1 Introduction

High redshift radio galaxies play an important role in the study of the early universe: thanks to their extreme luminosity at different wavelengths it has been possible to use them as cosmological probes already for several decades. Currently there are more than 150 radio galaxies known with redshift greater than 2, and recently a powerful radio source at a redshift of 5.19 has been discovered by van Breugel et al. (1999), becoming the most distant known AGN.

High redshift radio galaxies (HzRGs) comprise a different population to high redshift radio-quiet galaxies, e.g. Ly-dropouts: there is evidence that they are older and more massive, and will evolve into brightest cluster galaxies rather than L_* ellipticals (Best et al. 1997; van Breugel et al. 1998).

In the past few years a number of studies have concentrated on the properties of HzRG host galaxies, at visual and near infrared wavelengths (e.g. Pentericci et al. 1999; van Breugel et al. 1998; Best et al. 1997; Eales et al. 1997) with the main goal of studying the morphological evolution of these host galaxies, understanding the nature of the radio-optical alignment effect and discerning the various components (stellar light, scattered light, etc.) that contribute to the optical and infrared continuum emission.

However, one of the potentially most important results from recent studies on powerful radio galaxies came from radio observations and with the discovery that a significant fraction ($\sim 20\%$) of HzRGs have extremely large Faraday rotation, of the order of several thousands rad m^-2 (Carilli et al. 1997; Athreya et al. 1998), similar to low redshift powerful radio galaxies residing at the center of X-ray clusters with extreme cooling flows (Taylor et al. 1994). This makes HzRGs potential excellent targets for finding and studying high redshift (proto) clusters.

Therefore the main purpose of these new high resolution radio polarimetric imaging observations was to enlarge the number of known HzRGs with high intrinsic Faraday rotation, and to provide cluster targets for future observations with facilities such as the new X-ray telescopes, Chandra and XMM. High resolution radio imaging is important not only for finding high Faraday rotation radio galaxies, but also for a number of other important issues, such as the identification of the location of the active nucleus, the study of the correlation between the optical morphology and the radio jets, or the line emission gas and the radio jets, and the study of the evolution of radio size structure.

Throughout the paper we adopt a cosmology with H₀=50 km s^-1 Mpc^-1and q₀=0.5.