Effects of Lewis Acidic Metal Ions (M) on Oxygen-Atom Transfer Reactivity of Heterometallic Mn₃MO₄ Cubane and Fe₃MO(OH) and Mn₃MO(OH) Clusters

Abstract

The modulation of the reactivity of metal oxo species by redox inactive metals has attracted much interest due to the observation of redox inactive metal effects on processes involving electron transfer both in nature (the oxygen evolving complex of Photosystem II) and in heterogeneous catalysis (mixed-metal oxides). Studies of small molecule models of these systems have revealed numerous instances of effects of redox inactive metals on electron and group transfer reactivity. However, the heterometallic species directly involved in these transformations have rarely been structurally characterized and are often generated in situ. We have previously reported the preparation and structural characterization of multiple series of heterometallic clusters based on Mn₃ and Fe₃ cores and described the effects of Lewis acidity of the heterometal incorporated in these complexes on cluster reduction potential. To determine the effects of Lewis acidity of redox inactive metals on group transfer reactivity in structurally well-defined complexes, we studied [Mn₃MO₄], [Mn₃MO(OH)], and [Fe₃MO(OH)] clusters in oxygen atom transfer (OAT) reactions with phosphine substrates. The qualitative rate of OAT correlates with the Lewis acidity of the redox inactive metal, confirming that Lewis acidic metal centers can affect the chemical reactivity of metal oxo species by modulating cluster electronics.

Graphical-abstract

1. INTRODUCTION

Lewis acidic, redox inactive metal ions are known to influence electron- and group-transfer processes in both natural and synthetic systems.^1–3 In biology, the preeminent example is that of the redox inactive Ca²⁺ center found in the oxygen evolving complex (OEC) of Photosystem II (PSII), a Mn₄CaO_x cluster responsible for water oxidation to dioxygen in plants, algae, and cyanobacteria. Although its role has not been fully elucidated, Ca²⁺ is an essential cofactor in this system.^{4, 5} Redox inactive metals likewise affect the catalytic behavior of heterogeneous mixed-metal oxides in reactions such as water oxidation and dioxygen reduction.^6–8 The reactivity of inorganic oxidants (e.g. MnO₄^–, Cr₂O₇^2–) with organic substrates can be modulated by Lewis acidic additives.^9–15 Redox inactive metals have been reported to affect the rate of phosphine oxygenation by a Mn^V–oxo complex.¹⁶ Lewis acidic, redox inactive metal ions show a range of effects on the oxygen atom transfer (OAT) reactivity of high-valent Fe-, Mn-, and Co-oxo species including rate enhancement and shifts in reaction mechanism.^17–25 Lewis-acid induced O–O bond cleavage or release of O₂ from a Fe(III)-bound peroxide species has also been reported.^{26, 27} The effects of Lewis acids on group transfer reactions by high-valent Mn, Fe, and V complexes, including under catalytic conditions, have been described.^28–33 Sc³⁺ has been reported to affect the formation and reactivity of Fe(III)-(hydro)peroxo species.^{34, 35} While the observation of substantial effects of redox inactive metals on oxygen related redox chemistry is well demonstrated, structural characterization of the precursor Lewis acid-bound species active in these transformations remains a challenge. In most cases, reactive species are generated in situ by addition of large excesses of Lewis acid additives to high-valent metal oxo complexes. Although spectroscopic studies of the interaction of metal oxo complexes with metal ions in solution have been described in select cases, structural characterization of these species has rarely been reported.^{17, 36} The only crystallographically characterized high-valent metal oxo/redox inactive metal ion adduct, a [Fe(IV)–O–Sc³⁺] complex,¹⁷ has recently been reassigned as a [Fe(III)–(OH)–Sc³⁺] species of spectroscopic characterization and structural comparisons with other Fe(III) and Fe(IV) complexes.^{37, 38} Therefore, conclusions drawn from the majority of these studies must account for both the effect of Lewis acid binding to metal–oxo species and for the dynamic equilibrium processes responsible for these interactions.

Incorporation of Lewis acidic metal centers in well-defined multimetallic complexes precludes the need for use of excess Lewis acid additives by providing structurally unambiguous precursors for reactivity studies. This is an appealing strategy for such studies, as it enables a less ambiguous interpretation of any observed heterometallic effects on reactivity. Recently, studies of the effect of redox inactive additives on reactivity related to oxygen or atom-transfer processes^44–48 and on photocatalytic water oxidation⁴⁹ have been reported. Our group has studied the effects of Lewis acidity of a redox inactive metal center on the reduction potential of well-defined heterometallic clusters.^{3, 39–43} Tetrametallic complexes were supported by a multinucleating ligand framework and contained a Mn₃ or Fe₃ core, as well as a fourth, redox inactive metal (M). Tetraoxo [Mn₃MO₄] cubane clusters, as well as oxo-hydroxo [Mn₃MO(OH)] (previously assigned as [Mn₃MO₂] dioxo species, vide infra) and [Fe₃MO(OH)] clusters were investigated. A linear correlation was discovered between the Lewis acidity of the redox inactive metal and the reduction potential of the cluster,^{40, 42, 43} supporting the plausibility of a potential-tuning role for redox inactive metals in biological and heterogeneous systems. Although comparisons of these complexes have focused on their one-electron reduction potentials (a thermodynamic parameter), these clusters are well suited to further exploration of the effects of redox inactive metals on chemical reactivity. We have previously reported initial studies of reactions of cubane clusters with phosphines,⁵⁰ including investigations into the mechanism of OAT from these species. However, an extensive structure-function study involving a broader range of cluster structures containing redox-inactive metals of varying Lewis acidity has not previously been carried out. Importantly, our structurally characterized heterotetrametallic clusters provide well-defined precursors for OAT reactions, and the availability of structurally analogous complexes containing different redox inactive metals allows for direct comparison of their effects. Herein, we describe the qualitative assessment of OAT reactivity with phosphine substrates (Scheme 1) of three series of complexes: [Mn₃MO₄] cubanes and [Mn₃MO(OH)], and [Fe₃MO(OH)] oxo-hydroxo complexes. The reassignment of Mn₃ oxo-hydroxo clusters, previously identified as the corresponding [Mn₃MO₂] dioxo complexes, on the basis of spectroscopic and computational data is also discussed.

