ESR of Radicals in Organic & Bio organic Systems, Cardiff, 1988 Burkhard Kirste, Michael Grimm and Harry Kurreck

ENDOR Studies of Organic Multispin Systems Ladies and Gentlemen, organic oligoradicals or multispin systems are characterized by scalar exchange interaction and dipolar coupling of the unpaired electrons in addition to Zeeman splitting and hyperfine interaction. The exchange interaction gives rise to an increasing number of hyperfine components in the ESR spectra, and the dipolar coupling, also known as zero field splitting, is a source of line broadening because of electron electron dipolar relaxation. For both reasons, fluid solution ESR spectra of oligoradicals are usually poorly resolved. Consequently the higher resolution of ENDOR spectroscopy should be useful for unravelling the hyperfine interactions in these systems. In my talk I want to discuss the application of ENDOR spectroscopy to the investigation of galvinoxyl bi-, tri- and tetraradicals in fluid solution.

(Slide 1) The first slide shows the formula of a 13C labelled bisgalvinoxyl. The ESR spectrum of the respective monoradical (with one OH instead of O dot) exhibits a doublet of quintets. The doublet splitting of about 28 MHz is due to one 13C nucleus, whereas the quintet pattern with a splitting of about 3.6 MHz is due to the four galvinoxyl ring protons. In the biradical, the scalar electron exchange interaction is large, and hence there is hyperfine interaction with two 13C nuclei and eight galvinoxyl ring protons. Thus, the ESR spectrum of the biradical exhibits a triplet of nonets. The hyperfine splittings are half of those of the monoradical. The ENDOR spectrum of the monoradical which is shown at the top right was recorded at 210 K. It shows only the signals of the galvinoxyl ring protons, which are slightly inequivalent, and those of the tert butyl protons close to the free proton Larmor frequency. 13C ENDOR signals can only be detected at higher temperatures, for instance at 250 K. The ENDOR spectrum of the biradical, recorded at the same temperature, is depicted at the bottom right. Note that all signals appear at exactly the same positions as in the monoradical; but there is no signal at or near the free proton Larmor frequency. Moreover, the signals are significantly broader than in the case of the monoradical. This different relaxation behaviour is a consequence of the electron electron dipolar interaction. In summary, the ENDOR spectrum of a biradical which is made up of two identical monoradical moieties differs from the monoradical spectrum only with respect to the relaxation behaviour. So how can we be sure that the spectrum is really due to the biradical? By virtue of 13C labelling and the selectivity of the ENDOR method, this is fortunately granted. In the ENDOR experiment, a field setting at the centre of the ESR spectrum was employed. With this setting, we cannot pick up any response from 13C labelled monoradical. Since we do see 13C ENDOR signals, these must stem from the biradical.

(Slide 2) The next slide shows an energy level diagram for a biradical, detailing the electron Zeeman, nuclear Zeeman, and hyperfine interactions. The NMR transitions labelled 1 and 3, occurring in the electron spin manifolds MS = ±1, give rise to the ENDOR signals at the free nuclear Zeeman frequency νn ± at (the triplet coupling constant). Since at equals ad/2 (the doublet coupling constant) in our case, the signals should indeed occur at the same positions as in the monoradical. The energy level diagram predicts a third ENDOR signal at the free nuclear Larmor frequency, denoted by NMR 2, arising from NMR transitions within the sublevel MS = 0. This signal cannot be observed in isotropic fluid solution, because the ESR transitions connected to the level MS = 0 are pumped equally strongly, thus leaving the thermal nuclear spin polarization within this level undisturbed. It can be detected, however, in liquid crystalline or solid solutions.

After discussing biradical ENDOR in some detail, I will look at triradicals more quickly. (Slide 3) The ESR and ENDOR spectra of a 13C labelled galvinoxyl triradical are depicted on the next slide. The quartet pattern of the ESR spectrum is due to hyperfine interaction with three 13C nuclei. Proton hyperfine splittings are not resolved in the ESR spectrum. On the other hand, proton ENDOR signals are well resolved. In fact, the galvinoxyl ring protons give rise to four ENDOR signals with a spacing of 1.2 MHz, which is one third of the monoradical splitting. Moreover, three 13C ENDOR signals can be detected. A fourth signal should appear at the inconveniently low frequency of about 1 MHz.

(Slide 4) An energy level diagram for a quartet state triradical is given on the next slide. In agreement with experimental observations, it predicts four ENDOR signals for each set of equivalent nuclei, denoted by NMR 1 to 4.

