EPR studies on carboxylic esters, 23 [1]. Preparation of new dialkyl azulenedicarboxylates and EPR-spectroscopic study of their radical anions

2.1 Preparations

In order to facilitate the interpretation of the EPR spectra and the assignment of the proton hyperfine structure (hfs) coupling constants to specific positions in the azulene system and in the side chain, we prepared tert-butyl azulenedicarboxylates possibly with additional tert-butyl substituents in the ring, besides the corresponding ethyl esters.

Mainly the two methods which we had used previously [1] were also applied for the preparation of the dialkyl azulenedicarboxylates, i.e., carbo-bromination of the monoesters 1, 3 and 5 with subsequent direct alcoholysis [6] (Scheme 1), and ring enlargement (alkoxycarbonylcarbene insertion) of ethyl indole-2-carboxylate (7) [1] with alkyl diazoacetates [7] (Scheme 2).

Scheme 1:

Preparation of dialkyl azulene-1,x-dicarboxylates by carbo-bromination.

Scheme 2:

Preparation of dialkyl azulene-2,x-dicarboxylates by ring enlargement.

Carbo-bromination of the monoesters expectedly occurred in the electron-rich 1- or 3-position of the unsubstituted five-membered ring. Thus, we obtained dialkyl azulene-1,4-dicarboxylates (2) from ethyl azulene-4-carboxylate (1). Starting with ethyl azulene-5-carboxylate (3) gave diethyl azulene-1,7-dicarboxylate (4a), diethyl azulene-1,5-dicarboxylate (4b) and the mixed dialkyl azulene-1,5-dicarboxylate (4c). Starting with ethyl azulene-6-carboxylate (5) led to diethyl azulene-1,6-dicarboxylate (6a) (Scheme 1).

The structure of the products was proved by ¹H and ¹³C NMR spectroscopy. Diethyl azulene-1,4-dicarboxylate (2a) exhibited a pronounced down-field shift of the signal at δ = 9.75 ppm for 8-H (d, ³J = 9.7 Hz), which compares with δ = 9.64 ppm for 8-H of ethyl azulene-1-carboxylate [1] but cannot be assigned to any proton of the isomeric azulene-1,8-dicarboxylate ester. Apparently, due to steric hindrance neither dialkyl azulene-1,8-dicarboxylates nor 1-tert-butyl 7-ethyl azulene-1,7-dicarboxylate were formed from the corresponding starting esters 1 and 3.

Separation of the isomers 4a and 4b, which were produced as a mixture, was achieved by column chromatography. The pure isomers could be discriminated on account of the splitting pattern of their ¹H signals. The shielding effect of each ethoxycarbonyl group causes down-field shifts to δ = 9.24 ppm for 4-H (d, ⁴J = 1.5 Hz) and δ = 9.63 ppm for 8-H (d, ³J = 10.2 Hz) in 4b whereas the 8-H signal (d, ⁴J = 1.6 Hz) of 4a is extremely shifted to δ = 10.42 ppm due to the two neighboring ethoxycarbonyl groups in the 1- and the 7-position. The structure of 4a was also corroborated by an X-ray structure analysis (Fig. 1). Selected bond lengths, bond angles, and dihedral angles of 4a are compiled in Table 1.

Fig. 1:

Ortep-style plot of diethyl azulene-1,7-dicarboxylate (4a). Non-H atoms are drawn as spheres with radii relative to their U_equiv values, H atoms as spheres with arbitrary radii.

Insertion of alkoxycarbonylcarbenes into different bonds of the benzene ring of ethyl indane-2-carboxylate (7) led to mixtures of the regioisomers 8, 9 and 10 (Scheme 2), which could be separated by column chromatography and characterized by NMR spectroscopy.

Dimethyl azulene-1,3-dicarboxylate (12a) [8] and its 6-tert-butyl derivative 12b were obtained via cleavage of the acetyl groups of the 1,3-diacetylazulenes 11a and 11b by in-situ-generated sodium hypoiodite accompanied by iodoform extrusion and subsequent methylation with diazomethane (Scheme 3). A small amount of methyl 6-tert-butyl-3-iodoazulene-1-carboxylate (12c) was formed as by-product.

Scheme 3:

Preparation of dimethyl azulene-1,3-carboxylates by iodoform splitting.

The ester 12b crystallized in beautiful red needles which were suitable for an X-ray structure analysis. The Ortep plot (Fig. 2) demonstrates the high symmetry of this molecule with a perpendicular crystallographic mirror plane through C2 and C6. Selected bond lengths, bond angles and dihedral angles are compiled in Table 1.

