A Borane Sandwich Analogue of Ferrocene

ABSTRACT

Sandwich complexes built from planar aromatic ligands are a central class of compounds in organometallic chemistry, yet a boron‐ring analogue of ferrocene has remained elusive despite decades of work. Here, an extensive global exploration of the FeB₁₀H₂₀ potential energy surface identifies the homoleptic iron sandwich η⁵,η⁵‐Fe(B₅H₁₀)₂ as the global minimum for this composition. This structure constitutes a fully inorganic boron‐ring analogue of ferrocene.

Iron coordination reverses the stability ordering of the borane framework by stabilizing borane rings over the preferred nido motifs, thereby enabling η⁵ coordination of two borane ligands. Bonding, magnetic response, and real‐space analyses indicate that the electronic structure is dominated by ligand‐centered delocalization, reorganized and stabilized upon Fe (II) coordination, and that this delocalization is shaped primarily by multicenter B─H─B bonding rather than direct B─B σ bonding. Ligand‐redistribution energetics further indicate that Cp‐mediated routes do not provide a thermodynamic driving force toward the homoleptic complex and, together with known temperature constraints for borane anions and borane‐ring rearrangements, support a Cp‐free, temporally decoupled capture─then─anneal strategy. Beyond identifying η⁵,η⁵‐Fe(B₅H₁₀)₂ as a plausible synthetic target, these results show that transition‐metal coordination can invert the stability preferences of electron‐deficient boron frameworks and thereby enable aromatic borane ligands that are otherwise in accessible in isolation.

The first ferrocene analogue with two boron‐based ligands is identified through a global exploration of the FeB₁₀H₂₀ potential energy surface. The η⁵,η⁵‐Fe(B₅H₁₀)₂ complex emerges as the global minimum, showing that metal coordination inverts borane stability and enables aromatic boron rings inaccessible in isolation.

1. Introduction

The structural elucidation of ferrocene, Fe(η⁵‐C₅H₅)₂ (FeCp₂; Cp = cyclopentadienyl) [1, 2, 3, 4, 5], marked a turning point in organometallic chemistry [6, 7] by establishing the sandwich architecture and showing how delocalized metal–ligand interactions can stabilize transition‐metal centers in unprecedented ways. This discovery initiated the broad family of metallocenes and sandwich complexes, in which the interplay between metal d orbitals and the π‐aromatic character of the coordinating rings governs structure and reactivity [8, 9, 10, 11, 12, 13, 14]. Subsequent isolobal and isoelectronic ligand substitutions extended this paradigm to heteroelement‐containing sandwiches [15, 16, 17, 18], including borazine, boroxine, phospholyl, silolyl, and borolyl ligands, and thereby broadened the range of accessible electronic structures and bonding patterns [19, 20, 21]. Within this context, boron‐based analogues are particularly interesting because the electron‐deficient and multicenter bonding of boranes can support coupled σ and π delocalization, which in principle may sustain metal–boron interactions analogous to those in ferrocene. However, the extension of the ferrocene motif to genuinely aromatic boron‐based rings has not yet been achieved.

Weiss and Grimes [22] reported (η⁵‐C₅H₅)FeB₅H₁₀, a mixed Cp–borane sandwich complex that represents a partial boron‐based variation of ferrocene, identifying a dynamic equilibrium between an η³‐bound species (I) and the thermodynamically preferred η⁵ isomer (II). Although crystallographic data were unavailable, the near‐planarity of the B₅H₅ fragment in the isoelectronic η⁵‐[B₅H₁₀]⁻ ring was inferred from ¹H/¹¹B NMR symmetry patterns [23, 24]. Ferrer et al. [21] later supported this structural assignment, showing that the Fe(η⁵‐C₅H₅)⁺ fragment drives the isomerization and that the η⁵ geometry corresponds to the global minimum. In parallel, Co‐ and Ni‐containing metallaboranes were experimentally established as cage‐type boron frameworks (see Scheme 1) [25]. More recently, global‐minimum screenings predicted B₇Be₆B₇ as a homoleptic inorganic sandwich [26], showing that fully inorganic sandwich architectures are viable in principle. Scheme 1 places these developments in context, from early heteroleptic ferraboranes to fully boron‐based sandwich architectures.

SCHEME 1.

From early organometallic‐borane hybrids to fully inorganic boron‐based sandwich complexes.

