Spectroscopic and magnetic investigations of actinide-based nanomagnets



Magnetic exchange is an essential feature of transition-metal nanomagnets because it combines the relatively low spin-only moments of several ions into a “giant spin” ground state, which can make slow magnetic relaxation very favorable in an axially anisotropic environment. In contrast, most of the early research on lanthanide-based complexes focused on single-ion magnets, where the required large moment is generated by the unquenched orbital contribution (which is parallel to the spin in heavy rare earths). With their unfilled 5f electronic shell being on the verge between localization and itinerancy, actinides are expected to combine the best of both 3d and 4f metals in terms of exchange and anisotropy, and are therefore under consideration as potential building blocks for the next generation of single-molecule magnets. In this Perspective, a review of the recent development in this field is given, and some discrepancies between the spectroscopic and magnetic data are discussed. © 2014 European Commission. International Journal of Quantum Chemistry published by Wiley Periodicals, Inc.


Single-molecule magnets (SMMs) are zero-dimensional complexes which can retain all or part of their magnetization M for a long period of time after the external magnetic field H used to magnetize them is removed. Under these conditions, the M(H) curve for each single molecule displays an hysteresis cycle similar to that of a bulk permanent magnet; however, instead of the coercivity associated with domain-wall motion in a three-dimensional structure, the hysteresis of a SMM arises from the slow relaxation which results from bistable magnetic configurations at the molecular level. In other words, once the external field is reversed or removed, the fast direct transition between the nonequilibrium magnetized state and the ground state is forbidden and a different relaxation pathway must be followed. If the latter is a thermally activated process, the relaxation time τ is expected to follow an Arrhenius law of the form τ = τ0exp(+Ueff/kBT) and can, therefore, become long enough to ensure bistability at low temperatures.[1]

During the first decade of SMM research, the strategy to improve the effective anisotropy barrier Ueff (and, therefore, the figures of merit such as the blocking temperature) mainly focused on increasing the total spin of the molecule using a larger number of coupled magnetic centers rather than their individual spins; this implied sticking with transition metals of the 3d group as the radial extension of their unfilled electronic shell is very large, which in turn usually translates to a strong magnetic exchange. The anisotropy itself, however is generally not very favorable for this class of elements as pure spin moments are not affected by the crystal-field potential (the main source of magnetic anisotropy in modern bulk permanent magnets).[2] Rare earths, conversely, maintain the orbital degrees of freedom due to the strong electronic repulsion, which maximizes the spin and orbital atomic moment as stated by Hund's rules. It is, therefore, not surprising that the effective barrier Ueff = 230 cm−1 for the first lanthanide-based SMM to be reported, [Pc2Tb]·TBA+ (Pc = phthalocyanine dianion; TBA = N[C4H9]4), was larger than for any 3d-based SMM to date.[3] This complex is made by a single Tb center “sandwiched” between two Pc rings and obviously does not rely on magnetic exchange in any way, hence the alternative definition “single-ion magnet” (SIM) was coined. Similar sandwich structures with different organic cycles have been investigated in the following years: as an example, Np(cyclooctatetraenyl [COT])2 (also known as neptunocene), a similar complex involving a NpIV ion and two COT (C8H8) cycles, was the first transuranic SIM ever reported.[4]

The main mechanisms leading to magnetic relaxation are sketched in Figure 1. The axial part of the crystal-field potential should isolate a suitable doublet ground state such that under a Zeeman splitting the direct relaxation process (dashed arrow) is forbidden (in principle, for a Kramers ion one-phonon processes are not allowed at zero field[5]; to achieve bistability for non-Kramers ions, a strongly axial ligand field which fulfills stringent requirements is crucial).[2] The magnetization will then relax through thermally activated Orbach processes, following the path indicated by full arrows: excited crystal-field states are populated by the absorption of phonons with the right energy, until the system reaches a level which is connected to the ground state by an allowed transition. Direct processes are the equivalent of quantum tunneling of the magnetization (QTM),[6] and when they are strictly forbidden the observed energy barrier Ueff is at least as large as the splitting between the ground and the lowest excited crystal-field levels.

Figure 1.

