Multiple-decker and ring sandwich formation of manganese–benzene organometallic cluster anions: Mn_nBz_n⁻ (n = 1–5 and 18)

Abstract

Organometallic multiple-decker sandwich clusters are topics of great interest due to their unique electronic and magnetic properties originating from anisotropic structures. We report a joint anion photoelectron spectroscopic and computational study on a new family of manganese (Mn)–benzene (Bz) anionic clusters Mn_nBz_n⁻. In stark contrast to the most widely studied vanadium–Bz sandwich clusters, it is found that Mn_nBz_n⁻ (n = 1–5) clusters exhibit unprecedented multiple-decker structures with a tilted Mn–Bz stacking and a monotonically increasing behavior of their high spin multiplicities. Furthermore, a couple of closed ring forms of Mn₁₈Bz₁₈⁻ and its neutral state are computationally anticipated as an intriguing “cluster of Mn₁Bz₁ clusters” in which the neutral Mn₁₈Bz₁₈ has extremely high C_18h symmetry with an uncommon spin state of 2S + 1 = 55. The extensively delocalized electron environment of Mn₁₈Bz₁₈ allows the simple Hückel model to reveal the strong intra-atomic exchange interactions within the Mn 3d electrons.

---

1. Introduction

Molecular magnetism has received considerable attention due to potential technological applications for high-density information storage,¹ quantum computing,² and molecular spintronics.^3,4 A variety of single-molecule magnets (SMMs), whose individual molecules act as a magnet, have been developed since the discovery of the first SMM, a manganese (Mn)–oxo cluster [Mn₁₂].⁵ Mn-based SMMs have been most investigated due to the significant contribution of Mn electrons to their spin states and magnetic anisotropy.⁶ More recently, one-dimensional (1D) polymer chains called single-chain magnets (SCMs) have been demonstrated to show magnetic properties analogous to those of SMMs.^7,8 They can be viewed as 1D systems composed of high-spin magnetic units when compared with 0D SMMs.

A popular approach towards making better performing SMMs/SCMs is to realize a high-spin ground state with a large magnetic anisotropy. Major progress in this direction has been often made through appropriate modulation of existing systems. For example, a strategic choice of bridging ligands has been demonstrated to induce effective exchange interactions with a large anisotropy in the metal–ligand system.^9–12 On the other hand, their local electronic structures have to be characterized not only to understand their molecular magnetism but also to gain a concept for future SMM/SCM developments. At this point, there are several fruitful contributions from the field of cluster research.^13–16 They have shown that clusters are appropriate models for nanoscale magnetism, which can be described as a function of cluster size.

Clusters are also expected as a potential building block for designer SMMs/SCMs. Organometallic sandwich clusters are topics of great interest due to their unique electronic and magnetic properties originating from anisotropic structures.^17,18 For example, vanadium (V) atoms and benzene (Bz) molecules are piled up alternately to form multiple-decker sandwich structures (V_nBz_m; m = n − 1, n, and n + 1),¹⁹ exhibiting a monotonic increase in their magnetic moments with increasing number of sandwich layers.^20,21 While this observation has stimulated numerous theoretical attempts to explore metal–ligand sandwich based SMM/SCMs,^22–26 experimental approaches remain challenging.

Here we report the size-dependent geometric and electronic structures on a new class of Mn–Bz cluster anions, Mn_nBz_n⁻ (n = 1–5 and 18), by means of anion photoelectron spectroscopy in combined with theoretical calculations. We first show experimental and theoretical evidences of multiple-decker formation for Mn_nBz_n⁻ (n = 1–5) through quite different stacking forms that contrast with the linear V–Bz clusters. Then we computationally investigate extremely high-spin ring Mn₁₈Bz₁₈^−/0 clusters, whose motifs can intriguingly be recognized as a “cluster of Mn₁Bz₁ clusters”.

