A neat study gives clear-cut evidence that when a wire made of a magnetic material such as iron is squashed to the atomic scale, the material's magnetism disappears via an exotic physical process.
Since the 1960s, it has been known that a magnetic impurity in a non-magnetic host metal is subject to a mechanism known as the Kondo effect. That is, below a characteristic temperature, the Kondo temperature, the metal's electrons screen the magnetic moment of the impurity1,2. In this context, ferromagnetic metals, such as those commonly used to make magnets for holding notes on refrigerator doors, are intriguing. These materials, which are marked out by their ability to retain magnetization, can in principle screen a magnetic impurity3. But can a ferromagnet screen one of its own atoms? On page 1150 of this issue, Calvo et al.4 show that this can be achieved when a ferromagnetic wire is reduced to atomic dimensions.
In their experiment, Calvo and colleagues created atomic-size ferromagnetic wires using two techniques known as scanning tunnelling microscopy and electromigration. These techniques allowed the authors to reduce the size of a junction between two metallic conductors made of pure ferromagnets — iron, cobalt or nickel — to the scale of a single atom, creating ferromagnetic, atomic-size contacts with full mechanical and thermal stability. Next, the authors measured how the conductance of the fabricated atomic contacts changed as a function of an applied voltage. For all three ferromagnets, the atomic contacts showed anomalies, in the form of peaks or dips, in their conductance at around zero volts. This is a sign of the Kondo effect, although not all zero-voltage conductance anomalies are indicative of it.
To confirm that the atomic contacts indeed showed Kondo behaviour, Calvo and co-workers studied the evolution of the conductance anomalies with temperature and found that it agreed with that expected for a system in the Kondo regime: the peaks and dips in the conductance curves disappeared with increasing temperature. What's more, they showed that the energy of the system, which is related to the temperature by the Boltzmann constant, followed a log-normal distribution — that is, the logarithm of the energy obeyed a Gaussian distribution. This observation is, again, in accord with the predictions of Kondo physics. Hence, the authors give clear proof that the Kondo effect is at work in these systems.
But their results are surprising. How can an atom that is a constituent member of a ferromagnet behave as a magnetic impurity screened by its identical neighbours? The Kondo effect usually requires two species of atom. Figure 1 shows a simple scheme for the screening process that takes place when a bulk ferromagnet is pulled into an atomic-size neck. In the bulk (Fig. 1a), all atoms contain electrons that have magnetic moments and are bound to the atomic sites; these 'd electrons', so-called because they reside in the d atomic orbitals, where their magnetic moments tend to align, give rise to the total magnetic moment of the ferromagnet. In much smaller quantities, there are also 'sp electrons', acting as if they are moving freely, which occupy the s and p orbitals in the isolated atom. These electrons interact weakly with the localized d electrons. Electronic conduction mainly takes place through sp electrons.
But as Calvo et al.4 demonstrate, when the ferromagnet is constricted (Fig. 1b), the interaction between d electrons is greatly reduced. Moreover, the sp electrons increase their interaction with the d electrons because of the smaller number of neighbouring atoms, efficiently screening the magnetic moment of the atomic contact. In other words, the magnetic moment of the atomic contact is effectively eliminated.
Calvo and colleagues' work shows that measurements of conductance can yield valuable information on the electronic properties of matter on the nanometre scale. Their results pave the way to studies of magnetism in atomic contacts and prompt further questions. Do non-magnetic atoms develop magnetism in non-magnetic materials as their dimensions are reduced? How does electronic transport take place? Does spin transport occur in addition to charge transport? We expect that, besides its fundamental implications for our understanding of how electrons interact in solid-state materials, this work will provide new insights into the properties of nanostructures and their use in fields such as magnetoelectronics, or indeed any technology operating on the scale where quantum mechanics rules.
Yosida, K. Theory of Magnetism 155–304 (Springer, 1996).
Fulde, P. Electron Correlations in Molecules and Solids 3rd edn 290–295 (Springer, 1995).
Pasupathy, A. N. et al. Science 306, 86–89 (2004).
Calvo, M. R. et al. Nature 458, 1150–1153 (2009).