Phosphaalkenylidene bridged ferrocenes.

The lithiation of ferrocenylphosphane Fc-PH(2) (Fc = -C(5)H(4)FeC(5)H(5)) has been reinvestigated and both Fc-PHLi and Fc-PLi(2) have been identified by NMR-spectroscopy. The lithiated phosphanides have been converted to the corresponding mono and bis(silylated) species the latter of which gave synthetic access to an oligomer in which three ferrocene units are symmetrically connected by phosphaalkene units. The charge distribution within this oligomer and its isomers has been analyzed using DFT calculations which indicates that the iron atom of the central metallocene unit is slightly more positive than the terminal ones. These findings are supported experimentally by Mößbauer spectroscopy and cyclic voltammetry.

Introduction

Main group based π-conjugated materials have recently attracted a lot of attention in the field of molecular electronics. The unique electronic nature of phosphorus in particular endows molecules with different and interesting properties [1], [2], [3]. Phosphaalkene units alternating with arylene moieties have been incorporated into conjugated oligomers and polymers without the requirement for strict steric protection as in the case of diphosphenes [4], [5], [6], [7]. Also functional arenes, pyridine or metallocenes and heteroatoms with π-donor abilities have been used to bridge bis(phosphaalkenes) [8], [9], [10], [11], [12], [13]. By extending this concept we were interested to connect several ferrocene units via phosphaalkene bridges to an extended conjugated π-system. Besides extended conjugation also the redox-activity of ferrocene moieties should allow unusual properties. For instance closely related carboxamide bridged oligoferrocenylenes show charge dependent conformational changes as has been nicely demonstrated recently [14], [15].

Here we report the preparation of a phosphaalkene bridged triferrocenyl system and the investigation of its electronic situation with spectroscopic and computational methods.

Results and discussion

Employing the Becker route to phosphaalkenes we have recently synthesized the first metallocene bridged bis(phosphaalkene) [9]. Generally, this approach is based on the reaction of a bis(trimethylsilyl)phosphane with a carboxylic acid chloride with trimethylsilyl chloride (TMSCl) elimination and subsequent rearrangement of the corresponding acylphosphane to the phosphaalkene [16]. In order to get a bis(phosphaalkene) which is C-bonded to the central ferrocenylene unit, the reaction of the corresponding 1,1′-ferrocenylene bis(carboxylic)acid chloride (Fc′(COCl)2) with ferrocenyl bis(trimethylsilyl)phosphane 5 would be a straightforward synthetic approach (Scheme 1). Since compound 5 is not available in the literature we first set out to explore its preparation. Silylated phosphanes should be accessible via initial lithiation and subsequent silylation starting from the known ferrocenylphosphane 1. Reaction of 1 with 1 eq. of n-BuLi in THF at −78 °C affords the monolithiated phosphane 2. The latter shows a 31P NMR chemical shift at −161.7 ppm which is exactly the value reported for the dilithiated species 3 by Cowley et al. [17]. Our assignment of this NMR signal to compound 2 is further corroborated by a 1JPH-coupling of 170 Hz. Lithiation of 1 with 2 eq. of n-BuLi in DME indeed affords the dilithiated phosphane 3 which shows a resonance δ(31P) at −243.2 ppm with no PH-coupling. Therefore we assume that FcPLi2 (3) has actually not been observed in ref. [17] and was rather confused with FcPHLi (2) as indicated by the δ(31P) shift values.

Scheme 1

Synthetic route to bis(phosphaalkene) 6.

Silylation of 2 does not cleanly afford monosilylated phosphane 4. Instead a mixture of mono- and bis(silylated) compounds 4 and 5 is obtained besides unsilylated 1 (Scheme 1). This silyl-redistribution can be attributed to the increased acidity of the P–H bond in 4 compared to 1 which means in turn that in situ formed 4 is deprotonated by 2 affording FcPTMSLi which is subsequently silylated to 5. However bis(silylated) 5 can be easily obtained in a clean reaction from dilithiated 3 in DME. Compound 5 shows a δ(31P) shift at −167.0 ppm which differs significantly from that of its monosilylated congener 4 which resonates at −149.2 ppm (1JPH = 200 Hz).