Fig. 2:

Ortep plot of dimethyl 6-tert-butylazulene-1,3-dicarboxylate (12b). Displacement ellipsoids are drawn at the 50 % probability level, H atoms as spheres with arbitrary radii. The atoms are numbered according to the symmetry of the molecule and do not agree with the IUPAC rules which are applied for 12b in the text.

The perimeter bond lengths of the two esters 4a and 12b show only minor alternation except the bonds next to the ester substituents which are slightly longer [4a: d(C7–C8) = 1.464(4) Å; 12b: d(C1–C81) = 1.423(5) Å]. Also, the central C–C bonds exhibit the typical elongation (4a: d(C3A–C8A) = 1.491(6) Å; 12b: d(C81–C81A) = 1.467(7) Å); see Table 1 and Figs. 1 and 2. This is in agreement with the structures of azulene itself (1) [9, 10] and related derivatives such as 6-tert-butyl 2-ethyl azulene-2,6-dicarboxylate, the isomer of 10b, [11, 12], azulene-1,3- and -5,7-dicarboxamides [13, 14], and -5,7-dicarbothioamides [13, 14]. The planes of the ester groups of 4a and 12b are nearly coplanar with the azulene ring plane (see Table 1) just as in the isomer 10b [12] but contrary to the significant torsion angles in N,N’-dibutyl-azulene-1,3-dicarboxamide (21°/35.7°) [13], N,N’-dibutyl-azulene-5,7-dicarboxamide (34.9°/30.3°) [13, 14] and N,N′-dibutyl-azulene-5,7-dicarbothioamide (−148.8°/−138.1°) [13, 14].

Reaction of ethyl 6-methylazulene-1-carboxylate (7) with oxygen in the presence of potassium acetate and subsequent methylation with diazomethane [15] yielded the mixed diester 1-ethyl 6-methyl azulene-1,6-dicarboxylate (6b) (Scheme 4).

Scheme 4:

Preparation of 1-ethyl 6-methyl azulene-1,6-dicarboxylate by air oxidation.

2.2 Electroanalytical results

Reduction potentials E_red of the diesters (Table 2) in aprotic medium (dry DMF) were measured by use of the differential pulse polarographic (DPP) method [16, 17]. The peaks with the lowest cathodic reduction potential E_red(1) are assigned to a single electron transfer (SET) step, i.e. the formation of a radical anion. Thus it is most important in view of the EPR measurements. Another reduction peak E_red(2) at a more negative reduction potential (see Table 2) is observed if one ester substituent is located in the seven-membered ring of the azulene moiety. It can be assigned to the formation of a diamagnetic dianion. Most dialkyl azulenedicarboxylates exhibit a third peak E_red(3) at even more shifted potentials (see Table 2) which should be due to the formation of a radical trianion. We have not studied these latter sensitive species by EPR spectroscopy.

The reduction potentials E_red of the first SET of the diesters are less negative by 10–30 mV than those of related azulenemonoesters (see Table 2); compare, e.g., diethyl azulene-1,3-dicarboxylate (12a, E_red = –0.51 V) with ethyl azulene-1-carboxylate (E_red = –0.73 [1]); diethyl azulene-1,5-dicarboxylate (4b) (E_red = –0.68 V) with ethyl azulene-5-carboxylate (3, E_red = –0.79 V [1]) or diethyl azulene-2,4-dicarboxylate (8a, E_red = –0.37 V) with ethyl azulene-2-carboxylate (E_red = –0.68 V [1]). Diethyl azulene-2,6-dicarboxylate (10a) exhibits the least negative half-wave reduction potential of all compounds studied for the first SET [E_red(1) = −0.27 V] and even for the second SET [E_red(2) = −0.79 V]. This may be due to the high symmetry of 10a and the longest distance between the two ester groups and hence the best separation of the negative charges in the anions of all isomeric azulenedicarboxylate esters. This effect, which is also observed for dimethyl naphthalene-2,6-dicarboxylate [2], is illustrated by the resonance formulae in Scheme 5.

Scheme 5:

Resonance formulae of dialkyl azulene-2,6-dicarboxylate and naphthalene-2,6-dicarboxylate, and their anions.

