J. Chem. SOC., Faraday Trans. I , 1987, 83, 57-68 Electron Nuclear Double Resonance of S = 1/2 Defects in a Single Crystal of the Morpholinium-TCNQ 1 : 1 Complex Anna Lisa Maniero, Ornella Priolisi and Carlo Corvaja* Department of Physical Chemistry, University of Padova, Padova, Italy Single crystals of the rnorpholinium-TCNQ 1 : 1 complex give e.s.r. spectra showing a central resonance at g = 2 in addition to the lines of thermally excited triplet excitons. ENDOR spectra recorded by saturating the central line show that it is due to isolated TCNQ anions which are produced at two different crystal defects. The hyperfine tensors of the TCNQ anion obtained from ENDOR give a spin density distribution in the solid much different from that in solution. The polarization of the spin distribution is consistent with a rearrangement of the morpholinium cations around the anion in the defect.The 1 : 1 morpholinium (M) salt of the TCNQ radical anion crystallizes with the TCNQ molecules stacked in columns with alternate large and small distances between them.l The morpholinium cations occupy lattice positions between the stacks. In such a crystal structure the ‘molecular unit’ in the crystallographic unit cell is a (TCNQ)2- dimer having a singlet ground state and a thermally accessible triplet state. The triplet excitation migrates through the crystal and constitutes a mobile triplet exciton whose e.s.r. lines are very narrow (200 mG), any hyperfine interaction being averaged out by the fast motion. Besides triplet exciton lines, the e.s.r.spectra of the M-TCNQ salt show lines due to spin-1/2 species which are the result of crystal defects or impurities. Much speculation has appeared in the 1iteratu1-e~~~ as to the nature of this species and its temperature behaviour.2v 4 7 Little can be said from the only analysis of the e.s.r. spectrum, since no hyperfine structure is resolved, and the small anisotropy of the g factor is difficult to correlate witb the molecular structure. In this work we have used ENDOR spectroscopy6 to investigate the nature of the spin-1/2 defect in M-TCNQ. Owing to the inherently high spectral resolution of ENDOR we were able to measure several hyperfine couplings of this species which in the e.s.r. spectrum give rise only to an inhomogeneous broadening. We have determined by ENDOR the complete hyperfine tensors (isotropic and anisotropic parts) of eight protons which were found present in two different defects both containing TCNQ.This knowledge allows us a detailed insight of the structure and of the electron spin distribution of the paramagnetic species. A knowledge of the spin distribution is also of interest in relation to another important aspect. The fine-structure parameters D and E of the triplet exciton spin hamiltonian DS; + E(S: - S i ) are often calculated by taking the (TCNQ); dimer geometry as found by X-ray diffraction studies and an appropriate spin density distribution.’ The latter is obtained by assuming it to be the same as for the TCNQ- anion in liquid solution. The discrepancy between calculated values (always larger than experimental values) was attributed to delocalization of the triplet state over more than two molecules*~ or by assuming a triplet species formed by the more distant TCNQ neighbours instead of the near neighbours.1° Flandrois and Boissonade7 proposed that the discrepancy had to be attributed to a different spin density distribution in the solid.They showed that there is a quite large variation of calculated zero field splitting parameters on different choices 5758 ENDOR Spectra of the Morpholinium-TCNQ Complex of the spin density distribution. Their proposal is relevant in relation to another problem, the determination of the charge-transfer character of the charge-transfer complexes in their excited triplet states from the measured hyperfine coupling constants.ll7 l2 The charge-transfer character is usually obtained from the ratio of the hyperfine coupling of the complex to that of the uncomplexed molecule.This procedure is correct only when there is no spin redistribution in the solid owing to the charge interaction. Finally, there is another need for accurate data concerning the spin densities and isotropic coupling in crystals of TCNQ salts. Their knowledge is a prerequisite for obtaining information on the exciton motion and its dimensionality through measure- ments of nuclear spin relaxation times, T.l3 Experimental Single crystals of morpholinium-TCNQ complexes were obtained by the method of Melby et a1.14 Reddish-purple prisms of the 1 : 1 complex were grown from acetonitrile solutions with molar ratios of TCNQ to morpholinium iodide between 0.3 and 0.4.The solutions were prepared with acetonitrile that had previously been distilled under nitrogen and kept in a nitrogen atmosphere, with due care being taken to minimize the presence of water and oxygen.l59 l6 TCNQ was repeatedly vacuum-sublimed while the morpholinium iodide, prepared from morpholine and HI, was purified by recrystallization from ethanol. M-TCNQ crystallizes in the triclinic system, space group P I , with two molecules in the unit cell of dimensions a = 9.54 A, b = 8.78 A, c = 10.60 A, a = 98.4", j? = 122.9", y = 93.5O.l Single crystals of M-TCNQ were mounted inside a small Plexiglas cube to facilitate their mounting on a gonionieter rod along three orthogonal axes. E.s.r. and ENDOR spectra were recorded at 300 K for magnetic-field directions in the three orthogonal planes.The spectrometer was a Bruker ER 200 D X-band spectrometer equipped with an ENDOR accessory and an EN1 300 W r.f. amplifier. TRIPLE resonance17 experiments were performed using a Systron Donner 5000 A r.f. sweeper to supply the additional r.f. field at the ENDOR transition frequencies. The r.f. frequency was measured by a Hewlett-Packard 5342A frequency counter. Results The e.s.r. spectra of the M-TCNQ complex have already been discussed, as well as the variation of intensity and linewidth with The room-temperature e.s.r. spectra show a pair of motionally narrowed lines due to thermally activated triplet excitons. The magnitudes of the zero-field splitting parameter, D and E, determined from the angular dependence of the line separation, agree with the previously reported values.2T In addition to the triplet lines a central resonance is present composed of two components, a broad one (AH from 3 to 12 G depending on orientation) and a narrow one (AH = 0.8 G).Both have a thermally activated intensity.2 ENDOR experiments were performed by setting the magnetic field at the centre of the central e.s.r. resonance where both components are saturated, even though the ENDOR spectra are due to the broad component only. In fact, the same ENDOR spectrum, although weaker, was obtained by setting the magnetic field at the wings of the broad resonance where the narrow one does not contribute to the e.s.r. intensity. A typical ENDOR spectrum in the range 10-20 MHz in the region of the protons' absorption is shown in fig.1. It consists of eight lines whose frequency dependence on the crystal orientation with respect to the magnetic field was fitted to the equation18 v2 + - 1 + cos2 # + B , sin2 # - 2C+ sin # cos 41. - - - -A . L. Maniero, 0. Priolisi and C. Corvaja 59 Fig. 1. ENDOR spectra of an M-TCNQ single crystal for arbitrary orientations of the magnetic field. The spectra are due to protons (1&20 MHz) and to nitrogen nuclei (4-9 MHz). The spectra were recorded by saturating the broad central line of the e.s.r. spectrum. A + , - - B , and C+ - are related to the hyperfine tensor elements by the equations where the indexes i andj (i, j = x, y , z ) refer to a particular plane where the magnetic-field direction is rotated.vH is the free proton frequency at the given magnetic field intensity. Fig. 2 shows the variation of the ENDOR frequency with the crystal orientation. Eight tensors were determined whose principal values and directions are reported in table 1. In addition to the abovementioned ENDOR lines, the spectra show other lines with smaller (ca. 1 MHz) hyperfine separation. These are due to protons not directly bonded to the paramagnetic centre, and their analysis has not yet been made in detail. Moreover, many other lines were recorded with frequencies ranging from 2 to 10 MHz (see fig. 1). Their intensity is strongly dependent on the crystal orientation. These resonances are to be attributed to the hyperfine interaction of the unpaired electron with the N nuclei of the CN groups.Discussion Models of Spin-1/2 Defects The central resonance line in the e.s.r. spectrum of M-TCNQ is typical of many TCNQ The concentration of the species giving rise to this anomalous resonance appears to be a characteristic of the salt and not of the preparative method.lg Thus the central resonance is not to be simply ascribed to a doublet-state chemical impurity, since the60 ENDOR Spectra of the Morpholinium-TCNQ Complex Fig. 2. Frequencies of the