J. Chem. Soc., Faraday Trans. I , 1987,83, 37-42 A Single-crystal ENDOR Study of y-Irradiated Pyridoxine Hydrochloride Neil M. Atherton* and Wendy A. Crossland Department of Chemistry, The University, Shefield S3 7HF Hyperfine coupling tensors for five protons have been determined from room-temperature ENDOR measurements on a single crystal of y-irradiated pyridoxine hydrochloride. Their interpretation supports the identification of the radical made by Masiakowski, Krzyminiewski and Pietrzak from analysis of the e.s.r. spectra using resolution enhancement. The identification of the radical formed on y-irradiation of pyridoxine hydrochloride is of interest because of the molecule's close relation to vitamin B,. In a recent study Masiakowski, Kryzyminiewski and Pietrzak used a resolution enhancement technique to enable the e.s.r.spectra of an irradiated single crystal to be ana1ysed.l They were able to distinguish six hyperfine couplings, and concluded that the radical had the structure I, formed by loss of hydrogen atom from c6 (cf. fig. 1). It seemed desirable to confirm the conclusions of the resolution enhancement technique using ENDOR spectroscopy. In this paper we report the results of a room-temperature ENDOR study of an irradiated crystal : five proton hyperfine tensors have been determined and their interpretation does indeed confirm that the trapped radical has the structure I. CHOH I I Experimental Well formed single crystals of pyridoxine hydrochloride (Aldrich) were grown from aqueous solutions at room temperature.The crystal structure has been accurately determined from neutron diffraction by Bacon and Plant,2 and the axes of our crystal were kindly determined for us by Dr N. A. Bailey of this department. The crystals are triclinic, space group p i , the two molecules in the unit cell being related through a centre of symmetry. The crystals grow as tablets and the morphology is such that a* is perpendicular to the well defined face, so a* was taken as a reference axis for our measurements. Our b*c* reference axes thus lie in the bc plane. After overnight irradiation in a ,OCo y-ray source, ca. 250 Ci, a crystal was glued to an accurately machined Perspex cube of 3 mm side. Thus by rotating the cube about the three fourfold axes perpendicular to its faces three orthogonal planes could be explored, the b*c* plane being located accurately but with the relation between the b*c* and bc axes determined by the mounting.E.s.r. and ENDOR spectra were taken every 10" during rotation about each of the 3738 ENDOR Study of y-Irradiated Pyridoxine Hydrochloride Fig. 1. Molecular structure of the pyridoxine hydrochloride cation and atomic numbering of ref. (2). three orthogonal axes using a Bruker ER 200D e.s.r. spectrometer with a Bruker ENDOR unit (100 W r.f.) and an Aspect 2000 computer. Good-quality proton ENDOR was obtained without difficulty. The crystal was not cooled, so the temperature of measurement was a little way above room temperature on account of r.f. heating. The e.s.r. spectra are broad doublets for many orientations, and for some show more complex, but very poorly resolved, hyperfine features.In general the ENDOR response was taken at the maximum of the e.s.r. absorption but the setting was not critical. The ENDOR spectra show a strong narrow feature which arises from distant protons and the centre of this was taken as an internal calibration of the free proton frequency. The spectra were taken from scans in the range 0.1-50 MHz. This limited the digital resolution to 50 kHz, which is adequate for our present purpose whose principal objective is the identification of the radical. Derived hyperfine parameters are quoted to 0.1 MHz, which is more than sufficient to account for mounting errors arising from non- orthogonality of the reference axes. The ENDOR transition frequencies were analysed assuming that the hyperfine interaction could be treated to first order and that the anisotropy of g could be neglected.Thus the orientation dependence of each transition frequency, v, was fitted to3 (1) v 2 * - V ~ + ( 1 / 4 ) ( t l ’ A . A . t l } T V ~ , - ( t l . A . t t } 2 where the upper and lower signs refer to the M, = +f and M , = -; electron spin states, respectively, vN is the free proton frequency, A the hyperfine tensor, and a unit vector defining the orientation of the applied field in the a*b*c* axis system. Transition frequencies were linearly corrected to a constant free-proton frequency of 15 MHz before fitting to eqn (1) using a least-squares procedure. An example of the fit for the b*c* plane is shown in fig. 2. Elements of A were then obtained from the difference between the parameters for the pairs of curves corresponding to each proton. Results and Discussion Couplings to five protons, which we initially label A-E, have been completely charac- terised.Their resolution into isotropic and principal dipolar components, and the orientation of the principal axes in a*b*c*, are shown in table 1 . Smaller proton couplings can be tracked to some extent but have not been analysed for the present. The spectra also show a number of weak features whose intensities are markedly orientation dependent; many of these lie at frequencies which are close to sums and differences ofN . M. Atherton and W. A . Crossland 39 18 17 16 15 14 13 0 ’2 g 11 % 10 2 9 ---. .