Keywords

1 Structure and Bonding

When one of us was a postgraduate at Imperial College, the physical appearance of the Structure and Bonding series was part of the reason I started to read it. They were thin, approachable volumes with the title and volume number in very large font at the top of the friendly green front cover. The reason I read more than one volume was that the content surveyed modern inorganic chemistry quickly and authoritatively, saving me a great deal of time. I found much that influenced my eventual research from these thin volumes, for example, reviews on polymetallic compounds [1, 2] and their magnetic properties [3] and reviews on the seemingly incomprehensible magnetic properties of high-spin cobalt(II) [4]. And then when working in Texas on gold clusters, I spent a great deal of time pouring over a lovely article by Schmid on how to build larger clusters [5] and an article by the editor of this volume on theoretical treatments of the bonding in such clusters [6]. The volume containing that second article also contained a fascinating review by the late Olivier Kahn concerning how to design magnetic molecules with predictable properties [7]; this review hugely influenced what I chose to do once I had an independent position. The juxtaposition of the review I needed to read preceding a review that pointed in a new direction was always part of the charm of S&B. It is one of the losses of the computer age that now we always find exactly the article we want quickly, and it is much more difficult to become productively distracted.

It was therefore with a great deal of pleasure that I accepted Peter Day’s invitation to edit a volume of Structure and Bonding, on Single-Molecule Magnets and Related Phenomena [8]. The authors of the various chapters wrote excellent reviews, two are officially “highly cited”, and the volume has received over 1,200 citations in total. It was also a great pleasure when the volume was updated in 2015 under the excellent editorship of Prof. Song Gao [9].

This short perspective on molecular nanomagnets (and related phenomena) has been co-written with Dr. Rodolphe Clérac – who contributed superb reviews to both of the previous volumes of Structure and Bonding.

2 Importance and Significance of Single-Molecule Magnets and Related Phenomena

The discovery in the early 1990s that a single molecule could retain magnetisation in the absence of a magnetic field was extremely exciting [10, 11]. At the time it was hoped that such molecules could revolutionise information storage; most magnetic information is stored in arrays of magnetic nanostructures, and, in 1993, a molecule was around 105 times smaller than the smallest feature in use. Sadly, in the first decade after the initial discovery, progress towards any technological applications was very slow.

The major challenge was the temperature at which these novel materials – rapidly named “single-molecule magnets” (SMMs) – operated. In the best examples, an extensive family of {Mn12} cages [12], the energy barrier to magnetic relaxation (U eff) is around 70 K. This means that the magnetic information could only be stored at temperatures that require liquid helium for cooling. Commonly, such a temperature is considered to be too low for any real application, but it is important to understand that if extraordinary behaviour or performance of SMM-based devices could be discovered, then low temperatures would not be a problem; for example, liquid He is used to cool the superconductor coils in modern NMR spectrometers and magnetic resonance imagers.

The energy barrier arises from a combination of a high-spin ground state (S = 10 for {Mn12}) and significant axial anisotropy of that ground state, normally reported as the axial zero-field splitting parameter D. For integer spins U eff = |D|S 2, if D is negative, which conventionally is the sign chosen for easy-axis anisotropy (with H = DS z 2). Until 2003, this largest energy barrier observed did not change very much. By contrast, the number of techniques available routinely for studying the magnetic behaviour increased dramatically. In the 1990s, characterisation of SMMs involved low-temperature direct current (d.c.) susceptibility measurements, and, as the temperatures involved were so low, there were few centres in the world that could discover SMMs [13, 14]. At the beginning of this century, the use of alternating current (a.c.) susceptibility measurements became common which became an alternative method to discover SMMs; the availability of a.c. susceptibility measurements led to a huge increase in the number of reported SMMs. High-frequency and high-field EPR spectroscopy became more common and even the use of other spectroscopic techniques such as magnetic circular dichroism (MCD) spectroscopy [15]. The use of MCD on a solution of a {Mn12} cage proved that the slow relaxation was due to individual molecules rather than any intermolecular interaction.

