Uranium and manganese assembled in a wheel-shaped nanoscale single-molecule magnet with high spin-reversal barrier

Uranium and manganese assembled in awheel-shaped nanoscale single-moleculemagnet with high spin-reversal barrierVictor Mougel1, Lucile Chatelain1, Jacques Pecaut1, Roberto Caciuffo2, Eric Colineau2,

Jean-Christophe Griveau2 and Marinella Mazzanti1*

Discrete molecular compounds that exhibit both magnetization hysteresis and slow magnetic relaxation below acharacteristic ‘blocking’ temperature are known as single-molecule magnets. These are promising for applications includingmemory devices and quantum computing, but require higher spin-inversion barriers and hysteresis temperatures thancurrently achieved. After twenty years of research confined to the d- block transition metals, scientists are moving to thef-block to generate these properties. We have now prepared, by cation-promoted self-assembly, a large 5f–3d U12Mn6

cluster that adopts a wheel topology and exhibits single-molecule magnet behaviour. This uranium-based molecular wheelshows an open magnetic hysteresis loop at low temperature, with a non-zero coercive field (below 4 K) and quantumtunnelling steps (below 2.5 K), which suggests that uranium might indeed provide a route to magnetic storage devices.This molecule also represents an interesting model for actinide nanoparticles occurring in the environment and in spentfuel separation cycles.

In the quest for systems that can function as molecular nanomag-nets, and find application in information storage, quantum infor-mation processing, spintronics and magnetocaloric refrigeration1–5,

a number of increasingly larger molecular clusters containing one ormore types of d-block transition metals have been synthesized.

The development of single-molecule magnets (SMMs) requiresthe association of high-spin ground states (S) with a large magneticanisotropy (D). Together, these properties create a barrier to magne-tization reversal—and thus a magnetization hysteresis—below a‘blocking’ temperature TB that is specific to each system. Withinthe 3d-block, manganese(III) clusters are the most studied SMMcompounds because of the high uniaxial anisotropy and spinground state of the Mn(III) ion, and have provided the highestreported relaxation barriers (Ueff¼ S2|D| up to 86.4 K with S¼12) and blocking temperatures (�4.3 K)4. High-spin ground statesup to S¼ 83/2 have been obtained by associating high-spinMn(II) to Mn(III) in large clusters, but in these systems the presenceof the isotropic Mn(II) ion and the geometry of the anisotropicMn(III) ions result in a low magnetic relaxation barrier(a hysteresis below 0.5 K has been measured for the Mn19S¼ 83/2)6.

Although high spin states can be achieved with d-block ions, felements have higher single-ion anisotropy, which makes themvery attractive for the development of SMMs with improved proper-ties. Notably, the molecular compounds showing the highest relax-ation barriers reported to date are mono- or multimetalliclanthanide complexes, with a record barrier of 530 K having beenachieved for a Dy6 cluster7–10. However, only a few complexeshave shown hysteresis in the magnetization: a bis-phthalocyaninato(Pc) rare earth(III)7 compound, and two dinuclear complexes con-sisting of Dy(III)8 or Tb(III)11 ions linked by a N2

32 radical, whichshowed blocking temperatures of 8.3 and 14 K, respectively.

SMMs based on actinide ions, such as uranium, have not beenstudied to such an extent, and the first examples, U(III) andNp(IV) mononuclear complexes12–14, have only recently beenreported. A dinuclear complex, for which the presence of magneticcoupling between the U(III) ions remain ambiguous, also showsSMM behaviour15,16. A combination of slow relaxation of the mag-netization and effective superexchange interactions (that is, occur-ring between two magnetic centres through a non-magneticbridge) between 5f ions has been observed only in a trinuclearheterovalent neptunyle trimer17. So far, however, magneticmemory effects in 5f-block clusters have been reported only in theform of butterfly-shaped hysteresis loops, with negligibleremanent magnetization at zero applied field, even at the lowestobservation temperature.

Actinides are particularly attractive for attaining higher relax-ation barriers because, in contrast to lanthanide ions, they can estab-lish partially covalent interactions and therefore be involved inmagnetic communication18–25, leading to concerted magnetic behav-iour. As well as focusing on their potential applications,magnetic actinide complexes are of high fundamental interest inthe investigation of the role of 5f orbitals in bonding and magneticproperties. However, the supramolecular chemistry of actinides ispoorly developed26, with only a few examples of large paramagnetichomometallic clusters described in the literature27–29. A lack ofappropriate synthetic approaches means that heterometallicsystems containing 5f and 3d metals are even rarer, being limitedto a few dinuclear and trinuclear examples20,30–32. For some ofthem20,30, clear evidence of 5f–3d magnetic coupling has beenreported, but to date there are no examples of 5f–3d complexesshowing SMM behaviour.

The development of synthetic strategies leading to large 5f–3dassembly is also of high relevance to nuclear technology and

1Laboratoire de Reconnaissance Ionique et Chimie de Coordination, SCIB, UMR-E 3 CEA-UJF, INAC, CEA-Grenoble, 17 rue des Martyrs, F-38054 GrenobleCedex 09, France, 2European Commission, Joint Research Centre, Institute for Transuranium Elements, PO Box 2340, D-76125 Karlsruhe, Germany.

*e-mail: [email protected]

ARTICLESPUBLISHED ONLINE: 11 NOVEMBER 2012 | DOI: 10.1038/NCHEM.1494

NATURE CHEMISTRY | VOL 4 | DECEMBER 2012 | www.nature.com/naturechemistry 1011

© 2012 Macmillan Publishers Limited. All rights reserved.

associated environmental clean-up strategies. 3d transition-metalions are present in the environment and in spent nuclear fuelstreams, and therefore nanosized clusters formed by actinides and3d elements provide a good model of species involved in actinidemigration and of colloidal species affecting the technology ofnuclear fuel reprocessing.