proton ENDOR lines for magnetic field orientation in three orthogonal planes. Only half the experimental points are shown in the figure. Table 1. Principal values and directions of protons in the I and I1 defects in M-TCNQ salt proton . Ia Ib Ic Id IIa IIb IIC IId isotropic value/MHz - 6.29 - 5.35 -2.10 - 2.20 -5.14 - 5.29 - 3.00 - 2.54 anisotropic principal values/MHz direction cosines x Y -7 4.05 -2.58 - 1.47 3.63 - 2.00 - 1.63 2.52 - 0.88 - 1.64 2.62 - 0.92 - 1.70 3.53 - 2.00 - 1.53 3.67 -2.19 - 1.48 2.92 - 1.31 - 1.61 2.69 - 1.01 - 1.68 0.1646 0.7477 0.2582 0.6820 0.0640 0.6006 0.2035 0.6175 0.1073 0.6984 0.2884 0.601 5 0.0808 - 0.6028 0.7938 0.22 10 - 0.6540 0.7235 - 0.6433 - 0.6843 - 0.7970 - 0.7598 - 0.7076 -0.7450 0.6266 - 0.5879 -0.5 173 - 0.8338 - 0.2005 - 0.5 1 44 0.6752 0.4777 0.5620 0.1947 0.5148 0.6226 -0.8349 - 0.602 1 - 0.4999 -0.8364 - 0.2205 -0.5018 0.6598 0.6293 0.4 107 0.28 12 0.5036 -0.8169 -0.7618 -0.3180 - 0.5644 0.4880 - 0.7034 - 0.5 169 - 0.7348 0.3695 0.5687 0.51 14 0.6203 0.5948 -0.7752 -0.3869 - 0.4994 0.466 1 - 0.7679 -0.4395 -0.7471 0.4905 0.4486 0.5328 0.7023 0.472 1A .L. Maniero, 0. Priolisi and C. Corvaja 61 latter would be expected to be present in a concentration strongly dependent on the preparation of the crystals. This observation favours the hypothesis that the paramagnetic species are intrinsic crystal defects. Also, the fact that the intensity of the line increases with increasing temperature2* and that this variation is reversible, points to the presence of intrinsic crystal defects. Buckman et al. measured the variation of the e.s.r. linewidth with the crystal orientation and they found the same linewidth anisotropy as for the triplet excitons.20 Since the linewidth of the latter ones is due to unresolved hyperfine structure the authors suggested that the central e.s.r.line derived from a species, possibly TCNQ-, with the same orientation as the TCNQ, dimers. Several models have been proposed in order to describe the defects. Their validity is examined in the following and checked on ihe basis of the ENDOR results. (1) The defects consist of isolated TCNQ, dimers with a single unpaired electron and unit negative charge. These species could be formed from disproportionation according to the following equation :5 2[TCNQ];- e [TCNQ]; + TCNQ2- + TCNQ-. In this hypothesis the obvious question arises as to the role of the two individual molecules in the dimers : are both molecules equivalent, or in other words is the unpaired electron evenly shared by the two? If not, how much is the electron spin transfer from one molecule to the other? (2) The defects consist of isolated TCNQ- molecules. In addition to the above equation, TCNQ- radicals could arise as solitons.These may be formed in the alternating chain of large and small distances between the TCNQ planes as discussed later in more detail. (3) Another model which was proposed for the species giving the central resonance is that of Wannier-Mott excitons, elementary excitations with loosely bonded electrons and holes.21 The linewidth would be due to unresolved electron-electron dipolar interactions. The ENDOR experiments immediately rule out this model, since they show that the cause of the e.s.r. linewidth is a hyperfine interaction with a limited number of nuclei. (4) Quite recently it was shown that the powder spectra of weakly perturbed triplet states, such as the triplet excitons in the complex (&-AsCH,t)(TCNQ);, are dominated by a sharp central line which is derived from the effects of Heisenberg spin exchange.22 The authors also suggested that in single crystals of other materials the central line could have the same origin and be due to the presence of amorphous regions in the crystals.This interpretation is discarded by the ENDOR experiments for the same reason as in point (3). Models (1) and (2) remain for discussion in the following section. The Structure of the Spin-1/2 Defects The ENDOR spectra of M-TCNQ single crystals in the proton region, recorded by saturating the central resonance line for an arbitrary orientation of the magnetic field, consist of eight pairs of lines due to the hyperfine interaction of eight different protons.Although by using ENDOR one cannot determine the number of nuclei giving rise to a particular resonance line, we exclude the assumption that more than one proton contributes to each ENDOR line on the basis of the determined values of the spin densities, as we will show later. The