Inspection of Table 2 reveals that the reduction potentials E_red of azulenes with ester substituents in the 2,4- (8a), 2,5-, (9a) and 2,6-position (10a) are significantly less negative than those of derivatives with ester substituents in the 1,3- (12a), 1,5- (21b), and 1,7-position (21a). This thermodynamic effect does not mean that the corresponding radical anions are more persistent. Cyclovoltammetric measurements showed, that the radical anions formed at the first SET step in most cases should be stable enough to allow EPR spectra to be recorded successfully after in-situ electroreduction, since peak current ratios of i_pa/i_pc(1) ≥ 0.4 (Table 2) are observed. But there are deviations from that rule. For example, 12a^·– with i_pa/i_pc(1) = 0.40 could not be detected by EPR spectroscopy. The corresponding 6-tert-butyl derivative 12b^·–, on the other hand, with i_pa/i_pc(1) = 0.73 is obviously stabilized because the bulky substituent inhibits decomposition reactions via the high-spin-density 6-position. Hence, its EPR spectrum can be observed although the SET reduction potential is more negative. Unexpectedly and inexplicably, an EPR spectrum of 21a^·– was obtained although its i_pa/i_pc(1) is only 0.07 (see below).

2.3 Electron paramagnetic resonance results

Radical anions of the diesters were generated by internal electroreduction as previously described [1] for the monoesters. In most cases the recorded EPR spectra exhibit large numbers of lines (see Figs. 3 and 4), since the symmetry of the azulene derivatives is low. Nevertheless, valid sets of proton hfs coupling constants a^H could be determined by simulation of the spectra. The EPR spectroscopic data obtained for the dialkyl azulenedicarboxylates are compiled in Table 3.

Fig. 3:

Experimental (top) and simulated (bottom) EPR spectra of the diethyl azulene-1,4-dicarboxylate radical anion (2a^·–).

Fig. 4:

Experimental EPR spectra observed after in-situ electroreduction of diethyl azulene-1,5-dicarboxylate (4b, top), diethyl azulene-1,6-dicarboxylate (6a, middle), and simulated EPR spectrum (bottom) as calculated with the proton hfs coupling constants of 6a^·–: a^H₂ = 0.407 mT, a^H₃ = 0.063 mT, a^H₄ = 0.413 mT, a^H₅ = 0.030 mT, a^H₈ = 0.472 mT, a^H_OCH2 = 0.059 mT (2 H), see Table 3.

Scheme 6:

Rearrangement of alkyl azulene-1,5- and azulene-2,5-dicarboxylate radical anions to alkyl azulene-1,6- and azulene-2,6-dicarboxylate radical anions.

The assignment of the proton hfs coupling constants, in particular of the asymmetric diesters, was not always straightforward. In most cases, however, a satisfying result was obtained by comparison with tert-butyl substituted derivatives and with values calculated from MO spin densities by use of the McConnell equation a^H_μ = Qρ^π_μ (see below).

Like ethyl azulene-1-carboxylate, the radical anions of dimethyl azulene-1,3-dicarboxylate (12a^·–) were not persistent enough to give an EPR spectrum. However, its 6-tert-butyl derivative 12b^·– provided a well-resolved spectrum. The high spin density in the 6-position which is presumably responsible for the low stability, i.e., the follow-up reactions of 12a^·–, consequently leads to a measurable splitting (0.02 mT) due to the γ-protons of the tert-butyl substituent in 12b^·–. Spin coupling with the OCH₃ protons in the 1,3-positions is not observed.

The proton hfs coupling constants of the azulene-1,4-carboxylate radical anions 2a^·– and 2b^·– resemble those of 12b^·–. Expectedly, the EPR spectra exhibit a large doublet splitting of 0.59 mT for the proton in the 6-position. Also, spin coupling with the two OCH₂ protons probably of the ester group in the 4-position, but not in the 1-position, is observed (Fig. 3).

In-situ electroreduction of 1,5-diethyl (4b) and 1-tert-butyl 5-ethyl azulene-1,5-dicarboxylate (4c) did not yield the corresponding radical anions. Instead, EPR spectra were recorded which were identical with the spectra of the 1,6-isomers 6a^·– and 6c^·– (see Fig. 4). The same was observed for the 2,5-isomers 9a^·– and 9b^·– which rearranged to the 2,6-isomers 10a^·– and 10b^·– and gave rise to the corresponding EPR spectra. We have of course scrutinized the esters 4 and 9 by high-amplification ¹H NMR spectroscopy in order to exclude any contamination with the isomers 6 and 10.

We assume a 5 → 6 rearrangement to take place but we cannot put forward a fully convincing reason for the observed rearrangement. Apparently, the azulene-1,6- and -2,6-dicarboxylate radical anions are considerably more stable and persistent compared with the 1,5- and 2,5-isomer which effect may be associated with the maximum spin density in the 6-position of azulene radical anions. Scheme 6 shows a proposed rearrangement mechanism.

