1. Introduction

When a star of low to intermediate main sequence mass has reached the stage where its central part is built up of a degenerate nucleus of carbon and oxygen, it starts ascending the asymptotic giant branch (AGB) of the Hertzsprung-Russell diagram. While burning hydrogen and helium in a shell around the nucleus it will expand its envelope. The star pulsates and starts losing mass at a high rate (10^-7 to ). The star will be obscured in the visible wavelengths by its dense shell of circumstellar matter. In this slowly expanding and cooling shell molecules form and with regularity conditions are met, such that the molecules in the circumstellar shell support a maser.

Since the discovery of the double peaked profile of the hydroxyl maser satellite line at 1612 MHz in late-type infrared stars, the so-called OH/IR stars have been studied to investigate stellar evolution and Galactic dynamics. The characteristic double spectral feature originates from hydroxyl (OH) masers in the expanding circumstellar shells of oxygen rich AGB stars. It is easy to recognise the object as an OH/IR star; one obtains the position and line-of-sight velocity of the star as well as the shell expansion velocity directly from the double peaked spectrum.

As the variable stellar radiation at visible and near-infrared wavelengths is absorbed, the only means of investigating the properties of the underlying star is by observing radiation that is re-emitted by the shell: infrared and sub-mm emission of the dust and gas, in which the molecules can show maser emission (OH, H₂O and SiO). Reviews on the star and its envelope can be found in Iben & Renzini (1983) and in Habing (1996).

Apart from studying the underlying star, one can also make use of OH/IR stars to probe the Galaxy for its structure, evolution and dynamics. For example, Whitelock et al. (1991) derived a period-luminosity relationship similar to the Mira variables and Blommaert et al. (1994) compared the OH/IR stars in the center, bulge and outer part of our Galaxy. Good reviews about using OH/IR stars as tools, are from Habing (1993) and Dejonghe (1993).

The Galactic center

Lindqvist et al.\ (1992a; hereafter LWHM) surveyed the Galactic Center (GC) region for OH/IR stars with the Very Large Array (VLA) in 1984 and 1985. The OH/IR stars in the GC are one of the few stellar samples in the GC that can be directly observed despite the enormous visual extinction in front of the GC. The total of 150 OH/IR stars found (see appendix) are all at about the same distance and thus can be studied as a sample without assumptions on their individual distances (e.g. Baud et al. 1981; Jones et al. 1994; Blum et al. 1996). In Lindqvist et al. (1992b) the stars are investigated for their spatial and kinematic properties and used as tracers of the central potential and mass distribution. It is shown that the distribution of the GC sample generally depends on the shell expansion velocity at 1612 MHz and that the surface density increases strongly towards Sagittarius A* (Sgr A*). Sevenster et al. (1995) have shown that the OH/IR stars in the GC consist of a global Galactic component and a separate, strongly rotating disk of "younger" OH/IR stars, possibly formed at a distinct event. Unfortunately the number of OH/IR stars known is too low to do a conclusive dynamical study, especially within about 10-20 pc of Sgr A*. Lindqvist et al. (1997) show that there should be many more OH/IR stars in the GC with apparent weak OH masers. The reason why they have not been found in the LWHM survey is mainly a matter of sensitivity. Also the OH masers vary in luminosity, as they are indirectly pumped by the variable stellar radiation. To find all stars one should preferably observe and search the same region for more than one epoch.

By monitoring the variability of some of the LWHM OH sources with the VLA, Van Langevelde et al. (1993; hereafter vLJGHW, or the "monitor") have tried to measure phase-lag distances to these OH/IR stars in order to get a direct estimate of the distance to the GC. However, a highly scattering interstellar medium was discovered in the direction of the GC, making it impossible to achieve their primary goal (Van Langevelde & Diamond 1991). Nevertheless, 20 observations of the GC region had been done, each in sensitivity comparable to the original LWHM survey observations. A "cheap" way of finding faint OH/IR stars is by analysing the concatenated data set. In that way the search can be done in a high-sensitivity data cube, and one is able to detect OH/IR stars which were in a minimum of their OH maser luminosity at the time of the LWHM survey. By averaging many different epochs taken over a time longer than the typical periods of the stars, the detection becomes effectively a function of the time averaged flux density in the spectral peaks. With 20 epochs, the most sensitive way to find stars is by using the concatenated data rather than to search all epochs separately (under the assumption that the OH/IR stars vary typically a factor of two during the monitor).

We intend to use the new, extended sample of OH/IR stars for testing the location of, and probing the potential in the very center of our Galaxy, and, secondly, to study the sample of OH/IR stars in the GC compared to all other known samples of OH/IR stars. To overcome asymmetry problems, introduced by the particular pointing of the VLA data (optimised for the monitoring program; see below), we used the Australia Telescope Compact Array (ATCA). The bandwidth for the ATCA observations has been chosen to include an equal sensitive search for high-velocity OH/IR stars (velocity up to ), which might add important clues for future dynamical modelling.

Outline of this paper

In this paper we describe the data reduction procedure of the VLA monitor data set and of the ATCA observations in Sect. 2. Section 3 presents a list of both known and suspected OH/IR stars in the GC. Here, when we refer to an OH/IR star, we actually refer to circumstellar OH maser emission. The large interstellar visual extinction prevents us to make a clear distinction between optically thick circumstellar shells, as for genuine OH/IR stars, and optically thin circumstellar shells as for the evolutionary closely related Mira variables. We may even have picked up an individual supergiant; however, see Blum et al. (1996) for recent evidence that the number of supergiants in the GC is low. Section 3 also includes an error budget for our measurements. In Sect. 4, we comment on some of the detections. We discuss the survey sensitivity and anomalous features in some OH/IR star spectra. Briefly, we compare the previously unknown OH/IR stars with the known OH/IR stars and derive the OH luminosity distribution. From this we conclude in Sect. 5, that the central stars of the new detections are not different from the AGB stars that constitute the known population in the GC.

We do not attempt to measure the periods of the individual detections as the objects are too faint to be detected in each epoch separately. Also, detailed discussion of the spatial, kinematic and physical properties of the new, extended sample of OH/IR stars in the GC, as well as the issue whether Sgr A* is the dynamical center of the OH/IR star sample, is deferred to additional papers. A preliminary result on the survey can be found in Sjouwerman & Van Langevelde (1996).