Scheme 1.

Oxidation of Phosphines Investigated in the Present Study.

2. RESULTS AND DISCUSSION

2.1. OAT Reactivity of [Mn₃MO₄] Cubanes (M = Ca²⁺, Gd³⁺, Sc³⁺, Mn³⁺).

The OAT reactivity of [Mn^IV₃CaO₄], [Mn^IV₃ScO₄], and [Mn^III₂Mn^IV₂O₄] (Figure 1) with trialkylphosphines has previously been described.⁵⁰ Reaction of [Mn^III₂Mn^IV₂O₄] with trimethylphosphine (PMe₃, 2 equiv) led to rapid generation of trioxo cluster [Mn^III₄O₃], in which the bottom oxide ligand is absent (Scheme 2). Cubane complex [Mn^IV₃CaO₄] was instead found to be unreactive towards PMe₃. This observation is consistent with the greater oxidizing power of [Mn^III₂Mn^IV₂O₄] as indicated by its more positive one-electron reduction potential (E_½ = –0.70 vs. –0.94 V vs. Fc^+/0 for [Mn^IV₃CaO₄]). In contrast with this model, [Mn^IV₃ScO₄] displayed only slow reactivity with PMe₃, generating an intractable mixture of metal-containing products, despite being a stronger oxidant than [Mn^III₂Mn^IV₂O₄] (E_½ = –0.24 vs. –0.70 V vs. Fc^+/0, respectively; see Table 1).

Figure 1.

Left: Heterotetrametallic clusters investigated for OAT reactivity. Right: plots of cluster redox potentials vs. pK_a of M(aqua)ⁿ⁺ ion (as measure of Lewis acidity) summarizing the previously reported redox inactive metal effects on redox processes of heterometallic clusters (red squares: [Mn₃MO₄] cubane clusters;^39–41 green circles: [Fe₃MO(OH)] oxo-hydroxo clusters;⁴² blue diamonds: [Mn₃MO(OH)] oxo-hydroxo clusters,⁴³ vide infra; the red square labeled as “Mn³⁺” corresponds to the [Mn^IIIMn^IV₃]/[Mn^III₂Mn^IV₂] couple in the all-manganese cubane cluster⁴¹).

Scheme 2.

OAT Reactivity of [Mn₃MO₄] Cubane Clusters.

Table 1.

One-electron Redox Potentials of Selected [Mn^IV₃MO₄] and [Mn^IIIMn^IV₂MO₄] Clusters

Compound	E½ (V) vs. Fc+/0	Redox Couple
[MnIII2MnIV2O4]41	−0.70	MnIII2MnIV2/ MnIII3MnIV
[MnIV3CaO4]41	−0.94	MMnIV3/MMnIIIMnIV2
[MnIV3ScO4]40	−0.24	MMnIV3/MMnIIIMnIV2

These seemingly contradictory observations can be rationalized by considering the previously studied mechanism of transfer of an oxygen atom from the cubane cluster to PMe₃. According to earlier computations,⁵⁰ OAT involving cubane clusters [Mn^IV₃MO₄] and [Mn^IIIMn^IV₂MO₄] proceeds via initial dissociation of a bridging acetate ligand from one of the three core Mn centers. The Mn^III center (high spin, d⁴) in complexes [Mn^IIIMn^IV₂MO₄] is more substitutionally labile due to electrons occupying a Mn–O σ–antibonding orbital. Thus, metal–carboxylate interactions in complexes [Mn^IIIMn^IV₂MO₄] are weaker than in the case of [Mn^IV₃MO₄] clusters, in which all Mn centers are in the Mn^IV oxidation state (d³). The difference in Mn oxidation states between [Mn^IV₃MO₄] and [Mn^IIIMn^IV₂MO₄] affects the kinetics of ligand dissociation, and only indirectly the OAT process, resulting in slower reactivity of [Mn^IV₃ScO₄] despite a larger driving force for electron transfer.

