Osmium(IV) Tetraaryl Complexes Formed from Prefunctionalized Ligands

Abstract

Transition metal(IV) tetraaryl, M(aryl)₄, complexes hold great promise as functional building blocks for complex organometallic materials, yet their widespread utility depends on the development of improved synthetic protocols and a deeper understanding of their chemical structure–property relationship(s). Here we show that Os(aryl)₄ complexes with preinstalled functional groups (–F, –Cl, –Br, –I, and –SMe) can be prepared from reactions between (Oct₄N)₂[OsBr₆] and Grignard reagents formed by halogen/magnesium insertion or exchange. This approach provides access to compounds that may otherwise prove challenging to prepare through postfunctionalization strategies, such as those comprising halogens in the 2- or 5-positions. We characterize these, as well as previously reported, materials using single-crystal X-ray diffraction, solution voltammetry, and UV–vis spectroscopy. Through a comparison of 13 differently substituted complexes, we identify correlations between their electrochemical and optical gaps, and between the E_1/2 of their 0/1+ redox event and an adjusted Hammett parameter that accounts for all aryl ligand substituents. The electronic property trends observed are further rationalized through a series of first-principles calculations based on density functional theory. Together, this work lays a foundation for advancing new preparative methods to further derivatize such species, and a robust experimental data set to help benchmark future experimental and computational compound characterization.

Short abstract

Os(aryl)₄ complexes with preinstalled functional groups can be prepared from reactions between (Oct₄N)₂[OsBr₆] and Grignard reagents formed by halogen/magnesium insertion or exchange. Our approach provides access to compounds that may otherwise prove challenging to prepare through postfunctionalization strategies, laying a foundation for advancing new preparative methods to further derivatize such species. Through a comparison of 13 differently substituted complexes, we identify structure-electronic property relationships that may serve to benchmark future experimental and computational compound characterization.

Introduction

Though the chemistry of transition metal(IV) tetraaryl, M(aryl)₄, complexes¹ has developed relatively slowly over the past ∼50 years, significant milestones may now be identified. Among the first homoleptic, charge neutral compounds shown to be stable at room temperature were the group six M(mesityl)₄ (M = Cr,² Mo³) complexes introduced by Seidel from 1976. Robust air-stable analogues, for example, featuring Os(IV)⁴ and Ru(IV)⁵ centers with 2-methylphenyl ligands were subsequently advanced by Wilkinson in the 1980s.¹ Here the presence of an aryl ortho-methyl substituent was found essential to improve thermal stability, where Os(phenyl)₄ was found to decompose above −20 °C.⁶ Other compounds with Ti(IV),⁷ V(IV),⁸ Cr(IV),⁹ and Mo(IV)³ centers have been reported to be air-stable. In recent years, targeted applications have been identified for these complexes which provide motivations for their study that move beyond fundamental exploration. Notably, their tetrahedral geometries and associated distinct electronic structures has driven interest in their potential utility as redox-active, electronically transparent building blocks for extended polymeric materials or molecular electronic components,¹⁰⁻¹² or as molecular qubits for quantum information science.^13,14 Despite this progress, the robust synthetic framework needed to prepare advanced derivatives of such complexes remains largely underdeveloped, providing inspiration for new studies focused on advancing preparative methods to expand their potential utility. These investigations will likely also play a foundational role in determining chemical structure–property relationships useful in the identification of specific M(aryl)₄ complexes targeted for distinct applications.

Among the most established members of the M(aryl)₄ family are the complexes with osmium(IV) central atoms. This is attributable to their stability under ambient conditions in solution,⁴ and with respect to purification by chromatography,¹⁵ as well as their reversible redox chemistry.¹⁶ While these compounds are known to react, for example, with small nucleophiles such as PMe₃ or CO to form osmium(II) η⁶-biaryl complexes,¹⁷ or with 3-chloroperoxybenzoic acid or ^tBuOOH to form the osmium(VI) tetraaryl monooxo species,¹⁵ they have the capacity to serve as excellent model substrates to develop σ-aryl ligand chemistry.¹⁰ Notably, these complexes are reported to withstand Friedel–Crafts acylation,^10,18 bromination,^10,12 and subsequent Suzuki,¹⁰ Ullmann-like,¹⁰ or Negishi¹² cross-coupling reactions. Such transition metal-catalyzed processes have facilitated the isolation of σ-biaryl complexes functionalized with carboxylic acid, amine, fluorine, hydroxy, aldehyde, or acyl substituents,¹⁰ oligomeric materials which have to date only been partially characterized,¹⁹ and a tetrakis(4-ferrocenyl-2,5-xylyl)osmium(IV) complex that exhibits a particularly rich redox chemistry.¹² Conversely, attempts to introduce phenylethynyl or pinacol boronate substituents, through Stille coupling and Miyaura borylation, respectively, have to date provided only low/trace yields of mono- and disubstituted products.¹⁹

While the functionalization of bound σ-aryl ligands remains an important area of development, such approaches require parent complexes to survive a range of different reaction and workup conditions. Low yielding reactions that result in product mixtures comprising heteroleptic complexes (having both substituted and nonsubstituted aryl ligands) may also prove difficult to separate (noted below). In contrast, the use of prefunctionalized ligands to form target complexes directly offers several advantages. Intact homoleptic complexes can be prepared in a single step using established M-aryl bond forming strategies and may be more easily isolated. By incorporating distinct, or more reactive, functional groups than those which can be introduced through other means (i.e., –I instead of –Br), milder conditions may be utilized to realize subsequent chemical transformations from the resulting M(aryl)₄ complex. To date, however, the only substituents incorporated into Os(aryl)₄ complexes using prefunctionalized aryl ligands are alkyl (–Me, –Et, –ⁱPr), and fluoride—in the preparation of tetrakis(4-fluoro-2-methylphenyl)osmium(IV) (Os1–4F) using 4-fluoro-2-methylphenyl magnesium bromide.²⁰ The synthesis of tetrakis(4-methoxy-2,5-dimethylphenyl)osmium(IV) (Os2–4OMe) was attempted, but only the monooxo species could be isolated.¹⁵ We note that to help facilitate discussions of these compounds, in this work we use the prefixes Os1-and Os2-as part of our naming convention to indicate that the given Os(IV) complex comprises functionalized 2-methylphenyl (2-tolyl) or 2,5-dimethylphenyl (2,5-xylyl) ligands, respectively.

