Investigations by Exchange Spectroscopy of 54d

For the purpose of gaining quantitative information about the dynamic behavior of 54d, EXSY experiments were performed at various temperatures. As for the exchanging species a syn/syn configuration is assumed, two options for their interconversion exist. First option would be a twofold syn/anti exchange (π-σ-π-process) via a short-lived (though not detectable) syn/anti intermediate. This is however very unlikely, as this mechanism is rarely found for N-donor ligands. Thus, the apparent allyl rotation, which was evidenced for the analogous (methallyl)palladium complexes 46a-d most likely takes place.^[101]

No exchange processes for isomer C could be observed. As this isomer could only result from a π-σ-π-process, both major rotational processes for allyl palladium complexes are operative for 54d, which has already been reported for other complexes.[142,164a,194a,195]

In the following discussion species C will not be excluded.

The A-B exchange process via apparent allyl rotation is observed for all protons of isomer A and B, but only the signals of the protons in α-position of the oxazoline ring were chosen due to their complete baseline separation (primarily the B(+α)/A(α) protons, see Figure 7.19).

Figure 7.19: Part of an EXSY spectrum (400 MHz, 273 K, CDCl3, d8 = 0.5 s) of 54d, observed cross-peaks originate from the exchange process.

entry c [mg/ml] T [K] kAB [s^-1] kBA [s^-1]

Table 7.11: Determined exchange rate constants of the apparent allyl rotation for 54d from EXSY spectra (400 MHz, CDCl3). Calculation were performed with EXSYCalc.,^[176] using signals of B(+α)/A(α) if not stated otherwise; a) average value of two measurements; b) average however in all cases consistent with the A:B ratios of roughly 2:1 determined by integration.

Previous work of other groups has revealed various factors influencing the apparent allyl rotation. First of all, the impact of potentially coordinating impurities such as water or chloride anions (from the precursor complex or by decomposition of CD2Cl2 or CDCl3) has to be addressed.[146b,195a-c,196]

For instance, no recognizable decrease of the rate constants was observed, when anhydrous CDCl3 was used (entry 3). For solutions containing 4 equivalents of [PPh4]Cl per complex, only an insignificant increase for the exchange rates was observed (entry 5). CDCl3 was therefore used without purification, as only trace amounts of chloride ions and water were expected therein.

Another reported effect is the formation of an ‘‘ion pair’’ between the cationic complex and the anion that should be favored in concentrated solutions.^[146b] This is expected to be more pronounced in non-coordinating solvents, but depending on temperature and concentration

range used; its influence can be rather small. Therefore, samples of various concentrations were measured (entries 1-3), yielding not statistically significantly different values.

Further, the influence of strongly coordinating additives, such as PPh3 was evaluated (entry 4).^[197] After addition of 0.5 equivalents per palladium atom, no increase could be observed.

However, when an excess of about two equivalents per palladium center was used, no more exchange cross peaks were observed, probably due to displacement of the PyrBOX ligand.

Finally, as anticipated, a decrease of the rate constants by lowering the temperature to 273 and 248 K was observed (entries 6-8). Entries 6 and 7 further demonstrate the reliability of the method, as irrespective of which of the resonances of the asymmetric units is used, (average) reaction rate constants (Table 7.11. entry 2, 6, 8) using the Eyring equation.

As the rate constants were determined at different temperatures, further information on the activation enthalpy ΔH^‡ and entropy ΔS^‡ can be obtained, applying the Eyring-Polanyi equation. (k = rate constant, kB = Boltzmann constant, T = absolute temperature, h = Planck’s constant, R = gas constant)

The free activation energy is further defined by of the Gibbs-Helmholtz equation as

‡

From the plot of ln(k/T) versus 1/T a straight line should result, with a slope –ΔH^‡/R, from which the activation enthalpy can be derived. From the intercept ln(kB/h) + ΔS^‡/R, the activation entropy can be calculated. The derived values from the linear fit of the plot for the investigated temperature range are depicted in Table 7.13.