A key implication of the work by Ferrer et al. is that metal coordination can reverse the intrinsic stability preference of the borane framework, stabilizing a pentagonal ring and allowing it to function as a π‐delocalized ligand. This raises a central question: can two such borane rings coordinate simultaneously to a single metal center to form a true carbon‐free sandwich architecture? Whereas B₇Be₆B₇ contains a homoleptic boron‐ring sandwich supported by a main‐group Be₆ core, an (η⁵‐B₅H₁₀)₂Fe complex would constitute the first homoleptic boron‐ring sandwich centered on a transition metal. To examine this possibility, we carried out a systematic exploration of the FeB₁₀H₂₀ potential energy surface (PES). Our results identify the ferraborane sandwich as the global minimum structure, indicating that metal coordination can overcome the intrinsic instability of isolated cyclic C _{5 v} [B₅H₁₀]⁻ fragments. This finding establishes a true boron‐based ferrocene analogue and, more generally, shows that metal coordination can stabilize ligand frameworks that are not preferred in isolation.

2. Results and Discussion

2.1. Coordination‐Induced Stability Inversion of B₅H₁₀ ⁻

Figure 1 summarizes how the relative stability of nido and cyclic C _{5 v} [B₅H₁₀]⁻ motifs changes with the chemical environment. The cyclic C _{5 v} [B₅H₁₀]⁻ fragment is isoelectronic with the Cp⁻ anion and therefore constitutes an inorganic analogue of the prototypical η⁵‐aromatic ligand in metallocene chemistry. In the absence of metal coordination, however, the nido [B₅H₁₀]⁻ structure (I) is the global minimum, whereas a C_5v ‐symmetric [B₅H₁₀]⁻ fragment with a five‐membered boron ring (II) is much less stable, lying 78.2 kcal/mol higher in energy. This large energy penalty reflects the intrinsic instability of pentagonal borane rings in the absence of a metal center. Upon coordination to the Fe(η⁵‐C₅H₅)⁺ fragment, as reported experimentally by Weiss and Grimes and later analyzed by Ferrer et al. [21], the stability ordering is inverted: the η⁵‐bound complex Fe(η⁵‐C₅H₅)(η⁵‐B₅H₁₀) (II) becomes 7.2 kcal/mol more stable than the η³‐bound, nido‐like analogue Fe(η⁵‐C₅H₅)(η³‐B₅H₁₀) (I). This inversion indicates that metal coordination can compensate for the energetic penalty associated with isomerization of the borane framework.

FIGURE 1.

Coordination‐induced stabilization of B₅H₁₀ motifs: relative energies for isolated [B₅H₁₀]⁻ (top), CpFe‐coordinated borane (middle), and FeB₅H₁₀ ⁺ (bottom). Energies are relative within each pair (kcal/mol).

To disentangle the role of the metal center from that of the cyclopentadienyl ligand, we also explored the PES of FeB₅H₁₀ ⁺ (lowest‐energy isomers shown in Figure S1). The resulting energy ordering again favors an η⁵‐coordinated motif, now by 17.0 kcal/mol relative to the nido‐like alternative (Figure 1), indicating that stabilization of the C _{5 v} B₅H₁₀ ring arises from coordination itself rather than from a specific effect of Cp co‐ligation. This result motivates the main question addressed here: whether two such borane rings can be stabilized simultaneously at a single metal center to yield a thermodynamically viable homoleptic FeB₁₀H₂₀ sandwich complex. All geometries and relative energies discussed here were computed at the same level as used for the global PES screening (see computational details in the Supporting Information).

2.2. Global PES exploration of FeB₁₀H₂₀

The FeB₁₀H₂₀ PES was explored globally across the singlet, triplet, and quintet manifolds using AUTOMATON [27, 28]. During the screening stage, candidate structures were optimized at the PBE0 [29]/SDDALL [30, 31, 32, 33, 34] level, and all distinct minima were then refined at the PBE0‐D3 [35]/def2‐TZVP [36, 37] level. Harmonic frequency calculations confirmed that the optimized structures correspond to true minima. For isomers lying within 20 kcal/mol of the global minimum, relative energies were further refined by DLPNO‐CCSD(T) [38, 39, 40]/CBS [41, 42] single‐point calculations on the PBE0‐D3/def2‐TZVP geometries. To evaluate whether the η⁵,η⁵‐Fe(B₅H₁₀)₂ global minimum requires a multireference treatment, we also performed CASSCF [43] calculations with progressively expanded active spaces. The largest space considered, CAS (12,10), includes the five Fe‐centered 3d orbitals together with their correlating counterparts and selected ligand‐based contributions. The wavefunction remains dominated by a single closed‐shell singlet configuration, with a reference weight of about 82%, and the natural orbital occupations show only moderate deviations from ideal values. No qualitative changes in orbital ordering or in the metal–ligand bonding pattern are found (Figures S2–S4). Full computational details are provided in the Supporting Information.