Based on this interpretation, it was thus postulated that one-electron reduction of other [Mn^IV₃MO₄] clusters, resulting in generation of a kinetically labile Mn^III center, would provide more reactive precursors for OAT reactivity, enabling more extensive comparisons of the effects of redox inactive metal of varying Lewis acidity on this transformation. One-electron reduction of [Mn^IV₃MO₄] complexes was achieved using cobaltocene (CoCp₂) or decamethylferrocene (FeCp*₂) as reductant. Reduction of [Mn^IV₃CaO₄] with CoCp₂ generated a highly unstable material, which could not be isolated cleanly, and was therefore not investigated further. Reduction of [Mn^IV₃ScO₄] and [Mn^IV₃GdO₄], on the other hand, successfully yielded the corresponding [Mn^IIIMn^IV₂ScO₄] and [Mn^IIIMn^IV₂GdO₄] clusters as reported previously.^{39, 40} Treatment of [Mn^IIIMn^IV₂ScO₄], with PMe₃ (10 equiv) results in clean conversion to a new species over 1.5 weeks at room temperature as determined by proton nuclear magnetic resonance (¹H NMR, Figure S6). The reaction could be accelerated by heating to 50 °C in benzene, resulting in full conversion to the same product in 35 h (Figure S7). As a control, heating of [Mn^IIIMn^IV₂ScO₄] in the absence of PMe₃ led to negligible (<10%) conversion. The product of reaction of [Mn^IIIMn^IV₂ScO₄] with PMe₃ was characterized by single-crystal X-ray diffraction (XRD) as the trioxo complex [Mn^III₃ScO₃] (Figure 2), analogously to the previously reported reaction of [Mn^III₂Mn^IV₂O₄] to generate [Mn^III₄O₃].⁵⁰ The average Mn–oxo and Mn–O(acetate) distances in [Mn^III₃ScO₃] are 1.88 and 2.10 Å, respectively, consistent with the assignment of the Mn₃ core as [Mn^III]₃. No signal for trimethylphophine oxide (OPMe₃), the expected byproduct OAT, was detected by ³¹P NMR spectroscopy during the reaction. The OPMe₃ generated in the reaction is thus proposed to be bound to the nascent [Mn^III₃ScO₃], at least in solution under the reaction conditions. Formation of OPMe₃ was however confirmed at the end of the reaction via gas chromatography mass-spectrometry (GC-MS) following workup (see Experimental Section).

Figure 2.

Left: truncated solid-state structure of [Mn^III₃ScO₃]. Right: relevant M–O and Mn–O bond distances in Å.

Treatment of [Mn^IIIMn^IV₂GdO₄], containing a significantly less Lewis acidic Gd³⁺ center (pK_a = 8.4 for Gd³⁺ vs. 4.8 for Sc³⁺)with PMe₃ (10 equiv) at 50 °C in benzene led to complete conversion of the starting material over 50 h to a stable new product (heating in the absence of PMe₃ did not lead to detectable conversion). Although crystallographic characterization of this product was unsuccessful, this species was assigned as the trioxo cluster [Mn^III₃GdO₃] on the basis of the similarity of its ¹H NMR features with those for structurally characterized [Mn^III₃ScO₃] (Figure S8). As a signal for OPMe₃ was detected during the reaction by ³¹P NMR, the reaction was monitored by NMR spectroscopy upon treating [Mn^IIIMn^IV₂ScO₄] with a smaller excess of PMe₃ (3 equiv) at 50 °C to improve accuracy of integration of NMR signal. This experiment was performed in the presence of PPh₃ (2 equiv) as an internal standard. A total of 1 equiv of PMe₃ was consumed throughout the reaction (67 h), and formation of the same metal species was observed (¹H NMR, Figure S9). The reaction stoichiometry (1:1 [Mn^IIIMn^IV₂GdO₄]/PMe₃) is consistent with transfer of a single O-atom equivalent during the reaction, corroborating the assignment of the metal-containing product as trioxo species [Mn^III₃GdO₃]. Notably, [Mn^III₃GdO₃] is stable even in the presence of a large excess of PMe₃ (10 equiv.), and formation of reduced clusters with <3 oxide ligands–the products of further OAT processes–was not observed.

The observed rate of OAT from [Mn^IIIMn^IV₂MO₄] clusters to phosphine substrates correlates with the Lewis acidity of M (Table 2, Entries 1–7). To generate these correlations, we have used the pK_a of the M(H₂O)_n ion⁵¹ as a measure of Lewis acidity; in our previous work, this scale of Lewis acidity has enabled analogous correlations between Lewis acidity and reduction potential. Analysis of the qualitative, relative rates of OAT of [Mn^IIIMn^IV₂MO₄] complexes with PMe₃ as substrate yields the following trend:

Table 2.

