Coordination Clusters of 3d-Metals That Behave as Single-Molecule Magnets (SMMs): Synthetic Routes and Strategies

The area of 3d-metal coordination clusters that behave as Single-Molecule Magnets (SMMs) is now quite mature within the interdisciplinary field of Molecular Magnetism. This area has created a renaissance in Inorganic Chemistry. From the synthetic Inorganic Chemistry viewpoint, the early years of “try and see” exercises (1993–2000) have been followed by the development of strategies and strict approaches. Our review will first summarize the early synthetic efforts and routes for the preparation of polynuclear 3d-metal SMMs, and it will be then concentrated on the description of the existing strategies. The former involve the combination of appropriate 3d-metal-containing starting materials (simple salts with inorganic anions, metal cardoxylates, and pre-formed carboxylate clusters, metal phosphonates) and one or two primary organic ligands; the importance of the end-on azido group as a ferromagnetic coupler in 3d-metal SMM chemistry will be discussed. The utility of comproportionation reactions and the reductive aggregation route for the construction of manganese SMMs will also be described. Most of the existing strategies for the synthesis of SMMs concern manganese. These involve substitution of carboxylate ligands in pre-formed SMMs by other carboxylate or non-carboxylate groups, reduction procedures for the {Mn8IIIMn4IV} SMMs, spin “tweaking,” “switching on” SMM properties upon conversion of low-spin clusters into high-spin ones, ground-state spin switching and enhancing SMM properties via targeted structural distortions, the use of radical bridging ligands and supramolecular approaches. A very useful strategy is also the “switching on” of SMM behavior through replacement of bridging hydroxide groups by end-on azido or isocyanato ligands in clusters. Selected examples will be mentioned and critically discussed. Particular emphasis will be given on the criteria for the choice of ligands.


INTRODUCTION
Today there is a multi-billion euro industry concerning magnets due to their applications in many useful and modern technologies, such as televisions, credit/debit/ATM cards, motors, computer hard drives, switches and medical equipment, among others. Magnetism has its roots in Ancient Greece; in the province of Magnesia at central Greece, the first discovery of the attraction between lodestone and iron was made. This, however, is a matter of debate. The first observations were made near the city of "Magnesia" (close to the Sipylum mountain) in the old, Asian Minor Greek province of Lydia. This old city was near the modern city Manisa, close to Smirna (Izmir). Sophocles and other Ancient Greek writers reported on the "Lydian stone" in reference to the lodestone (Guimaraes, 2005). There is also a strong piece of evidence that Chinese sailors used the compass. The principles of Magnetism were established through the excellent work of Ampeére, Oersted, Faraday, Maxwell and Pierre Curie, just to name few legendary scientists who contributed in this field of Physics. The discovery of the electron, the seminal ideas of Lewis who wrote simple electronic structures for molecules and ions, and the development of quantum mechanics provided scientists with the appropriate tools for the in-depth understanding of the magnetic properties of the matter. Until ∼1960, the popular magnets consisted mainly of metallic alloys and ceramics, e.g., Alnico and ferrites.
Chemists entered the field in the late 50s developing the field of Magnetochemistry. They used experimentally determined magnetic moments and their variation with temperature to draw chemical conclusions and make structural proposals in molecular compounds. Famous examples include: (i) the distinction between the two stereochemistries (tetrahedral, square planar) in 4-coordinate Ni(II) complexes based on the value of the effective magnetic moment, and (ii) the determination by Bleaney and Bowers of the singlet-triplet gap in copper(II) acetate hydrate, before the crystallographic determination of its dinuclear [Cu 2 (O 2 CMe) 4 (H 2 O) 2 ] structure.
Molecular Magnetism (Kahn, 1993) came into the scientific scene in the early 80s. It is an interdisciplinary field in which chemists, physicists, material scientists, and chemical engineers, both experimentalists and theoreticians, actively participate. A NATO ASI in Italy in 1983 is widely considered as the beginning of the Molecular Magnetism era. The chemists presented magnetic properties of dinuclear, polynuclear, and polymeric molecular species, whereas the contribution from physicists was based on infinite atom-based arrays with a special focus on 1D systems which were popular at that time (mainly due to the availability of spin Hamiltonians which could be solved). The ASI, which gave the birth of a common scientific language between the two communities, convinced (a) chemists to dare to synthesize complex molecular compounds, and (b) physicists that it was possible to discover new magnetic phenomena working with molecule-based materials (Benelli and Gatteschi, 2015). The development of Molecular Magnetism has been tremendous since then. Milestones include-among others-the synthesis of molecular ferrimagnets, examples of molecular ferromagnets, the discovery that pure organic matter can order ferromagnetically, the attainment of room temperature molecular magnets, the development of spin-crossover systems, the discovery that the magnetic state of several molecular magnets can be photoinduced in a reversible manner, the isolation of systems exhibiting a coexistence of magnetic and other (e.g., conductivity) properties, the development of "magnetic nanowires" (Single-Chain Magnets) and the revolution with Single-Molecule Magnets (Coronado and Gatteschi, 2006;Coronado and Dunbar, 2009;Dunbar, 2012;Palacio, 2014;Benelli and Gatteschi, 2015). Some synthetic aspects of 3d-metal coordination clusters that behave as SMMs are presented in this Review.
An important milestone in the field of Molecular Magnetism was the discovery of the magnetic properties of complex [Mn III 8 Mn IV 4 O 12 (O 2 CMe) 16 (H 2 O) 4 ] (1) (Caneschi et al., 1991;Sessoli et al., 1993a), which indicated that this molecule can be considered as a small piece of manganese oxide whose further growth is inhibited by the presence of the acetate groups. The molecule possesses 12 spin centers; the 8 Mn III atoms have their spins up and the 4 Mn IV ones down and this coupling scheme result in a S = 10 spin ground state (8 × 2 -4 × 3/2 = 10). The impressive discovery was that, at very low temperatures, the magnetization relaxes slowly and the complex displays hysteresis (of molecular origin) like a tiny magnet. Most remarkably, a negative value for the axial zero-field splitting, D, results in the loss of the degeneracy of the corresponding M S sublevels, and so the M S = +10 and M S = −10 levels become lowest in energy. Since the selection rule is M S = ±1, an energy barrier of U = S 2 |D| results for "spin up" to "spin down" conversion in the ground state. In the case of 1, the S and D values are 10 and −0.5 cm −1 , respectively; these values give a U ≈ 50 cm −1 and this results in magnetic blocking below 4 K, as proved by magnetic hysteresis loops. The behavior of 1 gave the idea that information could in the future be stored as the magnetization direction in discrete molecules. Assuming that a diameter for each bit is ∼1 nm, a dense array of molecules might achieve a very high surface density (∼200,000 gigabits/in 2 ), and this is an advantage of several orders of magnitude over the existing magnetic alloy film technologies, a very important consequence for computer hard drives (Beltran and Long, 2005). The ability of a single molecule to behave as a tiny magnet represents another approach, the so-called "bottom-up" approach, to nanoscale magnetic materials, in which the need for smaller and smaller magnetic particles increases continuously (the so-called "topdown" approach). The "bottom-up" route provides this scientific area with many advantages of molecular chemistry, for example good solubility (and not colloid formation observed in magnetic nanoparticles), monodispersity, crystallinity, and avoidance of close contacts between the magnetic cores due to the intervention by an organic shell from the ligands (Bagai and Christou, 2009).
Polynuclear complexes (or coordination clusters) with properties similar to those of 1 are called Single-Molecule Magnets (SMMs); however, some scientists prefer the term Molecular Nanomagnets. In brief, SMMs are individual polynuclear molecules that function as magnets (i.e., they display magnetization hysteresis) below a characteristic blocking temperature, T B (which depends on the sweep rate of the magnetic field). T B is the highest temperature at which an SMM displays hysteresis in plots of magnetization (M) vs. magnetic field (H). However, since the T B value depends strongly on the sweep rate of H, comparisons of blocking temperatures in different SMMs should be done with caution (Woodruff et al., 2013). In the now classical book of Gatteschi, Sessoli, and Villain (Gatteschi et al., 2006), the authors define T B as the temperature at which the time for the magnetization relaxation is 100 s. This property arises from a simultaneous presence of a high-spin ground state, S, and a large and negative Ising (or easy-axis) type of magnetoanisotropy, as measured by the parameter D. This combination gives a significant energy barrier to magnetization reversal (U), the maximum value of which is given by S 2 |D| (see above) and (S 2 -1/4) |D| for integer and non-integer total ground state spins, respectively. However, quantum tunneling of the magnetization vector (QTM) (Friedman et al., 1996;Thomas et al., 1996) through the barrier via M S levels of higher energy leads to a smaller effective energy barrier (U eff ) than the calculated U. In addition to QTM, some SMMs display other interesting quantum properties such as quantum phase interference (Wernsdorfer and Sessoli, 1999;Lecren et al., 2005;Wernsdorfer et al., 2005), exchange-biased QTM (Wernsdorfer et al., 2002;Pinkowicz et al., 2015) and others with potential technological interest (Leuenberger and Loss, 2001;Affronte et al., 2007;Bogani and Wernsdorfer, 2008;Affronte, 2009;Urdampilleta et al., 2011). For all these reasons, SMMs are candidates for a number of specialized applications (Bagai and Christou, 2009). Excellent books Benelli and Gatteschi, 2015) and reviews (Gatteschi and Sessoli, 2003;Beltran and Long, 2005;Aromi and Brechin, 2006;Bagai and Christou, 2009;Wang et al., 2011;Milios and Winpenny, 2015) for the properties of transition-metal SMMs are available.
In the last 10 years or so, there has been strong evidences that the most important parameter when designing SMMs is the single-ion anisotropy which should be large. Thus the research focus has been shifted to lanthanides (Ln) and actinides (An) which, in general, have large anisotropies. In Ln(III) SMMs, the required electronic ground-state bistability is due to the [2J + 1] M J microstates within the ground term 2S+1 Lj which is spin-orbit-coupled (Woodruff et al., 2013). The advantages of working with An-based SMMs are their large magnetic anisotropy (like the Ln III ions) and the possibility for covalency (like the transition metal ions), advantages which, however, add some complexity (Meinhaus and Long, 2015). Very interesting reviews have been published on the SMM properties of dinuclear and polynuclear Ln(III) (Sessoli and Powell, 2009;Guo et al., 2011;Rinehart and Long, 2011;Habib and Murugesu, 2013;Woodruff et al., 2013;Zhang et al., 2013aZhang et al., ,b, 2015Liddle and van Slageren, 2015;Pedersen et al., 2015;Pointillart et al., 2017;Liu et al., 2018), An (Meinhaus and Long, 2015), 3d/4f (Polyzou et al., 2013;Feltham and Brooker, 2014;Sharples and Collison, 2014;Liu et al., 2015;Rosado Piquer and Sañudo, 2015), and organometallic Ln(III) and An (Layfield, 2014) SMMs.
Before going to the main "menu" of this Review, we would like to say few words about the so-called Single-Ion Magnets (SIMs), which is currently a "hot" research area. These are mainly mononuclear d-or f-metal complexes whose magnetization relaxes slowly not because of cooperative exchange phenomena (as in dinuclear and polynuclear SMMs), but due to the intrinsic properties of the metal ion under an appropriate ligand field. The origin of the barrier here is the magnetic anisotropy. Excellent reviews on transition- (Craig and Murrie, 2015;Bar et al., 2016;Frost et al., 2016), 4f-(Woodruff et al., 2013Gupta and Murugavel, 2018;Liu et al., 2018) and 5f-metal (Meinhaus and Long, 2015) SIMs are available in the literature. The very recent report of the SIM [(C ttt p ) 2 Dy][B(C 6 F 5 ) 4 ] (2) (Goodwin et al., 2017a,b;Guo et al., 2017) with a record T B value of ∼60 K (U eff = 1277 cm −1 in zero field) is a revolutionary result, and it predicts a brilliant future in this area; C ttt p is the 1,2,4-tri(tert-butyl)cyclopentadienide, i.e., {C 5 H 2 t Bu 3 -1,2,4} with t Bu = C(Me) 3 . It should be emphasized at this point that the SMM properties of many dinuclear and polynuclear Ln(III) complexes are in fact due to singleion properties and such molecules can be regarded as a collection of SIMs. This is a consequence of the weak or very weak Ln III . . . Ln III exchange interactions due to the wellknown internal character of the 4f orbitals which are severely shielded from the environment; however, most Ln III ions easily produce large anisotropy due to unquenched orbital moment.
The above mentioned developments in the areas of SMMs and SIMs created an explosive growth in synthetic Inorganic (both coordination and organometallic) Chemistry in the first quarter-century of this new "chapter" in the "book" of Molecular Magnetism. New reaction schemes and new types of metal complexes have been discovered. This Review covers the most important (a subjective opinion!) synthetic routes and strategies that have led to SMMs that belong to the family of 3dmetal coordination clusters. This means that synthetic aspects related to SIMs and other types of SMMs (4d-, 5d-, 4f-, 5f-, and d/f-metal-based) are not covered. The content of the Review is purely chemical and it is assumed that the readers have a basic knowledge of structural inorganic chemistry, as well as of the properties of SMMs and the methods of their study. Structural and magnetic descriptions will be confined to the necessary minimum. To avoid long synthetic descriptions, balanced chemical equations (written using molecular-and not ionic-formulae) will be used. In the text, we shall try to explain the synthetic rationale and philosophy behind the reactions with particular emphasis on the choice of reactants (ligands and metal-containing starting materials). Synthetic schemes, routes and strategies for the preparation of 3d-metal SMMs have appeared in several books, reviews, and papers along with detailed structural and magnetic descriptions; this is, however, the first review which is concentrated exclusively on the synthetic and reactivity chemistry of 3d-metal SMMs. Since there are thousands of papers describing the syntheses and properties of polynuclear 3d-metals SMMs, it is inevitable that we cannot cover all the existing literature. Thus, we express our apologies to the scientists whose excellent research will not be cited here.
The popular Harris notation (Coxall et al., 2000) will be used to describe the ligands' coordination modes. The Harris notation describes the ligation mode as X.Y 1 Y 2 Y 3 . . . Y n , where X is the total number of metal centers bound by the ligand, and each of Y refers to the number of metal ions attached to the different donor atoms. The order of the Y groups follows the well-known Cahn-Ingold-Prelog priority rules; therefore, for the most ligands reported in this work O comes before N. From time to time the traditional η/µ notation will be also used.