Two new groups of compounds were discovered in the early 2000s, which both offered higher energy barriers for magnetic relaxation. Firstly, one-dimensional (1D) systems were discovered in 2001 to exhibit slow dynamics of their magnetisation [16] as predicted in 1963 by R. J. Glauber in the frame of the Ising model [17]. These compounds were named single-chain magnets (SCMs) [18] by analogy of their properties to SMMs.

In contrast to SMMs for which the slow relaxation of the magnetisation is the signature of isolated anisotropic complexes (vide supra), the magnetisation dynamics in SCMs arises from the magnetic interactions between anisotropic repeating units along a single chain [19, 20]. As a consequence, the energy barriers found were immediately higher than for the SMMs as both exchange and anisotropy energies contribute to them. Because intrachain magnetic interactions are easier to control experimentally than the intrinsic magnetic anisotropy in SMMs, SCMs appear to be a promising alternative for applications. Nevertheless, the physics of the SCMs is still much less understood than for SMMs; it is particularly intriguing [17, 20] as it often falls between physics that can be assigned to single molecules and physics due to extended lattices.

The second group was based on monometallic complexes of the lanthanides, and there have been huge developments here since 2006; these are covered in the next section.

While technological applications were not forthcoming for SMMs, the field required chemists and physicists to work closely together to understand the physics of existing systems and to then synthesise new compounds to test the new theories developed. SMMs lie at the boundary of quantum systems, and it became possible to perform novel studies of quantum phase interference and parity effects [21] and other novel quantum physics [22]. The close collaborations also created an influx of new ideas around the physics of low-dimensional magnetic materials, including proposals to use molecule-based nanomagnets in quantum computing [23] and in magnetic cooling [24]. These ideas have largely inspired much of the work performed since 2006. The strong interactions between chemists and physicists are probably the most important outcome of the first two decades of studies of molecule-based nanomagnets.

3 Major Developments from 2006 to 2015

It is somewhat dangerous to pick out the major developments over the last decade or so; we restrict ourselves to seven headings, recognising we will have missed something. We apologise for any offence caused.

3.1 Lanthanide Single-Molecule Magnets

Beyond argument, the major change between 2006 and 2015 is the change in the elements studied; to 2006, the vast majority of SMMs involved polymetallic cage complexes of Mn(III), with an occasional report of an SMM involving a cage complex of another 3d-metal [15].

A report from 2003 by Ishikawa and co-workers changed the focus by reporting that the U eff value for a terbium-phthalocyanine (Pc) sandwich complex was an order of magnitude higher than that for {Mn12} [25]. This has led to a huge amount of work studying the dynamic magnetism of lanthanide complexes, especially complexes of Dy(III) [26, 27]. The work coincided with the increase use of a.c. susceptibility as a method to study SMMs, and this has perhaps polluted the area a little; almost any Dy(III) complex can be made to show some slow relaxation of magnetisation if an expert is involved in designing the experiment.

The slow relaxation of the lanthanides arises because of their electronic structure. The strong spin-orbit coupling leads to ions with very high total angular momentum, J, and large magnetic moments, especially for the heavier lanthanides Dy(III), Ho(III), Tb(III) and Er(III). In the correct coordination geometries, and hence crystal fields, this can stabilise ground-state doublets with very high m J values. For Dy(III), the most studied 4f-ion in this context, the ground-state doublet is often m J  = ±15/2 with the other doublets at higher energy. This is a very similar energy spectrum to that observed for 3d-SMMs, except that we are discussing m J levels, not m S levels, and the splitting of the doublets can, in principle, be controlled by the symmetry of the crystal fields.

The highest U eff values now reported are 652 cm−1 for a derivative of {TbPc2} [28] and 585 cm−1 for Dy(III) doped into an yttrium alkoxide cage [29]. However, higher energy barriers have not led to the same increase in the “blocking temperature” T B, i.e. the temperature at which the magnetic properties of the system show a dependence on history. The highest value published is around 14 K, for a radical-bridged Tb(III) dimer [30].

