摘要:

J. Chern. Soc., Faraday Trans. I, 1989, 85(12), 3987-3994 Experimental Evidence for the Hyperfine Interaction between a Surface Superoxide Species on MgO and a neighbouring Hydroxylic Proton E. Giamello,* E. Garrone and P. Ugliengo Dipartimento di Chimica Inorganica, Chimica Fisica e Chimica dei Materiali, Universita di Torino, Via P. Giuria 7, 10125 Torino, Italy Michel Che Laboratoire de Chimie des Surfaces et Structure UA I106 CNRS, Universite' P. et M. Curie, 75252 Paris, Cedex 05, France the late A. J. Tench Chemistry Division, A.E.R.E., Harwell, Didcot, Oxfordshire The superoxide 0; radical ion has been generated at the surface of MgO by two different methods both involving the parallel formation of surface hydroxyls. EPR spectra at 77 K reveal the presence of only one species with a hyperfine structure due to weak dipole4ipole interaction with a nearby hydroxyl (B, = 0.37 mT, B, = 0.21 mT, B, = 0.10 mT).The set of B data is satisfactorily reproduced by a simple model for the dipolar interaction envisaging a T-shaped O;..-H species. Calculated geometrical features are in excellent agreement with a plausible structure for the superoxide species sitting on low-coordinate Mg2+ and interacting with the hydrogen of an adjacent hydroxyl. The chemistry of the superoxide 0; ion is receiving increasing attention owing to the role played by this species in many catalytic and biological systems. A recent example in heterogeneous catalysis is the methane oxidative coupling that occurs on alkaline-earth- metal oxides and on other oxide systems,' involving 0, and other related oxygen species as intermediates.Stable 0; species have been observed on several ionic oxide surfaces by EPR:, the commonly accepted description is that given by Che and Tenth,, who adapted the model advanced by Kanzig and Cohen3 for 0; in the bulk of alkali-metal halides. Experiments employing "0-enriched oxygen4 have shown that the superoxide has equivalent oxygen atoms : accordingly, it is described as lying flat on the surface in electrostatic interaction with one positive metal ion.' In the EPR spectra of a series of oxides the g,, value is indeed observed to depend on the effective charge of the coordinating ion.2 To account for the variety of species observed under particular circumstances on the ionic surface of polycrystalline MgO, some of us5 have recently discussed the possible locations of adsorbed 0, and shown that these are restricted to three- and four- coordinate Mg2+ ions present at the morphological defects of the microcrystals such as edges, corners, kinks etc.Two different methods (A and B) for 0; generation on MgO are possible. According to method A,6 the outgassed sample is first y- or UV-irradiated in an H, or D, atmosphere. Surface oxygen anions react to yield a hydroxyl and an electron trapped at a surface anionic vacancy (F: centre Id), according to the scheme: O2-+;H2 -+ OH-+M. (1) 39873988 Successive reaction with oxygen brings about superoxide formation : Surface Superoxide Species on MgO o,+ I=] + 0,. (2) Method B envisages the reaction at the surface of a hydrogen-containing, slightly acidic molecule HX with 0,.The first steps of reaction are5' the dissociative adsorption of HX to yield a hydroxyl and an anionic species X-: HX + 0'- + X- +OH- (3) X-+O,-+X'+O,. (4) which then reacts with oxygen by electron transfer: The radical species X' further reacts, and indeed the overall process is the oxidation of the HX molec~le,~ whereby the superoxide species acts as an intermediate. Some superoxide species, however, survive at the surface. Depending on the extent of reaction, this method for 0; generation leads to two distinct spectral shapes. In the former (type I, generated under mild conditions, e.g. at low pressures of hydrocarbon) a multiplicity of peaks is observed in the g,, region, indicating the presence of a variety of superoxide radicals differing in the local environment of the Mg2+ adsorption site.5 The second spectral shape (type 11, recorded at high extent of reaction) is much simpler and provides the theme of the present work.Experimental Polycrystalline, high-surface-area MgO samples were prepared by slow decomposition of the hydroxide as described elsewhere.8 The 0, ions were generated at the surface as follows. Method A : 0.1 kPa oxygen was dosed on samples of MgO containing ca. 1 YO surface excess electrons (F: centres) obtained by y-irradiation of the oxide under hydrogen or deuterium.6 Method B: a mixture of molecular oxygen and a hydrocarbon (HX = propene or toluene) was allowed to react on MgO outgassed at 1073 K. Both hydrocarbons yield the same results.An excess of the hydrocarbon was employed (HX:O, = 6.0, total pressure = 3.85 kPa) to obtain type I1 spectra. EPR powder spectra were recorded both at room temperature and at 77 K on a Varian E-109 equipped with a dual cavity. Varian Pitch ( g = 2.0028) and DPPH ( g = 2.0036) were used as standards for g value calibration. which perform molecular graphics on a personal computer, making use of the standard P L U T O ~ ~ program.9b Fig. 4 (later) was obtained by means of the MOLDRAW Results and Discussion Spectroscopic Evidence The spectra in fig. 1 ( a ) and 1 (b), recorded at room temperature and 77 K, respectively, were obtained by method B (reaction of toluene and oxygen). Fig. l ( a ) is a type I1 spectrum as it exhibits a single peak in the g,, region and was therefore ascribed to a single type of superoxide ions.5 The same spectrum recorded at 77 K reveals, however, a multiplicity of lines in the high-field region [fig.l(b)], seen better in fig. 2 where an expanded view of the spectrum of fig. l ( b ) is shown. Four resonance lines are clearly observed in the g,, and gzlv regions; the g,, line also results in a doublet with a separation of ca. 0.1 mT between the two components. The presence of doublets in the EPR spectrum cannot in this case be explained in terms of two similar superoxide species by analogy with what was done for type IE. Giamello et a1 3989 Fig. 1. EPR spectra of 0; on MgO (obtained by method B, see text). (a) T = 298 K, (b) T = 78 K. Q10 mT n I 2.&89 2.0018 Fig. 2. Expanded view of the spectrum in fig.1 (b). The hyperfine coupling constants are shown in the figure.3990 Surface Superoxide Species on MgO I I 2.0089 2.0018 f? Fig. 3. EPR spectra of 0; on MgO obtained by method A (see text). (a) Sample irradiated under H,, (b) sample irradiated under D,. spectra. This hypothesis is indeed in conflict with the equations by Kinzig and Cohen3 for the principal g-tensor values, which, neglecting second-order terms, are : gxx = ge ; guy = ge + 2A/E; g z z = ge + 2RlA ( 5 ) where R is the spin-rbit coupling constant for oxygen, A is the separation between the two n, orbitals of oxygen caused by the electric field of the adsorption site and E is the separation between the highest of the two n, orbitals and the 0, orbital. Taking into account that E is ca.an order of magnitude larger than A,2 g,, is most sensitive to the electric field exerted by the adsorption cationic site, followed by guy. The g,, term, in the approximation adopted, is constant and thus independent of the adsorption site. This is exactly the opposite of what is observed in fig. 2, where the highest separation is observed in the g,, region and the lowest separation is observed in the g,, region. The doubling of the spectral lines may be interpreted in terms of a hyperfine interaction of the superoxide radical with a nucleus with I = $. The only possibility in the present case is the proton of a hydroxyl OH- group located near the 0, radical: OH- ions at the MgO surface are formed via reaction (1) or (3) and their presence is documented by IR spectra.’ This interpretation is strongly supported by the results shown in fig.3. Fig. 3(a) reports the high-field part of the spectrum obtained by method A (oxygen adsorption on MgO containing F,+ centres produced in an H, atmosphere). In spite of the different method followed, the spectrum is strictly similar to that in fig. 2 (except for the features indicated by asterisks in the figure, not to be discussed here) and in particular shows the same splittings between the g,, and guy lines observed in the case of method B. The spectrum in fig. 3(b) was obtained by method A in a deuterium atmosphere:E. Giamello et al. 399 1 surface hydroxyls are thus replaced by deuteroxyls. The substitution of deuterium for hydrogen alters the spectrum, apparently causing the disappearance of the hyperfine structure. As a consequence of the magnetic properties of the two nuclei (pH/pD = 3.26) the doublets are substituted by much narrower triplets that are not easily resolved. To summarize, in contrast with type I ~pectra,~, lo where the simultaneous formation of slightly different species takes place, type I1 spectra are due to only one superoxide species coupled to a proton of a nearby hydroxyl.The spin-Hamiltonian parameters are : g,, = 2.0018; gyy = 2.0089; g,, = 2.0773 B, = 50.37 mT; B, = k0.21 mT; Bz = fO.O1 mT with an uncertainty in B values of ca. 0.02mT. This labelling of the spectral features corresponds to assuming for the principal directions of the g tensor the z axis as the internuclear axis of the superoxide and the x axis as the direction of the n orbital hosting the unpaired electron.The reported B values are those measured along the principal directions of g and, therefore, they are not necessarily the principal values of the proton hyperfine coupling tensor B as the directions of the latter can, in principle, not coincide with those of g. Finally we note that EPR spectra very close to those in fig. 2 and 3 have been reported by Derouane et al., l1 who assigned them to a V,," centre, i.e. a triangular array of 0- ions at the (1 11) plane of MgO interacting with a nearby proton. We feel we have provided sufficient evidence that, in our case at least, the radical species interacting with the proton is indeed the superoxide 0;. Models for the 0;. - .H Species The magnitude of the hyperfine coupling constants reported above indicates that the interaction between the superoxide radical and the proton is weak. A useful reference for a strong interaction is indeed the HO, molecule, which has been matrix-isolated and studied by Catton and Symons.12 The hyperfine coupling constants are much larger and the highest value (1.72 mT) was associated with the g,, component, where we observe the smallest coupling (0.1 mT).The present data more closely recall the findings of Wertz and co-workersL3 concerning the V,, centre in an MgO monocrystal. The centre is constituted by a linear array of a hole trapped on an oxide ion (i.e. an 0- ion), an Mg2+ vacancy and a hydroxyl OH- (0-OHO-). The maximum coupling observed is 0.085mT and is almpt entirely due to dipolar interaction.The 0-se-H distance is calculated to be 3.22A. Although the present values are somewhat larger than those for the V,, centre, they are small enough to assume that the hyperfine coupling is due to the dipolar interaction without significant contributions from direct spin polarisation. The precise calculation of the dipolar interaction between the unpaired electron and the nucleus is a difficult task and requires exact knowledge of the charge distribution. In the literature several examples are available for these calculations at different level of approximation. In the case of EPR spectra information is limited to the triplet of principal B values, so that sophisticated