With the bis(silylated) ferrocenylphosphane in hands we set out to synthesize bis(phosphaalkene) 6 according to the Becker route. The reaction was carried out in pentane as a solvent and gives 6 in 55% yield. Owing to the possibility of E-Z-isomers for each phosphaalkene unit, three diastereomers can be anticipated for 6 (i.e. EE, ZZ and EZ) for which different 31P NMR resonances can be expected. Out of the three possible resonances only two could be observed, which may be attributed to the close proximity of the δ(31P) chemical shift values of the phosphaalkene units in the mixed isomer EZ to those of isomers EE and ZZ respectively. Therefore the observed values of 124.3 and 127.4 ppm have been attributed to the E and Z units in 6. This interpretation is backed by our findings of a related metallocene bridged bis(phosphaalkene) (t-Bu(TMSO)CP–)2Fc′ and its experimental 31P NMR shift values in comparison with values obtained by GIAO quantum chemical calculations which indicate, that the E/Z configuration has little influence on the isotropic shift values, but significant impact on the shift anisotropy [9].

The likewise occurrence of these isomers suggests a similar thermodynamic stability for these diastereomers. This is corroborated by DFT calculations which show only small energy differences between the three isomers of 6. Employing the B3LYP functional and a 6-311g(d) basis set, we found EZ-6 to be the most stable isomer while ZZ-6 is only 1.6 kcal/mol higher in energy and the energetically most disfavored isomer EE-6 is 10.3 kcal/mol higher in energy. According to our TD-DFT calculations the isomers of 6 exhibit striking differences in their electronic transitions with respect to their nature, energies and intensities (Fig. 1). For isomer EE-6 one major transition at 550 nm is observed which can be associated with an electronic transition from mainly π-orbitals of both PC bonds and the central ferrocene to their corresponding anti-bonding orbitals. Furthermore weaker transitions at around 500 nm and 480 nm are attributed mainly to Fe(d) → (Fc)* transitions where (Fc)* refers to linear combinations of Fe(d) and Cp(π) states. Isomer EZ-6 shows several major transitions, which are generally weaker than those for EE-6 in the range 500–630 nm according to our calculations (Fig. 1). Interestingly, the absorption around 630 nm corresponds also to a (π(PC) + Fc) → (π(PC) + Fc)* transition in which all three ferrocene units are involved leading to a significant bathochromic shift compared to EE-6. Slightly higher in energy the next band around 580 nm, consisting of three electronic transitions, displays mainly a Fe(d) + π(PC) → (Fc)* character.

Fig. 1

Calculated electronic transitions in EE-6, EZ-6 (top) and ZZ-6 (bottom panel).

The energetically lowest transitions around 570 nm in ZZ-6 display the same characteristics as those found for EZ-6, having contributions from all ferrocene units. Similarly, absorptions around 490 nm comprise Fe(d) + π(PC) → (Fc)* transitions. To our surprise, the ZZ isomer of 6 features an extremely strong absorption around 420 nm which is attributed to a π(PC) → π*(PC) transition with minor contributions from Fe(d) orbitals.

The calculated transitions fit well to the experimental UV–vis spectra in which overlapping absorption bands at 530, 542 and 553 nm have been recorded for the isomeric mixture of 6.

The significant differences of the calculated absorption spectra display the different possibilities of communication through the PC linking units, as seen by the hypsochromic shift of the longest absorption in EE-6. The rich transitions in the EZ isomer of 6 might also result from the larger asymmetry introduced in the mixed isomer. Importantly, most of the acceptor orbitals are ferrocene based, which indicates further possibilities in the tuning of such systems.

A possible explanation for the increased number of transitions in the Z isomer may be owing to the proximity of the lone pair at phosphorus to the carbon bonded ferrocenylene moiety which is changed to –O–TMS in the E isomer. This arrangement may facilitate MLCT-transitions in the Z form similar to those in ferrocenyldiphosphenes [18], [19], [20], [21].

Based on Mulliken population analysis the partial charge distribution within the ferrocene units differs only slightly. According to this method, the iron atom of the inner ferrocenylene unit is slightly more positive than the outer iron atoms. The influence of the isomers is only marginal. While ZZ-6 and EE-6 give almost identical partial charges, EE-6 exhibits slightly diminished partial charges for all iron centers. A summary of the charge analysis is provided in Table 1. Nevertheless, the difference of partial charges on the iron atoms (qinner − qouter) is Δq = 0.02, and hence nearly identical for all studied isomers.

Table 1

Mulliken charges on iron atoms of the ferrocene units derived from DFT calculations (B3LYP//6-311g(d)). Δq: Charge difference between averaged outer and inner iron atoms.

