1,1′-Dibromo-2,2′,5,5′-tetrakis(methylthio)ferrocene

Abstract

The reaction of 1,1′-dibromoferrocene with three equivalents of LiTMP and Me₂S₂ produces a product mixture, from which 1,1′-dibromo-2,2′-bis(methylthio)ferrocene (4) can be isolated in moderate yield. Further treatment of the isolated compound 4 with two equivalents of LiTMP and Me₂S₂ also produced a mixture, from which the title compound 7 could be isolated. A crystal structure determination of 7 shows that the two cyclopentadienyl rings are oriented in such a way that a maximum number of S…Br contacts results. Intermolecular S…Br contacts, supported by S…S contacts, lead to a 1D-chain polymer in [1 0 0] direction.

1. Introduction

Ferrocenyl thioethers have been known for almost 65 years [1]. The first ferrocenediyl bisthioether [Fe(C₅H₄SMe)₂] was reported in 1959 [2]. While the first preparations were performed by reduction of the corresponding sulfonyl chlorides [2,3], an alternative synthetic pathway was established, which used the dilithiation of ferrocene followed by electrophilic quench with dimethyl disulfide [4]. As it was recognized that such a bis-thioether could act as a ligand to many transition metal fragments and that these dinuclear complexes showed catalytic activity [4], the syntheses of ferrocenediyl tris- and tetra-thioethers and their corresponding complexes was reported only a couple of years later [5,6]. Again, the synthetic protocol used lithiation followed by electrophilic quench with dimethyl disulfide. Besides the direct use of these thioethers as ligands, some reactivity studies were undertaken with the purpose of creating mixed thioether-amine and -phosphine ligands [3,7]. While for the former the ferrocenediyl bisthioether was treated with bisdimethylaminomethane in acetic acid, for the latter first an oxidation to the bis-disulfoxide was performed followed by lithiation and electrophilic quench with PR₂Cl. Most of these studies showed that it was extremely difficult to steer the reaction only in one direction. Usually tri- and tetra-substituted products [Fe{C₅H₃X(SR)}{C₅H₄SR}] and [Fe{C₅H₃X(SR)}₂] were formed together, sometimes also with both 1,2- and 1,3-regioisomers formed [3,5,6]. Butler reported in 1999 the ortho-lithiation of 1,1′-dibromoferrocene using 1.0 eq. LDA in THF followed by electrophilic quench with dimethyl disulfide to give tri-substituted [Fe{C₅H₃Br(SMe)}(C₅H₄Br)]. In this paper the formation of “some disubstituted product” was mentioned with no further details given (the announced “full paper on this work” apparently never got published) [8]. In a related paper on poly(diphenylphosphino)ferrocenes, the reaction of 1,1′-dibromoferrocene with increasing amounts of LDA followed by reaction with PPh₂Cl gave mixtures of tri- and di-isomeric tetra-substituted products: [Fe{C₅H₃Br(PPh₂)}{C₅H₄Br}], [Fe{C₅H₂Br(PPh₂)₂}(C₅H₄Br)] and [Fe{C₅H₃Br(PPh₂)}₂], the latter exclusively as the meso-isomer [9]. Some other authors reported that careful control of not only reaction stoichiometry, but also the absolute concentrations was necessary for the lithiation of 1,1′-dibromoferrocene [10]. Previously, we reported the dilithiation of 1,1′-dichloroferrocene with n-BuLi/TMEDA, followed by electrophilic quench with dimethyl disulfide to give the tetra-substituted [Fe{C₅H₃Cl(SMe)}₂] as a mixture of meso- and DL-diastereoisomers [11]. Further lithiation with butyl lithium and quench with dimethyl disulfide gave a mixture of the penta- and hexa-substituted compounds [Fe[C₅H₂Cl(SMe)₂}[C₅H₃Cl(SMe)}] and [Fe[C₅H₂Cl(SMe)₂}₂]. We also had found that a sequence of deprotonations with lithium tetramethylpiperidinide (LiTMP) followed by quench with dimethyl disulfide starting from monobromoferrocene gave 1-bromo-2,3,4,5-tetrakis(methylthio)ferrocene [12], which on reaction with butyl lithium and MeSSMe gave 1,2,3,4,5-pentakis(methylthio)ferrocene. We wondered if a similar reaction sequence starting from 1,1′-dibromoferrocene would allow access to the elusive decakis(methylthio)ferrocene. Here we report on our studies on the synthesis of 1,1′-dibromo-2,2′,5,5′-tetrakis(methylthio)ferrocene.