approaches14* l5 have been discarded in favour of simpler methods. In the one adopted here, the distribution of the unpaired electron in 0, is simulated by two charges .Of - 0.5 located at the oxygen nuclei, the distance 2d between which is kept fixed at 1.35 A.Reasons for this are: (i) the spin distribution is known to be ~ymmetric;~ (ii) the HOMO has antibonding nature and is made up of two lobes. The value of 1.35 A agrees with both crystallographic determinations2 and ab initio calculations on the gas- phase radical anion. l6 Although the hydrogen atom may a priori be located in any position with respect to the 0 atoms in O,, only the highly symmetrical structures I and I1 (linear and T-shaped,3992 Surface Superoxide Species on MgO Scheme 1. respectively) have been considered. This assumption implies that the g and B tensors have the same principal directions.Assumptions similar to those above have been made by Gordyl' for a system with the same symmetry and involving the dipolar interaction of a proton perpendicular to a 2p orbital that contains an unpaired electron. The energy of the dipole-dipole coupling between the electron-spin and a nuclear-spin moments aligned along an external field is, in magnetic field units: B(e) = g , pN(3 cos2 6 - 1 )/r3 = ~ ' ( 3 C O S ~ 6 - I (6) where r is the distance between the two moments and 8 is the angle of the separation from the external magnetic field. This expression applies when the perturbation given by the hyperfine interaction is small compared to the external magnetic field. Eqn (6) also holds for structures I and I1 (scheme 1).B' assumes, however, two different values: (7) B; = g, pN[l / r 3 + 1 / ( r + 2d)3]/2 The components of the dipolar tensor are in the former case: Bz = 2B; (6 = 0); B, = B, = -Bi (6 = ~ / 2 ) (8) B, = B;,(3cos2p- 1); By = -ql; B, = B;,(3c0s26- I) (9) and in the latter case: 6 = c0s-l ( d / r ) ; P = ~ / 2 - 6. Structure I is characterized by an axial tensor, in which B, is the largest term, in clear conflict with the experimental evidence. Structure 11, instead, is characterized by a traceless non-axial tensor whose largest term, for reasonable values of 6, is B,, in agreement with experiment. A least-squares fit with the triplet of experimental B values has beep performed by varying r, keeping d fixed. The best result is obtained with r = 2.41 A (from which 6 = 37.7"), which corresponds to the set of B values: B, = +0.36 mT, By = -0.20 mT and B, = -0.15 mT.The agreement with the experiment is excellent and encourages us to consider structure I1 as representative of the O;-**H species. The distance from the H atom to the ceFtre of mass of 0, is 2.31 A and is of the order of the Mg2+-02- distance in MgO (2.10 A), as expected, because the H atom sits on an oxygen anion and the 0; ion sits on a magnesium cation. We consider this result to be a strong piece of evidence in support of the assignment. A possible structure for the O,-.-H species on the MgO surface is depicted in fig. 4. Following the conclusions in ref. (9, the site of adsorption is a four-coordinate Mg2+ at the (100, 010) edge of MgO, on which 0, sits symmetrically, in agreement with the observed equivalence of the two oxygen nuclei3 The hydroxyl OH group is assumed toE.Giamello et al. 3993 Fig. 4. Different views of the arrangement of 0; interacting with a nearby hydroxyl at the (100, 010) edge of an MgO crystal. Magnesium iqns as small spheres, hydrogen atom as filled sphere. Bond lengths in A, angles in degrees. have a bond length of 1.0 and to lie in the symmetry plane of the (100,010) dihedron as is clearly seen in fig. 4. The bulk value of 2.10 A is assumed for the Mg2'-02- dista;ce. Using for the H. e . 0 (superoxide) distance the previously computed palue of 2.41 A, the distance between Mg2+ and each of the oxygen atoms in 0; is 2.07 A, which coincides with the Mg2+-02- distance in the solid.The 0; species thus appears to be located at the surface so as to mask most efficiently the exposed Mg2+ cations. The structure in fig. 4 suggests that further refinement of the model of magnetic interaction is possible. Indeed, whereas the g tensor principal directions are imposed by the underlying crystal, and coincide with those usually assumed (e.g. y axis as the line joining the underlying Mg2+ cation and the 0; centre of mass etc.), the B tensor has different x and y axes because the H atom does not lie at the same level as 0;. We refrain, however, from more sophisticated models, because the agreement with experiment is already of the order of the experimental uncertainty. A final comment concerns the simplicity of type I1 spectra, discussed in the present paper, in sharp contrast with type I ~pectra.~ The latter spectra correspond to low extent of reactions (3) and (4), so that the surface morphology of the MgO microcrystals is not yet strongly perturbed. Accordingly, the 0, species formed act as a sensitive probe of various surface sites5 and several components are seen in the EPR spectrum.When the reaction of HX with oxygen has progressed extensively, many kinds of sites for 0; adsorption are depleted, and only the most abundant type of sites (four-coordinate Mg2+ ions at the edges) is available. Moreover, hydroxyl species are present in a large quantity at the surface as a product of the oxidation of HX, so that it becomes highly probable I33 F A R I3994 Surface Superoxide Species on MgO that each site for 0; adsorption has a nearby hydroxyl.A similar explanation holds in the case of method A as the hydroxyls are produced uia reaction (1). Financial support from the Italian Ministry of Education (Progetti di rilevante interesse nazionale) is acknowledged. Thanks are also due to CSI Piemonte (Consorzio per il Sistema Informativo) for an allowance of computer time. E. Giamello acknowledges also the receipt of a NATO Collaborative Research Grant (no. 86/560) for the cooperation with the P. and M. Curie University (Paris). References 1 (a) J. X. Wang and J. H. Lunsford, J. Phys. Chem., 1986, 90, 5883; (b) C. H. Lin, T. Ito, J. X. Wang and J. H. Lunsford, J . Am. Chem. Soc., 1987, 109,4808; (c) C. Mirodatos and G. A. Martin, in Proc. 9th Int. Congr.Catal., ed. M. J. Phillips and M. Ternan (The Chemical Institute of Canada, 1988), vol. 2, p. 899. 2 M. Che and A. J. Tench, Adv. Catal., 1983, 32, 1. 3 W. Kanzig and M. H. Cohen, Phys. Rev. Lett., 1959, 3, 509. 4 (a) A. J. Tench and P. Holroyd, Chem. Commun., 1968,471 ; (b) M. Che, E. Giamello and A. J. Tench, 5 E. Giamello, P. Ugliengo and E. Garrone, J . Chem. SOC., Faraday Trans. 1, 1989, 85, 1373. 6 A. J. Tench and R. L. Nelson, J. Colloid. Interface Sci., 1968, 26, 364. 7 E. Garrone, E. Giamello, S. Coluccia, G. Spoto and A. Zecchina, in Proc. 9th Int. Congr. Catal., ed. 8 A. Zecchina, M. G. Lofthouse and F. S . Stone, J. Chem. SOC., Furuday Trans. 1 , 1975, 71, 1476. 9 (a) P. Ugliengo, G. Borzani and D. Viterbo, J. Appl. Crystallogr., 1988,21,75; (b) W. D. S . Motherwell and W. Clegg, PLUTO78 (University of Cambridge, 1978). 10 (a) D. Cordischi, V. Indovina and M. Occhiuzzi, J. Chem. SOC., Faruday Trans. I , 1978, 74, 456; (b) J. H. Lunsford and J. P. Jayne, J . Chem. Phys., 1966,44,1487; (c) E. G. Derouane and V. Indovina, Chem. Phys. Lett., 1972, 14, 455. 11 E. G. Derouane, V. Indovina, A. B. Walters and M. Boudart, Proc. 7th Int. Symp. on the Reactivity of Solids, Bristol, 1972. 12 R. C. Catton and M. C. R. Symons, J. Chem. Soc. A, 1969, 1393. 13 P. W. Kirklin, P. Auzins and J. E. Wertz, J. Phys. Chem. Solids, 1965, 26, 1067. 14 H. M. McConnel and J. Strathdee, Mol. Phys., 1959, 2, 129. 15 W. Derbyshire, Mol. Phys., 1962, 5, 225. 16 P. Ugliengo, unpublished results. 17 W. Gordy, in Theory and Application of Electron Spin Resonance (John Wiley, New York, 1980), p. 207. Colloids Surf, 1985, 13, 231. M. J. Phillips and M. Ternan (The Chemical Institute of Canada, 1988), vol. 4, p. 1577. Paper 9/01604J; Received 17th April, 1989

ISSN:0300-9599    

DOI:10.1039/F19898503987

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

J. Chem. Soc., Furuduy Trans. I, 1989, 85(12), 39954009 Mixed-ligand Complexes of CuII- 1,lO-o-phenanthroline and its Analogues Characterized by Computer-aided Electron Spin Resonance Spectroscopy Yan Yang,? Rebecca Pogni and Riccardo Basosi" Department of Chemistry, University of Siena, Italy An extensive reinvestigation based on computer-aided ESR spectro- scopy, for Cu- 1,lO-o-phenanthroline (CuOP) and its analogue Cu-2,9- dimethyl- 1,lO-o-phenanthroline (CuDMP) has been performed in the liquid and solid phases under a large range of pH, temperature and mole ratio conditions. Equilibria involved in solution and coordination behaviour as a function of pH have been characterized by a careful analysis of magnetic ESR parameters. Selective isotopic enrichment and computer-aided ESR spectroscopy have been used for precise determination of spin-Hamiltonian parameters and correlation times.CuOP and its analogue CuDMP have been reported to possess very different chemical and biological char- acteristics despite being conformationally homologous. The 2,9-DMP form possesses a marked specificity and stereochemistry that confers upon the bidentate cupric chelate potent oxidative properties. This could be attributed to the steric hindrance to normal planar configuration caused by the presence of methyl groups close to the metal coordination site, a feature not shared by 1,lO-phenanthroline. A structure-activity relationship is proposed on the basis of difference in magnetic parameters and consequent evaluation of molecular orbital coefficients K and a2 under the assumption of an effective D,, tetragonally distorted octahedral symmetry.Our results are in favour of two distinct structures for OP and 2,9-DMP copper(I1) complexes in solution : a tetragonal structure probably with rhombic distortion for OP and a flat tetrahedral structure for DMP. The presence of one or two ligands coordinating the metal ion does not seem to play a determining role in the structural arrangement. On the contrary the presence of substituent methyl groups in positions 2- and 9- is fundamental for explaining the striking differences shown by the analogue compounds and can reconcile the manifestation of unexpected properties. Although it has been known for years that 1,lO-phenanthroline inhibits cell proliferation1 the mechanism for its antineoplastic activity is somewhat obscure.Indeed, several recent studies have suggested that a copper([) phenanthroline complex is the active form of the ligand23 and that it plays a role in the cleavage of DNA under certain conditions.*~ A likely picture of the overall mechanism assumes that a phenanthroline-Cu complex cycles between Cu" and Cu' to catalyse the reduction of oxygen to reactive radical species by reducing agents like ascorbate or cellular thiols. Cu'I - Phen --+ Cu' - Phen CU' - Phen + H202 + CuII - Phen + OH' + OH- The highly reactive hydroxyl radicals, OH' generated through the site-specific Fenton t On leave from School of Pharmacy, West China University of Medical Sciences, Chengdu, People's Republic of China. 