we defined the exact location of our center by examining each image with the following methods. First, we combine all the JHK images into one near-IR image and then apply visual inspection. Secondly, we use the BLKAVG task in IRAF to project the two dimensional image into X- and Y-directions and locate the maximum. Finally, by interactive profile extraction in X- and Y-direction, using IRAF/imexam, while avoiding bright stars scattered in the cluster field.

The centers determined by the above methods do not differ much. Some exceptions do exist however when there are strong clumps of stars near the general area of cluster center. These clusters are marked as such in Table 2.

After identifying the location of cluster center, we used our privately written program of digital aperture photometry to measure the flux within several apertures of different radii. The results, all converted to standard magnitudes and color are presented in the following section.

4.2 Results

(a) Magnitude and color profile for each cluster

In order to investigate the radial propagation of integrated magnitudes and colors, we have made multi-aperture concentric photometry with aperture size of 2, 4, 7, 10, 15, 20, 30, 40, 60, 80, 100 arcsec. The J, H and K magnitudes are tabulated in Table 3.

The graphical illustration of magnitude and color are shown in Fig. 2. In the upper panel, the lines with data points are the growth curve of J, H and K, i.e. encircled flux plotted against log R. The error bars are mostly very small and do not show up in the figure except for a few data points. In the lower two panels, the integrated color J-K and H-K are plotted against log R.

As shown in Fig. 2, most clusters show only a small color variation with no big change around diameter 60 arcsec. This tells us that the color with this aperture is representative of each cluster and useful for our purpose.

(b) Comparison with Persson83

In order to see how our data compare with the photometry of Persson83, we performed another set of aperture photometry with apertures of the same size as used by Persson83. The results are illustrated in Fig. 3, which shows the residuals in J, H, K, and also in J-K and H-K.

Because Persson83 uses the CIT/CTIO photometric system, we converted their data into our SAAO system using the transformation equations given by Carter (1990).

If there had been serious problem with centering, the discrepancy will be bigger for small apertures. This is because of the strong central concentration of light in clusters and also because the small area coverage will be more subject to an error in using the wrong center. However we do not see any systematically bigger scatter for small aperture measurements. The residuals in H and K bands have mean values close to zero with a scatter less than 0.1 mag, indicating that our photometry is consistent with Persson83.

5 Discussion

With the optical data collected in Table 1 and near-infrared data from Table 3, we present in Fig. 4 the optical and near-IR color variation as a function of cluster age. SWB group boundaries are indicated in the upper part of this diagram. It is apparent that the optical colors U-B and B-V show very strong relation with age. As clusters get older the optical colors get redder almost monotonically. Beyond 1 Gyr, the slope becomes rather flat and even falls off towards the blue. But the latter is probably caused by the oldest cluster NGC 1786 with SWB VII, which has very low metallicity. Compared to the optical colors, the changes in near-IR colors J-K and H-K are not significant nor monotonic.

\end{figure} Figure 4: Optical color evolution (from Searle et al. 1980) and near-infrared color evolution of LMC clusters. Our data (filled circles) are sampled at 60 arcsec diameter and log t is from our Table 1. Persson83 data are shown for comparison (open circles). All color are reddening corrected by Bessell & Brett(1988)'s extinction law.
Open with DEXTER

\end{figure} Figure 5: Our data plotted against log t from our Table 1. For near-infrared colors, the data are sampled at 60 arcsec diameter, but for V-K we resampled our near-infrared data to match the V aperture given in Table 1. Horizontal dashed line are defined by Persson83 for IR enhanced clusters, shown here after Carter (1990) correction to the SAAO system. Cross symbols indicate 4 clusters classified as IR enhanced by Persson83.
Open with DEXTER

Persson83 defined the globular clusters with $(J-K)_0 \geq 0.8$, $(H-K)_0 \geq 0.2$ and $(V-K)_0 \geq 2.5$ as "IR enhanced clusters". Within each group of globular and open clusters, there is a tendency that the color becomes bluer as age sequence increases. However, it is not at all the case in our data. Figure 5 shows our data re-plotted for the same colors and for the same scale as in Persson83's Fig. 2. We particularly note the following points;

As Persson83 noted, we also find a transition between younger and older clusters. This appears to happen at the age of 100 Myr, which corresponds to the border between SWB II and III. Persson83 on the other hand observed that this happens in SWB IV and V. Our transition is also less abrupt than Persson83's. Among early SWB I and II, the near-infrared colors get somewhat bluer with increasing age, while among late SWB types the trend seems to get reversed.