To broaden the configurational space sampled for FeB₁₀H₂₀ and reduce any bias toward predefined coordination motifs, the unbiased global search was complemented by a fragment‐based assembly strategy. The singlet exploration of FeB₅H₁₀ ⁺ yielded five low‐energy minima (Figure S1). For each of these motifs, the B₅H₁₀ framework was duplicated, mirror‐reflected, and placed on the opposite side of the Fe center without imposing hapticity constraints, thereby generating Fe(B₅H₁₀)₂ starting geometries. These assembled structures were then optimized and refined using the same protocol applied to the AUTOMATON‐generated candidates. This combined strategy allowed systematic sampling of both symmetric and asymmetric sandwich‐like topologies.

The low‐energy structures of FeB₁₀H₂₀ obtained from the PES exploration are depicted in Figure 2. The global minimum is a symmetric Fe(η⁵‐B₅H₁₀)₂ sandwich structure (D _{5 d} , ¹A₁), in which both borane ligands coordinate through η⁵ binding to C _{5 v} ‐[B₅H₁₀]^– frameworks. The next singlet minimum lies 12.7 kcal/mol higher in energy (C_s , ¹A^′) and exhibits mixed η³/η⁵ coordination, whereas the remaining singlet structures within 16 kcal/mol all display η³/η³ binding modes (C_2h , ¹A_g; C₁, ¹A). At higher energies, the triplet manifold is dominated by η³/η³‐coordinated motifs above 27 kcal/mol, together with a cage‐like structure containing localized Fe‐H bonds at 30.9 kcal/mol. Harmonic‐frequency analyses confirm that all reported stationary points are true minima, and the consistently small DLPNO‐CCSD(T) T₁ diagnostic values (0.02–0.04) are consistent with a predominantly single‐reference description, indicating that dynamic correlation provides the dominant contribution to the electronic structure. These results show that dual η⁵ coordination is uniquely stabilized in FeB₁₀H₂₀, whereas partial or complete loss of η⁵ binding is associated with pronounced energetic destabilization.

FIGURE 2.

Lowest‐energy structures of FeB₁₀H₂₀. Relative energies (kcal/mol) are given with respect to the GM (at DLPNO‐CCSD(T)/CBS//PBE0‐D3/def2‐TZVP level). T ₁ diagnostics are shown in braces, and the lowest harmonic frequencies (cm⁻¹) are given in parentheses.

Born–Oppenheimer molecular dynamics (BO–MD) [44] simulations at 400, 500, and 1000 K provide additional information on the kinetic robustness of the Fe(η⁵‐B₅H₁₀)₂ complex. At 400 and 500 K, the trajectories remain within the basin of the global minimum, with small RMSD fluctuations that indicate hindered relative rotation of the two B₅H₁₀ ligands in the C _{5 v} structure. At 1000 K, thermally activated reorientation events corresponding to relative rotation of the borane rings between symmetry‐equivalent staggered arrangements are found, but the Fe–borane framework remains intact (Figure S5). These simulations indicate that the sandwich topology is preserved even under strong thermal excitation.

2.3. Chemical Bonding

As shown in Figure 3, the two η⁵‐coordinated B₅H₁₀ ligands bind symmetrically to the central iron atom. The Fe–B distance of 2.03 Å is close to the single‐bond reference value estimated from Pyykkö covalent radii (2.01 Å) [45], supporting the presence of a Fe–B bonding interaction rather than a purely long‐range electrostatic contact. The corresponding Wiberg bond index [46] (WBI _Fe‐B = 0.29) is consistent with delocalized η⁵ metal‐ring bonding. Rotation of one borane ligand toward an eclipsed D _{5 h} arrangement proceeds through a transition state 2.9 kcal/mol above the staggered ground state, indicating a modest but well‐defined rotational barrier. A similar Fe–B distance of about 2.0 Å is obtained for the heteroleptic η⁵,η⁵‐B₅H₁₀FeCp complex, showing that the η⁵ borane–iron interaction is only weakly affected by ligand substitution.

FIGURE 3.