2. Experimental and computational methods

2.1 Experimental methods

Mn_nBz_m⁻ clusters were generated in the gas phase using a laser vaporization and supersonic expansion method.^27,28 Briefly, a Mn metal rod sample was ablated with the second harmonic (532 nm) light pulses of a Nd³⁺:YAG laser at a 10 Hz repetition rate. The Mn vapor was reacted with helium-diluted Bz gas (at ∼3 atm) that was introduced using a pulsed valve (General Valve Series 9), and then very intense supersonic pulses of helium carrier gas at 40 atm, obtained by using a high-pressure Even–Lavie pulsed valve,²⁹ were used to cool the hot species mixture of atoms and clusters. The resulting cluster anions were then extracted to a linear time-of-flight mass spectrometer. The anions were mass-selected and photodetached using the second (532 nm, 2.33 eV), third (355 nm, 3.49 eV), and fourth (266 nm, 4.66 eV) harmonic outputs of another Nd³⁺:YAG laser. The kinetic energies of the detached photoelectrons were measured using a magnetic bottle-type photoelectron spectrometer. The electron binding energies were then obtained using the energy-conservation relationship hν = EBE + EKE, where hν, EBE, and EKE are the photon energy, electron binding energy, and electron kinetic energy, respectively.

2.2 Computational methods

In order to explore the structures of Mn–Bz clusters, density functional theory (DFT) calculations were performed at the unrestricted BP86^30,31 level, implemented in the Gaussian 09 program package³² using the def2-TZVP³³ basis sets. The Molekel program³⁴ was used for molecular visualization. The vertical detachment energies (VDEs) of each of the optimized anions were calculated to compare with the experimental values. Details of the computational method are provided in the ESI.†

It is important to correctly evaluate the energy orderings of different spin states of Mn_nBz_n⁻ clusters, as it has been pointed out that the current DFT methods for transition metal-based organometallic systems have a specific problem that the spin multiplicity of a global-minimum structure depends on the functional employed.³⁵ In this study, Mn₁Bz₁^−/0 were optimized using various functionals to calculate the ADE/VDE values of Mn₁Bz₁⁻, and the BP86 was adopted on the basis of the reasonable agreements with the experimental values (see Tables S2 and S3 in the ESI†).

3. Results and discussion

3.1 Mass distribution of Mn_nBz_m⁻

Fig. 1 shows a typical mass spectrum of gas-phase produced Mn_nBz_m⁻ cluster anions. It was taken by using a pulse duration of 280 μs for the Bz pulsed valve. Changes of the duration were found to significantly affect both the mass intensity and distribution (vide infra). The mass assignments were reconfirmed by the experiment using d₆-benzene (C₆D₆; see Fig. S1 in the ESI†). The spectrum indicates that various compositions of Mn_nBz_m⁻ are generated; in particular, Mn_nBz_n⁻ and Mn_nBz_n–1⁻ are more intense than Mn_nBz_n+1⁻ and the other Mn_nBz_m⁻ with m ≤ n − 2. Other mass species are mostly classified into either pure Mn_n⁻ (marked by ■) or incomplete Mn–Bz cluster anions (marked by ●) which have a single Bz molecule and which cannot be assigned to Mn_nBz_m⁻.

Fig. 1 Mass spectra of Mn_nBz_m⁻. The peak assignments for (n, m)⁻ = (n, n − 1)⁻, (n, n)⁻, and (n, n + 1)⁻ are indicated with vertical lines. Several dominant (n, m)⁻ clusters with m ≤ n − 2 are also labeled. Coexistent naked Mn_n⁻ clusters and incomplete Mn–Bz cluster anions are marked by ■ and ●, respectively.

We assessed how the mass distribution is affected by kinetic conditions (i.e., Bz introduction time) in the cluster source, which were qualitatively controlled by changing the duration of the Bz pulsed valve. Relative abundances of Mn_nBz_m⁻ with different valve durations employed are depicted in Fig. 2, showing that the production of Mn_nBz_m⁻ with large m (e.g., m = n) is much more diminished by limiting the Bz introduction time than that of other Bz-poor Mn_nBz_m⁻ (Mn₁Bz₁⁻, Mn₂Bz₁⁻, and m ≤ n − 2 for n ≥ 3). In contrast, the mass distribution of the Bz-poor Mn_nBz_m⁻ (for each n) is not shifted irrespective of the experimental conditions. Taken into account that the mass distribution of kinetically generated species must be controlled by the concentration of reactants,³⁶ these characteristics suggest that the production of most Mn_nBz_m⁻ (m = n − 1, n, and n + 1) can be ruled by a kinetic process while the others (i.e., Mn₁Bz₁⁻, Mn₂Bz₁⁻, and m ≤ n − 2 for n ≥ 3) would experience a thermodynamic cluster growth. The difference in cluster growth mechanisms could also cause (co)existence of energetically higher structural isomers for some sizes of Mn_nBz_n⁻, which is discussed in the following subsection.

