J. Chem. Soc., Faraday Trans. I , 1987,83, 69-75 An Electron Nuclear Double Resonance Study in a Glassy Matrix of Nitroxide Radicals with Delocalized Spin Density Marina Brustolon," Anna Lisa Maniero and Ulderico Segre Dipartimento di Chimica Fisica, Universita di Padova, Via Loredan 2, 35131 Padova, Italy Lucedio Greci Dipartimento delle Scienze dei Materiali e della Terra, Universita di Ancona, Via Brecce Bianche, 60131 Ancona, Italy Indolinone (l), benzimidazole (2) and quinoline (3) nitroxide radicals have the nitroxide function in a conjugated position with respect to the 71 system. Their electron nuclear double resonance (ENDOR) spectra in a [2H,]toluene glassy matrix have been detected and the principal values and the orientation of the proton hyperfine coupling tensors have been obtained by comparison with computer-simulated spectra.Typical features of ENDOR spectroscopy in frozen solution, and in particular its more general feasibility with respect to use in solution, are discussed. Proton electron nuclear double resonance (ENDOR) of organic radicals in solution is a useful tool for the determination of isotropic hyperfine coupling c0nstants.l However, in many cases the application of ENDOR spectroscopy in solution is severely limited by the low sensitivity of the technique, the ENDOR signal being in the best cases some fraction of the e.s.r. one. Moreover, to obtain the optimum ENDOR signal it is necessary to find the appropriate arrangement for a number of experimental conditions (e.g. solvent viscosity, concentration, intensity of the time-dependent fields), so that it is difficult to use this technique in solution as routinely as e.s.r.On the other hand, organic radicals trapped in solids can be routinely studied by the ENDOR technique, owing to its more favourable spin relaxation properties2 Many radicals giving very weak or undetectable ENDOR spectra in liquid solution can give strong signals when the solvent is frozen as a glass. In a previous paper it was shown that from proton ENDOR spectra of organic radicals in a polycrystalline or glassy matrix one can easily obtain the principal values of the hyperfine coupling tensors of the protons, together with information on the principal directions. The radicals studied were aliphatic nitroxide radicals, showing ENDOR spectra of methyl and methylene proton^.^ In this paper an ENDOR investigation in frozen solution is reported on organic radicals with delocalized spin density.The radicals studied ( l a d , 2, 3) have the nitroxide function in a conjugated position with respect to the molecular n system.* The hyperfine coupling tensors of all protons in the a position with respect to the carbon atoms bearing unpaired spin density (a-protons) have been measured, and in some cases the relative orientations of the nitrogen and proton hyperfine tensors have been obtained by computer simulation of the ENDOR spectra. Theory The spin hamiltonian for a nitroxide radical is given by 6970 Electron Nuclear Double Resonance Study where I , and A , are the nitrogen spin and the hyperfine (h.f.) tensor, and Ii and Ai are the spin and the h.f.tensor of the ith proton. The shape of the polycrystalline e.s.r. spectrum is given by the superposition of the lineshapes at different orientations, i.e. where B,(R) and p,.(Q) are, respectively, the resonance field and the probability of the rth transition for a radical at orientation n, andflx) is a shape function. The values of B, and pr can be computed for nitroxide radicals according to standard first-order expressions.' The nitrogen h.f. tensor A , gives the largest anisotropic interaction of the hamiltonian (l), so that it is convenient to choose the A , principal axes as the molecular frame, with the z axis along the nitrogen p-orbital direction. Since the tensor A , is quite anisotropic, it is possible to obtain selectively oriented ENDOR spectra of radicals embedded in glassy disordered samples.2 Simulated e.s.r.absorption spectra for radical (1 a) are shown in fig. 1 (a). They have been calculated by taking the magnetic field along