PerspectivePhysics

Hitting the limit of magnetic anisotropy

See allHide authors and affiliations

Science  30 May 2014:
Vol. 344, Issue 6187, pp. 976-977
DOI: 10.1126/science.1254402

Single magnetic atoms adsorbed on surfaces, or so-called adatoms, provide a viable ground for realizing information storage and processing at ultimate length scales (13). Such concepts hinge on the magnetic anisotropy energy (MAE), which energetically favors a preferential spatial orientation of the adatom's magnetic moment m, where m is the sum of coupled spin and orbital momenta. Unlike for a free atom, where m can rotate freely in any direction without energy cost (see the figure, panel A, MAE = 0), large MAE (panel B, MAE ≠ 0) enables m to maintain a fixed orientation for a sufficient amount of time. For stable and robust magnetic memory storage, large values of MAE are desirable. On page 988 of this issue, Rau et al. (4) show that a suitable choice of substrate and adatom pairing can result in the further enhancement of MAE, thus providing a possible route toward realizing information storage at the atomic scale.

MAE originates from the so-called ligand field, which is the electrostatic energy generated by the shape of the orbitals of both the adatom and the neighboring substrate atoms, and which breaks the degeneracy of the atomic levels producing the magnetic moment. This energy landscape, combined with spin-orbit coupling (SOC), which locks the spin moment to the orbital moment, generates the resultant MAE. One crucial hurdle toward stabilizing a single adatom is to maximize the MAE, thus protecting m from thermal fluctuations. Consequently, the three important ingredients for large MAE are a large SOC energy, a large orbital moment, and a strong ligand field. The first condition can be maximized by a proper choice of adatom species. Unfortunately, the other two are usually inversely correlated to each other (i.e., a strong ligand field typically quenches the orbital moment).

Atomically controlled magnetism.

Orbital (blue arrow) and spin (purple arrow) moments, and the corresponding MAE for (A) a free Co atom, (B) a Co adatom on a Pt(111) surface, and (C) a Co adatom on an MgO surface. The zero-field splitting energy ΔE is a measure of the MAE. (D) Atomic spin detection techniques based on SP-STS and XAS.

In 2003, experiments using x-ray adsorption spectroscopy (XAS) and x-ray magnetic circular dichroism (XMCD) showed that Co adatoms on a platinum(111) metallic surface (see the figure, panel B) exhibited a MAE of 10 meV resulting from strong SOC and an incomplete quenching of the orbital moment (5). With the advancement of single-atom spin detection (see the figure, panel D) based on inelastic scanning tunneling spectroscopy (ISTS) (6, 7) and spin-polarized STS (SP-STS) (8, 9), it was revealed that even in the presence of large MAE, both single Fe and Co adatoms do not exhibit magnetic stability, even at extremely low temperatures. Although the insufficient MAE is one reason for this instability, strong interactions of m with the underlying metallic substrate electrons further destabilize m by spin-flip scattering. Therefore, questions remained whether the MAE could be further enhanced by a choice of substrate that supports a large adatom orbital moment, but at the same time provides a sufficiently strong ligand field, and that limits the coupling of m to the substrate.

Rau et al. show that these challenges can be solved by placing a Co adatom on a thin insulating MgO film (see the figure, panel C). Because of the particular symmetry of the Co adatom on top of an oxygen atom, there is a strong ligand field that keeps particular degeneracies of the Co orbitals. As a result, the orbital moment retains its free atom value and the MAE is pushed toward the SOC limit of the free Co atom of about 60 meV.

Probing the magnetic excitations of the Co adatom with ISTS (see the figure, panel D), Rau et al. show that the energy needed to excite the Co adatom is ∼58 meV. These observations are corroborated by measurements in a magnetic field that confirm the magnetic nature of the excitation. Because ISTS yields little information about the magnetization dynamics of the Co adatom, Rau et al. also apply SP-STS based on a pump-probe scheme (7) to probe the relaxation time of the excited Co adatom. They find that the time for the magnetic moment to relax to the ground state after excitation over the anisotropy barrier is 0.2 ms, by far the largest lifetime seen for a 3d transition metal atom on any surface. The insulating MgO film evidently decouples the Co moment from interacting with the underlying electron bath.

To gain deeper insight into the orbital configuration that drives the gigantic MAE, and to determine m of the Co adatom, Rau et al. used XAS and XMCD, which are extremely sensitive to the magnetic orientation and orbital configuration of the Co adatom. Such experiments were quantitatively compared to ligand field theory simulations and ab initio calculations, which can precisely determine the multiplet energies of the Co adatom and nicely reproduce the XAS spectra. These results corroborate the STS experiments illustrating that m strongly favors an out-of-plane orientation with a gigantic MAE. The work of Rau et al. presents the first application of all established atomic spin detection techniques, combined with insights from state-of-the-art calculations, to unravel the magnetic properties of an adatom.

Although the MAE reported by Rau et al. is record breaking, the observation that the Co adatom still fluctuates on a time scale of hundreds of microseconds certainly places in doubt its feasibility for information storage. Therefore, it is important to ascertain whether such fluctuations are weaker for other adatoms, like Fe, and whether such behavior can be preserved on metallic surfaces. One promising route would be to expand such studies to rare earth metal adatoms or to study whether atomic alloys formed between rare earth and 3d transition metal adatoms can lead to stability. Finally, the viability of single-atom information technology will hinge on the coupling of such single adatoms to arrays that form bits (2, 3) or logic elements (1), and on a minimally disturbing input and readout, leaving room for a bunch of further challenging investigations.

References

View Abstract

Navigate This Article