Fig. 2 Relative abundance of Mn_nBz_m⁻ clusters with different conditions of the cluster generation source employed. Each Mn_nBz_m⁻ is denoted as (n, m)⁻. To guide the eyes, Mn_nBz_n⁻ clusters are colored in red. The generation conditions were qualitatively controlled by changing the duration of the Bz pulsed valve. The average uncertainty on the vertical axis (estimated from the standard deviation) is ±20%. Note that the photoelectron spectra in the present study were measured under the condition of ≥280 μs unless otherwise noted.

The present study hereafter concentrates on the structural and electronic properties of Mn_nBz_n⁻ (i.e., m = n) according to their high abundances in the mass spectra.

3.2 Photoelectron spectra and structural assignments of Mn_nBz_n⁻ (n = 1–3)

Fig. 3a shows the photoelectron spectra of Mn_nBz_n⁻ (n = 1–3) taken with 355 nm radiation. The additional spectra measured with different wavelengths (532 nm and 266 nm; Fig. S2 in the ESI†) exhibit a photon energy-independent feature. Each peak maximum in the spectra provides VDE and the first VDEs of Mn₁Bz₁⁻, Mn₂Bz₂⁻, and Mn₃Bz₃⁻ are defined from the 532 nm spectra as, respectively, 0.76 ± 0.02, 1.21 ± 0.02, and 1.11 ± 0.01 eV. In addition, higher VDEs of other spectral peaks are summarized in Table 1. The Mn₁Bz₁⁻ spectrum has a vibrational feature with a spacing of 240 ± 40 cm⁻¹ (inset of Fig. 3a), which is assignable to a stretching mode between Mn and Bz in the excited state of neutral Mn₁Bz₁. The first sharp peak of Mn₁Bz₁⁻ also determines an adiabatic detachment energy (ADE) of 0.58 ± 0.07 eV, which reflects the electron affinity of Mn₁Bz₁.

Fig. 3 (a) Photoelectron spectra of Mn_nBz_n⁻ (n = 1–3) taken with 355 nm (3.49 eV) radiation. The calculated VDEs of the plausible structures are indicated below the spectra with vertical bars. Bold-faced labels right to the VDE bars are denoted to identify the structures. For Mn₁Bz₁⁻, a vibrational progression is also shown. (b) Front and side views of the most plausible structures for each Mn_nBz_n⁻ (n = 1–3) with their symmetries and electronic states. Typical bond lengths are indicated in Å. Bonds are visible for C–C, C–H, and Mn–C with distances of <2.4 Å only to guide the eyes. They do not necessarily represent single bonds. (c) Structural isomers with possible contributions to the photoelectron spectra and their symmetries, electronic states, and energies relative to 2a (Mn₂Bz₂⁻) and 3a (Mn₃Bz₃⁻).

Table 1 Experimental and calculated VDEs (in eV) for Mn_nBz_n⁻ (n = 1–5)

Cluster	VDE
	Expt.^a	Calc.^b
a Obtained from the 532 nm spectra (Fig. S2 in the ESI) unless otherwise noted because photoelectron resolution is better at smaller photon energies. b Labels of anions (Fig. 3b, c and 4b) and final electronic states are denoted in parentheses. For n ≥ 3, higher VDEs (indicating excited final states) are not shown because several TD-DFT calculations met fatal problems such as “Fatal Problem: The smallest beta delta epsilon is 0.54335749D-03”. c Obtained from the 266 nm spectra (Fig. S2 in the ESI). d Obtained from the 355 nm spectra.
Mn1Bz1^−	0.76 ± 0.02 3.47 ± 0.03^c	0.69 (1 → ^4A2), 1.22 (1 → ^6A2) 2.98 (1 → ^4A1)
Mn2Bz2^−	1.21 ± 0.02	1.10 (2a → ^5A)
	1.75 ± 0.07	1.47 (2c → ^9Ag), 1.51 (2c → ^9Ag), 1.58 (2b → ^7A′), 1.62 (2c → ^9Au), 1.64 (2a → ^7A), 1.76 (2b → ^9A′), 1.85 (2c → ^9Au), 2.01 (2a → ^7A), 2.09 (2b → ^7Au)
	2.60 ± 0.06^d	2.72 (2a → ^7A)
Mn3Bz3^−	1.11 ± 0.01	1.25 (3b → ^8A)
	1.75 ± 0.03	1.48 (3c → ^2A), 1.88 (3a → ^10A)
Mn4Bz4^−	2.16 ± 0.08	2.25 (4 → ^15A)
Mn5Bz5^−	2.16 ± 0.07	2.24 (5 → ^18A)