Given their capacity to undergo transition metal-catalyzed transformations, the formation of osmium(IV) σ-aryl complexes comprising different halogens is an important initial target. In support of this assertion, we recognize that the development of robust methods to synthesize halogenated metallocenes, most prominently ferrocene, has and continues to prove instrumental in the development of complex organometallic materials comprising metallocene units.²¹ Accordingly, here we synthesize and characterize seven new homoleptic osmium(IV) tetraaryl complexes with halide (X = –F, –Cl, –Br, –I) substituents in the 2-, 4-, or 5- positions (Figure 1). In addition, a thioether (–SMe)-substituted complex was prepared as these groups can be used to connect this complex between nanoscale gold electrodes for single-molecule conductance measurements.²² These compounds are formed through reactions between (Oct₄N)₂[OsBr₆] and the appropriate aryl Grignard reagents prepared by halogen/magnesium insertion or exchange, in yields which range from 10 to 77%. We emphasize that several of the compounds prepared by magnesium exchange (including Os2–4I, Os1–4Br, and Os1–5Br) may not be easily accessed by other pre or postfunctionalization synthetic methods, clearly distinguishing the present strategy from previous preparative approaches.^10,11 Single-crystal X-ray diffraction studies of five of the new halogenated compounds confirm they exhibit a pseudotetrahedral structure in the solid state. These complexes, as well as those previously reported by our group, are further characterized in a series of solution electrochemical and UV–vis experiments. We show that while all 13 of these compounds are panchromatic and strongly absorb in the visible region, to the extent that they appear as darkly colored or black solids, they exhibit a rainbow of different colors in solution. Notably, the color of a given compound appears to be influenced by their σ-aryl ligand substituents through small changes to the λ_max and relative intensity of their four primary absorbance features. Further analysis of this data reveals correlations between their electrochemical and optical gaps, as well as between the first oxidation potential and an adjusted Hammett parameter (considering all the substituents on the aryl ring), that may facilitate useful qualitative property predictions for new compounds and help identify complexes with unusual properties. To support our experimental results, we perform density functional theory (DFT)-based calculations for selected complexes which provide additional insights into their electronic structure.

Figure 1.

Left: Synthetic route to homoleptic osmium(IV) σ-aryl complexes. Right: Targeted aryl Grignard reagents are prepared through halogen/magnesium insertion or exchange reactions using “Turbo Grignard” reagents.²³ Bottom: Chemical structures of the prefunctionalized aryl ligands used in this study. Os1 is included here as a control. The prefixes “Os1–” and “Os2–” indicate that the given complex comprises functionalized 2-methylphenyl or 2,5-dimethylphenyl ligands.

Results and Discussion

Synthesis

Recent work from our group has established that (Oct₄N)₂[OsBr₆] (an Os(IV) salt) is a more effective starting material for the synthesis of Os(IV)(aryl)₄ complexes than the previously used OsO₄ (comprising an Os(VIII) center). Here, the supporting tetraoctylammonium cations are used primarily to facilitate the dissolution of the salt in common organic solvents (e.g., THF). Through reaction of the Os(IV) salt with aryl Grignard reagents, known complexes can be prepared in higher yields, with reduced amounts of OsO(aryl)₄ and OsO₂(aryl)₂ side products that can complicate chromatography purification of the target homoleptic material. Our approach also provides access to Os(aryl)₄ species that could not easily be isolated from reactions with OsO₄, such as Os(mesityl)₄ (Os3).^11,24 We reasoned that this same strategy could be used to form Os(aryl)₄ complexes with aryl Grignard reagents bearing useful functional groups rather than simple alkyls, if these could be accessed through halogen/magnesium insertion or exchange reactions (Figure 1).

Aryl Grignard reagents are formed through halogen/magnesium insertion through reaction of activated magnesium metal with an appropriate aryl halide, typically a bromide.²⁵ Accordingly, for successful incorporation into the tetraaryl complex, any additional aryl functional groups must exhibit a reduced reactivity under these conditions (as is known for –F or –Cl substituents) or prove inert to substitution (such as is the case for –SMe groups). Using this approach we found we could indeed synthesize tetrakis(4-fluoro-2,5-dimethylphenyl)osmium(IV) (Os2–4F), tetrakis(4-chloro-2,5-dimethylphenyl)osmium(IV) (Os2–4Cl), and even tetrakis(4-bromo-2,5-dimethylphenyl)osmium(IV) (Os2–4Br) in yields of 77, 60, and 59%, respectively (entries 1–3, Table 1). Tetrakis(2,5-dimethyl-4-(methylthio)phenyl)osmium(IV) (Os2–4SMe) was also obtained, albeit in 23% yield (entry 8, Table 1). All aryl Grignard reagents used in these syntheses were prepared from the corresponding aryl bromides. Conveniently, 4-bromo-2,5-dimethylphenyl magnesium bromide can be formed from the reaction between 1,4-dibromo-2,5-dimethylphenyl and one equivalent of magnesium.²⁶ While this single-step preparation of Os2–4Br from (Oct₄N)₂[OsBr₆] is useful, we note the same method may not easily be applied to form other bromoalkyl complexes such as tetrakis(4-bromo-2-methylphenyl)osmium(IV) (Os1–4Br). Any attempts to form the necessary aryl Grignard reagent from 1,4-dibromo-2-methylbenzene would likely activate the less sterically hindered, and less electron rich 4-position. With the methyl group positioned meta to the installed -MgBr group, the resulting tetraaryl complexes would not comprise the stabilizing aryl ortho-substituents required to impart high thermal stability.

Table 1. Summary of Os(aryl)₄ Synthetic Yields^a.

entry	compound	yield (%)
		halogen/magnesium insertion	halogen/magnesium exchange
1	Os2–4F	77	-
2	Os2–4Cl	60	-
3	Os2–4Br	59	36
4	Os2–4I	-	10
5	Os–2Cl	-	68
6	Os1–4Br	-	33
7	Os1–5Br	-	10
8	Os2–4SMe	23	-
9	Os1	73b	35

^aIsolated yields obtained after purification by column chromatography.

^bYield previously reported by Parr et al.¹¹

Knochel and others have elegantly demonstrated that a wider array of functional aryl Grignard reagents, including esters, amide, nitrile, and nitro groups that are typically incompatible with halogen/magnesium insertion strategies,^27,28 may be formed through halogen/magnesium exchange.²⁹ Such processes have also been shown to be chemoselective, converting iodoaryl substituents to the Grignard reagent in the presence of aryl bromide groups, and forming the most stable magnesiated compound from dibromoaryls comprising halides in different chemical environments.³⁰ In a commonly utilized scheme, a functionalized aryl iodide is reacted with iPrMgCl•LiCl, or “Turbo-Grignard” reagent, to form the target aryl-MgCl•LiCl species.²⁹ In the scope of the present study, we reasoned that bromo/iodoaryl Grignard reagents formed through such exchange processes could be used to prepare the corresponding tetrahaloaryl osmium(IV) complexes.

