Intramolecular Hydrogen Atom Transfer in Aminyl Radical at Room Temperature with Large Kinetic Isotope Effect

Abstract

We report a large kinetic isotope effect at 298 K, k_H/k_D ≈ 150, associated with an intramolecular 1,5-hydrogen atom transfer (1,5-HAT) in the decay of PEGylated carbazyl (aminyl) radical in solution. The experimental observations surprisingly combine the hallmarks of tunneling, including large KIEs and unusual activation parameters, with linear Arrhenius and Eyring plots over an exceptionally wide temperature range of 116 K.

Graphical Abstract

1,3,6,8-Tetra-tert-butyl carbazyl (TTBC) is one of very few nitrogen-centered (aminyl) radicals that are stable at ambient conditions and can be isolated as a solid (Figure 1).^1,2 Although this structural motif is attractive as the building block for organic magnetic materials,^3–7 TTBC has received little attention since its report by Neugebauer and Fischer in 1971.¹ With our recent development of polynitroxide scaffolds as organic radical contrast agents (ORCAs) for magnetic resonance imaging,⁸ we have been exploring the feasibility of using aminyl radicals in the next generation of ORCAs. We were interested in TTBC as a framework in the design of stable, water soluble radicals. We envisioned that hydrophilic derivatives of the hydrophobic t-butyl groups in TTBC, as in the methoxy polyethylene glycol (mPEG) substituted 1-H, would provide solubility in water while maintaining adequate stability.

Figure 1.

TTBC and aminyl radicals 1-H and 1-D.

We prepared radical 1-H and found that it had a short half-life, τ_1/2 ~ 1 min, at 298 K. We suspected that an intramolecular 1,5-hydrogen atom transfer (1,5-HAT)^9–11 from the -CH₂-O- fragment^12,13 might be the factor in the short half-life of 1-H. We investigated the D₈-isotopomer 1-D and found that it had a much longer half-life, τ_1/2 > 1 h, with extremely large room-temperature kinetic isotope effect (KIE), e.g., k_H/k_D ≈ 150 in acetone. This magnitude of KIE suggests the possibility of quantum mechanical tunneling (QMT),^14,15 a fascinating phenomenon. Intramolecular HAT, has been widely investigated^9–11,16 and recognized in regioselective radical C-H activation. 1,5-HAT involving nitrogen-centered radicals may be relevant to organocatalysis.¹⁷

Here we report the synthesis and decay kinetics of aminyl radical 1-H and its isotopomer 1-D.

The synthesis of 1-H and 1-D, starting from 1,8-dibromo-3,6-di-tert-butylcarbazole¹⁸ (2), is outlined in Scheme 1.

Scheme 1.

Synthesis of 1-H and 1-D.

Pd-catalyzed cross-coupling of malonate with carbazole 2 afforded 3, which was alkylated with methyl iodide¹⁹ to provide 4. Reduction of 4 with LiAlH₄ or LiAlD₄ afforded either 5-H or 5-D. Subsequent Williamson etherification gave the carbazoles 6-H and 6-D. ¹H NMR spectra indicated >99% deuteration in 6-D. Treating 6-H and 6-D with n-BuLi, followed by one-electron oxidation of the resultant N-centered anion using ferrocenium cation ([FeCp₂][BF₄]) at low temperature provided aminyl radicals 1-H and 1-D. The blue solution of the radical was diluted with cold pentane and purified by column chromatography at −95 °C (for 1-H) and −40 °C (for 1-D) using deactivated silica gel.²⁰ After removal of the solvents at −40 °C, 1-D was characterized by paramagnetic ¹H NMR spectroscopy, which shows broadened peaks of 1-D accompanied by sharp peaks of unreacted carbazole 6-D. EPR spin counting determinations indicated spin concentration of 32–45% for 1-H and 51–63% for 1-D. The spin concentration of 1-H notably correlated inversely with the length of time for removal of solvents at −40 °C. These results provide initial clues about the stabilities of 1-H and isotopomer 1-D.

The decays of 1-H and 1-D were studied by EPR spectroscopy and superconducting quantum interference device (SQUID) magnetometry. First order rate constants, k, were obtained by monitoring the decay of the mid-field EPR peak height and SQUID paramagnetic susceptibility of the aminyls. Data derived from the double integrated EPR spectra were found to produce similar values of k (see: the Supporting Information (SI)). Arithmetic means of multiple EPR measurements and, at selected temperatures, of two or more different samples were obtained to ensure reproducibility of k values. At least 10 data points were fit to the first order rate equation in each case. For 1-H and 1-D, k values were determined at 13 and 11 different temperatures in the 182–298 K and 232.5–316 K ranges (spanning 116 K and 93 K). The temperatures of the measurements were limited by the liquid range of solvent and decay half-lives (τ_1/2). Long τ_1/2 in the low temperature range, e.g., τ_1/2 106 days for 1-D at 232.5 K, required long measurement times and good accuracy.