---

Our DFT calculations found the most plausible structures for Mn_nBz_n⁻ (n = 1–3) as depicted in Fig. 3b. The calculated VDEs of these structures are in good agreement with the experimental data (Fig. 3a and Table 1). Mn₁Bz₁⁻ was identified as the quintet spin state having a C_2v (pseudo-C_6v) half-sandwich structure 1 (Fig. 3b). The other spin states are higher in energy and are ruled out due to inconsistency of VDEs (Table S3 in the ESI†). This observation of the lowest-lying 1 is in line with the thermodynamic behavior of Mn₁Bz₁⁻ (Fig. 2). For neutral Mn₁Bz₁, the quartet state with perfect C_6v symmetry (Fig. S3 in the ESI†) was obtained and the calculated ADE of Mn₁Bz₁⁻ (0.59 eV) agrees well with the experimental value. Both the higher VDEs of 1.22 and 2.98 eV are ascribed to electronic excited states of the neutral Mn₁Bz₁.

Both Mn₂Bz₂⁻ and Mn₃Bz₃⁻ were identified as 2a and 3a (Fig. 3b), respectively. Both exhibit alternating sandwich structures with C₁ (pseudo-C_s) symmetry, where Mn atoms and Bz molecules are alternately stacked. They adopt relatively high spin states of 2S + 1 = 6 (2a) and 11 (3a). Energetically closer 2b and 3b (Fig. 3c) might be minor products which contribute to the photoelectron spectra. Our extensive searches further found structural isomers (Fig. S4 and S5 as well as Tables S4 and S5, ESI†), some of which are even lower in energy than 2a or 3a. However, the observation of the alternating sandwich structures will be reasonable if kinetic processes dominates their production. To confirm whether both 2a and 3a were kinetically produced, the photoelectron spectra of Mn₂Bz₂⁻ and Mn₃Bz₃⁻ were measured under a different kinetic condition. As shown in Fig. S6 in the ESI,† the limited introduction of Bz gas only diminishes the photoelectron features related to the alternating sandwiches (i.e., 2a, 2b, 3a, and 3b). This change apparently corresponds the decreasing population of Mn₂Bz₂⁻ and Mn₃Bz₃⁻ shown in Fig. 2, thus affirming that Mn₂Bz₂⁻ and Mn₃Bz₃⁻ are mostly the kinetically generated alternating sandwich clusters. In contrast, the remaining spectral features at higher binding energies (>2 eV) indicate coexistence of structural isomers for both Mn₂Bz₂⁻ and Mn₃Bz₃⁻, which are determined as non-alternating structures 2c and 3c (Fig. 3c), respectively. In particular, 3c is a so-called rice-ball cluster with a metal core surrounded by Bz molecules. It is generally concluded that such rice-ball clusters would become dominant under the thermodynamic conditions regarding Bz-poor (e.g., n = 1) Mn_nBz_m⁻, which are not capable of taking alternating sandwich structures due to smaller m (<n − 1).

3.3 Photoelectron spectra and structural assignments of Mn_nBz_n⁻ (n = 4 and 5)

The photoelectron spectra of Mn₄Bz₄⁻ and Mn₅Bz₅⁻ are shown in Fig. 4a. The VDEs of Mn₄Bz₄⁻ and Mn₅Bz₅⁻ are determined to be 2.16 ± 0.08 and 2.16 ± 0.07 eV, respectively, according to the peak maxima of the broad features in the 532 nm spectra (Fig. S2 in the ESI†). As the kinetic generation is also the case for Mn₄Bz₄⁻ and Mn₅Bz₅⁻ (Fig. 2), we focus on the possibilities of multiple-decker sandwich structures. Assuming a monotonic increase of the Mn_nBz_n⁻ spin multiplicities, the calculations found Mn₄Bz₄⁻ (4) and Mn₅Bz₅⁻ (5) with 2S + 1 = 14 and 17, respectively, as depicted in Fig. 4b. It is noted that both 4 and 5, in addition to 3a, have an unprecedented “tilted” form, representing continuous η² binding of Mn₁(η⁶-Bz)₁ units. Both the calculated VDEs (Table 1 and Table S6 in the ESI†) are in line with the experimental values, showing the high spin-state multiple-decker formation of Mn₄Bz₄⁻ and Mn₅Bz₅⁻.