2. Results

2.1. Synthesis and Reactivity

The synthesis of the title compound was performed in three steps. In the first step 1,1′-dibromoferrocene was transformed to 1,1′ dibromo-2,2′-bis(methylthio)ferrocene via double deprotonation with lithium tetramethylpiperidinediide (LiTMP) and electrophilic quench with dimethyldisulfide (Me₂S₂). (Scheme 1).

As might have been expected from the reaction stoichiometry as well as from literature precedent [5,13] the reaction yielded complex mixtures, the separation of which turned out to be very difficult, and yields of compounds with purity >95% were very low. The formation of 1,1′-dibromo-2′-methylthioferrocene that had been reported as the main product from the reaction of 1 with 1.0 eqiv. LDA and Me₂S₂ [8] was not observed, although it cannot be fully excluded due to the very similar NMR shifts of other compounds formed in the reaction. The isomeric compound 5 resembles the 1,1′-dibromo-2,5-bis(diphenylphosphino)ferrocene reported as a side product from the reaction of 1 with >1.0 equiv. LDA and PPh₂Cl [9]. Products 2 and 3 might also have been expected, as 2 and 1,2-dibromoferrocene were reported as products of the lithiation of 1 with LiTMP and electrophilic bromination [14]. Compound 4 exists as a diastereomeric (R_pR_p/S_pS_p) (”DL”) and (R_p/S_p) (“meso”) pair. Its dichloro counterpart was reported to be formed as a 1:2 mixture of the meso and DL forms [11]. In the present reaction, the DL isomer is formed exclusively or at least in large excess, as could be proved by an X-ray structure determination (see Supplementary Information). Although there are rather many weak signals in the NMR spectra, no complete set of signals attributable to the meso isomer was observed. A recent publication by Butler suggests that “the product ratios obtained depend on the quenching reagent effectiveness, the crystallization process, and the ratio of the lithium reagents themselves” [15].

In the next step compound 4 (purity ca. 90%) was treated with ca. 2.0 equivalents LiTMP and Me₂S₂. Again, a rather complex product mixture was obtained, in which compounds 3, 4, 6 and 7 could be identified (Scheme 2).

While the desired title compound 7 was already formed in this reaction and could be isolated (in a rather impure form), its yield could be improved by the separate reaction of isolated compound 6 with 1.5 equivalents LiTMP and quench with Me₂S₂.

For a short reactivity study, compound 7 was treated with 3 equivalents LiTMP and Me₂S₂. Mass spectroscopic analysis of the obtained product showed an extremely complex mixture of compounds of the formula types [C₁₀H_xBr_y(SMe)_zFe] with 0 ≤ y ≤ 3 and 2 ≤ z ≤ 7. It was not possible to separate these mixtures sufficiently to allow for the assignment of NMR signals, except for [Fe{C₅HBr(SMe)₃}₂] (8), (Scheme 3). Repeating this reaction with different stoichiometries led only to even more complicated mixtures with no clear indication for the formation of the elusive deca(methylthio)ferrocene.

2.2. Molecular and Crystal Structure of 7

Compound 7 crystallizes in the monoclinic space group P2₁/c with one molecule in the asymmetric unit. Figure 1 shows an ORTEP3 plot of 7. For comparison, Figure S14 shows the molecular structure of (R_pR_p/S_pS_p)-4, which crystallizes in the monoclinic space group C2/c with half a molecule in the asymmetric unit.

There is a (non-crystallographic) twofold rotation axis through the iron atom, parallel to the C3–H3 (or the C9–H9) vector, but no mirror plane. The cyclopentadienyl rings are eclipsed (torsion angle C3-CT1-CT2-C9 1.73°, with CT1 and CT2 being the centroids of both cyclopentadienyl rings) and parallel (interplanar angle 4.32(10)°). Two methylthio groups have the methyl group in an equatorial position, while the other two have axial methyl groups, directed away from the iron atom. The bromine atoms and the two “equatorial” sulfur atoms are shifted slightly (ca. 0.10 Å) to the distal side of the corresponding Cp rings while the two “axial” sulfur atoms are shifted by the same amount towards the Fe atom. This leads to rather short Fe…S contacts, of ca. 3.315 Å. The two Br…S distances of the substituents on the same ring are slightly shorter (3.45–3.59 Å) than the Br…S distances between the substituents on different rings (ca. 3.70 Å).

