Probing the magnetizability distribution of ferrocene as determined via anisotropic contributions to the NMR shielding and its application to several substituted ferrocenes

Abstract

Molecular magnetic anisotropies of ferrocene derivatives are derived from solution-state measurements of the Cotton–Mouton and Kerr constants. A calculation is presented of chemical shifts induced at the hydrogen and carbon nuclei, arising from the experimental magnetic anisotropy. Ferrocene, ruthenocene and analogously substituted benzenes were also examined. Comparisons of the chemical shifts were used to probe the effects of metal–carbon bonding in the metallocenes. The experimental chemical shifts for pentamethylferrocene have been interpreted, allowing for magnetic anisotropy contributions, to give carbon atom charges that are in close agreement with predictions from the measured electric dipole moment.

---

Following recent work¹ in this series,^1–6 the mean magnetizability (or susceptibility), χ, for a number of substituted ferrocenes was investigated in terms of a perturbation, λ, to the metal–ring bonding introduced to bring the experimental and predicted results into agreement. It was considered that a more sensitive probe of the magnetic environment of the ferrocenyl nucleus would be the anisotropy in the magnetizability, Δχ, rather than the λ corrections to the isotropic value. Considerable electron reorganization is known to occur along the metal–ring axis in ferrocene and ruthenocene,⁶ M(η⁵-C₅H₅)₂, M=Fe or Ru, respectively, following coordination. The purpose of this paper is to explore the magnetic anisotropy and other physical properties of both free and complexed rings in these systems. One such property is the NMR chemical shift, which for the ring nuclei of the cyclopentadienyl rings, displays a significant shielding contribution compared to equivalently substituted benzenes. This shielding can arise from a number of causes, including diamagnetic shielding from the metal or diminution of the classic aromatic ring current.

Experimental

The solute molar Cotton–Mouton constant at infinite dilution, _∞(_mC₂), from which the solute magnetizability anisotropy can be derived, is obtained from experimental observables by eqn. (1)^6,7

_∞(_mC₂)=_sC₁(1−β−Fγ+δ^⊥)M₂ (1)

where subscripts 1 and 2 refer in turn to solvent and solute; _sC₁ is the specific Cotton–Mouton constant of the solvent; β and γ are derived from the incremental densities and refractive indices of the solutions of graded concentration,⁸ which have been presented earlier;^6,9δ^⊥ derives from the incremental Cotton–Mouton constants for the solutions;¹⁰F is a solvent constant;¹¹ and M₂ is the molecular weight of the solute. Some of these parameters for the metallocenes studied here are collected in Table 1.

Table 1 Infinite-dilution molar Cotton–Mouton constant, _∞(_mC₂), and the magnetizability anisotropy, Δχ, of metallocenes as solutes in cyclohexane^a at 298 K and 632.8 nm

Metallocene	10^15(ΔC12/w2)w2=0^b/m A^−2	δ^⊥	10^27∞(mC2)w2=0/m^5 A^−2 mol^−1	10^29 Δχ/J T^−2 molecule^−1
a Reliable measurements could not be made for decachloroferrocene due to solubility limitations.b w2 refers to the solute weight fraction.c Data from ref. 6.
Fe(η^5-C5Me5)2	2.25±0.09	−12.9±0.5	33.4±1.4	77.6±3.6
Fe(η^5-C5H5)(η^5-C5Me5)	−2.52±0.21	14.5±1.2	−32.1±2.8	−101±11
Fe(η^5-C5H5)2^c	−4.33±0.24	24.9±1.4	−39.3±2.1	−92.2±0.6
Fe(η^5-C5H5)(η^5-C5Cl5)	−1.07±0.26	6.2±1.5	−19.5±5.2	−140±40
Ru(η^5-C5H5)2^c	−2.06±0.22	11.8±1.3	−23.5±2.4	−53.3±6.0

---

The substituted ferrocenes were purified² by vacuum sublimation and/or recrystallization. Solutions were prepared and manipulated under nitrogen. Cyclohexane, an inert solvent for such studies, was purified and dried by distillation under nitrogen from phosphorus pentoxide.

