Introduction
Semiconductor quantum dots are materials systems in which the electron motion is confined in all spatial directions. This confinement gives rise to a discreet energy spectrum for the electrons similar to that of atoms; hence the term 'artificial atoms' is often used to refer to quantum dots, even if they actually consist of thousands, or even tens-of-thousands, of atoms. Quantum dots are undoubtedly one of the most studied systems in materials science: they can be used as models for fundamental studies of the quantum properties of electrons; they also have strong potential for diverse applications, ranging from electronic devices to biomedical imaging. There are several methods of quantum-dot fabrication, of which wet chemistry is probably the cheapest and therefore the most interesting in terms of wide-scale application1, 2, 3. Because of their typical size (a few nanometres in diameter), quantum dots obtained in this way are typically referred to as 'nanocrystals'.
A common approach to strengthen confinement in a quantum dot, as well as improve its optical properties, is to enclose the nanocrystal in a layer of a different material (usually with a wider bandgap), effectively creating a core–shell structure. On page 35 of this issue, Bussian and colleagues report on such a core–shell structure in which the shell is used to stretch the electronic wavefunction rather than squeeze it even more4. The quantum dots they examined consist of ZnSe cores doped with Mn and enclosed in a CdSe shell. The addition of magnetic atoms actually transforms the quantum dots in the nanoscale analogues of bulk-diluted magnetic semiconductors (DMSs) — the workhorse materials of emerging spintronics. Each Mn atom carries a large magnetic moment, S, and through an exchange interaction with the electron spin it influences the energy spectrum of the electrons5. In particular, it can induce the so-called 'giant spin splitting' of the energy levels, which corresponds under favourable conditions to a g-factor of several hundred.
Bussian et al. used a CdSe shell with a narrower gap than the ZnSe core. The effect of this unusual design is that although the electrons in the conduction band are still confined in the core, those in the valence band — and therefore any optically excited hole — have a wavefunction peaked in the shell and in proximity to the core–shell interface (Fig. 1). The degree to which the valence electrons are pulled out depends on the thickness of the outer shell and may be precisely controlled during preparation of the nanocrystals. On the other hand, introducing Mn ions into nanocrystals occurs only during the early stages of growth, so they tend to be buried in ZnSe. The fact that the valence electrons reside in the outer shell, rather than in the core, means that they interact more weakly6 with Mn through the exchange interaction than they would had the overlap of their wave function been greater (for example, such as in bulk DMS). Incidentally, the weaker confinement also pushes the quantum dot photoluminescence below that ascribed to internal Mn optical transitions, which avoids part of the electron–hole pairs recombining through the states around the dopants — an approach also used in self-assembled quantum dots of CdMnTe (ref. 7).
Figure 1: Core/Shell Mn:ZnSe/CdSe quantum dots.
The graphs on the right show that although the conduction electron wavefunction (
e) has its maximum in the centre of the core, that of the holes (h) has peaks in the CdSe shell.
Modifying the overlap between wavefunction and magnetic atoms can have greater effects than simply altering the strength of the exchange interaction, as shown by Merkulov and colleagues with quantum wells8. The reduced symmetry can activate the so-called 'kinetic component' of the exchange interaction, which in turn can lead to the sign of the spin splitting being reversed. This sign is closely linked to the polarization properties of the light emitted by the quantum dots. In other words, we could say that by simply controlling the thickness of the outer layer, Bussian et al. are able to control the strength, and even the sign, of the exchange interaction that is at the heart of spintronics.
Although this study has been performed on an ensemble of nanocrystals, the experiments are likely to be repeated soon on single nanocrystals, providing even more convincing evidence of the observed tunable exchange interaction. Another approach may be to further dilute the content of Mn in nanocrystal cores and obtain a single dopant atom. Two magnetic ions in a nanocrystal is probably the least favourable situation from the perspective of giant spin-splitting: the interaction of the localized Mn spins in ZnSe happens to be antiferromagnetic, and so the total magnetic moment of the pair vanishes. Dealing with a single magnetic ion per nanocrystal would be much more desirable from that perspective and should be feasible, as suggested by studies of epitaxial, self-assembled quantum dots9, 10.
A final consideration is that decreasing the overlap between electrons and holes also reduces the probability of radiative recombination. On the downside, this makes the nanocrystals a less efficient light source. However, they are more useful for logic and memory applications.
The importance of the study by Bussian et al. is it demonstrates that intrinsic quantum properties of DMS nanocrystals can be controlled by as simple a parameter as overlayer thickness. This is comparable to the bandgap engineering revolution made possible by adjusting the width of the quantum wells in optoelectronic devices. Before realistic applications can be considered, there are several obvious issues that need to be solved — for example, the fact that the operation of such systems only occurs at low temperature.
