In 1961, White and Van Vleck showed that the anomalous sign reversal of the samarium contribution to the 27Al Knight shift in the cubic Laves-phase compound SmAl2 follows from the second-order Zeeman (Van Vleck) paramagnetism of the Sm3+ 4f5 shell: because the Landé factor of the 6H5/2 ground multiplet is g = 2/7, the spin polarization that drives the conduction electrons vanishes at a crossover temperature T_CO = 5a/2b (free-ion estimate about 300 K, observed near 150 K), and, by a reciprocal relation, an exchange field alone induces no Sm moment at that same temperature — which they offered as an explanation of the near-zero difference between the magnetic moments of yttrium and samarium iron garnets measured by Pauthenet. They stressed that a quantitative account requires the cubic crystal field (which splits J = 5/2 into a doublet and a quartet and reduces T_CO) and that, at low temperatures in the garnet, their own attempts at the required theory had 'not yet met with quantitative success'. The literal experimental requests of 1961 have been answered (the SmAl2 crystal-field scheme was determined by 149Sm nuclear forward scattering in 2005; the garnet 'inertness' is qualitatively a small, sign-reversing Sm sublattice moment), but a precise open core survives, stated as unresolved in the 2005 and 2018 literature: no microscopic theory explains the anomalously small ordered moment of SmAl2 (about 0.2 Bohr magnetons per formula unit — mean-field calculations either overestimate it fourfold or predict no order at all) or the anomalous sixth-order crystal-field coefficient an order of magnitude above the RAl2-series trend, and the samarium garnet moment retains a roughly 50% theory–experiment discrepancy with no directly measured crystal-field parameters for Sm3+ in SmIG itself.

Background

The Knight shift and the samarium anomaly. In a metal, the resonance frequency of a nuclear magnetic resonance line is shifted (the Knight shift) chiefly because conduction electrons at the Fermi surface are spin-polarized and couple to the nucleus through the hyperfine interaction. In intermetallic compounds of rare earths with aluminum, the dominant contribution to the 27Al Knight shift — the shift of the nuclear resonance of the 27Al aluminum nuclei — is indirect: the localized 4f shell polarizes the conduction electrons through the exchange coupling between 4f spins and conduction-electron spins (the Zener–Ruderman–Kittel–Kasuya–Yosida, or s–f exchange, mechanism), so the sign of the rare-earth-induced shift reveals whether the conduction-electron spin polarization is parallel or antiparallel to the 4f spin. Measurements on the cubic Laves-phase compounds RAl2 (R denoting a rare-earth element; the Laves phase is a cubic AB2 crystal structure, designated C15) showed large, temperature-dependent shifts of either sign — and in SmAl2 the samarium contribution reverses sign between 77 K and room temperature.

White and Van Vleck traced this reversal to the peculiar electronic structure of the Sm3+ ion (4f5 configuration). Its ground multiplet 6H5/2 has total angular momentum J = 5/2 and an unusually small Landé factor g = 2/7, because the orbital (L = 5) and spin (S = 5/2) moments couple antiparallel; the free-ion moment is only gJμB=5μB/7≈0.71 μBgJ\mu_B = 5\mu_B/7 \approx 0.71\,\mu_B (μB\mu_B is the Bohr magneton). The first excited multiplet 6H7/2 lies only about 1500 K above, so a magnetic field also mixes excited multiplets into the ground state, producing a temperature-independent 'Van Vleck' (second-order Zeeman) susceptibility that is essential at all relevant temperatures. Writing the Sm moment in an applied field H0H_0 and an exchange field HexH_{\rm ex} (which acts on the spin only, not on the full moment) as

MSm=aNβ(H0−5Hex)T−1+bNβ(H0+2Hex),M_{\rm Sm} = aN\beta(H_0 - 5H_{\rm ex})T^{-1} + bN\beta(H_0 + 2H_{\rm ex}),