We first evaluated the utility of this strategy through a control study targeting the synthesis of tetrakis(2-methylphenyl)osmium(IV) (Os1). Conditions to form the aryl Grignard reagent were optimized by reacting 1 equiv of iPrMgCl•LiCl with 2-iodotoluene at different temperatures, followed by quenching the reaction with 2,2-dimethylpropanal (pivaldehyde). Analysis of the crude product mixture by ¹H NMR showed that while incomplete conversion was observed at – 78 and −42 °C, high conversion could be obtained after 0.5 h at 0 °C. A representative synthetic protocol and ¹H NMR spectrum is provided in the Supporting Information, in which we confirm we achieve >90% conversion of 2-iodotoluene to the quenched product, 2,2-dimethyl-1-(2-methylphenyl)propan-1-ol, using 1,3,5-trimethoxybenzene³¹ as an internal standard (Supporting Information, Figure S1). Iodoaryls that are more electron deficient relative to 2-iodotoluene are expected to react more rapidly,²³ ensuring the suitability of these conditions for the preparation of haloaryl Grignard reagents in good yields. We subsequently reacted the resulting 2-tolyl magnesium chloride-lithium chloride with (Oct₄N)₂[OsBr₆], using conditions that were otherwise identical to the reactions using Grignard reagents formed by insertion. Remarkably, we were able to isolate Os1 in a reasonable, albeit significantly lower, yield than that obtained with Grignard reagents formed through insertion (35% vs 73%, entry 9, Table 1). Encouraged by this result, we directly applied the same methodology toward the synthesis of different tetrakis(haloaryl)osmium(IV) complexes. Tetrakis(4-bromo-2,5-dimethylphenyl)osmium(IV) (Os2–4Br), tetrakis(4-iodo-2,5-dimethylphenyl)osmium(IV) (Os2–4I), tetrakis(2-chlorophenyl)osmium(IV) (Os–2Cl), tetrakis(4-bromo-2-methylphenyl)osmium(IV) (Os1–4Br), and tetrakis(5-bromo-2-methylphenyl)osmium(IV) (Os1–5Br) were isolated in 36, 10, 68, 33, and 10% yields, respectively (entries 3–7, Table 1).

The reduced yields of Os(aryl)₄ typically observed using aryl-MgCl•LiCl prepared by exchange compared to aryl-MgBr generated through insertion may be attributable to several factors (entries 3 and 9 provide a direct comparison, Table 1). We hypothesize that the presence of chloride salts in the reactions involving aryl-MgCl•LiCl could lead to halide exchange at the Os(IV) center, and modify the solubility and equilibria associated with the [Oct4N]X/MgX₂ salts expected to form during halide substitution of the OsBr₆^2– ion by σ-aryl ligands, slowing reaction progress. Indeed, we previously found that yields of Os(aryl)₄ formed from (Oct₄N)₂[OsCl₆] are lower (≤40%) than those obtained from (Oct₄N)₂[OsBr₆] (≤73%).¹¹ Furthermore, the identity of the halide and aryl substituents is likely to influence the solution structure (monomer, dimer, tetramer) and complex equilibria (e.g., between different solution structures, the Schlenk equilibrium) of the aryl Grignard reagent, which could modulate its nucleophilicity.²⁵ We also recognize that unreacted ⁱPrMgCl•LiCl may attack the Os(IV) center directly, forming heteroleptic complexes which may not be stable upon reaction workup. Product yields may be improved through future studies focused on careful controls and reaction optimization, for example, using appropriate conditions to form aryl-MgX from other exchange reagents such as ⁱPrMgBr.²⁸ This work could further help to rationalize the unusually high yield of Os–2Cl, beyond noting that this aryl is the most electron poor of the series studied and so is expected to form the most stable Grignard reagent. In contrast, the particularly low yields of Os1–5Br and Os2–4I are attributed to competing magnesiation at the 5-bromo or 4-iodo sites, respectively (forming the mono or bis-Grignard reagent). This could result in, for example, heteroleptic Os(aryl)₄ side products comprising unsubstituted 2-methylphenyl or 2,5-dimethylphenyl ligands after protic work up that complicate isolation of the target material. We note that any such reduced chemical selectivity is expected to be less problematic in the case of Os1–4Br than for Os1–5Br (as reflected in the comparable yields of Os1–4Br and Os1 prepared by this route). The 5-bromo position of the 5-bromo-1-iodo-2-methylbenzene precursor is more activated toward substitution than the 4-bromo position of 4-bromo-1-iodo-2-methylbenzene, as indicated by the Hammett σ_m(I) and σ_p(I) constants for iodide of 0.35 and 0.18, respectively.³²

The realization of this synthetic approach, and the corresponding characterization of the first complete structurally homologous series of halogenated Os(aryl)₄ complexes, motivates several further points of discussion. Counterintuitively, ¹H NMR studies of Os2–4X (X = F, Cl, Br, I) show the chemical shift of the resonance assigned to the 6-proton (established by NMR 2D correlation spectroscopy for Os2–4F and Os2–4SMe), shifts upfield with increasing halogen electronegativity (from 7.25 ppm for Os2–4I to 6.62 ppm for Os2–4F). Significant J_C–F (and to a lesser extent J_H–F) couplings are observed in the spectra of Os2–4F, as previously noted for Os1–4F.²⁰ The isolation of Os–2Cl demonstrates that the stabilization of Os(aryl)₄ complexes is not limited to 2-alkyl groups, and that halogens can be introduced in the 2- as well as the 4- (and 5-) position(s). Analogous neutral, homoleptic metal(IV) complexes with V,³³ Nb,³⁴ and Cr^35,36 centers and aryl ligands comprising ortho-chloride substituents have been isolated. For completeness, we note that Ti(III)(C₆Cl₅)₄, [Pt(III)(C₆Cl₅)₄]⁻, and Pt(IV)(C₆Cl₅)₄ complexes have also been reported, exhibiting approximate tetrahedral, square planar, and octahedral (with two coordinating ortho-chloride atoms) geometries, respectively.³⁷⁻³⁹ Building on Wilkinson’s preparation of Os1–4F, in which a fluorinated 2-methylphenyl ligand is installed directly at the osmium center, our synthesis of Os1–4Br, and Os1–5Br shows that analogous brominated complexes can be realized through a similar approach, and that these functional groups can be installed in either the 4- or 5-positions.

We note there are now four distinct methods to form the synthetically useful precursor Os2–4Br. Previously this has been prepared by (1) reaction of tetrakis(2,5-dimethyl)osmium(IV) (Os2) with pyridinium tribromide/Fe powder (52% yield),¹⁰ and (2) reaction of Os2 with N-bromosuccinimide/H₂SO₄ (73%).¹² Here we show it can be produced by (3) direct interaction of aryl-MgBr (insertion product, 59%) or (4) aryl-MgCl•LiCl (exchange product, 35%) with (Oct₄N)₂[OsBr₆]. In future, with appropriate scale-up of one or more of these reactions, we propose that Os2–4Br can be used to demonstrate a range of new σ-aryl ligand derivation strategies⁴⁰ and a wide array of Os(aryl)₄-based materials with interesting properties. One potentially rich area of study is lithium–bromide exchange at the σ-bromoalkyl ligands, followed by subsequent reaction of the lithiated tetraaryl species with an appropriate electrophile. In the Supporting Information, Figure S2, we present ¹H NMR spectra that provides clear evidence for the formation of Os2 from the reaction between Os2–4Br and n-BuLi, followed by addition of MeOH. While this result is encouraging, our attempts to quench such lithiated tetraaryl intermediates with other electrophiles such as chlorodiphenylphosphine (PClPh₂) have to date resulted in multiple reaction products that prove difficult to isolate via column chromatography. Clearly, access to larger quantities of Os2–4Br will help facilitate an expanded and systematic series of postfunctionalization studies to drive further synthetic advances.

