Acyl chloride carbon insertion into dicarbaborane cages – new route to tricarbollide cages

Introduction

Tricarbollides (eleven-vertex nido tricarbaboranes) [1–3] have been discovered in 1995 by our group in collaboration with F. Teixidor’s team. The method for their synthesis consists in the insertion of the cyanide or isocyanide carbon into the structure of the [nido-5,6-C₂B₈H₁₁]¯ anion [4–9]:

The carbon-insertion processes, in which the amino group has been retained on the cluster carbon, can be classified as skeletal aminocarbonation, which is typical for nido clusters [10, 11]. The reason for RNH-retention is that the lone extra hydrogen atom in [nido-5,6-C₂B₈H₁₁]¯ is incapable of interrupting completely the C≡N or N≡C bond in reactions (1) and (2) to release NH₃ or ^tBuNH₂, respectively, under CH incorporation into the cluster. Further chemical transformations of the amino derivatives generated in turn a long series of tricarbollide compounds, inclusive of the parent nido compounds [7,8,9-C₃B₈H₁₁]¯, 7,8,9-C₃B₈H₁₂, and [7,8,10-C₃B₈H₁₁]¯ and their derivatives [1–8].

The works by Brellochs at al. [12–15] on degradative insertion of the aldehyde carbon into the structure of the [6-HO-arachno-B₁₀H₁₃]^2- dianion and some metal-carbonyl C-insertion reactions [16–18] suggested that some specific C≡O – carbon incorporation procedures might be, in principle, applicable to other cluster systems. This viable synthetic approaches have been recently followed by our group, which led to developing a new carbon-incorporation strategy based on interaction between acyl chlorides (RCOCl) and the arachno-6,9-C₂B₈H₁₄ dicarbaborane in the presence of amine bases [19, 20]. These so called skeletal alkylcarbonation (SAC) reactions resulted in unprecedent incorporation of the CO–carbon into the dicarbaborane cage and formation of the symmetrical 8-substituted 7,8,9- tricarbollides. Herein we would like to feature some general aspects of this chemistry along with the first results on Fe – complexation reactions involving arylsubstituted ferratricarbollides.

Acyl chloride carbon-incorporation reactions

Treatment of the arachno-6,9-C₂B₈H₁₄ (1) dicarbaborane [21–23] with excess RCOCls in the presence of two equivalents of amine base (PS or Et₃N) (proton scavenger) in refluxing CH₂Cl₂ for 24 h, followed by adding concentrated H₂SO₄ at 0 °C (Scheme 1), led to the isolation of a series of neutral tricarbollides 8-R-nido-7,8,9-C₃B₈H₁₁ (2) [19, 20]. The reaction seems to be of general character, as tested for RCOCls of varying R (alkyls and aryls) with yields of 2 ranging from 60–70 % and 80–95 % for alkyls and aryls, respectively. The reactions are restricted only by the availability of the starting acyl chloride.

Scheme 1

Tricarbollide compounds from the SAC cross-coupling via incorporation of the RC unit into the structure of arachno-6,9-C₂B₈H₁₄ (1).

The acyl chloride reactions are in accord with a simple stoichiometry of Eqs. (3 and 4) comprising the deprotonation of 1 along with the elimination of HCl and H₂O:

Scheme 1 also suggests that these reactions are consistent with regiospecific insertion of the 3-electron carbyne RC≡ unit into the open-face section of the [arachno-6,9-C₂B₈H₁₃]¯ anion (1¯) identified by B(5), C(6), C(9), and B(10) vertices to form the [8-R-nido-7,8,9-C₃B₈H₁₀]¯ (2¯) anion. This must proceed under extraction of all the three extra hydrogen atoms which are then eliminated as H₂O and HCl. It is also readily seen that the reactions result in an effective cross-coupling between R and the carborane cage of 1¯ which is then leaving the coupling process enriched in one more carbon vertex. From the viewpoint of cluster chemistry, the whole RC-insertion process of Eq. 4 can be envisaged as an arachno → nido oxidation.

The reaction was proposed [11, 19, 20] to start with endo-acylation of 1¯ at B(5) to form the neutral acyl derivative endo-5-(RCO)-arachno-6,9-C₂B₈H₁₃ (1a) (see Scheme 2) which then extracts the endo hydrogen atom from the cage C(6)H₂ group to form the aldolization product 1b. Entailing H₂O elimination, followed by C-insertion, would then lead directly to the formation of the neutral 2.