Fig. 4 (a) Photoelectron spectra of Mn₄Bz₄⁻ and Mn₅Bz₅⁻ taken with 355 nm (3.49 eV) radiation. (b) Optimized structures of Mn₄Bz₄⁻ and Mn₅Bz₅⁻ in a multiple-decker sandwich fashion, with their symmetries and electronic states. The indications are shown in the same manner as Fig. 1.

3.4 Computational anticipation of ring-structured Mn₁₈Bz₁₈^−/0

According to the bottom-up approach presented, the size evolution of the Mn_nBz_n⁻ clusters was assumed to be terminated at a finite number n, in stark contrast to the linear V–Bz system.¹⁹ Here we computationally anticipate a couple of ring-structured Mn₁₈Bz₁₈^−/0, as it was estimated from the extrapolation of the structures 3a, 4, and 5 that the cyclization of Mn_nBz_n⁻ occurs at n = 18. Starting from a C_18h geometry, Mn₁₈Bz₁₈^−/0 were optimized as given in Fig. 5a (details are summarized in Table S7 in the ESI†). Surprisingly, the neutral Mn₁₈Bz₁₈ retains C_18h symmetry. The Mn₁₈Bz₁₈^−/0 adopt extremely high spin states: the anion and neutral take 2S + 1 = 54 and 55, respectively. In both cases, the calculations on the neighboring spin states found higher energies or even met convergence problems within C_18h or C₂ symmetry. In fact, both the spin state and the cohesive energies per Mn₁(η⁶-Bz)₁ unit of Mn₁₈Bz₁₈⁻ are very consistent with the extrapolation from those of Mn_nBz_n⁻ (n = 1–5), as shown in Fig. 5b and Fig. S7, respectively. These findings could suggest synthetic feasibility of the large Mn₁₈Bz₁₈^−/0 rings via possible exothermic aggregations of small Mn_nBz_n⁻.

Fig. 5 (a) Top and side views of the optimized structures for Mn₁₈Bz₁₈^−/0 ring clusters with their symmetries and electronic states. Typical bond lengths are indicated in Å (those of the neutral are given in parentheses). (b) Size-dependent spin multiplicities of Mn_nBz_n⁻ (n = 1–5 and 18). Because the structure determined for Mn₂Bz₂⁻ (2a) has a different bonding scheme from others, 2b is explicitly indicated with a white square, even though it is a minor product. The linear extrapolation of the spin multiplicities for tilted multiple-decker sandwich structures is expressed by the red line.

Fig. 6a shows the frontier Kohn–Sham (KS) orbitals in the α spin part of the neutral Mn₁₈Bz₁₈. The α-HOMO has an extensively delocalized bonding character, and can be divided into contributions from the α-HOMOs of each quartet-state Mn₁Bz₁ unit (inset of Fig. 6a), as if the Mn₁₈Bz₁₈ exists as a cluster of the Mn₁Bz₁ clusters. The degenerate α-HOMO−1 and α-HOMO−2, and several other α-valence orbitals (6 sets of degenerate and 1 nondegenerate; see Fig. S8 in the ESI†) are isomorphous to the α-HOMO. Interestingly, each delocalized Mn₁₈Bz₁₈ orbital allows us to define the number of its orbital nodes k, analogously to π conjugated hydrocarbon systems. Particularly unusual for Mn₁₈Bz₁₈ is that the smaller number of nodes is rather associated with the energetically higher orbital. These characteristics are further highlighted by comparing with the frontier orbitals of [18]annulene (C₁₈H₁₈), which has a 18-membered ring structure with 18 π electrons (Fig. 6b and Fig. S9 in the ESI†).³⁷ An intelligible explanation of such delocalized electron systems can be provided by a simple Hückel model: for cyclic conjugated systems, each molecular orbital energy ε_k is given as

where the Hückel parameters α and β are employed. The number of basis functions N is set to 18 in both cases of Mn₁₈Bz₁₈ and C₁₈H₁₈, in which fragment (or atomic) orbitals of each Mn₁Bz₁ and C are treated, respectively. To apply this model, KS orbital energies were plotted against k and fit by eqn (1); namely, the plots for the Mn₁₈Bz₁₈ α-HOMO and lower-lying isomorphous orbitals are shown in Fig. 6c. The same analysis was done for the C₁₈H₁₈ frontier orbitals as plotted in Fig. 4d. From the figure the simple Hückel model can describe the Mn₁₈Bz₁₈ plot but claims a positive β which contrasts the conventional cases as demonstrated for C₁₈H₁₈.