Different views of η⁵, η⁵‐Fe(B₅H₁₀)₂. Selected bond lengths (Å), Wiberg bond index (in blue), and CM5 charges (in green).

Within the borane ring, the B─B bond length of 1.77 Å is only moderately longer than the single‐bond reference value of 1.70 Å, and the corresponding WBI of 0.62 indicates that covalent character is retained within the boron framework. Atomic charges were evaluated using NPA [47], CM5 [48], QTAIM [49], Hirshfeld [50], and VDD [51] schemes (Table S1), revealing the expected method dependence for a delocalized transition‐metal complex. NPA assigns a negative charge to Fe (q(Fe) = −0.7), but this result is a known artifact in transition‐metal systems caused by the exclusion of Fe 4p functions from the NPA valence space [52]. For that reason, NPA charges were not used for chemical interpretation. Hirshfeld and VDD yield smaller absolute values, reflecting the sensitivity of atomic charge partitioning in highly delocalized bonding environments [53, 54, 55, 56]. The discussion below, therefore, uses CM5 as the principal descriptor. Within this scheme, Fe carries a positive charge of +1.1, balanced by a negative charge on the borane framework (q(B) = −0.4, q(H) = +0.1). This distribution is consistent with η⁵,η⁵ coordination, in which each B₅H₁₀ ⁻ unit acts as a π‐donor ligand toward Fe, in direct analogy to Cp⁻ and to the behavior established for the experimentally characterized heteroleptic complex (η⁵‐C₅H₅)Fe(η⁵‐B₅H₁₀) [21]. Overall, the charge analysis supports a positively charged Fe center embedded in a strongly delocalized Fe–borane framework.

The AdNDP [57, 58] analysis (Figure 4) provides a chemically intuitive picture of the bonding framework. In addition to three localized one‐center two‐electron (1c–2e) lone pairs at Fe and the expected B─H σ framework in the two borane ligands, including both 2c–2e B─H bonds and 3c–2e B─H─B bonds, six delocalized 11c–2e orbitals are recovered that extend across the full sandwich framework. Analysis of the isolated [B₅H₁₀]⁻ ligand (Figure S6) yields three 15c–2e π‐type delocalized orbitals associated with the cyclic crown‐like B₅H₅ framework, in addition to the B─H σ bonds. When two such ligands are combined in the same relative geometry found in η⁵,η⁵‐Fe(B₅H₁₀)₂ but without the metal center, six delocalized multicenter orbitals are already recovered (Figure S7). This indicates that the delocalized electronic framework is largely ligand‐derived, while metal coordination reorganizes and stabilizes these ligand‐based delocalized orbitals within the assembled η⁵,η⁵ iron–bis(borane) structure.

FIGURE 4.

AdNDP orbitals of η⁵,η⁵‐Fe(B₅H₁₀)₂ computed at the PBE0‐D3/def2‐TZVP level.

2.4. EDA and IQA analysis of Metal–Ligand and Intra‐Ligand Bonding

To obtain a more detailed orbital description of the Fe–ligand interactions, we analyzed η⁵,η⁵‐Fe(B₅H₁₀)₂ and, for direct comparison, D _{5 d} Fe(Cp)₂ using energy decomposition analysis combined with natural orbitals for chemical valence (EDA‐NOCV) at the PBE0‐D3/TZ2P‐ZORA level [59, 60]. This approach provides a quantitative extension of the qualitative orbital interaction framework established by Hoffmann and later developed for transition‐metal sandwich complexes by Albright et al. [61], and follows the EDA‐based methodology proposed by Frenking and coworkers for quantitative analyses of transition‐metal bonding, including ferrocene [62, 63]. The bonding was described using the singlet fragment pair Fe²⁺ and (L₂)²⁻ (L = Cp or B₅H₁₀), following the fragmentation scheme previously applied to ferrocene [63], which allows direct comparison between the two systems. The resulting EDA‐NOCV values are summarized in Table 1.

TABLE 1.

EDA‐NOCV results of Fe(Cp)₂ and η ⁵ ,η⁵‐Fe(B₅H₁₀)₂ at PBE0–D3/TZ2P–ZORA level. All energy values are given in kcal/mol. L₂ ²⁻ = Cp₂ ²⁻, (B₅H₁₀)₂ ²⁻.