the principal directions of the A , tensor. Note that, when Bo is parallel to z, the spectrum extends over a much wider range, and the three hyperfine components are well separated. The conventional first-derivative e.s.r. spectrum for a disordered sample is shown in fig. 1 (b). Absorption in the low- and high-field zones is mainly due to molecules with Bo nearly parallel to the z axis. It is therefore possible in principle to saturate these molecules selectively by taking the magnetic field equal to the low or high value: crystal-like ENDOR spectra can therefore be observed, provided that the saturation is not transferred by spin diffusion to molecules having different orientations.On the other hand, it is possible to saturate only the probes with Bo lying in the (x,y) plane by setting the magnetic field at the value B,, z 3410 G, as can be seen from fig. 1. The ENDOR signal for the ith proton is therefore given by3 where v is the radiofrequency, vi+(n) is the proton frequency corresponding to M, = f 1/2, g(x) is a shape function and w(n) is the orientation-dependent intensity factor given by a Gaussian distribution: w ( n ) = 2 exp { - [B- B,(R)]2/2a2}. T (4) The distribution width CT is taken to be equal to the e.p.r. inhomogeneous linewidth, (i = 2.G. Since each proton contributes additively to the ENDOR signal, the total lineshape is obtained as I(v) = z [z,+(v)+ri-(v)].( 5 ) i Hyperfine enhancement effects5 have not been included in the simulations, since the radiofrequency field used in the experiments is strong enough to saturate the nuclear transitions.6 The ENDOR spectra obtained for a model system of a proton with the hyperfine interaction principal axes parallel to the A , principal axes are reported in fig. 2. These calculations show the dependence of the lineshapes upon the parameter (i when the value of the magnetic field is chosen to pick up molecules with zllBo [fig. 2(a)] or molecules with z l B o [fig. 2(b)]. The positions of the ENDOR peaks are in any case determined by the principal values of the Ai tensor. Therefore, on picking out the three pairs of lines corresponding to the same proton, the proton hyperfine tensor Ai should be extracted straightforwardly from this type of ENDOR spectrum, and some infor- mation on its orientation should also be obtained.M .Brustolon et al. 71 3410 3 4 4 0 BIG Fig. 1. Computer simulated e.s.r. spectra for radical (1 a). The magnetic interaction values are as follows: g = (2.008, 2.005, 2.002); A,/MHz = (1 1, 1 1 , 65); for Ar see table 1. (a) Oriented ‘ single-crystal ’ spectra, with B,, aligned along the three principal axes; (b) ‘powder ’ first-derivative spectrum. , I I , 1 I I I I 1 -3 - 2 -1 0 1 2 3 - 3 -2 -1 0 1 2 3 ( Y -vH )/MHz Fig. 2. Computer-simulated ENDOR spectra for a model proton with non-axial hyperfine interaction in a nitroxide radical: g and A , are the same as in fig.1, A,/MHz = (2, 4, 6). (a) B = 3440.G, (b) B = 3410.G. a/G = (i) 2, (ii) 10 and (iii) 40.72 Electron Nuclear Double Resonance Study Experimental Nitroxides (l),' (2)* and (3)9 were prepared as described in the literature. The concen- tration of each sample was ca. mol drnp3. E.s.r. and ENDOR spectra were recorded with a Bruker ER 200 spectrometer equipped with a 300 W radiofrequency amplifier. (la): R2 = Ph, R5 = R' = H (lb): R2 = Me, RS = R7 = H Ph 1 I a Ph Ph I (lc): R2 = Ph, RS = H, R7 = OBuf (Id): R2 = Ph, Rs = OBut, R7 = H 0 (3) Results In fig. 3 the e.s.r. and ENDOR spectra for radical (1 a) in frozen [2H,]toluene are reported as an example. ENDOR spectra of comparable intensity were detected for the other radicals.We tried without success to detect ENDOR signals for the same radicals in [2H,]toluene fluid solution, at different concentrations and on varying all the relevant experimental parameters. When the magnetic field is set on positions A and B of the e.s.r. spectrum, markedly different ENDOR spectra are obtained. According to the previous discussion, spectrum A [in fig. 3 (a)] is given by the pairs of lines due to the z components of the proton tensors A , ; spectrum B [in