eight protons do not belong to the same paramagnetic species, as we showed by general TRIPLE res0nance.l' This experiment consists of irradiating the spin system with three radiation fields. One saturates an e.s.r. transition and at the same time irradiates the sample with r.f. radiation at the frequency of an ENDOR line and then observes the effect of a variable-frequency third r.f. field. The latter has an effect only when its frequency matches the value corresponding to62 END0 R Spectra of the Morpholinium-TCNQ Complex Table 2.Experimental spin densities" on C-H bonded carbon of TCNQ- ion radical and of I and I1 defects H H H H TCNQ- I I1 ref. (24) proton positionb Pi Pi Pi ~~ . -~ a 2 0.0948 0.0775 0.0662 b 6 0.0806 0.0797 0.0662 C 5 0.03 17 0.0452 0.0662 d 3 0.033 1 0.0383 0.0662 ZPii 0.2402 0.2407 0.2648 a From e.s.r. and ENDOR hyperfine data, with pi = laHl/lQ&I, taking Q& = -23.7 G = 66.36 M H z . ~ ~ For attribution, see text. another ENDOR line of a nucleus coupled to the same unpaired electron. In fact only in this case does the third r.f. field induce transitions in the same manifold of levels. In this way we were able to distinguish two sets of four protons.The pumping of one ENDOR transition has an effect only on the ENDOR lines of the protons in the same set. The hyperfine tensors obtained from an analysis of the variation of the ENDOR frequencies with the crystal orientation and reported in table 1 are accordingly collected in two different sets (I and 11). The ENDOR lines of the protons of set I1 are always (at all crystal orientations) lower in intensity by a factor of ca. Q than those of set I. Although the ENDOR line intensity is not in general a direct measure of the concentration of the species involved, as the two defects are expected to have similar relaxation properties, this indicates that the ratio of the concentrations of the two species is ca. 3 : 1 . Table 1 shows that all proton tensors have a principal axis (the last given in table 1) along the same direction.This one turns out to be almost parallel to the principal axis of the triplet exciton electron-electron dipolar interaction corresponding to the largest eigenvalue. This is perpendicular to the planes of the TCNQ molecules. The above observations support the idea that both defects I and I1 consist of TCNQ- molecules with the same orientation as those in the dimer stack. Further support for this comes from the consistency of the following analysis and from the fact that nitrogen ENDOR lines were also recorded. The isotropic hyperfine couplings of ring protons are related to the 71.-carbon spin density by the McConnell equation, a = Qp.23 With the assumption of a Q value of 67 M H z , ~ ~ the spin densities reported in table 2 are obtained.For comparison the values for a free TCNQ anion in solution are also reported in the same table. If one considers the total spin density on the four ring positions one obtains for both I and I1 almost the same value as for the free TCNQ anion. The assumption of a defect structure containing two equivalent TCNQ molecules connected by an inversion centre, and that the ENDOR lines are due to the couplings of pairs of magnetically equivalent protons, would give an unrealistically high total spinA . L. Maniero, 0. Priolisi and C . Corvaja 63 density on the eight (four for each molecule) 2, 3, 5, 6 positions. Moreover, if the ENDOR lines were assigned to pairs of protons the e.s.r. linewidth would be larger than the measured one.In this way we exclude the symmetric dimer model and come to the conclusion that the defects giving rise to the broad e.s.r. line are isolated TCNQ anions; if another (or more) TCNQ molecule is involved in the formation of defects I and I1 the spin density transfer from one molecule to the other(s) should be very small. Spin Density Distribution Inspection of table 2 shows that the spin distribution is strongly polarized. Before coming to the possible cause of this polarization we first examine how the different couplings and spin densities are to be attributed to the single positions. This is done on the basis of the anisotropic part of the hyperfine tensors. The anisotropic hyperfine tensor of protons in a n-radical can be computed from a known spin density distribution.The tensor elements have contributions from the 'local' spin density on the carbon atom bonded to the considered proton, as well as 'non-local' contributions coming from spin densities on the other n-centres. McConnell and Strathdee have given the expressions far the dipolar interaction of a proton with the unpaired spin density in a Slater p - orbital. 