SYNTHESIS OF 3d-METAL SMMS: GENERAL CONSIDERATIONS
The 3d-metal systems that have been used in the preparation of SMMs are V(III), V(IV), Mn(II/III/IV), Mn(III), Mn(II/III), Mn(III/IV), Fe(II), Fe(III), Co(II), Co(II/III), Ni(II), and Cu(II). The most effective and well-studied SMMs are those containing high-spin Mn(III). The presence of high-spin octahedral Mn III atoms in molecules is valuable for making SMMs; the reason is that this 3d 4 , Jahn-Teller elongated ion has a large easy-axis anisotropy (negative D) which is desirable. Thus, many of the routes and strategies below involve examples from Mn chemistry.
All 3d-metal SMMs are dinuclear or polynuclear complexes. We are doing a parenthesis here to mention that the exchange coupling, J, is an important parameter to consider in the preparation of SMMs. This should be strong, because otherwise the spin ground state is not energetically isolated. Singlemolecule magnetism is a ground-state phenomenon, and thus in order SMM properties to be observed at a certain temperature, the ground state must be ideally occupied at this temperature. The "translation" of this is that the bridging ligand or ligands should propagate strong ferromagnetic exchange interactions between the metal centers resulting in high-spin ground states (large S values). However, large S values can also arise from competing antiferromagnetic interactions in some topologies that prevent ("frustrate") exactly antiparallel spin alignments. Organic ligands discussed in this Review are shown in Figure 1. Many of these ligands can provide monoatomic bridges which often promote ferromagnetic exchange interactions.
A great development in synthetic chemistry during the twentieth century has been the ability of organic chemists to make large, complicated molecules in a systematic, step by step controlled manner, based on the known behavior of carbon and well-established reaction patterns. Sometimes, three or more different parts of a molecule are synthesized separately before the whole species is assembled in two or three final steps. Unfortunately, this is not possible for the transitionmetal synthetic chemists, whose efforts are complicated by the non-predictable nature of the metal (e.g., variable oxidation state and coordination number), and the influence of reaction stoichiometry and solvent on product identity. For this reason, many synthetic schemes for SMMs are based on one-or twostep reactions. The lack of control in transition-metal chemistry has led to the adoption of the neologism "self-assembly" (Winpenny, 1999). The situation is becoming worse for the inorganic chemist who must crystallize her/his products (and SMM candidates) and determine their structures by singlecrystal X-ray crystallography; this is often a difficult task. On the contrary, organic chemists can use all the "arsenal" of NMR and mass spectroscopic methods (which have undergone an explosive growth in lase decades) to firmly establish the structures, with no need to crystallize the compounds.

THE YEARS OF INNOCENCE: "TRY AND SEE" EXERCISES
The discovery of the exciting magnetic properties of 1 ignited many efforts to synthesize dinuclear and polynuclear (coordination clusters) 3d-metal complexes with SMM properties. Most research groups, including our group, used simple 3d-metal "salts" with coordinating (e.g., Cl − , Br − , NO − 3 , SCN − , SO 2− 4 ,. . . ) or non-coordinating (e.g., ClO − 4 , PF − 6 , BF − 4 ,. . . ) inorganic anions and flexible polydentate organic ligands. Such reaction schemes permit a "conversation" between the preferred coordination geometry of the metal ion and the coordination mode of the ligand. If exclusive formation of 5or 6-membered chelating rings is avoided, the coordinative flexibility of the polydentate ligand is large. This allows the isolation of coordination clusters with a variety of unpredictable structures (and hopefully SMM properties), almost always incorporating further auxiliary bridging groups formed in situ, e.g., OH − , RO − , O 2− , depending on the solvent and the "pH" of the reaction medium. Organic ligands possessing one, two, three or more ionizable hydrogens exhibit better bridging capabilities, if fully or in part deprotonated (with the help of an external base). The fate of the inorganic anion, present in the 3d-metal starting material, depends on the reaction conditions. It can participate in the cluster (either as a ligand or a counterion) or it can be absent. If there is evidence (or suspicion) for the presence of anionic cluster species in solution, bulky cations, e.g., Bu n 4 N + , should be used to precipitate the product.
In a number of reaction schemes, the 3d-metal starting material contains a strongly basic anionic group, e.g., a carboxylate group (MeCO − 2 , PhCO − 2 ,. . . ) or a β-diketonate ion (acac − , bzac − , dbm − ,. . . ), where Me = methyl, Ph = phenyl, acacH = acetylacetone, bzacH = benzoylacetone, and dbmH = dibenzoylmethane. If the primary organic ligand LH x (x = 1,2,3,. . . ) is not in excess, then it can be deprotonated by the auxiliary organic anion while an amount of the latter remains in solution. This organic "blend, " which contains the deprotonated primary organic ligand and the anionic auxiliary organic group, leads often to dinuclear or polynuclear 3d-metal complexes which often exhibit interesting magnetic properties. This route is particularly useful when external bases for ligand deprotonation perplex the reaction, e.g., by precipitating amorphous hydroxido-or/and oxido-containing materials.
It would be naïve to believe that the above described "try and see" approach leads directly to SMMs. Considerable forethought and hindsight in the metal ions, their inorganic or auxiliary organic anions, primary organic ligands and reaction conditions (reaction ratio, solvent, "pH, " nature of external base, temperature, pressure, crystallization technique, etc.) are necessary to "switch on" or improve the SMM properties in the products.
A final point of general discussion is the importance of the presence of terminal ligands in the reaction media; these are necessary for the termination of a possible polymerization process which would lead to 1D, 2D, or 3D coordination polymers. In the chemistry of 3d-metal coordination clusters, this role is often fulfilled by the solvent molecules, the inorganic anions of the 3d-metal starting materials, the auxiliary organic ligands and the "chelating" part of the bridging organic ligands, or by the purposefully addition of a chelating ligand in the reaction systems.