Cation–cation interactions, a term used to describe the inter-action of the actinyl oxo groups with the metal of another actinylgroup, or with metal cations from the alkali, 4f or 3d series are akey feature of solid-state and molecular actinide chemistry, whichprovide an attractive route to supramolecular structures and mag-netic communication14,21,33,34. Notably, we have recently isolatedstable dinuclear21, trinuclear22 and tetranuclear21,23,35 complexes ofpentavalent uranyl, assembled via UO2

þ–UO2þ and UO2

þ–M(M¼ K, Rb) interactions, which present unambiguous magneticcommunication, rarely found in actinide ions. These polynuclearcomplexes are noteworthy because of their stability with respectto the disproportionation reaction that is commonly observed forpentavalent uranyl. Only dinuclear complexes showing cation–cation interaction between UO2

þ and Fe(II), Zn(II) or Ln(III)cations have been reported to date. In the last case, the magneticdata were analysed in terms of UO2

þ–Ln(III) antiferromagneticinteraction31,34. In contrast to these dinuclear systems, here wereport the formation of a large cluster resulting from the interactionof UO2

þ with Mn(II) cations. We show that the ability of the Schiffbase complexes of pentavalent uranyl to form cation–cation inter-actions with elements of the 3d block provides a versatile route tothe assembly of a U12Mn6 wheel, which is the largest reported hetero-metallic 5f–3d complex. We also demonstrate that the topology of thepolynuclear assembly is tuned by the nature of the cation (UO2

þ–Ca2þ interaction yields a tetramer). Whereas all the previously reportedexamples of actinides-based SMMs show butterfly-shaped hystereticloops14,15, the U12Mn6 wheel presents, below TB¼ 4 K, an open stair-case-like hysteresis with non-zero remanent magnetization, a necessaryrequirement for information storage. The coercive field Hc (the mag-netic field required to switch the magnetization from saturation tozero) increases with decreasing temperature and reaches a value of�1.4 T at 2.25 K. This behaviour does not originate from

intermolecular cooperative interactions as in long-range magneticallyordered systems, but is of purely molecular origin and is related tothe presence of the energy barrier hindering the relaxation of the mag-netization towards equilibrium. Abrupt steps in the hysteresis loop,appearing below �2.5 K at m0H¼ 0 and 1.65 T, reveal that at thesefields the relaxation rate is strongly enhanced by quantum tunnellingof the magnetization through the relaxation barrier, providing evidenceof quantum-mechanical properties on a macroscopic scale, as observedin several transition-metal and rare earth systems but never reported foran actinide complex2.

Results and discussionSynthesis and structural characterization. We have previouslyidentified a convenient route to salen-based heterometallictetranuclear uranyl(V)–uranyl(V)–M (M¼ K, Rb) cation–cationclusters that consists in reacting the monomeric [UO2(salen)(Py)][Cp*2Co] (salenH2¼N,N′-ethylenebis(salicylimine); Cp*¼pentamethylcyclopentadienyl; Py¼ pyridine) complex with thedesired MI salt (M¼K, Rb)35. Here, we have used this strategy toprepare cation–cation clusters with divalent cations. At first, westudied the influence of the presence of a divalent alkaline-earthmetal (which has no preferential coordination number orgeometry) on the final structure. We then used paramagneticdivalent manganese to assemble a cation–cation cluster containinga UO2

þ–Mn interaction and to promote magnetic couplingbetween the isotropic Mn2þ ion and the anisotropic uranyl(V) ion.The most common geometry for manganese(II) is octahedral(although it can be found in other geometries depending on theligand set). The preference of Mn(II) for an octahedralcoordination geometry in the reaction conditions used in thiswork affords a cation–cation cluster with a new wheel structure.Thus, we demonstrate that the presence of a transition metal witha specific geometric preference can be used to control the finalstructure and to design new cluster topologies.

The reaction of 2 equiv. of monomeric uranyl(V) complex[UO2(salen)(Py)][Cp*2Co] with 1 equiv. of CaCl2(DME)(DME¼ dimethoxyethane) in pyridine results in the formation ofthe tetrameric complex {[UO2(salen)]4Ca2} (1) in 70% yield (Fig. 1).

[Cp*2Co]

Mn(NO3)2

Pyridine

Salen2−

CaCl2(DME)

Pyridine

N N

O O

1

2N

N

O

O

O

U

O

N

N

O

OO U O

NN

OO

O

U

O

N

N

O

OOUO

N N

O O

Ca

CaO

U

O

N

N

O

O

O

U

O

N

N

O

O

O

U

ON

NO

OO

U

O N

NO

O

O U O

NN

OO

OUO

N N

O O

O

U

O

N

N

O

O

O

U

O

N

N

O

O

O

U

O N

NO

OO

U

ON

NO

O

OUO

N N

O O

O U O

NN

OO

Mn

Mn

Mn

Mn

Mn

Mn

Py

Py

Py

Py

Py

Py

Py

Py

Py

Py

Py

Py

Py

Py

Py

Py

Py

Py

O

U

O

N

N

O

O

=–

–

–

–

Figure 1 | Reaction scheme. The mononuclear pentavalent uranyl complex [UO2(salen)(Py)][Cp*2Co] reacts with CaCl2(DME) to produce tetrameric

complex 1 and with Mn(NO3)2 to yield [{[UO2(salen)]2Mn(Py)3}6] (2), a dodecanuclear uranyl(V) complex containing six manganese(II) centres.

ARTICLES NATURE CHEMISTRY DOI: 10.1038/NCHEM.1494

NATURE CHEMISTRY | VOL 4 | DECEMBER 2012 | www.nature.com/naturechemistry1012

© 2012 Macmillan Publishers Limited. All rights reserved.