The reason for the disparity is that the relaxation mechanisms involved in lanthanide SMMs are far more complex than for 3d-SMMs. This is probably caused by the very large thermal barrier; in 3d-SMMs, the U eff value is normally 60 cm−1 at the best, and hence other routes to relaxation do not really need to be considered to explain the dynamics. For lanthanides, other processes are clearly important, and a feature in the next few years will be the development of methods to measure, model and understand the relaxation processes in these new compounds.

The close relationship between the crystal field and the splitting of doublets in lanthanide complexes also allows challenges to be laid down to synthetic chemists in this area. For Dy(III) or Tb(III), the ideal crystal field is strongly axial, and the most axial crystal field that could be achieved is D ∞h. It is difficult to imagine feasible lanthanide complexes that could have such high symmetry, but the extraordinary developments of metal-organic chemistry have allowed low coordination number 4f-complexes to be made. Unfortunately, the only such complex to approach linearity, a two-coordinate Sm(II) complex, contains an ion which is diamagnetic at low temperature [31]. However, calculations suggest that if a two-coordinate Dy(III) complex could be made, the U eff values found would be double any thus far reported [31].

3.2 Organometallic and Low Coordination Number SMMs

Most lanthanide complexes are found in the +3 oxidation state, and they can exist in this oxidation state almost regardless of the ligands present. This allows ligands to be used that would create diamagnetic, or at least very low spin, complexes if used with d-block metals. Thus, it is possible to make SMMs involving cyclopentadienyl ligands [32] or bridging hydrides [33], or even bridging N2 3− ligands [30], if the paramagnetic centre involved is a 4f-ion. This opens up many more ligands, and perhaps more importantly has brought molecular magnetism to the attention of extraordinarily talented synthetic chemists. Magnetic studies are now being reported of lanthanide complexes in the +2 oxidation states due to the synthetic skills of the Evans group [34].

This has led in turn to studies of low-coordinate 3d-metal complexes as SMMs [35]. Perhaps, the most interesting is a two-coordinate Fe(I) complex which has a huge U eff value of 226 cm−1 [36], which is a little higher than the energy barrier in the equivalent two-coordinate Fe(II) complex. It is striking that this massive barrier has been seen in a linear compound, which has more recently been predicted to be the ideal geometry for 4f-SMMs [31].

The involvement of chemists capable of handling ragingly air-sensitive materials has also led to actinide SMMs, chiefly of uranium as both U(III) and U(V) [37] but also neptunium [38] and plutonium [39]. There are still too few examples known for it to be clear how far this work can extend. Understanding the magnetic properties of the 5f-ions could make a major contribution to understanding the electronic structure of these fascinating elements.

3.3 Single-Chain Magnets and Related Systems

While before 2006, the number of papers dedicated to SCMs or closely related systems was less than 30 [40], the last decade has seen the publication of more than 500 papers on this emerging field of research [20, 41–44]. Most of these reports are dedicated to the preparation of new SCMs for which serendipity plays a key role that contrasts with the designed SCM materials reported at the infancy of the field [16, 40]. Among this massive amount of results since 2006, it is worth emphasising the SCMs assembled from paramagnetic radical [16, 45–49] that promote strong magnetic interaction along the chain and thus often lead to the high energy barriers and blocking temperatures. This approach, trying to increase the intrachain interactions as much as possible, is definitely an interesting synthetic strategy to explore further in order to make SCMs at high temperatures, while attempting to organise one-dimensional coordination network of lanthanide ions or more generally weakly interacting spin carriers seems to be a dead end.