Fig. 6 (a) Frontier Kohn–Sham (KS) orbital pictures in the majority (α) spin part of the Mn₁₈Bz₁₈ neutral cluster. The number of nodes k is given for each delocalized orbital. Inset shows top and side views of the partially magnified α-HOMO. The isosurface value of 0.01 was adopted unless otherwise noted. (b) KS orbital pictures of [18]annulene (C₁₈H₁₈) with an isosurface value = 0.01, having the same k values as depicted in (a) for comparison. (c) KS orbital energies plotted against k for the Mn₁₈Bz₁₈ α-HOMO and lower-lying isomorphous orbitals, and (d) those for the C₁₈H₁₈ frontier orbitals composed of delocalized π electrons. Blue lines are obtained by a curve-fitting analysis using eqn (1). Note that two plots for k = 3 and 6 are excluded from (d) because the orbitals are not degenerate under the D_3d symmetry of C₁₈H₁₈ in our geometry optimization.

In the following, we indicate that the positive β of Mn₁₈Bz₁₈ originates from its strong intra-atomic exchange interactions within the Mn 3d electrons. Let us recall the definition of the Hückel parameter , where two fragments (or atoms) having basis functions χ_i and χ_j are next to each other. The Hückel Hamiltonian Ĥ is composed of the kinetic term and the electrostatic potentials by nuclei and other electrons; as its resonance integral name implies, β is in general not significantly affected by two-electron repulsive interactions (the last term in Ĥ; i.e., , where r_ij is the distance between two electrons), thus resulting in β < 0. Therefore the positive β implies significant contributions from the two-electron interactions in the Mn₁₈Bz₁₈ frontier orbitals. According to previous studies,^21,38,39 the intra-atomic exchange interactions within the metal 3d orbitals induce localized positive and negative spin densities on each metal atom and benzene molecule, respectively, thus stabilizing the ferromagnetic spin ordering in multiple-decker V_nBz_n+1. Indeed, this should be also the case for Mn₁₈Bz₁₈ due to a pretty large negative spin density (ρ_s = −0.52) for each Bz and the positive spin density of ρ_s = 3.52 per a Mn atom.

3.5 Magnetism of Mn₁₈Bz₁₈ in solid-state physics

The periodic nature of Mn₁₈Bz₁₈ could allow further discussion on its magnetism from a solid-state point of view. When the aforementioned intra-atomic exchange interactions are restated as the on-site Coulomb repulsion U in the Hubbard model,⁴⁰ the Mn₁₈Bz₁₈ could be then treated as a U ≫ t case (where t is a transfer integral). Despite the high spin state of the Mn₁₈Bz₁₈, such U ≫ t systems (often referred to as Mott insulators) are believed to usually exhibit an antiferromagnetic ordering of spins.⁴¹ In this sense, the magnetism within the Mn₁₈Bz₁₈ cluster might be uncommonly understood from Heisenberg’s original picture of ferromagnetism,⁴² which is based on Coulomb interactions combined with the Pauli exclusion principle. Interestingly, several attempts^43,44 have employed the Hubbard model to prove that, in certain lattices, the spin-independent Coulomb interactions can cause a ferromagnetic alignment. Taken together, the low-dimensional and periodic nature of the Mn₁₈Bz₁₈ cluster could be therefore exploited as a substantial model for investigating nanoscale magnetism, which allows not only the bottom-up (molecular) but also top-down (solid-state) approaches.

4. Conclusions

We have investigated a series of Mn–Bz cluster anions, Mn_nBz_n⁻ (n = 1–5), through a joint photoelectron spectroscopic and DFT study. We have shown that Mn_nBz_n⁻ clusters form multiple-decker sandwich structures with the characteristic tilted Mn–Bz stacking and the high-spin states. Furthermore, a couple of ring-structured Mn₁₈Bz₁₈^−/0 have been found with the aid of a linear extrapolation from the Mn_nBz_n⁻ clusters. Importantly, the neutral Mn₁₈Bz₁₈ cluster exhibits several striking features, namely, the high symmetry (C_18h) structure, the high spin state (2S + 1 = 55), and the extensively delocalized electrons.