(Slide 5) The next member in the series of oligoradicals would be a tetraradical. We were interested in a tetraradical with tetrahedral symmetry for the following reasons. First, the scalar electron exchange interaction between all pairs of monoradical moieties should be equal, and therefore the formation of a thermally populated quintet state may be expected. Second, the zero field splitting should vanish for symmetry reasons, thus removing an unwanted source of relaxation. We prepared tetraphenylmethane tetrakisgalvinols, with and without 13C labelling. The tetrakisgalvinol can be converted to the tetraradical, but lower oxidation steps can also be generated. The slide shows the formula of the 13C labelled tetraradical, but first the unlabelled radicals shall be considered.

(Slide 6) From top to bottom, the slide shows the ESR and ENDOR spectra of the mono-, bi-, tri-, and tetraradical generated from unlabelled tetraphenylmethane tetrakisgalvinol. The resolution of the ESR spectra deteriorates progressively from the mono- towards the tetraradical; the tetraradical spectrum shows only an unresolved line. In contrast, the

ENDOR spectra of all species are well resolved. The calculated signal positions are indicated by triangles. Obviously all expected ENDOR signals can be observed, except those at the free proton Larmor frequency. In particular, the tetraradical gives four ENDOR signals; the fifth line, at the free proton frequency, is missing. This result establishes the presence of a quintet state. Considering the bi- and the triradical, it is clear that other oxidation steps are also present because of redox equilibria.

Additional evidence should be available from the ESR and ENDOR spectra of the 13C labelled tetraradical. (Slide 7) The ESR spectrum of the tetraradical exhibits in fact the expected quintet pattern due to hyperfine interaction with four 13C nuclei. The ENDOR spectrum shows again four proton signals, labelled by outlined triangles. In the frequency range shown, three 13C ENDOR signals are expected, indicated by full triangles. The two signals between 10 and 11 MHz could not be resolved experimentally. In the case of the bi- and the triradical, 13C labelling now allows a selective desaturation of the ESR; the field settings for the ENDOR experiments are indicated by arrows. It should be pointed out that the 13C ENDOR signal at about 8 MHz in the triradical spectrum is either missing or very broad. Actually, in each spectrum only one pair of 13C ENDOR signals is definitely present, with positions independent of the spin state.

(Slide 8) Energy levels for the tetraradical are given in the next slide. Five proton and five 13C ENDOR signals are predicted. The signals at the free nuclear Larmor frequencies are not observable, of course, and the 13C ENDOR signal at about 3 MHz could not be detected either.

(Slide 9) The experimental ESR spectrum of the biradical is actually dominated by contributions from the monoradical, and some amount of triradicals is also present. These contributions have been subtracted by computer manipulations, and the result is shown below the original spectrum. The difference spectrum now exhibits the expected triplet of nonets. However, the intensity pattern is not quite the binomial one (1:2:1). In order to account for this behaviour, computer simulations for different values of the scalar exchange integral J were performed. The best match to the experimental spectrum was obtained for J ≈ 600 MHz. Calculated spectra for other values of J are depicted on the right hand side.

(Slide 10) Similarly, a difference spectrum of the triradical has been generated. A computer simulation is depicted at the bottom left. A closer inspection of the tetraradical spectrum reveals an indication of proton hyperfine splittings. They can be emphasized by resolution enhancement techniques. A computer simulation is shown at the bottom.

(Slide 11) On the next slide, the ENDOR spectra of two 13C labelled triradicals are compared. The top spectrum belongs to the tetraphenylmethane series currently discussed, and the bottom spectrum is a copy of the spectrum from the third slide, belonging to 1,3,5-benzenetrisgalvinoxyl. Now the different behaviour with respect to the signal at about 8 MHz becomes evident. The fact that this signal is either absent or very

broad in the top spectrum may be attributed to the influence of the exchange interaction, which is not very much larger than the 13C hyperfine interaction.

(Slide 12) The last slide shows solid solution ESR spectra of the unlabelled oligoradicals. Whereas measurable zero field splittings are observed for the bi- and the triradical, the zero field splitting of the tetraradical is vanishingly small, in accord with the tetrahedral symmetry of this radical. In conclusion it can be stated that the presence of a thermally populated quintet state in a tetrakisgalvinoxyl with tetrahedral symmetry has been established by ESR and ENDOR spectroscopy. A value of about 600 MHz can be estimated for the exchange integral.