Table 1. Important intramolecular distances [Å] and angles [°] of compound 7.

Fe1–Ccp1	2.021(2) – 2.087(2)	Fe1–Ccp2	2.023(2)– 2.084(2)
Fe1–CT1	1.647(1)	Fe1–CT2	1.645(1)
Fe1…S2	3.3176(6)	Fe1…S10	3.3125(6)
C1–Br1	1.875(2)	C6–Br2	1.874
C2–S2	1.754(2)	C7–S7	1.751(2)
C5–S5	1.746(2)	C10–S10	1.753(2)
Br1…CP1	0.089(1)	Br2…CP2	0.114(1)
S2…CP1	−0.078(1)	S10…CP2	−0.089(1)
S5…CP1	0.118(1)	C7…CP2	0.116(1)
C2-S2-C21	100.4(1)	C10-S10-C101	101.0(1)
C5-S5-C51	100.8(1)	C7-S7-C71	101.1(1)
CT1-Fe1_CT2	177.35(4)
C1-C2-S2-C21	90.7(2)	C6-C10-S10-C101	94.3(2)
C4-C5-S5-C51	−16.4(2)	C8-C7-S7-C71	−12.9(2)

CT1 is the centroid of ring Cp1, CT2 of ring Cp2. C_cp1 are the carbon atoms of ring Cp1, C_cp2 of ring Cp2. CP1 is the best plane through the carbon atoms C_cp1, CP2 through carbon atoms C_cp2.

collects important geometrical parameters of 7. The corresponding parameters of compound 4 can be found in Table S2.

Besides these intramolecular Br…S interactions, there are also many such intermolecular interactions: Br1…S7′ (1−x, 1−y, 1−z) and Br2…S5′ (−x, 1−y, 1−z) are only 3.375(1) and 3.357(1) Å apart, well under the sum of their van-der-Waals radii (3.65 Å). These contacts between neighboring inversion-related molecules lead to an infinite 1D-chain with the base vector [1 0 0]. Figure 2 shows a packing plot of compound 7, which displays the occurrence of S…S contacts as well (S2…S10′ (1 + x, y, z) are only 3.405(1) Å apart), which support this 1D-chain. There are no short Br…Br contacts and also no π stacking occurs. However, there are some C–H…π contacts: atom H9 is only 2.74 Å away from the centroid CT1′ of a neighbor molecule (x, ½ − y, ½ + z).

For comparison, the structure of compound 4 (see also Figure S14) was also examined for intra- and intermolecular interactions. Figure 3 shows packing plots viewed both along a and along b. The intramolecular Fe…S contacts are with 3.378(1) Å, slightly longer than in 7. The Br…S distances between neighboring substituents on the same Cp ring and between those of the two rings of the same molecule are almost identical (3.573(1) Å vs. 3.592(1) Å) and longer than intermolecular S…Br contacts (3.463(1) Å). The intermolecular S…Br contacts join the molecules in the c direction. There are no intermolecular Br…Br or S…S contacts. However, there are intermolecular C–H…S contacts of 2.93 Å (ca 0.07 Å shorter than the sum of van-der-Waals-radii), which join the molecules in the b direction.

Another way of looking at intermolecular interactions is Hirshfeld analysis. For this purpose, we used the program CrystalExplorer [16]. Figure S15 shows the Hirshfeld surface of 7, Figure S16 a “Fingerprint Plot” and Figure S17 the HOMO and LUMO of compound 7, as calculated by some of the subroutines of this program. As a result of the “Fingerprint” analysis the relative importance of the individual atom interactions can be quantitatively estimated. This shows that Br…Br, S…S and Br…S interactions are relatively unimportant (contributions of 1.6, 1.4 and 5.2%) The major contributions come from H…X interactions (H…H 44.0, H…Br 17.3, H…S 18.9 and H…C 11.3%).