The in-situ electroreduction of the diesters 6a, 6b, 4a, 8a, 8b, 10a and 10b, on the other hand, was very straightforward and led to the respective radical anions and the corresponding EPR spectra (see Table 3). Similarly as for the 1,4-diester radical anions 2^·–, the protons of only one of the two ester groups in the 1,6-diesters, i.e., the two 6-OCH₂ protons of 6a^·– and the three 6-OCH₃ protons of 6b^·−, led to splitting whereas the 2,6-isomers exhibited splitting due to the four OCH₂ protons in 10a^·– and the two 2-OCH₂ protons in 10b^·–.

The radical anions exhibit g-factors of 2.0051 ± 0.0001 if one ester substituent is located at the seven-membered ring. A lower value of 2.00475 is found for the 1,3-diester radical anion 12b^·–. This range is significantly higher than the g-factors of the related naphthalene esters which are found at 2.0032−2.0035 [2]. This effect may be due to a more pronounced transfer of spin density from the azulene hydrocarbon skeleton into the oxygen containing ester substituents.

2.4 Spin density distributions

In our recent publication on alkyl azulenemonocarboxylate radical anions, we have pointed out that an influence of an electron-withdrawing ester substituent on the spin density distribution does exist [1], although it is not as pronounced as in the naphthalene series. We present here corresponding results on the spin density distribution of related dialkyl azulenedicarboxylate radical anions.

Table 4 shows the spin densities ρ^π_μ obtained from the experimentally observed proton hfs coupling constants a^H_μ by use of the McConnell relationship ρ^π_μ = a^H_μ/Q, together with theoretical values determined by McLachlan type as well as by more elaborate MO calculations. Due to significant deviations of all bond angles from 120° the choice of one suitable spin attenuation factor Q, which is valid for both the seven- and the five-membered ring of the azulene skeleton is not possible [18]. The agreement between the experimental and the theoretical spin densities is thus only moderate. In most cases, it is more satisfying for the McLachlan type results compared with the sophisticated semi-empirical (PM3) or ab-initio (DFT, B3LYP 1 STO 3-G21) calculations – just as in the case of the azulenemonoester radical anions [1], see Table 4.

Particularly the high spin densities which are generally observed at the even-numbered centers ρ^π₂, ρ^π₄, ρ^π₆ and ρ^π₈ [1, 4, 18, 19] are calculated significantly too low compared with the experimental values. The influence of ester substituents at the five-membered ring on the spin density distribution of azulene radical anions is only slight. Azulene (1^·–) itself, the 1-ester, the 2-ester and the 1,3-diester (12b^·–) closely resemble each other except for a minor perturbation of the symmetry in the 1-ester radical anions (see Scheme 7). This is somewhat unexpected since the electron distribution in azulene should be markedly influenced by strongly electron-withdrawing ester substituents. It may be due to the fact that the ester substituents of 12b^·– are located at positions with very low spin densities. The change is more pronounced if ester substituents are located at the seven-membered ring. The spin density distribution in the 1,6- (6a^·–) and the 2,6-diester (10a^·–) radical anions are similar to each other and to the 6-monoester (5^·–) but not to the 1-monoester or 2-monoester (Scheme 7). Ester substituents in the 1,4- (2^·–), the 2,4- (8^·–) and, most noticeable, in the 1,7-position (4^·–) lead to significant effects on the spin density distribution. Apparently, the deviation of these esters from the symmetry of the azulene system causes a marked disturbance of the spin density distribution. For instance, diethyl azulene-1,7-dicarboxylate radical anions (4a^·–) exhibit the largest proton hfs coupling constant (0.974 mT) in the 4-position, i.e. opposite to the ester substituent.

Scheme 7:

Proton hfs coupling constants a^H_μ (mT) in alkyl azulenemono- and -dicarboxylate radical anions. a^H₆ = (0.907) was measured for the ethyl 3-tert-butylazulenecarboxylate radical anion, and a^H₆ = (0.766) for 12b^·– was calculated from the corresponding theoretical spin density ρ^π₆ = 0.333.

