J. Chem. Soc., Faraday Trans. I, 1989, 85(12). 3927-3937 Structural Information from Powder ENDOR Spectroscopy Possibilities and Limit at ions Donato Attanasio Istituto di Teoria e Struttura Elettronica e Comportamento Spettrochimico dei Composti di Coordinazione del C.N.R., P.O. Box 10, 00016 Monterotondo Staz., Roma, Italy The methodology for extracting structural information from powder ENDOR spectra is briefly reviewed. The procedure, based upon tensor transformation of hyperfine data obtained from spectra recorded at the EPR ‘turning points ’, is described in some detail, whereas other techniques, such as fitting of spectra obtained as a function of the observing field, are only mentioned. Examples from the recent literature are presented and discussed. It is shown that in many cases single-crystal measurements do not offer appreciable advantages, as far as the geometry of magnetic nuclei surrounding the paramagnetic probe is concerned.Continuous-wave electron nuclear double resonance spectroscopy (ENDOR),14 when performed on magnetically diluted single-crystals, typically yields highly accurate data concerning weak magnetic interactions between a probing paramagnetic centre and surrounding magnetic nuclei. Often these hfs tensors measured by ENDOR are dominated by a simple dipolar contribution, and the coordinates of the interacting nucleus, with respect to the magnetic axes of the central unpaired spin, can be extracted from the experimental data. In the past beautiful examples of ‘ENDOR crystallography’ have been provided by the work of Hutchison Jr et al.,5-7 who applied this technique in the determination of proton coordinates both in model compounds and biomolecules.However, in many cases it is convenient or essential to obtain this kind of information from the spectra of non-oriented samples, even though considerable loss in resolution and accuracy may result. ENDOR spectroscopy is often one of the few techniques available to give an insight into the electronic and geometrical structure of metal-containing, complex materials of biochemical or catalytic interest. A conventional powder ENDOR experiment is performed by choosing a suitable value of the external magnetic field B, (i.e. a suitable position within the powder EPR absorption envelope) and then sweeping a radiofrequency field. At certain values of the applied r.f., nuclear transitions, detected as intensity changes of the EPR absorption, may be observed, assuming favourable values of the electron and nuclear relaxation times.** Clearly the ENDOR response of a powder sample arises only from the subset of molecules which contributes to the EPR intensity at the chosen value of the observing field B,.If EPR turning points are selected, namely magnetic field values which correspond to defined molecular orientations, so-called single-crystal-like ENDOR spectra are obtained.1° The possibility of using this selection technique depends upon the anisotropy and relative orientation of the various interacting tensors. In some favourable cases three components of the magnetic tensor can be obtained for each set of equivalent, interacting nuclei.However, these experimental parameters do not correspond to 39273928 Powder ENDOR Spectroscopy Fig. 1. General orientation of the principal axis system for g and proton hyperfine tensors. principal values of the magnetic tensors, but rather represent the extrema of the interaction along the magnetic axes of the central unpaired spin. Therefore a further step in data analysis or additional experimental data, such as collection of spectra as a function of B,, is required before the principal values and hence structural data may be obtained. Recently several groups have shown renewed interest in powder ENDOR spectro- scopy, with the aim of taking full advantage of the wealth of information contained in the spectra.11-20 Proper interpretation of the experimental data may yield the principal values of the magnetic interactions in general cases, thus providing access to the geometrical arrangement of ligand nuclei in the vicinity of the paramagnetic centre. Several, slightly different approaches have been proposed and are briefly discussed below.Data Analysis Fig. 1 shows the principal axis systems of a generic proton hfs tensor and