Synthesis of Polysubstituted Ferrocenesulfoxides

The purpose of the study is to design synthetic methodologies, especially directed deprotometalation using polar organometallic reagents, to access polysubstituted ferrocenesulfoxides. From enantiopure 2-substituted (SiMe3, PPh2) S-tert-butylferrocenesulfoxides, a third substituent was first introduced at the 5 position (SiMe3, I, D, C(OH)Ph2, Me, PPh2, CH2NMe2, F) and removal of the trimethylsilyl group then afforded 2-substituted ferrocenesulfoxides unreachable otherwise. Attempts to apply the “halogen dance” reaction to the ferrocenesulfoxide series led to unexpected results although rationalized in light of calculated pKa values. Further functionalizations were also possible. Thus, new enantiopure, planar chiral di- and trisubstituted ferrocenes have been obtained, in addition to several original 2-substituted, 2,3- and 2,5-disubstituted, 2,3,5-trisubstituted and even 2,3,4,5-tetrasubstituted ferrocenesulfoxides, also enantiopure.

Since the discovery of their parent compound in 1952 [20,21], ferrocenes have established themselves as one of the most important families of organometallics. These three-dimensional compounds in which the iron is surrounded by two cyclopentadienyl rings are often stable to air, water, heat and light, and also exhibit a reversible redox behavior. Therefore, they can bring specific physical and chemical properties to the molecules in which they are included [22]. Thus, they have found many applications in fields such as catalysis [23][24][25], medicinal chemistry [26][27][28], and material science [29,30].
While the first ferrocene sulfoxide was reported in 1964 [31], the first stereopure members of this family were obtained 20 years later by the diastereoselective oxidation of sulfide derivatives of Ugi's amine [32,33]. However, it was only from 1993 that the key methodologies to access enantiopure ferrocene sulfoxides and convert them by diastereoselective deprotolithiation were developed, in particular by Kagan and co-workers [34,35]. 4-Tolylsulfinyl [36][37][38] and tert-butylsulfinyl [39] have been mainly employed as directing In the examples above, the deprotolithiation always occurred on the side toward which the oxygen of the sulfoxide directing group is pointed. In 2012, Tokitoh's group showed that an S-phenylsulfinyl group could direct an initial deprotonation-trapping sequence before being converted to the R-phenylsulfinyl, able to direct the functionalization on the other side, leading to a stereopure 2,5-disubstituted ferrocenesulfoxide [44,45] (Figure 1d). During the synthesis of enantiopure ferrocene-1,2-disulfoxides that we recently reported, we showed that these reduction/oxidation steps could be avoided by adapting the reaction conditions [46]. This possibility of realizing the direct functionalization of the unfavorable position next to the sulfoxide has here been applied to the synthesis of many original derivatives, otherwise unreachable.
In 2014, Šebesta rationalized the diastereoselective deprotolithiation of S-(4-tolyl) ferrocenesulfoxide by lithium amide by using DFT calculations [50]. However, to our knowledge, a related study has never been reported starting from S-tert-butylferrocenesulfoxide. Therefore, to explain the diastereoselectivity observed during the deprotolithiation of this substrate, we compared the thermodynamic acidity of different hydrogen atoms of the cyclopentadienyl ring. Since the pK a values of (S)-S-tert-butylferrocenesulfoxide (S-FcSOtBu) have never been determined, they were calculated within the DFT framework (see Section 3.12 for details), as we had already done for other substrates [51][52][53][54]. For free S-FcSOtBu, the position closest to the sulfinyl oxygen has the largest pK a value (Figure 2a which contradicts the results of deprotolithiation of S-FcSOtBu. This indicates that, as already observed with ketone directing groups [55], the coordination of the sulfinyl oxygen to lithium has a significant effect on deprotolithiation [50] since it can lead to transition states (and lithiated products) stabilized by the formation of cyclic structures involving a lithium ion. To account for this effect, we performed pK a calculations for the complex S-FcSOtBu·LiNMe 2 , in which the sulfinyl oxygen is coordinated to the lithium ion of a base, resulting in a 3-4 unit decrease in pK a values (Figure 2b). At the same time, for the positions close to the sulfinyl group, the anions generated can be stabilized by the formation of cyclic structures involving a lithium ion. For the position closest to the sulfinyl oxygen, the formation of such an anionic cyclic structure leads to a decrease in pK a of more than 11 units compared to the free S-FcSOtBu (Figure 2c). The formation of a similar anionic structure with the other position next to the sulfinyl requires the rotation of the bulky tert-butyl moiety out of its exo position (Figure 2d). However, our calculations showed that the energy costs for such a rotation overcome the energy gain due to the formation of a cyclic anionic structure. As a consequence, the calculated pK a value (34.8; Figure 2d) is slightly higher than in the case of the formation of an anion in which tBu remains in the exo position (34.2; Figure 2b). Thus, the sulfinyl oxygen directs the deprotolithiation only at one of the two neighboring sites, leading to a stabilized cyclic structure involving a lithium ion in which the bulky part of the sulfinyl (here tBu) remains in the exo position, in agreement with the experimental results. The required 2-substituted derivatives of S-tert-butylferrocenesulfoxides were prepared, in the majority of cases, by treating the substrate with n-butyllithium in tetrahydrofuran (THF) at room temperature (rt) for 1 h before the interception by the electrophile (Method A) [34]. Under these conditions, the expected derivatives 1 (substituted on the position closest to the oxygen side of the sulfinyl) were isolated with good yields, whether by trapping with chlorotrimethylsilane ( Table 1, entries 1-2), iodine (entries 3 and 4), heavy water (entry 5) and iodomethane (entry 6). The use of tert-butyllithium (1.5 equiv), this time at −80 • C (Method B) [56,57] in order to avoid decomposition of THF and consequent formation of undesirable products during the trapping step, difficult to separate, has also proved to be suitable for carrying out the deprotonation step; after quenching, the expected phosphine (entry 7) and stannane (entry 8) were isolated in high yields.   1 See the Materials and Methods section for more details on the electrophilic trapping and subsequent hydrolysis. 2 Yields are given after purification, as described in Materials and Methods.