3995 133-23996 Computer-aided ESR of Mixed-ligand Complexes Table 1.Chemical and biological characteristics of copper phenanthrolines CU(OP), Cu(DMP), ref. turnover rate constant for superoxide dismutation pH 7" redox potentials photoredox chemistry" stability constants of complexes 1:1 1:2 inhibition of proliferation 5.1 x lo8 dm3 mo1-1 s-' 2.6 x lo8 dm3 mo1-1 s-l 12, 13 + 174 mV unreactive + 594 mV reactive 14 15 16 log Kl = 9.0 log K, = 6.7 log K , z 6.1 log K, = 5.6 CCRF-CEM lymphoblasts, - 1,17 gram + and - bacteria, myco bac terium tuberculosis, yeast, fungi, viruses, mammalian cells such as HeLa, monkey kidney epitheleal ; human amnion and chick embryo fibroblasts - marginally active for 18 L 1210 cells in vitro; P388 murine lymphoma in vivo For complexes of Cu:OP 1 : 2 and Cu:DMP 1 :2. reaction are suggested to be ultimately responsible for the DNA damage in the presence of Cu-phenanthroline.6-s Furthermore the proposed ability of copper complexes to facilitate copper-dependent processes in antineoplastic activityg> lo is an attractive possibility.In this connection the understanding of copper-phenanthroline complex structures and dynamics in solution is crucial if their mechanism of action at molecular level is to be studied. In fact very subtle differences in the chemical arrangement can result in striking differences in bio-pharmacological activity. This was recently proposed in a series of tetradentate copper(I1) complexes of bis(thiosemicarbazones) l1 and is our aim in the present study for bidentate copper-phenanthrolines.Cu- 1,lO-o-phenanthroline (CuOP) and its analogue Cu-2,9-dimethyl- 1,lO-o- phenanthroline (CuDMP) have been reported to possess very different biological characteristics despite being conformationally homologous. In table 1 some of the significant dissimilarities for CuOP and CuDMP are summarized on the basis of data reported in the 1iterature.l2-ls The most striking discrepancy is related to the tendency to damage DNA already mentioned for CuOP2* '-', l9 and the lack of any toxicity and even inhibition of the damage, in the case of CUDMP.'~~*-~~ This behaviour can be related to differences in the 'turnover' rate constant for superoxide dismutation, which for CuOP is double that of CuDMP,12, l3 or may possibly be due to their redox potentials ( + 174 mV for CuOP and + 594 mV for CuDMP). l4 Other differences between the two phenanthroline complexes can be found in their toxicity to algae,23 tumour cells,24 and bacteria and fungi.25 Whether all these dissimilarities should be attributed to a difference in molecular rearrangement of the two ligands around copper in solution needs to be clarified. The 2,9-DMP form possesses a marked specificity and stereochemistry that confers upon theY.Yang, R. Pogni and R. Basosi 3997 bidentate copper(I1) chelate potent oxidative properties. This could be attributed to the steric hindrance to normal planar configuration caused by the presence of methyl groups proximate to the metal coordination site, a feature not shared by 1,lO-phenanthroline or other methylated derivatives.If a structure-activity relationship holds for the two compounds, we can anticipate finding traces of it in the ESR magnetic parameters which we could hope to discuss in terms of a simple molecular-orbital scheme. Although this has already been attempted26, 27 in single crystals, definitive results in solution have so far been impeded by the poor resolution of the ESR spectra obtained at room temperature and the multiple equilibria involved in the liquid phase which did not allow the precise determination of spectral parameters. In this paper an extensive reinvestigation based on computer-aided ESR spectroscopy, for Cu- 1,lO-o-phenanthroline (CuOP) and its analogue Cu-2,9-dimethyl- 1,lO-o- phenanthroline (CuDMP) was performed in liquid and solid phases over a large range of pH, temperature and molar ratio conditions, in order to compare homologous species in identical circumstances. Although in neither case was nitrogen superhyperfine splitting resolved, computer simulation and computational processing of spectral profiles led to additional information about speciation of copper complexes.The complexes were isotopically enriched with 65Cu in order to avoid complications arising from extra absorption peaks. Heavy reliance on the use of computer was required to overcome the relative lack of information of the ESR spectra at room temperature and to avoid constraints connected with changes in the physical state of the sample due to freezing. 28-31 Experimental 1,lO-o-phenanthroline (Merck) and 2,9-dimethyl- 1’10-o-phenanthroline (Aldrich) were recrystallized from water and from several organic solvents until there was no further change in absorption spectra. The solutions were made in doubly distilled H20, and the uncorrected pH was adjusted with either HC1 or NaOH (Baker).pH values were measured with a Metrohm Herisau E603 potentiometer. Isotopically pure 65Cu0 from Oak Ridge National Laboratory, TN(USA), was used for all ESR experiments. X-Band ESR spectra ( v = 9.87 GHz) were obtained with a Bruker ER 200D-SRC spectrometer equipped with a high-sensitivity ER 4108 TMH cavity and 100 kHz field modulation. Microwave frequency was measured with an EIP 301 counter and temperature was calibrated with an ER 4111 VT accessory thermocouple. The spectrometer was interfaced with a PS/2 Technical Instruments Hardware computer with a CPU 80386 Intel 32 bit, a clock of 26 MHz and a RAM of 2048 kilobytes, which allows acquisition, processing, simulation and comparison of the ESR spectra.The programs CUKIVL and CUCHI for the simulation of ESR spectra of fast-tumbling copper complexes in an isotropic environment are described in a previous paper32 and were modified in order to take account of the contemporaneous presence of multiple species in solution (CUKOS). All the programs are based on Kivelson’s theory for the l i n e ~ i d t h ~ ~ and include the second-order shift equation of Bruno et and the further assumption of Lorentzian lineshapes. With the input parameters giso, Ag, Aiso, AA and the rotational correlation time z,, the programs calculated the parameters of Kivelson’s equations, the linewidths, the line positions and the lineshapes.They are additionally able to accommodate two sets of nitrogens with up to four in each set. The programs run in a few seconds on a PDP-l123/plus with 512 kbytes of memory and floating-point processor or on the computer interfaced to the ESR instrument. More details of the computer simulation of copper complexes in solution at room temperature can be found in the l i t e r a t ~ r e . ~ ~ UV-visible spectra were measured with a Perkin-Elmer 200 spectrophotometer3998 Computer-aided ESR of Mixed-ligand Complexes 0 2 4 6 8 10 12 microwave frequency /GHz Fig. 1. Computed hyperfine width for the high-field M , = +$ line of the ESR spectrum of the s5Cu(OP), complex us.microwave frequency (ESR parameters given in table 2 species C). Table 2. ESR parameters of 65Cu11 complexesa of 1,lO-o-phenanthroline (CuOP) and 2,9-dimethyl- o-phenanthroline (CuDMP) giso g, CUOP A 2.150 2.317 2.066 -194.6 -503.3 -40.2 12 4 55 3 B 2.118 2.285 2.034 -225.7 -551.2 -62.9 11 2 35 7-9 C 2.121 2.259 2.052 -205.2 -512.7 -51.4 11 4 40 11 D 2.134 2.260 2.065 -129.6 -481.6 +46.3 11 6 60 3-11 CuDMP A' 2.161 2.335 2.073 -148.6 -481.1 +17.6 11 4 55 3 B' 2.130 2.296 2.046 -174.2 -562.5 + 14.3 11 2 40 7-11 a 65Cu magnetogyric ratio 0.759 58 x A values f 3.0 MHz, for z, f 15 ps. rad s-' G-l. Calculated error for g values +0.001, for Number of coordinated nitrogen atoms. equipped with a Hitachi recorder. 1 cm pathlength quartz cells were used for all the determinations in solution.CuO for UV-visible determinations was obtained from Prokeme. Buffer solutions of acid phthalate and phosphate were made according to the U.S. Pharma~opeia.~~ Results and Discussion Determination of Magnetic Parameters In their pioneering work on the role of microwave frequency in ESR spectroscopy of copper complexes,36 Hyde and coworkers stressed the point that even in the liquid phase, there exists an optimum microwave frequency to minimize the linewidth, given a properY. Yang, R. Pogni and R. Basosi - I OPPH 3999 \ Fig. 2. Computer-simulated spectra (---) for 65Cu(OP), complex (species C in table 2) at four microwave ESR frequencies. The spectrum at X-band is paired with the experimental spectrum (-) obtained at pH 11, 298 K for a 1 :2 Cu:OP molar ratio [Cu"] = 5 x mol dm-3.(a) v = 1.1 GHz, L-band; (b) v = 3.6 GHz, S-band; (c) v = 9.87 GHz, X-band; (d) v = 35.0 GHz, Q-band. value for the correlation time of the complex. Often this situation occurs at S-band where a superior resolution of the hyperfine structure appears as a consequence of cancellation of MI- and frequency-dependent terms in the expression for the linewidth. The frequency, around 8 GHz, which gives narrowest lines for a typical complex of Cu(Phen) can easily be anticipated from fig. 1 on the basis of a computational approach. Fig. 1 shows the computed width for the high-field MI = + $ hyperfine line of the ESR4000 Computer-aided ESR of Mixed-ligand Complexes ,516_ I O I P H ----- ------ : ; I t I t I I Fig. 3.Experimental ESR spectra (-) obtained at pH 3.4 (a), 7.4 (b) and 9.4 (c) for a Cu:OP metal ligand ratio equal 1 : 1 at 298 K. [Cu"] = 5 x mol dm-3. Dashed lines represent simulated spectra plotted using parameters reported for species A (a) and species B (c) in table 2. spectrum of the 65Cu(Phen), complex [C in table 21 plotted against microwave frequency. Input parameters and z, used for simulating the ESR spectrum are reported in table 2. In fig. 2 ESR spectra simulated at four different frequencies, keeping constant the spin Hamiltonian magnetic parameters and correlation time, are shown and compared at X- band with the experimental spectrum obtained at pH 11 for a Cu: OP molar ratio of 1 : 2. The fit at X-band can be considered quite good and lends credence to the tentative interpretation detailed below.Fig. 1 and 2 suggest that in this 'unfortunate' case multifrequency ESR can be of little help because lineshapes of the S- and X-band turn out to be very similar and no superhyperfine structure due to coordinated nitrogens can be resolved in either case.Y. Yang, R. Pogni and R. Basosi 400 1 1006 - 1 OPPH Fig. 4. (a) experimental ESR spectrum obtained at pH 3.4 and T = 150 K for a Cu: OP metal ligand ratio of 1 : 1 ; (b) second-derivative computer-processed ESR spectrum for the g, region. Computer-aided spectroscopy can still be used to overcome problems arising in solution from the overlap of different species characterized by similar magnetic parameters. Fig. 3 shows experimental X-band ESR spectra obtained at room temperature and at several pH values (pH 3.4, 7.4, 9.4) for a metal/ligand ratio of 1 : 1 compared in the two limit cases with the simulated spectra which gave the best fit.The magnetic parameters for the two complexes indicated as A and B in table 2 are interpreted in terms of equilibrium between two species : Cu(Phen), Cu(Phen). Besides the magnetic parameters which can be differentiated by careful computer analysis of the ESR spectra in solution, a meaningful piece of information can be inferred from the values of the correlation time definited for A (7, = 5.5 x s) and B (z, = 3.5 x lo-" s). As is evident from table 2, the parameters which eventually gave the best fit for species A are consistent with a coordination of four nitrogen atoms around copper, whereas those for species B are consistent with only two nitrogens.The small difference in the N unresolved coupling constants falls within the experimental error and can by no means be considered univocal. Fig. 4(a) shows the ESR spectrum obtained at liquid-nitrogen temperature (150 K) in identical molar ratio (1 : 1) and pH conditions to those of