Simple 3d-Metal Starting Materials With Inorganic Anions
Commonly used 3d-metal "salts" with inorganic anions include halides, nitrates, sulfates, perchlorates, etc. Ni(II) is promizing in the synthesis of SMMs due to its significant single-ion anisotropy. For example, the reaction between Ni(NO 3 ) 2 ·6H 2 O, 1,1,1-tris(hydroxymethyl)ethane (H 3 thme; G with × = 1 and R = Me in Figure 1) and NaOMe in MeOH gives complex [Ni 4 (H 2 thme) 4 (MeCN) 4 ](NO 3 ) 4 (3), Equation (1) (Moragues-Cánovas et al., 2004). The initial reaction solution was evaporated to give a solid residue that was extracted with MeCN to give blue crystals of the product. The cation has a cubane structure. Each Ni II atom is surrounded by two neutral oxygen atoms of H 2 thme − that act as terminal donors and three alkoxidolike deprotonated oxygen atoms from three ligands; thus, each monoanion behaves as a 3.311 ligand (AA in Figure 2); one MeCN molecule completes six coordination at each metal center. Intramolecular ferromagnetic Ni II . . . Ni II exchange interactions give an S = 4 ground state which is 40 cm −1 lower in energy from the first S = 3 excited state. Single-crystal magnetization vs. field (H) measurements reveal an SMM behavior below 0.5 K with fast magnetization relaxation due to QTM. The width of the hysteresis loop depends on the "history" of the sample, i.e., the starting magnetic state before varying the magnetic field.
Another member of the ligand family G (see Figure 1), but in Co chemistry this time, gives a structurally and magnetically interesting mixed-valence coordination cluster. The D parameter for the Co II single ion in a tetragonally distorted octahedral environment is positive, because the M = ±1/2 ground state is lower in energy than the M = ±3/2 state, these two states being the Kramers' doublets that derive from the 4 A 2g ground state in ideal D 4h symmetry due to spin-orbit coupling However, the orientation of the axial zero-field tensors on the Co II atoms, that are part of a cluster, may be favorable with respect to the molecular easy axis giving an overall negative D-value (Ising type anisotropy) and leading to SMM behaviors (Kostakis et al., 2012). The 1:1 reaction between Co(NO 3 ) 2 · 6H 2 O and 2-amino-2(hydroxymethyl)propane-1,3-diol (H 3 ahmpd; G with R = NH 2 in Figure 1) in H 2 O, followed by increase of the pH to 7.5 by addition of (Bu n 4 N)OH· 5H 2 O gives a solution; removal of water under vacuum and dissolution of the residue in MeOH leads to cluster [Co II 4 Co III 3 (Hahmpd) 6 (NO 3 ) 3 (H 2 O) 3 ](NO 3 ) 2 (4), Equation (2) (Ferguson et al., 2007). The yield is very low (∼5%) but the reaction is reproducible. The cation is disk-like with the octahedral metal ions approximately coplanar and bridged exclusively by the alkoxido groups of the doubly deprotonated ligands, which adopt the 4.3201 coordination mode (BB in Figure 2). One alkoxido oxygen atom of each ligand is µ 3 , bridging the central Co II to one outer Co II and one Co III , with the second alkoxido oxygen bridging one outer Co II to a Co III center; the amine group is bound to Co III . The coordination at each outer Co II is completed by a H 2 O molecule and a monodentate nitrato group. The complex is an exchange-biased SMM. For a well-isolated SMM, the first step in the hysteresis loop due to QTM should occur at H = 0 T. For 4, due to weak intermolecular interactions, this step is shifted to ±30 mT, suggesting a small antiferromagnetic exchange bias between cations of about 30 mT.  (Kitos et al., 2011), the reaction of Co(ClO 4 ) 2 · 6H 2 O and 2-acetylpyridine [(py)(me)CO; J in Figure 1] in the presence of (Bu n 4 N)OMe (∼1:1:1) in Me 2 CO at room temperature under aerobic conditions gave a brown solution, from which the green compound [Co II 6 Co III (OH) 6 (L) 6 ](ClO 4 ) 3 (5) was isolated in a moderate yield, Equation (3). The ligand L − which participates in the complex is the anion of 2-(pyridine-2-yl)pentane-2-ol-4-one, (py)(me)C(CH 2 COCH 3 )(O) − , formed in solution (in situ) through a metal ion-assisted, crossedaldol reaction in Me 2 CO under strongly basic conditions (see Figure 3). The disk-like cation of 5 constists of a central octahedral Co III atom linked to six peripheral Co II atoms (also octahedral) by six 3.3 hydroxido groups; the six Co II centers on the rim are held together by six 2.211 L − ligands (CC in Figure 2) and the oxygen atoms of the hydroxido groups. Each heptanuclear cation is a weak SMM but there are also intermolecular interactions (proven by crystallography) leading to a 3D architecture; this gives magnetization hysteresis at 5 K. 7 Co II (ClO 4 ) 2 · 6H 2 O + 6(py)(me)CO + 11 (Bu n 4 N)OMe When 3d-metal fluorides are used as starting materials in SMM chemistry, their insolubility in common organic solvents is a problem. In order to overcome this disadvantage, "melt" reactions are often used. For example, Collison, Brechin, and  Figure 2). Single-crystal magnetic studies at low temperatures performed on 6 (S = 4, D = −0.90 cm −1 ) revealed the time-and temperature-dependent magnetization hysteresis loops indicative of SMM behavior. The SMM parameters are U eff = 10.4 cm −1 and τ 0 = 3 × 10 −9 s. Below ∼0.2 K, there is strong evidence for the presence of QTM in the ground state because the relaxation rates are independent of the temperature.  (7) can be prepared successfully (and in a serendipitous way) by adding a variety of highly charged anions or cations to a reaction solution comprising Mn(O 2 CMe) 2 ·4H 2 O, mesaoH 2 and Et 3 N in MeOH, Equation (5); mesao 2− is the dianion of the R = Me derivative of salicylaldoxime (L in Figure 1). Equivalent reactions without the presence of highly charged anions or cations affords hexanuclear Mn(III) complexes. It has been suggested ) that these ions simply enforce changes in the selfassembly process (thus avoiding the Mn III 6 clusters) by their mere presence in the reaction mixtures. The metal topology can be described as a partial supertetrahedron in which the upper vertex is absent. The complex is an SMM with a U eff value of ∼21 cm −1 and-at the time of publication-represented the first chiral SMM obtained from achiral starting materials. Fitting of the dc magnetization data give the parameters S = 6, g = 1.98, and D = −0.60 cm −1 . 9 Mn II (O 2 CMe) 2 · 4H 2 O + 6 mesaoH 2 + 4 MeOH + 15 Et 3 N The use of α-methyl-2-pyridine-methanol (mpmH; M in Figure 1) in Mn carboxylate chemistry has provided access to the first {Mn 31 } coordination cluster (Abasi et al., 2017).  (6). The core can be described as a consecutive array of edge-sharing {Mn 4 (µ 4 -O)} tetrahedra and {Mn 3 (µ 3 -O)} triangles that are connected to each other via bridging oxido and methoxido groups. Complex 8 is not only structurally novel due to its nano-size dimensions, but also behaves as SMM with one of the largest U eff values (∼40 cm −1 ) yet reported for a 3d-metal SMM; moreover, it presents magnetization loops below 5 K, one of the highest T B temperatures at which molecular hysteresis has been observed for 3d-metal-based SMMs. Ac magnetic susceptibility measurements indicate an S = 23/2 ground state for this giant cluster. 31 Mn(O 2 CPh) 2 · 2H 2 O + 2 rac−mpmH + 46 Et 3 N + 24 MeOH A common synthetic route for preparing Mn coordination clusters and SMMs is the reaction of a pre-formed carboxylate cluster with a polydentate organic ligands (either bridging or chelating). Two frequently used Mn(II) carboxylate sources are the family of triangular oxido-centered complexes of the general formulation [Mn 3 O(O 2 CR) 6 L 3 ] 0,+ (Vincent et al., 1987), where L is a neutral monodentate ligand or solvent and R = Me, Ph,. . . , and the butterfly anionic complex (Bu n 4 N)[Mn III 4 O 2 (O 2 CPh) 9 (H 2 O)] (Wang et al., 1991). Two examples (from hundreds in the literature) are illustrated below.
The reaction of [Mn II,III,III 3 O(O 2 CPh) 6 (py) 3 ] (py=pyridine) and 1,1,1-tris(hydroxymethyl)ethane (H 3 thme; G with x = 1 and R = Me in Figure 1) in MeCN gives complex [Mn II 2 Mn III 10 O 4 (OH) 2 (O 2 CPh) 12 (thme) 4 (py) 2 ] (9) in 20% yield, Equation (7). The molecule has a rod-or ladder-like structure and its core can be described as consisting of 10 edge-sharing {Mn 3 (µ 3 -O)} triangles or five edge-sharing {Mn 4 (µ 3 -O) 2 } units. The four triply deprotonated ligands sit above and below the {Mn II 2 Mn III 10 O 4 (OH) 2 } plane and are two types: two use two of their arms in a µ 2 -fashion with the third arm acting as a µ 3bridge adopting the 5.322 coordination mode (FF in Figure 2); the other two thme 3− ligands have two µ 3 -arms and one µ 2 -arm adopting the 5.332 coordination mode (GG in Figure 2). The complex has an S = 7 ground state with g = 1.98 and D = −0.09 cm −1 . There are low-lying excited states with S values greater than the S = 7 ground state. The appearance of single-crystal magnetization hysteresis loops and relaxation measurements shows this complex to be an SMM .
butterfly and a {Mn III 3 Mn IV O 4 } cubane sharing a Mn III ion and the 5.5 (µ 5 ) oxido group. The complex possesses an S = 6 ground state spin (D = −0.18 cm −1 , g = 1.86). Single-crystal magnetization vs. dc field scans show only very weak hysteresis at 0.1 K.

