In this project, we wish to take advantage of a fundamental figure-of-merit of the CMOS transistors, the doping modulation, to propose new functionalities arising from the control of a single charge and spin on individual dopants in silicon. The ultimate electrical switch is an atomic point contact. It has been realized and operated several times in laboratories at low temperature under the form of Quantum Point Contacts (QPC), metallic break junctions, molecules placed in an air gap. However a silicon atomic switch has not been realized yet. The devices will be manufactured within a mature technology on state-of-the-art CMOS platforms. Contrarily to bottom-up approaches, there is an unavoidable dispersion in the average number and location of dopants, using masking and implantation CMOS techniques. Nevertheless several approaches are now addressing this question for top-down devices, and this option will be considered. Therefore, we will study single atomic devices, either real (i.e. dopant) or artificial (i.e. quantum dots), with a manageable dispersion by considering three generic cases: devices without dopant, devices with a targeted concentration of one dopant and devices with many dopants. Devices without dopant will be based on ultimate silicon SET. A targeted size of 10 nm is realistic, allowing operation at low temperature (but much above 4.2K). These devices are fully controlled and scalable. Devices with one dopant or two dopants will be identified and selected from their electrical characteristics and then studied thoroughly. Relatively high operating temperatures up to room temperature are expected using donors with large ionization energy. These devices are the smallest possible switches using the silicon technology. Because our ultimate SET present the decisive advantage of an immediate integration in the CMOS, the AFSID project will prove the validity of hybrid SET-CMOS approach by building a SET-FET hybrid device on chip.