fig. 3 (a). From these findings, suitable g , , and A , , parameters can be derived. Fig. 4(b) shows a computational plot of the second derivative of the high-field part of the same spectrum which magnifies the features of superhyperfine splittings due to four coordinated nitrogens and allows precise definition of the experimental value. By this method preliminary values for magnetic parameters are obtained which can be inserted into room-temperature simulation programs. The only remaining unknowns are thus the number of coordinating nitrogens and the rotational correlation time which are somewhat related4002 Computer-aided ESR of Mixed-ligand Complexes Phen 0.2 0.4 0.6 0.8 c u + + X DMP 0.2 0.4 0.6 0.8 Cu++ X Fig.5. Plots of UV-visible absorbance ( A ) us. mole fractions of ligand (X) for CuOP system (a) and CuDMP system (b) at pH 3.4 (@) and 7.0 (0). The raw absorbance data for the complexes at pH 3.4 were corrected by subtraction of the absorbance due to free ICu2+l ion according to Job's method.37 Spectra at pH 7.0 were recorded on the supernatant after centrifugation to exclude effects of precipitate. because of the change in molecular weight of the complex. A routine best-fit procedure eventually yields the parameters given in table 2 under a simple assumption of Lorentzian lineshapes and a limited contribution of residual linewidth.The asymmetry of fig. 4(b) is attributed to overlap of 65Cu perpendicular components and nuclear superhyperfine splittings due to coordinating nitrogens. Furthermore, a limited contribution due to the contemporaneous presence of species B, which is still evident even at room temperature, cannot be ruled out. The attribution to species A at pH 3.4 and species B at pH 7.4 have been confirmed by UV-visible spectra recorded using the Continuous Variation Method (Job's method). By this method a series of solutions are prepared such that the sum of the total metalY. Yang, R. Pogni and R. Basosi 4003 Fig.6. Experimental ESR spectra obtained at pH 7.4 and T = 298 K for Cu(0P) molar ratios of 1 : l (a), 1:2 (b) and 1:lO (c). and total ligand molar concentrations is constant (5 x mol dm-3). The absorbances of the solutions at a given wavelength ( A ) plotted against mole fraction of ligand ( X ) gives the maximum value of A for the ligand/metal ratio of the predominant species in solution at different pH values. Fig. 5(a) shows the results obtained at 678 and 680 nm for Cu(0P). The dissimilarity in the two maxima at pH 3.4 and 7 is completely consistent with the above interpretation of ESR parameters based on experimental and computational approaches. Experimental ESR spectra obtained at 298 K and pH 7.4 for Cu(0P) molar ratios of 1 : 1, 1 :2 and 1 : 10 are shown in fig.6. Attribution to species B (CuOP) and D Cu(OP), in table 2 are reliable for the spectra of fig. 6(a) and (c), respectively. Species D exists for all molar ratios over 1 : 3 virtually under any condition of pH confirming previous assignments by other (data not shown). The interpretation of the spectrum in fig. 6(b) is intriguing because of the simultaneous presence of two species in equilibrium (B and C) and it can be achieved only by careful comparative analysis of the spectra obtained for the same molar ratio at higher pH. Under these conditions in fact the equilibrium shifts towards species C which becomes predominant at pH 1 1.4, allowing simulation of the ESR spectrum already shown in fig. 2, An ESR approach similar to that developed above for 1-10-o-phenanthroline was used to study the complexes formed by copper with 2,9-dimethyl-o-phenanthroline, and the results are summarized in table 2 together with the relative magnetic parameters.On the whole, the DMP spectra are shapeless in comparison with those of OP. In some cases the equilibria involved in solution at room temperature even prevent the4004 Computer-aided ESR of Mixed-ligand Complexes I t I t I t : I ;j Fig. 7. Experimental (-) and simulated (---) ESR spectra of CuDMP system obtained for molar ratio 1 : 1 and 298 K at pH: (a) 3.4, (b) 7.4, (c) 9.4, ( d ) 11.4. (a) was simulated with parameters of species A’ in table 2 and ( d ) assuming a superposition of 24 YO species A‘ and 76 % species B’. appearance of a single pure species. This is in reasonable agreement with the proposed stepwise stability constants for the complexes CuDMP and Cu(DMP), which are very close (see table 1).UV-visible spectra recorded using the previously cited method give, in the case of DMP, an analogous result at pH 3.4 and 7, suggesting a 1 : 1 form at both pH for the copper complexes fig. 5(b). It should be mentioned that the 3, values were slightly different and that the absorbance values at pH 7 were ca. twice those at pH 3.4. This finding ensures the involvement of two different species. However, an inconsistencyY . Yang, R. Pogni and R. Basosi gIs0 = 2.130 4005 Fig. 8. Experimental ESR spectra of CuDMP system at 298 K and 1 : 1 molar ratio compared by computer acquisition at different pH: (a) (-) pH 3.4, (---) pH 7.4; (6) (---) pH 7.4, (-) pH 9.4.between results suggested by UV and our ESR experiments at pH 3.4 exists and we are unable to explain it. In fig. 7 experimental ESR spectra recorded at molar ratio 1 : 1 at four different pH values (3.4, 7.4, 9.4, 11.4) at 298 K for CuDMP complexes are shown and compared in the limit case with the simulated spectra of best fit. Fig. 7 ( d ) was simulated using a program which in the frame of Kivelson’s theory permits the contemporaneous presence of up to three different species and their weights to be taken into account. The best fit of fig. 7 ( d ) has been obtained assuming a superposition of 24% species A’ and 76% species B’. Magnetic parameters related to species A’ and B’are given in table 2. Fig. 8 (a) shows how intensive computer acquisition of experimental ESR spectra yielded characterization of the two different species in solution at pH 3.4 (A’) and pH 7.4 (B’) for a molar ratio 1 : 1 in a preliminary step.In fig. 8(b) a similar procedure makes it possible to conclude that the same species is present at 1 : 1 molar ratio and all the basic pH values, with an increasing contribution of B’ despite the fact that a 100% pure species never appears. As the low exchange equilibria occurring in solution give rise to very complicated patterns, the spectra recorded at low temperature are of limited help and serve to rule out possibilities rather than suggest them. No evidence was found of a Cu(DMP), species under any pH, molar ratio or temperature conditions. Full speciation of copper complexes of OP and DMP, together with reliable spin- Hamiltonian parameters and correlation times are summarized in table 2.With these parameters computer-simulated spectra can be generated which agree well with the experimental data and must therefore be regarded as a possible solution for room-4006 Computer-aided ESR of Mixed-ligand Complexes Table 3. Molecular-orbital coefficients for the complexes A Cu(OP), 0.340+0.005 0.87f0.01 0.357f0.002 308.7+ 3 B Cu(0P) 0.325 f 0.005 0.88 f 0.01 0.370 & 0.002 325.5 & 3 A’ Cu(DMP), 0.296 & 0.005 0.91 f 0.01 0.401 & 0.002 332.5 f 3 B’ Cu(DMP) 0.295 f 0.005 0.96f 0.01 0.472 f 0.002 388.3 & 3 temperature spectra. Of course it has not been shown conclusively that the ESR spectra reported in this study cannot be simulated using another combination of parameters, but the overall agreement does not seem fortuitous and gives us confidence in the proposed interpretation.Calculation of Bonding Parameters from ESR Data Once we have confidence in the ESR parameters derived from the experimental data in table 2, it is possible, using the theoretical basis of Maki and M c G a r ~ e y ~ ~ for copper(I1) complexes, to apply a semiempirical LCAO MO scheme and express anisotropic g- values and hyperfine constants as a function of molecular-orbital coefficients and certain atomic constants. The construction of one-electron molecular orbitals reflects the local symmetry of copper(I1) complexes and we can hope to obtain interesting information about the differences previously detected for homologous complexes of CuOP and CuDMP.In the majority of cases for copper, the general symmetry corresponds to an octahedral structure with strong tetragonal elongation, but additional rhombic or tetrahedral distortion can arise from differences in donor atoms of the ligand or steric repulsion between chelating molecules. Chemical considerations and previous suggest that it is reasonable to confine ourselves to the case of a tetragonally distorted octahedral effective D,, symmetry, with the unpaired hole in the d,2-yZ orbital. In this case, the symmetry-adapted antibonding molecular orbitals lead to analytical equations which can be used to measure the extent of ionic bonding (a2) and the extent of the contact term (K) from the magnetic parameters in table 2: 3(g1 - 2.0023) 7 14 where P is taken as 0.036 crn-I2’ and hyperfine splittings are expressed in consistent units.In table 3 values calculated for copper complexes of OP and DMP are reported and compared. K values are lower for A’ and B’ (CuDMP) than for A and B (CuOP). The terms in K arise from the Fermi contact interaction which has its origin in a non-vanishing probability of finding the unpaired electron at the site of the nucleus. This term is assumed to be independent of the direction of the magnetic field and the maximum value is attained at an intermediate covalency of a2.,l The Kvalues for CuDMP complexes are consistent with a tetrahedral distortion of the symmetry around the metal ion. This type of distortion can lead to 3d4p hybridi~ation,,~ which explains the reduction of the anisotropic hyperfine term and accounts for a strong repulsion between the chelating groups.It is noteworthy that the tetrahedral configuration seems to be favoured even in the CuDMP 1 : 1 complex,Y. Yang, R. Pogni and R. Basosi 4007 suggesting that 2- and 9-methyl groups interfere sterically with water molecules occupying any coordination sites around the metal ion. These data confirm a very early observation by Perrin and hawk in^^^ who explained on this structural basis the striking difference in standard redox potentials for substituted phenanthrolines in copper(I1) complexes. A similar conclusion has also been drawn on the basis of electronic spectra4* and IR, electric conductivity meas~rements.