An X-ray diffraction study carried out on single crystals of1 grown from a dichloromethane solution showed the presenceof a square-shaped tetranuclear structure (Fig. 2a,c), similar tothat found for the previously reported23,35 tetranuclear uranyl(V)salen complexes {[UO2(salen)]4[m8-K]2}.2[K(18C6)(Py)] and{[UO2(salen)4][m8-Rb]2[Rb(18C6)]2} (18C6 is 18-crown-6, or1,4,7,10,13,16-hexaoxacyclooctadecane)23,36.

Similar to the reactivity observed with calcium, the reactionof 2 equiv. of monomeric uranyl(V) complex [UO2(salen)(Py)][Cp*2Co] with 1 equiv. of Mn(NO3)2 in pyridine produced ahighly insoluble dark violet microcrystalline powder. The presenceof Mn(II) results in a lower solubility and stability with respectto complex 1; attempts to recrystallize it from dichloromethaneresulted in partial decomposition, as indicated by the NMR spec-trum showing the presence of uranyl(VI) salen in the resulting sol-ution. However, crystals of reasonable quality were obtained byslow diffusion of a solution of 1 equiv. of Mn(NO3)2 in pyridineto a solution of 2 equiv. of [UO2(salen)(Py)][Cp*2Co] in pyridine.X-ray diffraction studies revealed the presence of a U12Mn6 clusterof uranyl(V) with a wheel topology (Fig. 2b,d).

Complex 2 provides a new example of a uranyl(V) cluster that isstable with respect to the disproportionation reaction and is thelargest reported to date and the first containing UO2

þ–Mncation–cation interactions. In contrast with the uranyl(V) clusterspreviously reported, 2 does not contain UO2

þ...UO2þ interactions;

only the phenolate oxygens from the salen ligand bridge theuranium centres (Supplementary Fig. S13).

The structure of 2 is described as a centrosymmetric hexamerassembled from six triangles consisting of two salen-bound UO2

þ

cations, mutually coordinated through two salen–phenolatebridges, which are both involved through the uranyl oxygen in acation–cation interaction with the same Mn2þ ion. This structurediffers significantly from those of complex 1 and the few othercharacterized discrete polynuclear complexes of pentavalenturanyl. In all these systems, the oxo group of the uranyl moietyacts as a bridging group between two U atoms, producing differentgeometrical arrangement (T-shaped21,35, diamond-shaped21,34 andbutterfly-shaped37, as shown in Supplementary Fig. S13) with U–Udistances ranging from 3.35 to 4.19 Å. In contrast, in the U12Mn6wheel, two phenolate oxygens (each from a different salen ligand)bridge two uranyl(V) centres at 3.92(1) Å and one oxo group fromeach uranyl(V) complex binds a Mn(II) ion to produce a triangle.

The six triangles are connected together to yield the final U12Mn6wheel through the cation–cation interaction of the manganese ionfrom one triangle with the uranyl oxygen of an adjacent triangle.As a result, both oxygens of six uranyl(V) complexes are bound toa Mn(II) ion; for the remaining six uranyl(V) complexes only oneof the two oxygens is Mn-bound. Each Mn(II) ion is six-coordinatedby three pyridine nitrogens and by three uranyl(V) oxo groups fromthree different uranyl(V)–salen complexes, of which two belong to

O(2U1)O(1U1)

O(2U2)

O(1U2)159.43º

178.56º

4.259 Å

4.247 Å

U(2)

U(1)

O(2U1)

U(1)

O(1U1)

O(2U4)

O(1U4)

U(4)

3.921 Å

147.51º

3.895 Å

3.904 Å3.91

4 Å

3.920 Å155.75º

169.74º

154.18º148.03º

145.85º168.95º

155.55º

Mn(1)

4.009 Å3.926 Å Mn(2)

O(1U5)

U(5)

O(2U5)

O(2U6) O(1U6)Mn(3)

U(6)

O(1U3)

O(2U3)

O(1U2)U(2)

O(2U2)

U(3)

Mn(2)

167.70º Mn(3)

Mn(1)

4.047 Å

4.028 Å

3.873 Å

3.93

7 Å

3.918 Å

a c

bd

Figure 2 | Solid-state structure of {[UO2(salen)]4Ca2} (1) and [{[UO2(salen)]2Mn(Py)3}6] (2). a–d, Ellipsoid plots at 50% probability of 1 (a) and 2 (b)

and detail of the cores in ball-and-stick representations of 1 (c) and 2 (d). Co-crystallized solvent molecules and H are omitted and ligands are represented

with pipes for clarity. C atoms are represented in grey, O atoms in red, N atoms in blue, Ca atoms in turquoise, Mn atoms in magenta and U atoms in green.

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1494 ARTICLES

NATURE CHEMISTRY | VOL 4 | DECEMBER 2012 | www.nature.com/naturechemistry 1013

© 2012 Macmillan Publishers Limited. All rights reserved.

the same triangle. The twelve U and six Mn ions are coplanar (meandeviation from the mean plane¼ 0.19(3) Å) and are arranged in alarge circular array with a diameter of 2 nm (longest distancebetween two U ions). The 2:1 UO2

þ:Mn2þ ratio ensures thebalance of charges and gives a neutral cluster. The mean value ofthe Mn–Mn distances (7.89(3) Å) is much longer than thosereported for Mn6 clusters presenting magnetic interaction betweenthe Mn ions (3.2–3.4 Å)38.

The mean Mn–Oyl bond distance (where Oyl is a uranyloxygen)—2.15(2)Å—is similar to that found in a heterodinuclearuranyl(VI)–Mn(II) complex of a tetra-anionic pyrrole-imine macro-cycle (often called ‘pacman ligand’) (Mn–O¼ 2.163(4) Å)39. Thedistance falls in the range of Mn–OPh (where OPh is a phenyloxygen of the salen ligand) distances reported for Mn(II) ions inmanganese clusters (2.135–2.500 Å)6,40. A similar distance (takinginto account the difference of 0.128 Å between the Sm(III) andMn(II) ionic radii) was also found for a uranyl(V)–Sm(III) complex(2.238(5) Å), showing strong magnetic coupling between uraniumand samarium39.

