Quantitative and Mechanistic Mechanochemistry in Ferrocene Dissociation

Abstract

Ferrocene is classically regarded as being highly inert owing to the large dissociation energy of metal-cyclopentadienyl (Cp) bonds. We show that the Fe-Cp bond in ferrocene is the preferential site of mechanochemical scission in the pulsed ultrasonication of main-chain ferrocene-containing polybutadiene-derived polymers. Quantitative studies reveal that the Fe-Cp bond is similar in strength to the carbon-nitrogen bond of an azobisdialkylnitrile (bond dissociation energy < −0 kcal/mol), despite the significantly higher Fe-Cp bond dissociation energy (approximately 90 kcal/mol). Mechanistic studies are consistent with a predominately heterolytic mechanism of chain scission. DFT calculations provide insights into the origins of ferrocene’s mechanical lability.

Graphical Abstract

Since ferrocene was discovered in the early 1950s, its captivating structure, reactivity and properties have led it to become one of the most classical organometallic compounds to date.^1–4 Ferrocene’s broad utility is due in large part to the stability of the iron-cyclopentadienyl (Fe-Cp) bond, the bond dissociation energy (BDE) of which (reported up to be 91 kcal/mol) approaches that of conventional covalent C-C bonds.⁵ As such, the ferrocene core is often assumed to be inert, enabling its utility in a wide range of material, electrochemical, and catalytic applications.^{4, 6–13}

Inspired by recent work in covalent polymer mechanochemistry,^14–22 we wondered whether the Fe-Cp coordination bond could be broken efficiently and selectively through the application of mechanical force. And if so, what is the mechanism of the mechanochemical dissociation, and how does the mechanical strength of the organometallic Fe-Cp bond compare to that of organic covalent bonds of similar BDE? Not surprisingly, the allure of ferrocene as a potential mechanophore has captured the attention of others as well. Giannantonio et al. recently reported that main-chain ferrocene acts as a preferred site of mechanochemical chain scission in polyurethane and polyacrylate polymers, leading to the release and oxidation of iron to Fe³⁺.²³ Here, we report our own studies of ferrocene mechanochemistry and demonstrate that ferrocene is an efficient mechanophore in polybutadiene- derived polymers as well. We reasoned that the further development of this promising new class of mechanophores will benefit from quantitative mechanistic insights, and so we additionally set out to answer to several open questions: (1) Whether ferrocene dissociation might lead to the release and capture Fe²⁺ rather than Fe³⁺ as observed by Giannantonio et al.? (2) What is the fate of the dissociation Cp ligand? (3) How does the mechanical strength of the Fe-Cp bond compare to that of other mechanically labile bonds? (4) More generally, what is the mechanism of the mechanochemical dissociation? Is it heterolytic or homolytic? A combination of experimental and computational techniques is used to probe these questions, and the results suggest strategies for expanding the utility of this promising class of multifunctional metallomechanophores.

Multi-mechanophore polymers are known to facilitate the characterization and quantification of mechanophore reactivity, in ways that provide advantages over a single mechanophore labeling strategy,^23–25 and we followed a well-established methodology of using entropy driven ring-opening metathesis polymerization (ED-ROMP) to synthesize polymers of varying molecular weight and mechanophore content (Table 1; see SI for synthetic details).^{26, 27}

Table 1.

Polymers used in this study

Polymer	Architecture	Mn (Da)	Đ	Labeling molar ratio
P1		111,500	2.1	2%
P2		90,400	2.2	3%
P3		75,000	1.6	--
P4		97,000	1.6	10%
P5		92,000	1.7	9%
P6		75,000	1.5	8%
P7		80,000	1.6	2%
P8		73,100	1.6	--

We first investigated the sono-mechanochemistry of P1 (M_n = 111,500 Da; 1 mg/mL in THF), a copolymer of 5-methoxycyclooctene and a ferrocene macrocycle 4 (Figure 1b). The molecular weight decreases with sonication time (Figure 1c) until reaching an apparent limiting molecular weight of M_lim = 17,600 Da (Figure S1). The formation of dissociated Cp ligand is revealed by¹H NMR spectroscopy (δ=6.4~7.5 ppm, Figure 1d),^{28, 29} and the peak intensity increases with sonication time. We hypothesize that the Cp originated from protonation of mechanically generated cyclopentadienyl anions.³⁰ Support for this hypothesis is found through the addition of isopropyl alcohol (10 vol%) to the sonication solution, which significantly increased the yield of the free ligand (Figure S2). Thus, this observation suggested that the mechanism of ferrocene dissociation involves the heterolytic dissociation of Cp⁻ from [CpFe]⁺ and inspired further mechanistic studies (vide infra).

Figure 1.

(a) Schematic illustration of a main-chain ferrocene-containing polymer exposed to external mechanical stress; (b) Representative synthetic scheme for main-chain ferrocene-containing polymer; (c) GPC traces of P1 after sonication for different time. The arrow indicates the increase of sonication time; (d) Time-dependent¹H NMR spectra of P1 exposed to sonication.