3. Discussion

The reaction of 1,1′-dibromoferrocene with 3 equivalents of LiTMP followed by electrophilic quench with Me₂S₂ leads to a mixture of several compounds, which are difficult to separate, and complete purification of the desired 1,1′-dibromo-2,2′-bis(dimethylthio)ferrocene succeeds only with substantial decrease in yield. It was therefore decided to continue the study with products of only 90% purity. Therefore, it cannot totally be excluded that the fact that the subsequent reaction steps produced even more complex product mixtures is a consequence of the reduced purity of the starting materials. However, the nature of the obtained side products, particularly the formation of compounds with higher and lower bromine content, suggests that extensive “halogen dance” reactions occur, and it appears also very likely that a “methylthio dance” is also taking place. Both phenomena had previously been observed by us when studying analogous reactions in the pentachloroferrocene system [17]. It seems very likely to us that the simultaneous presence of substituents that both can be regarded as “labile” with respect to reactions with lithiated species makes it impossible to obtain clean products in an acceptable yield.

The molecular structures of both compounds 4 and 7 show that intramolecular, interannular Br…S interactions govern the relative orientation of the cyclopentadienyl rings. While it might be expected that large substituents would avoid each other and lead to structures with staggered rings and a maximal number of interannular Cp-H…X contacts, this is not the case. To what extent these interactions are responsible for the formation of particular isomers (as the apparent exclusive formation of the (RR/SS) diastereomer of 4, or the apparent exclusive dominance of the ortho-directing effect of Br over that of the SMe group), can unfortunately not be decided at this moment due to the fact that so far it was not possible to obtain any of the products with 100% purity.

4. Materials and Methods

4.1. Starting Materials, Reagents and Solvents

Ferrocene was obtained commercially and was used as obtained. 1,1′-dibromoferrocene 1 was prepared from ferrocene, n-butyl lithium, TMEDA and tetrabromoethane according to a slightly modified literature procedure [18]. 2,2,6,6-tetramethylpiperidine (HTMP), n-butyl lithium (2.5M solution) and dimethyl disulfide (Me₂S₂) were obtained commercially and used as obtained. Anhydrous tetrahydrofuran (99.9%, saturated with N₂, stabilized with BHT) was obtained commercially and was transferred via cannula to the reaction flasks under N₂, using standard Schlenk techniques. Solvents used for chromatography (hexane, petroleum ether (PE), diethyl ether (Et₂O), dichloromethane) were of analytical grade.

4.2. Reaction of 1,1′-Dibromoferrocene (1) with LiTMP and Me₂S₂

A freshly prepared solution of LiTMP (from 1.47 mL HTMP and 3.49 mL 2.5M n-BuLi solution, corresponding to 8.73 mmol) in THF (85 mL) was treated at −30 °C with 1 (1.00 g, 2.91 mmol) with stirring for 5 h. After cooling to −78 °C Me₂S₂ (0.77 mL, 8.73 mmol) was added, and the solution was gradually warmed to room temperature within 16 h. Then the reaction mixture was filtered through a plug of silica gel and evaporated in vacuo. ¹H-NMR examination of the residue showed more than 10 signals in the SMe region of the spectrum (2.15 ppm < δ ≤ 2.80 ppm, see Figure S1 of the Supporting Information). A first chromatography on silica gel, using petroleum ether as eluent, gave six fractions. The first contained compounds 2 and 3 alongside traces of ferrocene (δ = 4.16 ppm). A second fraction contained a mixture in which signals attributable to compound 4, and presumably 6 and 7 could be identified (Figure S2). The third fraction consisted of a mixture of mainly 4 with its isomer 5 (combined purity ca. 95%, Figure S3). Fraction 4 consisted of a mixture of compounds 5 and 6 with at least one other unidentified compound (Figure S4). The final two fractions showed a plethora of NMR signals and were therefore discarded. The third fraction was re-chromatographed twice, yielding (still impure) compounds 4 (ca. 150 mg, purity ca. 90%, Figure S5) and 5 (76 mg, purity ca. 60%, Figure S6).