As observed in other UO2þ cation–cation clusters, the mean value

of the U–Oyl distance is longer for the Mn-functionalized oxo group(U–Oyl(Mn)¼ 1.89(1) Å) than for the oxo group not involved incation–cation interactions (U–Oyl¼ 1.83(3) Å). These U–Oyldistances are longer than those found in the uranyl(VI)–Mn(II)‘pacman’ complex with the tetra-anionic pyrrole-imine macrocycleligand (U–Oyl¼ 1.768(5) Å and U–Oyl(Mn)¼ 1.808(4) Å), inagreement with the presence of UO2

þ.The new topology of the structure of 2 compared to the

previously obtained dinuclear, trinuclear and tetranuclearcation–cation complexes is most probably the result of a combi-nation of structure-directing parameters — the 2:1 UO2

þ:Mn2þ

ratio used, the divalent charge of the Mn2þ ion, and the strongpreference of divalent manganese for a octahedral geometry. Here,the UO2

þ–Mn2þ cation–cation interaction plays the structure-directing role.

Although several uranyl(V) mononuclear, polymeric and oligo-meric complexes with different topologies have been reported thatcontain alkali ions23,35,41, Fe(II)42, Zn(II)42, Sm(III)38 or Y(III)38, weare not aware of any other uranyl(V)–Mn(II) clusters. In contrast

to previously reported uranyl(V) cluster compounds, 2 does notcontain UO2

þ–UO2þ interactions but is exclusively built from the

functionalization of the uranyl-oxo group by a Mn(II) ion. This pro-vides further insight into the structure-directing parameters andshould open the way to a rich variety of fascinating topologies.Moreover, 2 is the largest uranyl(V) cluster reported to date, withan original wheel topology that complements the previouslyreported diamond21,34, square23,35 and triangular structures22.

Magnetic characterization. The temperature dependence of thed.c. magnetic susceptibility of the tetrameric U4Ca2 complex 1(Supplementary Fig. S10) is very similar to that reported for the[UO2(salen)]4[m8-K]2K2 analogue, showing a cusp at �5 K, whichinitially suggested the presence of oxo-mediated antiferromagneticcoupling between the two uranyl ions. However, further magneticcharacterization did not show features that would be consistentwith a single-molecule magnet behaviour for 1, as we hadanticipated from the antiferromagnetic character of the U–Uinteraction.

Figure 3 shows the temperature-dependent d.c. magnetic suscep-tibility of the U12Mn6 wheel, xM(T), measured with a supercon-ducting quantum interference device (SQUID) magnetometer anddisplayed as the product TxM(T). Below �60 K, the TxM(T)curve shows a strong deviation from Curie behaviour. The increasewith decreasing temperature observed between �60 and �30 K andthe field variation of the magnetic response are very similar to thosereported for the triangular-shaped {NpVIO2Cl2}{NpVO2Cl(thf )3}2complex (thf¼ tetrahydrofuran)17. In that case, the observed behav-iour was understood as a combination of ligand field and superex-change interactions between the 5f centres. We suggest that asimilar scenario is realized in 2.

The ground-state degeneracy of the ions coupled by superex-change interactions is lifted by the magnetic field, leading to ahigher energy state with parallel U and Mn magnetic momentsand a lower energy state with antiparallel orientations. The suscep-tibility first increases with decreasing temperature because of thehigher energy level contribution, then drops down when only thelower energy level is thermally populated.

The finite value of 1.5 e.m.u. K mol21 for TxM(T), observed at�2 K for B¼ 1 T, suggests a magnetic ground state for the wheel,

0

5

10

15

20

25

30

0 50 100 150 200 250 300

1 T3 T5 T7 T

TX M

(T)

(e.

m.u

. K m

ol–1

)

Temperature (K)

0

0.6

1.2

0 20 40 60 80 100

X M (

e.m

.u. m

ol–1

)

T (K)

Figure 3 | Temperature dependence of the molar d.c. magnetic

susceptibility xM(T ) of complex 2. The TxM(T) data were collected after

zero-field cooling in a magnetic field of 1 T (red circles), 3 T (green squares),

5 T (open olive circles) and 7 T (brown squares). Inset: xM(T) curve

between 2 and 100 K in a field of 3 and 7 T. A contribution due to a

ferromagnetic impurity has been subtracted, as discussed in the

Supplementary Section S5.1.

–12

–3

0

6

12

–8 –6 –4 –2 0 2 4 6 8

2.25 K

4 K

Magnetic field (T)

M (μ B

/mol

ecul

e)

–8 –4 0 4 8

–12

–6

0

6

12

4.5 K

M (μ B

/mol

ecul

e)

Magnetic field (T)

Figure 4 | Low-temperature magnetic hysteresis loops showing an open

cycle. Magnetization versus applied d.c. field scan measured at 2.25 and 4 K

while sweeping the field from 7 to 27 T and back, with a sweep rate of

0.004 T s21. Step-like changes at periodic field values are due to quantum

tunnelling of the magnetization. Data collected at 4.5 K are shown in

the inset.

ARTICLES NATURE CHEMISTRY DOI: 10.1038/NCHEM.1494

NATURE CHEMISTRY | VOL 4 | DECEMBER 2012 | www.nature.com/naturechemistry1014

© 2012 Macmillan Publishers Limited. All rights reserved.

which is expected in the presence of strong antiferromagnetic inter-actions between the UV and the MnII centres and a weaker antifer-romagnetic interaction between the two UV centres within atriangle. The former coupling is mediated by the uranyl O atoms,forming almost linear bonds with mean values of U–Mn distancesof 3.92(1) and 3.89(2) Å within a given triangle, and of 4.03(1) Åbetween adjacent triangles. The interaction between two U(V)ions belonging to a given triangle is mediated by the salen–phenolate oxygens with mean U–U distance at 3.921 Å. Such anexchange topology would result in parallel coupling between theU moments, and antiparallel between the U and Mn moments.The difference between the magnetic moment of the MnII ions inthe high-spin (S¼ 5/2) state and that of the 5f1 ions with a G7doublet stabilized by a strong axial ligand field means that theground state of the wheel is magnetically uncompensated. A mag-netic ground state would also be obtained in the case of low-spinMnII ions and a quasi-quartet ground state for the U ions.