Scheme 2

Proposed aldolization mechanism for the insertion of the acyl chloride RC unit between the C(6) and C(9) positions in [arachno-6,9-C₂B₈H₁₃]¯ (1¯)

All the 8-R-nido-7,8,9-C₃B₈H₁₁ (2) compounds can be deprotonated (Scheme 1), for example, by dissolution in 10 % aqueous NaOH, followed by precipitation of the solution with a suitable counter-cation (e.g., Tl⁺, NEt₄⁺, etc.), to generate very stable anions of general constitution [8-R-nido-7,8,9-C₃B₈H₁₀]¯ (2¯). However, the simplest way for ready preparation of these anions is based on treatment of compounds 2 with NEt₃ in dichloromethane, which leads to quantitative formation of the corresponding Et₃NH⁺ salts on evaporation of the solvent. These salts represent a very stable storage form of tricarbollides that can be converted easily into any other salt on dissolution in aqueous NaOH, followed by precipitation with a suitable counter-cation.

The structures of the anions [8-R-nido-7,8,9-C₃B₈H₁₀]¯ (2¯) (for R=C₆H₅, 4-I-C₆H_4, and 1-C₁₀H₇) and of the neutral 8-(1-C₁₀H₇)-nido-7,8,9-C₃B₈H₁₁ compound have been established unambiguously by X-ray diffraction analyses [19, 20] together with that of the parent [nido-7,8,9-C₃B₈H₁₁]¯ [24]. The data confirm the overall C_s-symmetry and 8-substitution on the central carbon vertex of the pentagonal open face. The cluster C–C distances in the substituted anions approximate 1.530 Å and are slightly longer than the equivalent bond vector in the parent anion (mean 1.515 Å), and the C(7)–C(8)–C(9) angles approximate the ideal pentagonal angle.

Multinuclear ¹¹B, ¹H, and ¹³C NMR measurements on compounds of structures 2 and 2¯ also clearly confirm the 8-substitution. Because of the same C_s-symmetry, the NMR characteristics resemble those reported earlier for the parent nido tricarbollides [7,8,9-C₃B₈H₁₁]¯ and 7,8,9-C₃B₈H₁₂ [4–6]. As shown in Fig. 1, the ¹¹B NMR spectra of compounds of structures 2 and 2¯ exhibit striking similarities, indicating that electronic effects exerted by individual substituents do not affect too much the chemical shifs at B-positions of the cluster. The spectra of the neutral compounds 2 entirely differ from those of the corresponding 2¯ counterparts due to the presence/absence of the bridging hydrogen. As exemplified by the spectra of the Ph anions in Fig. 2, those of their Et₃NH⁺ and Tl⁺ salts rather differ. The latter salts exhibit a downfield shift (∼5 ppm) of the B(10,11) and B(1) resonances, which points to at least partial coordination of the Tl⁺ ion to the open pentagonal face in structure 2¯. The Tl centre would then appear in a position antipodal to the B(1) vertex – this type of coordination reminds that observed in the structurally similar compound Tl₂[nido-C₂B₉H₁₁] [25]. Similar effect was also observed for Tl⁺ salts of 7-aminosubstituted derivatives of the [7,8,9-C₃B₈H₁₁]¯ anion [4–8]. Figure 3 shows that, the ¹H NMR spectra of compounds 2 and 2¯ consist of a singlet of relative intensity two attributed to the cage CH(7,9) unit; this position being remarkably shielded (∼1.5 ppm) for anions 2¯ when compared to their conjugated acids 2 that also exhibit a typical high-field resonance (∼–2 ppm) due to the bridging hydrogen. Figure 3 also demonstrates that conversion into Tl⁺ salts is associated with a marked downfield shift of the cage CH(7,9) signals.

Fig. 1

Inter-comparison of δ(¹¹B) NMR shifts for selected anions [8-R-nido-7,8,9-C₃B₈H₁₀]¯ (2¯) and the corresponding neutral compounds 8-R-nido-7,8,9-C₃B₈H₁₁ (2) showing dramatic NMR changes caused by the absence or presence of one bridging hydrogen in the open pentagonal face. Assignments by [¹¹B-¹¹B]-COSY NMR measurements.