Summary of OAT reactivity of heterometallic Mn₃ and Fe₃ clusters

Entry	Compound	E1/2 (vs. Fc+/0)a	Lewis Acid	pKa51	Substrate (equiv.)	Timeb	T (°C)
137	[MnIV3ScO4]	–0.24 V40	Sc3+	4.8	PMe3 (2)	N.R.	r.t.
237	[MnIV3CaO4]	–0.94 V41	Ca2+	12.9	PMe3 (2)	N.R.	r.t.
3	[MnIII2MnIV2O4]	–0.70 V41	Mn3+	0.1	PMe3 (2)	15 min	r.t.
4	[MnIIIMnIV2ScO4]	-	Sc3+	4.8	PMe3 (10)	1.5 weeks	r.t.
5	[MnIIIMnIV2ScO4]	-	Sc3+	4.8	PMe3 (10)	35 h	50 °C
6	[MnIIIMnIV2GdO4]	-	Gd3+	8.4	PMe3 (10)	>2 weeks	r.t.
7	[MnIIIMnIV2GdO4]	-	Gd3+	8.4	PMe3 (10)	50 h	50 °C
8	[MnIII3YO(OH)]2	+0.42 V43	Y3+	8.6	PEt3 (10)	15 min	r.t.
9	[MnIII3YO(OH)]2	+0.42 V43	Y3+	8.6	PPh3 (10)	30 min	r.t.
10	[MnIII3CaO(OH)]	–0.07 V43	Ca2+	12.9	PEt3 (10)	3 h	r.t.
11	[MnIII3CaO(OH)]	–0.07 V43	Ca2+	12.9	PPh3 (10)	20 h	r.t.
12	[MnIIMnIII2YO(OH)]	-	Y3+	8.6	PEt3 (10)	30 min	r.t.
13	[MnIIMnIII2YO(OH)]	-	Y3+	8.6	PPh3 (10)	1 h	r.t.
14	[MnIIMnIII2CaO(OH)]	-	Ca2+	12.9	PEt3 (10)	36 h	r.t.
15	[MnIIMnIII2CaO(OH)]	-	Ca2+	12.9	PPh3 (10)	> 2 weeks	r.t.
16	[FeIII3LaO(OH)]	–0.08 V42	La3+	9.06	PPh3 (10)	24 h	r.t.
17	[FeIII3CaO(OH)]	–0.48 V42	Ca2+	12.9	PPh3 (10)	N.R.	r.t.
18	[FeIIFeIII2ScO(OH)]	-	Sc3+	4.8	PMe3 (10)	14 h	r.t.
19	[FeIIFeIII2LaO(OH)]	-	La3+	9.06	PMe3 (10)	130 h	r.t.
20	[FeIIFeIII2CaO(OH)]	-	Ca2+	12.9	PMe3 (10)	270 h	r.t.

^aFc^+/0 = ferrocenium/ferrocene couple; only potentials for relevant reduction events included;

^bN.R. = no reaction observed

[Mn^III₂Mn^IV₂O₄] > [Mn^IIIMn^IV₂ScO₄] > [Mn^IIIMn^IV₂GdO₄]

The decrease in OAT rate correlates with the progressively higher pK_a (i.e. lower Lewis acidity) across this series of metals (pK_a(M(H₂O)_n): Mn³⁺ = 0.1, Sc³⁺ = 4.7, Gd³⁺ = 8.4).⁵¹ Most notably, changing from Mn to Gd results in decrease in the rate of OAT for more than three orders of magnitude.

2.2. Structural Reassignment of [Mn₃MO₂(H)] Clusters (M = Na⁺, Sr²⁺, Ca²⁺, Zn²⁺, Y³⁺).

Comparisons of the structural parameters between initially reported Mn₃MO₂ clusters and more recently prepared Fe₃MO(OH) show strong similarities. Through a combination of EPR, XRD and computational studies, these Mn clusters were reassigned as Mn₃MO(OH), with oxidation state of [Mn^III₃MO(OH)] as [Mn^III₃] and of [Mn^IIMn^III₂MO(OH)] as [Mn^III₂Mn^II]. A detailed discussion of these data and their interpretation is presented in SI, page S3. It should be noted that the changes in assignment of the oxidation states and identity of bridging ligands in clusters [Mn^III₃MO(OH)] and [Mn^IIMn^III₂MO(OH)] bear no effect on the conclusions of earlier studies on these complexes regarding the effect of the redox inactive metal on reduction potentials.⁴³ All comparisons within this series of complexes remain valid, as changes affect compounds across the entire series. The reactivity studies described next were focused on the available oxo-hydroxo complexes.

2.3. OAT Reactivity of [Mn₃MO(OH)] Clusters (M = Ca²⁺, Y³⁺).

OAT reactivity of complexes [Mn^III₃MO(OH)] and [Mn^IIMn^III₂MO(OH)] was investigated using phosphines as substrates (Scheme 3; Table 2, entries 8–15). Clusters [Mn^III₃MO(OH)] and [Mn^IIMn^III₂MO(OH)] displayed faster reactivity with phosphines than cubane complexes [Mn^IV₃MO₄] and [Mn^IIIMn^IV₂MO₄]. Triethylphosphine (PEt₃) and triphenylphosphine (PPh₃) were used, as PMe₃ resulted in reactivity too rapid for comparisons to be possible by ¹H NMR spectroscopy. Treatment of [Mn^IIMn^III₂CaO(OH)] with 10 equiv. PEt₃ led to conversion, over 36 h, to a product with very broad ¹H NMR features (Figure S16), typical of [LMn^II₃X₃] (X = ^–OAc or ^–OTf) species.⁵² This observation is consistent with loss of the redox inactive metal center following OAT. Similar to the case of cubane clusters, triethylphosphine oxide (OPEt₃), the product substrate oxidation, was not observed spectroscopically during the reaction. Removal of solvents and excess PEt₃ in vacuo after completion, followed by methanolysis and analysis by GC-MS allowed for identification of the OPEt₃ product. [Mn^IIMn^III₂CaO(OH)] also reacts with the weaker O-atom acceptor PPh₃, albeit much more slowly (>2 weeks; Figure S17).

Scheme 3.