of the g tensor of the central paramagnetic probing ion. Generally speaking, the two sets of magnetic axes are completely independent and the effective hyperfine coupling constant can be writ ten2' where and g2 = c g i " 1: a (3) Our coordinate reference frame is the principal axis system of the g tensor, with i = 1-3 = x-z. The arbitrary orientation of the hyperfine tensor principal axis system, (x', y', z'), with respect to the reference frame, is represented by the polar angles (01, O,, e3).M (el, O,, 0,) is the Euler transformation matrix 22 and Zi are the direction cosines defining the orientation of the external magnetic field. If we restrict ourselves to the simple case of an axially symmetric g tensor the above Euler transformation reduces to a simple rotation of an angle 8 around the x' axis.D. Attanasio 3929 II(III A195.97 MO ( I = 5/2) I I I I I 310 330 350 BO /mT Fig. 2. Frozen-solution X-band EPR spectrum of [MoOCl,(H,O)] in 1 : 1 glycerol : HCl (2 mol dm-3). The spectrum consists of a central intense g feature and satellite hyperfine lines, respectively due to the presence of non magnetic (ca.75 %) and magnetic (ca. 25 %, I = p) Mo nuclei (see text). z and xy indicate the turning points saturated to obtain parallel and perpendicular ENDOR spectra. Approximate positions of the perpendicular hyperfine lines are given. Fig. 2 shows a typical MoV, dl, EPR powder spectrum for such an axially symmetric situation and in the absence of any metal hyperfine interaction. Note that the central g pattern is due to molecules containing the non-magneticg6* Mo nuclei (natural abundance 75%). ENDOR spectra recorded with the external field set at the two EPR turning points, as indicated in the figure, selects subsets of molcules with their z axes respectively parallel and perpendicular to the external magnetic field. These spectra may yield, for each proton hfs, three different couplings A,,, A,,, and A,, which correspond to the extrema of the hyperfine interaction along the z-axis and within the xy plane.Fig. 1 clearly shows that A,, coincides with a principal value of the the hfs tensor, whereas rotation of the unknown angle 0 must be performed on the two other components to obtain A , and A,. Rotation formulae can easily be obtained on the basis of the above equations and below we quote the results reported by Hofmann et al. :21a AEy - (Atu +A&) sin2 9 1 -2 sin2 9 We now assume the electronic spin to be 1a;gely localized at the central metal ion and the metal-proton distance to be at least 2 A: Under these conditions the point dipole approximation is valid, i.e. the interacting tensors are axially symmetric, and the additional condition : A , = A , = A,, (7) 131 FAR I3930 Powder ENDOR Spectroscopy is valid.Thus an estimate of the rotation angle 8 can be obtained: Alternatively, direct measurement of the hyperfine principal value may be possible by recording ENDOR spectra as a function of the observing field. Moving the field from B, to B, the ENDOR lines corresponding to A,, move outwards until they reach a maximum splitting when B, coincides with the direction of the principal value A,(see fig. 1). The measured splitting gives the A, value, whereas the rotation angle can be easily obtained from the measured effective g value. In the literature such a direct procedure has been found to be viable for the spectra of very simple paramagnetic systems.2o Otherwise fitting of field-variation spectra is necessary. l6 In addition sensitivity problems often complicate or completely hamper recording of ENDOR spectra outside the EPR turning points. The above techniques allow derivation of the hfs principal values. If the relative signs of the tensor components can be assessed their average gives the isotropic contribution, whereas the anisotropic principal value may be used to derive the metal-proton distance according to the simple dipole formula,22 which assumes unit spin density at the metal : (9) In this way the proton spectroscopic coordinates are obtained as r, the distance of the proton from the paramagnetic ion, and 8, the angle that the metal-proton vector makes with the g, reference axis. In the case of axial systems the second polar angle @ is not accessible, but estimates of its value have been obtained for strongly rhombic