Phosphonates, a Useful Alternative to Carboxylates
Phosphonates (RPO 2− 3 ) are ideal ligands for stabilizing 3dmetal coordination clusters, some of which are SMMs. They possess three O donor atoms and can, in principle link to a maximum of nine metal centers. However, polymeric 3dmetal phosphonates are very insoluble and the isolation of phosphonato complexes is problematic. Three routes have been used to overcome the insolubility problem and stabilize discrete molecular species (Langley et al., 2008). First, solvothermal techniques can improve the solubilities. Second, a "co-ligand" can be used to maintain some solubility. And third, pre-formed carboxylato clusters can be linked in various ways by displacing some carboxylate groups by phosphonato ligands. For preparing 3d-metal phosphonate SMMs, there is one disadvantage and one advantage. The disadvantage is that phosphonato ligands propagate weak magnetic exchange interactions between metal centers, but the presence of bridging co-ligands (including monoatomic ones) can improve the situation. The advantage is that most metal phosphonates form 2D lattices; if polymerization can be stopped, e.g., by using other ligands, this would be ideal for creating structural anisotropy which, in turn, can lead to the desirable magnetic anisotropy.
Compound (Et 3 NH)[Co II 8 (NO 3 ) 3 (O 3 PPh) 2 (chp) 10 (Hchp) 2 ] (11), where Hchp is 6-chloro-2-hydroxypyridine (A with R = Cl in Figure 1) and PhPO 2− 3 is the dianion of phenylphosphonic acid, is prepared by the reaction of Co(NO 3 ) 2 ·6H 2 O, Hchp and PhPO 3 H 2 in MeCN in the presence of Et 3 N, Equation (9); the yield is ∼50%. To retain the reagents in solution, the phenylphosphonic acid: Co(II) ratio should be low and Hchp has to be added before PhPO 3 H 2 . The base is necessary to deprotonate the two ligands. The two P atoms and the four Co II ions lie on the vertices of a trigonal prism. The remaining four metal ions lie above and below the triangular faces of the prism. The phosphonato ligands both adopt the 4.211 mode (HH in Figure 2), while eight of the chp − ligands bind with the 2.21 mode (II in Figure 2) and two with the 3.21 mode (JJ in Figure 2). The two neutral Hchp molecules are terminally coordinated exhibiting the 1.10 mode. The structure also contains two chelating nitrato groups (1.110) and a bridging 2.110 nitrato group (Langley et al., 2008). Study of the dynamic behavior of the magnetization shows that 11 displays slow magnetization relaxation. It was found that different samples gave different U eff values and a range of pre-exponential factors, τ 0 . For example, one sample gave U eff = 58.4(±2) cm −1 and τ 0 = 1.8(±3) × 10 −12 s, and another gave U eff = 55.6(±2) cm −1 and τ 0 = 2.1 (±3) × 10 −11 s.
The reaction between the pre-formed oxido-centered  (10). The core of the complex consists of a central trigonal prism. Two of the square faces of the prism are each bridged to one edge of a peripheral triangular unit by two 5.221 phosphonato ligands (KK in Figure 2) and two 2.2 methoxido groups. The Li I atom sits in the cavity of the central trigonal prism and forms six bonds to oxygen atoms. The remaining edges of the peripheral triangular units are unchanged from those in the starting material, i.e., are bridged by 2.11 PhCO − 2 groups. The oxido atom in the triangular Mn units are µ 3 (3.3), again as found in the starting Mn triangle. Interestingly, the crystal lattice contains two different {Mn II 2 Mn III 10 } molecules which have the same connectivity, but which are Jahn-Teller isomers of one another. The complex has a spin ground state of S = 6, but the small anisotropy of the system makes the compound a poor SMM with a very low U eff (Shanmugam et al., 2006  "Dancing" With Azides The azide (N − 3 ) ion is a very popular inorganic ligand in transition-metal cluster chemistry (Escuer and Aromi, 2006;Stamatatos and Christou, 2009;Escuer et al., 2014). In addition to its terminal coordination mode (1.10), it often participates as a bridging ligand in transition-metal chemistry. No less than 8 bridging modes have to-date been crystallographically characterized for the azido ligand. These are end-to-end (EE) (syn, syn-2.11, and syn, anti-2.11), endon (EO) (2.20, 3.30, 4.40) and mixed EE/EO (3.21, 4.22, 6.33) modes. The bridging azido group is an effective magnetic coupler and it is thus not surprising that it is an ingredient in many 3d-metal SMMs. EE azido groups generally promote antiferromagnetic exchange interactions, whereas the EO ligation mode is responsible for ferromagnetic exchange interactions; however, there are exceptions to this general rule. There have been reported many reactions schemes that lead to azidobridged 3d-metal coordination clusters and SMMs. These include, among others, systems intended to provide azides as the only bridging ligands in the cluster, primary organic ligand/azide "blends, " and reaction mixtures containing two different organic ligands and N − 3 ions. In the following, we give a representative example of SMMs from each reaction system.  (Alexandropoulos et al., 2014). The presence of Et 3 N is necessary for the isolation of the cluster, but its role appears unclear. The heptanuclear cations of 13 and 14 consist of an hexagon of M II atoms (M = Co, Ni) surrounding a seventh metal ion in the center of the hexagon. The disk-like core is completed by 12 N atoms of six 3.30 and six 2.20 EO N − 3 groups. The 3.30 azido groups bridge the M II 6 hexagon with the central metal ion, while the 2.20 ones bridge the M II 2 pairs of the hexagon. Two MeCN molecules are coordinated to each external metal center (see Figure 5). The cations have a layered structure, with layers of azido N atoms above and below the M II 7 plane. There are strong intracluster ferromagnetic Co II . . . Co II interactions in 13, which leads to an effective spin ground state of S = 7/2; the complex is SMM under an external dc field of 0.1 T (U eff = ∼20 cm −1 , τ 0 = 8.0 × 10 −8 s). The cluster cation has a very well-isolated groundstate spin value as indicated by the similar EPR spectra at 4 and 23 K.
The ligand "blend" sao 2− /N −, 3 where sao 2− is the dianion of salicylaldoxime (saoH 2 ; L with R = H in Figure 1), in Mn chemistry has provided access to a structurally interesting SMM FIGURE 5 | Schematic molecular structure of the heptanuclear cations of complexes 13 and 14; coordination bonds are drawn in bold (Schematic drawing inspired by Alexandropoulos et al., 2014). (Yang et al., 2007). The reaction of MnCl 2 ·6H 2 O, NaN 3 , saoH 2 , and (Et 4 N)OH in MeOH gives a dark green solution from which (Et 4 N) 3 [Mn II 2 Mn III 3 O(N 3 ) 6 Cl 2 (sao) 3 ] (15) can be crystallized in low yield (∼15%), Equation (12). The five metal ions define a trigonal bipyramid in which the two tetrahedral Mn II atoms occupy the axial positions and the distorted octahedral Mn III atoms are in the equatorial plane linked by a 3.3-O 2− group. Each Mn II atom is connected to the equatorial Mn III centers with three 2.20 azido groups, while a terminal chlorido group completes four-coordination at each Mn II . The Mn III atoms are bridged by three 2.111 sao 2− ligands (LL with M = Mn III and R = H in Figure 2). The anion has a ground-state spin of S = 11, with a first excited state S= 10 being 3.2 cm −1 higher in energy. Fit of the magnetization data gave a D value of−0.22 cm −1 . The cluster is SMM with a U eff value of 28 cm −1 .
5 Mn II Cl 2 · 6 H 2 O + 6 NaN 3 + 3 saoH 2 + 5 (Et 4 N)OH A modern trend in 3d-metal azide cluster chemistry is the employment of two organic ligands (a primary and an ancillary) and N − 3 ions in the reaction systems. The loss of a degree of synthetic control seems to be, at first glance, disadvantageous; however, this loss is more than compensated for by the large diversity of structural types expected. We present an example in which the ancillary organic ligand is a carboxylate ion. The reaction of Ni(ClO 4 ) 2 6H 2 O, phenyl 2-pyridyl ketoxime (ppkoH; H with R = Ph in Figure 1), NaN 3 and NaO 2 CH in a ∼1:1:1:1.5 ratio in MeOH gives a red-brown solution, from which can be subsequently isolated complex [Ni 5 (N 3 ) 4 (O 2 CH) 2 (ppko) 4 (MeOH) 4 ] (16) in ∼50% yield, Equation (13). The structurally novel {Ni 5 (2.20-N 3 ) 2 (3.30-N 3 ) 2 } 6+ core (see Figure 6A) contains the five metal ions in a bowtie topology (two Ni 3 triangles with a common vertex). Each diatomic oximate group of a 2.111 ppko − (MM with R = Ph and M = Ni in Figure 2) bridges the central Ni II atom and a peripheral one (see Figure 6B), while a 2.11 formato ligand spans the base of each isosceles triangle. Magnetic data indicate an overall ferromagnetic behavior (S = 5 ground state). Ac magnetic susceptibility studies reveal non-zero, frequencydependent out-of-phase (χ " M ) signals at temperatures below ∼3.5 K. However, single-crystal magnetization vs. dc field scans at variable temperatures and variable sweep rates down to 0.04 K show no significant hysteresis loops, except minor ones at 0.04 K that are assigned to weak intermolecular interactions (Papatriantafyllopoulou et al., 2010).

Comproportionation Reactions in Manganese SMM Chemistry
A commonly used synthetic route for the preparation of highoxidation state Mn clusters and SMMs is the comproportionation reaction of a Mn II source and Mn VII O − 4 (i.e., Mn VII ) in the presence of simple carboxylate ions and/or chelating/bridging ligands. Such reactions are very useful because they normally give products containing some Mn(IV), thus preventing complication from low-lying excited states which is a common situation with Mn II . . . Mn III and Mn III . . . Mn III couplings. This was the original method for the synthesis of the {Mn III 8 Mn IV 4 } complex 1 which, in a sense, initiated the area of SMMs (Lis, 1980 (14), in the form of reddish-black crystals. The crystals were of X-ray quality and the structure of the cluster was solved by single-crystal, X-ray crystallography.
An ideal high-oxidation Mn(VII) reagent for such reactions in organic solvents has been documented to be (Bu n 4 N)(MnO 4 ) (Vincent et al., 1986); this oxidizing reagent is soluble in many organic solvents. It is easily prepared by mixing aqueous solutions of KMnO 4 and (Bu n 4 N)Br in slight excess; it is pure enough for use without recrystallization and storage at −5 • C increases its stability (storage at room temperature leads to slow decomposition in a period of ∼1 month give a brown sticky material  (Sessoli et al., 1993b). Although the Mn II :MnO − 4 ratio was identical to that in Equation (14), the major product of this reaction is [Mn II,III,III 3 O(O 2 CPh) 6 (py) 2 (H 2 O)], i.e., in this case only a small amount of higher (>III) average oxidation product is obtained.
An interesting comproportionation reaction to manganese (III/IV) pivalato clusters gave a half-integer spin SMM (Mukherjee et al., 2013 (18) in ∼25% yield. The average metal oxidation state in the reaction is +4.5, Equation (16), but in the product it is less (∼+3.1). The low yield of 18 is an evidence that there are probably other higher oxidation-state products in the mother liquor which are not precipitated. Use of MnCl 2 or Mn(NO 3 ) 2 , instead of the manganese(II) perchlorate salt, also give 18, but in lower yields. It was seemed possible that the low Mn(IV) content in 18 is due to the fact that MnO − 4 oxidizes not only Mn(II) but also the solvent. The use of an even lower Mn(II):Mn(VII) reaction ratio of 2:3 to increase the average oxidation state to +5, Equation (17) (2) cm −1 . The ac susceptibility data revealed non-zero, frequency-dependent χ" M signals for this complex below 3 K, providing a strong evidence for SMM behavior; this behavior was confirmed by the appearance of hysteresis loops in single-crystal magnetization vs. dc field studies. The loops display two wellresolved QTM steps. The complex is a half-integer cluster and it is not expected to display QTM in the absence of transverse fields; the latter are provided by the transverse components of dipolar and exchange fields from neighboring molecules, and hyperfine fields from 55 Mn (I = 15/2) nuclei. The relaxation rate at zero applied field is very fast at several temperatures, and this prevented the determination of U eff . Complex 19 does not exhibit χ" M signals above 1.8 K.

