Nanotechnology is impacting fundamental science by giving researchers a means to engineer model systems. With semiconductor processing and growth techniques that offer atomic precision, it is now possible to make and manipulate ‘ideal’ quantum systems.

Fig. 1: Topography of manganese phthalocyanine atoms on a lead surface.

Writing in Physical Review Letters, Qi-Kun Xue, Xi Chen and colleagues at Tsinghua University in China are utilizing the tools of nanotechnology to control electron-spin interactions in a silicon-supported device1. The team adsorbs magnetic molecules on ultra-thin ‘islands’ of lead grown on a silicon surface (Fig.1). They show that the thickness of the island – on the scale of a single atomic layer - determines the strength of the interaction between the molecular spin and the lead electrons.

In metals containing dilute magnetic impurities, such as cobalt, the conduction electrons are scattered by the localized spins. This leads to an increase in resistance of the metal below the so-called Kondo temperature, named after the physicist that first explained the effect in 1964.

The Kondo effect has since been reincarnated in devices in which a semiconductor quantum dot or a magnetic molecule acts as the magnetic impurity and metallic leads provide the conduction electrons. The Kondo physics in these systems can be manipulated by either adding an additional electron to the quantum dot or substituting a different type of molecule.

The Tsinghua University group is taking a different approach in their devices by instead controlling the properties of the conduction electrons. The lead islands are only ten to twenty atomic layers thick and the electrons that move in them experience the effects of quantum confinement, similar to the problem of a particle in a box. As a result, the density of electrons near the Fermi level of the lead islands—that is, those electrons that interact with the molecular spin—oscillates up and down with each additional monolayer of lead atoms. The strength of the Kondo interaction thus oscillates in the same way.

The group’s route to controlling the Kondo interaction may prove to be easier to extend to other spin-based devices than previous methods and could be an important step toward applications in spintronics.