Above 100 K, fitting TxM(T) as the sum of a Curie and a vanVleck term gives a T-independent contribution of�0.01 e.m.u. mol21 (due to the population of excited ligand-fieldstates) and a paramagnetic moment meff¼ 13.5+0.2mB, that is,5.5mB per each triangular unit. This value is about half that expectedfor a system formed by one MnII and two UV free ions, suggestingthat the overall exchange and ligand field splitting is much largerthan 300 K. No attempts to quantify ligand field and exchange inter-actions have been put forth because the complexity of the systemprevents a quantitative analysis in the absence of further infor-mation. The successful synthesis of an isostructural analogue withMn replaced by a diamagnetic ion (such as CdII) and its magneticcharacterization would allow the separate quantification of theU–U interactions in the complex and will be the subject offurther studies.

The presence of a magnetic ground state is confirmed by theobservation of magnetic hysteretic loops. As shown in Fig. 4, mag-netic bistability is observed in the magnetization versus applied d.c.field scan taken at 4 K. With decreasing temperature the coercivefield increases, reaching a value of �1.5 T at 2.25 K. This behaviouris typical of a single-molecule magnet below its blocking tempera-ture TB (refs 43–45). Moreover, highly resolved step-like featuresare observed below 2.5 K, revealing the occurrence of quantum tun-nelling of the magnetization increasing the relaxation rate44,46,47.These phenomena have been reported previously for several tran-sition-metal complexes48, in mononuclear lanthanide phthalocya-nines7 and in mixed 3d–4f complexes49,50.

The in-phase component of the a.c. susceptibility, xM′, shows a

peak at a frequency-dependent temperature, reaching �10 K at

�1 kHz, accompanied by a maximum in the out-of-phase com-ponent xM

′′ clearly indicating the occurrence of slow magneticrelaxation (Fig. 5). The overall behaviour of the peaks in xM

′′

closely resembles the data of Ishikawa et al. on diluted rare-earth bis-phthalocyanine samples rather than on pure ones7,and together with the ratio between the peak amplitudes in theout-of-phase and in-phase susceptibility components14 confirmsthat intermolecular interactions are extremely weak. The crystalstructure of this complex clearly shows that there are no strongintermolecular contacts, with the shortest intermetallic distancesbeing 8–10 Å, ruling out the presence of strong intermolecularmagnetic interactions.

The relaxation behaviour can be fitted (see Methods andSupplementary Section S5.2) to an Arrhenius relation, t¼t0 exp(D/kBT), corresponding to a thermally activated regime,and a linear regression of the experimental data provides a pre-exponential factor of t0¼ (3+2) × 10212 s and a barrier to relax-ation of D¼ 142+7 K (Fig, 5c), larger than for any previouslyreported manganese cluster4. The value of t0 is smallerthan for typical small transition-metal SMMs (for instance, t0 isof the order of 1 × 10210 s for the manganese compound[Mn(III)6O2(Etsao)6{O2CPh(Me)2}(EtOH)6]) (ref. 4), but similarvalues are commonly found in high-nuclearity SMMs51.Moreover, a much smaller value of t0 in 5f-block SMMs than intransition-metal SMMs can be expected because active orbitaldegrees of freedom can affect the magnetoelastic interaction.

ConclusionsOur results demonstrate that cation–cation interaction betweenactinyl complexes and 3d transition-metal cations provides an effec-tive way to build large heterometallic 5f–3d assemblies. The U12Mn6wheel is the largest heterometallic 5f–3d cluster reported to date. Arich variety of topologies can be anticipated by the reaction betweenactinyl complexes in different ligand environments and in the pres-ence of different transition-metal cations. Future efforts will bedirected in this direction for the development of new stable polyme-tallic clusters based on pentavalent uranyl and its NpO2

þ analogues.The U12Mn6 wheel prepared in this work exhibit superparamagneticbehaviour with a relaxation barrier higher than that of any molecu-lar wheel reported so far. In contrast with the few previous reportson 5f-block organo-metallic SMM complexes, which all present but-terfly-shaped hysteresis with zero coercive field, this U12Mn6 clustershows open staircase-like magnetization hysteretic loops with non-zero coercive field (below about 4 K) and clear evidence of quantumrelaxation phenomena (below 2.25 K). The interesting magneticproperties of the U12Mn6 cluster suggest that the use of the highly

0

0.3

0.6

a b c

18 Hz 311 Hz987 Hz 2,791 Hz6,217 Hz9,887 Hz

Temperature (K)

0

1.2

2.4

–15

–10

–5

0

5

ln(τ

/s)

0 5 10 15 20

Temperature (K)

0 5 10 15 20

T –1 (K–1)

0.1 0.14 0.18 0.22

X M≤

(e.m

.u. m

ol–1

)

X M¢ (

e.m

.u. m

ol–1

)

Figure 5 | Dynamic magnetic data and magnetization relaxation time data for compound 2. a,b, Temperature dependence of the out-of-phase (xM′ ′, a) and

in-phase (xM′, b) components of the a.c. magnetic susceptibility measured in a 10 Oe a.c. field oscillating at the indicated frequencies, under zero d.c. field. c,

Temperature dependence of the magnetic relaxation time t under zero d.c. field is shown as ln(t) versus T21, as obtained from data collected in temperature

(filled circles) and frequency (open square) variation regimes. The values for the two lowest temperatures (filled squares) were obtained from time relaxation

measurements of the d.c. magnetization assuming a monomodal distribution of the characteristic relaxation rate. The straight line is a fit to the Arrhenius

relation, giving a thermal energy barrier for the relaxation of D¼ 142+7 K and a pre-exponential factor t0¼ (3+2)× 10212 s.