X-ray Crystallography

In Figure 2 we present molecular structures for five of the seven new complexes reported here, as determined using single-crystal X-ray crystallography. Additional structural parameters are provided in the Supporting Information, Table S1. All complexes exhibit average Os–C bond lengths typical of Os(aryl)₄ complexes, between 1.9965(8) (Os1–4Br) and 2.0560(10) Å (Os–2Cl). The ligand environment around each Os(IV) center exhibits a distorted tetrahedral geometry, which we further characterize through calculation of the “tetrahedricity,” or T-value (analogous to an “octahedricity” metric^41,42 introduced previously).¹¹ We determine the T-value for a given complex by taking the root-mean-square deviation of the set of C–M–C angles from 109.5°, which quantifies the degree to which the observed structure deviates from that expected for a perfect tetrahedral geometry (where T-value = 0). For comparison, a T-value of 43.71 is obtained for a square planar geometry. The T-values of the new complexes reported here follow expectations based on previous studies, with Os1-X (T-value ≥4.38) exhibiting larger distortions from perfect tetrahedral geometries than Os2–4X (T-value ≤2.25). While it remains difficult to establish robust relationships between T-values and the nature, position, and number of substituents on the aryl ligands, in an earlier study we attributed the large tetrahedral distortions observed for M(mesityl)₄ (M = Os (Os3), Ru; T-value ≥7.42) to the significant steric bulk of this ligand.

Figure 2.

X-ray crystal structures for (a) Os2–4F, (b) Os2–4Cl, (c) Os2–4I, (d) Os–2Cl, and (e) Os1–5Br (50% probability ellipsoids). Hydrogen atoms are omitted for clarity (Os = teal, C = gray, F = light blue, Cl = green, I = purple, Br = orange). Selected structural parameters for these and selected, previously characterized, tetrahedral compounds are provided in the Supporting Information, Table S1.

Given the challenges in correlating T-values with ligand structure, we also previously suggested that the extent of tetrahedral distortion observed for a given Os(aryl)₄ complex may be influenced by crystal packing effects in the solid state.¹² The crystal structure of Os–2Cl provides evidence in support of this hypothesis. In contrast to other structurally characterized Os(aryl)₄ complexes, the asymmetric unit for the Os–2Cl structure comprises >1 symmetry-independent chemical entities (Z′ = 4). Each of these entities exhibits a different T-value, ranging from 2.48 to 3.70 (these values are intermediate between those obtained for complexes with 2-methylphenyl and 2,5-dimethylphenyl based ligands). Clearly, the through-bond, and intramolecular, influences of the σ-aryl ligand substituents are not the only contributing factors to the geometry around the central metal atom. We note that the Os–2Cl structure is of additional interest given the close proximity of the ortho-chloride substituents to the Os(IV) center. While 4-coordinate tetrahedral geometries and moderate M-Cl distances were observed for structurally related Cr(IV) or V(III) tetraaryls comprising ligands with ortho-chloride substituents, Cr(III)(aryl)₄ complexes (aryl = 2,6-dichlorophenyl, 2,3,4,5,6-pentachlorophenyl) have been observed to form 6-coordinate geometries with coordinating Cr–Cl contacts at distances <3 Å.^35,43,44 A similar structural change from a square planar to an octahedral geometry was observed upon oxidation of [Pt(III)(C₆Cl₅)₄]⁻ to Pt(IV)(C₆Cl₅)₄.³⁸ Critically, we find that the minimum Os–Cl distance in Os–2Cl is 3.266 Å (average = 3.316 Å), providing no evidence that such bonding interactions are present in this Os(IV) species. However, we recognize that the possibility of such an interaction may have implications for the redox chemistry of Os–2Cl upon reduction (discussed below).

UV–vis Spectroscopy and Solution Electrochemistry

To supplement our spectroscopic and electrochemical analysis, we compile characterization data for previously reported complexes in addition to the compounds reported here for the first time. Specifically, data for Os1, Os2, and Os3 enables direct comparisons of functionalized and parent complexes and illustrates the influence of increasing numbers of aryl methyl substituents. Data for tetrakis(2-ethylphenyl)osmium(IV) (Os–2Et) and for tetrakis(2-isopropylphenyl)osmium(IV) (Os-2ⁱPr) can be compared with that of other complexes with 2-substituents, Os1 and Os–2Cl.

While the dark coloration of Os(aryl)₄ complexes has previously been reported,^6,10,20,45 here we emphasize that these species are so strongly absorbing that they can appear black in the solid state (Figure 3a-left). We further show, in Figure 3a-right, that as dilute solutions these compounds can generate a rainbow of colors which vary depending on the σ-aryl substituents. In Figure 3b,c we present overlaid normalized UV–vis spectra for selected complexes measured in CH₂Cl₂. Additional UV–vis spectra are provided in Supporting Information, Figure S3, with spectroscopic data given in Supporting Information, Table S3. Each spectrum exhibits the four primary absorption features with ε ∼ 10³ M^–1 cm^–1 that are characteristic of this class of complexes.⁴⁵ These bands have been attributed to d–d transitions, tentatively involving transitions from a singlet ground state to a triplet excited state that is quadruply split by spin–orbit coupling resulting from the presence of the heavy osmium center.²⁰ In support of this assignment, it was noted that these features lack charge transfer character, exhibiting weaker intensities that would be otherwise expected (i.e., ε < 10⁴–10⁵ M^–1 cm^–1), with invariant λ_max when measured in different solvents.

Figure 3.

(a) Left: Tetrakis(aryl)osmium(IV) complexes are typically isolated as dark powders (the example shown is Os2–4Br). Right: These compounds readily dissolve in organic solvents such as CH₂Cl₂ to produce intensely colored solutions of various colors, as illustrated in these representative examples (1 mM). (b–c) Overlaid UV–vis spectra for selected new compounds prepared here, in addition to those reported previously,¹¹ show each exhibits four primary absorption features in the visible region. The varying color of each solution is dependent on both the λ_max and relative ε of each feature, which vary depending on the aryl ligand substituents. Additional UV–vis spectra and spectroscopic data for all compounds are provided in Supporting Information, Figure S3 and Table S3.