Top: sketches of the two main mechanisms through which paramagnetic ions achieve relaxation of the magnetization. Direct processes (dashed arrow) involve the emission of one phonon and are strongly influenced by the symmetry and composition of the ground state; otherwise, the magnetization can relax through two-phonon processes (full arrow) involving excited crystal-field states. Bottom: molecular structure of two “double-decker” f-electron SIMs, TbPc2 (left) and Np(COT)2 (right); carbons are black, nitrogens are dark-gray, Tb and Np are light gray, hydrogens are not shown for clarity. See main text for more details.

The 5f shell of actinide ions has a much larger radial extension than the lanthanide 4f shell, but at the same time retains most features of localized electrons (as opposed to the largely itinerant nature typical of transition metals). In this Perspective, I will review the recent efforts of several groups toward the realization of actinide-based SMMs with improved properties. For the sake of clarity, instead of strictly following chronological order, each of the next two sections deals with isostructural complexes, respectively, 5f1 (UV and NpVI) and 5f3 (UIII and NpIV).

The 5f1 Configuration: UV, NpVI

Despite being rarer and less stable than its tetravalent and hexavalent counterparts, pentavalent uranium has been the object of renewed interest during the last few years.[7] From the physical point of view, this particular oxidation state is remarkable because the open 5f shell in the electronic configuration of UV only contains one electron, so that the combined effect of ligand-field potential and spin-orbit interaction gives rise to a relatively simple spectra. For example, the splitting of the cubic orbitals due to spin-orbit coupling (schematized on the left-hand panel of Fig. 2) has been thoroughly studied for octahedral and pseudo-octahedral UV complexes by a combination of spectroscopic (UV-vis-NIR and EPR), and magnetic measurements[8]; this analysis succeeds in reliably describing the fundamental properties from a microscopic point of view, and, therefore, gives valuable information on the bonding and covalency in these systems.[9]

Figure 2.

Schematic energy spectra of a f1-configuration ion. The central panel represents the strong crystal-field case, switching from a purely cubic to purely axial symmetry. The leftmost and rightmost panels show the effect of the spin-orbit interaction on the one-electron orbitals.

A rich subset of the recent advances in UV chemistry deals with the [UO2]+ cation[10]; The uranium centers in uranyl complexes tend to conserve an almost linear O[BOND]U[BOND]O arrangement with short distances, which is expected to give rise to a strongly axial field (and, therefore, to an energy spectra similar to that in the right-hand panel of Fig. 2). However, the local geometry of the additional ligands near the equatorial plane can significantly affect the ground-state composition (and, therefore, the magnetic behavior) by mixing the lowest-energy wavefunctions in a way which depends on their arrangement and on the coordination number of UV; for example, this is clearly seen in the very different effective g-factors measured for several different monometallic uranyl complexes by EPR.[11]

Pentavalent uranium has recently been shown to support SIM behavior in the monometallic complex [U(O)[BOND]{N(Ch2Ch2 NSiiPr3)3}].[12] A butterfly-shaped hysteresis cycle is observed at 1.8 K indicating slow relaxation of the magnetization; despite the relatively low magnetic moment of UV (around 1.8 μB at room temperature) the effective energy barrier extracted from ac susceptibility measurements Ueff = 21.5 K is comparable with that of several standard SMMs, possibly because of the significant anisotropy arising from the strongly axially coordinated environment. Conversely, polymetallic UV-based systems are good candidates to display strong communication between different centers. Mazzanti's group exploited cation–cation interaction to synthesize a series of clusters with different nuclearities, which have been shown to be antiferromagnetic at low temperatures.[13] A stronger superexchange interaction between two UV ions through oxygen ligands was later found in one of Arnold's group “Pacman-shaped” molecules as a result of a rearranged uranium-oxo motif (an oxygen on a uranyl group moved from a trans to a cis position).[14]