As observed by Giannantonio et al., the chain scission is quite specific to the embedded ferrocene units.²³ A comparison of the number of chain scission events (quantified by gel permeation chromatography) and the number of ferrocene scission events (quantified by¹H NMR) revealed that 97 ± 3% of the scission events occur at ferrocene (see Figures S3).

This selectivity hints at unexpectedly high lability of the Fe-Cp bond, and we quantified the mechanical bond strength of ferrocene by adding gem-dichlorocyclopropane (gDCC) mechanophores into the same polymer backbone (P4, Table 1). As some of us have reported previously,^{31, 32} the gDCC mechanophores serve as an internal standard for estimating the scission bond strength, through the competing mechanochemical processes of gDCC ring opening and ferrocene scission as shown in Figure 2a. The fraction of gDCCs that ring open per number of scission cycles of the parent polymer, Φ_i, has been shown to be a robust indicator of the mechanical strength of the bonds that break (here, the Fe-Cp bond) during polymer sonication.³³ As seen in Figure 2b, Φ_i of P4 (Φ_i = 0.22) is considerably lower than that of dichlorocyclopropanated polybutadiene P3, which lacks the ferrocene (Φ_i = 0.79); ferrocene is demonstrably more mechanically labile than the carbon-carbon bonds of polybutadiene. Similarly, control polymer P5 (Φ_i = 0.62) confirms that the ferrocene mechanophore is also more labile than the bonds associated with the aryl ester linkages

Figure 2.

(a) Competing mechanochemical processes of gDCC ring opening and ferrocene scission; (b) Fraction of ring opening of gDCC versus scission cycle for P3, P4 and P5.

Because they are robust over a range of experimental sonication conditions, the Φ_i values allow us to compare the mechanical strength (in ultrasound) of ferrocene to that of other scissile, covalent bonds. Lee et al. recently reported the mechanical strength of the carbon-nitrogen bond of an azobisdialkylnitrile (BDE < 30 kcal/mol; Φ_i = 0.18), the carbon-sulfur bond of a thioether (BDE = 71–74 kcal/mol; Φ_i = 0.43), and the carbon-oxygen bond of a benzylphenyl ether (BDE = 52–54 kcal/mol; Φ_i = 0.62).³¹ Interestingly, ferrocene has a mechanical strength that is very similar to that an azobisdialkylnitrile, despite the fact that its reported thermodynamic bond strength is nearly three times as large and even greater than that of a thioether or benzylphenyl ether. We note that the trend in Φ_i values are verified by a similar trend in M_lim (see Table S2).

We next consider further the chain scission mechanism of ferrocene, and in particular whether Cp dissociation involves back electron transfer from the Cp anion to Fe. Giannantonio et al. trapped released Fe³⁺ ions using KSCN, suggestive of the same heterolytic dissociation mechanism implied by our above observation of protonated Cp.²³ This mechanism is also consistent with the accepted mechanism for formation of ferrocene, although under different conditions. The generation of Fe³⁺, however, requires an unspecified oxidative process, which leaves some ambiguity as to the oxidation state of the iron formed directly from the mechanical dissociation. We therefore probed further the contributions from two possible mechanisms: heterolytic or homolytic,³⁴ as shown in Figure 3a.

Figure 3.

(a) Proposed heterolytic and homolytic dissociation mechanisms of a ferrocene-containing polymer by ultrasound-induced chain scission. (b) Ring-opening ratio of gDCC versus scission cycle for P4 and organic salt-stabilized P4; (c) Ring-opening ratio of gDCC versus scission cycle for P3 and organic salt-stabilized P3; (d) UV-vis absorption spectra of P7 mixed with phenanthroline before and after sonication; (e) UV-vis absorption spectra of P8 mixed with phenanthroline and cyclic ferrocenyl olefin monomer 6 before and after sonication.

If the mechanism was heterolytic, as suggested by the observations noted above, then the dissociation energy would include a contribution from the charge separation of [CpFe]⁺ and Cp⁻ in the relative low dielectric THF solvent. We therefore added soluble tetrabutylammonium bromide (TBAB) salt to the mixture in order to stabilize any charge separation and/or assist in anionic ligand exchange. As shown in Figure 3b, the addition of TBAB lowers Φ_i further to 0.12 (the ferrocene is now measurably weaker than the azobisdialkylnitrile), consistent with a heterolytic mechanism. Importantly, a control experiment demonstrates that the influence of TBAB is specific to the ferrocene, Φ_i does not change for poly gDCC (Figure 3c).