OAT Reactivity of Mn₃ Oxo-Hydroxo Clusters [Mn^III₃MO(OH)] and [Mn^IIMn^III₂MO(OH)]

As expected, the more oxidized cluster [Mn^III₃CaO(OH)] reacts more rapidly with both PEt₃ and PPh₃; full conversion was achieved in 3 and 20 h, respectively (Figures S12 and S13). Similar to [Mn^IIMn^III₂CaO(OH)], OAT from [Mn^III₃CaO(OH)] results in formation of reduced Mn^II₃ species whose structural characterization was unsuccessful. Clusters displaying the more Lewis-acidic Y³⁺ ion yield faster reactivity. Reaction of [Mn^IIMn^III₂YO(OH)] with PEt₃ (10 equiv.) proceeded to completion in 30 min (Figure S14), whereas reaction with PPh₃ was complete in 1 h (Figure S15). As before, generation of Mn^II₃ species was suggested by the broad ¹H NMR features observed. Although isolation of monomeric oxidized complex [Mn^III₃YO(OH)] has not been achieved to date via either direct synthesis or transmetallation from [Mn^III₃CaO(OH)], a dimeric form of this complex, [Mn^III₃YO(OH)]₂, has been prepared from precursor complex LMn₃(OAc)₃ (¹H NMR, elemental analysis). XRD studies have only yielded a poor-quality structure (Figure 3), which is nonetheless sufficient for basic connectivity determination. In this complex, two [Mn₃YO(OH)] cores are bridged by two ^–OAc and one ^–OTf ligands, with five outer-sphere triflate counteranions. [Mn^III₃YO(OH)]₂ reacts rapidly with phosphine substrates: reaction with PEt₃ (10 equiv./Y) was complete in less than 15 minutes (Figure S10), and even the more electron deficient PPh₃ led to complete reaction in 30 minutes (Figure S11).

Figure 3.

Solid-state structure of [Mn^III₃YO(OH)]₂. Five molecules of outer sphere triflate anions and hydrogen atoms are omitted for clarity.

Qualitative analysis of the relative OAT rates for clusters [Mn^III₃MO(OH)] and [Mn^IIMn^III₂MO(OH)] reveals the following trend:

[Mn^III₃YO(OH)]₂ > [Mn^IIMn^III₂YO(OH)] > [Mn^III₃CaO(OH)] > [Mn^IIMn^III₂CaO(OH)]

As observed for the cubane clusters, the complexes containing more Lewis acidic Y³⁺ (pK_a = 8.6) undergo faster OAT processes than those containing Ca²⁺ (pK_a = 12.9), in some case by more than two orders of magnitude.⁵¹ Unlike their cubane analogues, which displayed clean transfer of a single oxygen atom to give trioxo clusters, oxo-hydroxo complexes appear to transfer both oxygen atom equivalents to yield [Mn^II₃] species. It is probable the first OAT event (presumably on the more sterically accessible μ ₂-bridge) relieves steric congestion in the cluster, increasing flexibility and allowing access to the second O-atom. Deprotonation of the hydroxide may precede OAT, though independent deprotonation studies were inconclusive. Formation of mono-oxo complexes en route to the [Mn^II₃] products is supported by the observation of intermediate species during the reaction of [Mn^IIMn^III₂YO(OH)] with PPh₃. After 30 minutes, ¹H NMR features consistent with formation of a [Mn₃MO] species were detected (Figure S3),⁴¹ leading to the proposed generation of tetrametallic monooxo species from [Mn^III₃MO(OH)] and [Mn^IIMn^III₂MO(OH)] clusters prior to the second OAT event and formation of the fully reduced products. Although quantification of the phosphine oxide produced in these processes was attempted in situ (³¹P NMR) as well as following workup (see the Experimental Section for the workup protocol), these procedures led to incomplete accounting of phosphine oxide product even in most cases where the stoichiometry of the reaction was well-established (e.g. the reactivity of cubane clusters [Mn^IIIMn^IV₂MO₄]), likely due to strong interaction of the phosphine oxide product with metal-containing byproducts. Thus, the stoichiometry of the reaction of Mn oxo-hydroxo clusters [Mn^III₃MO(OH)] and [Mn^IIMn^III₂MO(OH)] (or their Fe counterparts, vide infra) could not be conclusively assigned, except in the case of [Mn^IIIMn^IV₂ScO₄] which reacted to generate cleanly 1 equiv. of (O)PMe₃ (³¹P NMR).