The Reductive Aggregation Route and Its Modifications in Manganese SMM Chemistry
The reductive aggregation route in Mn chemistry represents a modification of the comproportionation procedure. The Mn(II) source is omitted and MeOH is used not only as a solvent, but also both as the reducing agent for Mn(IV) and as a potential source of bridging MeO − groups, in the presence of an excess of carboxylic acid that prevents precipitates of Mn hydroxides and/or oxides, and provides carboxylato ligands.
The reaction of phenylacetic acid (PhCH 2 CO 2 H) with (Bu n 4 N)(MnO 4 ) in MeOH leads to a dark brown solution and isolation of [Mn III 10 Mn IV 6 O 16 (OMe) 6 (O 2 CCH 2 Ph) 16 (MeOH) 6 ] (20) (King et al., 2004), eqn. (18). In Equation (18), it is assumed that MeOH is oxidized to formaldehyde. The yield is low (∼10%), but the reaction is perfectly reproducible and gives crystalline material. Such reactions are certainly complicated with several cluster species in solution; the final outcome depends on several factors, such as relative solubilities, kinetics of crystallization and lattice energies. An important issue of this reaction (and several reactions of this type) is the necessity of using small volumes of MeOH; dilute solutions lead to Mn hydroxides and/or oxides. In the reaction that leads to 20, use of EtOH instead of MeOH leads also to hydroxides/oxides, whereas 2-propanol leads to pale yellow solutions indicative of full reduction of Mn(VII) to Mn(II  (King et al., 2005). No sign of any steps was observed in the hysteresis loops as a result of broadening effects due to extensive disorder of most Bu t groups. The reductive aggregation route has also led to anionic {Mn III 8 Mn IV 4 } 2− SMMs (Tasiopoulos et al., 2005).

Procedures Involving Two Different Primary Organic Ligands
The combined use of two primary organic ligands with different functionalities in the reaction systems is another route for the construction of clusters and SMMs. This method has developed mainly by the groups of Tasiopoulos and Brechin. In the most of the reported examples, one of the primary organic ligands does not appear in the final product, but its role in solution appears essential for the reaction. This route has led, among others, to a giant {Mn II 18 Mn III 14 } double-decker wheel with an S = 11 or 12 ground state and SMM behavior (U eff = ∼31 cm −1 , τ 0 = 3.5 × 10 −12 s); the small τ 0 value has been attributed FIGURE 7 | A simplified view of the {Mn III 3 Na I } repeating unit that forms the core of complex 24 (R ′ NO = saph 2− ); coordination bonds are drawn in bold (Schematic drawing inspired by Manoli et al., 2016).
to low-lying excited states and weak intermolecular interactions (Manoli et al., 2011). We give below a representative, simpler example to illustrate the method; the product however, lacks SMM behavior, even to temperatures as low as 30 mK. The structure of 24 consists of a {Mn III 18 Na I 6 } wheel-like cluster comprising oximate-based {Mn III 3 O} 7+ triangular units linked through six Na + ions. The {Mn III 3 Na I } repeating unit possesses an oxo-centered triangular arrangement of Mn III atoms and a Na + ion attached to it via oxygen atoms (see Figure 7) of oximate groups which occupy the edges of the triangle. Two 3.211 (NN in Figure 2) and one 4.221 (OO in Figure 2) sao 2− ligands, as well as one 2.2 H 2 O molecule connect each Mn(III) triangle with two neighboring Na + ions, which in turn are linked to the next triangular subunits forming the wheel.

THE YEARS OF DESIGN: STRICT AND LESS STRICT SYNTHETIC STRATEGIES
All the routes described in section The Years of Innocence: "Try and See" Exercises depend on serendipity rather than on strict design; this means that the final product and its properties could not be predicted. However, those routes (and others not mentioned) were the basis of the great developments in the area of SMMs. After ∼2,000, scientists were more interested in designing the chemical and structural identity of the coordination clusters resulting from their efforts, and in predicting and controlling the magnetic properties of the products. Design principles and examples, which have been developed mainly in Mn SMM chemistry, will be presented in this chapter. The strategies are based on the modification of structural types already known that may or may not already be SMMs in their own right. From the structural viewpoint, two are the important parameters to be considered for such modifications: (a) the nuclearity of the coordination cluster, and (b) the core. Each of these parameters, or both, may change or may not change in the product with respect to the starting material (which is always a pre-formed cluster). From the magnetic viewpoint, the starting material may or may not be an SMM. In the former case, the primary goal of the synthetic strategy is to improve the SMM properties. In the latter, the main goal is to "switch on" SMM properties.
The choice of the below mentioned strategies is subjective and again we owe our apologies to many synthetic chemists whose elegant work will not be illustrated here.

Substitution of Carboxylato Ligands by Other Carboxylato Groups
Several groups (mainly Christou's group) have developed procedures to replace RCO − 2 ligands in Mn SMMs with a variety of other R ′ CO − 2 groups. The replacement can be complete or partial (vide infra). The most studied SMM for such reactivity studies is 1. As described in section Comproportionation Reactions in Manganese SMM Chemistry, complex 17 can be prepared using a comproportionation reaction, but in low yield (∼10%). Treatment of 1 with 32 equivs of PhCO 2 H (100% excess) in CH 2 Cl 2 leads to the exchange of most MeCO − 2 ligands. For full replacement another ligand substitution cycle can be performed; the partially exchanged solid from the first cycle can be treated with other 32 equivs of PhCO 2 H giving the desired product in an overall >50% yield (Sessoli et al., 1993b), Equation (23) In general, such reactions are equilibria, Equation (24). The substitution reaction is facilitated by the higher acidity of the incoming carboxylic acid, R ′ CO 2 H, and its presence in excess.
In the case of R = H (i.e., 1), MeCO 2 H can be removed as its toluene azeotrope shifting the equilibrium represented by Equation (23) to the right. The distillation of the azeotrope of MeCO 2 H and toluene is a very valuable method for introducing R ′ CO − 2 groups whose conjugate acid has a pK a value higher than that of MeCO 2 H or comparable to this (Bagai and Christou, 2009). The nuclearity and the core in the product are the same with those existing in the starting material. Such carboxylate substitution reactions often lead to better solubility in organic solvents and altered redox properties of the products compared to 1.
This type of carboxylate substitution is of important utility in several technological aspects of the chemistry of 1 and its derivatives (Domingo et al., 2012). We are doing a parenthesis here to mention that applications of SMMs require the development of strategies to move from bulk crystals to molecules that can be grafted on surfaces, sensors or other systems able to behave as a device; the challenge here is that the SMMs should retain their properties in the device. There are three routes for the nanostructuration of SMMs on surfaces from solution: (a) SMM direct deposition on a bare surface in order to immobilize the SMM through weak non-covalent interactions, (b) pre-functionalization of SMMs with groups that are able to interact chemically with the bare surface, and (c) pre-functionalization of the surface with appropriate groups that are able to interact with the SMM. An early example of pre-functionalization of SMMs for deposition onto an Au (111) surface involves replacement of the MeCO − 2 groups of 1 by carboxylato ligands containing thiol groups (Cornia et al., 2003). The 16-sulfanylhexadecanoate (P in Figure 1) derivative of 1 was designed, in which the thiol groups were acetyl-protected (Q in Figure 1) to avoid undesirable reactivity in the presence of the Mn III and Mn IV centers. The U eff value of the product is ∼38 cm −1 , close to that of 1. The deposition was achieved by incubating the Au substrate in an SMM solution in a basic environment to ensure the deprotection of the thiol groups. The SMMs covered completely the surface with disordered layers, which could be disrupted by a continuous scanning of the area allowing the first imaging of discrete dodecanuclear Mn molecules by Scanning Tunneling Microscopy.
The carboxylate substitution can be partial and site-selective.  (24). The site-selective (or specific) substitution results from the different reactivity to electrophiles of some carboxylate ligands in 1, 17 and related SMMs. Of the 16 carboxylates, 4 (type I) have both their O atoms on Jahn-Teller elongation axes of Mn III atoms, 4 (type I ′ ) have one, and 8 (type I ′′ ) have none being equatorial. Since the Mn III -O bonds on the Jahn-Teller axis are weaker than those on the axes of the equatorial plane, the relative susceptibility of the carboxylate groups to electrophilic attack would be expected to be I>I ′ >I ′′ (Bagai and Christou, 2009) and this has been nicely demonstrated in several cases. Another relevant synthetic hypothesis is that the more basic carboxylate groups would thermodynamically prefer to bind at the non-Jahn-Teller equatorial sites (type I ′′ ) where they can form the strongest Mn III -O bonds.  CH 2 , 31). The four nitrato groups are coordinated in the same 2.11 mode as the carboxylato ligands at the type I (section Substitution of Carboxylato Ligands by Other Carboxylato Groups) Mn III axial Jahn-Teller positions . The reaction represented by Equation (30) can be reversed; reactions of 30 and 31 with 4 equivs of NaO 2 CR in CH 2 Cl 2 /MeOH give complexes 17 and 25, respectively, Equation (31). This makes complexes such as 30 and 31 candidates for reactivity centered at the nitrato positions, taking advantage of the good leaving properties of bound NO − 3 ions. Complex 31 (S = 10) is an SMM with a U eff value of 50.0 cm −1 . The complex also shows hysteresis in magnetization vs. dc field scans; the hysteresis loops show steps at regular intervals of magnetic field, the diagnostic evidence of QTM. [ (Chakov et al., 2003). The PhSO − 3 ligands are at the eight axial (I and I ′ , see in the "Substitution of Carboxylato Ligands by Other Carboxylato Groups" section above) Mn III Jahn-Teller positions, again as expected on the basis of relative basicities (pK a of MeCO 2 H = 4.76, pK a of PhSO 3 H = 2.55). In other words, the more basic (stronger donors) MeCO − 2 ligands prefer the equatorial sites where shorter, stronger Mn-O bonds can be formed stabilizing better the molecule. The less basic PhSO − 3 ligands are located at axial positions where they bridge either Mn III /Mn III or Mn III /Mn IV pairs and therefore have one or both of their O atoms on the Jahn-Teller elongation axes of the Mn III atoms. The mixed-ligand cluster 32 retains both the high ground-state spin (S=10) and SMM behavior (U eff = ∼47 cm −1 ) of the parent SMM 1 and its carboxylato derivatives. Hysteresis loops were observed below 4.0 K; their coercivities increase with decreasing temperature, a typical behavior of SMMs. The synthetic importance of reactions, such as that represented by Equation (32)  . In other words, the large steric bulk of the Ph 2 PO − 2 ligands has, as a consequence, their equal distribution between axial and equatorial sites. The SMM properties are retained in 33 (U eff = ∼42 cm −1 ). Magnetization hysteresis loops were observed for oriented crystals of 33·12CH 2 Cl 2 .