Unfortunately, this evolution towards the production of more and more SCM systems does not help to understand the tricky physics of these 1D systems that is much less developed than the chemistry and needs simple model compounds. While SCMs falling in the Ising limit (for which the magnetic anisotropy energy dominates the exchange energy) are now relatively well understood from a theoretical point of view [20], this is still not the case for more complicated materials in terms of spin and interaction topologies. In most of the reports, the characterisation of SCM properties relies mainly on the observation of the dynamics of the magnetisation by a.c. susceptibility, while detailed static (thermodynamic) and dynamic magnetic measurements are both required to prove unambiguously the SCM properties. The future developments of this field of research will necessary need the strong involvement of physicists working closer to chemists to understand the fascinating physics behind these materials.

Another interesting development in this field was to study the influence of the interchain magnetic interactions on the magnetisation dynamics of these systems. As suggested by Sessoli [50] commenting the work of Ishida and co-workers on a radical CoII chain system [51], the intrinsic magnetisation dynamics of SCM units might induce magnetisation hysteresis loop with very large coercivity in 3D magnetically ordered materials. Indeed, Coulon, Miyasaka and co-workers demonstrated that the magnetisation dynamics of SCM is preserved for materials showing a 3D magnetic order, at least when the interchain couplings are weaker than the intrachain interactions [52, 53]. This conclusion was drawn for antiferromagnets and later for canted antiferromagnets [54], which also exhibit large M vs. H hysteresis loops (like classical magnets) due to the presence of SCMs in antiferromagnetic interactions in the crystal packing. These reports point towards the possibility to design new materials with hard magnet properties at high temperature incorporating SCMs but also any units possessing an intrinsic slow magnetisation dynamics like SMMs [52]. Detailed magnetic characterisation (static and dynamic) involving the application of a d.c. magnetic field are now essential to fully understand if a system is an SMM or SCM or possesses a magnetically ordered ground state.

3.4 SMMs on Surfaces and on the Route to Spintronic Devices

In 2006, the first experiments on deposition of SMMs on surfaces were being reported, and this preliminary work looked interesting but clearly had problems [55]. The chief problem was that {Mn12}, the prototype SMM, decomposed when deposited on any surface.

In the following decade, two SMMs have been studied which are stable on surfaces which led to the idea that SMMs could become components of future molecular spintronic devices [56]. The two most studied SMMs in this context are derivatives of an {Fe4} stars [57], where beautiful X-ray MCD experiments show that the SMM properties are retained when the molecules are bound to surfaces [58]. Unfortunately, {Fe4} does not have a very high U eff, but clearly this work shows that SMMs could be placed on surfaces without destroying their properties.

The other molecules studied are the derivatives of {TbPc2} where some astonishing results have been reported [59]. These reports show a real promise for esoteric spintronic devices into the future. The most extraordinary results have come from the team of Ruben and Wernsdorfer and have included reports of molecular spin valves [60] and single-molecule transistors [61]. Perhaps, the most extraordinary work involves driving Rabi oscillations in the nucleus of a single Tb(III) ion [62]. This is revolutionary work. It is interesting that such results mirror activities in IBM laboratories, where studies are being pursued on the magnetism of individual metal atoms deposited by scanning tunnelling microscopes [63]. These results point towards devices that do not presently exist, i.e. they are not just a scaled-down version of existing technology. It is in such disruptive new devices that molecular nanomagnets could have impact and could justify liquid helium cooling.

3.5 Molecular Nanomagnets for Magnetic Cooling

The research in the 1990s on 3d-metal cages as SMMs produced a very large number of molecules with high-spin ground states. Unfortunately, not all of these molecules had significant anisotropy of the spin in the ground state, and hence quite a number of the highest-spin molecules were not SMMs. One such molecule is an {Fe14} cage made by the McInnes group [64]. This molecule has an S = 25 ground state and vanishingly small anisotropy. These parameters are ideal for use as molecular magnetic coolants, using the magnetocaloric effect (MCE).