The NMR chemical shifts of the metallocenes were recorded on either a Joel FX-60Q, XLFT-100 or a Bruker WM-400 spectrometer; all chemical shifts are referenced to TMS. The measured shifts (δ) for the metallocenes and reference benzene compounds are given in Table 2.

Table 2 ¹H and ¹³C NMR solution-state chemical shifts, δ, of the metallocenes studied^a

Molecule	Nucleus	^13C shift/ppm^b	^1H shift/ppm
a In C6D12 solvent except as noted otherwise.b The ^13C shifts, apart from those of Ru(η^5-C5H5)2 and C6Me6, are from ref. 9.c In C6D6 solvent.
Fe(η^5-C5Me5)2	*C5Me5	78.73	—
Me	9.84	1.64
Fe(η^5-C5H5)(η^5-C5Me5)	*C5Me5	80.29	—
C5H5	71.53	3.55
Me	11.66	1.87
Fe(η^5-C5H5)2	C5H5	68.28	4.04
Fe(η^5-C5H5)(η^5-C5Cl5)	C5Cl5	88.18	—
C5H5	79.64	4.26
Fe(η^5-C5Cl5)2^c	C5Cl5	90.56	—
Cr(η^6-C6H6)2	C6H6	74.90	4.26
Cr(η^6-C6H6)2^c	C6H6	74.93	4.213 (T=323 K)
4.223 (T=298 K)
4.227 (T=278 K)
Ru(η^5-C5H5)2	C5H5	70.42	4.43
C6H6	C6H6	128.79	7.16
C6Me6	*C6Me6	131.78	—
Me	16.68	2.14
C6Cl6	C6Cl6	133.15	—

---

Theoretical

For the axially symmetric molecules investigated, each of which has a threefold or higher-order rotation axis (subscripted z), the classical statistical mechanical expression for the Cotton–Mouton constant reduces to:^10,12

where N_A, k_B, ε₀, μ₀ and T have their usual meanings,²Δα(=α_zz−α_xx) and Δχ(=χ_zz−χ_xx) are the anisotropies in the optical polarizability and susceptibility, respectively. The tensor Δη, describing the temperature-independent contribution to the Cotton–Mouton constant, has been shown to be negligible for such highly anisometric systems.^12–14

To obtain the magnetizability anisotropy from _mC₂, the sign of the polarizability anisotropy must be known with reference to molecular orthogonal axes. Earlier analysis of the polarizabilities of methylated ferrocenes⁹ enabled the sign of Δα to be specified for decamethylferrocene, Fe(η⁵-C₅Me₅)₂. For ferrocene and ruthenocene, the sign of Δχ had earlier been established from crystal state studies.⁶

In order to develop a physical model that would provide insight into the relationship between experimentally determined Δχ values and NMR shifts, the following argument was adopted. The chemical shift is often deconvoluted into contributions arising from purely empirical components, e.g. local atomic properties or neighbouring environment factors. Of the latter, those originating from the presence of delocalized electrons or due to the presence of a magnetically anisotropic group are the most relevant. We define the term σ_Δχ to represent the contribution to the shielding arising from the magnetic anisotropy of a group or bond. This includes the magnetic field arising from electrons not centred on the atom in question or its bonding electrons. McConnell¹⁵ and Pople¹⁶ were the first to quantify this factor. For an axially symmetric group or bond, G, the magnetic point dipole induced at the origin by the randomly oriented magnetic field will produce a secondary field at some nucleus, N, remote from it. Choosing a coordinate system with the major axis along the bond direction, McConnell has advanced the following expression (in SI units)

where Δχ is the magnetic anisotropy of the group or bond G; R is the distance between the resonant nucleus and the centre of the induced magnetic dipole of G; θ is the angle between R and the symmetry axis of G. Note that for a dipolar field to be an accurate approximation, the distance R must be large in relation to the orbital radius of G, typically R>0.3 nm.