Reduction Pathways for the {Mn III 8 Mn IV 4 } SMMs
Detailed electrochemical studies on many members of the {Mn III 8 Mn IV 4 } family of SMMs (1, 17, 25, 26,. . . ) have revealed several oxidation and reduction processes, with some of the latter being reversible. The oxidation processes are at rather high potentials, but the reduction processes are accessible by chemical means. Electron-withdrawing R groups in the carboxylato ligands (e.g., R = CHCl 2 or C 6 F 5 ) favor the reduction processes, because they reduce the electron density in the core and make reductions easier. An ideal reducing agent is the iodide (I − ) because: (i) Its reducing strength is sufficient, but mild; (ii) there are many organic salts C + I − , the majority of which have good solubility in a variety of organic solvents; and (iii) the only byproduct of the redox processes is elemental I 2 which can be easily removed.
Reduction  (33). The structure reveals that the extra electron is localized on an outer (originally Mn III ) ion rather than an inner (cubane) Mn IV center (see Figure 8), and the product is a trapped-valence {Mn II Mn III 7 Mn IV 4 } − ion (Eppley et al., 1995). Cluster 34 has an S = 19/2 ground state being SMM. The complex exhibits well-developed hysteresis loops. The benzoate analog of 34, i.e., cluster (Ph 4 P)[Mn 12 O 12 (O 2 CPh) 16 (H 2 O) 4 ] (35) can be prepared in a similar way (Aubin et al., 1999). For the eight "external" Mn ions (formerly Mn III ), it was not possible to decide whether a trapped-valence {Mn II Mn III 7 } or an electronically delocalized description is the most scientifically correct. Complexes such as 34 and 35 (both with S = 19/2) are ideal candidates for the study of QTM in half-integer spin systems; QTM is not allowed in such compounds in the absence of an applied magnetic field.  represents Mn IV , the symbol represents Mn III , the symbol represents Mn II , and the asterisk illustrates the triply bridging O 2− group (Schematic drawing inspired by Bagai and Christou, 2009). Coordination bonds are drawn in bold.
of the bulk material from CH 2 Cl 2 /hexanes yields a mixture of two crystals forms, 36·4CH 2 Cl 2 ·H 2 O (36a) and 36·6CH 2 Cl 2 (36b), both of which have been structurally characterized by single-crystal, X-ray crystallography. The molecular structures are almost identical, with the two extra electrons localized on former Mn III centers (see Figure 8) generating a trapped-valence {Mn II 2 Mn III 6 Mn IV 4 } 2− ion. Bulk dc magnetization data for dried 36 in the 1.8-4.0 K and 1-7 T ranges were fit to give S = 10, D = −0.275 cm −1 and g = 2.00. The U eff values for 36a and 36b are ∼13 and ∼21 cm −1 , respectively. Magnetization vs. dc field scans on single crystals of the two samples gave hysteresis loops containing steps due to QTM. The step separations yielded |D|/g values of 0.087 and 0.14 cm −1 for 36a and 36b, respectively, indicating that the differences in U eff are caused by changes to D.   (Bagai and Christou, 2007), Equation (35) (C = Pr n 4 N, Me 4 N); the yields are high (∼80%). The complexes are not stable in solution and all the attempts to grow singlecrystals of the salts were unsuccessful. Dc magnetization data on dried microcrystalline samples were fit by matrix diagonalization methods to give S = 17/2, D = −0.25 cm −1 and g = 1.91 for 37 and S = 17/2, D = −0.23 cm −1 and g = 1.90 for 38. The two complexes exhibit frequency-dependent out-ofphase magnetic susceptibility signals at ≤2.5 K, indicating them to be SMMs, albeit at lower temperatures compared with the {Mn 12 } SMMs (6-8 K range), {Mn 12 } − SMMs (4-6 K range), and {Mn 12 } 2− SMMs (2-4 K range); the shifts to lower temperatures are indicative of the decreasing S and D-values upon successive reduction and therefore the decreasing U eff value. The decreased |D| values in the order {Mn 12 }>{Mn 12 } − >{Mn 12 } 2− >{Mn 12 } 3− indicate that the third added electron is localized on a formerly Mn III ion, since the Jahn-Teller-distorted Mn III ions are the main source of the molecular anisotropy (see Figure 8). The preferential reduction of Mn III rather than Mn IV can be interpreted in terms of the different environments which these metals have within the {Mn 12 } unit. Thus: (i) The Mn IV centers are each bound to five hard O 2− groups which favor higher oxidation states and preclude reduction to Mn III ; and (ii) the reduction of a Mn IV would generate a Mn III center whose Jahn-Teller distortion axis would have to include at least one Mn III -O 2− bond (an inherently strong bond) and would introduce a severe strain into the rigid {Mn IV 4 O 4 } subcore.
"Switching On" SMM Behavior Through Substitution of Bridging Hydroxido Groups by End-on Azido or Isocyanato Ligands in Pre-formed Clusters One of the prerequisities of a 3d-metal cluster to behave as SMM is a high or a relatively high total spin in the ground state. The high-spin ground state can result from either ferromagnetic (or ferrimagnetic) metal···metal exchange interactions and/or topologically frustrated antiferromagnetic exchange interactions. In general, however, it is difficult to predict which topology will lead to high-spin ground states, and more difficult to design and achieve the syntheses of such compounds. In many coordination clusters, the exchange interactions between the paramagnetic metal ions is propagated by hydroxido (OH − ), oxido (O 2− ), alkoxido or alkoxido-type (RO − ), or carboxylato (RCO − 2 ) ligands, or a combination of two or more such ligands. These ligands most often propagate antiferromagnetic exchange interactions. Our group (Papaefstathiou et al., 2001a,b;Boudalis et al., 2004Boudalis et al., , 2008, in collaboration with Escuer's group, have developed a general strategy for the synthesis of high-spin clusters, which often "switches on" SMM behavior. The strategy is based on the substitution of bridging hydroxido or/and alkoxido groups in pre-formed coordination clusters (generally low-spin due to antiferromagnetic interactions) by EO azido groups (2.20, 3.30, 4.40) or N-bonded cyanato (isocyanato) bridging groups (2.02, 3.03, 4.04). The nuclearity does not change, but the core does. The entering bridging groups introduce ferromagnetic components in the superexchange scheme of the cluster, always increasing the ground-state spin and sometimes "switching on" SMM properties. The method works better for divalent 3d metals. The strategy is illustrated with one example (Boudalis et al., 2004(Boudalis et al., , 2008. Cluster [Fe II 9 (OH) 2 (O 2 CMe) 8 {(py) 2 CO 2 } 4 ] (39), where (py) 2 CO is di-2-pyridyl ketone (C in Figure 1) and (py) 2 CO 2− 2 is the dianion of the gem-diol form of (py) 2 CO (D in Figure 1) is prepared under an inert atmosphere by the reaction outlined in eqn. (36) in ∼30% yield. The nine Fe II atoms adopt the topology of two square pyramids sharing a common apex and are held together by four 5.3311 (py) 2 CO 2− 2 ligands (PP in Figure 2). Each Fe II ···Fe II edge of the bases of the pyramids is further bridged by one syn, syn-2.11 MeCO − 2 group. The four MeCO − 2 groups create a concave cavity in the base of each pyramid, into which a rare 4.4 OH − group is trapped capping the square base. The magnetic study of 39 reveals an overall antiferromagnetic behavior with an S = 2 ground state.  (37). Complexes 40 and 41 have striking structurally similarity to 39, the main difference being the replacement of the 4.4 OH − groups in the latter by the 4.40 N − 3 and 4.04 OCN − groups in the products (see Figure 9). 1 H paramagnetic NMR spectroscopic studies in CD 3 CN reveal that the three enneanuclear Fe(II) complexes are stable in solution. Dc magnetic susceptibility studies show that the ground-state spin values are higher in 40 and 41 than in 39; variable-field experiments have shown that the ground state is not well isolated from the low-lying excited states, and it cannot thus be determined with accuracy. X-band EPR spectroscopy at 4.2 K reveals characteristic integer-spin signals for 39, but not for the azido-and isocyanato-bridged clusters. Complexes 40 and 41 are SMMs with U eff values of 28.5 and 30.5 cm −1 , respectively. The slow magnetic relaxation has also been observed by 57 Fe Mössbauer spectroscopy.
"Tweaking" the Spin The strategy reported in section "Switching on" SMM Behavior Through Substitution of Bridging Hydroxido Groups by Endon Azido or Isocyanato Ligands in Pre-formed Clusters involves conversion of a low-spin 3d-metal cluster into a high-spin one with the simultaneous "switching on" of SMM properties. This approach changes the core. An alternative approach is to start with a pre-formed, high-spin molecule (and perhaps an SMM) and then perturb it without significant core change by targeted modification of the peripheral ligands, in order to modify some exchange parameters, and slightly increase the ground-state spin and "switch on" or enhance the SMM properties. This strategy has been termed "spin tweaking" (Stamatatos et al., 2007a,b). We give an example of this approach from Mn chemistry. The starting material is [Mn II 6 Mn III 18 Mn IV (OH) 2 O 18 (N 3 ) 12 (pdm) 6 (pdmH) 6 ](N 3 ) 2 (42) (Murugesu et al., , 2008, where pdm 2− and pdmH − are the di-and the monoanion, respectively, of pyridine-2,6-dimethanol (pdmH 2 ; R in Figure 1). The core is held together by 12 4.4 O 2− , 6 3.3 O 2− and two 3.3 OH − groups, as well as 6 2.20 N − 3 ions and the deprotonated alkoxido arms of the 5.331 pdm 2− (QQ in Figure 2) and 3.211 pdmH − ligands (RR in Figure 2). The complex has a total spin of S = 51/2 in the ground state, being an SMM (U eff = ∼8 cm −1 ). The target was to replace the peripheral 6 2.20 and 6 terminal (1.10) azido ligands by 6 2.21 hmp − ligands (SS in Figure 2), where hmp − is the monoanion of 2-(hydroxymethyl)pyridine (hmpH; S in Figure 1). The synthetic idea was that the bridging deprotonated oxygen atom of hmp − (known to promote ferromagnetic exchange between Mn ions) can be considered as the equivalent of the bridging N atom of the 2.20 N − 3 ligands, while the 2-pyridyl nitrogen atom of hmp − is the equivalent of the terminal N atom of the 1.10 N − 3 ligands. The reaction of 42, Na(hmp) and NaClO 4 in 1:6:6 molar ratio in MeCN/MeOH gives a dark brown solution, from which is subsequently crystallized [Mn II 6 Mn III 18 Mn IV (OH)(OMe)O 18 (hmp) 6 (pdm) 6 (pdmH) 6 ] (N 3 ) 2 (ClO 4 ) 6 (43) in ∼65% yield, Equation (38). The crystal structure shows the core of 43 to be isostructural with that of 42. The main differences are that the 12 coordinated azido groups of 42 are replaced by 6 2.21 hmp − ligands in 43 (see Figure 10) and also a 2.2 OH − group of the former is substituted by a 2.2 MeO − group in the latter. The result is that all bridging atoms in the product are oxygen atoms and there are thus changes in many exchange interactions. The ground-state spin increases from S = 51/2 in 42 to S = 61/2 in 43. Despite the intended increase of S, 43 is not a better SMM compared to 42 (Stamatatos et al., 2007a,b). The conversion of 42 to 43 is the first time in which removal of EO azido groups leads to an increase in the ground-state S-value in a 3d-metal coordination cluster; usually their addition is a method of increasing the S value (section "Switching on" SMM Behavior through Substitution of Bridging Hydroxido Groups by End-on Azido or Isocyanato Ligands in Pre-formed Clusters).    al., 2005, 2007c), Equation (39). The 1:3 ratio was chosen to allow for the addition of one mpko − ligand onto each edge of the triangle. The process is a simple substitution reaction involving replacement of 3 RCO − 2 groups and 3 py ligands by three mpko − ones. As in 44a-44c, the cations of 45a-45c contain a {Mn III 3 (µ 3 -O)} 7+ triangular core, but with each Mn III 2 edge now bridged by a 2.11 RCO − 2 group and a diatomic oximato group of one mpko − ligand (see Figure 11). The 3 RCO − 2 groups lie on one side of the {Mn III 3 } plane and the three oximato groups on the other. The 2.111 coordination mode of the mpko − ligands (MM with R = Me and M = Mn III in Figure 2) results in a buckling of the formerly planar {Mn III 3 (µ 3 -O)} 7+ core; the consequences are a relative twisting of the three metal octahedra and the displacement of the central O 2− group ∼0.3 Å above the {Mn III 3 } plane on the same side as the RCO − 2 groups. The Mn III ···Mn III exchange interactions are ferromagnetic in 45a-45c resulting in an S = 6 ground state. The complexes are SMMs with U eff values of ∼8 cm −1 . Magnetization vs. dc field experiments on single crystals of 45a·3CH 2 Cl 2 display hysteresis loops that exhibit steps due to QTM. The loops are temperature-independent below 0.3 K, suggesting only QTM in the ground state between the lowest-lying M S = ±6 levels. High-frequency EPR spectra of single crystals of 45a·3CH 2 Cl 2 provide strong evidence for a non-negligible transverse anisotropy of E ≥ 0.015 cm −1 .
[Mn III 3 O(O 2 CR) 6 (py) 3 ](ClO 4 ) + 3 mpkoH (39) DFT calculations reveal that the unusual ferromagnetic exchange interactions in complexes 45 may originate from a combination of the "non-planarity" (with respect to the {Mn III 3 } plane) of the bridging oximato groups and the non-parallel alignment of the Jahn-Teller axes; however, several factors are at play (Cano et al., 2008;Atanasov et al., 2011) including the {Mn III 3 } out-of-phase shift of the triply bridging oxido group.
The molecular structure of the complex, which is common to all members of this family (vide infra) consists of two offset, stacked triangular {Mn III 3 (µ 3 -O)} subunits held together via six sao 2− ligands, which span each Mn III ···Mn III edge of the two triangles, forming Mn III -N-O-Mn III bridges, as well as dimerizing the two {Mn III 3 (µ 3 -O)} subunits via their phenolato and oximato O atoms (see Figure 12). The two MeCO − 2 groups are 2.11 (this also occurs in the firstly reported members of the family with formato, benzoate etc. groups) and sit "above" the plane of the symmetry equivalent triangles. Four of the sao 2− ligands show the common 2.111 coordination mode (LL with R = H and M = Mn III in Figure 2), while the other two adopt the rather rare 3.211 mode (this is similar to the coordination mode NN observed in cluster 24, section 3.8, with a Mn III atom replacing Na I ). The core is {Mn III 6 (3.3-O) 2 (3.21-ONR ′′ ) 2 (2.11-ONR ′′ ) 4 } 8+ , where R ′′ NO − = sao 2− . The four "central" Mn III ions are 6-coordinate with distorted octahedral geometries, while the outermost Mn III ions exhibit square pyramidal geometries with an axial weak bonding interaction of ∼3.5 Å to a phenolato O atom. Magnetic studies reveal that the compound has an S = 4 ground state as a result of ferromagnetic exchange between the two antiferromagnetically coupled {Mn III 3 } triangles; it is also an SMM with U eff = ∼20 cm −1 . While the S and U eff values are not spectacular, the D-value (−1.2 cm −1 ) certainly is.
In an elegant series of studies (Milios et al., 2007a(Milios et al., ,b,c,d, 2008), Brechin's group showed that by employing derivatised versions of the salicylaldoximate ligand (Rsao 2− ; R = Me, Et) and bulky carboxylates [R ′ CO − 2 ; R = Ph, CMe 3 , CPh 3 , Ph 2 OPh, Ph 4 OPh, PhBr, Ph(Me) 2 , 3-thiophene. . . ], It should be mentioned at this point that the strategy detailed in this section is different from the approaches described in sections "Switching on" SMM Behavior through Substitution of Bridging Hydroxido Groups by End-On Azido or Isocyanato Ligands in Pre-formed Clusters, "Tweaking" the Spin, and "Switching on" SMM properties upon Conversion of Low-Spin Complexes Into High-Spin Ones Without Changing the Core. The present strategy is strictly speaking not a switching on of the SMM property (sections "Switching on" SMM Behavior Through Substitution of Bridging Hydroxido Groups by Endon Azido or Isocyanato Ligands in Pre-formed Clusters and "Switching on" SMM properties upon Conversion of Low-Spin represents Mn III . The dashed lines represents the phenolato-o to square-based pyramidal Mn III interaction (Schematic drawing inspired by Milios et al., 2008).
Complexes Into High-Spin Ones Without Changing the Core), since the S = 4 {Mn III 6 } cluster is also an SMM. Also, the increase of the total ground-state spin here is large (∼200%) and not small (∼20%) as in the "spin tweaking" strategy (section "Tweaking" the Spin); in both strategies the starting materials are SMMs.