S-FcSOtBu
Deprotometalation of 2-substituted S-tert-butylferrocenesulfoxides is not favorable because the oxygen of the sulfinyl group of these compounds is now directed to the substituent and not to the neighboring free site. However, we have recently shown, using (S,S P )-S-tert-butyl-2-(phenylthio)ferrocenesulfoxide, that Method A is appropriate for such a functionalization [46]. Therefore, we applied these conditions to 1a, a compound that benefits from pK a values close to those of S-FcSOtBu ( Figure 2e) and that contains an easily removable trimethylsilyl protecting group.
This worked satisfactorily since, after interception with chlorotrimethylsilane ( Table 2, entries 1 and 2), iodine (entries 3 and 4) or heavy water (entry 5), the corresponding trisubstituted derivatives 2a were obtained with yields ranging from 63 to 88%. By applying Method B followed by trapping by benzophenone at S,S P -1a, the expected derivative 2ag was also obtained, although with a moderate yield of 35% (entry 6). This last result seems to indicate that a contact of 1.5 h with tert-butyllithium at −80 • C is not sufficient to efficiently generate the lithiated intermediate; this is consistent with the lower reactivity predicted for the 2-substituted S-tert-butylferrocenesulfoxides.
As diphenylphosphino has already been used as a protecting group in the ferrocene series [58], we here explored this ability and tested Method A on the substrate S,S P -1e. Under these conditions, after iodolysis, the product 2eb was isolated as phosphine oxide with a moderate 40% yield (entry 7). However, when the reaction was performed by using lithium 2,2,6,6-tetramethylpiperidide (LiTMP) at −80 • C for 1 h (Method C) before adding chlorotrimethylsilane as an electrophile, the expected ferrocene silane 2ea was formed in 70% yield (entry 8). This better result could be explained by the higher compatibility of this electrophile toward hindered lithium amide, making it possible to shift the supposedly equilibrated deprotolithiation reaction by in situ trapping toward the formation of the silane [59]. Indeed, Method C proved to be unsuitable in the case of electrophiles incompatible with LiTMP, such as iodine.  2 Yields are given after purification, as described in Materials and Methods. 3 S,S P -1a was recovered in 40% yield. 4 Presumed degradation as no S,S P -1e was recovered in this case.
Since S-(4-tolyl)ferrocenesulfoxides are subject to sulfoxide/lithium exchange in the presence of alkyllithiums, their deprotolithiation requires a lithium amide. While lithium diisopropylamide (LiDA) is generally used for this purpose [34,35], we have recently shown that the stronger LiTMP [63] is equally or even more suitable [46]. We therefore prepared the 2-silylated derivative by reacting rac-FcSO-p-Tol with LiTMP (1.2 equiv) in THF at −80 • C before trapping the lithiated intermediate with chlorotrimethylsilane (Table 3, entry 1). By increasing the amount of base to 1.5 equiv in order to improve the yield, we obtained the expected product (43% yield) but also isolated the disilylated derivative rac-3a' (19% yield; entry 2). The formation of the latter shows that the sulfoxide of the first is capable of directing a second deprotolithiation toward a neighboring position located on the tolyl group. Using iodine, an electrophile capable of instantaneously quenching the excess of LiTMP employed, no deprotonation on the tolyl ring was observed (entries 3 and 4), and the monoiodinated product was obtained in up to 81% yield. From S-FcSO-p-Tol, it was even possible to increase the amount of base to 2 equiv in the case of deuteriolysis (conc. DCl) with a similar result (entry 5). Table 3. Deprotometalation of racemic S-(4-tolyl)ferrocenesulfoxide (rac-FcSO-p-Tol) or (S)-S-(4tolyl)ferrocenesulfoxide (S-FcSO-p-Tol) followed by electrophilic trapping. Even though the competitive formation of a disilylated product during deprotolithiationtrapping from rac-FcSO-p-Tol was hardly engaging (Table 3, entry 2), we attempted a similar reaction from R,R P /S,S P -3a by using iodine as an electrophile (Scheme 1a). As feared, the sulfoxide directed the reaction toward the tolyl group, a result evidenced by the formation of the corresponding iodinated derivative R,R P /S,S P -4. Cleavage of the silyl group quantitatively led to S-(2-iodo-4-tolyl)ferrocenesulfoxide (rac-5) from which derivatives of interest could be prepared (Scheme 1b). Scheme 1. Unsuccessful attempts to access 2,5-disubstituted S-(4-tolyl)ferrocenesulfoxides.
Our objective being here to access 2,5-disubstituted S-(4-tolyl)ferrocenesulfoxides, we tested in the same way R,S P /S,R P -3b which bears an electron-withdrawing iodine that should promote the deprotonation on the ferrocene ring. However, after subsequent trapping with chlorotrimethylsilane, the expected product R,S P /S,R P -6 was only isolated in a low 11% yield due to a high recovery (80%) of the starting material (Scheme 1c). By introducing chlorotrimethylsilane before the base, in order to displace by in situ trapping the supposedly equilibrated deprotolithiation reaction toward the formation of the silane [59], the substrate R,S P /S,R P -3b was completely recovered. These unsuccessful experiments led us to abandon the S-(4-tolyl)ferrocenesulfoxide route and turn again to S-(tert-butyl)ferrocenesulfoxides.
Interestingly, removal of the trimethylsilyl group from these 2,5-disubstituted S-tertbutylferrocenesulfoxides 2 can lead to 2-substituted S-tert-butylferrocenesulfoxides 1, diastereoisomers of those obtained by the direct deprotometalation-trapping from S-tertbutylferrocenesulfoxides. This desilylation was easily achieved by using tetrabutylammonium fluoride (TBAF; 2 equiv) in THF at rt [64], providing novel 2-substituted derivatives of S-tert-butylferrocenesulfoxides in high yields ( Table 5, entries [1][2][3][4][5]. From the S-tert-butyl-2,5-bis(trimethylsilyl)ferrocenesulfoxides 2aa, reducing the amount of TBAF to 1 equiv in order to avoid bis-desilylation selectively afforded the product monodesilylated on the side of the sulfur lone pair (entries 6 and 7). Therefore, its diastereoisomer (S,R P )-S-tertbutyl-2-(trimethylsilyl)ferrocenesulfoxide (S,R P -1a) cannot be reached by this approach. To our knowledge, there is no precedent on the topic. However, this result seems to suggest that the oxygen of the tert-butylsulfinyl group prevents to some extent an attack of the trimethylsilyl group by the fluoride. Table 5. Desilylation of 2,5-disubstituted S-tert-butylferrocenesulfoxides. 1 Yields are given after purification, as described in the Materials and Methods section. 2 1 equiv of nBu 4 NF was used in this case.

