We present initial results of a study of low surface brightness dwarf galaxies within eight different clusters (A262, A400, A426, A569, A1367, A1656, A2635, A3526) as part of our program to determine the clustering properties, luminosity functions, and morphologies of dwarf galaxies in a wider range of cluster environments. Deep V-band images covering up to 1 square degree in each cluster were obtained in addition to control fields which are used to statistically correct for background galaxy contamination. In addition, fiber spectroscopy data has been obtained for over 500 galaxies in A3526. This data will be used to study the mass contribution of and the effect of the cluster environment on these low mass galaxies.

1. Introduction

Dwarf galaxies are the most common type of galaxy thus far discovered and may be substantial contributers of mass to the universe; with a high enough M/L ratio, dwarfs and their associated dark matter halos may even compose a significant portion of the overall cluster dark matter fraction. Our work addresses three issues related to dwarf galaxies:

Dwarf Galaxy Luminosity Functions - First, the data will be used in determining the cluster luminosity functions. The faint-end of the luminosity function in clusters has been difficult to establish due to complex selection effects related to the detection of low luminosity, low SB systems. Different cluster environments may play a significant role in determining the numbers and morphologies of member galaxies causing cluster-to-cluster variations in the luminosity function. A thorough understanding of the faint-end slope will determine if dwarfs (and their associated dark matter halos) can account for a substantial fraction of the cluster dark matter content.

Environmental Influences on Dwarf Properties - Because the dwarfs are of low mass, they are more fragile than giant galaxies and thus are a better indicator of environmental effects. Morphology, luminosity, structure (such as existence of a nucleus), and surface brightness may all be affected by environmental factors. Specifically, we shall look at the dwarf properties with respect to their radial cluster distributions (and proximity to larger galaxies) and between different clusters. For example, previous work has revealed that the dwarf/giant ratio may increase with increasing cluster richness (Ferguson & Sandage 1991) and Thompson and Gregory(1993) find a drop in the density of dwarf spheroidals toward the Coma cluster core. Preliminary data analysis of A426 appears to support this latter trend while A2634 does not.

Dwarfs at the Current Epoch - Finally, this data set will be used to determine the cluster-to-cluster variations in dwarf properties expected at the current epoch, which is essential to interpret observations at higher redshift.

2. Observational Overview

Observations of A3526 were acquired on 13-20 May 1993 on the LCO 1m with a TEK 3 2048 x 2048 CCD camera at prime focus. The seven northern clusters were observed between Oct. 1994 and Dec. 1996. These were all taken at the Michigan-Dartmouth-MIT (MDM) 1.3m telescope with a Loral 2048 x 2048 CCD. All data were taken in the V-band for maximum light transmission and clusters were observed in a mosaic pattern with each field exposed 4 X 15 minutes for a total exposure of 1 hour. Control fields were also taken in the same manner 3 - 4 degrees away from the cluster center. All data were calibrated directly from observations of standards taken on each observing run. For fields in which same night standards were not available, stars in overlap regions of neighboring fields were used to obtain the calibration. Total field coverage for each cluster is about 1 square degree. The grey scale image (Fig. 2) displays the spatial coverage of A3526. While all data is currently in the V band, B-band data for at least A1367 will be obtained in a future observing run. In May 1995, a spectroscopic followup was made of the brightest ~ 500 galaxies in A3526 using the LCO 2.5m with a 2D Frutti + fibers. A 600 line/mm grating was used to obtain a spectral resolution of 8.5 angstroms over the region 3800-6500A. Spectra were taken in 5 different setups and include cluster galaxies as faint as Mv = -14. The table below provides a summary of the CCD data obtained.

