When two ferromagnetic layers are separated by a nonmagnetic transition-metal spacer layer, their magnetizations couple by an indirect (RKKY-type) exchange interaction that oscillates between ferromagnetic and antiferromagnetic as the spacer thickness is varied. Parkin's 1991 systematic study through essentially all 3d, 4d, and 5d transition metals established the empirical rules that still frame the field: an oscillation period of roughly 10 Å in nearly every spacer in which coupling is observed (Cr being the striking exception with a much longer period), and a coupling strength that increases systematically from 5d to 4d to 3d metals and approximately exponentially with the number of d electrons along each period. Parkin's report left open the microscopic origin of these systematics and the lack of any theory connecting realistic Fermi surfaces, interface electronic structure, and disorder to the measured periods, phases, and magnitudes. Later literature has largely discharged the period part of the question: the oscillation periods, including Cr's distinct long period, are now understood to be set by the extremal (stationary) spanning vectors of the spacer Fermi surface, within the quantum-interference picture of spin-dependent electron confinement. The surviving open core is quantitative: no accepted microscopic theory reproduces, or explains, the measured coupling magnitudes and phases across the transition-metal series — in particular the systematic 5d→4d→3d strength increase and its approximate exponential dependence on d-electron count — for the real, rough, intermixed multilayers in which the systematics were measured (the model theories soon proposed for the strength systematics, a confinement-based theory by Mathon, Villeret and Edwards and an electron-correlation model by Shi, Ding and Lin, did not become quantitative, parameter-free accounts); and it remains unsettled whether spacers in which no antiferromagnetic coupling was seen lack intrinsic coupling or merely fall below the detection limits of the growth and measurement conditions used.

Background

Magnetic multilayers — alternating ferromagnetic and nonmagnetic metallic films a few atoms thick — became a central platform of condensed-matter physics after Grünberg and co-workers (1986) observed antiferromagnetic coupling of Fe layers across Cr interlayers, and Parkin, More and Roche (1990) found that the coupling and the giant magnetoresistance it produces (giant magnetoresistance: the large change in electrical resistance between parallel and antiparallel alignment of the layer magnetizations) oscillate with spacer thickness in sputtered Co/Ru, Co/Cr and Fe/Cr superlattices — periodically repeated stacks of these bilayers, grown by sputtering, i.e. deposition of atoms ejected from a solid target, a process that yields dense but polycrystalline, atomically rough films. The coupling in question is the interlayer exchange coupling (IEC): the energy per unit area that favors parallel (ferromagnetic) or antiparallel (antiferromagnetic) alignment of the magnetizations of two ferromagnetic films separated by a nonmagnetic 'spacer' layer. Its microscopic origin is an indirect, Ruderman–Kittel–Kasuya–Yosida (RKKY)-type exchange, in which conduction electrons mediate an oscillatory interaction; in the original free-electron RKKY picture the oscillation period is fixed by the Fermi wavevector, which for a free-electron gas of typical metallic density would give a period of only a few ångströms and no reason for universality across different materials.

In 1991 Parkin reported measurements on sputtered multilayers in which Fe, Co, Ni or Ni-alloy ferromagnetic layers sandwich spacers drawn from nearly the whole 3d, 4d, and 5d transition series. Three empirical rules emerged, quoted here from the published abstract: the oscillatory indirect coupling is a general phenomenon; the oscillation period is approximately the same, about 10 Å, in all the transition metals in which coupling was observed, with the single exception of Cr, for which it is much longer; and the coupling strength increases systematically from the 5d to the 4d to the 3d metals, and exponentially with the number of d electrons along each period. These systematics were surprising: simple RKKY or free-electron models give spacer-specific periods tied to a single Fermi wavevector, and none of the obvious electronic-structure metrics (density of states, occupied bandwidth, susceptibility) tracked the strength trend. Parkin's report accordingly left the microscopic origin of the periods, of the strength systematics, of the strong sensitivity of the oscillation phase to the spacer and magnetic-layer identity and to interfacial conditions, and of the absence of antiferromagnetic coupling in several spacers (among them Ti, Zr, Hf, Pd, Pt, Ag, Au) as open issues, and noted the lack of a theory linking realistic transition-metal Fermi surfaces, interface electronic structure, and disorder to the measured periods, phases and magnitudes.