Attempts to Apply the "Halogen Dance" Reaction to the Ferrocenesulfoxide Series
The "halogen dance" is a reaction in which halogen-substituted aromatics are isomerized [65][66][67][68][69][70][71][72][73]. Requiring a hindered lithium amide such as LiTMP, the reaction is driven by the stability of the arylmetal formed. As evidenced in the ferrocene series in the 2010s [60,74], it has since evolved to currently represent a valuable synthetic tool [51][52][53][54]64,[75][76][77][78]. In the continuity of this work, we sought to implement this reaction by using the sulfoxide as a stabilizing/directing group and the trimethylsilyl as a protecting group. The pK a values calculated for the complexes between 2-and 3-iodinated S-tertbutylferrocenesulfoxides and LiNMe 2 seem to indicate that the S,S P (or R,R P ) stereoisomer would only be a suitable substrate after the protection of the free position next to the sulfoxide (Figure 3a,b). Indeed, as the position activated by the sulfoxide is the most acidic for both S,S P -1b·LiNMe 2 and its isomerized derivative, a migration is not expected without protection. Regarding the S,R P (or R,S P ) stereoisomer, the pK a values indicate that this substrate could be tested protected or even as is (Figure 3c,d). Since the pK a values indicated that (S,R P )-S-tert-butyl-2-iodoferrocenesulfoxide (S,R P -1b) might be used in "halogen dance" without trimethylsilyl protection, the reaction was attempted by using our standard conditions (1.1 equiv of LiTMP, THF, −50 • C, 2 h; Method F) [51][52][53][54]64,75,77,78]. However, after methanolysis, a complex mixture was obtained within which only S,S P -1b could be identified. When chlorotrimethylsilane was used as the electrophile after 1 h of contact, a mixture was obtained, from which S,S P -2ab, S,S P -7, S,S P -1b and S-FcSOtBu were isolated with yields of 20, 18, 9, and 8%, respectively (Scheme 2). While the structure of S,S P -7 might appear as atypical for a classic "halogen dance" reaction, it is in agreement with the formation of the compound S,S P -1a by iodine/lithium exchange, and was further unambiguously confirmed by X-ray diffraction. Scheme 2. Attempt to perform "halogen dance" from (S,R P )-S-tert-butyl-2-iodoferrocenesulfoxide (S,R P -1b).
While we already evaluated the behavior of (S,S P )-S-tert-butyl-2-iodo-5-(trimethylsilyl) ferrocenesulfoxide (S,S P -2ab) in the "halogen dance" reaction, it was of interest to test its diastereoisomer S,R P -2ba under similar conditions. As it is possible to deprotolithiate the sulfoxide adjacent position of S,R P -1b (Scheme 2), it was treated successively with LiTMP in THF at −80 • C for 1 h (Method C) and then with chlorotrimethylsilane. However, instead of the expected product, we again isolated S,S P -7 (this time in 49% yield) as well as S,S P -2ab in 16% yield (Scheme 4a). Additionally, the use of LiTMP at −80 • C in the presence of chlorotrimethylsilane as an in situ trap (both used in excess) also led to the products S,S P -7 (24% yield) and S,S P -1a (30% yield) (Scheme 4b). Taken together, a putative reaction pathway including two successive "halogen dance" reactions toward the compound S,S P -7 could be proposed (Scheme 5). Scheme 4. Unsuccessful attempts to access (S,R P )-S-tert-butyl-2-iodo-5-(trimethylsilyl)ferrocenesul foxide (S,R P -2ba).
Although S,R P -2ba cannot be reached directly from S,R P -1b, it can be prepared by following our recently reported work [46]. However, when it was engaged into a "halogen dance" reaction (Method F), we only observed an iodine/lithium exchange, leading to the compound S,R P -1a (39% yield), while 50% of starting material was recycled. Replacement of iodine from S,R P -2ba with bromine in S,R P -2ja (prepared by using Method A) did not change the outcome of the reaction, as S,R P -1a was this time isolated in 33% yield (Scheme 6). Scheme 6. Access to (S,R P )-S-tert-butyl-2-(trimethylsilyl)ferrocenesulfoxide (S,R P -1a) and reactivity.