A more rigorous approach has been advanced by Stiles and co-workers.^17–19 However, for a description of σ_Δχ that is suitably tractable in relation to the available data, the treatment of ApSimon and co-workers.^20–23 was preferred. According to this scheme, the dipole is no longer considered a point, but rather is assigned a length S, which is taken to be 75% of half the bond length. This empirical choice of S was adopted for C–C and C–H together with the directional bond magnetic susceptibilities of Davies.²⁴ The non-point dipole equation in SI units is

where χ_‖ is the magnetic susceptibility of the bond along the bond axis, χ_⊥ is the magnetic susceptibility in a direction perpendicular to the bond axis and Δχ is the magnetic anisotropy, in units of J T⁻² molecule⁻¹. The distances R and S are in metres, θ is the inclination of the vector R to the axially symmetric bond axis to which χ_‖, χ_⊥ and Δχ refer. The point of origin of the induced dipole, of length S, is chosen to be the centre of the bond. This same origin determines the distance R to the nucleus being investigated. For the metallocenes the delocalized bonding between the cyclopentadienyl rings and the metal commends the metal–ring distance as the appropriate measure of the bond length. The cyclopentadienyl–metal bond, which contains the magnetically anisotropic C–C linkages as well as bonding to the electron-rich metal, would dominate the molecular anisotropy.

Hence, by using the experimentally determined molecular magnetic anisotropy and mean magnetizability to represent the delocalized metal–ring bond, a reasonable estimate of the chemical shift arising from the magnetically anisotropic metal–ring bond can be ascertained. As was applied to the electric polarizability studies⁹ of these axially symmetric molecules (i.e., α_xx=α_yy or χ_xx=χ_yy), the parallel and perpendicular magnetizability components of the metal–ring bond are derived from the following molecular relations:

where χ=(χ_zz+2χ_xx)/3 and Δχ=(χ_zz−χ_xx) for the axial symmetry present in these molecules.

Results and discussion

The NMR chemical shift has a dependence upon both the local electric field due to neigbouring dipolar bonds and to the magnetic anisotropy of the metal–ring bond.^{25 –28} Contributions arising from higher orders of the electric field are expected to be small for the metallocenes of this study.²⁹ The geometry of ferrocene is such that the protons and iron atom are on a surface at which the interaction of their atomic charges with the molecular electronic quadrupole moment³⁰ is nullified, i.e. the cone about the iron atom for which θ= cos⁻¹(1/√3)=54.7°. Due to the structural similarities, this simplification is also expected to apply reasonably well to the persubstituted ferrocenes. The carbon Cp atoms lie very close³¹ to this surface (within 18°). This configuration also means that there will be negligible screening due to a magnetic point dipole centred on the iron atom. The overall result is to make the ring carbons, protons and/or substituents, sensitive probes of the magnetic environment associated with the delocalized metal–ring bond.

The purely electronic contribution to the ¹³C NMR shielding would be determined by the charge and associated bond polarization, while effects due to the magnetic anisotropy of bonds can play a significant role for hydrogen atoms.^32,33 For carbon nuclei the effect on the chemical shift due to the contri bution arising from σ_Δχ (being independent of the nature of the atom) will be a much smaller perturbation, amounting to between 2–10% of the observed chemical shift. The results are compiled in Table 3 for both the proton and carbon nuclei of these molecules.

Table 3 The chemical shift arising from the magnetic anisotropy, σ_Δχ, of the metal–ring bond and the aromatic benzenoid bond for hydrogen and carbon nuclei