From comparisons of spectra from the 13 different complexes characterized here, we find the observed changes in solution color appears to be due to small changes in λ_max and the relative ε of each feature. This is illustrated in Figure 3b, in which the overlaid spectra of Os1 and Os2 are shown to be highly comparable to each other (yielding purple solutions), yet clearly distinct from that of Os3 (green solution). For Os3 the three lowest energy bands undergo a bathochromic shift, with the features at ∼700 and ∼500 nm increasing and decreasing in relative intensity, respectively. As exemplified in Figure 3c, we find that the absorbance spectra of Os2–4X (purple solutions) are broadly independent of the identity of the halogen, but all features undergo a bathochromic shift of 7–33 nm when the 4-substituent is changed to a methanethiol group (blue solution).

We further characterized the redox properties of this series of complexes through solution electrochemical studies in 0.1 M (ⁿBu)₄NPF₆–CH₂Cl₂. Representative overlaid cyclic voltammograms for Os2–4X are provided in Figure 4a, and for Os–2Cl, Os1–4Br, and Os1–5Br in Figure 4b. We note that in these plots the voltammogram of Os2–4I was measured at 1 V/s, rather than the 0.1 V/s used for all other compounds, due to the improved reversibility of redox features for this species at faster scan rates. As previously found for analogous Os(aryl)₄ complexes,¹¹ the new compounds reported here typically exhibit reversible oxidation and reduction features corresponding to 0/1+ and 1–/0 events (i_pa/i_pc ∼ 1, i_p ∝ V_s^1/2; Supporting Information, Table S2). In addition, we find that all complexes with substituents in the 4-position (Os2–4X, Os1–4Br, and Os3) exhibit an electrochemically accessible second oxidation corresponding to the 1+/2+ redox transition. While this transition is found to be reversible for Os2–4SMe and Os2–4F (and for Os3, as noted previously¹¹), a robust reversibility analysis is precluded for other compounds due to the proximity of these features to the oxidation limit of our electrochemical window.

Figure 4.

Overlaid representative cyclic voltammograms for (a) Os2–X and (b) Os–2Cl, Os1–4Br, Os1–5Br. For clarity, currents for Os2–4F, Os2–4Cl, Os2–4SMe, and Os1–5Br are scaled by a factor of 5, 10, 3, and 2, respectively. All voltammograms are measured in 0.1 M (ⁿBu)₄NPF₆–CH₂Cl₂ at 0.1 V/s apart from Os2–4I which was obtained at 1 V/s. Potentials are reported relative to the [FcH]⁺/FcH redox couple.

The measured redox potentials for all 13 Os(aryl)₄ complexes are collected in Table 2, reported relative to [FcH]⁺/FcH. As might be expected, all the halogenated compounds are significantly more difficult to oxidize than their parent Os1 or Os2 species. The complexes with the highest first oxidation potential, E_1/2^(0/1+), are Os–2Cl and Os1–5Br with E_1/2^(0/1+) = +0.696 and 0.676 V, respectively. Here, the σ-aryl ligands of Os–2Cl comprise an electron-withdrawing ortho-chloride substituent (σ_p(Cl) = 0.23; using Hammett σ_p constants as a proxy for σ_o constants) and no electron-donating methyl substituents (σ_p(Me) = −0.17, σ_m(Me) = −0.07).³² The ligands of Os1–5Br comprise a strongly electron-withdrawing meta-bromide substituent (σ_m(Br) = 0.39) and only a single electron-donating methyl substituent ortho to the Os(IV) center. The E_1/2^(0/1+) values for Os2–4X exhibit a reasonable correlation with the Hammett σ_p constant of each halogen, where σ_p(Br) = σ_p(Cl) (= 0.23) > σ_p(I) (= 0.18) ≫ σ_p(F) (= 0.06). However, Os2–4SMe appears to be a clear outlier based on reported Hammett constants: E_1/2^(0/1+) = +0.018 is the lowest redox potential of any Os(aryl)₄ complex reported to date. With σ_p(SMe) = 0.00, this potential is > 200 mV lower than the expected value of E_1/2^(0/1+) ∼ + 0.244 (the potential measured for Os2).

Table 2. Redox Potentials for Selected Os(aryl)₄ Complexes^a.

entry	compound	E1/2
		1–/0	0/1+	1+/2+	reference
1	Os2–4F	–1.927	+0.318	+1.215	this work
2	Os2–4Cl	–1.797	+0.432	+1.363d	this work
3	Os2–4Br	–1.759b	+0.453b	+1.326	this work
4	Os2–4Ic	–1.707	+0.448	+1.264	this work
5	Os–2Cl	–1.348	+0.696	-	this work
6	Os1–4Br	–1.679	+0.558	+1.49d	this work
7	Os1–5Br	–1.562	+0.676	-	this work
8	Os2–4SMe	–1.955	+0.018	+0.650	this work
9	Os1	–1.961	+0.326	-	(11)
10	Os(2,5-xylyl)4 (Os2)	–2.008	+0.244	-	(11)
11	Os(mesityl)4 (Os3)	–2.028	+0.153	+1.117	(11)
12	Os(2-ethylphenyl)4 (Os–2Et)	–2.026	+0.349	-	(11)
13	Os(2-isopropylphenyl)4 (Os-2iPr)	–2.075	+0.336	-	(11)

^aScan rate = 0.1 V s^–1 (unless stated); in CH₂Cl₂ solutions with 0.1 M (ⁿBu)₄NPF₆ supporting electrolyte; working electrode: glassy carbon; reference electrode, counter electrode: Pt. analyte solutions were between 0.1 and 1 mM. All potentials reported relative to [FcH]⁺/FcH.

^bRedox couples 1–/0 and 0/1+ were previously reported at −1.77 and 0.45 V, respectively.^10,19

^cAt scan rate = 1 V s^–1.

^dE_pa of irreversible redox feature.

Given the apparent good correlations between most E_1/2^(0/1+) and Hammett substituent constants for these complexes, we sought a simple mechanism to quantify this relationship. In the Supporting Information, Table S4, we calculate an adjusted Hammett constant for the different σ-aryl ligands of each compound, using the 2-methylphenyl ligand of Os1 as a reference (= 0). As a worked example, the adjusted constant for Os2–4Br is derived by adding one meta-methyl (σ_m = −0.07) and one para-bromide (σ_p = +0.23) to give +0.16. Similarly, the adjusted constant for Os–2Cl is calculated by subtracting one ortho-methyl (σ_p(Me) = −0.17) and adding one ortho-chloride (σ_p(Cl) = +0.23) to give +0.40. In Figure 5a we plot this adjusted Hammett constant against E_1/2^(0/1+) for all the complexes studied here. We observe an excellent positive linear correlation for 11/13 (85%) of these data points. While a linear fit to all the data provided a R² (goodness-of-fit) value of 0.81 (solid black line), a fit to all data points except 8 (Os2–4SMe) and 11 (Os3) gives R² = 0.99 (red dashed line). While this straightforward analysis may prove useful for quickly estimating the electrochemical properties of as yet uncharacterized complexes, it also serves to help identify potential outliers within our data set and/or the limitations of applying Hammett constants in this context (see below).