On the Way to Polysubstituted Ferrocenesulfoxides
Since fluorine is also a group able to direct deprotometalation [80,81], we then studied the reactivity of the fluorinated ferrocenesulfoxide S,R P -1i (see Table 5). Inspired by previous studies in fluoroferrocene series [52,64], we chose sec-butyllithium to carry out the reaction in THF at −75 • C (Method G). Trapping with iodine or chlorodiphenylphosphine yielded the halide S,R P -2bi and the phosphine S,S P -2ei, a result consistent with a sulfoxidedirected deprotolithiation (Scheme 7). Scheme 7. Deprotometalation-trapping sequences from (S,R P )-S-tert-butyl-2-fluoroferrocenesulfoxide (S,R P -1i) and pK a values.
We then thought interesting to observe the behavior of these (tert-butylthio)ferrocenes in deprotometalation as there is only one mention of such a reaction in the literature [83].
In their study, Brown and co-workers employed sec-butyllithium in THF under different conditions to deprotonate (4-tolylthio)ferrocene quite regioselectively: either at C3 by performing the reaction at 0 • C, or at C1 at 30 • C, or even at C2 at −75 • C in the presence of potassium tert-butoxide. A similar regioselectivity in the case of (tert-butylthio)ferrocene was also claimed, but without reporting the electrophiles used and the yields obtained. 3-(tert-Butylthio)ferrocenecarboxaldehyde has since been mentioned in a scheme of a patent [84], but again without further details.
In our hands, despite the presence of fluorine as a directing group, S P -8ai could not be deprotometalated by using sec-butyllithium in THF at temperatures between −75 and 0 • C, or even at −75 • C in the presence of potassium tert-butoxide. Indeed, after subsequent iodolysis, the starting material was always recovered. This reluctance to deprotometalation is surprising since fluoroferrocene is functionalized under similar conditions. From R P -8i, which benefits from pK a values very close to those of fluoroferrocene [52] (Scheme 8), an inseparable mixture of iodides and starting material was obtained when sec-butyllithium was employed in the presence of potassium tert-butoxide at −75 • C.
Thus, we turned back to the fluorinated ferrocenesulfoxide S,S P -2ai to consider its further functionalization. This time, the use of sec-butyllithium in THF at −75 • C for 1 h (Method G) and subsequent quenching with iodine or iodomethane resulted in the clean formation of the products functionalized next to fluorine with excellent yields (Scheme 9a). From the methylated product S,S P -9d, desilylation gave S,R P -10. Deprotolithiation of the latter logically occurred next to the sulfoxide group by using Method G to give, after trapping with iodine or hexachloroethane, the sulfoxide-containing tetrasubstituted ferrocenes S,R P -11 (Scheme 9b).
In order to reach the first sulfoxide-based pentasubstituted ferrocene, the chlorinated compound S,R P -11k was subjected to Method G. Surprisingly, instead of the expected deprotometalation next to chlorine, we observed a sulfoxide-induced chlorine/lithium exchange [85,86]; this was demonstrated by subsequent trapping with iodine, leading to S,R P -11b. This iodide was therefore subjected to an iodine/lithium exchange-hexachloroethane trapping sequence in order to recover the starting chloride S,R P -11k. This could finally be converted to the expected original 2,3,4,5-tetrasubstituted ferrocenesulfoxide S,S P -12 by using Method C with chlorotrimethylsilane as the electrophile (Scheme 10). Scheme 10. Deprotometalation-trapping sequences from (S,R P )-S-tert-butyl-5-chloro-2-fluoro-3methylferrocenesulfoxide (S,R P -11k).
The only difference between the compounds S,S P -7 and S,S P -9b is an additional fluorine atom for the latter. Therefore, they have almost identical structural characteristics in terms of C-S, C-Si, C-I bond lengths, and coplanarity between the oxygen of the sulfinyl group and the substituted cyclopentadienyl ring ( Figure 5). Furthermore, both ferrocene cores were found in an eclipsed conformation with a torsion angle C10-Cg1···Cg2-C1 of −22.47 • for S,S P -7 and a torsion angle C9-Cg1···Cg2-C1 of −19.42 • for S,S P -9b (Cg1 being the centroid of the C6-C7-C8-C9-C10 ring and Cg2 being the centroid of the C1-C2-C3-C4-C5 ring). Finally, due to the presence of iodine remote from the sulfoxide, interactions between the former and the oxygen of the latter were identified, leading to chains of molecules at the solid-state ( Figure 6) [53,54,89]. For the two compounds, their I···O bond lengths and their S-O···I and O···I-C angles indicate an interaction between the lone pair of the oxygen and the σ-hole of the iodine atom, characteristic of halogen bonds [90,91].