Lastly, the completely planar (2D) structure of Mn₁₈Bz₁₈ at a finite size, when compared to infinite sandwich chain cases, might be advantageous for controlling the cluster’s surface morphology (e.g., adsorption states), whose importance in keeping the cluster’s functionality has been demonstrated through soft-^45,46 and reactive-landing techniques.⁴⁷ We hope that the current work will stimulate synthetic and functionalization efforts on Mn₁₈Bz₁₈^−/0 as a building block of future SMMs/SCMs.

Acknowledgements

We are grateful to Profs. S. Yabushita (Keio University) and K. Kanoda (The University of Tokyo) for fruitful discussion. We acknowledge Prof. S. Saito (Institute for Molecular Science) and Dr K. Iwahashi (Research Center for Computational Science) for their technical assistance with the computations. This work is partly supported by JSPS KAKENHI of Grant-in-Aids for Scientific Research (A) Grant Number 15H02002. T. M. acknowledges the financial support of Keio Leading-edge Laboratory of Science and Technology (KLL) PhD. Program Research Grant 2015. The computations were performed using Research Center for Computational Science, Okazaki, Japan.

References

1. A. Ardavan, O. Rival, J. J. L. Morton, S. Blundell, A. M. Tyryshkin, G. A. Timco and R. E. P. Winpenny, Phys. Rev. Lett., 2007, 98, 057201.
2. M. N. Leuenberger and D. Loss, Nature, 2001, 410, 789.
3. M.-H. Jo, J. E. Grose, K. Baheti, M. M. Deshmukh, J. J. Sokol, E. M. Rumberger, D. N. Hendrickson, J. R. Long, H. Park and D. C. Ralph, Nano Lett., 2006, 6, 2014.
4. L. Bogani and W. Wernsdorfer, Nat. Mater., 2008, 7, 179.
5. R. Sessoli, D. Gatteschi, A. Caneschi and M. A. Novak, Nature, 1993, 365, 141.
6. D. Gatteschi, R. Sessoli and J. Villain, Molecular Nanomagnets, Oxford University Press, New York, 2006.
7. R. Clérac, H. Miyasaka, M. Yamashita and C. Coulon, J. Am. Chem. Soc., 2002, 124, 12837.
8. W.-X. Zhang, R. Ishikawa, B. Breedlove and M. Yamashita, RSC Adv., 2013, 3, 3772.
9. G. S. Papaefstathiou, A. Escuer, C. P. Raptopoulou, A. Terzis, S. P. Perlepes and R. Vicente, Eur. J. Inorg. Chem., 2001, 1567.
10. C. Dendrinou-Samara, M. Alexiou, C. M. Zaleski, J. W. Kampf, M. L. Kirk, D. P. Kessissoglou and V. L. Pecoraro, Angew. Chem., Int. Ed., 2003, 42, 3763.
11. D. E. Freedman, W. H. Harman, T. D. Harris, G. J. Long, C. J. Chang and J. R. Long, J. Am. Chem. Soc., 2010, 132, 1224.
12. F. Habib, O. R. Luca, V. Vieru, M. Shiddiq, I. Korobkov, S. I. Gorelsky, M. K. Takase, L. F. Chibotaru, S. Hill, R. H. Crabtree and M. Murugesu, Angew. Chem., Int. Ed., 2013, 52, 11290.
13. M. R. Pederson and S. N. Khanna, Phys. Rev. B: Condens. Matter Mater. Phys., 1999, 60, 9566.
14. J. Jellinek, P. H. Acioli, J. Garciá-Rodeja, W. Zheng, O. C. Thomas and K. H. Bowen, Phys. Rev. B, 2006, 74, 153401.
15. S. Peredkov, M. Neeb, W. Eberhardt, J. Meyer, M. Tombers, H. Kampschulte and G. Niedner-Schatteburg, Phys. Rev. Lett., 2011, 107, 233401.
16. V. Zamudio-Bayer, K. Hirsch, A. Langenberg, M. Kossick, A. Ławicki, A. Terasaki, B. v. Issendorff and J. T. Lau, J. Chem. Phys., 2015, 142, 234301.