25 For the local contributions better results are obtained by taking Ti = 35.0 MHz, qj = 5.3 MHz and Tkk = -40.4 MHz, where i and j are, respectively, two perpendicular axes along the C-H bond and along the p-orbital, with k completing the orthogonal set.26 Comparing the principal values and the directions of the calculated anisotropic tensors with the experimental values of table 1, one has a test of the goodness of the assumed spin distribution and of the correctness of the attribution of the couplings to the individual positions.A first guess for the attribution is based on the fact that at least for the positions bearing the largest spin density we expect the anisotropic tensor to be to a large extent dominated by the local contribution and therefore to have one principal axis approximately along the C-H bond. This principal direction should correspond to the largest eigenvalue. This consideration and the data of table 1 allow us to assign protons a and b to positions 2 and 3 (case A) or 2 and 6 (case B), excluding the other possibilities (case C) (fig. 3) since the corresponding principal directions are not parallel, making an angle of ca.150" (or 30"). On the other hand, these principal directions are almost parallel (within 5") to the corresponding directions of the protons c and d. This is true for both defects I and 11. We have performed calculations of the dipolar tensor elements for the two cases A arid B by taking the spin density at the carbon centres bonded to a proton as given by the isotropic coupling constant and the McConnell equation. For the spin densities on the other centres we had to rely on the values calculated for the TCNQ anion by semi-empirical methods24 (set 1) or by an ab initio method27 (set 2) (table 3). In order to facilitate a discussion of the results of the calculations, we denote by ti the principal directions of the dipolar tensor of the proton i corresponding to the largest principal value.For a 7~ free radical with unit spin density on the carbon atom, as mentioned before, t corresponds to the C-H bond direction. When the spin density is also present on other carbon atoms the 5 direction is rotated away from the CH bond; this is why the experimental tensors have directions making angles other than 0 and 60" (120") as is the case for the C-H bond directions. Both choices of spin distrib~tions,~~? 27 when applied to model A, give dipolar tensors with reasonable principal values, but the angles are far from the experimental ones. In particular, the calculated angle between rz and c5 and that between r3 and (close to zero according to the experiments) are too large. Apparently there is no way to improve the situation 3 FAR 164 ENDOR Spectra of the Morpholiniurn-TCNQ Complex A B C Fig.3. Different models of spin density distributions for TCNQ- in M-TCNQ according to the possible attributions of the proton tensors. The dimensions of the circles indicate the magnitudes of the spin density on the carbon atoms. by a slight change in the spin density distribution. Conversely, model B gives satisfactory results not only for the principal values but also for the angles between the principal axes with both spin density distributions. We note that the largest coupling tensors were best accounted for by the spin densities of set 1, while the smallest ones were best accounted for by the other set of spin densities, which involves a larger spin density on C4. An improvement is therefore obtained by allowing a change in the spin density distribution, and the best agreement is obtained by using set 3 shown in table 3.The principal values of the proton dipolar tensors are reported in table 4, while the angles between the principal directions of the tensors are given in table 5. Formation of TCNQ- Defects in M-TCNQ Crystals Isolated TCNQ- radicals in the M-TCNQ crystal may arise as solitons in the alternating chain of large and small distances in the TCNQ stack, in a manner similar to that in a polyacetylene polymer chain.28 The model is schematically shown in fig. 4. It is worth mentioning that in an isolated polyacetylene chain the ground state is degenerate, and this fact allows for a certain freedom of movement of the soliton.In the case of a stack of charged TCNQ molecules there is a strong interaction with the surrounding cations not