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1494 ARTICLES

NATURE CHEMISTRY | VOL 4 | DECEMBER 2012 | www.nature.com/naturechemistry 1015

© 2012 Macmillan Publishers Limited. All rights reserved.

anisotropic uranium ions is a very promising route in the quest forbetter performing single-molecule magnets.

MethodsSynthesis of complex 1 is described in Supplementary Section S2.1.

The proton NMR spectrum of solutions of 1 in pyridine displays features similarto those observed for the K and Rb adducts, in agreement with the retention of thetetranuclear structure. The Fourier transform infrared (FTIR) spectrum of 1 in KBrpellets (Supplementary Fig. S1) contains a band at 756 cm21 assigned to U–Ostretches that are weakened with respect to the uranyl(VI) analogue [UO2(salen)(Py)]complex (asymmetric U–O stretch at 892 cm21)23. These data support thepentavalent oxidation state of the isolated compound.

2. A dark brown solution of Cp*2Co (53.5 mg, 0.162 mmol, 1 equiv.) in pyridine(1 ml) was added under stirring to give a bright orange solution of [UO2(salen)(py)](100 mg, 0.162 mmol, 1 equiv.) in pyridine (2 ml), resulting in a dark greensolution, which was stirred for 1 h. The dark green solution was filtered, and asolution of Mn(NO3)2 (14.5 mg, 0.081 mmol, 0.5 equiv.) in pyridine (5 ml) wasadded dropwise to the filtrate under stirring, resulting in the precipitation of a darkviolet powder. The suspension was stirred for 3 h at room temperature, and the darkviolet precipitate was filtered out and washed with pyridine (10 × 1.5 ml) and driedthoroughly under vacuum to yield 82 mg of a violet powder of[{[UO2(salen)]2Mn(Py)3}6] (0.010 mmol, 74%).

Elemental analysis (%) calculated for [{[UO2(salen)]2Mn(Py)3}6](C282H258N42Mn6O48U12 8189.38 g mol21) C 41.36, H 3.18 and N 7.18, found C41.02, H 3.18 and N 7.08.

Crystals suitable for X-ray diffraction were obtained using a slow diffusionmethod, as described in Supplementary Section S2.2.

The FTIR spectrum in KBr pellets of X-ray quality crystals of 2 prepared by theslow diffusion method is identical to that for the bulk dark violet microcrystallinepowder. The spectrum shows similar features to 1 with a band at 752 cm21 assignedto uranyl(V) U–O stretches (Supplementary Fig. S2). Elemental analysis andmagnetic data (see below) also confirm the formula of the complex and that thesame species is obtained using either method (slow diffusion and direct reaction).

Crystallographic data were collected using a Oxford Diffraction Xcalibur-Skappa geometry diffractometer (Mo-Ka radiation, graphite monochromator, l ¼0.71073 Å) and have been deposited in the Cambridge Structural Database as CCDC871784 (1) and CCDC 871785 (2).

Temperature-dependent d.c. magnetic susceptibility data of the U12Mn6 wheelwere collected from 2 to 300 K at different fields up to 7 T, after zero-field coolingfrom room temperature. The raw experimental data were corrected by subtractingthe calculated diamagnetic contribution and a temperature-independentmagnetization term, Mimp¼ 6.7 × 1023mB, as described in SupplementarySection S5. To characterize the relaxation of the magnetization at low temperature,a.c. magnetic susceptibility measurements were performed on polycrystallinesamples in a 10 Oe a.c. field oscillating at a frequency f varying between 18 and9,887 Hz. Data were collected either as a function of temperature T for a given f(Fig. 5) or by sweeping f at constant temperature (Supplementary Figs S7–S9).

The characteristic relaxation time t(T) can be estimated from the inverse of thedriving field angular frequency, v¼ 2pf, at the peak temperature of the xM

′ ′ curves.Alternatively, t(T) can be determined by fitting a.c. susceptibility isothermsmeasured as a function of v to a generalized Debye model providing the averagerelaxation time and a parameter a, determining the width of the distributionfunction of relaxation times (Supplementary Information). The values obtained fora suggest a more complex relaxation scenario than in transition-metal SMMs, with awide distribution of relaxation times. The results obtained are shown in Fig. 5. Thevalue corresponding to the lowest temperatures (T¼ 4.5 K and 5 K) were obtainedby fitting to a single stretched-exponential behaviour the time dependence of the d.c.magnetization measured with the SQUID, giving t¼ 140 s for T¼ 4.5 K andt¼ 30 s for T¼ 5 K. Additional details on magnetic measurements are providedin the Supplementary Information.

Received 10 May 2012; accepted 4 October 2012;published online 11 November 2012

References1. Wernsdorfer, W., Aliaga-Alcalde, N., Hendrickson, D. N. & Christou, G.

Exchange-biased quantum tunnelling in a supramolecular dimer of single-molecule magnets. Nature 416, 406–409 (2002).

2. Gatteschi, D., Sessoli, R. & Villain, J. Molecular Nanomagnets (Oxford Univ.Press, 2006).

3. Affronte, M. et al. Linking rings through diamines and clusters: exploringsynthetic methods for making magnetic quantum gates. Angew. Chem. Int. Ed.44, 6496–6500 (2005).

4. Milios, C. et al. A record anisotropy barrier for a single-molecule magnet. J. Am.Chem. Soc. 129, 2754–2755 (2007).

5. Bogani, L. & Wernsdorfer, W. Molecular spintronics using single-moleculemagnets. Nature Mater. 7, 179–186 (2008).

6. Ako, A. et al. A ferromagnetically coupled Mn-19 aggregate with a recordS¼ 83/2 ground spin state. Angew. Chem. Int. Ed. 45, 4926–4929 (2006).