* + k CI : 8 ; 7 6 5 4 3 2 1 0 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 4 0 42 ENDOR frequency/MHz Fig.2. Angular dependence of the ENDOR frequencies in the b*c* plane. The curves are least-squares fitted as described in the text. Table 1. Proton coupling tensors principal values/MHz direction cosines proton isotropic dipolar in a*b*c* A -40.3 18.5 0.803, - 0.553, - 0.222 3.0 -0.158, -0.557, 0.812 -21.6 0.575, 0.619, 0.535 B - 13.7 9.6 0.807, 0.333, 0.487 -9.4 -0.523, 0.787, 0.328 -0.2 -0.274, -0.519, 0.810 C 7.8 - 1.2 0.853, 0.206, 0.478 3.7 -0.332, 0.923, 0.193 -2.5 -0.403, -0.324, 0.886 D 9.6 - 2.4 0.126, 0.715, -0.688 3.7 -0.527, 0.636, 0.564 - 1.3 0.841, 0.291, 0.457 E 2.4 - 5.4 0.946. - 0.122. 0.302 0.6 -0.087, 0.799, 0.596 4.7 -0.314, -0.589, 0.745 those of strong transitions and we take them to be ‘forbidden’ transitions arising from simultaneous flips by two protons.No 14N ENDOR has been detected in these high-temperature measurements. Its value is of interest, the more so because ENDOR would reveal the quadrupole coupling, and we intend to pursue a low-temperature study. So far as the comparison between ENDOR and resolution enhancement is concerned40 ENDOR Study of y-Irradiated Pyridoxine Hydrochloride the latter appears to have the edge for this nucleus if one is restricted to room temperature. The pattern of hyperfine coupling for an a-proton, i.e. a proton in a ,C-H fragment whose C atom is part of a conjugated system, is well known:4 if the isotropic coupling constant is - 2a then the principal dipolar components are about (+a, 0, - a ) with the positive component lying along the C-H bond and the near-zero component along the axis of the 2p orbital on the C atom.A similar pattern is expected for an >NH fragment. The couplings to protons A and B conform to this pattern so these are clearly a-protons. The angle between the positive dipolar components is 69" and that between the small components is 8". This is consistent with the radical having structure I with proton A attached to C6, and lying close to the bisector of the H5C6H6 angle of the parent molecule, and proton B being the N-H proton, Hlo. However, these relative orientations would also hold for structure 11, which is the radical resulting from the loss of a hydrogen atom from C,. \ CHOH CH3 I1 It is useful to note that both I and I1 are iso-n-electronic with the benzyl radical. Now one readily sees that if the radical has structure I1 then proton B must be H,, the ring proton attached to C,.The choice between the two structures has to be settled by relating the orientations of the tensor components to crystal axes and hence to bond directions in the molecule. In our measurements the best-defined crystal axis is a*, which is perpendicular to the bc plane, and consideration of the crystal structure2 shows that a clear test is provided by the angles between the N-H,, and C,-H2 bonds and this axis. From the crystal structure we calculate the angles to be 37.5 and 89.5", respectively, while for the positive dipolar component of proton B the angle is 36". This essentially confirms that the radical is correctly assigned as I.The N-HI, centre is well removed from the site of the radiation damage, so one would expect no change of geometry on forming the radical and can confidently take the orientations of the tensor axes for proton B as defining the orientation of the ring in our axis system. Our results for proton A indicate that the plane of the O,C,H, group is not quite coincident with that of the ring, a point already remarked by Masiakowski et al. and discussed by them.' The distortion is not a small rotation about c3c6. Assignment of the remaining three proton tensors can be made by using the comparison with benzyl. We thus expect significant spin population at C, and C, and small, probably negative, spin population at C , and C,.The hydroxy proton at C , lies close to the plane of the ring in the undamaged crystal,2 and if this geometry is preserved in the radical the spin density at it should arise almost exclusively from spin polarisation and be small. On the other hand, in the CH,OH group at C,, O3 lies close to the plane of the ring so protons H, and H, should bear relatively high-spin densities from hyperconjugation. We assign protons C and D to these two positions. The isotropic couplings to such P-protons are often described by4 a = B,+B, cos20 (2) where 8 is the dihedral angle between the plane containing the axis of thep-orbital and the C-CH bond and that containing the C-C-H atoms. The contribution from spinN . M. Atherton and W. A . Crossland 41 polarisation is given by B,: it is much smaller than B, and often neglected.If this approximation is made the experimental isotropic couplings would indicate cos2 I!?