	ZZ-6	EZ-6	EE-6
qouter (Fe6)	0.965	0.965	0.944
qinner (Fe22)	0.986	0.981	0.962
qouter (Fe38)	0.965	0.963	0.945
Δq	0.021	0.017	0.017
Ratio	1.022	1.018	1.018

Temperature-dependent Mössbauer effect (ME) spectroscopy has proven itself a valuable technique for the elucidation of charge state, charge distribution, metal atom dynamics and vibrational anisotropy of organo-iron complexes, especially those involving the bis(cyclo-pentadienyl) iron moiety (see for example lit. [22], [23], [24]. and references therein). As is true of most neutral FeCp2 complexes, the ME spectra of 6 consist of doublets, characterized by their isomer shift (IS), quadrupole splitting (QS), and recoil free fraction (f) related to the metal atom vibrational amplitude. The ME spectra of 6 acquired at 172 K indicate that two iron species are present which show an IS of 0.511 ± 0.005 and 0.505 ± 0.009 mm s−1 and a quadrupole splitting of 2.365 ± 0.005 and 2.169 ± 0.009 mm s−1 at 90 K. This spectrum is shown graphically in Fig. 2. The IS(90) value is only slightly less than that reported for the parent ferrocene suggesting a minor increase in the s-electron density at Fe center when a hydrogen atom of the Cp ligand is replaced by the phosphaalkene group in 6. Matrix inversion least squares fitting analysis shows that the temperature dependencies of the IS for the two sites are nearly identical (−2.36 ± 0.20 mm s−1 K−1). The QS parameter is essentially T insensitive over the temperature range of the ME measurements (96 < T < 240 K). The area ratio for the two Fe sites is close to 2:1, as expected from the sample stoichiometry and is not T dependent. The area ratio of each site is also T insensitive, indication essentially isotropic metal atom motion relative to the principal symmetry axis including the metal atom. The hyperfine parameter systematics, as well as the area ratio parameter leads to an unambiguous identification of site 1 with the two “outer” Fe atoms and site 2 with the single “inner” site. As noted above, the near identity of the IS for the two metal sites suggests that the s-electron density at the Fe is independent of whether the Cp2Fe moiety is ligated to one (as in the case of the “outer” atom) or to two P atoms. This observation is consistent with the earlier observations that sigma-bonded substituents on the Cp ring have relatively little effect on the s-electron density at Fe [25], [26], [27]. In this connection it is to be noted (Table 1) that the calculated charges based on B3LYP/6-311G calculations show a slightly larger charge on the “inner” Fe atoms than the “outer”, consistent with the IS parameters referred to above. The temperature-dependence of the IS parameter permits an approximate calculation of the Meff parameter [28]. The difference between these values for the two Fe atoms in 6 and the “bare” iron mass of 57 Da is due to the covalent ligation of the metal atoms in the present structure.

Fig. 2

Mößbauer spectra of neat 6 at 172 K.

From the average temperature dependence of the IS (−dIS/dT) and the average temperature dependence of the logarithm of the recoil-free fraction (−dln A/dT) as summarized in Table 2, it is possible to calculate a “Mössbauer Lattice Temperature”, ΘM, of 75 ± 5 K, indicating that with respect to the metal atom dynamics, 6 provides a relatively “soft” environment for the Fe atoms in this structure. The ME parameters and derived quantities are summarized in Table 2.

Table 2

Mössbauer hyperfine and derived parameters for 6.

Parameter	Fe1 (“outer”)	Fe2 (“inner”)	Units
IS(90)	0.514(5)	0.505(9)	ms−1
QS(90)	2.365(5)	2.169(9)	ms−1
−dIS/dT	2.31(17)	2.42(34)	×10−4 mm s−1 K−1
Meff	180 ± 12	172 ± 20	Daltons
–dln[A(T)/A(90)]/dT	7.97(62)a		×10−3 K−1
Area ratio	1.03(1)b

Correlation coeff. 0.997 for 8 data points for the “outer” Fe atom.

96 < T < 240 K.

