An atom diode, i.e., a device that lets the ground-state atom pass in one direction but not in the opposite direction in a velocity range, is devised. It is based on the adiabatic transfer achieved with two lasers and a third laser potential that reflects the ground state.The detailed control of internal and/or translational atomic states is a major goal of quantum optics. Optical elements in which the roles of light and matter are reversed such as mirrors, gratings, interferometers, or beam splitters made of laser light or magnetic fields, allow to manipulate atomic waves. Atom chips [1,2] and atom-optic circuits [3] have been also realized recently. The aim of this paper is to propose simple models for an "atom diode," a laser device that lets the neutral atom in its ground state pass in one direction but not in the opposite direction for a range of incident velocities. A diode is a very basic control element in a circuit and many applications are possible for atomic trapping or quantum information processing.More specifically, our goal is to model an atom-field interaction so that the ground-state atom is transmitted when traveling, say, from left to right, and it is reflected if coming from the right. We shall describe effective three-level and two-level atom models, for simplicity in one dimension, to achieve the desired behavior. The one-dimensional description is accurate if the atom travels in waveguides formed by optical fields [3], or by electric or magnetic interactions due to charged or current-carrying structures [2]. It can be also a good approximation in free space for atomic packets which are broad in the laser direction, perpendicular to the incident atomic direction, as demonstrated for time-of-arrival measurements by fluorescence [4].In our models the atom is in an excited state after being transmitted and, in principle, excited atoms could cross the diode "backwards," i.e., from right to left. Nevertheless, an irreversible decay from the excited state to the ground state, will effectively block any backward motion.Let us denote by R ␣ l ͑v͒ ͓R ␣ r ͑v͔͒ the scattering amplitudes for incidence with velocity v Ͼ 0 from the left (right) in channel ␣ and reflection in channel . Similarly, we denote by T ␣ l ͑v͒ ͓T ␣ r ͑v͔͒ the scattering amplitude for incidence in channel ␣ with velocity v Ͼ 0 from the left (right) and transmission in channel  to the right (left). The potential will be such that ͉T 31 l ͑v͉͒ 2 Ϸ 1, ͉R 11 l ͑v͉͒ 2 Ϸ 0 and ͉T 31 r ͑v͉͒ 2 Ϸ 0, ͉R 11 r ͑v͉͒ 2 Ϸ 1. The basic idea is to combine two lasers that achieve stimulated Raman adiabatic passage (STIRAP) with a state-selective reflecting interaction for the ground state. The STIRAP method is well known [5] and consists of an adiabatic transfer of population between levels 1 and 3 by two partially overlapping (in time or space) laser beams (see Fig. 1). The pump laser couples the atomic levels 1 and 2 with Rabi frequency ⍀ P , and the Stokes laser couples the states 2 and 3 with Rabi frequency ⍀ S . We assume here that these ...