~~ Higher values of K for A and B species of CuOP are to be expected if rhombic distortion of D,, symmetry induces 4s admixture of the ground state with the symmetry axes x and y orientated toward the ligand donors.This effect, similar to the spin- polarization of the 4s orbital, gives a negative contribution to the Fermi term, leading to an overestimation of the parameter E ’ ~ in table 3 whenever 3 d 4 orbital mixing is neglected. This type of analysis has already been successfully utilized by other authors in the investigation of copper complexes of amino a ~ i d s . ~ ~ ? ~ ’ Plots of E ’ ~ us. a2 show a linear relationship which reflects competition between a-bonds of different symmetries. The slight rhombic distortion of these compounds is also supported by smaller Aaniso(A ,, - Aiso) values than in parent complexes of DMP.In table 3 this value is 308.7 MHz (species A) against 332.5 MHz (A’) and 325.5 MHz (species B) against 388.3 MHz (B’). For the complexes with effective D,, symmetry, Aaniso tends to decrease with increasing s-character of the MO containing the unpaired electron, probably as a consequence of 3d4s orbital mixing. The terms containing the factor a2 arise from the dipole4ipole interaction between magnetic moments associated with the spin motion of the electron and the nucleus. This contribution is predicted to be reduced by delocalization of the unpaired electron to the neighbouring atoms. Its value decreases with increasing covalency to a minimum theoretical value of 0.5, increasing to a maximum value of 1.0 for a completely ionic copper-ligand bond.The values reported in table 3 show a trend from intermediate covalency (species A) to pronounced ionic character (species B’). Comparison of trends for a2 and E ’ ~ supports the finding that the covalancy of the in-plane a- and the 4s a-bond varies as expected within the series. However, as the a2 covalency factor describes the relative weight of the copper 3d orbital in the singly occupied ground state, the stronger covalency should result in a smaller hyperfine interaction. From this point of view our results deviate from the rule according to previous observation4** 49 in which this was ascribed to a strong contribution from ligand electronegativity. a2 can also be strongly affected by the solvent nature which weakens the covalency of metal ligand bonds, presumably interacting along the z-axis. This may account for a discrepancy with data obtained in chloroform-toluene The evaluation of the bonding parameter a2 as a function of pH variation for a given ligand and molar ratio does not provide sufficient data to establish anything specific as to coordination site, but clearly indicates a significant change in the metal-ligand binding over the range pH 6-7 for both CuOP and CuDMP.The more drastic variation in the case of CuDMP for A’ and B’ could be indicative of a more drastic change in the nature of DMP-copper(I1) binding and a significant deviation from the planarity connected with the tetrahedral distortion. Conclusion Our results are in favour of two distinct structures for 1,lO-OP and 2,9 DMP copper(r1) complexes in solution : a tetragonal structure probably with a rhombic distortion for OP and a flat tetrahedral structure for DMP.The presence of one or two ligands coordinating the metal ion does not seem to play a determinant role in the structural arrangement. On the contrary, the presence of substituent methyl groups in positions 2- and 9- is fundamental for explaining the striking differences shown by the analogue compounds in table 1 and can reconcile the manifestation of unexpected properties. For4008 Computer-aided ESR of Mixed-ligand Complexes instance the gain in stability, expected from the increase in basicity due to the inductive effect of the methyl groups, is easily outweighed by the decrease in stability brought about by the steric hindrance to coordination.If the differences in bio-pharmacological behaviour have their foundation in a structure-activity relationship, the approach proposed by us deserves to be developed for ternary complexes formed by copper-phenanthrolines and amino acids and/or peptides. Ternary complexes containing a protein constituent as second ligand are of great significance, because they are not only potential models for enzyme-metal ion-substrate complexes, but have also been proposed as carriers of antitumour agents. Although interesting preliminary papers have already been p~blished~’-~~ on this subject, definitive results have been impeded by the lack of exploitation of more penetrating methods like multifrequency and computer-aided ESR spectroscopy.Thanks are due to Dr Alicia Diaz Garcia for helpful discussion and Mr Castelli for technical assistance. The authors thanks the Italian National Council of Research (CNR) for financial support and Italian Ministry of Foreign Affairs for the grant provided to one of the authors. References 1 K. H. Falchuk, A. Krishan and B. L. Vallee, Cancer Research, 1977, 37, 2050. 2 V. D’Aurora, A. M. Stern and D. S. Sigman, Biochem. Biophys. Res. Commun., 1977, 78, 170. 3 V. D’Aurora, A. M. Stern and D. S. Sigman, Biochem. Biophys. Res. Commun., 1978, 80, 1025. 4 J. M. C. Gutteridge and B. Halliwell, Biochem. Pharmacol., 1982, 31, 2801. 5 L. E. Pope and D. S. Sigman, Proc. Natl Acad. Sci., Biochem., 1984, 81, 3. 6 D. R. Graham, L. E. Marshall, K. A. Reich and D.S. Sigman, J. Am. Chem. Soc., 1980, 102, 5419. 7 K. M. Downey, B. G. Que and A. G. So, Biochem. Biophys. Res. Commun., 1980, 93, 264. 8 L. E. Marshall, D. R. Graham, K. A. Reich and D. S. Sigman, Biochemistry, 1981, 20, 244. 9 J. R. J. Sorenson, in Metal Ions in Biological Systemhed. H. Sigel (Marcel Dekker, New York, 1982), vol. 14, p. 77. 10 W. M. Willingham and J. R. J. Sorenson, in Trace Elements in Medicine (Dustri-Verlag, Munich, 1986), vol. 3(4), p. 139. 11 R. Basosi, J. Phys. Chem., 1988, 92, 992. 12 S. Goldstein and G. Czopski, J. Am. Chem. Soc., 1983, 105, 1276. 13 S. Goldstein and G. Czopski, J. Inorg. Chem., 1985, 24, 1087. 14 B. R. James and R. J. P. Williams, J. Chem. Soc., 1961, 2007. 15 S. Sundararajan and E. L. Wehry, J. Phys. Chem., 1972, 76, 1528.16 H. Irving and D. H. Mellor, J. Chem. Soc., 1962, 5222. 17 A. Shulman and F. P. Dwyer, in Chelating Agents and Metal Chelates, ed. F. P. Dwyer and D. P. Mellor (Academic Press, New York, 1964), p. 415. 18 A. Mohindru, J. M. Fisher and M. Robinovitz, Biochem. Pharmacol., 1983, 32, 3627. 19 C. A. Reich, L. E. Mashall, D. R. Graham and D. S. Sigman, J. Am. Chem. Soc., 1981, 103, 3582. 20 D. S. Sigman, D. R. Graham, V. D’Aurora and A. M. Stern, J. Biol. Chem., 1979, 254, 12269. 21 B. G. Que, K. M. Downey and A. G. So, Biochemistry, 1980, 19, 5987. 22 S. Goldstein and G. Czapski, J. Free Radicals Biol. Med., 1986, 2, 3. 23 T. M. Florence, B. G. Lumsden and J. J. Fardy, Anal. Chim. Acta, 1983, 151, 281, and references 24 A. Mohindru, J. M. Fisher and M.Robinovitz, Nature (London), 1983, 303, 64. 25 F. Black, Nature (London), 1951, 168, 516. 26 D. E. Billing, R. J. Dudley, B. J. Hathaway and A. A. G. Tomlinson, J. Chem. Soc. A , 1971, 50, 691. 27 B. J. Hathaway, P. G. Hodgson and P. C. Power, Inorg. Chem., 1974, 13, 2009. 28 T. Vanngard, in Biological Applications of Electron Spin Resonance, ed. H. M. Swartz, J. R. Bolton and 29 C. E. Falk, E. Ivanova, B. Ross and T. Vanngard, Inorg. Chem., 1970, 9, 556. 30 J. S. Leigh and G. H. Reed, J. Phys. Chem., 1971, 75, 1202. 31 Y. Orii and M. Morita, J. Biochem. (Tokyo), 1977, 81, 163. 32 R. Basosi, W. E. Antholine, W. Froncisz and J. S. Hyde, J. Chem. Phys., 1984, 81, 4849. 33 R. Wilson and D. Kivelson, J. Chem. Phys., 1986, 44,4445. 34 G. V. Bruno, J. K. Harrington and M. P. Eastman, J. Phys. Chem., 1977, 81, 11 1 1. 35 U.S.P. XX National Formulary, 1980, 1101. therein. D. C. Borg. (Wiley, New York, 1972), p. 423.Y. Yang, R . Pogni and R. Basosi 4009 36 J. S. Hyde and W. Froncisz, Annu. Rev. Biophys. Bioeng., 1981, 11, 391. 37 F. R. Hartley, C. Burgess and R. Alcock, in Solution Equilibria (Ellis Horwood, Englewood Cliffs, NJ, 38 A. Ya. Sychev and I. Zamaraev, Zh. Fiz. Khim., 1970, 44, 2996. 39 A. H. Maki and B. R. McGarvey, J. Chem. Phys., 1958, 29, 31; 1958, 29, 35. 40 A. Abragam and B. Bleaney, in Electron Paramagnetic Resonance of Transition Ions (Clarendon Press, 41 A. Rockenbauer, J. Magn. Reson., 1979, 35, 429. 42 M. Sharnoff, J . Phys. Chem., 1964, 41, 2203; 1965, 42, 3383. 43 C. J. Hawkins and D. D. Perrin, J. Chem. Soc., 1963, 2996. 44 J. R. Hall, N. K. Marchant and R. A. Plowman, Aust. J. Chem., 1963, 16, 34. 45 Ph. Thomas, D. Rehorek and H. Spindler, 2. Anorg. Allg. Chem., 1973, 399, 175. 46 D. Kivelson and R. Neiman, J. Chem. Phys., 1961, 35, 149. 47 H. R. Gersmann and J. D. Swalen, J. Chem. Phys., 1962, 36, 3221. 48 H. A. Kuska, M. T. Rogers and R. E. Drullinger, J. Phys. Chem., 1967, 71, 109. 49 H. R. Gersmann and I. D. Swalen, J. Chem. Phys., 1962, 36, 3221. 50 C. J. Simmons, M. Lundeen and K. Seff, Inorg. Chem., 1978, 17, 1429. 51 S. V. Deshpande and T. S . Srivastava, Inorg. Chim. Acta, 1983, 78, 75. 52 W. L. Kwik and K. P. Ang, J. Inorg. Nucl. Chem., 1980, 42, 303. 1980), p. 45. Oxford, 1970). Paper 9/01 5956 ; Received 17 April, 1989

ISSN:0300-9599    

DOI:10.1039/F19898503995

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. I , 1989, 85(12), 4011-4017 Electron Spin Resonance, ENDOR and TRIPLE Resonance of some 9,lO-Anthraquinone Radicals in Solution Part 2 .-Ant hraquinonesulphonates t Reijo Makela and Mikko Vuolle Department of Chemistry, University of Jyvaskyla, Kyllikinkatu 1-3, SF 40100, Jyvaskyla, Finland We report studies on radical anions of 9,lO-anthraquinone- 1 -sulphonate, 9,lO-anthraquinone- 1,5-disulphonate, 9,l O-anthraquinone-2-sulphonate, 9,1O-anthraquinone-2,6-disulphonate and 9,lO-anthraquinone- 1,2-dihy- droxy-3-sulphonate in alkaline ethanol and water mixtures carried out by multiple resonance spectroscopy. The hyperfine coupling constants and relative signs were determined in ENDOR and TRIPLE resonance experiments. The temperature dependence of the coupling constants was measured and the g values were also estimated. Use of a modified additivity relationship suggested a new assignment of the coupling constants.Anthraquinones play a significant role as parent molecules for compounds used in biological, commercial and medical applications. 1, The photosensitivity of 9,lO- anthraquinonesulphonates in different solvents has been reported in many The analysis of coupling constants is difficult because of overlapping lines in ESR spectra and the assignment using a modified additivity relationship seems to be problematic. We have measured ESR, ENDOR and TRIPLE resonance spectra of five different radical anions of 9,lO-anthraquinonesulphonates in alkaline aqueous ethanol solution under conditions as similar as possible.The many coupling constants and rotational conformers of sulphonate groups make theoretical calculations of coupling constants tedious. The additivity relationship offers a convenient means to assign the coupling constants at specific sites in the m o l e c ~ l e . ~ - ~ ~ In this work we make use of modified calculation routine,'* which gives good results for radicals of the same type if the relative signs are known. Experimental Materials 9,lO-Anthraquinone (purum) from Fluka, 9,lO-anthraquinone- 1 -sodium sulphonate (practical grade, m.p. > 280 "C) from K & K Labs, 9,lO-anthraquinone- 1,5-disodium sulphonate (m.p. > 300 "C, techn) from EGA-CHEMIE, 9,1O-anthraquinone-2-sodium sulphonate (puriss p.a., 99 %) from Fluka, 9,l O-anthraquinone-2,6-disodium sulphonate (m.p.> 325 "C) from Aldrich and 9,lO-anthraquinone- 1,2-dihydroxy-3-sulphonate (pro analysi) from Merck were used without further purification. Other chemicals were Na,S,O, (87 %) from Merck, TCNE (98 YO) from Aldrich, ascorbic acid (p.a., 99.7 YO) from Merck and NaOH (p.a., 99%) from Merck and absolute ethanol (AaS) from OY ALKO AB. !' Part 1 : J. Chem. SOC., Faraday Trans. 1, 1987, 83, 51. 