symmetrically placed around the TCNQ anion (TCNQ is not placed at an inversion point in the crystal lattice).l This interaction removes the degeneracy of the ordering of the short and large distances and freezes the motion.A . L. Maniero, 0. Priolisi and C. Corvaja 65 Table 3. Spin densities on the n-centres of the TCNQ- radical ~ set 3" atom set l a set 2b I I1 set dd 1 0.0494 2 0.0539 3 0.0539 4 0.0494 5 0.0539 6 0.0539 7 0.2245 8 0.2245 9 0.0126 10 0.0126 11 0.0 126 12 0.0126 13 0.0465 14 0.0465 15 0.0465 16 0.0465 0.132 0.061 0.061 0.132 0.061 0.061 0.180 0.180 0.002 0.002 0.002 0.002 0.03 1 0.03 1 0.03 1 0.03 1 0.048 0.094 0.033 0.108 0.032 0.08 1 0.218 0.190 0.0 12 0.002 0.012 0.002 0.045 0.039 0.045 0.039 0.048 0.077 0.038 0.108 0.045 0.080 0.218 0.190 0.0 12 0.002 0.012 0.002 0.045 0.039 0.045 0.039 0.064 0.126 0.001 0.132 0.006 0.087 0.281 0.075 0.029 0.00 1 0.035 0.00 1 0.06 1 0.0 19 0.063 0.0 19 a Ref. (24).agreement with experimental tensors (see tables 4 and 5). densities calculated with McClelland method. Ref. (27). Spin densities which give the best Spin Table 4. Calculated principal values of proton dipolar tensor in the I and I1 defects in M-TCNQ salt proton T/MHz proton T/MHz Ia 4.11 IIa - 2.65 - 1.46 Ib 3.71 IIb -2.16 - 1.55 Ic 2.68 IIc -0.52 -2.16 Id 2.71 IId -0.51 -2.20 3.58 - 1.98 - 1.60 3.69 - 2.09 - 1.60 3.08 - 1.01 - 2.07 2.88 -0.77 -2.11 Furthermore, the cations are expected to rearrange their position around the defect. It is not surprising that two different arrangements, I and 11, can arise because of the low symmetry of the crystal lattice.The polarization of the spin distribution reflects the strength of the electrostatic cation interaction. This is more effective for defect I than for 11. The e.s.r. measurements2 have shown that the defect concentration is almost constant for temperatures <250 K and increases with increasing temperature in the range 250-350 K. The same behaviour is shown by the ENDOR lines intensity. At 425 K a phase transition occurs with a sudden decrease in e.s.r. intensity. Note that above the 3-266 ENDOR Spectra of the Morpholinium-TCNQ Complex Table 5. Matrix of the angles (in ") between the < principal directions of the calculated dipolar tensors (values obtained from the experimental data are in parentheses) defect I <, 131 (148) 6(7) 46(28) t b 42 (25) 3 (3) t C 140 (158) defect I1 5, 131 (148) 3(3) 43 (26) 5 6 45 (29) 6(6) t c 140 (157) t defect \ Fig.4. Models which show a defect of solitonic nature in trans-polyacetylene (a) and in a chain of TCNQ molecules stacked in dimeric units (b). transition temperature the e.s.r. spectra retain almost the same characteristics as for the low-temperature phase. In particular, triplet exciton lines are observed in both phases with very similar zero field splitting parameters.2 Analogous behaviour cannot be observed by ENDOR because the ENDOR lines at temperatures ~ 3 2 0 K are broader and become undetectable above 350 K.No further information is available on the nature of the phase transition; however, the abovementioned observations indicate that only minor changes in the crystal properties should be involved. This allows us to suggest that little energy is needed to create a defect in the crystal lattice, or regions with the molecules arranged as they are in the phase that is stable at high temperature. In this way another hypothesis can be formulated as to the origin of the isolated TCNQ- radicals which is in line with the temperature behaviour of the concentration. The defects may be attributed to TCNQ- radicals present at the boundary between the regular lattice and the defective region.A . L. Maniero, 0. Priolisi and C . Corvaja 67 Polarization of the Spin Density Distribution The polarization of the spin density in the TCNQ- anion radicals, which is caused by the electrostatic interaction of the cations, resembles those of alkali-metal radical ion pairs in solutions of low dielectric constant.When the radical anion contains strongly charged groups, as occurs for example in semiquinones, the alkali-metal cation is situated close to the negative charge and polarizes the spin distribution, making the latter asymmetric. Calculations of the latter have been carried out with success following a model proposed by M~Clelland.