17. A. Nakajima and K. Kaya, J. Phys. Chem. A, 2000, 104, 176.
18. A. Nakajima, Bull. Chem. Soc. Jpn., 2013, 86, 414.
19. K. Hoshino, T. Kurikawa, H. Takeda, A. Nakajima and K. Kaya, J. Phys. Chem., 1995, 99, 3053.
20. K. Miyajima, A. Nakajima, S. Yabushita, M. B. Knickelbein and K. Kaya, J. Am. Chem. Soc., 2004, 126, 13202.
21. K. Miyajima, S. Yabushita, M. B. Knickelbein and A. Nakajima, J. Am. Chem. Soc., 2007, 129, 8473.
22. V. V. Maslyuk, A. Bagrets, V. Meded, A. Arnold, F. Evers, M. Brandbyge, T. Bredow and I. Mertig, Phys. Rev. Lett., 2006, 97, 097201.
23. H. Xiang, J. Yang, J. G. Hou and Q. Zhu, J. Am. Chem. Soc., 2006, 128, 2310.
24. Y. Mokrousov, N. Atodiresei, G. Bihlmayer, S. Heinze and S. Blügel, Nanotechnology, 2007, 18, 495402.
25. H. Weng, T. Ozaki and K. Terakura, J. Phys. Soc. Jpn., 2008, 77, 014301.
26. X. Yao, X. Zhang and J. Wang, Int. J. Quantum Chem., 2015, 115, 607.
27. T. Masubuchi, K. Ohi, T. Iwasa and A. Nakajima, J. Chem. Phys., 2012, 137, 224305.
28. T. Masubuchi, T. Iwasa and A. Nakajima, J. Chem. Phys., 2014, 141, 214304.
29. U. Even, J. Jortner, D. Noy, N. Lavie and C. Cossart-Magos, J. Chem. Phys., 2000, 112, 8068.
30. J. Perdew, Phys. Rev. B: Condens. Matter Mater. Phys., 1986, 33, 8822.
31. A. D. Becke, Phys. Rev. A: At., Mol., Opt. Phys., 1988, 38, 3098.
32. M. J. Frisch, et al., Gaussian 09, Revision D.01, (see ESI† for full citation).
33. F. Weigend and R. Ahlrichs, Phys. Chem. Chem. Phys., 2005, 7, 3297.
34. U. Varetto, Molekel 5.4.0.8, Swiss National Supercomputing Centre, Manno, Switzerland, 2009.
35. L. Horváthová, M. Dubecký, L. Mitas and I. Štich, Phys. Rev. Lett., 2012, 109, 053001.
36. D. N. Shin, Y. Matsuda and E. R. Bernstein, J. Chem. Phys., 2004, 120, 4150.
37. F. Sondheimer, R. Wolovsky and Y. Amiel, J. Am. Chem. Soc., 1962, 84, 274.
38. T. Yasuike and S. Yabushita, J. Phys. Chem. A, 1999, 103, 4533.
39. A. Goto and S. Yabushita, Chem. Phys. Lett., 2008, 454, 382.
40. J. Hubbard, Proc. R. Soc. A, 1963, 276, 238.
41. P. W. Anderson, Phys. Rev., 1959, 115, 2.
42. W. Heisenberg, Z. Phys., 1928, 49, 619.
43. A. Mielke, J. Phys. A: Math. Gen., 1991, 24, 3311.
44. H. Tasaki, Phys. Rev. Lett., 1995, 75, 4678.
45. S. Nagaoka, T. Matsumoto, E. Okada, M. Mitsui and A. Nakajima, J. Phys. Chem. B, 2006, 110, 16008.
46. S. Nagaoka, T. Matsumoto, K. Ikemoto, M. Mitsui and A. Nakajima, J. Am. Chem. Soc., 2007, 129, 1528.
47. S. Nagaoka, K. Ikemoto, K. Horiuchi and A. Nakajima, J. Am. Chem. Soc., 2011, 133, 18719.

---

Footnote

† Electronic supplementary information (ESI) available: Full citation of ref. 32, computational details, additional photoelectron spectra of Mn_nBz_n⁻ (n = 1–5), all optimized structures for Mn_nBz_n⁻ (n = 1–5), Kohn–Sham orbital energy levels of Mn_nBz_n (n = 1 and 18) and [18]annulene, cohesive energies of Mn_nBz_n⁻ (n = 1–5 and 18), comparison of calculations employing different DFT functionals, and Cartesian coordinates. See DOI: 10.1039/c6cp05380g

---