Molecule	10^29 Δχ /J T^−2 molecule^−1	10^29χxx /J T^−2 molecule^−1^a	10^29χzz /J T^−2 molecule^−1^a	S/10^−10 m^b	RH/10^−10 m (θ/deg.)^c	σΔχ (^1H)/ppm^c	RH/10^−10 m (θ/deg.)^c	σΔχ (^13C)/ppm^c
a The components χxx and χzz are calculated using eqns. (5) and (6). Values of the mean magnetizability are given in ref. 1.b The induced dipole length S is calculated from 75% of half the metal–cyclopentadienyl ring bond length for the metallocenes with the origin located at the midpoint of this bond. For the benzene compounds, S is taken as the same ratio of the ring radius (equal to the C–C bond length). The electron diffraction-derived values are used for Fe(η^5-C5Me5)2 ,^37 Fe(η^5-C5H5)2^31,37 and Ru(η^5-C5H5)2^31 since they would better reflect the solution-state structure. For Fe(η^5-C5H5)(η^5-C5Me5) the crystallographic values derived earlier^2 were employed.c The chemical shifts are calculated using eqn. (4) and published geometries^2,31,37–39 where available, or assumed values as outlined in the following table notes d and e. The quoted uncertainties are based on the errors in Δχ and χ. The symbols Me and Cp refer to the methyl and cyclopentadienyl ring protons respectively, while the asterisked carbon refers to the resonant nucleus.d The ferrocene geometry is assumed in Fe(η^5-C5H5)(η^5-C5Cl5).e In this model the geometry of the C5H5–Fe system is taken to be the same as that of C5H5–Fe of pentamethylferrocene, while that of C5–Fe in C5Cl5–Fe is analogous to that of C5Me5–Fe in pentamethylferrocene.f Ref. 40.g Ref. 41.h Ref. 42.i Ref. 43.
Fe(η^5-C5Me5)2	77.6±3.6	−392±8	−314±9	0.6233	3.410 (74.2)	Me: 1.00±0.04	1.479 (55.8)	*C5Me5: −1.46±0.19
2.866 (71.3)	Me: 1.74±0.06
Fe(η^5-C5H5)(η^5-C5Me5)	−101±11	−253±12	−354±14	0.6236	2.249 (68.9)	Cp: −0.10±0.28	1.452 (55.1)	C5H5: −6.91±0.43
0.6154	3.169 (74.2)	Me: −0.34±0.11	1.451 (55.6)	*C5Me5: −6.47±0.43
0.6154	2.839 (72.5)	Me: −0.34±0.15
Fe(η^5-C5H5)2	−92.2±0.6	−179±5	−271±5	0.6229	2.309 (69.7)	Cp: −0.44±0.16	1.452 (55.6)	C5H5: −5.39±0.26
Fe(η^5-C5H5)(η^5-C5Cl5)	−140±40	−350±18	−490±29	0.6229^d	Cp: −0.16±0.86	C5H5, C5Cl5: −9.2±1.4
0.6236^e	Cp: −0.15±0.90	C5H5: −9.6±1.3
0.6154^e	C5Cl5: −9.0±1.3
Ru(η^5-C5H5)2	−53.3±6	−280±6	−334±7	0.6837	2.524 (68.8)	Cp: 0.60±0.11	1.526 (53.3)	C5H5: −5.82±0.21
C6H6	−106±4^f	−55.5±1.7^g	−161.5±2.8	0.5239	2.481 (90)	C6H6: −1.78±0.09	1.397 (90)	C6H6: −3.55±0.67
C6Me6	−136±7^f	−157±10^h	−294±11	0.5213	3.084 (90)	Me: −1.12±0.11	1.39 (90)	*C6Me6: 5.95±2.2
2.93 (90)	Me: −1.25±0.13
C6Cl6	−144±10^f	−194±12^i	−338±14	0.5224	1.393 (90)	C6Cl6: 9.5±2.8

---

The difference between the magnetic anisotropies of free and complexed rings can be probed by examining changes in σ_Δχ between similarly substituted ferrocene and benzene molecules. The latter represents a close, if imperfect, approximation to the free uncoordinated C₅H₅ ligand. For the benzene molecules the magnetic origin is taken to be the centre of the ring, while the induced dipole length is taken as 75% of half the ring radius. These results are presented in Table 4. In absolute magnitude, the largest calculated differences occur in the cyclopentadienyl framework carbon atoms, with a calculated shielding shift of from 1.8 ppm for ferrocene to 12.4 ppm for the methylated carbons of pentamethylferrocene, Fe(η⁵-C₅H₅)(η⁵-C₅Me₅), relative to the carbons of the free benzene ring. Since the ferrocenyl carbons are directly bonded to the iron, it is expected that they would be most affected by the magnetic anisotropy of the metal–ring bond. Examinin g the calculated chemical shift differences Δσ_Δχ, given in Table 4, we observe a relatively large variation upon complexation.