Energy terms	Interaction	Fe(Cp)2	η5,η5‐Fe(B5H10)2
ΔE int		−982.1	−1104.1
ΔE pauli		421.2	580.3
ΔE elstat		−666.9	−707.7
ΔE oi		−734.3	−970.5
ΔE oi(1)	(L2)2− → Fe2+ (3$d_{{\mathrm{x}}^2-{\mathrm{y}}^2}$) donation	−47.5(6.5%)	−114.2(11.8%)
ΔE oi(2)	(L2)2− → Fe2+ (3d xy) donation	−47.9(6.5%)	−114.3(11.8%)
ΔE oi(3)	(L2)2− → Fe2+ (3$d_{{\mathrm{z}}^2}$) donation	−186.6(25.4%)	−206.3(21.3%)
ΔE oi(4)	(L2)2− → Fe2+ (3d xz) donation	−117.5(16.0%)	−147.2(15.2%)
ΔE oi(5)	(L2)2− → Fe2+ (3d yz) donation	−117.0(15.9%)	−170.5(17.6%)
ΔE oi(rest)		−217.8(29.7%)	−218.0(22.5%)
ΔE disp		−2.1	−6.1

The EDA‐NOCV values obtained here for ferrocene at the PBE0‐D3/TZ2P‐ZORA level differ quantitatively from those reported by Rayón and Frenking at the BP86/TZP level [62, 63], as expected from the different computational setups. However, the bonding picture remains the same in both analyses: the Fe–Cp interaction is dominated by ligand‐to‐metal π donation, and the relative balance between electrostatic and covalent contributions is comparable, with values of 51.1% and 48.9% in the Rayón–Frenking study and 47.6% and 52.4% in the present analysis.

The total interaction energy is more stabilizing in η⁵, η⁵‐Fe(B₅H₁₀)₂ than in ferrocene (ΔE _int = −1104.1 vs. −982.1 kcal/mol). Although the borane sandwich shows a larger Pauli repulsion (ΔE _Pauli = 580.3 vs. 421.2 kcal/mol), this difference is compensated by more attractive electrostatic (ΔE _elstat = −707.7 vs. −666.9 kcal/mol) and orbital interaction terms (ΔE _orb = −970.5 vs. −734.3 kcal/mol). As a result, the orbital contribution accounts for a larger fraction of the total attractive interactions in the borane sandwich (57.8%) than in ferrocene (52.4%), indicating a more pronounced covalent character in the Fe–borane bonding. The dispersion term is small in both systems (ΔE _disp = −6.1 vs. −2.1 kcal/mol) and does not contribute significantly to the bonding differences between them.

The EDA‐NOCV deformation densities (Figure 5) indicate that the bonding in both complexes is dominated by ligand‐to‐metal donation, with no significant metal‐to‐ligand backdonation. In η⁵,η⁵‐Fe(B₅H₁₀)₂, the five principal interaction channels correspond to donation from the borane ligands into the Fe 3d acceptor manifold [61]. The largest contribution corresponds to donation into the Fe $3d_{{\mathrm{z}}^2}$ orbital ($\mathrm{\Delta }{E}_{\mathrm{orb}}^{(3)}$ = −206.3 kcal/mol, 21.3%), followed by donation into 3d _yz ($\mathrm{\Delta }{E}_{\mathrm{orb}}^{(5)}$ = −170.5 kcal/mol, 17.6%), 3d _xz ($\mathrm{\Delta }{E}_{\mathrm{orb}}^{(4)}$ = −147.2 kcal/mol, 15.2%), $3d_{{\mathrm{x}}^2-{\mathrm{y}}^2}$ ($\mathrm{\Delta }{E}_{\mathrm{orb}}^{(1)}$ = −114.2 kcal/mol, 11.8%), and 3d _xy ($\mathrm{\Delta }{E}_{\mathrm{orb}}^{(2)}$ = −114.3 kcal/mol, 11.8%). In symmetry terms, the d _yz and d_xz channels correspond to the dominant e_1g(π) donation, which is also the largest orbital contribution in ferrocene, whereas the $3d_{{\mathrm{x}}^2-{\mathrm{y}}^2}$/3d _xy and $3d_{{\mathrm{z}}^2}$ channels correspond to the e_2g(δ) and a_1g(σ) components, respectively. The remaining orbital contribution (ΔE _orb(rest) = −218.0 kcal/mol, 22.5%) arises from higher‐order polarization and charge‐transfer effects that are not resolved into individual NOCV pairs.

FIGURE 5.