compounds.16 In the case that the tensor principal values are obtained through tensor transformation two further points must be stressed.The first is the possibility of error propagation from the calculated value of 8 to the calculated value of r. Determination of 8 is based upon the difference between A,, and A,,, which are similar in value, and is therefore affected by rather large errors. However, it is easy to show, on the basis of eqn (4)-(8), that the value of the rotation angle affects only the isotropic part of the tensor, whereas it leaves completely unaffected the dipolar component A,,. This means that the calculated Y values do not depend upon the tensor transformation we perform.23 The second point to be noted is the ambiguity in the sign of 8.The rotation angle enters eqn (4)-(6) as a squared sine function. Therefore its absolute value can easily be derived, but we cannot discriminate between clockwise and anticlockwise rotations. Although reasonable guesses can often be made this appears to be the major drawback of this kind of analysis. A , = A,-A,, = pbpngeg,(3COS2~- 1 p 3 . Results VO(sa1en) Several characteristics make the dl V02+ ion one of the most suitable transition metals to be used as an ENDOR paramagnetic probe." Its o non-bonding ground-state implies high localization of the unpaired electron, thus extending the validity of the point-dipole approximation. In addition highly orientational selective and well resolved ENDOR spectra can often be obtained from powder samples of vanadyl compounds at liquid- nitrogen temperatures or above.This is due, respectively, to the large anisotropy of the A" tensor and to the fact that vanadyl EPR transitions can easily be saturated. lass together with the 11 and I ENDOR spectra recorded at the indicated B, settings,323gSix, out of seven sets of non-equivalent protons gave measurable couplings with the central Fig. 3 shows the frozen-solution EPR spectrum of VO(sa1en) in a dmf-d7/CDC1D. Attanasio 393 1 I 1 ' 1 I I I I I ] 280 320 360 T/K I -2 -1 0 1 2 (v - VP )/MHz 1 I I I 1 I -2 -1 0 1 2 (v - v p )/MHz Fig. 3. Frozen-solution parallel and perpendicular proton ENDOR spectra of VO(sa1en) at 100 K. Different proton lines are labelled as follows: a and e, axial and equatorial ethylenediamine protons , 7, 6, 3, and 5 are H(7), H(6), H(3) and H(5) protons. X has not been identified.Table 1. Experimental and transformed proton hyperfine splitting parameters (MHz) of VO(sa1en)" assign- men t A , A , A , A , , A l Aiso Adi, rb 9 r" 8' H3 1.03 1.01 2.02 2.03 -1.01 - 2.03 427 97 438 98 H, 1.66 1.55 3.14 3.21 -1.51 0.06 3.15 370 104 385 100 0.47 0.47 1.01 1.01 -0.47 0.02 1.03 535 90 528 89 0.26 nmd nm 0.52 -0.26 - 0.52 662 - 651 97 H6 H5 CHzes 2.26 2.3 3.91 4.03 -2.11 -0.06 4.09 338 103 324 108 CH,,, 2.68 2.59 4.40 4.74 -2.51 -0.09 4.81 310 78 305 79 a From ref. (23). In units of pm. Crystallographic values. nm = not measured. unpaired spin, and the experimental splittings were assigned to specific protons essentially on the basis of chemical substitution at selected positions.Table 1 summarizes the experimental data and includes the results of a tensor transformation performed according to the procedure outlined above. Keeping in mind the rather crude approximations used, comparison between spectroscopic and crystallographic results appears to be quite satisfactory. In addition it turned out that experiments with randomly oriented vanadyl complexes may provide structural information with accuracy comparable to detailed single-crystal 25 In the case of frozen-solution spectra, the use of deuterated solvents allows easy detection of lines due to quite distant protons, e.g. H, with r = 650 pm, which in the solid are generally obscured by strong matrix lines due to surrounding, non-interacting pro tons.Another outcome of the spectra concerns the conformation of the ethylenediamine bridge. X-Ray data26 for these complexes indicate a bridged conformation intermediate between gauche and eclipsed, with the two equatorial protons at largely different 131-23932 Powder ENDOR Spectroscopy A?