1 Introduction

It is now widely accepted that the energy source of active galactic nuclei (AGN) originates in the accretion of material onto a central massive black hole. However, the fuelling mechanism of the central engine remains unclear. The problem is how to transfer mass from the galaxy into its very central regions (parsec scale). Several mechanisms can be invoked to induce the potential perturbation that could initiate gas inflow. Among them, m=2 or m=1 modes tiggered via gravitational perturbation by a companion, instabilities developing in the galactic disc or both. It has also been proposed that the so called "minor mergers'', in which a gas rich galaxy and a satellite galaxy are involved could play an important role in triggering activity in Seyfert galaxies (de Robertis et al. 1998 and reference therein).

An efficient way to drag significant amount of mass in the central regions would be the presence of a large scale bar in the host galaxy, which could initiate strong inflows of gas (e.g. Athanassoula 1992; Friedli & Benz 1993). However, it has been shown in recent studies that if stellar formation is marginally enhanced in barred galaxies, the presence of an AGN is not correlated with the existence of a bar in its host galaxy (Mulchaey & Regan 1997; Ho et al. 1997). In fact, if the bar indeed initiates gas inflow, its inner Lindblad resonance (ILR), when present, stops the inflow and the gas is redistributed in a disc inside the ILR radius (e.g. Buta & Combes 1996). Thus a bar is clearly an efficient way to drag gas in the central regions ($\sim$100 s pc), but another mechanism must take over to allow this gas to finally fall onto the AGN.

Shlosman et al. (1989) proposed that instabilities such as secondary bars could develop in the inner disc, starting again the gas inflow. During the past five years, a large number of bars within bars have been detected and it becomes now possible to check statistically the impact of these structures on activity of galaxies. Whether or not there is a higher fraction of secondary embedded bars observed among Seyfert galaxies is still a matter of debate (Wozniak et al. 1995; Friedli et al. 1996; Jungwiert et al. 1997; Mulchaey & Regan 1997; Regan & Mulchaey 1999; Greusard et al. 2000). However, the lack of a clear correlation could be due to the different timescales in the involved processes (fuelling, AGN phase, bar dissolution). Still, bars within bars are often associated with bursts of star formation, confined within the nuclear bar, or the nuclear ring encircling it. Anyway, they do lead some significant evolution in the morphology and dynamics of the central regions of their hosts.

It is therefore important to trace both the dynamics and the stellar population of embedded bars, and examine potential links with the central AGN. So far, embedded bars have been observed essentially by optical or NIR imaging, whereas lack of collecting power has prevented any breakthrough from kinematic studies. Furthermore, if optical spectroscopy can be used to measure kinematics in dust-free regions, one has to move to less obscured wavelengths to map galaxy centres which are almost always very dusty (e.g. Valentijn 1990). As shown by Gaffney et al. (1995), the ¹²CO absorption features at 2.29 $\mu$m, if not widely used, are a very efficient tool for measuring stellar kinematics in dusty environments. The age of the stellar populations can also be approached through the equivalent widths of these absorption ¹²CO features (Doyon et al. 1994). Together with high-resolution NIR photometry, the kinematics can provide mass-to-light ratios, that also constrain the age of the populations.

We have thus undertaken a NIR spectroscopic study of a sample of Seyfert galaxies with and without double bars, using the spectrograph ISAAC mounted on the VLT/ANTU. The aim of this work, the DEBCA (Dynamics of Embedded Bars and the Connection with AGN) project, is to characterize the kinematics of stars and gas in the few 100 s inner parsecs, and to constrain the age of the stellar populations. We have so far obtained long-slit ISAAC spectroscopy of four Seyfert galaxies with double bars. In the present paper, the first of a series, we present the stellar kinematics extracted from these data, and discuss it in the light of simple dynamical models. A detailed discussion regarding the stellar populations is reported to a forthcoming paper (Greusard et al. 2001, in preparation, hereafter Paper II).