Fig. 2

Inter-comparison between δ(¹¹B) NMR shifts for the [8-Ph-nido-7,8,9-C₃B₈H₁₀]¯ (2¯) anions. Patterns for the Et₃NH⁺ salt (lower traces) markedly differ from those for the Tl⁺ salt (upper traces) due to the Tl-C₃B₂ coordination. Assignments by [¹¹B-¹¹B]-COSY NMR measurements.

Fig. 3

Inter-comparison between δ(¹H) (dotted lines) and δ(¹³C) (full lines) NMR shifts of cage CH or CPh units for the neutral 8-Ph-nido-7,8,9-C₃B₈H₁₁ and its Et₃NH⁺ and Tl⁺ salts.

[The ¹³C{¹H} NMR spectra of the 2 and 2¯ pair (see the ¹³C NMR scale in Fig. 3 for R=Ph) exhibit expectedly two broad (¹³C-¹¹B coupling) 1:2 singlets due to the cage C(8) and C(7,9) resonances. Deprotonation of 2 brings about remarkable shielding of the C(8) (∼28 ppm) and C(7,9) positions (∼9 ppm), which may be associated with partial localization of the negative charge on the open C₃B₂ face. The ¹³C{¹H} NMR spectra of the corresponding Tl⁺ salts exhibit a slight deshielding effect in comparison with their Et₃NH⁺ counterparts.

Iron-complexation reactions

The cluster constitution of anions 2¯ suggests straightforward analogy with the Cp anion (pentagonal open face and monoanionic character), which constitutes an excellent setting for Fe-complexation. Scheme 3 shows that high-temperature reactions between 8-R-nido-7,8,9-C₃B₈H₁₁ (2) (where R=Ph, 1-C₁₀H₇, and 2-C₁₀H₇) and [CpFe(CO)₂]₂ in mineral oil at ∼200 °C [26], followed by chromatographic separation of the metallatricarbollide fraction in a hexane/CH₂Cl₂ mixture led, indeed, to the isolation of an orange band containing a mixture of structurally different ferratricarbollide complexes which were identified as those of general closo constitution [1-Cp-10-R-1,2,4,10-FeC₃B₈H₁₀] (3), [1-Cp-12-R-1,2,4,12-FeC₃B₈H₁₀] (4), and [1-Cp-2-R-1,2,4,10-FeC₃B₈H₁₀] (5). For R=Ph, all the three types of complexes (3–5) were formed, while, for R=1-C₁₀H₇ and 2-C₁₀H₇, only the symmetrical complexes of types 3 and 4 have been formed, evidently due to sterical reasons (repulsion between the bulky R and the Cp ring might be expected). All the complexes were finally separated by semi-preparative HPLC chromatography in 100 % hexane and isolated in total yields ranging 80 %; for example the three Ph isomers were obtained in 15, 35, and 30 % yields as orange crystalline substances. Comparison of structure 2 with those of complexes 3–5 in Scheme 3 shows that the complexation process is associated with extensive rearrangement of cage C-vertices into mutually meta positions (for a similar effect see references [27–35]). All the complexes 3–5 are extremely stable orange solids that can be crystallized by slow evaporation of their hexane solutions. The structures of three complexes of different substituents, namely [1-Cp-10-(2-C₁₀H₇)-1,2,4,10-FeC₃B₈H₁₀] (type 3) [1-Cp-12-(1-C₁₀H₇)-1,2,4,12-FeC₃B₈H₁₀] (type 4), and [1-Cp-2-Ph-1,2,4,10-FeC₃B₈H₁₀] (type 5) have been established by X-ray diffraction analyses [26]. The 3 and 5 structures are characterized by 2,4,10-configuration of the cluster C-vertices, but differ in the positioning of the aryl substituent. In contrast, the complex of constitution 4 is a cluster isomer containing cage C-atoms in 2,4,12-positions with the aryl residing in para position with respect to the Fe-centre.