2.4. OAT Reactivity of [Fe₃MO(OH)] Clusters (M = Ca²⁺, La³⁺, Sc³⁺).

Studies of the OAT reactivity of clusters [Fe^III₃MO(OH)] and [Fe^IIFe^III₂MO(OH)], isostructural to Mn-containing complexes [Mn^III₃MO(OH)] and [Mn^IIMn^III₂MO(OH)], respectively, were carried out with phosphine substrates in an analogous manner to those of cubane and oxo-hydroxo Mn₃ clusters (Scheme 4; Table 2, Entries 16–20). Treatment of clusters [Fe^III₃MO(OH)], containing three Fe^III centers, with electron rich phosphines such as PMe₃ yielded complex mixtures. Therefore, OAT behavior of [Fe^III₃MO(OH)] complexes was investigated with the more electron deficient substrate PPh₃. [Fe^III₃CaO(OH)] displays minimal conversion in the presence of PPh₃ (10 equiv., Figure S19). Conversely, [Fe^III₃LaO(OH)] reacts completely with PPh₃ (10 equiv.) over 14 hours, generating OPPh₃ as determined via ³¹P NMR (Figure S18). OAT reactivity of reduced clusters [Fe^IIFe^III₂MO(OH)] was studied using PMe₃ as substrate. In the presence of a 10-fold excess of PMe₃, [Fe^IIFe^III₂ScO(OH)] displayed complete conversion in 60 hours – slower than for the more oxidized clusters. The OPMe₃ product generated was observed during the reaction via ³¹P NMR (Figure S20). Treatment of clusters [Fe^IIFe^III₂LaO(OH)] and [Fe^IIFe^III₂CaO(OH)] with 10 equiv PMe₃ resulted in slower reactivity, with full conversion in 130 h and 270 h, respectively. As in the case with [Fe^IIFe^III₂ScO(OH)], generation of OPMe₃ was confirmed via ³¹P NMR as the reaction proceeded (Figure S21 and S22). Unfortunately, characterization of the metal-containing products of these transformations has thus far been unsuccessful. Based on the observed reactivity of the Mn₃ analogues [Mn^III₃MO(OH)] and [Mn^IIMn^III₂MO(OH)], it is reasonable to expect that OAT from complexes [Fe^III₃MO(OH)] and [Fe^IIFe^III₂MO(OH)] would also generate mono-oxo complexes as well as reduced Fe^II₃ species. However, due to the similarity of the characteristic ¹H NMR features of these families of Fe₃ compounds, deconvolution of the spectroscopic data for the products of OAT has not been successful. Similar to what was observed for Mn₃ complexes, OAT reactivity of Fe₃ oxo-hydroxo complexes [Fe^III₃MO(OH)] and [Fe^IIFe^III₂MO(OH)] is affected by the Lewis acidity of the redox inactive metal. In both oxidation states, the clusters containing the more Lewis acidic heterometal display faster OAT to phosphine substrates. [Fe^III₃CaO(OH)] and [Fe^IIFe^III₂CaO(OH)], containing the least Lewis acidic Ca²⁺ (pK_a = 12.9), react slower than clusters [Fe^III₃LaO(OH)] and [Fe^IIFe^III₂LaO(OH)] and [Fe^IIFe^III₂ScO(OH)], respectively, which contain stronger Lewis acidic metals (pK_a: La³⁺ = 9.06, Sc³⁺ = 4.8).

Scheme 4.

OAT Reactivity of Fe₃ Oxo-Hydroxo Clusters [Fe^III₃MO(OH)] and [Fe^IIFe^III₂MO(OH)]

2.5. Reactivity Trends.

Access to structurally well-defined tetraoxo and oxo-hydroxo clusters displaying both redox active (Mn, Fe) and redox inactive metals has facilitated systematic structure-reactivity studies (Table 2). Although effects of Lewis acidic, redox inactive metals on redox processes such as group transfer reactivity have been previously reported, to our knowledge, this study is the first to employ heterometallic oxo clusters for which an unambiguous structural assignment of the OAT precursors is available. While in previous literature reports redox inactive metal centers were included as additives (often in suprastoichiometric amounts), generating the species responsible for OAT in situ, the use of well-defined heterometallic clusters simplifies the interpretation of the observed reactivity. When redox inactive metal additives are used in excess to generate adducts with high-valent metal oxo species in situ, the effects of the Lewis acidity on the dynamic equilibria involved in adduct formation must be accounted for when analyzing the reactivity of the proposed adducts. In the present systems, however, the redox inactive metal centers are stoichiometrically incorporated in clusters that can be isolated and structurally characterized. For each type of cluster reported here, the first OAT step is qualitatively faster with increasing Lewis acidity (decreasing pK_a of metal aquo complex) of the redox inactive, apical metal center in the cluster. For oxo-hydroxo complexes, clusters in higher oxidation states display faster OAT reactivity than their reduced counterparts. The case of cubane clusters is more complex, as the high kinetic barrier to dissociation of an acetate ligand from the all-Mn^IV core in oxidized clusters [Mn^IV₃MO₄] inhibits the OAT process (vide supra). Notably, in several classes of clusters, the metal-containing species generated by the reaction of the well-defined heterometallic precursors with phosphines were observed to themselves undergo reactivity with phosphines. However, as neither the transiently-generated metal-containing species nor the final metal-containing products of these processes could be conclusively isolated, no further information on the reactivity of these systems can be teased out from the available data. Nonetheless, the qualitative analysis of the kinetics of the first step in this reactivity (i.e. the reaction of the well-defined isolated clusters with phosphine substrates) indicates that the choice of Lewis acid plays a significant role in tuning the OAT reactivity of these clusters.

Comparison of the OAT activity of complexes [Mn^IIIMn^IV₂GdO₄], [Mn^IIMn^III₂YO(OH)], and [Fe^IIFe^III₂LaO(OH)], which contain heterometals with similar pK_a’s (Gd3⁺ = 8.4, Y3⁺ = 8.6, La3⁺ = 9.06), enables assessment of the effect of various cluster architectures on reactivity. Cubane cluster [Mn^IIIMn^IV₂GdO₄] reacts slower (>2 weeks at r.t.) with PMe₃ than [Fe^IIFe^III₂LaO(OH)] (120 h), which in turn displays slower OAT activity than [Mn^IIMn^III₂YO(OH)] (30 min, PEt₃). A similar reactivity ordering is observed between clusters containing Ca2⁺ (pK_a = 12.9) as the redox-inactive metal; [Mn^IIMn^III₂CaO(OH)] reacts faster (36 h) than [Fe^IIFe^III₂CaO(OH)] (270 h; the corresponding cubane cluster, [Mn^IIIMn^IV₂CaO₄], could not be isolated).

The observed qualitative rate of OAT correlates not only with the pK_a of the redox-inactive metal incorporated in the cluster, but also with the reported one-electron reduction potentials for the three series of clusters. Fe₃ oxo-hydroxo clusters display overall intermediate potentials in the three series in the present study–the reduction potentials of Fe oxo-hydroxo complexes are typically more positive than those of cubane clusters and more negative than those of Mn oxo-hydroxo complexes (Tables 1 and 2).^{40, 42, 43} In agreement with the electrochemical data, the [Fe₃MO(OH)] complexes display slower reactivity than the more oxidizing [Mn₃MO(OH)] complexes, and in turn undergo faster OAT than the less oxidizing cubane clusters.

