Astron. Astrophys. 359, 337-346 (2000)

1. Introduction

Supernova remnants (SNRs) have long been thought to be the dominant source of cosmic rays (CR) at energies below 100 TeV (for a review see e.g. Blandford & Eichler 1987). SNR, via the diffusive shock process, are able to accelerate electrons and hadrons and meet the energetics of the observed cosmic rays. The TeV gamma-ray flux predicted from SNR is the most accessible tracer of CR acceleration and its detection would be convincing evidence for the SNR origin of galactic CR. Models of the TeV gamma-ray emission from SNR predict distinct spectral features, according to the hadronic and/or electronic nature of the parent CR accounting for the gamma-ray flux (see Drury et al. 1994, Naito & Takahara 1994, Baring et al. 1999, and references therein for a summary).

Ground-based surveys of SNR at gamma-ray energies (TeV to PeV) have been carried out on several promising northern hemisphere candidates (e.g. IC443, Tycho's SNR, W51, W44, G78.2+2.1). The Whipple (Buckley et al. 1998, Lessard et al. 1999), HEGRA (Hess 1997at TeV energies, and Prosch et al. 1996at multi-TeV energies), CAT (Goret et al. 1999) and CYGNUS (Allen et al. 1995) groups have reported upper limits. Recently however, the HEGRA has seen marginal evidence for TeV gamma-rays from the young SNR Cas-A, after deep observation (Pühlhofer et al. 1999). In the southern hemisphere, the CANGAROO has reported the detection of TeV gamma-rays from SNR SN1006 (Tanimori et al. 1998) and SNR RX J1713.7-3946 (Muraishi et al. 2000), and if confirmed, will be strong evidence in favour of the production of cosmic rays electrons in SNR.

W28 (also SNR G6.4-0.1 from Green 1998) is considered an archetypal composite (mixed or M-type) supernova remnant, characterised by a centrally filled X-ray and shell-like radio morphology (Rho & Petre 1998, Long et al. 1991). The ROSAT X-ray emission appears best explained by a thermal model (Rho et al. 1996) although Tomida (1998) from the analysis of ASCA data, has suggested the presence of a weak a non-thermal component in the south west region. The limb-brightened radio emission (20, 6 & 2 cm) shows a synchrotron spectrum of varying spectral index (Andrews et al. 1983). A radio point source at (G6.6-0.1) is defined (Altenhoff et al. 1978, Andrews et al. 1983), hereafter referred as A83 in this paper. A glitching radio pulsar, PSR J1801-23 (PSR B1758-23, 416ms, ss^-1), lies at the northern radio edge (Kaspi et al. 1993). An upper limit to this pulsar's characteristic age is estimated at 58 000 years, and it's spin-down luminosity ( erg s^-1) is at the lower edge of luminosity values when compared to the known gamma-ray (EGRET & COMPTEL) pulsars.

The age of W28 is estimated (Kaspi et al. 1993) in the range 35 000 to 150 000 years, with upper and lower limits taken from the assumptions that W28 is currently in either the radiative or Sedov phases of expansion. According to Kaspi et al. 1993, the distance of PSR J1801-23 (9 to 16.5 kpc) derived from it's dispersion measure (DM) appears inconsistent with that derived for the remnant. Estimates for the remnant distance are set at 1.8 kpc (Goudis 1976-D relation) and 3.3 kpc respectively (Lozinskaya 1981, from mean optical velocities), indicating that the pulsar/W28 association is possibly a line-of-sight coincidence. However, Frail et al. (1993) have noted the large uncertainty in using the DM as a distance estimate for this pulsar due to the high concentration of ionised material in the line of sight, and conclude there is sufficient evidence for the pulsar/remnant association. The unidentified EGRET source 3EG J1800-2338 (95% error circle 0.32^o radius) (Hartman et al. 1999), listed as 2EG J1801-2312 in the second EGRET catalogue (Thompson et al. 1996), lies on the edge of the radio shell and was thought to be associated with the remnant (Esposito et al. 1996, Zhang & Cheng 1998). 3EG J1800-2338 has a relatively hard spectral index (Hartman et al. 1999) with no apparent sign of a turnover at 1 GeV (Merck et al. 1996). Lamb & Macomb (1997) also point out that 3EG J1800-2338 is visible above 1 GeV at 5.4 significance, and is centred very close to the A83 radio position. The 3EG position of the EGRET source is displaced by about 0.5^o relative to the 2EG position, yet still lies comfortably within the SNR radio shell, and remains a strong example of an EGRET source/SNR association (Romero et al. 1999). The 3EG error circle however, now excludes PSR J1801-23 and the molecular clouds.

W28 lies in a complex region of the galactic plane with many HII regions and dense molecular clouds (Wootten 1981) contributing to the ISM surrounding the SNR. Over forty OH (1720 MHz) maser emission sites are concentrated at the eastern and northern edges of the SNR (Claussen et al. 1997), along the SNR and molecular cloud interface. The distribution of shocked and unshocked gas in this region is also consistent with the idea of the SNR shock passing through the cloud (Arikawa et al. 1999). OH maser emission (1720 MHz) is considered a strong indicator of collisional pumping with matter densities cm^-3 (Claussen et al. 1997 , 1999). Enhanced levels of TeV -ray emission via the decay of neutral pions may be expected from such areas associated with the masers and molecular cloud (Aharonian et al. 1994). Fig. 1 indicates the sites of interest in relation to the radio continuum emission (327 MHz). The presence of these interesting objects make W28 a prime southern hemisphere candidate for study at TeV gamma-ray energies.

Fig. 1. Radio continuum (327 MHz) for W28, adapted from Claussen et al. (1997). Included are the positions (and error circles) of the EGRET source 3EG J1800-2238 (2EG J1801-2312 in the 2nd EGRET catalogue), OH maser sites A to F, PSR J1801-23 (also the tracking position for 1994 data) and the radio point source, labelled A83 (Andrews et al. 1983), is the tracking position for 1995 data. See Sect. 4 for a discussion.

We report here on the comprehensive analysis of data taken in 1994 and 1995 with the CANGAROO 3.8 metre telescope. This work follows analysis of data taken in 1992 (Kifune 1993) in which weak evidence for a gamma-ray signal was reported. At that time, only ON source data were collected, making an estimation of the background rate difficult. Mori (1995) reported briefly on an analysis of 1994 data centred on PSR J1801-23, in which a field of view was searched. Both ON and OFF source data were collected and no evidence for TeV -ray emission was seen from various point-like sources including the pulsar and both radio and X-ray maxima. The 1995 data were centred on the radio position A83, located away from PSR J1801-23. A search for point-like and diffuse sources of TeV emission was carried out on the 1994 and 1995 datasets out to from the tracking centre of each dataset, using an extended source analysis. We have used an improved set of cuts to those used in the analysis of data taken on the Vela Pulsar/Nebula (Yoshikoshi et al. 1997). These cuts were designed to minimise the loss of gamma-ray sensitivity for off-axis sources and in particular maintain reliable statistics over the search region.

© European Southern Observatory (ESO) 2000

Online publication: June 30, 2000
helpdesk.link@springer.de