We also consider the fate of the iron. Phenanthroline associates with ferrous ion to form a stable complex [Fe(phenanthroline)₃]²⁺ that produces a strong absorption in DMF solutions that,³⁵ at ~510 nm, is well separated from the characteristic absorption of ferrocene derivatives (~450 nm). We therefore synthesized a DMF-soluble copolymer of ferrocene and 5-hydroxycyclooctene (P7). P7 is almost identical in strength with P4 in THF (see Figure S4). Sonication of P7 alone produces no change in solution color or UV-Vis spectrum. Sonication of P7 in the presence of phenanthroline, however, leads to a peak at 510 nm (Figure 3d) that matches that of independently prepared [Fe(phenanthroline)₃]²⁺. A control experiment was carried out under the same conditions by mixing homopolymer P8 and ferrocene small molecules (Compound 6, see SI) at the same concentration of ferrocenyl units as P7 in DMF. Sonication produces no change in the absorption spectrum and provides strong evidence that the peak located at 510 nm in Figure 3d is due to the capture of mechanically generated ferrous ions. Interestingly, an additional peak (λ_max = 620 nm) also appears in response to sonication of P7, giving the solution a purple tint (Figure 3d). This peak might be attributed to a complex formed by half-sandwich ferrocenium and phenanthroline after chain scission, which is very similar to a reported half-sandwich ferrocenium acetonitrile intermediate with purple color and absorption band red shift by 100 nm relative to that of ferrocene.^{36, 37} This intermediate is unstable and decomposes to release ferrous ion gradually as shown in Figure S5, complicating further characterization. Nonetheless, these observations are consistent with initial heterolytic dissociation to [CpFe]⁺ and Cp⁻.

We also looked for evidence of homolytic dissociation, reasoning that we should be able to trap cyclopentadienyl radicals through radical addition reactions using methods some of us have employed successfully before.^{38, 39} P1 was carbon-centered radicals.⁴⁰ The rate of sonochemical molecular weight degradation did not change with addition of TEMPO, nor were any new peaks observed in the¹H NMR spectra that would support the incorporation of TEMPO moiety in the polymer chain ends (see Figures S6 and S7). Taken together, we conclude that mechanical dissociation of ferrocene proceeds through a predominately heterolytic dissociation pathway, which is scarcely reported in mechanically-induced bond cleavage.^{15, 24, 41–43}

To further probe the stretching and breaking process at an atomistic level, DFT calculations were performed with a constrained geometry simulates external force (COGEF) method.^44–46 As shown in Figure 4a, two Cp groups are aligned in an eclipsed geometry in the most energetically favorable state in the absence of applied force. As the ferrocene is stretched, the cyclopentadienyl rings start to twist until a fully staggered geometry is reached. Subsequent stretching distorts bond angles and bond lengths, and the energy of the system increases until Cp dissociates. Clearly, the dissociation process also supports the generation of Cp^‒ and [CpFe]⁺ as experimentally observed.

Figure 4.

DFT predictions of a model compound of ferrocene during mechanical stretching: (a) Structural evolutions as the chain termini are stretched; (b) COGEF potential with the chain terminal distance varied from fully relaxed (0 Å) to 8 Å; (c) COGEF force with a fixed stretching distance.

These calculations likely overestimate the dissociation energies, but nonetheless provide some key insights. Most sonicated in the presence of TEMPO (32 mM), which efficiently traps notably, the Fe-Cp is quite “soft,” and so the energy increases over a fairly long dissociation coordinate. One consequence of this is that the maximum force on the calculated potential energy surface (corresponding the force necessary to reach a barrier-less process) is only 3.1 nN, which suggests an upper limit for the ferrocene dissociation force that is much lower than that of conventional covalent bonds in organic compounds.⁴⁴ Similarly, as Cp dissociation occurs, the force is applied over a large distance and therefore contributes more energy through work to the reaction, effectively lowering the force-coupled activation energy to a greater extent. The calculations are consistent with the low force for dissociation on the microsecond timescale as revealed by our experimental observations.^{31, 44, 47} Applying models employed previously by Lee et al.,³¹ the characteristic breaking force for ferrocene is likely comparable to the relevant ring opening force of the gDCC mechanophore (~2 nN), although it is not clear how accurate these models are at low forces such as those observed here. Regardless, the forces required for ferrocene dissociation are very likely low enough that they can be probed using more precise techniques such as single molecule force spectroscopy (SMFS). Vancso and co-workers utilized SMFS experiments on ferrocene-containing polymers as a redox-driven single macromolecular motor.⁴⁸ However, they did not observe the dissociation of ferrocene due to insufficient mechanical stress. Further refinement on SMFS for main-chain ferrocene-containing polymers might be worthy for exploration. that are inert over long time in the absence of large mechanical forces but might be quite sensitive to high loads. Because the dissociation produces highly reactive and redox active products with differential properties, ferrocene might be a useful mechanophore for a range of material applications, including biomedical uses that capitalize on the rich biochemistry of iron. The mechanical susceptibility, mechanism, and products of mechanically triggered ferrocene scission are likely to depend very much on the specific use environment, and the quantitative and mechanistic studies reported here should provide baseline information and strategies for similar investigations on related systems. Finally, we speculate that striking mechanical sensitivity of the Fe-Cp bond might be general for the metallocene architecture, leading to opportunities to broaden this approach to a wide range of metals.