Synthesis of Radical-Bridged 3d-Metal SMMs
As it has been clearly stated in sections Introduction and Synthesis of 3d-Metal SMMs: General Considerations, the continuing research efforts to prepare 3d-metal SMMs with higher U eff values take into consideration the ground-state spin, S, and the axial zero-field splitting parameter, D; both these key parameters are correlated to the energy barrier height. A third important parameter, which is often overlooked, is the exchange coupling constant, J. The value of this physical parameter determines the separation between the spin of the ground state and the spins of the excited states. The J-value must be large (strong coupling) in an SMM to maintain slow magnetization relaxation at elevated temperatures and also to decrease fast magnetization relaxation through pathways involving excited spin states. An excellent synthetic strategy to achieve strong magnetic exchange is the incorporation of radical ligands into coordination complexes. The direct exchange between paramagnetic 3d-metal ions and paramagnetic radical ligands can result in very strong magnetic coupling, stronger than the more common superexchange interactions between the metal centers (Demir et al., 2015). In the coordination clusters of 3d-metal ions, the radical ligand is often bridging. Transition metal-based polynuclear SMMs possessing a radical ligand are extremely rare, most of them being dinuclear. We give an example of a dinuclear complex below.
Benzoquinonoid groups (A in Figure 13) are ideal ligands for the synthesis of radical-bridged SMMs with strong magnetic coupling. They can easily undergo redox reactions providing paramagnetic and diamagnetic electrontransfer isomeric species. The reaction of Fe(CF 3 SO 3 ) 2 with tris(2-pyridylmethyl)amine (tpya; T in Figure 1) in anhydrous MeCN, followed by treatment with a mixture of Li{N(SiMe 3 ) 2 } and NPh LH 2 (U in Figure 1) The molecular structures of both unreduced (52) and reduced (53) complexes show two {(tpya)Fe II } 2+ subunits connected by a deprotonated azophenine bridging ligand, with two N atoms occupying cis sites of each octahedral Fe II ion; the four N atoms of the tetradentate chelating tpya ligand complete the coordination sphere at each metal center (see B for 52 and C for 53 in Figure 13). While the gross structural features of the cations of 52 and 53 are similar, bond lengths in the backbone of the bridging ligand show key differences which permit the assignment of a trianionic bridging ligand with an unpaired electron ( NPh L 3−• ) in the case of 53. This assignment is supported by 57 Fe-Mössbauer studies; the identical isomer shifts, δ, in the two complexes suggest that the reduction is ligand-centered, while the larger quadrupole splitting, E Q , in 53 is due to the change of the ligand field at Fe II associated with ligand reduction and the larger distortion from the ideal octahedral environment at the metal ion in 53 compared to 52. Variabletemperature dc magnetic susceptibility studies for solid 52 show a weak antiferromagnetic exchange interaction between the highspin Fe II ions through the diamagnetic NPh L 2− ligand leading to an S = 0 ground state; the J value is −2.90(2) cm −1 based on the spin HamiltonianĤ= −2J(Ŝ Fe1 ·Ŝ Fe2 ). Data for 53 reveal an exceptionally strong antiferromagnetic exchange between Fe II centers and the NPh L 3−• radical ligand (J ≤ −900 cm −1 based on the same Hamiltonian). SMM behavior was confirmed for 53 through ac susceptometry with a U eff value of 50 cm −1 . The |J| ≥ 900 cm −1 value is the strongest magnetic exchange ever observed in an SMM.