Deformation densities (Δρ) plots from EDA‐NOCV analysis for (a) Fe(Cp)₂ and (b) η⁵, η⁵‐Fe(B₅H₁₀)₂. The isovalue used for the surfaces is 0.0003 a.u. Charge flows from red (donor) to green (acceptor). Energy values are given in kcal/mol.

Relative to ferrocene, the borane sandwich shows stronger donation into all five Fe 3d‐based acceptor orbitals, most clearly for the e_2g (δ) components. This result indicates that the B₅H₁₀ ⁻ ligands act as stronger overall π‐donor ligands toward Fe than Cp⁻.

To complement the orbital analysis, we performed interacting quantum atoms (IQA) calculations (Figure 6) [64, 65, 66]. The Fe–B interaction is moderately stabilizing (${\mathrm{V}}_{\mathrm{IQA}}^{\mathrm{int}}$= −41.9 kcal/mol) and results from an attractive exchange–correlation contribution that is partially offset by Coulomb repulsion (see Supporting Information for the full decomposition). This pattern is consistent with a weakly covalent and delocalized Fe–B interaction, as expected for η⁵ coordination to electron‐deficient borane ligands. The Fe–B delocalization index (δ = 0.51) further supports electron sharing between Fe and borane ligands.

FIGURE 6.

Energy components of IQA between atom pairs for η⁵,η⁵‐Fe(B₅H₁₀)₂. $V_{IQA}^{int}$, $V_C^{int}$, and $V_{XC}^{int}$ denote the interatomic interaction energy, Coulomb component, and exchange‐correlation component, respectively (kcal/mol).

Within the B₅H₁₀ units, B─B contacts are strongly repulsive (${\mathrm{V}}_{\mathrm{IQA}}^{\mathrm{int}}$ = +204.8 kcal/mol). Stabilization of the ligand framework instead arises mainly from B─H interactions. In particular, the B─H─B 3c–2e bonds involving bridging hydrogens are strongly stabilizing, and terminal B─H interactions contribute even more strongly to the cohesion of the borane framework. The delocalization indices are consistent with this description: δ(B─H₂) = 0.40 is characteristic of multicenter bonding, whereas δ(B─B) = 0.27 confirms limited direct electron sharing between boron atoms.

For comparison, IQA of ferrocene (Table S1) shows a more stabilizing Fe─C interaction, with both Coulombic and exchange–correlation contributions being attractive, and a ligand framework stabilized primarily by direct C─C σ bonding within the Cp rings. This comparison identifies an important difference between the two sandwich systems: although both display comparable metal–ligand electron sharing characteristic of η⁵ coordination, the σ framework responsible for ligand cohesion is fundamentally different.

The IQA results are therefore consistent with the EDA‐NOCV analysis and further show that, although η⁵ coordination in both systems involves comparable degrees of metal–ligand electron sharing, the intraligand σ framework is different in each case. According to the IUPAC definition of a sandwich complex as a compound in which a metal atom is located between the faces of two parallel and planar or nearly planar ring systems [67], η⁵,η⁵–Fe(B₅H₁₀)₂ satisfies the structural criterion for a genuine sandwich complex.

2.5. Aromaticity in the Borane Sandwich Framework

The aromatic character of η⁵,η⁵‐Fe(B₅H₁₀)₂ was analyzed using the magnetically induced current density (MICD)[68, 69] approach (see Supporting Information for computational details). The MICD isosurfaces and current‐density vectors (Figure 7) show a dominant diatropic current associated with the B₅ framework of each borane ligand, consistent with aromatic behavior. Quantification in terms of ring current strengths (RCS) gives a net diatropic contribution of 13.8 nA/T for the B₅ ring, partially offset by a weaker paratropic contribution of −3.9 nA/T localized closer to the metal center. The predominance of the diatropic term indicates a net aromatic response of the borane fragment. The magnitude of the borane ring current is comparable to that of benzene computed at the same level (11.9 nA/T), indicating that the cyclic B₅ framework sustains a clear aromatic response rather than a marginal delocalization.

FIGURE 7 (a).

MICD modulus for η⁵,η⁵‐Fe(B₅H₁₀)₂ with tangential current‐density vectors on the isosurface. (b) Selected planes used for the MICD vector plots. (c) Vector plot on planes in (b). (d) Identified ring current circuit in the B₅H₁₀ fragment with the corresponding RCS values (nA/T).