As a further experimental probe for the electronic situation of the ferrocene units in 6 also electrochemical measurements (CV) have been performed in acetonitrile solution using Bu4NBF4 as supporting electrolyte. Two overlapping waves at ca. +0.27 V (vs. FcH0/+) separated by ΔE1/2 = 270 mV can be observed in a 2:1 ratio. Unlike the chalkogen bridged oligoferrocenylenes where central and terminal ferrocene units show separate redox waves [4], [29], [30], the different ferrocene units in 6 show only tiny differences in their potentials. Therefore, a selective oxidation of either terminal or central ferrocene units employing a chemical oxidant seems difficult for 6. Nevertheless such mixed valent species would be attractive synthetic targets. Exploratory attempts to oxidize 6 with one or two equivalents of iodine in solution resulted in the formation of a dark insoluble solid material which we were unable to characterize. For well-defined mixed valent species of this type it will be necessary to modify the different types of ferrocene units in 6 electronically for instance via additional substituents. Following this approach it may be feasible to separate the redox potentials of the terminal and the central ferrocene units which may open the way for selective oxidation at only the inner or outer ferrocene units of the molecule.

Conclusions

In summary, we have developed a synthetic route to an oligomer in which three ferrocene units are symmetrically connected by phosphaalkene units. The charge distribution of the molecule have been analyzed using DFT calculations which indicate that the iron atom of the central metallocene unit is slightly more positive than the terminal ones. These findings are supported experimentally by Mößbauer spectroscopy in a solid glass at low temperature and cyclic voltammetry in solution. Interestingly the partial charges of the metal atoms are hardly affected by the E/Z configuration of the phosphaalkene units based on our calculations. In future work we will try to separate the overlapping potentials of the ferrocene units via electron withdrawing/releasing substituents aiming at well-defined mixed valent oligomers.

Experimental

Reactions are carried out under argon atmosphere using modified Schlenk techniques or in a glovebox. Solvents are degassed and dried using Puresolv solvent purification system. n-BuLi, diethylamine are purchased from Sigma Aldrich and used as received. PCl3 – from Sigma Aldrich – is distilled prior to use. FcPH2 (1) was prepared according to a modified procedure by Roesky et al. [31] starting from FcPCl2 [32]. 1,1′-Fc-(COCl)2 was prepared from 1,1′-Fc-(COOH)2 following a literature procedure [33]. NMR spectra were recorded with a Varian UnityInova 400 and a BrukerAvance 300 operating at a proton frequency of 399.983 MHz and 300.132 MHz, respectively. Chemical shifts are given in ppm, referenced externally to TMS, and H3PO4 (85%), for 1H, 13C, and 31P, respectively. Mass spectroscopy was performed on Agilent 5975C mass spectrometer equipped with a direct injection unit (DI-EI) or a Kratos MS-50 (DI-EI, HRMS). UV–vis spectra were recorded on a Varian Cary 50 conc. UV–vis spectrometer in quartz cells.

Synthesis of FcPHTMS (4)

A stirred solution of FcPH2 (1.31 g, 6 mmol) in diethylether (40 ml) is cooled to 0 °C and a solution of n-BuLi in hexanes (4.13 ml, 6.6 mmol) is added slowly via syringe. The temperature is maintained for 2 h after which the solution is slowly warmed to room temperature. The resulting deep red solution is slowly added to a cooled (0 °C) solution of an excess of freshly distilled TMSCl (1.5 ml) in diethylether (10 ml). Upon addition the mixture is warmed to room temperature and the volatile components are removed in vacuum. The residue is extracted with toluene (15 ml) three times and the solids are removed by filtering through a fritted funnel. Removal of the solvent in vacuum yields 7 as viscous oil as a mixture with 5 and FcPH2. Comparison with spectra of authentic samples of these other components allows the assignment of the spectral data of 4. Based on integration the yield of 4 is 59%. 31P NMR (C6D6): −149.2 (d, 1JPH = 200 Hz). 1H NMR (C6D6): 0.22 (dd, 3JHP = 4.3 Hz; 4JHH = 1.5 Hz), 3.19 (d, 1JHP = 200.3 Hz, I H, PH, 9 H, TMS), 4.12/4.07 (m, 2 H, Cp), 4.15 (m, 2 H, Cp), 4.18 (s, 5 H, Cp). 13C NMR (C6D6): 0.3 (d, 2JCP = 10.3 Hz, TMS), 70.2 (s, Cp), 70.7 (d, 3JCP = 7.2 Hz, Cp), 75.0 (d, 2JCP = 31.6 Hz, Cp). MS(EI, m/z): M+, 290.0344 (290.0343 calcd. for C13H19FePSi).