The kinetic observations (Table 1 and Fig. 2) have many remarkable features. The decay of 1-H still proceeds at 182 K, indicating a low ΔH^‡ (10.0 kcal mol⁻¹) despite a calculated ΔH^‡ (see below) of ≥20 kcal mol⁻¹. The KIE of ~150 at room temperature is unusually large and is one of the largest known at this temperature.^21–23 The most striking feature of the KIE, however, is the degree to which it increases at the temperature decreases. KIEs normally increase with decreasing temperature (with some interesting exceptions),^24,25 but the observed KIE here rises to ~360 at 253 K and 1380 at 232 K, far more than is consistent with an ordinary isotopic difference in activation parameters. The unusual nature of this rise is easily recognized from the Eyring plots where the slopes for 1-H versus 1-D differ substantially. This is reflected in the anomalously large difference in the enthalpies of activation, with ΔΔH^‡ = 4.90 ± 0.38 kcal mol⁻¹ (mean ± SEM). Analogous anomalous values are found for Arrhenius activation energies and pre-exponential factors, i.e., ΔΔE_a = 4.97 ± 0.38 kcal mol⁻¹ and A_H/A_D = 0.027 ± 0.020 (mean ± SEM). For reference, the difference in zero-point energy for an aliphatic C–H versus C–D stretching vibration is only ~1.1 kcal mol ¹ and A_H/A_D should be greater than about 0.7.²⁶ Thus, the observed difference in the activation parameters (ΔΔH^‡ and ΔΔE_a) upon isotopic substitution far exceeds the semiclassical maximum for an over-the-barrier reaction.^25,26 Unusual activation parameters have often been observed for isotope effects, and these are normally understood to be a consequence of the curvature of Arrhenius or Eyring plots.²⁵ However, the obtained Eyring and Arrhenius plots are indistinguishable from linear over the broad 116 and 93 K temperature range (Fig. 2). This is not in line with the typical signature of QMT and it is most likely indicative of thermally (vibrationally) activated tunneling,²⁵ e.g., analogous to that observed in 1,5-H-shift in derivatives of cyclopentadiene and cis-1,3-pentadiene.^27,28

Table 1.

Decay Kinetics of 1-H and 1-D in Acetone.^a,^b

	T (K)	Ea (kcal/mol)	ln A	ΔH‡ (kcal/mol)	ΔS‡ (e.u.)
1-H	182 – 298	10.5 ± 0.3	13.5 ± 0.8	10.0 ± 0.3	−33.3 ± 1.4
1-D	232 – 316	15.5 ± 0.8	17.1 ± 1.5	14.9 ± 0.8	−26.4 ± 2.9

^aFirst order rate constants determined by either mid-field peak height of the EPR spectrum or paramagnetic susceptibility measured by SQUID (Fig. 2).

^bError bars correspond to the 95% confidence intervals (Tables S5–S10, SI).

Figure 2.

Decay kinetics of 1-H and 1-D in acetone: numerical fits to the Arrhenius (top) and Eyring (bottom) equations. The first order rate constants (k) were measured using mid-field peak height of the EPR spectrum and paramagnetic susceptibility using SQUID.

A series of experimental observations were obtained to delineate the reactions of 1-H and 1-D and the specific step defining the isotope effect being observed. Decay of 1-H (32–45% spin concentration, containing 6-H) at room temperature showed a mixture of carbazole 6-H and aldehyde 7-H, as indicated by ¹H NMR and mass spectroscopic analyses. Isolation by preparative thin layer chromatography (PTLC) provided 6-H and the low polarity aldehyde 7-H in 59% and 17% yields. The structure of 7-H was confirmed using 2-D NMR spectroscopy and correlations between the DFT-calculated and experimental ¹H/¹³C NMR chemical shifts. Aldehyde 7-H is possibly the product of β-scission (fragmentation) of the C-centered radical 1-H-C1 formed by 1,5-HAT, or more likely, of disproportionation of 1-H-C1, followed by reaction of a product carbocation with trace amounts of water,²⁹ thus providing indirect evidence for the C-centered radical intermediate (Scheme 2).