The exact mechanism of phosphine oxidation–involving electron transfer or concerted O-atom transfer⁵³–cannot be conclusively assigned in all series of clusters based on the available data. The faster OAT observed with clusters displaying more Lewis acidic metals, M, could be a consequence of a more electrophilic oxo motif, or an oxidant that is more prone to undergo single electron transfer. As indicated by computational studies, oxidation of phosphine by cubane clusters [Mn^IIIMn^IV₂MO₄] proceeds via concerted oxygen-atom transfer rather than via electron transfer. This assignment is consistent with the observed slower rate of OAT to the more sterically hindered PEt₃ (vs. the less bulky PMe₃ substrate). In the case of oxo-hydroxo clusters [Mn₃MO(OH)] and [Fe₃MO(OH)], on the other hand, a stepwise mechanism involving reduction of the cluster species by the PR₃ substrate (i.e. an electron-transfer process) cannot be ruled out.

3. CONCLUSIONS

The Lewis acidity of redox inactive metal centers incorporated in heterometallic Mn₃ and Fe₃ clusters was found to have a substantial effect on the rate of oxygen atom transfer to phosphine substrates. An important feature of these systems is the stoichiometric incorporation a redox-inactive metal center. Moreover, these precursors can be isolated and structurally characterized. Qualitatively, within the same structural motif and redox state, the rate of OAT of cubane [Mn₃MO₄] and oxo-hydroxo [Mn₃MO(OH)] and [Fe₃MO(OH)] clusters correlates with the Lewis acidity of the apical metal, similarly to previously observed one-electron reduction potentials of the same series of clusters. OAT rates can be accelerated by between one and three orders of magnitude, depending on the nature of the cluster. The more Lewis acidic metals are proposed to increase the electrophilicity of the oxo moieties, promoting OAT. Higher oxidation state clusters further exacerbate this effect, except in the case of all-Mn^IV cubane clusters, which display slow ligand dissociation and, consequently, slower OAT. Overall, this study demonstrates that the OAT reactivity of the clusters can be tuned significantly by choice of the Lewis acid.

EXPERIMENTAL SECTION

General Considerations.

Unless otherwise specified, all compounds were manipulated using a glovebox or standard Schlenk line techniques with an N₂ atmosphere. Anhydrous tetrahydrofuran (THF) was purchased from Aldrich in 18 L Pure-Pac™ containers. Anhydrous benzene, dichloromethane, and diethyl ether were purified by sparging with nitrogen for 15 minutes and then passing under nitrogen pressure through a column of activated A2 alumina (Zapp’s). Methylene chloride-d₂ and acetonitrile-d₃ were purchased from Cambridge Isotopes, dried over calcium hydride, and vacuum transferred prior to use. Benzene-d₆ was also purchased from Cambridge Isotope Laboratories, Inc., dried over sodium/benzophenone ketyl, and vacuum transferred prior to use. Unless indicated otherwise, all commercial chemicals were used as received. LMn₃(OAc)₃,⁵² [Mn^IV₃ScO₄],⁴⁰ [Mn^IV₃GdO₄],³⁹ [Mn^IIIMn^IV₂GdO₄],³⁹ [Mn^III₃CaO(OH)], [Mn^IIMn^III₂YO(OH)], [Mn^IIMn^III₂CaO(OH)],⁴³ [Fe^III₃LaO(OH)], [Fe^III₃CaO(OH)], [Fe^IIFe^III₂ScO(OH)], [Fe^IIFe^III₂LaO(OH)], and [Fe^IIFe^III₂CaO(OH)]⁴² were prepared according to previously published protocols. ¹H and ³¹P NMR spectra were recorded on a Varian Mercury 300 spectrometer at room temperature. Elemental analyses were performed by Robertson Microlit Laboratories, Ledgewood, NJ.

Synthesis of [Mn^III₃ScO₃].

In an inert atmosphere glovebox, [Mn^IV₃ScO₄] (13.5 mg, 0.0103 mmol, 1 equiv) was initially dissolved in C₆D₆ in a J. Young NMR tube. To the wall of the J. Young tube was added PMe₃ (11 μL, 0.103 mmol, 10 equiv) quickly via a Hamilton syringe. The J. Young tube was sealed and brought outside the glovebox. The reaction mixture was heated to 50 °C using an oil bath. The reaction was monitored over time by ¹H NMR spectroscopy and was complete after 30 h. The solvent and excess unreacted PMe₃ were removed under vacuum. The residue was washed with Et₂O to remove trace PMe₃. The residue was resuspended in benzene and then filtered. Crystals were obtained via vapor diffusion of Et₂O into a concentrated benzene solution to give the clean product in 56% yield. ¹H NMR (C₆D₆, 300 MHz): 17.62 (v br), 10.69 (v br), 9.44 (br overlapped), 9.05 (v br overlapped), 4.57 (br), 3.53 (br), 2.34 (br), 1.84 (br), 1.63 (br), 1.35 (m overlapped), –18.98 (v br) ppm. Anal. Calcd. for: C₆₃H₄₈Mn₃N₆O₁₂Sc ([Mn^III₃ScO₃]) (%): C, 58.62; H, 3.75; N, 6.51. Found: C, 57.88; H, 3.70; N, 6.65.