fig. 3(b)] results instead from the x, y components and has twice as many lines as spectrum A. The ENDOR spectra for the radicals studied here are all given by a pattern centred at the free-proton frequency. Some general considerations help in obtaining the proton hyperfine tensors reported in table 1 from these spectra: (i) a pair of lines can be attributed to the lowest, intermediate or highest principal values (A,,A,,A,) on the basis of their lineshapes (see fig.2 ) ; (ii) if the isotropic hyperfine splitting Aiso is known, the (4) relations hip can be exploited; (iii) the dipolar tensor for an a-proton is generally determined mainly by the spin density on the bonded carbon atom, and therefore Tr A = ( A , + A , + A,)/3 = Aiso A,:A,:A, z 1 : 2 : 3 . (7) The hyperfine tensors obtained are reported in table 1. Discussion Radicals (1 a-d) The isotropic splittings for protons 4-7 are known from the e.s.r. spectra in CHCl,:' Ais0(5, 7) = 8.5 MHz and AiS0(4, 6) = 2.8 MHz for (1 a); similar values are obtained for (1 b). The four radicals have similar hyperfine tensors, and the traces of the hyperfine tensors for proton pairs 5,7 and 4,6 are in good agreement with the isotropic splittings obtainedM .Brustolon et al. 73 I . . , , , * . b , . ~ V V . 0 2 4 6 0 2 4 6 Fig. 3. E.s.r. (insert) and ENDOR spectra for radical (la) in [2H,]toluene at T = 105 K. (a) and (b) ENDOR obtained at positions A and B of the e.s.r. spectrum, respectively. Upper and lower traces correspond, respectively, to experimental and simulated ENDOR spectra. The simulated ENDOR spectra have been computed with the same parameters as in fig. 1 and CT = 2. G. The ENDOR lines near vH are due to the phenyl protons, which have not been included in the simulations. (v-vH)/MHz (v-vH)/MHz Table 1. Proton hyperfine tensor principal values = A j - & z= TrAa - radical proton Tj/MHz A/MHz l a 4, 6 1.7 5 4.9 7 2.5 l b 4, 6 1.7 5 4.9 7 2.5 l c 4, 6 1.7 5 5.0 I d 4, 6 1.5 7 2.5 2 4 1.1 5 6.1 7 3.6 3 3 2.0 5, 7 1.6 6 4.9 8 2.9 -0.2 0.1 -0.5 - 0.2 0 -0.5 -0.3 0.2 -0.1 -0.7 0.1 0.2 - 1.8 0.1 -0.1 0.5 0.6 - 1.5 - 5.0 -2.1 - 1.5 -4.9 - 2.0 - 1.4 - 5.2 - 1.4 - 1.8 - 1.2 -6.3 - 1.8 -2.1 - 1.5 - 5.4 - 3.5 2.8 - 8.9 - 8.3 2.9 -9.0 - 8.4 2.6 2.8 2.3 - 8.9 - 8.'2 - 10.7 - 9.8 - 3.9 2.9 - 8.9 - 8.0 a The signs are based on the spin density distribution assumed for the calculations.from the e.s.r. spectrum. The principal values for protons 5 and 7 are markedly different owing to the stronger dipolar interaction of proton 7 with the spin density on the N-0 group. Note that the relationships (7) do not hold for the principal values of proton 7, whereas they hold approximately for protons 4, 6 and 5 , which interact mainly with the spin density on the corresponding carbon atom.74 Electron Nuclear Double Resonance Study Table 2.Calculated proton tensorsa in radical (1 4 proton T/MHz direction cosinesb 4 1.50 0.42 - 1.92 5 4.40 - 0.20 - 4.20 6 1.70 0.40 -2.10 7 3.93 - 1.78 -2.15 0.3453 0 0.7073 0 0.7068 0.5585 0 0.7783 0 - 0.9385 0.8294 - 0.6279 0.9385 0.3453 0 0.7068 0 0.8294 0.5585 0 -0.7073 -0.6279 -0.7783 0 0 0 1 0 1 0 0 0 1 0 0 1 a The spin density distribution used in the calculation is as follows: pN = po = 0.310, psa = p7& = p3 = po = 0.050 (see text). In the molecular frame (x along the N-0 bond, z along the nitrogen p-orbital). p4 = p6 = - 0.044, p5 = 0.139, p7 = 0.130, In fig.3(a) the features due to protons 4, 6, 5 and 7 are indicated. They correspond, respectively, to the lowest (A4,61 = 1.3 MHz), the intermediate (A,, = 8.9 MHz) and the maximum (A7h = 10.4 MHz) principal values of the corresponding tensors. Since only these features appear for the protons, the conclusion must be drawn that the z axis of the nitrogen 2p orbital is nearly parallel to the principal directions corresponding to A4,61, A,, and A7h, as discussed in the previous section. To support our interpretation, we have performed a McConnell-StrathdeelO calculation for protons 4-7 in radical (1 a) by using a planar molecular geometry as determined by the crystal struct~re.