Using modern epitaxy methods, it is now possible to control the growth of semiconductor crystals with atomic-layer precision. In particular it is possible to incorporate clean "2-dimensional" planes and arrays of magnetic material into otherwise nonmagnetic semiconductor lattices. In this way the physics of (possibly interacting!) low-dimensional magnetic layers can be investigated.

All of the structures are made by our long-time collaborator Professor Nitin Samarth in the Department of Physics at Pennsylvania State University. Much of our work with Professor Samarth involves the study of the dynamical interplay between photoexcited electronic carriers (electron, holes, and excitons) and local magnetic moments (Mn, for instance) that are embedded in a semiconductor quantum structure. The carriers and local magnetic moments couple to one another via their respective magnetic spins, and the resultant spin-scattering occurs on femtosecond and picosecond timescales which require ultrafast laser techniques to resolve. Two such techniques are time-resolved Faraday rotation, which is sensitive to the net magnetization (spin) of the system under study, and time-resolved upconversion of photoluminescence, which reveals the picosecond evolution of photoexcited spin-polarized carriers.

A particularly attractive host in which to study these spin dynamics is a newly-developed class of magnetic semiconductor called digital magnetic heterostructures. In these, a fixed amount of magnetic material can be 'digitally' incorporated in different configurations (within a nonmagnetic quantum well, say) to study the effects of interfaces and reduced magnetic clustering on the underlying spin-scattering processes. Recent experiments have demonstrated the ability to observe terahertz spin precession and ultrafast transfer of angular momenta in these digital structures. These phenomena now enable all-optical spin resonance of single monolayers of magnetic semiconductor.

Digital Magnetic Heterostructures

Digital magnetic heterostructures incorporate discrete monolayer and sub-monolayer planes of purely magnetic semiconductor (Manganese-Selenide in this case, shown below in green) into otherwise nonmagnetic ZnSe-ZnCdSe-ZnSe quantum wells. In this way one can directly control, using a fixed number of magnetic spins, such factors as the exciton wavefunction within the well (shown in red), the local in-plane spin-density (affecting cluster formation), and the overlap of the wavefunction with the magnetic planes.

Three such digital structures are shown above (the schematics are of bandgap energy versus growth direction). Each incorporates only 3 monolayers of magnetic material, but in systematically varied distributions. The effect of this increased distribution is to radically increase the Zeeman splitting (see below) and the measured spin-scattering, and to decrease the statistical occurrence of antiferromagnetically-locked Mn clusters.

In the presence of an applied magnetic field, the planes of Mn become magnetized. The magnetic barriers are thus lowered for spin-down electrons and holes (excitons), and are increased for spin-up electrons and holes. Alternatively, spin-up excitons "see" a different barrier height than their spin-down counterparts. The net effect is to energy-split the exciton spin-states, shown schematically below. In these magnetic heterostructures, this Zeeman splitting can be made hundreds of times larger than in nonmagnetic quantum structures.

Terahertz Spin Precession and Transfer of Angular Momentum

In the absence of an applied magnetic field, spin-up and spin-down states are degenerate, and their spin relaxation may be seen as mirror images of each other (black curves), with a relaxation time in this sample of ~20 picoseconds. Upon the application of a magnetic field, the results could not be more different: the energy degeneracy is lifted, and the coherently-excited spin polarizations exhibit quantum beating between the Zeeman-split levels at terahertz frequencies, with a frequency proportional to the electron g-factor and a rapid dephasing time.

Although the electronic dynamics disappear after ~20 picoseconds, more careful examination of the data shows that the system has not returned to equilibrium. A second signal is seen with a different period, corresponding to a g-factor of 2, and is a signature of the manganese ions. Angular momentum is coherently transferred from the electronic system to the magnetic spins: unlike traditional semiconductors, the electronic carriers have left a "footprint" in the magnetic sublattice which subsequently persists long after the carriers have disappeared from the structure. Although it has returned to electronic equilibrium, the quantum well is still undergoing dynamical spin processes.

This magnetic spin precession persists over far longer timescales - hundreds of picoseconds - eventually relaxing due to phonons in the system. These types of studies allow us to disentangle electronic and magnetic spin phenomena in quantum structures, enabling us to perform spin resonance experiments in systems with near atomic thicknesses.