- ' VI -2 - 1 0 1 2 (v - v p Y M H Z Fig. 4. Frozen-solution parallel and perpendicular proton ENDOR spectra of [MoOCl,(CH,OH)]- at 15 K. Magnification shows the OH proton lines due to second solvent shell molecules. distances from the central metal ion. However, ENDOR spectra give a single set of lines, clearly suggesting equivalence of the two equatorial protons, therefore implying that in solution the bridge conformation relaxes to the more usual eclipsed configuration.[MoOCl,(S)]- Powder ENDOR spectroscopy of the dl MoV ion could give a substantial contribution to the structural characterization of the active sites in a number of Mo-based chemical and enzymatic catalysts. In this context single-crystal ENDOR2' and multifrequency EPR data2* have recently been reported. However, a number of additional problems may be expected from the powder ENDOR spectra of Mo-containing compounds. Among them the low g-tensor anisotropy, the superposition of different EPR lines, due to the presence of different magnetic isotopes, and the very low symmetry of most of these complexes may be recalled. Discussion of the ENDOR spectra of the particularly simple compounds derived from dissociation and solvent interaction of the [MoOC1J2- ion conveniently illustrates the actual problems.The frozen-solution and proton ENDOR spectra of different [MoOCL,(S)]- species, including S = H20, CH,OH, C2H,0H and CH,CHO, have been mea~ured.~' All the ENDOR spectra were recorded at temperatures between 10 and 20 K. Above this value their quality rapidly deteriorated, although the hydroxyl proton lines could still occasionally be measured up to ca. 110 K. Fig. 4 and 5 report the I( and I ENDOR spectra of [MoOCl],(CH,OH)]- as well as their variation as a function of B,. It turns out that the parallel spectra always show the presence of rather intense perpendicular lines. On the other hand fig. 5 clearly indicates that these perpendicular features can not be simply explained in terms of low g- tensor anisotropy, i.e.low orientational selectivity. Inspection of fig. 2 indicates thatD. Attanasio 3933 I 1 I I -3 -2 - 1 0 1 (v - v p )/MHz Fig. 5. Magnetic field dependence of the frozen-solution ENDOR spectrum of [MoOCl, (CH,OH)]-. The B, values used to saturate the EPR line are shown in the insert. Purely parallel ENDOR spectra are obtained only in a narrow field interval around gll. Outside this range intense perpendicular features appear. For B, .< BI, their intensity is ascribed to superposition of g,l and A , hyperfine lines due to magnetic Mo isotopes. superposition of the 11 g line, due to molecules containing non-magnetic Mo isotopes, with the M, = -$ I hyperfine line, due to molecules containing magnetic Mo isotopes ( 9 5 M ~ , I = 4, 15.72 % ; 9 7 M ~ , I = 4, 9.46 %), is a common feature of these EPR spectra. This means that nuclear irradiation at B, = hv/g,P simultaneously involves molecules containing both even and odd Mo isotopes, respectively oriented parallel and perpendicular to B,.Since superposition of the EPR lines is not perfect, small shifts in the observing field greatly alter the relative line intensities, allowing almost pure 11 or I spectra to be obtained. In spite of this, more complex spectra would inevitably require the use of isotopically pure molybdenum. Table 2 summarizes the experimentally obtained proton couplings. Axial coordination of the solvent molecule is easily deduced from the absolute values of the OH proton splittings and from the fact that the largest coupling is measured along the molcular axis.The I ENDOR should give two different pairs of peaks, measuring the extreme of the hfs interaction in the equatorial plane. Of course these two values are expected to be quite similar, and they are in fact unresolved, giving rise to a sharp doublet split by 2.38 MHz. This apparent axial symmetry of the interaction prevents determination of the rotation angle 8. Derivation of an approximate Mo-H distance implicitly assumes a small 8 value, so that measured and principal magnetic values approximately coincide. The couplings of the three inequivalent CH, protons could also be measured, showing the presence of one strongly interacting and two, quasi-equivalent, weakly interacting protons.In addition, variable-temperature measurements showed that at ca. 100 K the CH, line pattern reduces to a single broad pair of lines split by ca. 0.9 MHz, suggesting3934 Powder ENDOR Spectroscopy Table 2. Experimental and transformed proton hyperfine