3. Data Reduction

Great attention was paid to flat fielding and cosmic ray removal in order to remove intrinsic variations from the chip and to allow for the detection of very low surface brightness galaxies. Each image was first flattened with twilight sky flats. A second flat was then made by combining 10 flattened and normalized control fields. To avoid undesirable artifacts, we used a percentile clipping algorithm after removing the brightest pixels in each set of 10. All images were then divided by this flat and the 4 images in each field were summed after cosmic ray removal. The final sky fluctuations in the summed images were less than 0.2% of the mean sky level. We used the software package, FOCAS (Valdez 1989), to locate the dwarf galaxies. FOCAS locates the objects in an image by searching for a specified number of contiguous pixels with values greater than a specified threshold above the local sky. Best parameter values were determined to be 10 contiguous pixels and 3.5sigma for the detection of the lowest surface brightness galaxies without the detection of noise. The objects were classified by defining a PSF and all objects not classified as galaxies were removed from the catalog. The final list was then verified by eye to remove any spurious detections. Control fields were reduced in exactly the same manner and were used to determine the contribution of background galaxies in each magnitude bin. In both the cluster and control images, an assessment of completeness must be made. For this, we are using false galaxy simulations. These are added to cluster and control fields in a systematic manner by running simulations of galaxies of successive surface brightness and magnitude ranges. Fiber data was reduced using the HYDRA package in IRAF and radial velocities were determined via Fourier cross correlation using the IRAF task FXCOR on absorption line spectra. Velocity errors were typically in the range 50-100 km/s although they increased at fainter magnitudes. Most of the velocities obtained for low surface brightness dwarfs, however, were through emission line spectra. Each of the five setups contains at least 100 spectra and at fainter magnitudes the success rate at determining velocities drops to ~ 20%. We expect to obtain a greater percentage of reliable velocities with a higher S/N template in the future. A summary of the spectral results is given in the table below.

4. Discussion


Luminosity function

Figure 3 shows the raw log (number / mag) vs mag counts in Centaurus along with the background (control field) counts while in figure 4, background subtracted cluster counts scaled to a square degree in area are shown. For most bins, a significant excess over background counts is observed. A best fit straight line to the faint magnitudes (Mv > -15) yields a slope of -0.469. We have yet to correct the counts for completeness and it is readily apparent that incompleteness sets in by m = 22 (Mv = -11). We have also not taken into account extinction by cluster dust, excess counts due to globular clusters, differences in large scale structure between the control and cluster fields, or crowding in the fields. Completeness is currently being assessed by systematically adding false galaxies in successive magnitude and surface brightness bins to cluster fields and determining the recovery efficiency of FOCAS in each magnitude bin. Three half-magnitude bins show an excess of galaxies in the controls. This may be due to FOCAS splitting up large or diffuse galaxies, missing diffuse haloes altogether, or due to some physical cause such as inauspiciously observing control fields in a region of higher background density than the clusters. Of our five control fields, two were thrown out because they did land on a background cluster.



In figures 7 and 8, we show spectra of 5 galaxies (In fig. 8, the spectra correspond to the image 6 numbers:7,20,16). In the images(fig. 5 and fig. 6), members are shown in green rings while higher redshift galaxies are shown in red. Centaurus is at a redshift of .01. In several cases we find galaxies that would normally, through morphology, be classified as cluster dwarfs but which prove to be background galaxies instead. Some of these galaxies are low surface brightness while others are high surface brightness but M32-like. Most LSB dwarfs that we were able to obtain velocities for are from emission line spectra as can be seen here. In figure 6, there is at least one obvious LSB dwarf that we were unable to obtain a reliable velocity for. As can be seen in Table 2, at fainter magnitudes and surface brightnesses there is a sharp drop-off in the number of galaxies for which were are able to obtain velocities for and we appear to have reached the limits of the detector by m = 19.


Figure 1

A field in Centaurus in which several different galaxy morphologies are visible. Note the large LSB galaxy in the bottom left.

Figure 3

shows the number of raw counts in the Centaurus cluster fields and the counts in the control fields (scaled to the total cluster area). Three bins show an excess of objects in the controls which may be an error due to FOCAS splitting large and diffuse objects in this magnitude range.

Figure 4

shows the cluster counts backgroundcorrected and scaled to 1 square degree. The best fit slope of the faint magnitudes (m > 18) is .469 excluding the discrepant bins andthe region m > 22 where incompleteness sets in. We do reach a limitingapparent V magnitude of ~24 and Mv of -8.

Figure 5-8

show several examples of our spectral results. In Figs. 5 and 6, green rings around galaxies represent cluster members while red rings represent higher redshift galaxies. In Figs. 7 and 8 we show the spectra of 5 of these galaxies. Small bright galaxies often were determined to be at higher redshifts (ie 20) although this was not always the caseas can be seen in 17 and 13. The larger LSB galaxies that we were ableto obtain velocities for are predominantly from emission lines as is the case with 23 and 16. For these, the most obvious lines are the [OII]lambda3727 and the triplet of Hbeta and [OIII]\lambda\lambda4959,5007.