Persson83 also noted a substantial population of IR enhanced clusters among old clusters of SWB IV and V. Our data also confirm their existence; our sample includes four of Persson83's IR enhanced clusters and they are all IR bright. However, this IR enhancement does not appear to be limited to a small range of ages. Clusters of comparable IR colors do exist in early SWB types of 0, I and II. These clusters were not included in Persson83's sample.

The near-infrared color vs. age relation has a bigger scatter than in the optical. The origin of scatter in our near-infrared color vs. age relations will be studied in depth after we perform digital stellar photometry on our image data and also after running some color synthesis experiments based on the results of stellar photometry. The most likely cause of the scatter appears to be due to small number statistics for bright and cool giant stars, whose effect is much more pronounced for near-infrared colors.

To sum up, our near-infrared observations suggest that the age dependent variation of infrared color is minor. Also, even if it exists with a small slope, it would be difficult to define the trend directly from observations for small stellar systems such as LMC clusters. This does not mean that the construction of empirical infrared color evolution is impossible from this data. As suggested above, proper reconstruction of stochastic effects via digital photometry and simulations may prove still useful to define the near-infrared color evolution of stellar systems.

6 Summary

We carried out near-infrared imaging observations for 28 clusters in the LMC. These clusters cover a large span of ages. For each cluster, we performed digital aperture photometry with several different circular apertures and presented here the magnitudes and colors as a function of aperture size. The precision of our data has been checked internally with standard star observations and externally with previous photoelectric observations of Persson83 after proper filter correction.

The age dependence of near-infrared integrated color appears to be much smaller compared to that of optical colors. We also find the presence of a slope transient near 100 Myrs. Our color-age relationship shows a scatter much bigger than those of optical correlations. Presently we interpret this as a result of statistical effects involved with small numbers of bright and cool giant stars in the small stellar systems such as LMC clusters.

Population synthesis and Monte Carlo modelling of stochastic effects (e.g. Santos & Frogel 1997) can be a useful approach to the problem. In our subsequent work, however, we will estimate the stochastic effect in a more direct way by detailed study of digital stellar photometry using the present array data.

We gratefully acknowledge the careful reading and suggestions made by our referee R. A. Johnson. We also would like to thank the staff of the Australian National University for their help during our observing run at the Siding Spring Observatory. This research was supported by Korean Research Foundation Grant (KRF-2000-015-DP0445).


Online Material

\end{figure} Figure 2: Magnitude and color profiles of LMC clusters. In upper panel solid line represents J, dotted line for H, dashed line for K. Straight vertical line indicates the location of R=60 arcsec.
Open with DEXTER

\par\includegraphics[height=11cm,clip]{3939f2_2.ps} \end{figure} Figure 2: continued.
Open with DEXTER

  \begin{figure}\par\includegraphics[height=11cm,clip]{3939f2_3.ps} \end{figure} Figure 2: continued.
Open with DEXTER

  \begin{figure}\par\includegraphics[height=11cm,clip]{3939f2_4.ps} \end{figure} Figure 2: continued.
Open with DEXTER

  \begin{figure}\par\includegraphics[height=11cm,clip]{3939f2_5.ps} \end{figure} Figure 2: continued.
Open with DEXTER

  \begin{figure}\par\includegraphics[height=11cm,clip]{3939f2_6.ps} \end{figure} Figure 2: continued.
Open with DEXTER

Table 2: Observing log.

Table 3: Concentric aperture photometry for each cluster.

Copyright ESO 2003