~ The spin densities for nitrogen and oxygen have been assumed to be equal, and have been estimated by comparing the nitrogen isotropic splitting for radical ( l a ) with the corresponding average value (43 MHz) for nitroxide radicals with localized spin density on the N-0 gr0up.l' The spin densities for protons 4-7 have been obtained semiempirically from the hyperfine splitting values by making use of the McConnell relation with Q = -64 MHz.Negative spin densities have been taken on C(4) and C(6), in agreement with INDO calculation^.^ The residual spin density has been considered as equally distributed on the 'blind' positions C(3), C(3a), C(7a) and O(C-0). The calculated principal values and principal directions are reported in table 2. No attempt has been made to optimize the calculated values on adjusting the spin density distribution or the semiempirical parameters.The calculated values are in reasonable agreement with the experimental ones, and the principal directions corresponding to A4, 61, A,, and A7h are parallel to the z axis, as found experimentally. Radicals (2) and (3) The isotropic splittings for these two radicals are in good agreement with those obtained by e.s.r. spectroscopy.s Protons 5 and 7 in (2) have similar Aiso values being ortho and para with respect to the N-0 group, but they have different dipolar tensors owing to their different positions with respect to the same group. The same observation holds for protons 6 and 8 in (3). The considerations which have been made for radicals (1) areM. Brustolon et al. 75 also relevant for these species, since the same trend is found for the coupling tensors, as is reported in table 1.Some general considerations are worth mentioning. The detection of six ENDOR lines for each nucleus, instead of two as in liquid solution, complicates the analysis of the spectrum. On the other hand, more information is obtained from rigid-matrix spectra, since the dipolar tensor depends not only on the local spin density, as AiSO, but also on the relative position of the proton with respect to the other atoms bearing some spin density. Therefore an ENDOR investigation in frozen solution should be tried even when the ENDOR spectrum in fluid solution is obtainable, as an aid in the attribution of the hyperfine parameters to the different protons, and in order to obtain information on the total spin distribution on the radical.Conclusions We have shown that the determination of the dipolar tensors for a-protons in aromatic radicals in frozen solution by the ENDOR technique can be made relatively easily. ENDOR investigations in frozen solutions therefore appear to be a useful alternative technique to ENDOR in fluid solutions when the latter is difficult to perform. Moreover, the information obtainable in frozen solution on the dipolar hyperfine coupling of the protons can be exploited in determining the position of the proton in the radical, together with the spin distribution on the conjugated system. This work was supported in part by the C.N.R. through its Centro Studi sugli Stati Molecolari Radicalici ed Eccitati, and in part by the Minister0 della Pubblica Istruzione. References 1 N. M. Atherton, Electron Spin Resonance (Ellis Horwood, Chichester, 1973). 2 L. Kevan and P. A. Narayana, in Multiple Electron Spin Resonance, ed. M. M. Dorio and J. H. Freed 3 M. Brustolon, A. L. Maniero and U. Segre, Mol. Phys., 1985, 55, 713. 4 R. Benassi, F. Taddei, L. Greci, L. Marchetti, G. D. Andretti, G. Bocelli and P. Sgarabotto, J . Chem. 5 D. H. Whiffen, Mol. Phys., 1966, 10, 595. 6 J. H. Freed, J. Chem. Phys., 1965,43, 2312. 7 C. Berti, M. Colonna, L. Greci and L. Marchetti, Tetrahedron, 1975, 31, 1745; L. Greci, Tetrahedron, 8 C. Berti, M. Colonna, L. Greci and L. Marchetti, J. Heterocycl. Chem., 1979, 16, 17. 9 C. Berti, M. Colonna, L. Greci and L. Marchetti, Tetrahedron, 1976, 32, 2147. (Plenum, New York, 1979). SOC., Perkin Trans. 2, 1980, 786. 1982, 38, 2435. 10 H. M. McConnell and J. Strathdee, Mol. Phys., 1959, 2, 129. 1 1 L. J. Berliner, Spin Labelling (Academic Press, New York, 1976). Paper 6/328; Received 17th February, 1986