The theory that subsequently developed rests on a quantum-interference picture: electrons in the spacer are partially reflected at the ferromagnet/spacer interfaces with spin-dependent amplitudes, and the resulting confinement (quantum-well) interference modulates the density of states, which oscillates the total energy as the spacer thickness dd changes. Bruno and Chappert (1991) and Mathon (1991) generalized the RKKY result to arbitrary Fermi surfaces: the coupling is a sum of oscillatory components, one for each stationary (extremal) spanning vector of the spacer Fermi surface — a vector q\mathbf{q} that connects two points on the Fermi surface and is parallel to the layer normal — with each component oscillating as sin⁡(q⊥d+ϕ)/d2\sin(q_\perp d+\phi)/d^2, where q⊥q_\perp is the component of the spanning vector along the normal and ϕ\phi a phase set by the interface reflection phases. The periods therefore depend only on the spacer material and growth orientation, while the amplitudes and phases depend on the spin-dependent reflection probabilities at the ferromagnet/spacer interfaces and are thus sensitive to interface structure. Ab initio (from-first-principles, spin-polarized density-functional-theory) formulations — the magnetic 'force theorem' (which replaces expensive total-energy differences between magnetic configurations by sums of single-particle eigenvalue shifts), the Lloyd formula (a determinant identity that yields the coupling directly from the spacer's electronic Green function without summing eigenvalues), and the coherent-potential approximation (a mean-field treatment of substitutional disorder used to describe intermixed and alloyed interfaces) — were developed to compute these quantities without adjustable parameters; a complementary phenomenological 'interface-reflection model' fits measured coupling to extract reflection-related quantities. Interface roughness and intermixing are known to suppress and renormalize the amplitudes of the oscillatory components, so that in sputtered (polycrystalline, rough) multilayers only the longest-period components survive measurably, whereas epitaxial samples (grown as single crystals with a fixed crystallographic growth orientation) resolve several simultaneous periods per spacer.

Problem Statement

The problem, as left open by the systematic 3d/4d/5d measurements of Parkin (1991) and only partially answered since, is to convert the empirical systematics of the interlayer exchange coupling into predictive microscopic understanding, with one common objective: a quantitative microscopic theory of the coupling between ferromagnetic layers (Fe, Co, Ni and their alloys) across nonmagnetic transition-metal spacers that links the spacer Fermi surface, the electronic structure of the ferromagnet/spacer interfaces, and interfacial disorder and roughness to the measured periods, phases and magnitudes of the coupling. Concretely, the question asks: (i) the microscopic origin of the oscillation periods observed in sputtered multilayers — why a single period of roughly 10 Å dominated in nearly every spacer studied while Cr alone showed a much longer period (this part is now largely understood in terms of the extremal spanning vectors of the spacer Fermi surface, with interface roughness suppressing the additional, shorter-period components, and Cr's long period traced to the ellipsoidal sheets of its Fermi surface centered at the N point of its body-centered-cubic Brillouin zone); (ii) the physical mechanism behind the systematic increase of the coupling strength from the 5d to the 4d to the 3d metals and its approximately exponential dependence on the number of d electrons along each period, which no density-of-states, bandwidth or susceptibility metric tracks and for which no accepted quantitative microscopic account exists (the early model theories directed at it did not establish one); (iii) the strong sensitivity of the oscillation phase to the spacer and magnetic-layer identity and to interfacial conditions — intermixing, interface-specific interactions and local interfacial moments — and the role these play in masking short-period oscillations; (iv) whether, for the spacers in which no antiferromagnetic coupling was observed (such as Ti, Zr, Hf, Pd, Pt, Ag, Au) or only a single weak antiferromagnetic region was measured (such as Nb, Ta, W), the intrinsic indirect coupling is genuinely absent or simply too weak or short-ranged to be detected; and (v) the assembly of these elements into a parameter-free theory that reproduces and predicts the measured periods, phases and, above all, magnitudes of the coupling for the real, imperfect multilayers in which the systematics were established. The completion boundary is a parameter-free theory whose quantitative predictions for coupling strength and phase reproduce the measured magnitudes and phases across the 3d, 4d and 5d transition-metal series — including the systematic 5d→4d→3d strength increase and its approximate exponential dependence on d-electron count — for the real, rough, intermixed multilayers in which the systematics were measured, so that its predictions can be confronted with controlled measurements; of the enumerated parts, (i) is settled context to be incorporated rather than re-derived, (ii) and (iii) are discharged by a successful (v) rather than required as independent deliverables, and (iv) requires a stated position rather than a separate determination.