The subtle difference for the activation enthalpy indicates relatively enthalpically similar ground states for A and B. Thus, it would explain the relatively equal distribution for these isomers. As both, back and forth reaction show a negative activation entropy, an associative mechanism via a more organized transition state, can be assumed.^[145a]

Temperature dependent EXSY experiments in various solvents further corroborate an associative mechanism. As previously outlined, the simple rotation of the allyl moiety is orbitally forbidden, unless due to coordination of e.g. a solvent molecule, the coordination number is altered in the transitions state. Therefore, the efficiency of the initiation of the rotation process can be correlated with the coordination ability of the fifth donor.^[194a] As anticipated, in THF-d8, a broadening of all resonances for A and B was observed, while for C no significant acceleration of the exchange process was obvious (Figure 7.20).

Figure 7.20: Variable temperature ¹H-NMR spectra (400 MHz, THF-d8) of 54d; X marks residual solvent signals (THF).

Rate constants could subsequently be derived in the same temperature range, revealing an approximate twofold acceleration in THF compared to the non-coordinating CDCl3 (Table 7.14 and Figure 7.21, top).

entry T [K] kAB [s^-1] kBA [s^-1] ∆G _AB^‡ [kJ/mol]

∆G _BA^‡ [kJ/mol]

1 298 22.6 41.5 65.3 63.8

2 273 2.87 2.95 64.3 64.2

3 248 0.226 0.141 63.4 64.4

Table 7.14: Determined exchange rate constants of the apparent allyl rotation for 54d from EXSY spectra (400 MHz, THF-d8, c ≈ 35 mg/ml) and derived free activations energies using the Eyring equation. Calculation were performed with EXSYCalc., using signals of B(+α)/A(α).

Surprisingly, the A:B ratio inverts, from initially ≈ 2:1 at 298 K, over ≈ 1:1 at 273 K to ≈ 1:2 at 248 K, which is reflected in the rate constants for the back and forth reaction. The free activation enthalpies follow the observed trend, after which isomer B becomes favored at lower temperatures (Figure 7.21, bottom).

Figure 7.21: Schematic representation of the derived thermodynamic values of the A-B exchange process; top: CDCl3 and THF-d8 at 298 K; bottom THF-d8 in the investigated temperature range (298-248 K).

∆H _AB^‡ [kJ/mol]

∆H _BA^‡ [kJ/mol]

∆S _AB^‡ [J/(molK)]

∆S _BA^‡ [J/(molK)]

54.4 ± 0.5 67.6 ± 0.8 -36.5 ± 1.7 12.6 ± 3.0

Table 7.15: Derived values for activation enthalpy and entropy from the Eyring-plot (from data of Table 7.14).

This can be explained by the values determined from the Eyring-plot (Table 7.15). For the forward reaction (A to B), the activation enthalpy is lower, but the whole process is strongly negentropic. The exchange from B to A, however, possesses a significantly higher activation enthalpy, while being exotropic. Comparing the obtained values with the thermodynamic properties in CDCl3, it can be concluded that the ground state of A possesses a higher standard enthalpy. For B, the formation of the transition state requires a larger enthalpic contribution, accompanied by a loss of entropy. These observations are supported by the thermodynamic data of the ground state derived from the equilibrium constant. With the equilibrium constants K, which in the examined case represents the ratio of B to A.

]

Thermodynamic values can be derived using the van t’Hoff equation,

RT2 temperature in Kelvin and R the gas constant. From the definition of the Gibbs free energy ΔG^Θ (standard entropy change ΔS^Θ)

Therefore, the plot of ln K versus 1/T yields a straight line (Figure 7.22). From its slope, the standard enthalpy term can be derived by multiplication with -R. The standard entropy change can be calculated from the intercept by multiplication with the ideal gas constant. By this approach, values of ΔH^Θ ≈ -17 kJ/mol and ΔS^Θ ≈ -62 J/(molK) for the exchange from A to B can be approximated. This supports the previous assumption, that B possesses the lower standard enthalpy; meanwhile A is entropically favored.