Synthesis of [Mn^III₃YO(OH)]₂.

To a suspension of LMn₃(OAc)₃ (150 mg, 0.125 mmol, 1 equiv) in DME was added Y(OTf)₃ (67 mg, 0.125 mmol, 1 equiv). The mixture was stirred for 5 min then PhIO was added (58 mg, 0.263 mmol, 2.1 equiv). The mixture was stirred for 1 h then filtered. The purple filtrate was collected and the solvent was removed under vacuum. The DME fraction was recrystallized in CH₃CN/Et₂O to give clean product as precipitate in 71% yield. ¹H NMR (CD₂Cl₂, 300 MHz): 77.49 (v br), 67.83 (m overlapped), 52.93 (m br overlapped), 41.12 (v br), 38.51 (v br), 35.34 (v br), 25.17 (m br overlapped), 17.40 (br overlapped), 8.3 (m br overlapped),–21.81 (m br overlapped), –24.47 (d br overlapped), –28.05 (m br overlapped), –33.46 (m br overlapped) ppm. Anal. Calcd. for: C₁₃₂H₉₆F₁₈Mn₆N₁₂O₄₀S₆Y₂ [Mn^III₃YO(OH)]₂ (%): C, 44.89; H, 2.74; N, 4.76. Found: C, 44.89; H, 2.94; N, 4.57.

Protocol for reactivity studies of metal-oxo clusters with phosphine substrate.

In an inert-atmosphere glovebox, a J. Young NMR tube was charged with a solution ~12–15 mM) of the desired cluster complex in an appropriate NMR solvent (CD₂Cl₂, CD₃CN, or C₆D₆). Phosphine (PPh₃, PEt₃, or PMe₃; 2–10 equiv) was added neat via gas-tight microsyringe or as a stock solution in deuterated solvent. Reaction progress was monitored via ¹H and ³¹P NMR as appropriate until full (>95%) conversion was observed by disappearance of signal corresponding to starting material. Samples requiring heating were kept at elevated temperature in mineral oil baths. When phosphine oxide products were not observed during reaction via ³¹P NMR, samples were returned to a glovebox, where volatiles were removed in vacuo. On the benchtop, the resulting residue was dissolved in dichloromethane, filtered through a short silica plug, and further eluted with methanol. The resulting solution was analyzed by GC-MS to confirm the presence of phosphine oxide product.

EPR Spectroscopy.

Perpendicular-mode continuous-wave (CW) X-Band (9.33–9.37 GHz) spectra were collected using a Bruker E500 spectrometer with a super-high Q resonator (SHQE). Parallel-mode CW X-Band spectra (9.3 GHz) were collected using a dual-mode cavity (ER 4116DM). All EPR spectra were collected under non-saturating conditions. Temperature control was maintained with an Oxford Instruments model ESR900 helium-flow cryostat with an Oxford ITC 503 temperature controller. Temperature control was maintained with an Oxford Instruments CF935 helium-flow cryostat.

Computational Protocol.

Density functional theory (DFT) calculations were carried out on complexes [LMn^III₂ Mn^IVCaO₂(OTf)(DME)[OTf]₂ (oxCa’), [LMn^III₃CaO (OH)(OTf)(DME)][OTf]₂ (oxCa”), [LMn^III₃O₂Ca(OTf) (DME)[OTf] (redCa’), and [LMn^IIMn^III₂CaO(OH) (OTf)(DME)][OTf] (redCa’’ and redCa’’-noH) to elucidate the structure of complexes [Mn^III₃CaO(OH)] and [Mn^IIMn^III₂CaO(OH)]. Complexes oxCa’ and redCa’ were optimized using the solid-state structure without outer-sphere solvent molecules as the initial geometry input whereas complexes oxCa’’ and redCa’’ were modified by placing one triflate in the vicinity (~3.0 Å) of the bridging hydroxo motif for the initial geometry input. To decrease computational expenses, a “truncated ligand” approximation was employed in which the three carbon atoms connecting the bipyridine motifs were spatially constrained, and the remaining ligand scaffold connecting each carbon atom to the phenyl backbone was substituted with three protons. This approximation is expected not to influence bond parameters for the Mn₃O₂Ca core.

Schrodinger computational products Jaguar 8.4⁵⁴ and Maestro were employed respectively to perform gas-phase geometry optimization and measure the bond lengths and angles in the optimized complexes. Calculations were accomplished with the basis set/functional pairing of unrestricted dispersion-corrected B3LYP^55-57-d3⁵⁸ and basis set lacvp** for non-metals C, H, N, O, S, and F along with basis set lacv3p** for Mn.^{59, 60} Successful ground-state optimizations were assessed by the forces on the all atoms residing below 10^–5 hartrees/bohr as well as total energy remaining unchanged by 10⁻⁴ between the final SCF iterations. Oxidation states of clusters oxCa’, oxCa”, redCa’, and redCa’’ were tabulated through Mullikan spin populations and inspection of the spin-density plots. For structure redCa’’, wavefunctions with (redCa’’) and without (redCa’’-noH) hydrogen bonding between triflate and the hydroxide motif were isolated.

Synopsis:

The effects of Lewis acidic metal ions on the oxygen-atom transfer reactivity of series of structurally well-defined heterometallic clusters have been investigated. Complexes containing more Lewis acidic ions display faster oxygen-atom transfer to phosphines indicating that, even in cases where redox-inactive metal ions (M) are used, they can regulate group transfer reactivity.