Combining a significant exchange coupling with the Ising-like ground state arising from the strongly axial single-ion anisotropy is known to induce peculiar magnetic phenomena; uranium is no exception in this respect, and a recently reported triangular complex formed by three pentavalent uranyl groups displays a nonmagnetic ground state with a net toroidal moment, which results from the chiral arrangement of the moment direction on the different magnetic sites within the molecule.[15] Furthermore, being so prone to couple magnetically, UV has also been successfully paired with both transition-metal and rare-earth elements in heterometallic complexes. An example of the former is {[UO2(salen)]2Mn(Py)3}6, a large wheel-shaped SMM based on six triangular math formula Mn unit, where the pairing with UV improves the overall anisotropy, thus, yielding an effective barrier (Ueff = 99 cm−1) much larger than for any other previously reported manganese-containing cluster;[16] a similar self-assembly reaction has more recently expanded the system dimensionality from zero to one, and resulted in the successful synthesis of the first actinide-based single-chain magnet.[17] As for rare earths, heterobimetallic Pac-man uranyl-lanthanide complexes have been synthesized and studied by EPR spectroscopy and magnetic measurements. Pentavalent uranium seems to retain its Ising ground state, and one UV-SmIII complex displays a very strong superexchange coupling between the 5f1 and the 4f5 centers, which may sound surprising because of the largely reduced radial extent of the 4f shell (with respect to 5f and especially 3d) but might perhaps be explained considering the relative instability of the pentavalent state of uranium and the close divalent state of samarium.[18] Conversely, the dimeric (UV-DyIII)2 complex exhibits a butterfly-shaped hysteresis cycle, behaving as a SMM up to 3 K.[19]

The heterovalent neptunium trimetallic {NpVIO2Cl2} {NpVO2Cl(thf)3}2, to this day the homometallic actinide-based SMM with the highest effective barrier (Ueff = 97 cm−1) as well as the only transuranic one together with neptunocene, seems an appropriate choice to close this section. The ground state of this triangle-shaped complex is strongly influenced by the axial ligand field on the 5f1 NpVI magnetic center, and clear signs of superexchange coupling are visible in the dc susceptibility curves below 20 K. The dominant exchange interaction is an antiferromagnetic coupling between the two NpV-NpVI pairs, whereas the interaction between the two NpV centers is much smaller. Interestingly, this peculiar configuration seems to give rise to an effective ferromagnetic coupling, because a low temperature upturn in the χT-vs-T curve is clearly visible; the reason why this happens is that the NpV ions carry a much larger ground-state moment than NpVI, so that the dominant exchange interaction, albeit antiferromagnetic, actually favors a “parallel” alignment of their two spins.[20]

The 5f3 Configuration: UIII, NpIV

The first actinide-based complex for which slow magnetic relaxation was reported is the trivalent-uranium-based U(Ph2BPz2)3 (Pz = pyrazolyl). An effective relaxation barrier Ueff = 20 cm−1 was extracted from the ac susceptibility; however, the zero-field χ″ peaks amount to only a very small fraction of the total susceptibility and clear deviations from linearity in the Arrhenius plot indicate that direct relaxation processes akin to QTM are active at low temperature despite the Kramers ground state.[21] Shortly thereafter, Long's group also published a thorough investigation of the closely related dihydrobispyrazolylborate complex U(H2BPz2)3, which has a slightly lower anisotropy barrier but displays two distinct and well-resolved relaxation domains visible in the Cole–Cole plot, one of which results from short-range intermolecular interactions.[22]

To reduce the direct tunneling processes which are apparently hindering slow relaxation, several groups reported experiments on the organometallic COT sandwiches belonging to the actinocene family. The neutral uranocene U(COT)2 displays a D8h local symmetry at the magnetic site and its ligand-field Hamiltonian for 5f electron is, therefore, purely axial; however, it contains UIV, which is a non-Kramers ion and has the tendency to prefer a nonmagnetic ground state. To maintain the favorable 5f3 configuration of trivalent uranium, Long's group instead probed the magnetic behavior of its anionic analog, that is, of a K[U(COT)2] complex. However, ac measurements did not detect a sizeable relaxation barrier, whose absence was attributed to a lowering of the local symmetry at the uranium site caused by the influence on the K+ cation on the molecular structure of the complex.[21]

Reasoning along the same lines, Caciuffo's group turned their attention to the following member of the actinocene row, Np(COT)2, as the latter maintains its D8h symmetry while displaying a formal 5f3 configuration on its tetravalent neptunium site and a |MJ ± 5/2〉 Kramers' doublet ground state; moreover, dc susceptibility measurements suggest that the lower excited states may be about 1400 cm−1 above the ground doublet, which could translate into a huge energy barrier against magnetic relaxation in absence of other effective channels. However, even in this case only a small ac peak is observed without any static external magnetic field, although 0.5 T are enough to make this behavior change completely. Even if no ligand-field transverse terms are present, in fact, the electronic ground-state doublet can interact with the nuclear moment of 237Np via the hyperfine interaction, and this gives rise to several crossing points which are expected to provide very fast relaxation pathways in a way similar to QTM. Applying a 7 T magnetic field the Zeeman splitting of the electronic doublet is for most orientations so large that the hyperfine coupling becomes ineffective; indeed, the Arrhenius curve for this field is linear and practically vertical. As a result of these peculiar magnetic properties, the measured magnetization curve for neptunocene displays partial hysteresis starting from saturation as soon as the magnetic field is decreased but immediately closes back to the original curve at lower fields as soon as the fast relaxation processes kick in.[4]