401 14012 ESR of Anthraquinonesulphonates 1p MHz Fig. 1. (a) ENDOR and (b) general TRIPLE resonance spectra of 9,lO-anthraquinone- 1- sulphonate radical anion generated with sodium dithionite in aqueous alkaline ethanol at 283 K. Pumping frequency PF = 15.614 MHz. Equipment ESR spectra were measured on a Varian E-9 spectrometer with a field-frequency lock, a Varian variable-temperature unit, a Takeda Riken Industry Co.TR 521 1D microwave counter, a Varian E 500 gaussmeter and an Apple I1 computer.15 ENDOR and TRIPLE resonance spectra were recorded with a Bruker ER-200 D-SRC spectrometer connected to a Varian E-12 magnet. A laboratory-made ENDOR coil was used with a Bruker TM,,, cylindrical cavity. UV illumination was provided by an Airam HGU 300 W mercury lamp. Sample Preparation Samples were prepared either by high-vacuum techniques or by using a laboratory- developed flow system. Radicals were generated with sodium dithionite, ascorbic acid or TCNE in 0.01-0.15 mol dm-3 NaOH aqueous ethanol solution, where the ethanol concentration varied from 50 to 90 %. Results 9,10-Anthraquinone- 1-sulphonate The poor solubility of the compound hampers the generation of anion radicals of 9,lO- anthraquinone- 1 -sulphonate in the ethanol and water mixtures.The best chemical conditions were obtained in a solution containing 30 % 0.1 mol dm-3 sodium hydroxide, 70% absolute ethanol and some crystals of sodium dithionite. An excess amount of reducing agent destroys the short-lived radical. The ESR spectrum was recorded at room temperature after heating high-vacuum samples and the ENDOR spectra in the temperature range 263-283 K (fig. 1). The ENDOR spectrum shows five proton coupling constants [table 1 (a)], without appreciable temperature dependence. The simulated ESR spectrum contains of three doublets and two triplets and fits well with the experimental spectrum. Although theR.Makela and M. Vuolle 4013 Table 1 (a). Hyperfine coupling constants (a,/mT) (obsd) for 9,lO-anthraquinone and its sulphonates measured in aqueous alkaline ethanol solution with sodium dithionite at 283 K, and calculated constants (calcd) using parameters in table 2 [column (a)] position substituent /(g value) 1 2 3 4 5 6 7 8 - obsd - calcd 1 -sulphonate obsd (2.004 15) calcd 1,5-disulphonate obsd (2.004 24) calcd 1,8-disulphonate obsd" calcd 0.056 0.096 0.055 0.093 - 0.113 0.1 10 - 0.129 - 0.133 - 0.072 - - 0.071 0.096 0.056 0.093 0.055 0.074 0.032 0.080 0.033 0.039 <0.004 0.041 0.001 0.112 0.057 0.104 0.055 0.056 0.096 0.055 0.093 0.074 0.1 13 0.076 0.1 17 - 0.129 0.133 - 0.072 0.07 1 - - 0.096 0.093 0.056 0.055 0.039 0.041 0.1 12 0.104 0.056 0.055 0.024 0.023 < 0.004 0.001 0.057 0.055 a Ref.(11). Table l(6). Hyperfine coupling constants (a,/mT) (obsd) for 9,lO-anthraquinone and its sulphonates measured in aqueous alkaline ethanol solution with sodium dithionite at 283 K, and calculated constants (calcd) using parameters in table 2 [column (b)] position 1 2 3 4 5 6 7 8 - obsd 0.056 0.096 0.096 0.056 0.056 0.096 0.096 0.056 - calcd 0.054 0.094 0.094 0.054 0.054 0.094 0.094 0.054 2-sulphonate obsd 0.122 - 0.079 0.027 0.064 0.049 0.096 0.049 (2.004 05) calcd 0.123 - 0.081 0.027 0.065 0.051 0.096 0.050 2,6-disulphonate obsd 0.120 - 0.039 0.039 0.120 - 0.039 0.039 (2.004 07) calcd 0.1 19 - 0.037 0.038 0.119 - 0.037 0.038 radical anion is known to have seven magnetically non-equivalent protons, the measured coupling constants indicate that there are magnetically equivalent protons.The TRIPLE resonance spectrum gives the same relative signs for all coupling constants. 9,lO- Anthraquinone- 1,5-disulphonate The radical anion was generated under the same conditions as that of the 1-monosulphonate except that 0.3 mol dm-3 sodium hydroxide was used instead of 0.1 mol dm-3 NaOH. The radical anion has a lifetime of only a few hours and is very sensitive to sodium dithionite. The ESR, ENDOR and TRIPLE resonance spectra show only two coupling constants with the same relative sign [table l(a)]. The third coupling was not detected in the temperature range 278-298 K, because it is small and lies within the linewidth. The ENDOR spectrum (fig. 2) shows very broad (140-190 kHz) lines.9,10-Anthraquinone-2-sulphonate A small amount of sodium dithionite in a solvent mixture of 50% absolute ethanol and 50 % 0.1 mol dm-3 NaOH reduces 2-sulphonate, generating the corresponding radical anion. The 2-sulphonate radical anion was also produced at room temperature using our4014 ESR of Anthraquinonesulphonates PF I I l,6 MHz Fig. 2. (a) ENDOR and (b) general TRIPLE resonance spectra of 9,lO-anthraquinone- 1 3 - disulphonate radical anion generated with sodium dithionite in aqueous alkaline ethanol at 278 K, PF = 14.500 MHz. '12 1,6 MHr l? PF Fig. 3. (a) ENDOR, (b) general TRIPLE resonance spectra of 9,1O-anthraquinone-2-sulphonate radical anion generated with sodium dithionite in aqueous alkaline ethanol at 283 K, PF = 14.263 MHz. modified flow system, with the results agreeing well with those of high vacuum samples.The high-vacuum samples have the advantage of being stable for many weeks. The ENDOR and TRIPLE spectra (fig. 3) show six different coupling constants with the same sign. A special TRIPLE experiment and simulation of the ESR spectrum proved that two of the protons are magnetically equivalent [table 1 (b)]. The linewidth of the ENDOR spectrum varied from 70 to 90 kHz and the hyperfine coupling constants remained stationary between the temperatures used, 263-283 K. Results are in good agreement with those reported by Dodd and Mukherjee.' ,R. Makela and M . Vuolle 4015 ?* PF '14 1,6 MHz Fig. 4. (a) ENDOR and (b) TRIPLE resonance spectra of 9, IO-anthraquinone-2,6-disulphonate radical anion generated with sodium dithionite in aqueous alkaline ethanol at 283 K, PF = 13.462 MHz.9,10-Anthraquinone-2,6-disulphonate The radical anion of 9,l O-anthraquinone-2,6-disodiumsulphonate was generated by adding a small amount of reducing agent to the solvent mixture of 10 YO 0.1 mol dm-3 sodium hydroxide and 90 Yo absolute ethanol. When sodium dithionite was the reducing agent, the ENDOR spectrum (fig. 4) showed only two couplings, which were analysed as a quintet and a triplet in the ESR spectrum [table 1 (b)]. In the same solvent mixture, both ascorbic acid and TCNE with UV irradiation produced 9,1O-anthraquinone-2,6- disulphonate radical anion. The ENDOR spectrum clearly exhibited three proton couplings, which a TRIPLE experiment showed to have the same sign.The three systems gave nearly the same results. The quintet from the sodium dithionite generated anion splits to two triplets, whose average coupling is the same as the coupling of quintet. Temperature (248-298 K) apparently had only a slight effect on the coupling constants, with variations lying within the linewidth (70-90 kHz). Dodd and Mukherjee have measured the same radical anion in propan-2-01 at pH 13.' 9,lO- Anthraquinone- 1,2-dihydroxy-3-sulphonate The ESR, ENDOR and TRIPLE resonance spectra of radical anion generated with sodium dithionite in a solvent mixture of 80 YO absolute ethanol and 20 % 0.2 mol dm-3 NaOH are shown in fig. 5. Temperature (258-288 K) had only a slight effect on the ENDOR coupling constants. The radical anion is not very sensitive to sodium dithionite but it is necessary to use UV irradiation.The ENDOR spectrum gives four coupling constants 0.165,0.105,0.057 and 0.009 mT, which the TRIPLE spectrum shows to have the same relative sign. Calculations Table 2 shows the perturbations dij (mT) in the proton coupling constant at position j when a substituent is at position i, column (a) gives the perturbations when the substituent is at position 1, 4, 5 or 8 and column (b) the perturbations for the other401 6 ESR of Anthraquinonesulphonates 0.2 mT I12 "14 l,BMHz PF Fig, 5. (a) ESR, (6) ENDOR and ( c ) general TRIPLE resonance spectra of 9,lO-anthraquinone- 1,2-dihydroxy-3-sulphonate radical anion generated with sodium dithionite in aqueous alkaline ethanol solution at 278 K, PF = 14.185 MHz. Table 2.The calculated hyperfine coupling constants a, and a, for the reference compound and the perturbation parameters di, for sub- stituted 9,lO-anthraquinonesulphonate radical anions (in mT) a, = 0.0548 a, = 0.0933 a, = 0.0542 a, = 0.0942 (4 d,, 0.0163 d13 -0.0133 d14 -0.0217 d15 0.0215 dls 0.0237 d,, -0.0387 d18 -0.0320 (b) d21 0.0689 d23 - 0.0 1 3 5 d24 -0.0265 d25 -0.0041 d,, -0.0435 d26 0.0018 dZs 0.0105 positions. For instance, the coupling constant for 9,lO-anthraquinone- 1 ,Sdisulphonate at the position 2 is calculated using the perturbations given in column (a): a2+d12+d1, = 0.133. The values of parameters are chosen to minimize the sum of the squares of the differences between the observed coupling constants and the coupling constants calculated using the additivity re1ation~hip.l~ The reference compound is 9,lO- anthraquinone, and the additivity relationship assumes the correct assignment of thisR. Makela and M.Vuolle 4017 compound. For monosubstituted radicals there are 5040 possible ways of assigning the coupling constants. Discussion The ESR spectra of radicals generated in alkaline aqueous ethanol solution using sodium dithionite as reducing agent are complex because of the reduced multiplicity. Some of the observed lines represent two theoretical lines, so that instead of two triplets a quintet is observed. The spectra of monosubstituted radicals contain seven coupling constants and reliable assignments are tedious to carry out. We used a modified additivity relationship and the reference compound 9,lO-anthraquinone with Pedersenl' uniquely assigned coupling constants.The additivity relationship method can be applied only if coupling constants are detected under identical conditions with regard to solvent system, temperature and method of producing radicals. The calculated perturbations allowed the prediction of coupling constants of 9,lO-anthraquinone radical anions with a variety of sulphonyl substituents at different positions of the reference molecule. In the case of the 1,5-disubstituted anthraquinone the lines of the ESR spectrum were broadened and we never succeeded in detecting the third coupling constant despite the use of many different reduction systems. If heavy UV irradiation is applied for a long period, the radical is destroyed, while a shorter time causes loss of the sulphonate group.7 The ENDOR spectrum of the 2,6-disulphonate radical generated with sodium dithionite showed only two coupling constants, but three coupling constants were detected when other reduction methods were used and the difference between the two overlapping triplets was 0.006-0.007 mT depending on the system used? References 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 a M.B. Hocking and S. M. Mattar, J. Magn. Reson., 1982, 47, 1987. T. Julich, K. Scheffler, P. Schuler and H. B. Stegman, Magn. Reson. Chem., 1988, 26, 701. I. Loeff, A. Treinin and H. Linschitz, J. Phys. Chern., 1987, 87, 2536. I. Loeff, A. Treinin and H. Linschitz, J. Phys. Chem., 1984, 88, 4931. P. J. Baugh, G. 0. Phillips and J. C. Arthur Jr, J. Phys. Chern., 1966, 70, 3061. K. Kano and T. Matsuo, Bull. Chem. Soc. Jpn, 1974, 47, 2836. N. J. F. Dodd and T. Mukherjee, Biochem. Pharmacol., 1984, 33, 379. M. Vuolle and R. Makela, J. Chern. Soc., Faraday Trans. 1, 1987, 83, 51. B. Venkataraman, B. G. Segal and G. K. Fraenkel, J. Phys. Chem., 1959, 30, 1006. J. A. Pedersen, Mol. Phys., 1974, 28, 1031. J. A. Pedersen, Handbook of EPR Spectra from Natural and Synthetic Quinones and Quinols (CRC Press, Boca Raton, Florida, 1984). I. B. Goldberg and B. M. Peake, J. Phys. Chem., 1977, 81, 571. M. G. Bakker, R. F. C. Claridge and C. M. Kirk, J. Magn. Reson., 1987, 74, 503. H. Joela and R. Makela, J. Magn. Reson., 1989, in press. H. Joela and E. Salo, Acta Chem. Scand., Sect. B, 1985, 39, 131. J. A. Pedersen, J. Chem. Soc., Faraday Trans. 1, 1988, 84, 3223. R. Makela and M. Vuolle, to be published. Paper 9/01600G; Received 17th April, 1989

ISSN:0300-9599    

DOI:10.1039/F19898504011

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