Scheme 2.

Product Analyses for the Decay of Aminyl Radicals 1-H and 1-D.

For the decay of 1-H at 253.2 K and of 1-D at 295.2 K, first order rate constants are practically identical in acetone-h₆ and acetone-d₆ at 295.2 K, thus indicating that the solvent isotope effect is essentially negligible.^30a

We attempted a spin trapping³¹ experiment (Scheme 2). Addition of 10 equiv of t-BuN=O dimer (in the absence of light) does not significantly affect the rate of decay of 1-H at 253 K.^31c Formation of a relatively persistent nitroxide radical (g = 2.0058, a_N = 1.32 mT) was observed, and HR ESI- MS of the product 11-H (in the mixture with 6-H) is consistent with spin trapping of C-radical 1-H-C1.³¹

The decay of 1-D at room temperature provided aldehyde 7-D (d₇-isotopomer of 7-H) and a lower polarity minor product, which was tentatively assigned to a mixture of alkenes 8-D and 9-D (2:1 relative intensity in ESI MS). Isolated yields for 6-D, 7-D, 8-D/9-D were 49%, 11%, and 4%; the PTLC fraction of 7-D also contained a small amount of alkene 10-D. In addition, a more polar fraction was isolated in 8% yield (SI). The lower yield of 7-D, compared to 7-H, and the formation of alkenes 8-D – 10-D, suggests that the decay of 1-D includes not only the intramolecular 1,5-HAT (D-transfer) but also intermolecular or intramolecular HAT from other C-H bonds on the mPEG₃ chains through 8- and higher-membered TS (Scheme 2). Although the alternative HAT processes are not major, their contribution indicates that the KIEs based on the rates of decay of 1-H and 1-D, may be viewed as lower limits.

To support our experimental findings, we located relevant transition states on the potential energy surface (PES) of the simplified model aminyl radical 1a (N-radical 1a) and its decay products (Scheme 3).³² Two transition states (TS) for intramolecular HATs from N-radical 1a were identified. The lowest energy TS corresponded 6-membered ring with N-H-C angle of 147°, which had an activation energy, E_a = 21.8 kcal mol⁻¹, and led to C-radical 1a-C1, which was 5.7 kcal mol⁻¹ above 1a. Another TS corresponded to 8-membered ring with N-H-C angle of 163°, which had an E_a = 26.5 kcal mol⁻¹, and led to C-radical 1a-C2, which was 10.0 kcal mol⁻¹ above 1a. We modelled formation of aldehyde 7a and alkene 8a via β-scission from the corresponding C-radicals 1a-C1 and 1a-C2; as the relevant TS’s have E_a = 24.1 and 62.9 kcal mol⁻¹ (vs. 1a), they are not likely to be significant contributors to the decay of 1-H.

Scheme 3.

UB3LYP/6-311+G(d,p)+ZPVE Energies in Acetone Using the IEF-PCM-UFF Solvent Model.

We estimated contribution of QMT to the observed KIE using UM06-2X PES³³ and asymmetric Eckart barrier for 1,5-HAT step from 1a to 1a-C1;³⁴ the computed ratios of tunneling correction factors (Γ*_H/Γ*_D) for 1a and its D₂-isotopomer are 399 and 34 at 232.5 and 298 K (Table 2). If the corresponding over-the-barrier KIEs at 232.5 and 298 K are near their typical values of 8 and 5,^10c,11 then estimated k_H/k_D ≈ 3200 and 170 (Table S14, SI) are not far off from the experimental values k_H/k_D of 1380 and 156 (Table S11 and Figure S2, SI).³⁵

Table 2.

Eckart-Barrier Tunneling Correction Factors (Γ*_H and Γ*_D) for 1,5-HAT Step.^a

	Isotopomer	298 (K)	232.5 (K)
Γ*H	H2	683.9	317077
Γ*D	D2	19.9	794.4

^aUM06-2X/6-311+G(d,p)/IEF-PCM-UFF+ZPVE PES.³³

In summary, aminyl radical 1-H was found to decompose via 1,5-HAT, with unusual activation parameters for isotope effects (H vs. D) and a large contribution from QMT. Nevertheless, Arrhenius and Eyring plots remain linear over a 116 K range. These unusual results may provide inspiration for theoreticians to reproduce our experimental data.³⁵ In particular, it would also be interesting to see whether tunneling can affect stereoselectivity of 1,5-HAT, e.g., in 1-H, the C-H bonds of within each of methylene moieties are diastereotopic.