~Jcos~ I!?,, = 1.25. This is not pleasing agreement with the value 1.09 which we calculate from the neutron diffraction data:, taken at their face value the observed couplings imply that in the radical H, and H, are much less symmetrically disposed with respect to the plane of the ring than they are in the undamaged crystal. However, this analysis is severely oversimplified. The protons in question must be virtually as close to the substantial spin population on C, as they are to the smaller one on C,, and this must be reflected in the couplings, whose analysis is thus not trivial. This consideration also complicates the interpretation of the dipolar couplings to these protons, in particular any geometrical information one might hope to obtain from the orientations of the principal components. The full theoretical analysis of the observed couplings thus offers an interesting problem.The magnitudes of the dipolar components for proton E do not have the characteristic pattern for an a-proton, but nonetheless we assign the proton as H,, attached to the ring carbon C,. From the analogy with benzyl we expect small (negative) spin population on this ring atom and under these circumstances the much larger spin populations on C, and N should play a crucial role in determining the dipolar coupling to the proton on the intervening carbon atom. The effect has been well understood since the definitive study by Heller and Cole of the HOOCaCHCHCH-COOH radical in irradiated glutaconic acid., For a planar C-C(H)-C fragment symmetry demands that one principal component of the coupling should be perpendicular to the plane: for proton E the 4.7 MHz dipolar component is 6" from the intermediate component for proton B, the N-H proton.This is very strong support for this assignment. The different spin populations on N and C, cause the in-plane components to be skewed with respect to the C5-H2 bond. The protons of the methyl group substituted at ring atom C, are expected to have small hyperfine couplings if our analogy with benzyl is valid. They could well be responsible for the small couplings which we have observed but not analysed, as mentioned previously.Analysis of these couplings might be facilitated by studying a crystal grown from D,O, and this will be pursued. A more serious feature of the present analysis is our failure to observe a signal from the hydroxy proton at the C(H)OH radical centre. It is not obvious that its ENDOR transition probability should be so dramatically less than those of the other protons. If the radical had ionised the resulting neutral radical might be formally isoelectronic with the radical anion of benzaldehyde so one might take a qualitative discussion of the spin density using this anion as a prototype. in fact this would not affect the general conclusions in any way: the pattern of spin distribution in the benzaldehyde is quite similar to that in benzyl,; for example, there are negative spin populations at the meta positions and the aldehydic proton has the largest coupling.To conclude, we can say that, despite a number of unanswered questions which may arise because we have deliberately restricted ourselves to room temperature in order to make a fair comparison with the Polish work,l this ENDOR study essentially confirms the identification of the trapped radical. This is a pleasing result, for resolution enhancement facilities are probably more easily obtainable than those for ENDOR and so may become more generally available. We thank the S.E.R.C. for funding most of the purchase of the ENDOR spectrometer, Croydon Education Authority for undergraduate support for W. A. C., Mr Alan Hall and Mr Brian Watson for the construction of crystal mountings and goniometers, and Mrs Jean Stevenson for general technical assistance.42 ENDOR Study of y-Irradiated Pyridoxine Hydrochloride References 1 J. T. Masiakowski, R. Krzyminiewski and J. Peitrzak, Chem. Phys. Lett., 1985, 116, 387. 2 G. E. Bacon and J. S. Plant, Acta Crystallogr., Sect. B, 1980, 36, 1130. 3 See, e.g., N. M. Atherton and A. J. Horsewill, Mol. Phys., 1979, 37, 1349. 4 See, e.g., N. M. Atherton, Electron Spin Resonance (Ellis Horwood, Chichester, 1973). 5 C. Heller and T. Cole, J . Chem. Phys., 1962, 37, 243. 6 N. Steinberger and G. K. Fraenkel, J . Chem. Phys., 1964,40, 723. Paper 6/999; Received 22nd May 1986