.I. Chern. SOC., Faraday Trans. I, 1989, 85(12), 4019-4030 Electron Paramagnetic Resonance Spectra of the Fe,(CO), Radical trapped in Single Crystals of PPN+FeCo(CO),t John R. Morton, Keith F. Preston* and Yvon Le Page Division of Chemistry, National Research Council of Canada, Ottawa, Ontario, Canada KIA OR9 Paul J. Krusic* Central Research and Development Department, E. I. du Pont de Nemours and Company, Wilmington, Delaware 19898, U.S.A. Anisotropic, single-line EPR spectra observed in the temperature range 4 8 0 K in single crystals of PPN'FeCo(C0); doped with small amounts of the radical Fe,(CO), have been assigned to two electronically and geometrically distinct forms of the radical which exist in thermal equilibrium. The least energetic isomer has the orientation and the low symmetry of the host-crystal molecule with a single carbonyl bridge, one five-coordinate Fe and one six-coordinate Fe nucleus.The unpaired spin in this isomer is essentially confined to the five-coordinate Fe nucleus in a d,, orbital. The high-temperature species is believed to have a second carbonyl bridge and C,, symmetry with a SOMO of the representation B, and unpaired spin density shared equally between the metal nuclei. At temperatures in excess of 80 K the spectra of these two isomers are replaced by that of a third electronic doublet with g-tensor components which are approximately the average of those for the other two. The temperature dependence of the spectra is characteristic of exchange between two sites, and the third species is believed to be a fluxional variant of the low- temperature isomers.The great variety of structures found amongst polynuclear metal carbonyls undoubtedly stems from the ability of carbon monoxide to form bridging and capping bonds, as well as regular terminal bonds, to transition-metal atoms. In many instances, the energy barriers separating bridged and unbridged structures of the same carbonyl are small, and this can have important consequences for both dynamic and static structures. Recent NMR studies,'v2 for example, have shown that fluxional behaviour due to the rapid exchange of bridging and terminal CO ligands can persist into the solid state. Subtle changes in environment can sometimes lead to dramatic structural changes because of these same small energy differences between geometric isomers.A particularly striking example is provided by the isoelectronic series of binuclear carbonyls, Co,(CO),, FeCo(CO);, Fe,(CO)i-, which in the solid state have molecular structures containing two, one and no carbonyl bridges, re~pectively.~ Co,(CO), itself exists as a doubly bridged form in the pure solid and as unbridged forms in matrix-isolated Infra~ed~-~ and EPR'o-'2 studies suggest that this type of isomerism persists in coordinatively unsaturated and free-radical metal carbonyls. In particular, there are indications that Fe,(CO),, one of the first free-radical carbonyls detected by EPR, may exist in solution as two isomers, one of which is bridged." Intrigued by the possibility of isolating isomers of a free-radical metal carbonyl, we have initiated an EPR study of paramagnetic binuclear metal carbonyls doped into a t NRCC no.30522, du Pont Contribution no. 4573. 40194020 EPR of the Fe,(CO), Radical variety of single-crystal hosts. In this article, we present the results of our investigation of the Fe,(CO), radical doped into PPN+FeCo(CO); crystals. Experimental The diamagnetic host PPN+FeCo(CO), [PPN = (Ph,P),Nl was prepared according to the method published by Ruff.I3 It was recrystallized from ether-hexane containing traces of the radical Fe2(CO); in the following manner: to 30 mg PPN+FeCo(CO), dissolved in a minimal quantity of THF was added ca. 2 cm3 of ether followed by 0.1 cm3 of lo-, mol dm-3 Na+Fe,(CO); which was ObtainedlO by mixing equimolar solutions of Na,Fe,(C0),14 in THF and [ferrocenium] [BF,]/CH,Cl,.0.5 cm3 n-hexane was added as a layer, and the solution was stored in a refrigerator in a nitrogen glove box. Solutions of Na+Fe,(CO), in 2-MeTHF were obtained by the same method of one- electron oxidation. lo PPN+HFe,(CO); was prepared according to the procedure of Chin and Bau.15 EPR examination showed that the material naturally contained traces of a free radical (see below). Single crystals were selected with the aid of a polarizing microscope and oriented on a Nonius diffractometer according to the known3 structural parameters of the host. Crystals so oriented were sealed with epoxy glue into the ends of 4 mm quartz tubes such that each of the standard set16 of orthogonal axes X , Y, 2 in turn was aligned along the tube axis.A pointer was attached to the tube to indicate the direction of one of these axes in the perpendicular plane. The tube was placed in the Dewar insert of one of two cryostats centred in the rectangular cavity of a Varian El2 spectrometer. For temperatures in the range 4-100 K an Oxford Instruments helium cryostat was used; for temperatures in excess of 100 K a cold nitrogen-flow cryostat was employed. EPR spectra were recorded and measured with standard accessories as a function of angle between the tube pointer and the d.c. magnetic field. Results At temperatures below 10 K the EPR spectrum of the doped crystal of PPN+FeCo(CO), consisted of a single intense line (A in fig. 1) accompanied by weak satellite absorptions located ca. & 5 G from its centre.A second, much weaker single-line spectrum (B in fig. 1) was barely detectable at 10 K and below, but grew in intensity at the expense of the first signal as the temperature was raised. Both signals were markedly anisotropic, but neither split for any orientation of the crystal. This absence of site-splitting conformed with the known3 space-group ( P i ) of the host crystal. Resonant magnetic fields for both species A and B were measured as a function of orientation in all three crystal planes at 40 K. Plots of g2 against angle showed the expected sinusoidal behaviour (fig. 2), and least-squares fitting procedures were used to extract the tensor elements (table l).17 At temperatures in excess of 50 K the linewidths of both species increased (fig. 3, 4), and at temperatures in the vicinity of 80 K the spectrum was detectable only for certain limited orientations.However, above 100 K the linewidth .decreased and a new single- line spectrum appeared which corresponded to neither A nor B. The anisotropy of the new species C was explored at 120 K and its g2 tensor (table 1) established in the usual manner. By alternately raising and lowering the temperature of the crystal it was established that the ensemble consisting of A, B and C was thermally reversible : both line intensities and resonant magnetic fields were recovered unchanged for all three species after a thermal cycle. Using a microwave power sufficiently low to prevent saturation of either signal, the dependence of the intensities of A and B on temperature in the range 5-60 K was established (fig.5).J . R. Morton, K. I;. Preston, Y. Le Page and P. J, Krusic 402 1 Fig. 1. First-derivative EPR spectra of species A and B in a single crystal of PPN+FeCo(CO), doped with Fe2(CO); at 40 K for H, 165" from X in the X Y plane. A 71i g2 4.15 4.05 -- 0 30 60 80 UOlSOi 4.38 4.12 1 - 30 60 90 l201501 450 4.25 4.00 1 - 30 80 90 1201501 0 X Y -X Y z -Y X z -X 0 Fig. 2. Plots of g2 for A and B in the three crystal planes. Circles represent the experimental data, curves are the least-squares best-fit sine functions. Spectra of polycrystalline PPN+HFe,(CO); and of frozen solutions of Na+Fe,(CO), in 2-MeTHF were recorded (fig. 6) at 4 and 103 K. The spectra were very similar and clearly arose from the same carrier. In these matrices, unlike PPN+FeCo(CO),, no trace of isomer A was detected at any temperature of observation (2 4 K).Discussion Dealing first with the relationship between the three species A, B and C , we note that A and B exist in thermal equilibrium with each other. Furthermore, species C has an average g value identical to the mean of the average g values for A and B, and a g-tensor in the crystal-axis system (table 1) which is very nearly equal to the average of the tensors4022 EPR of the Fe,(CO); Radical Table 1. The g2 Tensors of species A, B and C in a single crystal of PPN+FeCo(CO),; the principal values of g and their direction cosines in the XYZ crystal axis systema g2 tensor principal values and direction cosines of g X Y 2 xx YY zz A 4.1407 0.0187 0.1158 B 4.0922 0.0207 - 0.0872 C 4.1238 0.0169 0.0000 0.0187 4.1485 0.0073 0.0207 4.2360 -0.0070 0.0169 4.1978 0.0068 0.1 158 0.0073 4.3652 - 0.0872 - 0.0070 4.1780 0.0000 0.0068 4.2570 2.0219 0.2347 0.3697 2.0092 0.0702 - 0.8990 -0.8501 -0.5219 2.0298 0.9769 0.0106 -0.2134 2.0375 0.1962 0.9707 -0.1391 2.0537 0.6544 - 0.3556 0.6673 - 2.0496 0.2132 0.9700 -0.1168 2.1012 0.391 5 0.0525 0.9187 2.06 19 0.3884 0.7528 0.5314 2.0634 0.0147 0.1 163 0.993 1 a The axis system is defined as follows: Z is parallel to the triclinic c-axis; Y is perpendicular to c in the bc plane; X is perpendicular to both Y and Z(c), forming a right-handed set.T/K 45 50 55 60 65 70 (x10) 75 (x10) 80 (x10) 85 (x10) 90 (x10) B A Fig. 3. First-derivative EPR spectra of the single crystal as a function of temperature.Arbitrary orientation of H,.J . R. Morton, K. F. Preston, Y. Le Page and P. J . Krusic 4023 lo 8 1 I 0 0 20 40 60 80 100 120 TIK Fig. 4. Peak-to-peak linewidths (G) of the three species A, B and C as a function of temperature. 0, A; b, B; ., c. + + 1 + I I 0 5 10 15 20 25 100 KIT Fig. 5. Natural logarithm of the ratio of intensities of A to B as a function of inverse temperature. for the other two. We conclude that A and B are isomers and that C is a fluxional variant of the same radical with their 'average' structure. The situation is that" of two-site chemical exchange in which the finite lifetimes of the species contribute to their observed linewidth. Incomplete averaging in the vicinity of 75 K (fig. 3) augments the linewidth to the point where the spectrum is hard to detect.Examination of the principal values of g for A and B (table 1) shows at a glance that we are dealing with two electronic doublets (S = $) which have distinct electronic configurations. Both g tensors are rhombic and have one component remote from the4024 EPR of the Fe,(CO), Radical 2.0582 Fig. 6. First-derivative EPR spectra of (a) polycrystalline PPN+HFe,(CO); at 4 K, and (b) a solution of 2-MeTHF containing Na+Fe,(CO); frozen (-) at 103 K, liquid (---) at 200 K. Table 2. Certain bond lengths and direction cosines in the orthogonal axis system of the PPN+FeCo(CO); crystal direction cosines length bond /nm X Y Z Fe-Co Fe-C, Fe-C, Fe-C, Fe-C, Fe-C, co-c, co-C6 co-c, co-c, 0.2585 0.221 1 0.1692 0.1711 0.1676 0.1673 0.1774 0.1731 0.I782 0.1720 0.2 176 - 0.053 1 - 0.798 1 - 0.4493 0.9028 0.2279 0.7730 0.3921 - 0.3833 - 0.6823 0.9472 0.8986 0.2352 - 0.4388 -0.0106 - 0.6936 - 0.2602 0.5322 0.6055 - 0.4754 perpendicular to plane Fe-CeC, - 0.9235 0.121 7 C ,-Fe< , - 0.8894 0.1374 c6-co-c8 0.4554 -0.0863 Fe-Co-C, -0.8721 0.0803 0.2356 0.5548 - 0.4355 - 0.778 1 - 0.4300 0.6833 - 0.8862 - 0.3454 0.7876 0.4096 0.3637 0.4827 0.4361 0.8861 other two which are 'nearly' equal. The clear distinction between the two, however, is that this unique g value is the maximum principal value for A, but the minimum value for B. The g tensor of A is typicallg of a d,, free radical, and a consideration of the principal directions bears this out. g,,, for A lies 9" from the perpendicular to the