: Topics : People

Semiconductors and magnetism

Diluted magnetic semiconductors | Hybrid structures

Incorporating magnetic properties into semiconductor heterostructures would allow one to manipulate not only the charge of the carriers (electrons in the conduction band, holes in the valence band) as in usual microelectronics, but also their spin. This is of interest in order to achieve semiconductor spintronics; it could even be a way towards quantum computing... But keep your old computer for a while, first we have to understand how this could be realized in a sample... In our lab, we try to do that:

in diluted magnetic semiconductors, i.e., usual semiconductors in which we add magnetic impurities (Mn) and free carriers to induce a ferromagnetic interaction between the Mn spins: hence ferromagnetism in inside;
in hybrid structures, i.e., structures with a ferromagnetic metal is grown on top of the semiconductor.

In both cases, samples are grown by molecular beam epitaxy, within the Joint CEA-CNRS-UJF Group "Nanophysique et semi-conducteurs".

1. Diluted magnetic semiconductors

Fig. 1: carrier induced ferromagnetic interaction between localized Mn spins

In II-VI semiconductors (such as CdTe or ZnTe), the Mn2+ ion forms an isoelectronic center: it substitutes the metal ion (Cd2+ or Zn2+) without being electrically active (no free electron in the conduction band or hole in the valence band). However, its d shell is half filled, so that it introduces a localized 5/2 spin into the semiconductor. It has been known for many years that the carriers in these "diluted magnetic semiconductors" are strongly coupled to the localized Mn spins, which gives rise to huge magneto-optic phenomena, such as a giant Zeeman effect as great as two orders of magnitude larger than in a normal semiconductor. If, in addition, we add free carriers to the diluted magnetic semiconductor, we can induce a ferromagnetic interaction between the Mn spins (fig. 1). This can be done by introducing holes into the valence band of in a thick ZnMnTe layer, or in a modulation-doped CdMnTe quantum well (fig. 2); in the first case, we have a 3D hole gas, in the second case it is 2D. While these materials are paramagnetic (and even weakly antiferromagnetic...) in the absence of carriers, a ferromagnetic behavior is observed on the doped samples either by direct measurement of the magnetization (fig. 3), or on the electronic properties (electrical transport, fig. 4 or spectroscopy). This is a proof that the semiconductor itself has been made ferromagnetic! We can even apply all the techniques which have made the success of the microelectronics and optoelectronics industry, to change the magnetic properties of our system. By inserting the CdMnTe quantum well in a p-i-n diode, we make a field effect transistor; applying a small voltage (fig. 5) drives the carriers out of the quantum well and back, making the system paramagnetic (without the carriers) or ferromagnetic (with the carriers). Similar effects are obtained when illuminating the sample with well chosen light. All that was obtained at low temperature in tellurides. However, if we extrapolate the simple model which explains this behavior to large gap semiconductors, higher critical temperatures are predicted (fig 6). These large gap semiconductors, like GaN, are widely studied for various applications in electronics. Making them ferromagnetic is thus a new challenge, and we try to do that in the frame of a European Program called FENIKS (http://www.feniks-project.org/).

Fig. 2: 2D hole gas in a modulation-doped CdMnTe quantum well; the 2D hole gas induces a ferromagnetic interaction between Mn spins.

Fig. 3: (left) remanent magnetization and susceptibility (inversed) of a doped ZnMnTe layer, as a function of temperature; (right) magnetization cycles at different temperatures on both sides of the ferromagnetic transition.

Fig. 4: Magnetic hysteresis measured on the electric transport (Hall effect) in a ZnMnTe layer strongly doped p-type.

Fig. 5: Electrostatic control of the magnetic properties of a CdMnTe quantum well; by applying a 1V voltage across the p-i-n diode, carriers are pushed out of the quantum well and back, making the quantum well paramagnetic or ferromagnetic; the spontaneous magnetization is measured by optical spectroscopy (right).

Fig. 6: Critical temperatures predicted by the mean field model of carrier indeuded ferromagnetism in various semiconductors with a concentration x=0.05 of spins 5/2, and 3×1020 holes / cm².

2. Hybrid structures

We can also make hybrid structures, in which a ferromagnetic metal is epitaxially grown on top of the normal semiconductor (this is a collaboration with the "Nanostructures magnétiques" group of CEA-Grenoble, and with Laboratoire de Minéralogie Cristallographie in Paris). An attractive configuration is realized when an anisotropic metal with perpendicular magnetization is deposited on a semiconductor quantum well. The metal can be an ordered alloy (FePd, FePt) or a multilayer (Au-Co, Cu-Ni). It features very nice magnetic domains which can be imaged using a Magnetic Force Microscope (fig 7). The spectroscopic properties of the quantum well can help us to understand the effect of the inhomogeneous magnetic field which results from the magnetic domains in the metal (fig. 8), or how the polarized carriers are transferred from the ferromagnetic metal to the semiconductor (spin injection).

Fig. 7: Magnetic domains in a FePd layer epitaxially grown on ZnSe. Light and dark domains correspond to the two orientations of magnetization normal to the plane. Image size is 8µm×8µm.

Fig. 8: Magnetic field at different depths in a semiconductor due to the domains in a ferromagnetic layer with perpendicular magnetization.