General Information
All reactions were carried out in Schlenk tubes under a dry argon atmosphere. THF was freshly distilled under argon from sodium-benzophenone. All alkyllithiums were titrated before use [92]. 2,2,6,6-Tetramethypiperidine was distilled over CaH 2 under reduced pressure and stored over KOH pellets. Room temperature (rt) refers to 25 • C. Column chromatography separations were achieved on silica gel (40-63 µm). All thin-layer chromatographies (TLC) were performed on aluminum-backed plates pre-coated with silica gel (Merck, Silica Gel 60 F254). They were visualized by exposure to UV light. Melting points were measured on a Kofler apparatus. Infrared (IR) spectra were taken on an ATR Perkin-Elmer Spectrum 100 spectrometer (Perkin-Elmer, Waltham, MA, USA) and the main absorption wavenumbers are given in cm −1 . 1 H and 13 C{ 1 H} nuclear magnetic resonance (NMR) spectra were recorded at 300 K either on a Bruker Avance III HD spectrometer fitted with a BBFO probe at 500 MHz and 126 MHz, respectively, or on a Bruker Avance III spectrometer fitted with a BBFO probe at 300 MHz and 75.4 MHz respectively (Bruker, Billevica, MA, USA). 1 H chemical shifts (δ) are given in ppm relative to the solvent residual peak, and 13 C{ 1 H} chemical shifts (δ) are given in ppm relative to the central peak of the solvent signal [93]. Cp refers to the unsubstituted cyclopentadienyl ring of ferrocene. The NMR data of all compounds described, selected NOESY correlations and the numbering used below for the NMR assignment is given in the Supplementary Materials. Optical rotations were determined on a Perkin Elmer 341 polarimeter (589 nm); the concentrations (c) are given in g/100 mL.