Table 4 Differences, Δ, in the predicted shielding shift, σ_Δχ, and the experimental chemical shifts, δ, between the ferrocenes and the analogous benzene, hexamethylbenzene or hexachlorobenzene molecules

Metallocene	Benzene	Differences^a/ppm	H in Me	CnHn^*^b	C in Me	Cn^*Men	Cn^*Hn	Cn^*Cln
a Δ refers to the difference given by the ferrocene shift minus the corresponding benzene shift; e.g. from Table 3, for Me(H) in Fe(η^5-C5Me5)2 relative to C6Me6, ΔσΔχ=1.00−(−1.12) ppm=2.12 ppm. Δδ is obtained in an analogous manner from the δ values in Table 2. The error in Δδ for protons is ±0.01 ppm, while for carbon it is ±0.05 ppm; ^* refers to the resonant nucleus.b n=5 refers to the cyclopentadienyl ring while n=6 refers to the benzene ring.c The geometry based on the pentamethylferrocene molecule is used for these calculations.d For this molecule the Cotton–Mouton constant could not be determined due to solubility limitations.
Fe(η^5-C5Me5)2	C6Me6	ΔσΔχ	2.12±0.12	—	2.99±0.14	−7.41±2.2	—	—
Δδ	−0.50	—	−6.84	−53.05	—	—
Fe(η^5-C5H5)(η^5-C5Me5)	C6H6, C6Me6	ΔσΔχ	0.78±0.14	1.68±0.29	0.91±0.20	−12.4±2.2	−3.36±0.8	—
Δδ	−0.27	−3.61	−5.02	−51.49	−57.26	—
Fe(η^5-C5H5)2	C6H6	ΔσΔχ	—	1.34±0.18	—	—	−1.84±0.72	—
Δδ	—	−3.12	—	—	−60.51	—
Fe(η^5-C5H5)(η^5-C5Cl5)	C6H6, C6Cl6 ^c	ΔσΔχ	—	1.63±0.90	—	—	−6.05±1.46	−18.5±3.1
Δδ	—	−2.90	—	—	−49.15	−44.97
Fe(η^5-C5Cl5)2	C6Cl6	ΔσΔχ	—	—	—	—	^d	—
Δδ	—	—	—	—	—	−42.59

---

When the Δσ_Δχ differences are compared to the experimental differences (Δδ) for the ¹³C resonances in Cp, which range from ca. 49 to 60 ppm, it is apparent that some other factor, apart from any magnetic anisotropy dependent shift, causes the major part of the shielding in the complexed species.

It is known, however, that eqns. (3) and (4) are valid within a series of homologous compounds.³⁴ To examine the validity of our approach, differences between the chemical shifts within the ferrocene derivatives themselves were also obtained and are presented in Table 5. In this Table the differences between calculated and experimental shifts are much smaller and they are of comparable magnitude. For the pentasubstituted molecules, the presence of internal metal–ligand dipole moments of opposite orientation with regard to the protonated cyclopentadienyl ring may generate chemical shifts arising from an electric field contribution that have not been quantified in this treatment.

Table 5 Differences, Δ′, in the predicted shielding shift, σ_Δχ, and the experimental chemical shifts, δ, between the pentasubstituted ferrocenes and the analogous persubstituted or unsubstituted ferrocene.