For comparison, MICD analysis of ferrocene (Figure S8) reveals a somewhat stronger diatropic ring current in the Cp ligands (RCS = 14.4 nA/T), consistent with their established π‐aromatic character. Despite the different chemical nature of the ligands (inorganic vs. organic), both sandwich systems display comparable global ring currents under an external magnetic field. This parallel supports the description of η⁵,η⁵‐Fe(B₅H₁₀)₂ as a boron‐based analogue of ferrocene from the perspective of magnetic aromaticity.

To complement the magnetic‐response analysis, we evaluated the electron density of delocalized bonds (EDDB) for η⁵,η⁵‐Fe(B₅H₁₀)₂ (Figure 8). The global EDDB_G population of 24.5 |e| indicates an extensive delocalized electron framework spanning the full sandwich complex. Fragment‐resolved analysis shows that this delocalization is predominantly ligand‐centered, with each B₅H₁₀ unit contributing the largest share (8.6 |e|), whereas the iron center mainly organizes the spatial distribution of the delocalized density. Inspection of the EDDB isosurfaces further indicates that a substantial fraction of the delocalized electron population arises from multicenter B─H─B bonding motifs intrinsic to the borane ligands, particularly the 3c–2e σ interactions absent in cyclopentadienyl‐based systems. Compared to ferrocene (EDDB_G = 13.3 |e|), the borane sandwich displays a much larger delocalized electron population, which indicates that the enhanced delocalization originates from the intrinsic multicenter bonding of the borane framework rather than from greater metal participation.

FIGURE 8.

Isosurface contour in η⁵,η⁵‐Fe(B₅H₁₀)₂ by global and local EDDB(r) functions (isocontour of 0.55 a.u.) with the corresponding population of delocalized electrons (PBE0‐D3/def2‐TZVP).

2.6. Energetics and Synthetic Feasibility

Formation energies of the relevant iron complexes were analyzed within a consistent fragment‐reference scheme based on Fe²⁺, Cp⁻, and the C _{5 v} ‐symmetric [B₅H₁₀]⁻ ligand (isomer II). Within this framework, formation of the homoleptic bis(borane) sandwich from two C _{5 v} borane rings and Fe²⁺ is highly exothermic (ΔE = ⁻717.4 kcal/mol), and more stabilizing than formation of ferrocene (ΔE = ⁻634.4 kcal/mol). Mixed‐ligand coordination is also strongly stabilizing (Cp⁻ + [B₅H₁₀]⁻ + Fe²⁺ → CpFe(η⁵‐B₅H₁₀), ΔE = ⁻676.8 kcal/mol). These comparisons identify η⁵,η⁵‐Fe(B₅H₁₀)₂ as the most strongly stabilized iron sandwich complex among the ligand combinations examined and show that, under η⁵ coordination, the borane rings can act as strong π‐donor ligands toward Fe(II).

However, ligand redistribution from the experimentally known mixed‐sandwich CpFe(B₅H₁₀) to the homoleptic bis(borane) complex is thermoneutral (2CpFe(η⁵‐B₅H₁₀) → η⁵,η⁵‐Fe(B₅H₁₀)₂ + Cp₂Fe, ΔE = ⁻0.2 kcal/mol). This near‐degeneracy explains why Cp‐mediated ligand exchange does not provide a viable synthetic route to η⁵,η⁵‐Fe(B₅H₁₀)₂, even though the latter is thermodynamically favored among the iron sandwich species considered here. The formation of the homoleptic bis(borane) sandwich is therefore better understood as a kinetic and operational problem rather than a thermodynamic one, and any realistic synthetic approach should avoid Cp‐mediated pathways.

Known experimental constraints from borane chemistry further define the accessible window for metallation. Deprotonation of arachno‐pentaborane(11) generates the [B₅H₁₀]⁻ monoanion only under cryogenic conditions because this species is thermally fragile and unstable above approximately −80°C [70]. Conversely, isomerization of the borane framework upon iron coordination, as reflected in the η³↔η⁵ interconversion observed in CpFe(B₅H₁₀), requires substantial thermal activation [71]. These opposing temperature requirements imply that borane capture and borane isomerization cannot be achieved under the same conditions.

These considerations motivate us to propose a two‐stage “capture–then–anneal” strategy. In a first stage, a labile Fe(II) center captures borane‐derived [B₅H₁₀]^– units under cryogenic conditions to preserve the integrity of the borane framework. In a second stage, controlled thermolysis supplies the energy required for the rearrangement of the coordinated borane units into C _{5 v} rings bound in an η⁵ fashion. The combination of established experimental temperature constraints, borane stability limits, and the present PES analysis thus indicates that formation of the homoleptic bis(borane) sandwich is thermodynamically viable and synthetically plausible, provided that borane capture and ligand isomerization are temporally and thermally decoupled.