where β\beta denotes the Bohr magneton and NN the number of Sm ions, with a=5β/21ka = 5\beta/21k (the Curie, 1/T1/T, term), b=20β/7hν7/2,5/2b = 20\beta/7h\nu_{7/2,5/2} (the Van Vleck term, hν7/2,5/2/k≈1500h\nu_{7/2,5/2}/k \approx 1500 K being the intermultiplet spacing), and analogously for the spin polarization ⟨Sz⟩\langle S_z\rangle that polarizes the conduction electrons, they showed that the 1/T1/T and constant terms have opposite sign patterns because g=2/7g = 2/7 makes 2(g−1)/g=−52(g-1)/g = -5. Hence ⟨Sz⟩\langle S_z\rangle, and with it the 4f-induced Knight shift, vanishes at a crossover temperature TCO=5a/2b≈300T_{\rm CO} = 5a/2b \approx 300 K for the free ion; experiment places it near 150 K. A reciprocal relation follows: the temperature at which an applied field produces no spin polarization is also the temperature at which an exchange field alone produces no Sm moment. Because the Sm site in SmAl2 has cubic symmetry, the crystal field (the electrostatic field of the surrounding ions acting on the 4f shell) splits the J = 5/2 manifold into a Kramers doublet Γ7\Gamma_7 (a pair of levels whose twofold degeneracy is protected by time-reversal symmetry) and a quartet Γ8\Gamma_8, separated by hνch\nu_c; this multiplies the 1/T1/T terms by a factor f(T)f(T) that in general reduces TCOT_{\rm CO} — significantly only if hνc≳kTCOh\nu_c \gtrsim kT_{\rm CO} — so an accurate TCOT_{\rm CO} would measure the crystal-field splitting. The small intermultiplet spacing has a further consequence: the crystal field and the exchange both mix J = 5/2 with J = 7/2 ('J-mixing'), so J is not a good quantum number for Sm3+ — a hallmark of samarium compounds that any quantitative theory must respect.

The garnet puzzle. In ferrimagnetic iron garnets R3Fe5O12 the net moment comes mainly from the two Fe3+ sublattices (5 μB\mu_B per formula unit in Y3Fe5O12, yttrium iron garnet, YIG); the rare-earth ions occupy dodecahedral sites of low (orthorhombic) local symmetry and couple to the Fe sublattice by exchange. Pauthenet's magnetization measurements showed that the difference between the moments of YIG and samarium iron garnet Sm3Fe5O12 (SmIG) is practically zero from 0 K to the Néel point (the temperature at which the ferrimagnetic order of the iron sublattices sets in) — the samarium appears 'magnetically inert' despite the exchange field of the iron sublattice. The reciprocal relation explains inertness only near one temperature, not at all temperatures; White and Van Vleck stated that at low temperatures the crystal field should be included, that their attempts to obtain the requisite inertness had 'not yet met with quantitative success', and that more complete experimental data on the samarium garnets were 'highly desirable'.

What followed. The crossover physics became experimentally concrete in the zero-magnetization ferromagnet Sm1-xGdxAl2 — a ferromagnet whose 4f orbital moment, 4f spin moment, and conduction-electron polarization cancel exactly at a compensation temperature — and the samarium sublattice in SmIG was shown to carry a small moment that reverses sign near room temperature. What remains unresolved, as stated under Problem Statement, is the quantitative single-ion theory behind both phenomena: the anomalously small ordered moment of SmAl2 and its anomalous crystal-field parameter, and the residual theory–experiment gap in SmIG.

Problem Statement

The open core of the question posed by White and Van Vleck (1961) is the quantitative microscopic theory of the Sm3+ ion response — crystal-field splitting including intermultiplet J-mixing (the mixing of the J = 5/2 and J = 7/2 multiplets by the crystal field and the exchange), exchange coupling, and conduction-electron polarization — in the two hosts of the 1961 Letter. Concretely, a quantitative theory must account for, within a physically consistent framework:

(1) SmAl2 (cubic Laves phase, ferromagnetic below TC≈120T_C \approx 120 K): reproduce (a) the 27Al Knight-shift crossover, including its detailed low-temperature behavior that deviates from the free-ion model, and (b) the anomalously small spontaneous ordered moment of about 0.2 μB\mu_B per formula unit at low temperature — the mean-field single-ion treatments available today either overestimate it by a factor of four (calculating 0.73 μB\mu_B while requiring an anomalously large molecular-field constant of order 950 T/μB\mu_B, i.e. an exchange field per unit moment far above the values found along the RAl2 series) or, in the full 1001-state ∣LS,Lz,Sz⟩|LS,L_z,S_z\rangle basis of the 4f5 configuration, predict no magnetic order at all — and account for the anomalous sixth-order cubic crystal-field coefficient A6≈−9.9 K a0−6A_6 \approx -9.9\,{\rm K}\,a_0^{-6} (a0a_0 being the Bohr radius) refined from 149Sm nuclear forward scattering (a synchrotron-based Mössbauer technique), which is about an order of magnitude larger in absolute value than the value expected from the trend across the RAl2 series of rare-earth dialuminides and 'has not yet received any clear explanation'.

(2) SmIG (samarium iron garnet): close the quantitative gap in the single-ion description of the samarium sublattice — the calculated 0 K Sm3+ moment of 0.21 μB\mu_B versus 0.14 μB\mu_B measured, the temperature-dependent sign reversal of the Sm moment, and the sequence of easy-axis transitions on cooling (the succession of preferred magnetization directions); crystal-field parameters transferred from samarium-doped gallium garnets reproduce the wrong phase sequence.