Two other U(III)-based SIMs, the cationic [U(TpMe2)2(bipy)]+ and its precursor [U(TpMe2)2I], were then reported by Almeida's group.[23, 24] The properties of these low-symmetry complexes–slow relaxation with Ueff = 18.2 and 21.0 cm−1, respectively,–are not dissimilar from those of U(Ph2BPz2)3, and the fact that significant deviations from the Arrhenius behavior are also present here at low temperature suggests that the transverse ligand-field terms for these nonaxial geometries facilitate faster relaxation processes.

A corrected crystal-field model was successfully used[25] to describe the static magnetic properties of the four UIII-based SIMs introduced so far as well as a fifth, UTp3 (Tp = trispyrazolylborate).[26] The model suggests that the local geometry of the magnetic center is a key factor to consider in order to obtain slow relaxation;[25] in particular, even in presence of deviations from the ideal symmetry, trigonal prismatic structures should display significantly less quantum tunneling (and, therefore, be more prone to magnetic bistability) with respect to tetragonal ones. Conversely, more or less at the same time another paper pointed toward somewhat different conclusions, experimentally showing that three novel UIII-based complexes all display slow magnetic relaxation with similar values of Ueff and τ0 despite being very different in terms of the crystallographic and electronic structure, and in particular of the local symmetry.[27]

One point which is still hampering the efforts toward understanding the fundamental properties of these actinide complexes is the marked discrepancy between the static and the dynamic magnetic behavior. This can best be seen in the case of UTp3,[26] which is particularly interesting because its unique UIII site has a very high, crystallographically exact D3h symmetry, and the availability of extensive optical spectroscopy data allowed Amberger and coworkers[28] to fit all its crystal-field parameters in the 5f3 single-ion Hamiltonian. In particular, the energy gap Δ between the lowest Г7 and the first excited Г9 doublets (270 cm−1) was directly measured, and its value agrees very well with the crystal-field calculations; furthermore, despite the fact that the crystal-field refinement was done using only the spectroscopic data, the temperature dependence of the dc magnetic susceptibility calculated with the obtained parameters agrees almost perfectly with the experimental curve (Fig. 3). These observations, however, are very difficult to reconcile with the dynamic properties, which are characterized by an Arrhenius behavior of the relaxation time as a function of temperature with an effective barrier Ueff = 3.8 cm−1, the lowest amongst all complexes discussed in this Perspective.[26]

Figure 3.

Open diamonds: experimental dc magnetic susceptibility measured for the UTp3 SIM (whose structure is depicted in the top-left corner; hydrogens not shown, other atoms from lighter to darker shade: boron, uranium, nitrogen, carbon). The black curve is calculated using the crystal-field parameters determined by optical spectroscopy, which points to an energy gap of 270 cm−1 between the ground and the first-excited levels. Bottom-right inset: temperature dependence of the relaxation time extracted from ac susceptibility curves. Fitting the data to the Arrhenius equation (dashed line in the log-versus-reciprocal plot in the inset) gives Ueff = 3.8 cm−1.

In fact, gathering together selected data from several literature sources (Table 1) one can see that this discrepancy between the electronic spectra and the magnetic relaxation behavior seems common to all the 5f3 SIMs studied so far. This is in stark contrast with the situation observed for archetypal lanthanide-based SIMs such as the phtalocyanine double-decker family (also included in Table 1 for comparison), whose zero-field energy gap is quantitatively very similar to the effective barrier against magnetic relaxation.[3, 29] The origin of this discrepancy is not yet clear; intermolecular interactions have been shown to be ineffective in many of these systems, and the isotopic composition of natural uranium is dominated (with an abundance of more than 99%) by 238U which, unlike 237Np, has no nuclear magnetic moment thus ruling out any hyperfine coupling with the electronic ground state. A fast relaxation process can be present even in Kramers ions (where time-reversal symmetry assures that no tunneling gap splits the ground electronic doublet) if internal moment fluctuations generate a small transverse magnetic field, but a small applied dc field is normally enough to eliminate it completely.[5] All these observation seem to suggest that actinide relaxation processes are more rich and complex when compared to transition metals and even lanthanides.