Synthesis of FcPTMS2 (5)

A stirred solution of FcPH2 (1.31 g, 6 mmol) in diethylether (40 ml) is cooled to 0 °C and a solution of n-BuLi in hexanes (8.25 ml, 13.2 mmol) is added slowly via syringe. The temperature is maintained for 2 h after which the solution is slowly warmed to room temperature. The resulting deep red solution is slowly added to a cooled (0 °C) solution of an excess of freshly distilled TMSCl (1.5 ml) in diethylether (10 ml). Upon addition the mixture is warmed to room temperature and the volatile components are removed in vacuum. The residue is extracted with toluene (15 ml) three times and the solids are removed by filtering through a fritted funnel. Removal of the solvent in vacuum yields 5 as viscous oil (2.17 g, 94%). 31P NMR (C6D6): −167.0 (s). 1H NMR (C6D6): 0.38 (d, 3JHP = 4.4 Hz, 18 H, TMS), 4.16 (m, 4 H, Cp), 4.19 (s, 5 H, Cp). 13C NMR (C6D6): 2.35 (d, 2JCP = 12.2 Hz, TMS), 69.59 (d, 3JCP = 2.7 Hz, C), 70.27 (s, Cp′), 73.72 (d, 2JCP = 8.8 Hz, C), 76.23 (d, 1JCP = 14.1 Hz, Cipso). MS(EI, m/z): M+, 362.0751 (362.0738 calcd. for C16H27FePSi2), 40%; 290, FcPHTMS+ (55%); 218, FcPH2+ (21%); 121, CpFe+ (35%); 73, TMS+ (100%).

Synthesis of [Fc-PC(OTMS)–]2Fc′ (6)

To a cooled (−30 °C) solution of 1,1′-Fc-(COCl)2 (154 mg, 0.5 mmol) in pentane (15 ml) two equivalents of Fc-P(TMS)2 (5) (360 mg, 1 mmol) are slowly added. Stirring is maintained at −30 °C for an hour and subsequently at room temperature over night. All volatiles are removed in vacuum and the product is extracted from the residue with pentane. After removal of the solvent at reduced pressure (6) is obtained as viscous oil in 55% yield (224 mg).

31P NMR (C6D6): 124.3 (s (broad) main isomer), 127.4 (s (broad) minor isomer). 1H NMR (CDCl3): 0.04 (s, OTMS, minor isomer), 0.09 (s, OTMS, main isomer), 4.13 (s (broad), Cp), 4.16 (s, Cp), 4.36 (s (broad), Cp), 4.51 (s (broad), Cp), 4.87 (s (broad), Cp).13C NMR (C6D6): 0.4 (OTMS), 0.7 (OTMS), 68.4–69.2 (br. Cp), 77.3–78.2 (br. Cp). EI-MS [m/z]: 818.0 M+ (7%), 746.1 Fc-PH–CO-Fc-PC-Fc+ (15%), 674.1 Fc-PH–CO-Fc–CO–PH-Fc+ (10%), 73.0 TMS (100%). UV–vis (λmax, pentane): 530, 542 and 553 nm. Elemental analysis (%) for C38H44Fe3O2P2Si2: C 55.77, H 5.42; found: C 54.60, H 5.16. The lower C, H values can be explained by additional LiCl from previous steps; hydrolysis and loss of TMS groups by contrast would lead to increased C, H values.

CV

CV spectra are recorded on a Gamry Reference 600 device in CH3CN solution (Bu4NBF4 0.06 M supporting electrolyte) of the respective substance under argon atmosphere. A standard three electrode system (Ag+/Ag-reference electrode; a Pt-auxiliary wire and a Pt-working electrode [Ø 3 mm]), referenced vs. Fc0/+.

Temperature-dependent ME spectroscopy

ME spectroscopy in transmission geometry was carried out, using a 50 mCi 57Co source (Rh) as described previously (see references above). Velocity calibration was effected using an α-Fe absorber foil at room temperature, and all IS are referenced to the centroid of such spectra. Temperature control and monitoring was effected using the Daswin program of Glaberson (www.megadq.com). Samples were transferred in O-ring sealed sample holders, protected in an inert atmosphere, and stored under liquid nitrogen until examined by ME methods.

Computational details

Calculations have been carried out with the G03 suite of programs (Rev. B.04) using a triple-ζ-basis set (6-311G(d)) and the widely used B3LYP functional [34]. All structures were confirmed as true minima by their frequency analysis having no imaginary frequency. TD-DFT calculations have been performed at the same level of theory considering up to 40 states, also including singlet and triplet states. Orbitals are visualized using the ChemCraft program and GausSum for creation of UV spectra [35]. Additional information for the computed structures of EE-6, EZ-6 and ZZ-6 (Cartesian coordinates) and their electronic transitions and selected frontier orbitals are available as electronic Supplementary material.