Ferrocene 1	Ferrocene 2	Differences^a/ppm	H in Me	C5H5^*	C in Me	C5^*Me5	C5^*H5	C5^*Cl5
a Δ′ refers to the difference given by the Ferrocene 1 shift minus the corresponding Ferrocene 2 shift (both from Table 3). The error in Δ′δ for protons is ±0.01 ppm, while for carbon it is ±0.05 ppm.
Fe(η^5-C5H5)(η^5-C5Me5)	Fe(η^5-C5Me5)2, Fe(η^5-C5H5)2	Δ′σΔχ	−1.34±0.12	0.34±0.32	−2.08±0.16	−5.01±0.47	−1.52±0.50	—
Δ′δ	0.23	−0.49	1.82	1.56	3.24	—
Fe(η^5-C5H5)(η^5-C5Cl5)	Fe(η^5-C5Cl5)2, Fe(η^5-C5H5)2	Δ′σΔχ	—	0.3±0.9	—	—	−4.2±1.3	—
Δ′δ	—	0.22	—	—	11.35	−2.38

---

An independent check is required to test the validity of the model presented. In assessing the contribution of an electric field arising from intramolecular charge separation to the chemical shift of a ring carbon atom, we must consider the factors affecting the observed resonance. As has been indicated by the significant differences between Δσ_Δχ and Δδ in Table 4, large shielding contributions arise from bonding to the metal. These factors will be dependant on: (i) the metal orbitals with which the ring carbon interacts, (ii) the nature of the atom bonded to the ring carbon, i.e. proton, carbon or chlorine, (iii) the contribution from the magnetic field of the metal–ring bond and (iv) any interannular charge displacement.

A comparison is made of intramolecular charge distributions from experimental dipole moments with those obtained by deconvoluting the ¹³C chemical shifts (δ) into component contributions according to eqn. (7):

δ=σ_Δχ+σ_Electronic+σ_Framework (7)

where δ is the experimentally determined NMR chemical shift, σ_Δχ, is given in Table 3 and equates to point (iii), σ_Electronic is the screening effect of molecular electronic charges as covered in point (iv), and σ_Framework covers the screening of the ring carbon arising from bonding to both the iron atom and the ring substituent. The σ_Framework term pertains to the electronic backbone of the molecule and encompasses both points (i) and (ii) above. For example, it can be accounted for in pentamethylferrocene by comparing the ¹³C resonances of this molecule with those of ferrocene and decamethylferrocene.

Due to the symmetry and planarity of the Cp ring and the ring substituents of ferrocene, pentamethylferrocene and decamethylferrocene, the experimental dipole moment³ can be taken to directly reflect the charge difference between the protonated and methylated rings of pentamethylferrocene. A similar approach could not be applied to the chlorinated series since the structures could not be determined†; the unavailability of the Cotton–Mouton constant for decachloroferrocene, Fe(η⁵-C₅Cl₅)₂, also prevented the determination of σ_Δχ for this molecule. The following argument is restricted to ferrocene and the methylated ferrocenes.

We define a corrected chemical shift, δ′, as the difference between the experimental NMR shift, δ, and the magnetic anisotropy contribution, σ_Δχ. The difference Δδ′ is that between δ′ of corresponding C₅H₅ or C₅Me₅ rings of pentamethylferrocene, decamethylferrocene and ferrocene resonances. The quantity Δδ′ should then represent the electric field effect σ_Electronic upon the chemical shift for each type of ring. The final difference between the Δδ′ of the protonated and methylated rings will directly report on the charge difference arising from any interannular charge donation. Using the proportionality constant of −160 ppm electron⁻¹,^32,35 the Δδ′ in Table 6 will translate to an equivalent electric charge of 0.011±0.004 electrons per carbon atom.

Table 6 Interannular charge separation and the resultant shielding of the ring carbon atoms of Fe(η⁵-C₅H₅)₂, Fe(η⁵-C₅Me₅)₂ and Fe(η⁵-C₅H₅)(η⁵-C₅Me₅)

Metallocene	Ring	δ^a /ppm	σΔχ^b /ppm	δ′^c /ppm	Δδ′^d /ppm	Δδ′(C5H5)−Δδ′(C5Me5)^e
a δ refers to the experimental chemical shift given in Table 2.b σΔχ is given in Table 3.c δ′ is defined to equal δ−σΔχ for either the C5H5 or C5Me5 ring systems.d Δδ′ is the difference between the pentamethylferrocene ring δ′ and the corresponding homoannular ferrocene δ′ value.e A proportionality constant of −160 ppm per electron was used to convert the magnetic resonance shift to a corresponding electric charge.
Fe(η^5-C5H5)2	C5H5	68.28	−5.39	73.67
4.77±0.50 (C5H5)
Fe(η^5-C5H5)(η^5-C5Me5)	C5H5	71.53	−6.91	78.44	−1.80±0.69 ppm
Fe(η^5-C5H5)(η^5-C5Me5)	C5Me5	80.29	−6.47	86.76	=0.011± 0.004 electrons
6.57±0.47 (C5Me5)
Fe(η^5-C5Me5)2	C5Me5	78.73	−1.46	80.19