The MCE is intrinsic to any paramagnet and occurs because the entropy of the system changes on magnetisation or demagnetisation [65]. On magnetisation, the system becomes more ordered; when an external magnetic field is switched off, this additional order is lost and hence the magnetic entropy goes up. If the system is ordered, then that additional entropy is gained by absorbing lattice phonons. This cools the system. The MCE is greatest for isotropic systems and is not so great for SMMs [24].

The observation of significant MCE for molecular nanomagnets has spawned a huge number of papers [66]. As with a.c. susceptibility measurements on Dy(III) complexes, it is quite straightforward, albeit expensive in time and liquid helium, to measure the MCE on isotropic polymetallic cage complexes. Unfortunately, the end result of all these studies is probably that the most important molecular parameter is the percentage, by weight of the isotropic metal present in the material studied; as a result, one of the highest MCE values reported is for [Gd(OH)(CO3)] n , which is not very far from the gadolinium oxides already used for this application [67]. However, some beautiful new physics has been seen, including signatures of quantum critical points in direct studies of the magnetic cooling by a {Gd7} disc [68].

3.6 Molecular Nanomagnets for Quantum Information Processing

The original idea by Leuenberger and Loss to use crystals of {Mn12} to perform quantum computation [23] has been heavily cited, but comparatively little progress has been made to implement the experiments proposed. This may be because they would be technically extremely demanding.

A later proposal by the Loss group to use molecular nanomagnets with S = 1/2 ground states for quantum information processing has led to much more new chemistry [69]. The idea is that an S = 1/2 molecule is a two-level system (m S  = +1/2 and m S  = −1/2) and could function as a quantum bit, invariably abbreviated as qubit. There are many molecules that could so function; one family that has been extensively studied to this end is {Cr7Ni} rings [70]. One question about using molecular nanomagnets in this context is the phase memory time T m (also called the coherence time and, confusingly, the decoherence time), which is the time the phase of the information is retained before it decays. Pulsed electron paramagnetic resonance (EPR) spectroscopic studies were reported in 2007 [71] that T m is around 3 μS if ligands are deuterated but is far shorter if the ligands have natural abundance 1H presence. This suggested that a major cause of decoherence was electron-nucleus hyperfine interactions, and since 2007 several groups have worked to maximise T m by minimising the number of H atoms present in molecules and by using simple complexes with rigid ligands. Such approaches have led to T m values approaching 70 μS, for example, in copper dithiolate complexes [72]. This is extremely promising.

Other implementation strategies have been examined. Probably the most advanced involves simple organic radicals and has been pursued by the Takui group, who has reported a C-NOT gate using an organic di-radical [73]. The Aromí group has also proposed using the Kramer’s doublets of hetero-bimetallic lanthanide complexes for two-qubit gates [74]. These two approaches have very different possible implementations but both involve g-engineering. In the organic di-radicals, the very high resolution possible for EPR spectroscopy of organic systems allows the different radicals to be addressed even though the g-values are quite close. In the dilanthanide case, the g-values are vastly different. Very recently, g-engineering has also been used for d-block spin centres, combining the {Cr7Ni} rings with Cu(II) complexes [75]. Implementation of another universal quantum gate, the √SWAP gate, was proposed using di-radicals within a polyoxometalate [76].

These proposals and experiments show a great deal of promise for this approach, but it is now becoming essential for the area to retain credibility for a simple algorithm to be performed using molecular electron spins. It is also noticeable that all the studies being reported involve simple paramagnets, and not SMMs. However, it was the collaborations established while SMMs were being explored that has led to this somewhat different science of paramagnetic compounds.