7. Ishikawa, N., Sugita, M., Ishikawa, T., Koshihara, S. & Kaizu, Y. Lanthanidedouble-decker complexes functioning as magnets at the single-molecular level.J. Am. Chem. Soc. 125, 8694–8695 (2003).

8. Rinehart, J. D., Fang, M., Evans, W. J. & Long, J. R. Strong exchange andmagnetic blocking in (N2)3–-radical-bridged lanthanide complexes. NatureChem. 3, 538–542 (2011).

9. Blagg, R. J., Tuna, F., McInnes, E. J. L. & Winpenny, R. E. P. Pentametalliclanthanide-alkoxide square-based pyramids: high energy barrier for thermalrelaxation in a holmium single molecule magnet. Chem. Commun. 47,10587–10589 (2011).

10. Blagg, R. J., Muryn, C. A., McInnes, E. J. L., Tuna, F. & Winpenny, R. E. P. Singlepyramid magnets: Dy5 pyramids with slow magnetic relaxation to 40 K. Angew.Chem. Int. Ed. 50, 6530–6533 (2011).

11. Rinehart, J. D., Fang, M., Evans, W. J. & Long, J. R. A (N2)3–-radical-bridgedterbium complex exhibiting magnetic hysteresis at 14 K. J. Am. Chem. Soc. 133,14236–14239 (2011).

12. Rinehart, J. D., Meihaus, K. R. & Long, J. R. Observation of a secondary slowrelaxation process for the field-induced single-molecule magnet U(H2BPz2)3.J. Am. Chem. Soc. 132, 7572–7573 (2010).

13. Antunes, M. A. et al. [U(Tp(Me2))2(bipy)]þ: a cationic uranium(III) complexwith single-molecule-magnet behavior. Inorg. Chem. 50, 9915–9917 (2011).

14. Magnani, N. et al. Magnetic memory effect in a transuranic mononuclearcomplex. Angew. Chem. Int. Ed. 50, 1696–1698 (2011).

15. Mills, D. et al. A delocalized arene-bridged diuranium single-molecule magnet.Nature Chem. 3, 454–460 (2011).

16. Mazzanti, M. Uranium memory. Nature Chem. 3, 426–427 (2011).17. Magnani, N. Superexchange coupling and slow magnetic relaxation in a

transuranium polymetallic complex. Phys. Rev. Lett. 104, 197202 (2010).18. Lam, O. P., Heinemann, F. W. & Meyer, K. Activation of elemental S, Se and Te

with uranium(III): bridging U–E–U (E¼ S, Se) and diamond-core complexesU–(E)2–U (E¼O, S, Se, Te). Chem. Sci. 2, 1538–1547 (2011).

19. Spencer, L. P. et al. M. Cation–cation interactions, magnetic communication,and reactivity of the pentavalent uranium ion [U(NtBu)2]þ. Angew. Chem. Int.Ed. 48, 3795–3798 (2009).

20. Kozimor, S. A., Bartlett, B. M., Rinehart, J. D. & Long, J. R. Magnetic exchangecoupling in chloride-bridged 5f–3d heterometallic complexes generated viainsertion into a uranium(IV) dimethylpyrazolate dimer. J. Am. Chem. Soc. 129,10672–10673 (2007).

21. Nocton, G., Horeglad, P., Pecaut, J. & Mazzanti, M. Polynuclear cation–cationcomplexes of pentavalent uranyl: relating stability and magnetic properties tostructure. J. Am. Chem. Soc. 130, 16633–16645 (2008).

22. Chatelain, L., Mougel, V., Pecaut, J. & Mazzanti, M. Chem. Sci. 3,1075–1079 (2012).

23. Mougel, V., Horeglad, P., Nocton, G., Pecaut, J. & Mazzanti, M. Stablepentavalent uranyl species and selective assembly of a polymetallic mixed-valenturanyl complex by cation–cation interactions. Angew. Chem. Int. Ed. 48,8477–8480 (2009).

24. Rosen, R. K., Andersen, R. A. & Edelstein, N. M. [MeC5H4)3U]2 [m-1,4-N2C6H4]—a bimetallic molecule with antiferromagnetic coupling between theuranium centers. J. Am. Chem. Soc. 112, 4588–4590 (1990).

25. Kiplinger, J. L. et al. Actinide-mediated cyclization of 1,2,4,5-tetracyanobenzene:synthesis and characterization of self-assembled trinuclear thorium anduranium macrocycles. Angew. Chem. Int. Ed. 45, 2036–2041 (2006).

26. Sigmon, G. E. & Burns, P. C. Rapid self-assembly of uranyl polyhedra into crownclusters. J. Am. Chem. Soc. 133, 9137–9139 (2011).

27. Evans, W. J., Kozimor, S. A. & Ziller, J. W. Molecular octa-uranium rings withalternating nitride and azide bridges. Science 309, 1835–1838 (2005).

28. Nocton, G., Pecaut, J. & Mazzanti, M. A nitrido-centered uranium azido clusterobtained from a uranium azide. Angew. Chem. Int. Ed. 47, 3040–3042 (2008).

29. Biswas, B., Mougel, V., Pecaut, J. & Mazzanti, M. Base-driven assembly of largeuranium oxo/hydroxo clusters. Angew. Chem. Int. Ed. 50, 5744–5747 (2011).

30. Le Borgne, T., Riviere, E., Marrot, J., Girerd, J. J. & Ephritikhine, M. Synthesis,crystal structure, and magnetic behavior of linear M2(II)–U(IV) complexes(M¼Co, Ni, Cu, Zn). Angew. Chem. Int. Ed. 39, 1647–1649 (2000).