---

This means that the ring carbon atoms are experiencing a chemical shift arising from a charge of 0.0112±0.0043 electrons. Now turning to the dipole moment data³ and the crystal structure² it is found that the assigned charges, required to reproduce the measured dipole moment of (3.00±0.07)×10⁻³⁰ C m, are not sensitive to the charge locations within the rings of the molecule. Thus, by fixing a single point charge in the ring or partitioning the charge between the ring carbon atoms, we determine a total charge of 0.056±0.001 electrons per ring. Assigning this charge to the ring carbons results in an average ring carbon charge of 0.0112±0.0003 electrons per ring carbon atom, in excellent accord with the value determined from the corrected ¹³C resonance differences.

The outcome of this analysis is that the model presented for the σ_Δχ determinations is self-consistent and reproduces the magnetic shielding fairly accurately for distances of the order of several bond lengths. Furthermore, the sense of the dipole moment is determined from the chemical shift analysis, such that the negative end is located in the C₅H₅ ring of pentamethylferrocene and the positive end lies in the C₅Me₅ ring. This is in accord with the hyperconjugative character of alkyl groups,³⁶ feeding electronic charge into the conjugated aromatic system across to the C₅H₅ ring.

Acknowledgements

Helpful discussions with Associate Professors F. Separovic and M. K. Cooper on the general trend of the NMR parameters and the syntheses, respectively, are gratefully acknowledged. Thanks must go to Professor N. J. Hush for discussing the electric field dependence, to Professor G. L. D. Ritchie for allowing the use of the Cotton–Mouton effect apparatus, and especially to Dr B. Cornell for support and the use of computing facilities.