Crystallography
The samples were studied with monochromatized Mo-Kα radiation (λ = 0.71073 Å). The X-ray diffraction data of the compounds S,R P -1b, S,S P -1d and S,S P -9b were collected at T = 150(2) K by using a D8 VENTURE Bruker AXS diffractometer equipped with a (CMOS) PHOTON 100 detector. The X-ray diffraction data of the compounds FcSOFc and S,S P -7 were collected at T = 150(2) K by using an APEXII Kappa-CCD (Bruker-AXS) diffractometer equipped with a CCD plate detector and a CCD-LDI-APEX2 detector, respectively. The crystal structures were solved by the dual-space algorithm using SHELXT program [94] and then refined with full-matrix least-squares methods based on F 2 (SHELXL program) [95]. All non-hydrogen atoms were refined with anisotropic atomic displacement parameters. H atoms were finally included in their calculated positions and treated as riding on their parent atom with constrained thermal parameters. The molecular diagrams were generated by Mercury 2020.3.0.

Safety Considerations
Due to its high pyrophoric character, tert-butyllithium has to be used under anhydrous conditions and nitrogen or argon atmosphere.

(4-Tolylthio)ferrocene
To a solution of ferrocene (9.30 g, 50.0 mmol) and potassium tert-butoxide (0.56 g, 5.0 mmol) in THF (375 mL) at −80 • C was added dropwise a 1.6 M hexane solution of tBuLi (62.5 mL, 100 mmol). The reaction mixture was stirred at −80 • C for 1 h before the addition of a solution of di-p-tolyl disulfide (7.8 mL, 40.0 mmol) in THF (60 mL). The reaction mixture was warmed to rt and then stirred for 1 h. Water was added and the reaction mixture was extracted with diethyl ether. The organic phase was washed three times with a 10% NaOH aqueous solution (3 × 10 mL). Drying over MgSO 4 and removal of the solvents under reduced pressure led to the crude product, which was partially purified by column chromatography over silica gel (eluent: petroleum ether-EtOAc-CHCl 3 100:0:0 to 98:1:1) to give the crude product used in the oxidation step. Recrystallization of the crude can afford pure (4-tolylthio)ferrocene as an orange solid for analysis. Mp 114 • C (lit. [97] 110. 5

General Procedure A: Deprotolithiation of S-tert-Butylferrocenesulfoxides Using nBuLi Followed by Electrophilic Trapping
This was adapted from a previously reported procedure [34]. To a solution of the ferrocenesulfoxide (1.0 mmol) in THF (5 mL) at 0 • C was added dropwise a 1.4 M hexane solution of nBuLi (0.86 mL, 1.2 mmol). After 15 min, the reaction mixture was warmed to rt and stirred at this temperature for 1 h. The electrophile (1.5 mmol unless otherwise specified; either pure for liquids or in solution for solids, as indicated below) was next added at 0 • C. The reaction mixture was kept at 0 • C for 15 min and warmed to rt. The addition of 1 M HCl (5 mL), or saturated aqueous Na 2 S 2 O 3 in the case of I 2 , extraction with EtOAc (3 × 20 mL), drying over MgSO 4 , and removal of the solvents under reduced pressure led to the crude product, which was purified by chromatography over silica gel (eluent given in the product description). When subsequent desilylation was performed, the protocol was as follows [64]. The silylated ferrocene (1.0 mmol) was treated by nBu 4 NF (1.0 M THF solution; 1.6 mL, 2.0 mmol) in THF (5 mL) at rt for 0.5 h. The solvent was removed under reduced pressure, and the product was purified by chromatography over silica gel (eluent given in the product description).

General Procedure B: Deprotolithiation of Enantiopure S-tert-Butylferrocenesulfoxides Using tBuLi Followed by Electrophilic Trapping
This was adapted from a previously reported procedure [56,57]. To a solution of the ferrocenesulfoxide (1.0 mmol) in THF (12.5 mL) at −80 • C was added dropwise a 1.6 M pentane solution of tBuLi (0.94 mL, 1.5 mmol), and the reaction was stirred at this temperature for 1.5 h before the addition of the electrophile (1.5 mmol unless otherwise specified; either pure for liquids or in solution for solids, as indicated below). The mixture was stirred at −80 • C for 0.5 h before being warmed to rt. The addition of 1 M HCl (5 mL), extraction with EtOAc (3 × 20 mL), drying over MgSO 4 , and removal of the solvents under reduced pressure led to the crude product, which was purified by chromatography over silica gel (eluent given in the product description).