Figure 5.

(a) A plot of the adjusted Hammett constant against E_1/2^(0/1+) for selected Os(aryl)₄ complexes. A linear fit to all data points provides R² = 0.81 (black solid line; y = 0.95x – 0.29), whereas a fit to all points except apparent outliers 8 and 11 gives R² = 0.99 (red dashed line; y = 1.06x – 0.34). Data point labels refer to the entries in Table 2. Adjusted Hammett constants are calculated in the Supporting Information, Table S4 using values reported by Hansch et al.³²(b) A plot of the optical gap against the electrochemical gap shows a negative correlation.

We further recognize that the data from our solution voltammetry and UV–vis spectroscopy studies can provide estimates of the energy gap between filled and unoccupied energy levels for each complex.⁴⁶ Accordingly, we calculate the electrochemical gap (E_elec) for each compound from the potential difference between the first oxidation and reduction: E_elec = E_1/2^(0/1+) – E_1/2^(1–/0). We define the optical gap (E_opt) as the longest wavelength at which absorbance reaches ∼10% of that of the peak with greatest intensity in the visible region. We find that E_elec and E_opt can be tuned by 438 mV and 85 nm, respectively; all values are compiled in the Supporting Information, Table S3. In Figure 5b, we plot E_opt against E_elec, which for most compounds exhibits the expected negative correlation. In this analysis, however, data point 5 (Os–2Cl) stands out as a potential outlier, suggesting Os–2Cl exhibits an E_elec that is >200 mV smaller than expected based on the observed E_opt. Given the good agreement between E_1/2^(0/1+) and the adjusted Hammett constant for this compound (Figure 5a), we reason the observed deviation in E_elec is due to a more positive E_1/2^(1–/0) (i.e., Os–2Cl is easier to reduce than predicted). We reiterate in this context that structurally related homoleptic tetrahedral Cr(IV) and square planar Pt(III) complexes comprising chlorinated σ-aryl ligands have been observed to form octahedral geometries with coordinated ortho-chloride atoms upon reduction or oxidation, respectively.^36,38 Accordingly, we tentatively suggest the distinct E_1/2^(1–/0) of Os–2Cl could result from an electrochemically induced structural change in which the ortho-chloride substituents of its σ-aryl ligands weakly interact with the osmium center upon one-electron reduction of the complex to [Os–2Cl]⁻. Such a geometric transition could reasonably be expected to modulate E_1/2^(1–/0), relative to its expected value according to the trend observed in Figure 5b, if the reduced 6-coordinate product was thermodynamically stabilized relative to the analogous 4-coordinate species. To evaluate this hypothesis, the isolation and structural characterization of [Os–2Cl]⁻ or related anionic species such as [Os(C₆Cl₅)₄]⁻ could be targeted in future experimental studies. As detailed below, preliminary calculations that probe the geometry of this complex provide additional support for this proposal.

Though we attempted to identify additional structure–property trends using the experimental data currently available to us, we found that a plot of the adjusted Hammett constant against E_1/2^(1–/0) exhibits only a weak correlation (Supporting Information, Figure S4a). As might be expected from this result, no clear relationships between the adjusted Hammett constant and either E_opt or E_elec could be observed (Supporting Information, Figure S4b,c).

Computational Studies

To help further rationalize these observations, we turned to preliminary first-principles calculations based on density functional theory (DFT; for methods, see the Experimental Section). Here we selected six compounds to study in a model data set: Os1, Os2, Os3, Os2–4Cl, Os–2Cl, and Os2–4SMe. Three of these followed the anticipated correlations shown in Figure 5 and spanned a wide range of E_1/2^(0/1+) (Os1, Os2, Os2–4Cl), the others were outliers in at least one plot. After geometry optimization, we calculated the ionization potential and HOMO–LUMO (Kohn–Sham) gap for each compound. We present, in Figure 6a, a plot of E_1/2^(0/1+) against the ionization potential,^47,48 which exhibits an excellent linear correlation (R² = 0.99). It is notable that data points corresponding to Os3 and Os2–4SMe are now well-integrated into the observed relationship, showing that the deviations they presented in Figure 5a were a result of inaccuracies associated with the application of simple empirical parameters to explain their measured properties. Critically, the trends in E_1/2^(0/1+) observed experimentally can be readily reconciled using more sophisticated (first-principles) analyses. For completeness, in the Supporting Information, Figure S5 we present a plot of the adjusted Hammett constant against the calculated ionization potential for these complexes, which exhibits a similar form to that observed in Figure 5a.

Figure 6.

(a) A plot of E_1/2^(0/1+) against the calculated ionization potential for selected Os(aryl)₄ complexes. A linear fit to all data points provides R² = 0.99 (black solid line; y = 1.37x + 4.58). Data point labels refer to the entries in Table 2. (b) A plot of the optical gap versus the calculated HOMO–LUMO (Kohn–Sham) gap. A linear fit to all data points provides R² = 0.98 (blue solid line; y = −0.01x + 9.74). (c) An orbital energy level diagram for Os2–4SMe, Os1, and Os–2Cl for HOMO–1 to LUMO+2, illustrating how substituents influence the energies of each occupied (solid) and unoccupied (dashed) state. (d) An isosurface plot (isovalue = 0.06 Å^–3) of the DFT-calculated HOMO for Os1 (see Supporting Information, Figure S6 for isosurface plots of the other Frontier orbitals).

We next plot, in Figure 6b, the optical gap versus the calculated HOMO–LUMO gap,⁴⁶ which again reveals a very good linear correlation (R² = 0.98). This result contrasts with the observed, albeit mostly minor, deviations from linearity in the relationship between experimental optical and electrochemical gaps (Figure 5b). We suggest this further highlights the possible influences of solvent or electrolyte, or geometric changes (as proposed above for Os–2Cl), on the properties of the positively or negatively charged complexes and so the exact potentials of their 0/1+ and 1–/0 redox events. We also performed additional calculations for Os–2Cl after removing one electron from the parent tetrahedral complex. In support of our hypothesis that the reduced species may undergo a change in geometry associated with metal coordination by two ortho-chloride groups, as observed for analogous Cr(III)/(IV) compounds (discussed above), we indeed find that the resulting structure of [Os–2Cl]⁻ after geometry optimization exhibits two short and two long Os–Cl distances (Figure S6). This structural change appears unique to [Os–2Cl]⁻; the optimized geometries of Os–2Cl, Os1, and [Os1]⁻ exhibit four similar Os–Cl or Os–C distances (where Os–C is the distance from osmium to the carbon of the tolyl methyl group), suggesting there are insignificant coordination interactions between these groups and the metal center in those compounds (Supporting Information, Table S5).