4.1 General

Melting points (corrected): Electrothermal. Boiling points were determined during distillation. UV/Vis spectra (solvent EtOH): Hitachi 200 spectrophotometer. NMR spectra (δ in ppm vs. Me₄Si) were recorded on a WM 400 spectrometer (Bruker) at 400 MHz (¹H) and 100 MHz (¹³C) in CDCl₃. The NMR data are compiled in Tables 5 and 6. Assignments of the ¹H NMR signals were performed by the DEPT method and consideration of the coupling patterns. The ¹³C NMR spectra of 2a, 4a, 4b, 6b, 8a, 9a, 10a, 12a, 12b, and 12c were calculated by use of the DFT method as described recently [1]. The effect of the iodo substituent in the 3-position of 12c was not properly captured [δ(C3) found: 73.7 ppm vs. calcd.: 110.5 ppm] since the smaller basis set LanL2D2 had to be used instead of the 6-311+G(2d,p) basis set. The ¹³C NMR signals of the other compounds were assigned with reference to the related derivatives. IR spectra (films for liquids, KBr pellets for solids, ν in cm^–1): FT-IR 1720X (Perkin-Elmer) and Genesis-FT-IR (ATI-Mattson). MS: CH 7 (EI, 70 eV, Varian). HRMS: 70-250S (VG-Analytical). Thin layer chromatography (TLC): Al-foils coated with Kieselgel 60F₂₅₄ (Merck). Column chromatography (CC): Kieselgel 60 (70–230 mesh, Merck).

Differential pulse polarograms were measured with the VA 663/Polarecord 626 (Metrohm), cyclic voltamograms: VA-Scanner E 612 (Metrohm) with plotter Servotec 7040 A (Hewlett Packard). Potentials were measured in DMF/0.1 n tetrapropylammonium bromide (TPAB) by use of a HMDE working cathode vs. an internal Ag wire as reference electrode, which corresponds to the Ag/AgBr couple in the TEAB solution [16, 17].

The in-situ generation of the radical anions in dry DMF at suitable temperatures by use of a Bank Electronic (Wenking type) potentiostat MP 31 and the measurement of the EPR spectra in quartz flat cells has been described earlier [1]. A Bruker ESP 300 E spectrometer was used. The spectra were processed by use of the WIN-EPR 9212101 software (Bruker). Simulation of the spectra was carried out with the Simfonia program (Bruker).

4.2 X-ray structure determinations

The crystal data of 4a and 12b and a summary of experimental details are given in Table 7. The structures were solved by Direct Methods (Multan) [20], and differential Fourier synthesis [21]. Refinement was performed by full-matrix least-squares methods. Crystallographic data of 4a and 12b have been deposited [22].

4.3 Preparations

4.3.2 Dialkyl azulenecarboxylates by ring enlargement of ethyl indane-2-carboxylate [23]

Caution! Alkyl diazoacetates are toxic, carcinogenic and explosive! All reactions with these reagents must therefore be performed behind a protective shield under a well-ventilated hood [24].

Ethyl diazoacetate [25] or tert-butyl diazoacetate [26] was dropped into an excess of ethyl indane-2-carboxylate (7) with stirring under N₂ at 20 °C. The mixture was heated to 140 °C for 6 h. The excessive indane was distilled off under vacuum. The residue (hydroazulene) was dissolved in a tenfold amount of benzene and refluxed with an equimolar amount of chloranil for 2 h. The color changed from yellow to green-blue. The solvent was removed under vacuum and the residue was extracted with hexane in a Soxhlet apparatus. Flash-CC (toluene-hexane 1:1) of the extract yielded the deeply blue colored azulene (mixture), which was purified by CC.

Reaction of 7 [1] (15.55 g, 82 mmol) with ethyl diazoacetate (4.90 g, 43 mmol) led to a mixture of diethyl azulene-2, 4-dicarboxylate (8a), diethyl azulene-2,5-dicarboxylate (9a) and diethyl azulene-2,6-dicarboxylate (10a). The components were separated by CC (AcOEt-hexane 1:20).

4.4 Quantum chemical calculations

DFT calculations of the NMR spectra [30] were performed at the Taito supercluster of the Finnish IT Center for Science (CSC), cf. http://www.csc.fi.

McLachlan-type MO calculations were performed by use of an unpublished Fortran-77 program Hueckel 88 [31, 32]. The Qcpe software Mopac 6.0 was used for the AM1 calculations. Semi-empirical PM6-type MO calculations were performed by use of the program package Mopac 2009 [33]. DFT-based geometry optimizations and spin density calculations were performed by the B3LYP method [34, 35] with 6-31G(p,d) basis sets. The program Firefly [36, 37] was used for the DFT calculations. The Qcpe programs Draw and Jmol (Open Project; http://sourceforge.net/projects/jmol/) were used for the graphical presentation of the results. A conventional PC (Pentium Dual Core CPU E5200, 2.5 GHz; 3.25 GB RAM) was applied for the calculations.