Finally, consistent with the electronic analogy between [B₅H₁₀]⁻ and Cp⁻, η⁵,η⁵‐Fe(B₅H₁₀)₂ has a slightly larger HOMO–LUMO gap (6.42 eV) than ferrocene (6.0 eV). Within the maximum hardness principle, this difference is compatible with a slightly higher electronic hardness of the inorganic analogue, which may be relevant to the operational stability of the target complex under synthetic and handling conditions. A detailed analysis of redox and substitution reactivity, however, would require explicit investigation.

3. Conclusions

An extensive global exploration of the FeB₁₀H₂₀ potential energy surface identifies the homoleptic sandwich complex η⁵,η⁵‐Fe(B₅H₁₀)₂ as the global minimum of this compositional space. The results uncover a coordination‐induced reversal of stability in which Fe(II) stabilizes borane rings relative to the intrinsically preferred nido motifs and thereby enables simultaneous η⁵ coordination of two B₅H₁₀ ligands. This behavior indicates that transition‐metal coordination can provide access to borane motifs that are not thermodynamically preferred in isolation.

Structural, bonding, and electronic analyses consistently support a delocalized η⁵ metal–ring interaction in which iron mainly acts as an electropositive center that enforces isomerization and reorganizes the electronic structure of the borane ligands, rather than forming localized two‐center bonds. The delocalized electronic framework of the sandwich complex originates primarily from the intrinsic π delocalization of the C _{5 v} [B₅H₁₀]⁻ ligands and is then reorganized and stabilized upon metal coordination. Magnetic‐response and real‐space delocalization analyses further indicate that aromaticity in this system is largely ligand‐centered and strongly shaped by multicenter B─H─B bonding motifs. In this sense, borane‐based metallocenes are electronically distinct from their carbon‐based counterparts despite the formal structural analogy.

Energetic analysis rationalizes why Cp‐mediated ligand‐exchange routes fail to access η⁵,η⁵‐Fe(B₅H₁₀)₂ despite its pronounced thermodynamic stabilization and, when combined with known experimental constraints on borane stability and rearrangement, supports a Cp‐free, temporally decoupled capture–then–anneal pathway as a plausible synthetic strategy. More broadly, this work shows that transition‐metal coordination can reverse the intrinsic stability preferences of electron‐deficient boron frameworks and thereby enable aromatic borane ligands that are otherwise inaccessible. This provides a basis for the design of a broader class of boron‐based sandwich complexes and extends the metallocene framework beyond carbon‐centered aromatic ligands.

4. Note added in proof

Shortly before the acceptance of our manuscript, an independent study by Mohapatra, Ghosh, Jemmis, and co‐workers reported the synthesis and crystallographic characterization of η⁵,η⁵‐Os(B₅H₁₀)₂ [72]. This achievement represents the culmination of a sustained experimental effort by the Ghosh group toward carbon‐free ferrocene analogues, building on earlier heteroleptic complexes such as [CpRu(η⁵‐B₅H₁₀)] [73, 74], [CpIr(η⁵‐B₅H₉)] [74], and [Cp*Os(η⁵‐B₅H₁₀)] [75]. The independent and concurrent emergence of these results validates the underlying bonding motif and is fully consistent with the global‐minimum predictions reported here.

Author Contributions

Viviana Roman‐Ventura: methodology, investigation, writing – original draft, visualization, validation. Luis Quispe‐Corimayhua: methodology, investigation, validation, writing – original draft. Dumer S. Sacanamboy: methodology, formal analysis, visualization. Diego Inostroza: methodology, software, validation. Luis Leyva‐Parra: conceptualization, methodology, formal analysis, supervision, writing ‐ original draft. Dayán Páez: methodology, software, supervision, writing – original draft. Kelling J. Donald: writing – review and editing, writing – original draft, formal analysis. Gabriel Merino: conceptualization, writing – review and editing, writing – original draft, formal analysis. William Tiznado: conceptualization, validation, resources, writing – original draft, supervision, writing – review and editing, project administration, funding acquisition.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File: anie72695‐sup‐0001‐SuppMat.docx

Data Availability Statement

The data that support the findings of this study are available in the Supporting Information of this article