31. Arnold, P. L., Patel, D., Wilson, C. & Love, J. B. Reduction and selective oxogroup silylation of the uranyl dication. Nature 451, 315–318 (2008).

32. Monreal, M. J., Carver, C. T. & Diaconescu, P. L. Redox processes in a uraniumbis(1,1′-diamidoferrocene) complex. Inorg. Chem. 46, 7226–7228 (2007).

33. Krot, N. N. & Grigoriev, M. S. Cation–cation interaction in crystalline actinidecompounds. Russ. Chem. Rev. 73, 89–100 (2004).

34. Arnold, P. et al. Single-electron uranyl reduction by a rare-earth cation. Angew.Chem. Int. Ed. 50, 887–890 (2011).

35. Mougel, V., Horeglad, P., Nocton, G., Pecaut, J. & Mazzanti, M. Cation–cationcomplexes of pentavalent uranyl: from disproportionation intermediates tostable clusters. Chem. Eur. J. 16, 14365–14377 (2010).

ARTICLES NATURE CHEMISTRY DOI: 10.1038/NCHEM.1494

NATURE CHEMISTRY | VOL 4 | DECEMBER 2012 | www.nature.com/naturechemistry1016

© 2012 Macmillan Publishers Limited. All rights reserved.

36. Mougel, V., Biswas, B., Pecaut, J. & Mazzanti, M. New insights into the acidmediated disproportionation of pentavalent uranyl. Chem. Commun. 46,8648–8650 (2010).

37. Arnold, P. L. et al. Strongly coupled binuclear uranium–oxo complexes fromuranyl oxo rearrangement and reductive silylation. Nature Chem. 4,221–222 (2012).

38. Milios, C. J. et al. Toward a magnetostructural correlation for a family of Mn6SMMs. J. Am. Chem. Soc. 129, 12505–12511 (2007).

39. Arnold, P. L., Patel, D., Blake, A. J., Wilson, C. & Love, J. B. Selective oxofunctionalisation of the uranyl ion with 3d metal cations. J. Am. Chem. Soc. 128,9610–9611 (2006).

40. Boskovic, C. et al. Single-molecule magnets: novel Mn8 and Mn9 carboxylateclusters containing an unusual pentadentate ligand derived from pyridine-2,6-dimethanol. Inorg. Chem. 41, 5107–5118 (2002).

41. Nocton, G. et al. Synthesis, structure, and bonding of stable complexes ofpentavalent uranyl. J. Am. Chem. Soc. 132, 495–508 (2010).

42. Arnold, P. L. et al. Uranyl oxo activation and functionalization by metal cationcoordination. Nature Chem. 2, 1056–1061 (2010).

43. Gatteschi, D. Molecular magnetism—a basis for new materials. Adv. Mater. 6,635–645 (1994).

44. Sessoli, R., Gatteschi, D., Caneschi, A. & Novak, M. A. Magnetic bistability in ametal-ion cluster. Nature 365, 141–143 (1993).

45. Sessoli, R. et al. High-spin molecules—[Mn12O12(O2Cr)16(H2O)4]. J. Am. Chem.Soc. 115, 1804–1816 (1993).

46. Thomas, L. et al. Macroscopic quantum tunnelling of magnetization in a singlecrystal of nanomagnets. Nature 383, 145–147 (1996).

47. Gatteschi, D. & Sessoli, R. Quantum tunneling of magnetization and relatedphenomena in molecular materials. Angew. Chem. Int. Ed. 42, 268–297 (2003).

48. Stamatatos, T. C. et al. ’Switching on’ the properties of single-moleculemagnetism in triangular manganese(III) complexes. J. Am. Chem. Soc. 129,9484–9499 (2007).

49. Stamatatos, T. C., Teat, S. J., Wernsdorfer, W. & Christou, G. Enhancing thequantum properties of manganese-lanthanide single-molecule magnets:

observation of quantum tunneling steps in the hysteresis loops of a {Mn12Gd}cluster. Angew. Chem. Int. Ed. 48, 521–524 (2009).

50. Rinck, J. et al. An octanuclear [Cr4(III)Dy4(III)] 3d–4f single-molecule magnet.Angew. Chem. Int. Ed. 49, 7583–7587 (2010).

51. Papatriantafyllopoulou, C., Wernsdorfer, W., Abboud, K. A. & Christou, G.Mn21Dy cluster with a record magnetization reversal barrier for a mixed 3d/4fsingle-molecule magnet. Inorg. Chem. 50, 421–423 (2011).

AcknowledgementsThe authors acknowledge support from the Commissariat a l’Energie Atomique, Directionde l’Energie Nucleaire, RBPCH programme and by the ‘Agence Nationale de la Recherche’,(ANR-10-BLAN-0729). The authors also thank F. Jacquot and L. Dubois for support andsuggestions regarding the magnetic measurements, A. De Geyer for recording the PXRDdiffractogram, N. Magnani, P. Santini and S. Carretta for useful discussions on theinterpretation of the magnetic data.

Author contributionsV.M. carried out the synthesis experiments, measured the d.c. magnetic data and analysedthe experimental data. L.C. performed the preliminary experiments. R.C., E.C. and J.C.G.collected and analysed the magnetic measurement data and created the magnetic model.J.P. and V.M. carried out X-ray single-crystal structure analyses. M.M. originated the centralidea, coordinated the work and analysed the experimental data. M.M., V.M. and R.C. wrotethe manuscript.

Additional informationSupplementary information and chemical compound information are available in theonline version of the paper. Reprints and permission information is available online athttp://www.nature.com/reprints. Correspondence and requests for materials should beaddressed to M.M.

Competing financial interestsThe authors declare no competing financial interests.

NATURE CHEMISTRY DOI: 10.1038/NCHEM.1494 ARTICLES

NATURE CHEMISTRY | VOL 4 | DECEMBER 2012 | www.nature.com/naturechemistry 1017

© 2012 Macmillan Publishers Limited. All rights reserved.