General Procedure C: Deprotolithiation of S-tert-Butylferrocenesulfoxides Using LiTMP Followed by Electrophilic Trapping
This was adapted from a previously reported procedure [54]. To a stirred, cooled (−15 • C) solution of 2,2,6,6-tetramethylpiperidine (0.28 mL, 1.6 mmol) in THF (2 mL) was added dropwise a 1.4 M hexane solution of nBuLi (1.1 mL, 1.5 mmol). The mixture was stirred for 5 min at −15 • C and then for 2 min at −80 • C and next cannulated onto a solution of the ferrocenesulfoxide (1.0 mmol) in THF (3 mL) at −80 • C. After 1 h at this temperature, the electrophile (1.5 mmol; either pure for liquids or in solution for solids, as indicated below) was introduced at −80 • C before warming to rt. The addition of MeOH (0.5 mL) and removal of the solvents under reduced pressure led to the crude product, which was purified by chromatography over silica gel (eluent given in the product description). The general procedure C, but using LiTMP (1.1 equiv) from (S,R P )-S-tert-butyl-2iodoferrocenesulfoxide (S,R P -1b; 0.42 g) and using ClSiMe 3 (0.14 mL, 1. CCDC 2152198. (S,S P )-S-tert-Butyl-2-iodo-5-(trimethylsilyl)ferrocenesulfoxide (S,S P -2ab) was similarly isolated in 16% yield (79 mg). The position of the trimethylsilyl group of S,S P -7 was confirmed by iodine/lithium exchange: by successively treating S,S P -7 with 2 equiv nBuLi in THF (0 • C then rt, 1 h) and MeOH in excess, (S,S P )-S-tert-butyl-2-(trimethylsilyl)ferrocenesulfoxide (S,S P -1a) was isolated in 77% yield. By using 1.3 equiv of LiTMP and 1.3 equiv of ClSiMe 3 to perform the iodine migration, S,S P -7 was isolated in 49% yield.

General Procedure D: Deprotolithiation of S-(4-tolyl)ferrocenesulfoxides Using LiTMP Followed by Electrophilic Trapping
This was adapted from a previously reported procedure [54]. To a solution of the S-(4tolyl)ferrocenesulfoxide (0.32 g, 1.0 mmol) in THF (3.5 mL) at −80 • C was added dropwise a solution of LiTMP [prepared at −15 • C by the addition of a 1.4 M hexane solution of nBuLi (0.86 mL, 1.2 mmol) to 2,2,6,6-tetramethylpiperidine (0.22 mL, 1.3 mmol) in THF followed by stirring for 5 min (1.5 mL)] cooled at −80 • C. After 0.5 h at this temperature, the electrophile (1.2 mmol unless otherwise specified; either pure for liquids or in solution for solids, as indicated below) was added. The mixture was stirred at −80 • C for 1 h before the addition of H 2 O (5 mL) or saturated aqueous Na 2 S 2 O 3 in the case of I 2 . Extraction with EtOAc (3 × 20 mL), drying over MgSO 4 , and removal of the solvents under reduced pressure led to the crude product, which was purified by chromatography over silica gel (eluent given in the product description). When subsequent desilylation was performed, the protocol was as follows [64]. The silylated ferrocene (1.0 mmol) was treated by nBu 4 NF (1.0 M THF solution; 1.6 mL, 2.0 mmol) in THF (5 mL) at rt for 0.5 h. The solvent was removed under reduced pressure, and the product was purified by chromatography over silica gel (eluent given in the product description).

General Procedure E: One-Pot Deprotolithiation-Trimethylsilylation-Deprotolithiation-Trapping of S-tert-Butylferrocenesulfoxides
To a solution of S-tert-butylferrocenesulfoxide (S-FcSOtBu or R-FcSOtBu; 0.29 g, 1.0 mmol) in THF (5 mL) at 0 • C was added dropwise a 1.4 M hexane solution of nBuLi (0.79 mL, 1.1 mmol). After 15 min, the mixture was warmed to rt and stirred at this temperature for 1 h. ClSiMe 3 (0.14 mL, 1.1 mmol) was introduced at 0 • C and, after 15 min, the mixture was warmed to rt and stirred at this temperature for 1 h. To this mixture, cooled at 0 • C, was next added dropwise a 1.4 M hexane solution of nBuLi (1.1 mL, 1.5 mmol). After 15 min at 0 • C, the mixture was warmed to rt and stirred at this temperature for 1 h. The electrophile (either pure for liquids or in solution for solids, as indicated below) was next added at 0 • C. The mixture was kept at 0 • C for 15 min and warmed to rt. The addition of 1 M HCl (5 mL), or saturated aqueous Na 2 S 2 O 3 in the case of I 2 , extraction with EtOAc (3 × 20 mL), drying over MgSO 4 , and removal of the solvents under reduced pressure led to the crude product, which was purified by chromatography over silica gel (eluent given in the product description). When subsequent desilylation was performed, the protocol was as follows [64]. The silylated ferrocene (1.0 mmol) was treated by nBu 4 NF (1.0 M THF solution; 1.6 mL, 2.0 mmol) in THF (5 mL) at rt for 0.5 h. The solvent was removed under reduced pressure, and the product was purified by chromatography over silica gel (eluent given in the product description). The general procedure E, from R-FcSOtBu, using I 2 (0.38 g, 1.5 mmol) in THF (3 mL) afforded (eluent: petroleum ether-EtOAc 80:20; Rf = 0.82) the title product in 53% yield (0.26 g). The analyses are as described above (see Section 3.5.12). To a stirred, cooled (−15 • C) solution of 2,2,6,6-tetramethylpiperidine (0.19 mL, 1.1 mmol) in THF (5 mL) was added a 1.4 M hexane solution of nBuLi (0.79 mL, 1.1 mmol). The mixture was stirred for 5 min at −15 • C and then for 2 min at −50 • C before the introduction of the iodoferrocene (1.0 mmol) in one portion. After 2 h at this temperature, methanol in excess (2 mL) was introduced at −50 • C before warming to rt and the addition of aqueous HCl (1 M, 10 mL). Extraction with EtOAc (3 × 20 mL), drying over MgSO 4 , and removal of the solvents under reduced pressure led to the crude product, which was purified by chromatography over silica gel (eluent given in the product description).