splitting parameters (MHz) of (NH,)[MOOC~,(CH,OH)]~ assignment Arb AYb Aiso A , rc OH 5.17 2.38 0.14 5.03 315 n.m.d 5.75 - n.m. 4.92 - n.m. 3.85 - - - - - - - 2.23 1.22 -0.07 2.30 410 0.79 0.51 -0.07 0.86 570 CH,’ 0.45 n.m. - - CH,” - CH3 From ref. (29). In units of pm. 11 and I refer to the g-tensor principal axes. n.m. = not measured. (v - vp U M H Z Fig. 6. Parallel (a) and perpendicular (6) proton ENDOR spectra of VIV/VOPO,(H,O), at 4 K. Labels 1, 2 and 3 simply show ordering from the largest to the smallest hfs.the presence of a freely rotating CH, group. Finally, additional broad lines ascribed again to hydroxyl proton interactions could be identified, although only in the parallel spectrum. Three different proton couplings were measured and they suggest the presence of a well ordered second solvation shell sphere. They correspond to at least three different CH,OH molecules, approximately oriented along the molecular plane and probably connected to the chlorine atoms via hydrogen bonds. VOPO,(H,O), The approach described above, tested in the case of simple coordination compounds of known crystal structure, has been applied to a more complex situation with the aim ofD. Attanasio 3935 Table 3. Experimental and transformed proton hyperfine splitting parameters (MHz) of VOPO,(H,O), at 4 K" assignment A , A , A , A,, A , Aiso Adi, rb 9 H(1) 2.58 0.70 0.70 2.58 -0.70 0.63 2.66 390 90+5 H(3Id 0.19 - 0.18 - 0.75 0.70 3.28 3.29 -0.70 0.39 2.19 416 c - - 0.19 940 - Ht2) a From ref. (32).determination of 9. different orientations (see text). In units of pm. ' Quasi-axial orientation of this protein prevented direct This coupling is assigned to different protons with similar distances but determining the geometrical arrangement of intercalated water molecules in the layered materials VOPO, (H,O), and VOPO, H20),. Vanadyl (V) phosphate dihydrate3', 31 consists of infinite layers of corner-sharing octahedra and tetrahedra. The vanadium atom lies on a fourfold axis and is surrounded by six oxygen atoms to give a distorted octahedron.The four equatorial oxygens are provided by four different tetrahedra, whereas the axial ones are a terminal 0x0 ligand and a water molecule. Each phosphate tetrahedron connects four different octahedra and vice versa, to make up the layers, which are then weakly connected, at a distance of 715 pm, by the two water molecules coordinated or hydrogen bonded to the vanadyl ion and to the layer oxygens. Interest in this and related materials has recently increased, after realizing that they readily swell and intercalate different guest molecules, providing novel expanded, pillared porous materials of catalytic interest. Of course detailed knowledge of the structural relationships between the host matrix and the intercalated or pillared molecules is of primary importance.On the other hand, VOPO, is particularly amenable to paramagnetic resonance techniques in that it always contains small amounts of VIV. Therefore determination of the weak interactions between the Vrv ion and the magnetic nuclei of the intercalated molecules is possible and may provide information upon their arrangement inside the layers. On this basis the EPR and proton ENDOR spectra of the simplest possible vanadyl phosphate intercalates, i.e. VOPO,(H,O), and VOPO,(H,O),, have been Fig. 6 shows the 11 and I ENDOR spectra of VOPO,(H,O), measured at 4 K, whereas no spectrum was detected from the dihydrate analogue in the temperature range 4- 100 K. This result was unexpected since axially coordinated and outer-sphere, hydrogen- bonded water molecules have been frequently and easily measured in the spectra of different vanadyl compounds.ll~ 2o Together with the rather poor quality of the spectra in fig.6 this is an example of the problems which can be met on going from simple model compounds to chemically relevant complex systems. From the fact that VOPO,(H,O), gives no observable spectrum, apparently because of unfavourable relaxation effects, we conclude that all the signals measured in the spectra of the pentahydrate are due to the three extra intercalated water molecules. Couplings due to three different proton sets, which we ascribe to at least two distinct interlayer water molecules are listed and analysed in table 3, according to the