The two hosts pose the same single-ion physics — a 6H5/2 manifold with strong J-mixing, split by crystal fields of order 100 K, driven by exchange of order 10–30 K — and the common objective, inherited from the 1961 Letter's call for a quantitative assessment of the crystalline field theory of the samarium ion, is a validated account of that physics.

A complete resolution accordingly requires one physically consistent framework that delivers every quantity enumerated above in both hosts. The scalar measured quantities — the SmAl2 ordered moment, the SmIG 0 K moment, the SmAl2 Knight-shift crossover temperature, and the SmIG sign-reversal temperature — must each be reproduced to within 10% of the corresponding measured value, whereas the low-temperature deviation of the Knight-shift curve from the free-ion model and the sequence of SmIG easy-axis transitions on cooling must each match qualitatively — the correct shape of the deviation and the correct succession of preferred magnetization directions, rather than a pointwise numerical fit. The anomalous sixth-order crystal-field coefficient A6A_6, taken at its value refined from 149Sm nuclear forward scattering, requires separately an explanation of why it lies about an order of magnitude above the RAl2-series trend. A result confined to a single host, or to a subset of these observables, does not by itself resolve the task.

References

  1. J. A. White and J. H. Van Vleck, 'Sign of Knight Shift in Samarium Intermetallic Compounds', Physical Review Letters 6, 412-413 (1961). DOI: 10.1103/PhysRevLett.6.412, https://doi.org/10.1103/PhysRevLett.6.412
  2. R. Pauthenet, 'Les propriétés magnétiques des ferrites d'yttrium et de terres rares de formule 5Fe2O3.3M2O3', Annales de Physique 13, 424-462 (1958). DOI: 10.1051/anphys/195811030424, https://doi.org/10.1051/anphys/195811030424
  3. K. H. J. Buschow, A. M. van Diepen, and H. W. de Wijn, 'Anomalous behaviour of the Knight shift in SmAl2', Physics Letters A 24, 536-537 (1967). DOI: 10.1016/0375-9601(67)90824-9, https://doi.org/10.1016/0375-9601(67)90824-9
  4. J. A. White, 'Theory of the magnetic torque anisotropy of the samarium iron garnet', Proceedings of the Physical Society 90, 1095-1109 (1967). DOI: 10.1088/0370-1328/90/4/321, https://doi.org/10.1088/0370-1328/90/4/321
  5. J. J. van Loef, 'On the sign reversal of the samarium sublattice magnetization in SmIG', Solid State Communications 6, 541-543 (1968). DOI: 10.1016/0038-1098(68)90507-3, https://doi.org/10.1016/0038-1098(68)90507-3
  6. A. M. Stewart, 'The Ordered Moment of Metallic Samarium Materials', physica status solidi (b) 52 (1972). DOI: 10.1002/pssb.2220520138, https://doi.org/10.1002/pssb.2220520138
  7. H. W. de Wijn, A. M. van Diepen, and K. H. J. Buschow, 'Effect of Crystal Fields on the Magnetic Properties of Samarium Intermetallic Compounds', Physical Review B 7, 524-533 (1973). DOI: 10.1103/PhysRevB.7.524, https://doi.org/10.1103/PhysRevB.7.524
  8. K. H. J. Buschow, A. M. van Diepen, and H. W. de Wijn, 'Moment Reduction in Magnetically Ordered Samarium Intermetallics', Physical Review B 8, 5134-5138 (1973). DOI: 10.1103/PhysRevB.8.5134, https://doi.org/10.1103/PhysRevB.8.5134
  9. H. W. de Wijn, A. M. van Diepen, and K. H. J. Buschow, 'Influence of crystal fields on the magnetic form factor of samarium', Solid State Communications 15, 583-585 (1974). DOI: 10.1016/0038-1098(74)91148-X, https://doi.org/10.1016/0038-1098(74)91148-X
  10. H. W. de Wijn, A. M. van Diepen, and K. H. J. Buschow, 'Samarium in Crystal Fields, a Case of Strong J-Mixing', physica status solidi (b) 76, 11-30 (1976). DOI: 10.1002/pssb.2220760102, https://doi.org/10.1002/pssb.2220760102
  11. S. Geller and G. Balestrino, 'Magnetic phase transitions in samarium iron garnet', Physical Review B 21, 4055-4059 (1980). DOI: 10.1103/PhysRevB.21.4055, https://doi.org/10.1103/PhysRevB.21.4055