General Procedure G: Deprotolithiation of Enantiopure Ferrocenes using sBuLi Followed by Electrophilic Trapping
This was adapted from a previously reported procedure [52,64]. To a solution of the ferrocene (1.0 mmol) in THF (3 mL) at −75 • C was added dropwise a 1.3 M cyclohexane solution of sBuLi (0.92 mL, 1.2 mmol), and the reaction was stirred at this temperature for 1 h before the addition of the electrophile (1.2 mmol unless otherwise specified; either pure for liquids or in solution for solids, as indicated below). The mixture was stirred at −75 • C for 15 min before being warmed to rt. The addition of 1 M HCl (5 mL), or saturated aqueous Na 2 S 2 O 3 in the case of I 2 , extraction with EtOAc (3 × 20 mL), drying over MgSO 4 , and removal of the solvents under reduced pressure led to the crude product, which was purified by chromatography over silica gel (eluent given in the product description). When subsequent desilylation was performed, the protocol was as follows [64]. The silylated ferrocene (1.0 mmol) was treated by nBu 4 NF (1.0 M THF solution; 1.6 mL, 2.0 mmol) in THF (5 mL) at rt for 0.5 h. The solvent was removed under reduced pressure, and the product was purified by chromatography over silica gel (eluent given in the product description).

Computational Details
The quantum chemical calculations were performed using GAUSSIAN 09 package [101]. All computations were conducted within the DFT framework. The CAM-B3LYP [102] hybrid functional was employed. The optimized geometries were obtained using the LANL2DZ basis set for both Fe and I and the 6-31G(d) basis set for the other atoms. No symmetry constraints were applied. To check stationary points and calculate zero-point vibrational energies (ZPVE) and thermal corrections, the Hessian matrices were calculated at the same level of theory. The single point energies of species were obtained at the CAM-B3LYP/LANL2DZ + 6-311 + G(d,p) level.
For the calculation of the CH acidities, we employed the approach successfully used before in the ferrocene series, including carboxamides [47], halides [48], sulfonates [71], and sulfonamides [72], and described therein. Briefly, the gas-phase acidity ∆G acid was calculated as the Gibbs energy of deprotonation of the corresponding substrate R-H (R-H(g) → R − (g) + H + (g)): The solvent influence was treated by using the polarized continuum model (IEF PCM) with the default parameters for THF [103].
The following isodesmic reaction was considered for the pK a values calculation: R-H(s) + Het − (s) → R − (s) + Het-H(s) where Het-H is furan. The latter was chosen as the reference compound due to its structural similarity and since its pK a (THF) = 35.6 reported by Fraser et al. [104] was expected to be close to the substrates under consideration.
Regarding the diversity of bases used, we chose LiNMe 2 as a model compound to track the influence of lithium coordination on the pK a values.
The calculated values of the Gibbs energies ∆G acid [kcal·mol −1 ] for deprotonation are given in the Supplementary Materials.

Conclusions
The purpose of this article was to show that the tert-butylsulfinyl group, already known to direct the deprotometalation on ferrocene to a privileged neighboring site, can be used more generally to access more substituted derivatives.
To this end, by starting from classical 2-substituted S-tert-butylferrocenesulfoxides, deprometalation conditions were found to introduce other substituents on the less activated ferrocene position next to the sulfoxide. Subsequent removal of the trimethylsilyl group led to 2-substituted S-tert-butylferrocenesulfoxides otherwise inaccessible. Their functionalization turned out to be easy, leading to many new stereopure di-to tetrasubstituted ferrocenesulfoxides.
Because the sulfoxide function can be reduced, these methodologies open the way to new planar chiral polysubstituted ferrocenes.