For additional insights we provide, in Figure 6c, an orbital energy level diagram for Os2–4SMe, Os1, and Os–2Cl that show how the calculated energies of frontier orbitals HOMO–1 to LUMO+2 vary with changing aryl ligand substituents. It can be seen, for example, that the trends in the calculated HOMO energies follow those found for the measured E_1/2^(0/1+) of these complexes. Representative isosurface plots for the HOMO of Os1, and the four other frontier orbitals highlighted here, are provided in Figures 6d and Supporting Information, S7, respectively. Each of these orbitals comprises the expected Os d-aryl π character and appears delocalized across the complex at this level of theory. Notably, for these frozen, distorted tetrahedral geometries we find that the degeneracy of the t_2g and e_g orbital sets expected for an ideal tetrahedral complex–comprising contributions from d_z2 (HOMO) and d_x2-y2, as well as d_xy (LUMO), d_xz, and d_yz orbitals–is lifted. Such analyses also provide a more comprehensive, albeit provisional, understanding of the trends in optical absorption observed for these complexes. For example, the bathochromic shift seen in the UV–vis spectrum of Os2–4SMe relative to Os1 or Os2 (Figure 3) corresponds to a decreased HOMO–LUMO gap that results primarily from an increase in the HOMO energy, while the energy of the LUMO remains largely unchanged. A closer examination of the relevant gas phase orbitals for Os2–4SMe (Supporting Information, Figure S8) reveals that there is a much smaller orbital density on the –SMe substituent in the LUMO than the HOMO. This indicates that the LUMO is delocalized over a smaller volume and suggests that the para-SMe substituents exert less influence over the energetic position of the LUMO through resonance contributions. Additional computational data for all complexes is provided in Supporting Information, Table S6.

Conclusion

In this study we have demonstrated the utility of both halogen/magnesium insertion and exchange strategies to synthesize Os(aryl)₄ complexes with prefunctionalized ligands. While additional work to improve the yields of most compounds prepared by exchange is needed (currently 10–36%), this strategy, in conjunction with traditional magnesium insertion, uniquely provides access to a complete series of halogenated Os2–4X complexes (X = –F, –Cl, –Br, –I, as well as –SMe), in addition to compounds with halogens positioned both ortho or meta to the Os–C bond, for the first time. With these key additions to the growing library of complexes characterized by both UV–vis spectroscopy and solution electrochemistry, we demonstrate that the E_1/2^(0/1+) of most compounds can be correlated with well-known empirical parameters such as Hammett constants using an adjusted metric that accommodates all aryl ligand substituents. We also find that trends in the HOMO–LUMO gaps of this family of complexes are consistent when estimated using different experimental techniques. Critically, computational studies using a model set of complexes show that both their E_1/2^(0/1+) and optical gap may be most accurately predicted using first-principles calculations based on DFT, which also provide deeper insights into their electronic structure.

The notable air- and temperature-stability of these differently functionalized complexes makes them important model precursors to help expand the reaction chemistry of σ-aryl ligands, including transition metal-catalyzed or lithium–halogen exchange strategies. While the utility of Suzuki¹⁰ and Negishi¹² cross-coupling reactions have already been demonstrated using Os2–4Br, the increased rates of reaction afforded by aryl iodides relative to aryl bromides⁴⁹ is expected to facilitate improved yields of desired products starting from Os2–4I. We note that when the halide group on all four σ-aryl ligands must be substituted to form a targeted tetrasubstituted compound, yields above 50% require a yield per aryl halide bond of ≥84%. Following the seminal work by Knochel and others,²⁹ we recognize that the magnesium/halogen exchange reactions exploited here can in principle be applied to prepare Os(aryl)₄ complexes with a range of other (reactive) groups preinstalled on their σ-aryl ligands. Such prefunctionalization approaches, using Grignard or aryl lithium reagents, could also be extended to install substituted σ-aryl ligands at other transition metal sites, including homoleptic σ-aryl complexes with different central atoms,¹ as well as heteroleptic compounds such as Ru(aryl)Cl(CO)(PPh₃)₂.⁵⁰ Beyond their potential reactivity to form other useful derivatives, these Os(aryl)₄ compounds may also be utilized directly. Notably, the thioether-substituted Os2–4SMe, serving as a short, conjugated wire comprising a tetrahedral Os(IV) center, will prove valuable for subsequent molecular conductance studies that seek to evaluate the role of the central atom on through-molecule charge transport.

Experimental Section

Synthetic Methods

All manipulations were carried out in oven-dried glassware under a nitrogen atmosphere using standard Schlenk line techniques. No special precautions were taken to exclude air or moisture during workup unless otherwise stated. Tetrahydrofuran and dichloromethane were sparged with nitrogen and dried using a two-column solvent purification system packed with alumina (Pure Process Technologies, Nashua, NH, USA). Grignard reagents prepared by halogen/magnesium insertion were obtained using methods reported previously,¹¹ and their solution concentration was determined through titration using a salicylaldehyde phenylhydrazone indicator prior to use.⁵¹ Flash chromatography was performed using a Pure C-850 FlashPrep chromatography system and FlashPure EcoFlex flash cartridges (silica, irregular 40–63 μm particle size, 55–75 Å pore size; BUCHI Corporation, New Castle, DE, USA), or manually using Alfa Aesar silica gel 60 (215–400 mesh). (Oct₄N)₂[OsBr₆]¹¹ was prepared using established methods from (H₄N)₂[OsBr₆] (available commercially or synthesized from OsO₄⁵²). All other chemical reagents were commercially available and used without further purification.

¹H and ¹³C{¹H} NMR spectra were recorded at room temperature on Varian VNMRS 500 (500 MHz), VNMRS 600 (600 MHz), or Mercury 400 (400 MHz) NMR spectrometers, unless otherwise stated. ¹H NMR data recorded in CDCl₃ is referenced to residual internal CHCl₃ (δ 7.26) solvent signal.⁵³¹³C{¹H} NMR data recorded in CDCl₃ is referenced to internal CDCl₃ (δ 77.16).⁵³¹H and ¹³C{¹H} resonances were assigned where possible using 2D correlation spectroscopy experiments. UV–vis spectra were obtained using a Cary 60 UV–vis Spectrophotometer (Agilent Technologies, Inc., Santa Clara, CA, United States) or an AvaSpec-ULS2048-EVO UV/vis Spectrometer integrated with an AvaLight UV/vis/NIR Light Source (Pine Research Instrumentation, Durham, NC, USA; Avantes North America, Lafayette, CO, USA). Mass spectrometry analyses were performed on a Waters GCT Premier (EI), or Bruker Autoflex Speed LRF (MALDI) at the Mass Spectrometry Lab, University of Illinois Urbana–Champaign. Microanalyses were carried out using a Thermo Flash 2000 CHNS Combustion Analyzer at USC, or on a Control Equipment Corp. CEC 440HA Elemental Analyzer at the Marine Science Institute, University of California Santa Barbara.