~~ Some recent results on the K+TCNQ- salt in tetrahydrofuran and dimethoxyethane as solvent seem to indicate that TCNQ- also forms asymmetric ion pairs.3o This would be demonstrated by a linewidth alternation in the e.s.r.spectra observed for temperatures lower than 249 K in DME, and produced by the K+ cation undergoing a jumping process between two equivalent potential minima. However, we observed only two very narrow lines (Av = 50 kHz) separated by 3.97 MHz (1.417 G) in the ENDOR spectrum of K+TCNQ- in liquid DME for temperatures (T = 190 K) well below that indicated as the onset of linewidth alternation. This fact means that the proposal of an asymmetric structure for TCNQ- in solution deserves further attention. According to an X-ray diffraction study, the TCNQ anion in M-TCNQ has four neighbouring M+ cations. Calculations of the spin distribution were performed by the McClelland method by considering the limiting situation where two cations close to the same C(CN), group of the TCNQ anion are absent.The results are reported in table 3 (set 4). Note that the spin densities have the same trend as those we found the best when calculating the dipolar tensor. Finally, we should mention that up to now there is no information on the nature of the species giving rise to the narrow resonance accompanying the broad one. Conclusions ENDOR spectra obtained by saturating the central broad resonance line of the M-TCNQ crystal e.s.r. spectrum show that this line is due to two different spin-1/2 defects with unresolved hyperfine structure. The defects consist of isolated TCNQ- molecules with a strongly polarized and asymmetric spin distribution, thus confirming the possibility of a different spin distribution in the solid with respect to the solution.This was invoked to explain the zero field splitting parameters of thermally excited triplet excitons in TCNQ salts. The defects are suggested to arise at the boundary regions between zones of the crystal lattice with different molecular arrangements. We thank Prof. Renato Bozio of this department for many helpful discussions. This work was supported by the Italian National Research Council (CNR) through its Centro di Studio sugli Stati Molecolari Radicalici ed Eccitati and in part by the Minister0 della Pubblica Istruzione. References T. Sundaresan and S. C. Wallwork, Acta Crystallogr., Sect. B, 1972, 28, 3507. 3. C. Bailey and D. B. Chesnut, J . Chem. Phys., 1969,51, 51 18. D. B. Chesnut and W. D. Phillips, J. Chem. Phys., 1961, 35, 1793. S.Nespurek, J. Pilar, P. Schmidt, M. Sorm, U. Langer, A. Graja and E. Sopa, Chem. Phys., 1986,101, 81. S. K. Hoffmann, P. J. Corvan, P. Singh, C. N. Sethulekshmi, R. M. Metzger and W. E. Hatfield, J . Am. Chem. Soc., 1983, 105, 4608. L. Kevan and L. D. Kispert, Electron Spin Double Resonance Spectroscopy (John Wiley, New York, 1976). S. Flandrois and J. Boissonade, Chem. Phys. Lett., 1978, 58, 596.ENDOR Spectra of the Morpholinium-TCNQ Complex 8 T. Hibma, P. Dupuis and J. Kommandeur, Chem. Phys. Lett., 1972, 15, 17. 9 T. Hibma, G. A. Sawatzky and J. Kommandeur, Chem. Phys. Lett., 1973, 23, 21. 10 A. J. Silverstein and Z. G. Soos, Chem. Phys. Lett., 1976, 39, 525. 11 N. S. Dalal, D. Haarer, J. Bargon and H. Mohwald, Chem. Phys. Letf., 1976, 40, 326. 12 C. Corvaja, A. L. Maniero and L. Pasimeni, Chem. Phys., 1985, 100, 265. 13 J. Avalos, F. Devreux, M. Guglielmi and M. Nechtshein, Mol. Phys., 1978, 36, 669. 14 L. R. Melby, R. J. Harder, W. R. Hertler, W. Mahlor, R. E. Benson and W. E. Mockel, J. Am. Chem. 15 L. Komorowsky and G. Malackowicz, J. Phys. C3, 1983,44, 1207. 16 A. Cehak, A. Chyla, M. Radomska and R. Radomsky, Mol. Cryst. Liq. Cryst., 1985, 120, 327. 17 R. Biehl, M. Plato and K. Moebius, J. Chem. Phys., 1975, 63, 3515. 18 A. Colligiani, C. Pinzino, M. Brustolon and C. Corvaja, J. Magn. Reson., 1978, 32, 419. 19 R. G. Kepler, J. Chem. Phys., 1963,39, 3528. 20 T. Buckman, 0. Griffith and H. M. McConnell, J. Chem. Phys., 1965,43, 2907. 21 Z. G. Soos, J. Chem. Phys., 1967,46,4284. 22 D. B. Chesnut and D. C. Meinholtz, J. Chem. Phys., 1985, 83, 5495. 23 H. M. McConnell and D. B. Chesnut, J. Chem. Phys., 1958, 28, 107. 24 P. H. Rieger and G. K. Fraenkel, J. Chem. Phys., 1962,37, 2795. 25 H. M. McConnell and J. Strathdee, Mol. Phys., 1959, 2, 129. 26 R. H. Clarke and C. A. Hutchinson Jr, J. Chem. Phys., 1971, 54, 2962. 27 H. Johansen, Int. J. Quantum Chem., 1975,9,459. 28 M. A. Collins, Adv. Chem. Phys., 1983, 53, 225. 29 B. J. McClelland, Trans. Faraday Soc., 1961, 57, 1458. 30 R. D. Rataiczak, M. Thomas Jones, J. R. Reeder and D. J. Sandman, Mol. Phys., 1985,56, 65. SOC., 1962,84, 3374. Paper 61849; Received 1st May, 1986