Figure 7.22: Plot of ln (K) versus T^-1; error bars represent an estimated error of 10 %;

equilibrium constants were determined by integration of the corresponding ¹H-NMR spectra at the corresponding temperatures.

Assuming an associative mechanism, a mechanistic model is proposed. The enthalpy of the transition state should be hardly affected by non-coordinating solvents such as chloroform or DCM. Thus, the previously determined values in these solvents of ΔH^‡ ≈ 62 kJ/mol may give a hint on the process with almost complete exclusion solvent effects. However, small solvent effects are observed, which tend in the same direction as the measurements in THF.

solvent ∆H _AB^‡ [kJ/mol]

∆H _BA^‡ [kJ/mol]

∆S _AB^‡ [J/(molK)]

∆S _BA^‡ [J/(molK)]

CDCl3 62.1 ± 2.1 62.8 ± 1.0 -17.2 ± 6.2 -8.8 ± 4.4 THF-d8 54.4 ± 0.5 67.6 ± 0.8 -36.5 ± 1.7 12.6 ± 3.0 Table 7.16: Derived values for activation enthalpies and entropies for CDCl3 and THF-d8.

By changing to THF, for the process A→B, the activation enthalpy decreases, which could arise from a relative stabilization of the transition state or less likely destabilization of A. The obtained values show further a significant enthalpic stabilization of ground state B (ΔΔH^‡ ≈ 13 kJ/mol, ΔΔS^‡ ≈ 48 (J/molK)) in THF-d8 compared to CDCl3 (Table 7.16).

As similar ground state geometries can be anticipated, the enthalpic differentiation must stem from sources other than the complex itself. As the complex is monocationic, a possible cause might be contributions of solvation process. Isomer A possesses a higher symmetry than B (C1-symmetry). It follows a higher polarity for B, thus, a higher degree of solvation. By passing the transition state starting from B, the polarity should be decreased along the reaction coordinate, resulting in desolvation (represented in a decrease of enthalpy and surprisingly even an increase of entropy). The opposite behavior should result from passing the transition state outgoing from A. Due to the increased polarity; the transition state might be more solvated, represented in the lower activation enthalpy and the strongly negative activation entropy. In conclusion, changing the solvent to THF accounts for an enthalpic stabilization of B of ΔΔH ≈ 13 kJ/mol and loss of entropy of ΔΔS ≈ -40 J/(molK).

Figure 7.23: Variable temperature ¹H-NMR spectra (400 MHz, CD3CN) of 54d; X marks residual solvent signals (MeCN, H2O).

Further experiments were performed in CD3CN (Figure 7.23), revealing, as anticipated, a substantially faster process.^[136b] While at room temperature only the average of A and B was observed, coalescence occurred at around 273 K. At 248 K the A:B ratio was determined to be approximately 1.2:1. However, the exchange was still too fast to unambiguously determine the rate constants by the applied method (expected k ≈ 100-500 s^-1, with free activation enthalpy ΔG^‡ ≈ 51-48 kJ/mol).

An overview of the most important rate constants obtained under various conditions is shown in Table 7.17. As anticipated, for an associative process, acceleration in coordinating solvents (such as THF-d8, entry 6-8) compared to non-coordinating solvents (such as CDCl3

Table 7.17: Overview of determined exchange rate constants of the apparent allyl rotation for 54d from EXSY spectra (400 MHz) and calculated free activation enthalpies. Calculation were performed with EXSYCalc., using signals of B(+α)/A(α) if not stated otherwise; a) average value of two measurements; b) average value of three measurements; c) c ≈ 75 mg/ml; d) c ≈ 10 mg/ml; e) c ≈ 35 mg/ml; f) no exchange cross peaks observed.

Further, a solvent dependency of the relative abundancies could be established. While in non-coordinating solvents only marginal differences were found, in coordinating more polar solvents, the equilibrium was considerably influenced. For this phenomenon a model was proposed, taking solvation processes into account, which might give an explanation.

7.2.3.4 Comparative Examination of the Solution Behavior of 54a-d