Supramolecular Approaches in Manganese SMM Chemistry
For several technological applications of SMMs, coupling of two or more SMMs to each other or to components of a device is very important. The coupling must be very weak in order to preserve the intrinsic single-molecule properties of each SMM. This coupling can be, in principle, achieved by H bonds and many H-bonded dimers, 1D, 2D, and 3D networks have been reported in the literature. However, reliance on H bonds to obtain the desired inter-SMM associations is a synthetic weakness, because it is difficult to control the oligomerization and to achieve retention of the supramolecular structure in solution. The answer to the problem is the linking of SMMs via coordination bonds. This approach gives often coordination polymers (1D, 2D, 3D) in which the coupling between the SMMs is strong leading to loss of the SMM property and appearance of single-chain magnetism (for 1D compounds) or ordering (for the 2D and 3D compounds). A synthetic challenge is thus the connection of "good" SMMs into discrete oligomers and the achievement of very weak exchange coupling between them through the bridging ligands, so that the properties of the oligomer can be considered as a small perturbation of the properties of the constituent SMM subunits. The desired oligomers can be prepared by direct mixing simple starting materials, but such a route offers limited synthetic control and there is no guarantee that the magnetic properties will be the targeted ones. A good strategy is the use of building-block approaches to link SMMs together employing carefully chosen linker groups that will provide weak inter-SMM interactions and ensure that discrete oligomeric species (and not polymers) are obtained. The groups of Papaefstathiou, Brechin, Christou, and Escuer, among others, have successfully contributed into this general synthetic goal Inglis et al., 2010;Cordero et al., 2011;Nguyen et al., 2011Nguyen et al., , 2015Nguyen et al., , 2016Mowson et al., 2013). We herein present examples of the supramolecular approaches which have been almost exclusively developed in Mn SMM chemistry. Complexes 45a, 45b, and 45c (section "Switching on" SMM properties upon Conversion of Low-Spin Complexes Into High-Spin Ones Without Changing the Core) are good SMM building blocks for the application of the supramolecular approach. They have an S = 6 ground state and are SMMs. Their triangular cations have C 3 symmetry with the carboxylate and mpko − FIGURE 13 | (A) Redox transformations of anionic benzoquinonoid ligands (E=O and NR; R is the Ph group in the present example). (B) Schematic representation of the molecular structure of the dication that is present in 52. (C) Schematic representation of the molecular structure of the monocation that is present in 53. N ∧ N ∧ N ∧ N ∧ represents the tetradentate chelating ligand tpya. Coordination bonds are drawn in bold (Schematic drawing inspired by Jeon et al., 2013;Demir et al., 2015).
ligands on opposite sites of the plane defined by the three metal ions. This tripodal arrangement of the three oximato groups suggested (Nguyen et al., 2011(Nguyen et al., , 2016) that their replacement with dioximate groups to link multiple {Mn III 3 } units together might give oligomeric complexes rather than polymeric species. The linker of choice has been 3-phenyl-1,5-bis(pyridin-2-yl)pentane-1,5-dione (pdpdH 2 ; V in Figure 1). This ligand consists of two mpkoH (H with R = Me in Figure 1) groups that are connected by a benzyl unit.
The cation of 54 contains 4 {Mn III 3 (µ 3 -O)} 7+ subunits connected by 6 pdpd 2− ligands to give a rectangular {Mn III 3 } 4 supramolecular species with each triangular unit at a corner. One of the two 2.11-MeCO − 2 groups bridging each edge of 44a has been replaced by a bridging oximato group from a pdpd 2− ligand. The short and long sides of the rectangle are bridged by two and one pdpd 2− ligands, respectively. In addition, the 2-pyridyl N atoms of pdpd 2− have replaced the terminal py molecules of 44a. The local structure of each {Mn III 3 } unit of 54 is similar to that of the mpko − -containing complex 45a. As in 45a, the 3 bridging oximato groups are on the same side of the {Mn III 3 } plane, and this leads to a tetramer of triangles rather than a coordination polymer. Solid-state dc magnetic susceptibility studies show that the {Mn III 3 } subunits of 54 possess an S = 6 ground state. Ac susceptibility studies indicate the dodecanuclear cluster to be a tetramer of {Mn III 3 } SMMs. Magnetization vs. dc field studies on a single crystal of 54·xCH 2 Cl 2 give hysteresis loops below 1 K that exhibit exchange-biased QTM with a bias field of 0.19 T. Simulation of the loops reveals that each {Mn III 3 } subunit is exchange-coupled to the two neighboring subunits connected to it by the pdpd 2− bridging ligands, with an antiferromagnetic inter-{Mn III 3 } exchange interaction of J/K B = −0.008 cm −1 . Thus, the synthesis of 54 confirms the feasibility of connecting multiple Mn x SMMs to isolate a discrete supramolecular "cluster of SMMs" with only weak coupling between them. Supramolecular aggregates of the {Mn III 3 } SMMs 45 can also be prepared through the carboxylato sites of the staring materials, thus complementing the reports of aggregation at the oximato positions using dioximate linkers (Mowson et al., 2013). The latter strategy is not by substitution on preformed 45, but the former is. The 2:3 reaction of 45a and fumaric acid (fumH 2 ; W in Figure 1) in MeCN gives complex [Mn III 12 O 4 (fum) 6 (mpko) 12 ](ClO 4 ) 4 (55) in ∼35% yield, Equation (44). The reaction of 45a and bulky α-truxillic acid (atxH 2 ; X in Figure 1) The cation of 55 consists of a {Mn III 3 } 4 tetrahedron with a 4.1111 fum 2− (TT in Figure 2) bridging each edge to give an adamantane-like motif, which is a common unit in many supramolecular systems. Each carboxylate group of fum 2− ligand is 2.11 and thus bridges within a {Mn III 3 } subunit like the MeCO − 2 group that it replaces. Cluster 55 can be described as a tetramer of 45a held together by fum 2− linkers. Complex 56 is a {Mn III 3 } 2 dimer (see Figure 14 for its core); for steric reasons (two bulky Ph substituents per ligand) it is bridged by only two 4.1111 atx 2− groups (UU in Figure 2), with two additional atx 2− groups coordinated only through one carboxylate at each {Mn III 3 } subunit (VV in Figure 2). The mpko − ligands in 55 and 56 bind in the familiar 2.111 mode (MM with R = Me and M = Mn III in Figure 2). Detailed magnetic studies indicate that the S = 6 {Mn III 3 } subunits within the "tetramer" (55) and "dimer" (56) aggregates are only weakly interacting and retain the intrinsic SMM properties of the "monomeric" {Mn III 3 } complex (45a). Ac magnetic susceptibility studies of 55 and 56 show the appearance of out-of-phase signals below 3.0 K, indicating the slow relaxation of SMMs.
Another attractive ligand for the supramolecular synthesis of Mn SMM is 1,3-di(pyridin-2-yl)propane-1,3-dione dioxime (dpdH 2 ; Y in Figure 1). This ligand represents a fusion of two mpkoH (H with R = Me in Figure 1) groups at the methyl group. The single sp 3 C atom at the center of the molecule reduces the conformational flexibility with respect to pdpdH 2 (V in Figure 1), and the combination of a ∼109 • ditopic dioximate with a tritopic {Mn III 3 } unit should yield a {Mn III 3 } 2 "dimer" with three linker groups and parallel {Mn III 3 } planes (to facilitate study of quantum properties). Another parameter for the selection of dpdH 2 is the fact that binding by tridentate 2-pyridyloximate groups of three doubly deprotonated linkers would provide robustness and rigidity in the "dimer, " resulting in retention of the structure in solution. Moreover, the single sp 3 carbon atom was expected to propagate a very weak (but non-zero) superexchange interaction between the two {Mn III 3 } SMMs and therefore a small perturbation on a two-SMM system (Nguyen et al., 2015). The reaction of 44a with dpdH 2 and I 2 in a 2:3:6 molar ratio in EtOH/CH 2 Cl 2 gives [Mn III 6 O 2 (O 2 CMe) 6 (dpd) 3 ](I 3 ) 2 (57) in 22% yield, Equation (46). I 2 is added to stop reduction of some of the Mn III to Mn II , as sometimes observed in such reactions.

[Mn III
3 O(O 2 CMe) 6 (py) 3 ] + + 3 dpdH 2 EtOH/CH 2 Cl 2 − −−−−−−− → [Mn III 6 O 2 (O 2 CMe) 6 (dpd) 3 ] 2+ + 6 pyH + + 6 MeCO − 2 (46) The structure of the hexanuclear dication consists of two {Mn III 3 (µ 3 -O)} 7+ subunits linked by three dpd 2− ligands to give a {Mn III 3 } "dimer, " with the two {Mn III 3 } planes being parallel. Each trinuclear subunit is structurally similar to that of "monomer" 45a. The dianionic ligand adopts the 4.111111 coordination mode (WW in Figure 2), each 2-pyridyloximate subgroup bridging two Mn III ions belong to a {Mn III 3 } subunit. Solid-state dc and ac magnetic susceptibility studies indicate that each {Mn III 3 } subunit of the "dimer" is a separate SMM with an S = 6 ground state and that the two SMM subunits interact very weakly in a ferromagnetic fashion. The ferromagnetic nature of the coupling arises from a spin polarization mechanism FIGURE 14 | An extended view of the core of cluster 56. The symbol represents Mn III . The dashed curve represents the diatomic oximato groups of mpko − . The line represents the aromatic and aliphatic backbone of the atx 2− ligands. Coordination bonds are drawn in bold (Schematic drawing inspired by Mowson et al., 2013 through the central sp 3 carbon atom, caused by spin density delocalized into the dpd 2− ligand π systems from the Mn III d π magnetic orbitals. High-frequency EPR spectra on a single crystal of 57 exhibit signal splittings that indicate quantum superposition of two SMMs. Studies on toluene/MeCN (1:1 v/v) solutions display the same spectra, indicating that the "dimer" retains its structure and the weak inter-{Mn III 3 } coupling in solution. This is of particular importance in the currently intense efforts for the deposition of SMMs on surfaces and other substrates.

BRIEF SUMMARY AND OUTLOOK
We hope that this review has provided the readers with a "flavor" of the synthetic chemistry and reactivity studies of 3d-metal coordination clusters with SMM properties. Empirical routes (section The Years of Innocence: "Try and See" Exercises) and strategies (section The Years of Design: Strict and Less Strict Synthetic Strategies) have been discussed and critically analyzed through selected examples; often the distinction between an empirical route and a strict strategy is blurred. Particular emphasis has been given in the synthetic rationale behind the reactions and in the criteria for the selection of the metal ions and the ligands. Most strategies refer to Mn chemistry, because the relatively large number of unpaired electrons (four) and the large easy-axis anisotropy (negative D) for the Jahn-Teller elongated high-spin Mn III make this ion an ideal candidate for the construction of SMMs. The extremely low temperatures at which the polynuclear 3d-metal complexes operate as SMMs (<4.5 K) make these molecules not suitable for technological applications. Their future utility is in low-temperature and specialized applications that take full advantage of their unique molecular characteristics (size, monodispersity, crystallinity, good solubility, feasibility of altering their peripheral ligation, and well-defined quantum properties). However, the inability of polynuclear 3d-metal SMMs for every day applications does not underestimate their importance. These compounds are responsible for great revolutions in synthetic (for their preparation and reactivity) and physical (for their characterization and study) Inorganic Chemistry, a field that has undergone a second renaissance (the first was in the 50's and early 60's) after the second world war.
Most research efforts in the chemistry of molecular compounds exhibiting slow magnetization relaxation (at zero field) are now focused on 4f-and 5f-metal SIMs for reasons outlined in section Introduction. However, we do believe that there is still enough room for the improvement of the existing strategies and the development of new ones in the chemistry of 3d-metal clusters with SMM properties. We expect that the synthesis and the study of radical-bridged 3d-metal SMMs (section Synthesis of Radical-Bridged 3d-Metal SMMs) will be intensified in the future because of the potential of huge J-values. We also predict new developments in the supramolecular chemistry of 3d-metal SMMs (section Supramolecular Approaches in Manganese SMM Chemistry), since this is related to the linking of SMMs on surface and other substrates. A final point of synthetic interest is that the larger radial extension of the 4d and 5d orbitals (compared to the 3d ones) allows for stronger exchange coupling through diamagnetic bridging ligands. Since the strong exchange is a key factor in SMM research, we anticipate an increased synthetic activity in the area of 4d-and 5d-metal cluster SMMs (not covered in this review).
We conclude by pointing out that it is hoped that this review will be proven useful for inorganic chemists who are already active in SMM research or just enter into this area, and we shall be happy if the readers enjoy the review as we enjoyed writing it.

AUTHOR CONTRIBUTIONS
DM and EP jointly studied the literature, proposed some of the cited examples, prepared the figures, and typed the manuscript. SP developed the concept of the review and wrote the manuscript.