References

  1. S. S. P. Parkin, 'Systematic variation of the strength and oscillation period of indirect magnetic exchange coupling through the 3d, 4d, and 5d transition metals', Phys. Rev. Lett. 67, 3598 (1991), DOI 10.1103/PhysRevLett.67.3598, https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.67.3598
  2. P. Grünberg, R. Schreiber, Y. Pang, M. B. Brodsky, and H. Sowers, 'Layered Magnetic Structures: Evidence for Antiferromagnetic Coupling of Fe Layers across Cr Interlayers', Phys. Rev. Lett. 57, 2442 (1986), DOI 10.1103/PhysRevLett.57.2442, https://doi.org/10.1103/PhysRevLett.57.2442
  3. S. S. P. Parkin, N. More, and K. P. Roche, 'Oscillations in exchange coupling and magnetoresistance in metallic superlattice structures: Co/Ru, Co/Cr, and Fe/Cr', Phys. Rev. Lett. 64, 2304 (1990), DOI 10.1103/PhysRevLett.64.2304, https://doi.org/10.1103/PhysRevLett.64.2304
  4. P. Bruno and C. Chappert, 'Oscillatory coupling between ferromagnetic layers separated by a nonmagnetic metal spacer', Phys. Rev. Lett. 67, 1602 (1991), DOI 10.1103/PhysRevLett.67.1602, https://doi.org/10.1103/PhysRevLett.67.1602
  5. J. Mathon, 'Theory of Oscillatory Exchange Between Magnetic Layers Through a Nonmagnetic Transition–Metal Spacer', MRS Proceedings 231, 179 (1991), DOI 10.1557/proc-231-179, https://doi.org/10.1557/proc-231-179
  6. J. Mathon, M. Villeret, and D. M. Edwards, 'Theory of the systematic variation of the exchange coupling strength in magnetic multilayers', J. Magn. Magn. Mater. 127, L261 (1993), DOI 10.1016/0304-8853(93)90040-9, https://doi.org/10.1016/0304-8853(93)90040-9
  7. Fa-jian Shi, Meng Ding, and Tsung-han Lin, 'Systematic variation of exchange coupling strength by electron correlation in magnetic multilayers', Solid State Commun. 97, 225 (1996), DOI 10.1016/0038-1098(95)00651-6, https://doi.org/10.1016/0038-1098(95)00651-6
  8. J. Mathon, M. A. Villeret, J. M. Mander, D. M. Edwards, and R. B. Muniz, 'Effect of Hybridization on Exchange Coupling in Magnetic Multilayers', MRS Proceedings 313, 171 (1993), DOI 10.1557/proc-313-171, https://doi.org/10.1557/proc-313-171
  9. M. D. Stiles, 'Oscillatory exchange coupling in Fe/Cr multilayers', Phys. Rev. B 54, 14679 (1996), DOI 10.1103/PhysRevB.54.14679, https://doi.org/10.1103/PhysRevB.54.14679
  10. Mark van Schilfgaarde, Frank Herman, Stuart S. P. Parkin, and Josef Kudrnovský, 'Theory of Oscillatory Exchange Coupling in Fe/(V,Cr) and Fe/(Cr,Mn)', Phys. Rev. Lett. 74, 4063 (1995), DOI 10.1103/PhysRevLett.74.4063, https://doi.org/10.1103/PhysRevLett.74.4063
  11. P. Bruno, 'Theory of interlayer exchange coupling', lecture notes (30th Ferienschule, Forschungszentrum Jülich, March 1999), arXiv:cond-mat/9905022, https://arxiv.org/abs/cond-mat/9905022
  12. J. Kudrnovský, V. Drchal, I. Turek, P. Bruno, P. Dederichs, and P. Weinberger, 'Ab initio theory of the interlayer exchange coupling', Lecture Notes in Physics ('Electronic Structure and Physical Properties of Solids'), pp. 313–346, DOI 10.1007/3-540-46437-9_9, https://doi.org/10.1007/3-540-46437-9_9; preprint arXiv:cond-mat/9811152 (1998), https://arxiv.org/abs/cond-mat/9811152
  13. M. D. Stiles, 'Interlayer Exchange Coupling', chapter in Ultrathin Magnetic Structures III, pp. 99–142, Springer, DOI 10.1007/3-540-27163-5_4, https://link.springer.com/chapter/10.1007/3-540-27163-5_4
  14. Tommy McKinnon, Bret Heinrich, and Erol Girt, 'Spacer layer thickness and temperature dependence of interlayer exchange coupling in Co/Ru/Co trilayer structures', Phys. Rev. B 104, 024422 (2021), DOI 10.1103/PhysRevB.104.024422, https://doi.org/10.1103/PhysRevB.104.024422
  15. S. Blizak, G. Bihlmayer, S. Blügel, and S. E. H. Abaidia, 'Interlayer exchange coupling between FeCo and Co ultrathin films through Rh(001) spacers', Phys. Rev. B 91, 014408 (2015), DOI 10.1103/PhysRevB.91.014408, https://doi.org/10.1103/PhysRevB.91.014408