Table 1. Values of the effective energy barrier against magnetic relaxation (Ueff), zero-field energy gap between the two lowest levels (Δ), and estimated ground-doublet wavefunction composition for several isoconfigurational (5f3) actinide-based SIMs discussed in this Perspective.
ComplexUeff (cm−1)Δ (cm−1)Electronic ground-state compositionReferences
  1. Two lanthanide-based complexes are also included for comparison.

  2. a

    Measured on a 1:90 (U:Y) diluted sample; reduces to 8 cm−1 for a pure sample.

  3. b

    Measured under a 3 T applied dc magnetic field.

  4. c

    The first value of Δ is calculated with the SO-CASPT2 method, whereas the second one is obtained by a corrected crystal-field model.

U(Ph2BPz2)32019079% |±5/2〉, 17% | math formula7/2〉21, 25
U(H2BPz2)316a23068% |±5/2〉, 24% | math formula7/2〉22, 25
Np(COT)228.5b1400100% |±5/2〉4
[U(TpMe2)2(bipy)]+18.2136, 138c31% |±3/2〉, 27% | math formula5/2〉, 15% | math formula7/2〉23, 24, 25
U(TpMe2)2I21.0110, 146c35% |±9/2〉, 34% |±5/2〉, 28% | math formula3/2〉24, 25
UTp33.827059% |±5/2〉, 38% | math formula7/2〉26, 28
[Pc2Dy]·TBA+2835100% |±13/2〉3
[Pc2Tb]·TBA+230220100% |±6〉3, 29

In terms of polymetallic complexes, so far the only relevant example which must be included in this section is the inverted-sandwich diuranium molecule {[U(BIPMTMSH)(I)]2(μ-η6: η6-C6H5CH3)} (BIPMTMS = C(PPh2NSiMe3)2), which has been studied to verify how the interaction between the two arene-bridged UIII centers would affect the magnetic properties. Slow magnetic relaxation has been detected in the ac susceptibility curves, and a butterfly-shaped hysteresis cycle has been measured, both indicating SMM behavior; however, no indication of strong magnetic coupling is evident in the dc susceptibility data, and also the effective magnetic moment at saturation seems similar to that of several UIII-based SIMs.[30]


Despite their less laboratory-friendly nature when compared to nonradioactive samples, actinide-based complexes have recently shown an enormous potential to pave the ground for the next generation of molecular nanomagnets. In particular, ions with 5f3 electronic configuration seem to display an uncanny ability to behave as SIMs in a way which is largely independent of the ligand geometry, whereas 5f1 ions support significant superexchange interactions when coupled with 3d, 4f, and 5f metals while retaining a strongly axial anisotropy. In addition to this, actinides have a larger coordination sphere, many more accessible oxidation states, and in general a much richer chemistry with respect to lanthanides, which in principle offers almost endless tailoring possibilities. However, to advance our understanding of magnetic relaxation in these systems, strong efforts will be required both toward expanding the number and variety of experimentally characterized actinide-based SMMs (e.g., moving one step up on the periodic table and focusing attention on plutonium, whose well-known chemistry could lead to SMMs with different configurations and a fuller 5f shell) and to refine the computational methods (e.g., exploiting ligand-field DFT calculation methods similar to those used for lanthanide-based systems,[31] or focusing on the apparent success of SO-CASPT2 methods in reproducing the static electronic spectra[4, 24] and on how to connect this with the dynamic magnetic properties).


  • Image of creator

    Nicola Magnani obtained his Ph D. at the University of Parma (Italy) in 2003 under the supervision of Prof. Giuseppe Amoretti. After postdoctoral experiences in the theoretical modeling of magnetic materials and nanomagnets his main research interests shifted to actinide-based molecules and compounds, and in 2006 he became a fellow of the Institute for Transuranium Elements in Karlsruhe (Germany). Between 2010 and 2012 he worked as a project scientist at the Lawrence Berkeley National Laboratory (USA).