Single-Crystal X-ray Diffraction

Crystals suitable for X-ray diffraction were grown by slow evaporation of n-hexane/CH₂Cl₂ or acetonitrile solutions, or by cooling a solution of analyte in n-hexane to −20 °C.

For Os2–4F, Os2–4Cl, Os2–4I, Os1–5Br: X-ray intensity data were collected at 100 K on a Bruker APEX DUO 3-circle platform diffractometer, equipped with an APEX II CCD detector, using Cu–K_α (λ = 1.54184 Å, multilayer optics monochromator) from a IuS microsource, or Mo–Kα radiation (λ = 0.71073 Å, TRIUMPH curved-crystal monochromator) from a fine-focus tube. The structures were solved by intrinsic phasing and refined on F² using the Bruker APEX3 Software Package and ShelXle.⁵⁴⁻⁵⁷

For Os–2Cl: Single crystal X-ray diffraction was performed at 100 K on a Rigaku-Oxford Diffraction XtaLAB Synergy-S diffractometer, equipped with a HyPix-6000HE CCD detector, using Cu–K_α (λ = 1.54184 Å) radiation from a microfocus sealed X-ray tube. The data set was recorded as ω-scans at 0.5° step width and integrated with the CrysAlis software package (Oxford Diffraction Ltd.: Abingdon, England, 2006). The structures were solved with ShelX⁵⁸ using the graphical interface provided by Olex2.⁵⁹ The final model was refined with ShelXL using full matrix least-squares minimization on F².⁵⁶

Further crystallographic details can be obtained from the Cambridge Crystallographic Data Centre (CCDC), registry numbers 2412630–2412634. Data collection and structure refinement parameters for each crystal structure are provided in the Supporting Information (X-ray Crystallography section).

Electrochemical Methods

Electrochemical measurements were performed under an argon atmosphere using a CHI760E bipotentiostat (CH Instruments, Austin, TX, USA) with argon or nitrogen-sparged 0.1 M tetrabutylammonium hexafluorophosphate ((ⁿBu)₄NPF₆) CH₂Cl₂ solutions (using nonanhydrous solvents, scan rate = 0.1 V s^–1). Plotted voltammograms are not corrected for iR_s unless otherwise stated. Studies employed glassy carbon disc working electrodes (Ø = 3 mm, CH Instruments), mechanically polished using an alumina slurry prior to use. Pt wire reference and counter electrodes were cleaned by annealing in an oxyhydrogen flame. Analyte solutions were between 0.1 and 1 mM. Potentials are reported relative to [Cp₂Fe]⁺/[Cp₂Fe], measured against an internal Cp*₂Fe reference (−0.532 mV vs [Cp₂Fe]⁺/[Cp₂Fe]).

Safety Considerations

Caution! Grignard reagents are pyrophoric. These compounds must be handled using proper needle and syringe techniques. All manipulations of these reagents were performed using standard air-free techniques.

Computational Methods

Density functional theory (DFT) calculations were performed using the Q-Chem 5.4.2 program, using the molecular editor and visualization package IQMol 2.15.1.⁶⁰ Input structures of Os(aryl)₄ complexes used, or were constructed from, experimental geometries determined through single-crystal X-ray diffraction (Table S1). The geometries of these input structures were subsequently optimized at the B3LYP level of theory, using in each case a 6–31G** basis for light atoms and LACVP for osmium. All calculations were carried out using Grimme’s DFT-D3 dispersion correction with Becke-Johnson damping,⁶¹ in a CH₂Cl₂ medium through application of a conductor-like polarizable continuum model (PCM). The PCM used atomic radii from the universal force field (UFF) to define the solute cavity, a relative permittivity (ε) of 8.9, and an optical dielectric (ε_opt) of 2.0.⁶²

Geometry optimizations were considered converged when the gradient, and either the energy or atomic displacement, satisfied the convergence criteria (energy = 10 × 10^–8, gradient = 10 × 10^–6, atomic displacement = 1200 × 10^–6; all values in atomic units [a.u.]). Calculations used the direct inversion in the iterative subspace (DIIS) self-consistent field (SCF) optimization algorithm, using an on-the-fly (automated) superposition of atomic densities (AUTOSAD) initial guess. The SCF cycle was considered converged when the wave function error between consecutive SCF cycles was less than 10^–8 a.u.

Vertical ionization potentials were calculated for each complex by subtracting the total energy of the geometry optimized, neutral complex from the total energy of the 1+ cation, fixed in the geometry of the neutral complex.⁶³ HOMO–LUMO (Kohn–Sham) gaps were calculated by subtracting the energy of the HOMO from that of the LUMO. Data for selected complexes in provided in Supporting Information, Table S6.

General Syntheses of Os(aryl)₄ Complexes from (Oct₄N)₂[OsBr₆]

Magnesium Insertion

The selected aryl Grignard reagent in THF was added dropwise to a stirred suspension of (Oct₄N)₂[OsBr₆] in THF. After stirring at room temperature for 2 h, nitrogen-sparged methanol (1 mL) was added to quench the reaction. The solvent was removed under vacuum, whereby the resulting material was exposed to air overnight to allow any residual solvent to evaporate. The crude product was dissolved in CH₂Cl₂, filtered, preabsorbed onto Celite, then purified by chromatography on a hexanes-packed column (SiO₂; hexanes/CH₂Cl₂, 1:0→4:1). To improve the resolution of eluting products we typically performed an initial (manual) chromatographic separation to remove extraneous material (collecting only the eluting black band(s)), followed by a second (automated) column focused on product isolation. If necessary, complexes could be further purified by recrystallization from hot acetonitrile.

Magnesium–Halogen Exchange

A stirred solution of the selected aryl-iodide in THF was cooled to 0 °C (ice–water bath), whereby iPrMgCl•LiCl (“turbo Grignard”) in THF was added dropwise. After ∼30 min, a solution of (Oct₄N)₂[OsBr₆] in THF was added dropwise to the reaction mixture. The ice bath was removed, and the solution was stirred for 2 h at room temperature. Nitrogen-sparged methanol (1 mL) was added to quench any unreacted Grignard reagent. Following solvent removal under vacuum, the resulting material was exposed to air overnight to allow any residual solvent to evaporate. The crude product was purified as described above for syntheses via magnesium insertion.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acs.inorgchem.4c05589.

• Electronic Supporting Information (ESI) available: Additional experimental details and synthetic methods, crystallographic, solution electrochemical, UV–vis, and computational data, ¹H and ¹³C{¹H} NMR spectra for all new compounds (PDF)

The authors declare no competing financial interest.