4.3. Reaction of 1,1′-Dibromo-2,2′-Bis(Methylthio)Ferrocene (4) with LiTMP and Me₂S₂

A freshly prepared solution of LiTMP (from 0.62 mL HTMP and 1.47 mL n-BuLi solution, corresponding to 3.68 mmol) in THF (35 mL) was treated at −30 °C with 4 (purity ca. 90%,0.83 g, < 1.84 mmol) with stirring for 5 h. After cooling to −100 °C Me₂S₂ (0.33 mL, 3.68 mmol) was added, and the solution was gradually warmed to room temperature within 16 h. After filtration through a plug of silica gel, the reaction mixture was evaporated in vacuo and examined by ¹H NMR spectroscopy. As there were seven major and many more minor signals in the SMe region of the spectrum (Figure S7), a chromatographic purification was attempted. Using a 98:2 mixture of n-hexane and diethyl ether, five fractions could be eluted from a silica gel column. The first fraction consisted mainly of compound 3, while the second contained apparently unreacted 4. Evaporation of the third fraction yielded rather pure compound 6 (100 mg, purity ca. 90%, Figure S8). Fraction 4 was a mixture of many compounds, of which only 7 could be identified (Figure S9). The last fraction was an even more complex mixture and was discarded. An attempted renewed chromatography of fraction 3, using a 99:1 PE/Et₂O mixture as eluent, did not improve the purity.

4.4. Reaction of 1,1′-Dibromo-2,5,2′-Tris(Methylthio)Ferrocene (6) with LiTMP and Me₂S₂

A freshly prepared solution of LiTMP (from 0.036 mL HTMP and 0.085 mL n-BuLi solution, corresponding to 0.21 mmol) in THF (2 mL) was treated at −30 °C with 6 (purity ca. 90%,0.068 g, < 0.11 mmol) with stirring for 5 h. After cooling to −78 °C Me₂S₂ (0.019 mL, 0.21 mmol) was added, and the solution was gradually warmed to room temperature within 16 h. The reaction mixture was filtered through a plug of silica gel with the addition of small amounts of Et₂O, and then chromatographed on silica gel using a 98:2 PE/Et₂O mixture as eluent. Only one band eluted. Evaporation in vacuo yielded impure 7 (61 mg, purity ca. 85%, Figure S10). A mass spectrum of the product showed that it was contaminated by [FeC₁₀Br(SMe)₄H₅] and [FeC₁₀Br₃(SMe)₄H₃] (Figure S11).

4.5. Reactivity Study with 1.1′-2,2′,5,5′-Tetrakismethylthioferrocene

A freshly prepared solution of LiTMP (from 0.091 mL HTMP and 0.216 mL n-BuLi solution, corresponding to 0.54 mmol) in THF (5mL) was treated at −30 °C with 7 (purity ca. 85%,0.095 g, <0.16 mmol) with stirring for 5 h. After cooling to −78 °C, Me₂S₂ (0.05 mL, 0.54 mmol) was added and the solution was gradually warmed to room temperature within 16 h. The reaction mixture was filtered through a plug of silica gel and then evaporated to dryness. The residue was placed on top of another silica gel column and eluted first with 9:1 PE/Et₂O. Five fractions were eluted. Then CH₂Cl₂ was used as eluent and two further fractions were collected. Fractions 2, 4 and 6 were examined by mass spectroscopy (Figure S12). Fraction 2 consisted mainly of [FeC₁₀Br_x(SMe)_yH_10-x-y] with x = 1, 2 and y = 5, 6; fraction 4 consisted of [FeC₁₀Br_x(SMe)_yH_10-x-y] with x = 0, 1, 2 and y = 5, in addition to a species that might be partially oxidized deca(methylthio)ferrocene [FeC₁₀(SMe)₁₀OH⁺]. The sixth fraction was clearly dominated by [FeC₁₀Br₂(SMe)₆H₂] (8). This finding was supported by the ¹H NMR spectrum, which showed only one signal in the Cp region and three in the SMe region (Figure S13).

The NMR data of compounds 4–8 as well as their HRMS data are collected in Table S1 of the Supporting Information.

4.6. Crystallography

Crystals of 4 and 7 were grown from petroleum ether by the slow evaporation of the solvent in a refrigerator, and were measured on Oxford Xcalibur diffractometers (4 on XCalibur 2, 7 on XCalibur Sapphire 3). The obtained data sets were solved using SHELXT [19] and refined using SHEXL 2018/3 [20]. Examination of the structure solutions was performed with the program PLATON as part of the WINGX program suite [21]. Graphics were prepared using either ORTEP3 FOR WINDOWS or MERCURY, both being part of the WINGX program suite as well. For further details of the crystal structure determinations see Table S3 of the Supporting Information.