References

1. L. Phillips, J. Chem. Soc., Dalton Trans., 1994, , 3257.
2. L. Phillips, A. R. Lacey and M. K. Cooper, J. Chem. Soc., Dalton Trans., 1988, 1383.
3. L. Phillips and S. W. Filipczuk, J. Mol. Liq., 1991, 47, 261.
4. S. W. Filipczuk and L. Phillips, J. Mol. Liq., 1994, 59, 13.
5. K. N. Brown, P. T. Gulyas, P. A. Lay, N. S. McAlpine, A. F. Masters and L. Phillips, J. Chem. Soc., Dalton Trans., 1993, 835.
6. G. L. D. Ritchie, M. K. Cooper, R. L. Calvert, G. R. Dennis, L. Phillips and J. Vrbancich, J. Am. Chem. Soc., 1983, 105, 5215.
7. R. J. W. Le Fèvre, P. H. Williams and J. M. Eckert, Aust. J. Chem., 1965, 18, 1133.
8. C. G. Le Fèvre and R. J. W. Le Fèvre, in Techniques of Chemistry, ed. A. Weissberger, Wiley–Interscience, New York, 1972, Pt. IIIc, vol. 1, ch. 6, p. 399..
9. L. Phillips and G. R. Dennis, J. Chem. Soc., Dalton Trans., 1995, 1469.
10. M. P. Brereton, M. K. Cooper, G. R. Dennis and G. L. D. Ritchie, Aust. J. Chem., 1981, 34, 2253.
11. M. R. Battaglia and G. L. D. Ritchie, J. Chem. Soc., Faraday Trans. 2, 1977, 73, 209.
12. P. B. Lukins, A. D. Buckingham and G. L. D. Ritchie, J. Phys. Chem., 1984, 88, 2414.
13. A. D. Buckingham, W. H. Prichard and D. H. Whiffen, Trans. Faraday Soc., 1967, 63, 1057.
14. H. Kling, E. Dreier and W. Hüttner, J. Chem. Phys., 1983, 78, 4309.
15. H. M. McConnell, J. Chem. Phys., 1957, 27, 226.
16. J. A. Pople, Proc. R. Soc. London, A, 1957, 239, 550.
17. A. D. Buckingham and P. J. Stiles, Mol. Phys., 1972, 24, 99.
18. J. W . Perram and P. J. Stiles, Proc. R. Soc. London A, 1976, 349, 125.
19. P. J. Stiles, Chem. Phys. Lett., 1975, 30, 259.
20. J. W. ApSimon, W. G. Craig, P. V. Demarco, D. W. Mathieson, L. Saunders and W. B. Whalley, Chem. Commun., 1966, , 359.
21. J. W. ApSimon, W. G. Craig, P. V. Demarco, D. W. Mathieson, A. K. G. Nasser, L. Saunders and W. B. Whalley, Chem. Commun., 1966, , 754.
22. J. W. ApSimon, W. G. Craig, P. V. Demarco, D. W. Mathieson, L. Saunders and W. B. Whalley, Tetrahedron, 1967, 23, 2339 and 2357.
23. J. W. ApSimon, P. V. Demarco, D. W. Mathieson, W. G. Craig, A. Karim, L. Saunders and W. B. Whalley, Tetrahedron, 1970, 26, 119.
24. D. W. Davies, Mol. Phys., 1963, 6, 489.
25. R. J. Pugmire and D. M. Grant, J. Am. Chem. Soc., 1968, 90, 697.
26. G. L. Nelson, G. C. Levy and J. D. Cargioli, J. Am. Chem. Soc., 1972, 94, 3089.
27. V. Graves and J. J. Lagowski, Inorg. Chem., 1976, 15, 577.
28. J. Evans and J. R. Norton, Inorg. Chem., 1974, 13, 3042.
29. J. G. Batchelor, J. H. Prestegard, R. J. Cushley and S. R. Lipsky, J. Am. Chem. Soc., 1973, 95, 6358.
30. A. D. Buckingham, in Physical Chemistry. An Advanced Treatise, ed. H. Eyring, D. Henderson and W. Yost, Academic Press, New York, 1970, vol. 4, ch. 8, p. 349..
31. A. Haaland and J. E. Nilsson, Acta Chem. Scand., 1968, 22, 2653.
32. H. Spiesecke and W. G. Schneider, Tetrahedron Lett., 1961, , 468.
33. P. C. Lauterbur, J. Am. Chem. Soc., 1961, 83, 1838.
34. H. Spiesecke and W. G. Schneider, J. Chem. Phys., 1961, 35, 722.
35. G. A. Olah and G. D. Mateescu, J. Am. Chem. Soc., 1970, 92, 1430.
36. C. A. Coulson, in Valence, ed. R. McWeeny, Oxford University Press, Oxford, 3rd edn., 1979, ch. 12, pp. 381–383..
37. A. Almenningen, A. Haaland, S. Samdal, J. Brunvoll, J. L. Robbins and J. C. Smart, J. Organomet. Chem., 1979, 173, 293.
38. P. Seiler and J. D. Dunitz, Acta Crystallogr., Sect., B, 1979, 35, 1068.
39. D. P. Freyberg, J. L. Robbins, K. N. Raymond and J. C. Smart, J. Am. Chem. Soc., 1979, 101, 892.
40. G. L. D. Ritchie and J. Vrbancich, Aust. J. Chem., 1982, 35, 869.
41. A. A. Bothner-By and J. A. Pople, Annu. Rev. Phys. Chem., 1965, 16, 43.
42. R. J. W. Le Fèvre and D. S. N. Murthy, Aust. J. Chem., 1966, 19, 179.
43. G. Foex, Constantes Sélectionées, Diamagnétisme et Paramagnétisme: Tables de Constantes et Données Numériques, Masson, Paris , 1957, ch. 1..

---

Footnote

† Several unseccessful attempts were made to determine X-ray crystal and electron diffraction structures in our own and other laboratories.

---