usual procedure. H(1), found at 416 pm in a quasi axial position, belongs to an outer-sphere water molecule, probably hydrogen bonded to the undetected, vanadyl-coordinated H20.Owing to the poor spectral resolution the H(l) lines may well account for both the quasi-equivalent water protons. H(2), at 390 pm and relatively close to the [vO(O),] equatorial plane (6' = 5"), identifies a second outer-sphere water molecule. Lines due to the second proton of H,0(2) could not be identified with certainty. Fig. 7, drawn taking into account the coordinates of VOP0,(H20)230 and the location of the two intercalated3936 Powder ENDOR Spectroscopy Fig. 7. Sketch of the expanded interlayer suggested to be present in VOPO,(H,O),. Re-drawn on the basis of the X-ray data reported for VOPO,(H,O),, [ref. (30)] and of the position of the two interlayer water molcules as derived from ENDOR.water molecules as derived above, reconstructs the expanded interlayer suggested to be present in VOPO,(H,O),. H,O( 1) and H,0(2), together with the V02+ coordinated water molecule, complete pillaring of the layers allowing for the 317 pm expansion observed on going from the dihydrate to the pentahydrate compound. The spectra of fig. 6 show the presence of a third proton coupling with an almost constant splitting of ca. 0.19 MHz both in the 1) and I spectra. We suggest that these lines are due to different protons having similar distances (ca. 940 pm) and different orientations. Possible assignments are the protons of H,O(3) as well as proton H(2) from an adjacent molecular unit in the layers. References 1 L.Kevan and L.D. Kispert, Electron Spin Double Resonance Spectroscopy (John Wiley, New York, 2 A. Schweiger, Struct. Bonding (Berlin), 1982, 51. 3 Multiple Electron Resonance Spectroscopy, ed. M. M. Dorio and J. H. Freed (Plenum Press, New 4 H. Kurreck, B. Kirste and W. Lubitz, Angew. Chem. Int. Ed. Engl., 1984, 23, 173. 5 C. A. Hutchison Jr and D. B. McKay, J. Chem. Phys., 1977, 66, 331 1. 6 C. A. Hutchison Jr and T. E. Orlowski, J. Chem. Phys., 1980, 73, 1. 7 C. A. Hutchison Jr and D. J. Singel, Proc. Natl Acad. Sci., 1981, 78, 6883. 8 G. Rist and J. Hyde, J. Chem. Phys., 1968,49, 2449. 9 G. Rist and J. Hyde, J. Chem. Phys., 1969, 50, 4532. 10 G. Rist and J. Hyde, J. Chem. Phys., 1970, 52,4633. 11 H. van Willigen, J. Magn. Reson., 1980, 39, 37. 12 B. Kirste and H.van Willigen, J. Phys. Chem., 1982, 86, 2743. 13 H. van Willigen and T. K. Chandrashekar, J. Am. Chem. SOC., 1983, 105, 4232. 14 R. A. Venters, M. J. Nelson,P. A. McLean, A. E. True, M. A. Levy, B. M. Hoffman and W. H. Orme- 15 A. E. True, M. J. Nelson, R. A. Venters, W. H. Orme-Johnson and B. M. HofEman, J. ,4m. Chem. 16 (a) G. C. Hurst, T. A. Henderson and R. W. Kreilick, J. Am. Chem. SOC., 1985, 107, 7294; (b) T. A. 17 P. J. O'Malley and G. T. Babcock, J. Am. Chem. Soc., 1986, 108, 3995. 1976). York, 1979). Johnson, J. Am. Chem. SOC., 1986, 108, 3487. SOC., 1988, 110, 1935. Henderson, G. C. Hurst, and R. W. Kreilick, J. Am. Chem. SOC., 1985, 107, 7299.D. Attanasio 3937 18 M. Baumgarten, W. Lubitz and C. J. Winscom, Chem. Phys. Lett., 1987, 133, 102. 19 D. Gourier and E. Samuel, J. Am. Chem. Soc., 1987, 109, 4571. 20 D. Mustafi and M. W. Makinen, Inorg. Chem., 1988, 27, 3360. 21 (a) B. M. Hoffman, J. Martinsen and R. A. Venters, J. Mugn. Reson., 1984, 59, 110; (b) B. M. Hoffman, R. A. Venters and J. Martinsen, J. Mugn. Reson., 1985, 62, 537. 22 A. Schweiger, G. Rist and Hs. H. Gunthard, Chem. Phys. Lett., 1975, 31, 48. 23 D. Attanasio, J. Phys. Chem., 1986, 90, 4952. 24 A. Schweiger and Hs. H. Gunthard, Chem. Phys. Lett., 1978, 32, 35. 25 S. Kita, M. Hashimoto and M. Iwaizumi, Znorg. Chem., 1979, 18, 3432. 26 D. Bruins and D. L. Weaver, Inorg. Chem., 1970, 9, 130. 27 N. H. Atherton and R. D. S. Blackford, Mol. Phys., 1987, 61, 443. 28 G. R. Hauson, G. L. Wilson, T. D. Bailey, J. R. Pilbrow and A. G. Wedd, J. Am. Chem. Soc., 1987, 29 D. Attanasio, M. Funicello and L. Suber, Chem. Phys. Lett. 1988, 147, 273. 30 H. R. Tietze, Aust. J. Chem., 1981, 34, 2035. 31 M. Tachez, F. Theobald, J. Bernard and W. Hewat, Rev. Chim. Miner., 1982, 19,291. 32 L. Alagna, D. Attanasio, T. Prosperi and A. A. G. Tomlinson, J. Chem. SOC., Faraday Trans. I , in press. 109, 2609. Paper 9/01634A; Received 18th April, 1989