Scheme 3

Insertion of the CpFe fragment into the structure of 8-R-nido-7,8,9-C₃B₈H₁₁ (2) (where R=Ph, 1-C₁₀H₇, and 2-C₁₀H₇) giving three isomeric ferratricarbollide structures, [1-Cp-10-R-1,2,4,10-FeC₃B₈H₁₀] (3), [1-Cp-12-R-1,2,4,12-FeC₃B₈H₁₀] (4), and [1-Cp-2-R-1,2,4,10-FeC₃B₈H₁₀] (5).

Inspection of NMR data shows marked similarities among compounds of individual types 3–5, which indicates that electronic effects contributed by the aryl substituents to the cage are very similar. As shown in Fig. 4 for R=Ph, the ¹¹B NMR spectra of complexes 3 and 4 exhibit 2: 1: 2: 1: 2 and 2: 1: 1: 2: 2 patterns of doublets, respectively, while the spectrum of type 5 displays eight different ¹¹B resonances due to asymmetrical disposition of the substituent.

Fig. 4

Correlation of the δ(¹¹B) NMR shifts for all the three [1-Cp-10-R-1,2,4,10-FeC₃B₈H₁₀] (3), [1-Cp-12-R-1,2,4,12-FeC₃B₈H₁₀] (4), and [1-Cp-2-R-1,2,4,10-FeC₃B₈H₁₀] (5) isomers. Exemplified for R=Ph, naphthyl-substituted analogues display very similar patterns.

Apart from typical low-field ¹H and ¹³C resonances due to aromatic and cyclopentadienyl CH groups in the NMR spectra of the isolated compounds, the most interesting is the area of resonances of the cage carbon vertices shown in Fig. 5. The ¹³C NMR spectra of C_s-symmetry complexes of types 3 and 4 contain 2:1 signals assigned to equivalent C(2,4) vertices and to the substituted C(10) and C(12) atoms, respectively. The corresponding spectrum of the asymmetrical 5 expectedly contains three resonances, of which that of the substituted C(2) shows strong deshielding (∼25 ppm) with respect to C(2,4) in structure 3. A similar deshielding effect (∼30 ppm) is observed for the substituted C(10) vertex in 3 when compared to C(10) in the structurally related 5. The corresponding ¹H NMR spectra of all complexes isolated contain cage CH signals within the range of 2.11–2.35 ppm.

Fig. 5

Correlation of the δ(¹H) and δ(¹³C) NMR shifts for all the three 1-Cp-10-R-1,2,4,10-FeC₃B₈H₁₀] (3), [1-Cp-12-R-1,2,4,12-FeC₃B₈H₁₀] (4), and [1-Cp-2-R-1,2,4,10-FeC₃B₈H₁₀] (5) isomers. Exemplified for R=Ph, naphthyl-substituted analogues display very similar patterns.

Conclusions

The novel acyl chloride (SAC) strategy for carbon insertion into the dicarbaborane cage is highly effective for the preparation of tricarbollide compounds bearing alkyl and aryl functionalities located symmetrically on the open face of the molecule. The method brings new synthetic features into the so far restricted area of eleven-vertex tricarbaboranes [1–9], being extremely flexible and allowing for variations either in molecular shape or substituent design. These synthetic tools may be exploited in cluster engineering and B-cage based biochemistry, for example, by varying substituents on carbon and boron positions in the starting carborane 1 and R in the RCOCl reagent; other tricarbaborane molecular shapes are also expected from thermal rearrangement reactions. It is also to be expected that the SAC strategy can be also applied to other-than-dicarbaborane cages, which may result in substantial synthetic extensions.

Complexes of types 3–5 represent the first examples of monoaryl substituted ferratricarbollides and there is no doubt that a broad area of the twelve-vertex aryl or alkyl substituted metallacarborane chemistry has been opened that parallels the work in the fields of eleven- and ten-vertex metallatricarbaborane chemistry [36–42]. The compounds reported in this study are cluster-boron analogues of ferrocene [43] and their extremely high stability predestines them for biological uses and as starting substrates for further metallacarborane syntheses, e. g., polyhedral expansion and contraction reactions [44]. Relevant experiments are currently underway in our laboratories. In progress are also the syntheses of alkyl substituted compounds [19, 20] and homoleptic (bis-tricarbollide)Fe(II)-complexes along with experiments aimed at the isolation of complexes with other transition metals.