3.7 Photomagnets

In the past decade, important research activity has been devoted to photoactive magnetic systems and in particular to magnets that can be controlled or induced by light irradiation, i.e. photomagnets. Three different kinds of photomagnets should be highlighted here: (1) photo-SMMs, (2) photo-SCMs and (3) photomagnets for which a 3D magnetic order is photoinduced. Interestingly, all these systems rely on two different photoactive mechanisms, the well-known spin-crossover properties of the FeII metal ion and the metal-metal electron transfer processes in cyanido-bridged metal ion assemblies (Prussian blue analogues). In 2012, Oshio and co-workers reported the first photoinduced SMM in an hexanuclear Fe/Co Prussian blue analogue [77], and shortly after in 2013, Long, Smith and co-workers discovered the same photomagnetic properties in mononuclear Fe(II) spin-crossover complexes [78, 79]. Photoinduced SCM properties were first demonstrated by Sato, Oshio and co-workers in one-dimensional Fe/Co and Fe/Fe Prussian blue analogues thanks to the intrachain electron transfer between Co and Fe metal ions through cyanido bridges [80–82] or to the spin-crossover properties of the Fe(II) site [83]. Related to these breakthrough results, it should be mentioned that Bogani and co-workers suggested that the magnetisation dynamics of SCM systems could be photo-controlled by domain-wall kickoff mechanism [84]. Even if this result requires further confirmation on different SCMs, it opens a completely new field of investigations with the ultimate goal to improve the optical switching methods for application.

In molecule-based materials, a photoinduced magnetic order was first discovered in 1996 by Hashimoto, Fujishima and co-workers in 3D Fe/Co Prussian blue analogues [85]; this system has subsequently been studied in great detail by Bleuzen, Verdaguer and co-workers [86]. Since then, only a few examples of photomagnets have been reported, most of them relying on the metal-metal electron transfer mechanism in Prussian blue analogues or related cyanido-based systems [86, 87]. In 2011, the discovery of the first photoinduced magnetic order based on a spin-crossover process is probably one of the most exciting results in the field of molecular magnetism for the past decade. By a judicious choice of the blocking ligand, Ohkoshi, Tokoro and co-workers were able to tune the ligand field of FeII metal ions in order to stabilise thermally and photoinduced spin-crossover processes at this site and at the same time incorporate this FeII unit into coordination networks that can promote magnetic interactions between spin carriers (in these materials, between FeII and NbIV through cyanido bridges) [88]. This inspiring result led to the design of new photomagnets modulating the FeII nitrogen-based ligands as reported by Sieklucka and co-workers [89] and Ohkoshi’s group that was able to implement chirality to a cyanido-based FeII/NbIV photomagnet inducing remarkable magneto-optical switching properties [90].

4 A Prediction of Future Developments

While reviewing progress over the previous 10 years is difficult, making predictions concerning future developments is largely impossible. There is one development that is needed, if the area of molecular nanomagnetism is to continue to thrive.

The field needs some form of device. The area arose from observations of magnetic information storage in individual molecules at very low temperatures [10, 11]. Recent developments have suggested that the temperature at which information could be stored in molecules could rise to temperatures achievable with liquid nitrogen cooling [30, 31], if we can understand relaxation processes in lanthanide SMMs and learn to control them. However, given the maturity of the information storage industry how technologically applicable this will be is debateable. The use of molecular electron spins in quantum information processing might be a more productive area, simply because it would not be competing with such mature technology. Again, however, there is a need to perform some simple algorithms simply to maintain credibility.

The MCE bubble may be coming to an end, simply because studies tend to lead back towards very simple compounds such as [Gd(OH)(CO3)]. However, possible applications of molecular MCE materials are much easier to see; for example, coating materials with magnetic coolants should be easier from molecular precursors than from inorganic solids. Again, the discovery of a useful device would make a massive difference to work in the area.

The extraordinary experiments reported by Ruben and Wernsdorfer and their co-workers [60–62] possibly show the most promise for exploitation, despite the fact that they are also the most technologically challenging. The justification for such an opinion is that such experiments cannot be performed by other means – the molecular nanomagnet is a key component, and there is no existing technology that would allow similar experiments at any temperature. The drawback of this approach is its lack of scalability; the method for making the devices is stochastic, and a major challenge must be to devise a method to make multiple equivalent single-molecule devices in a controlled and reproducible manner. Such an achievement could lead to implementation of molecular spintronic devices.