  12. M. Guillot, A. Marchand, Y. Nekvasil, and F. Tchéou, 'Anisotropic magnetic properties of samarium iron garnet in high magnetic field', Journal of Magnetism and Magnetic Materials 31-34, 777-778 (1983). DOI: 10.1016/0304-8853(83)90680-7, https://doi.org/10.1016/0304-8853(83)90680-7
  13. G. Balestrino, S. Geller, B. Antonini, P. Paroli, and A. Tucciarone, 'Lowest-temperature magnetic phase transition of samarium iron garnet', Physical Review B 30, 4010-4011 (1984). DOI: 10.1103/PhysRevB.30.4010, https://doi.org/10.1103/PhysRevB.30.4010
  14. V. Nekvasil, M. Guillot, A. Marchand, and F. Tcheou, 'Single-ion theory of the magnetic phase transitions in SmIG', Journal of Physics C: Solid State Physics 18, 3551-3555 (1985). DOI: 10.1088/0022-3719/18/18/019, https://doi.org/10.1088/0022-3719/18/18/019
  15. G. J. Bowden and V. H. McCann, 'On the Knight shift in samarium intermetallic compounds', Journal of Physics F: Metal Physics 16, 1855-1872 (1986). DOI: 10.1088/0305-4608/16/11/021, https://doi.org/10.1088/0305-4608/16/11/021
  16. H. Adachi and H. Ino, 'A ferromagnet having no net magnetic moment', Nature 401, 148-150 (1999). DOI: 10.1038/43634, https://doi.org/10.1038/43634
  17. H. Adachi, H. Ino, and H. Miwa, 'Separation of the 4f-spin, 4f-orbital, and conduction-electron magnetization from exotic thermomagnetic behavior for ferromagnetic Sm intermetallics', Physical Review B 59, 11445-11449 (1999). DOI: 10.1103/PhysRevB.59.11445, https://doi.org/10.1103/PhysRevB.59.11445
  18. H. Adachi, H. Kawata, H. Hashimoto, Y. Sato, I. Matsumoto, and Y. Tanaka, 'Zero-Magnetization Ferromagnet Proven by Helicity-Switching Compton Scattering', Physical Review Letters 87, 127202 (2001). DOI: 10.1103/PhysRevLett.87.127202, https://doi.org/10.1103/PhysRevLett.87.127202
  19. J. W. Taylor, J. A. Duffy, A. M. Bebb, M. R. Lees, L. Bouchenoire, S. D. Brown, and M. J. Cooper, 'Temperature dependence of the spin and orbital magnetization density in Sm0.982Gd0.018Al2 around the spin-orbital compensation point', Physical Review B 66, 161319 (2002). DOI: 10.1103/PhysRevB.66.161319, https://doi.org/10.1103/PhysRevB.66.161319
  20. H. J. Gotsis and I. I. Mazin, 'Ferromagnetism and spin-orbital compensation in Sm intermetallics', Physical Review B 68, 224427 (2003). DOI: 10.1103/PhysRevB.68.224427, https://doi.org/10.1103/PhysRevB.68.224427
  21. S. Qiao, A. Kimura, H. Adachi, K. Iori, K. Miyamoto, T. Xie, H. Namatame, M. Taniguchi, A. Tanaka, T. Muro, S. Imada, and S. Suga, 'Direct evidence of ferromagnetism without net magnetization observed by x-ray magnetic circular dichroism', Physical Review B 70, 134418 (2004). DOI: 10.1103/PhysRevB.70.134418, https://doi.org/10.1103/PhysRevB.70.134418
  22. A. Barla, J. P. Sanchez, F. Givord, J. X. Boucherle, B. P. Doyle, and R. Rüffer, 'Crystal fields, exchange, and conduction electron polarization in SmAl2', Physical Review B 71, 012407 (2005). DOI: 10.1103/PhysRevB.71.012407, https://doi.org/10.1103/PhysRevB.71.012407
  23. A. I. Popov, Z. V. Gareeva, and A. K. Zvezdin, 'Magnetoelectricity of domain walls of rare-earth iron garnets', Physical Review B 92, 144420 (2015). DOI: 10.1103/PhysRevB.92.144420, https://doi.org/10.1103/PhysRevB.92.144420
  24. T. Chatterji, A. Stunault, and P. J. Brown, 'Polarized single crystal neutron diffraction study of the zero-magnetization ferromagnet Sm1-xGdxAl2 (x = 0.024)', Physical Review B 97, 064417 (2018). DOI: 10.1103/PhysRevB.97.064417, https://doi.org/10.1103/PhysRevB.97.064417
  25. R. Michalski and J. Zygadło, 'Predictions of thermomagnetic properties of Laves phase compounds: TbAl2, GdAl2 and SmAl2 performed with ATOMIC MATTERS MFA computation system', Journal of Magnetism and Magnetic Materials 452, 415-426 (2018). DOI: 10.1016/j.jmmm.2017.12.069, https://doi.org/10.1016/j.jmmm.2017.12.069