C,-Co-C, plane, and gmin lies 7" from the perpendicular to the Fe-Co-C, plane in theJ.R. Morton, K. 1". Preston, Y. Le Page and P. J. Krusic 4025 Fig. 7. Structure of the host crystal anion in PPN'FeCo(C0); showing certain principal directions of g for the radicals A and B. Table 3. Comparison of g-tensor components of A and B with certain known species A 40 2.0219 2.0375 2.1012 B 40 2.0092 2.0537 2.0619 Fe,(CO),(p-PPh,)" 77 2.0132 2.0454 2.0973 Fe,(CO); in 2-MeTHF 103 2.0094 2.0501 2.0557 Fe,(CO); in HFe,(CO); 4 2.0087 2.0542 2.0582 a Data from ref. (20). diamagnetic host; the intermediate g value lies 22" from the Fe-Co bond (table 2, fig. 7). The situation here parallels that found2' in the phosphido-bridged Fe,(CO),(p-PPh,) radical, and the g tensors for the two species are in fact similar (table 3).In both cases, a close correspondence was found between principal directions of g and vectors emanating from the five-coordinate metal nucleus ; no close correlation was found for vectors originating at the six-coordinate metal nucleus. In structure A, the Fe2(CO), radical evidently adopts the low-symmetry singly- bridged geometry of the host,3 with the unpaired spin essentially confined to the five- coordinated end of the molecule. Just as in the pseudo-isoelectronic phosphido-bridged radical,20 the Co-Fe, Co-C, and Co-C, bonds define an equatorial plane of a distorted trigonal bipyramid ; the coordinatively unsaturated iron atom is formally FeI d7 and the unpaired electron occupies one of the e' pair (xy,x2-y2) of d orbitals.The bridge provides an intrinsic Jahn-Teller distortion which lowers the symmetry and removes the degeneracy (fig. 8); the unpaired electron then occupies the d,, orbital. This inference is supported by our extended Hiickel calculations21 for the radical using the 134 FAR 14026 EPR of the Fe,(CO); Radical 0 I b O3 h c, Fe' d7 Fig. 8. d-Orbital scheme for trigonal bipyramidal and distorted trigonal bipyramidal arrangement of ligands. geometry established3 for the host anion: the SOMO (fig. 9) closely resembles that deduced for the Fe,(CO),&-PPh,) radical with unpa3-ed spin density essentially confined to Fe d,, and p orbitals on C, and C,. As in the phosphido-bridged analogue, the large positive g shift perpendicular to the equatorial plane ( z ) arises through spin-rbit coupling between the erstwhile degenerate pair of e' orbitals.20 We also note the presence of a 'near'- or 'pseudo'-mirror-plane of symmetry in the singly bridged host anion,3 namely Fe-Co-C, (fig.7). It is not surprising, therefore, to find one g value (gmin) lying close to the perpendicular to that plane. Unlike species A, a close correlation between principal directions of g and bond directions in the host was not found for species B. This suggests that B arises through a considerable molecular rearrangement which makes correlation with the host crystal structure difficult. Certainly the drastic change in form of the g tensor shows that the electronic configuration undergoes a major alteration during the conversion of A to B. In spite of the lack of precise directional information, there are a number of clues which clearly demonstrate that B is a doubly bridged version (C2v symmetry) of the radical Fe,(CO); (fig.10). First we note that the g tensor for B is not axial and that minimum g lies almost perpendicular (104"), rather than parallel to the metal-metal bond of the host (fig. 7). The principal values of g are, in fact, strikingly similar to those2, of a free radical obtained by the one-electron oxidation of Fe,(CO)i- (table 3 and fig. 6) in 2- MeTHF solution, and to those of the radical naturally present in crystals of PPN+HFe,(CO); prepared by the method of Chin and Bau.15 Unquestionably, this same species was also observed'' upon reduction of Fe(CO),, although under conditions of poorer spectral resolution which obscured the slight non-axiality of g.It was also observed in several other ~ y s t e m s . ' ~ ~ * ~ ~ These observations are best accounted for by a common assignment to the doubly bridged structure of fig. 10 for B and the radicals observed1'* 22 in solution. The tendency to form a carbonyl bridge has already been notedJ. R. Morton, K. F. Preston, Y. Le Page and P. J . Krusic 4027 Fig. 9. Contour plots from two points of view of the singly occupied molecular orbital of Fe,(CO),, isomer A, deduced from extended Huckel calculations. in the phosphido analogue Fe,(CO),@-PPh,), where intramolecular exchange of carbonyls undoubtedly occurs via a CO-bridged intermediate. l2 Additional support for the assignment of a doubly bridged structure to B comes from the theoretical calculations of Thorn and Hoffman.24 Their predictions for the upper filled and semioccupied orbitals of such a structure are reproduced in fig.10. The remoteness of the filled b, orbital from the semioccupied b, orbital in this scheme ensures a g,, value (along the twofold axis) very close to that of a free spin. (The b, and lower a, levels belong to a cluster of six closely spaced level^.^^,^^). Positive g shifts are anticipated for the other two principal axes through spin-orbital interactions with the filled a, and a, orbitals. Evidently the latter two orbitals are close to each other and their precise order will determine which of the g components is the greater, along the metal-metal bond ( y ) or perpendicular to that bond (z).Unfortunately, our directional information for the two large g values of B is imprecise because of the proximity of the values. However, the direction of minimum g is very well defined and lies not too far from the twofold axis of the 'most likely' doubly bridged structure to arise from the host structure. Examination of a model of the host molecule (fig. 7) shows that carbons 2 and 4 are the most propitiously placed for the formation of a second carbonyl bridge, and that bridging with carbon 2 would require the least internal rearrangement. It is therefore with some satisfaction that we note that gEin lies only 11" from the plane which bisects Fe-Co-C, and Fe-Co-C,, and 104" from Fe-Co, i.e. close to the twofold axis 134-24028 EPR of the Fe,(CO), Radical t" OC-Fe7 -Fe -CO / 'co j 'co C c 0 0 Fig.10. Molecular-orbital scheme of Thorn and Hofmann for the doubly bridged C,, form of Fe,(CO),. Axis labels correspond to the labelling of the g principal axes in tables 1 and 3. of the proposed CZv structure. Evidently, bridging occurs for one carbonyl only, namely C,, since only one configuration of B is observed. Judging by the results of 13C0 exchange in the Fe,(CO),(p-PPh,) radical,', one would anticipate equally facile bridging via C, in thefree molecule. However, it would appear that within the confines of the crystalline environment, the imprisoned Fe,(CO), is subject to considerable constraint. Compelling evidence for the proposed geometry and electronic structure of isomer B comes from our ~reliminary,~ studies of Fe,(CO), in single crystals of PPN+HFe,(CO),.The principal g values of the radical in the latter host are very close to those determined (table 3) from a powder sample and to those of B in the PPN+FeCo(CO); host. More importantly, their location within the crystal structure shows that the radical adopts the C,, symmetry of the host hydridocarbonyl anion15 and that the g tensor is entirely in accordance with the predictions of theory.,, If we accept the notion that B arises from A through the formation of a second carbonyl bridge, then C corresponds to an averaged structure in which the bridge is continually forming and breaking. This is thought', to be the mechanism by which scrambling of the equatorial carbonyls occurs in Fe,(CO),(p-PPh,). However, we should emphasize that the equilibrium observed here was completely reversible in the sense that on recooling from temperatures where C was observed, A and B were recovered in their original conformations within the host crystal.In other words, the unpaired spin was localised on the same Fe nucleus in A after a complete cycle of heating and cooling. Presumably spatial restrictions within the confines of the host crystal prevent the effective interchange of the five- and six-coordinate ends of the radical which can take place in solution. Although the experimental evidence points to a substantial molecular rearrangement in the conversion of A to B, one would expect little movement of theJ. R. Morton, K. F. Preston, Y. Le Page and P . J . Krusic C 4029 A B Fig.11. Hypothetical potential-energy curve for A (2A" in C,) and B (,B, in C,,) as a function of the CO-bending reaction coordinate. Fe-Fe bond. In fact, in the thermally averaged species C, the minimum g value lies almost exactly perpendicular (89") to the host metal-metal bond. The temperature dependence of the intensities of A and B (fig. 5 ) clearly shows that in the crystal of PPN+FeCo(CO); the low-symmetry isomer of Fe,(CO);, A, is the more stable. In frozen solvents and in solid PPN+HFe,(CO),, however, that isomer is not detected : even at the lowest temperature (4 K), only species B is observed. Moreover, the measured isotropic g value of 2.0385 for the radical in THF at -80 "C and the equivalence of the 57Fe nuclei1* suggest that Fe,(CO); retains the doubly bridged structure in solution.The presence of bridging ligands in Fe,(CO), in solution is confirmed by the infrared which shows a characteristic vibration at 1730 cm-l. (This observation does not, of course, permit us to distinguish between the two isomers A and B.) We conclude that in the free-state isomer B has the lower energy; owing to constraints imposed by the crystal structure, however, the preferred structure in PPN+FeCo(CO), is the singly bridged isomer A. This unusual situation is almost certainly responsible for the pronounced curvature at low temperatures in the van't Hoff plot of fig. 5, although a detailed explanation escapes us. Suffice it for present purposes to note that above 20 K the van't Hoff curve approximates to the usual straight line with a slope that corresponds to an enthalpy difference of 730 J mol-1 (1 74 cal mol-l) between A and B in the crystal.Formally, one may envisage A and B as two distinct minima separated by a hump in the potential surface of C (fig. 11). Assuming a simple two-site exchange model, we estimate from Arrhenius plots of increased linewidth (taken from fig. 4) against inverse temperature activation energies of 6.8 f 0.4 and 5.1 f 0.4 kJ mol-l (1.6 and 1.2 kcal mol-l) for the forward (A -+ B) and backward reactions, respectively. Poor agreement between the difference in the activation energies (1.7 0.6 kJ mol-l) and the direct estimate of AH from the plot of fig. 5 clearly reflects our lack of understanding of the detailed thermodynamics and kinetics of this system.The energy scale suggested by these estimates (fig. 11) is certainly commensurate with a reaction coordinate consisting of CO bending. Unfortunately, it was not possible to extract useful information from the weak satellite resonances which accompanied the single-line spectra of A and B (fig. 1). The intensities of these satellites relative to the principal absorptions showed that they arose from interactions with several (at least two) 13C nuclei present in natural abundance in the radicals. The linewidth of the satellites showed substantial anisotropy, but resolution into individual 13C resonances was not achieved for any orientation. For species A the4030 EPR of the Fe,(CO), Radical I3C splitting varied between 8 and 10 G; for B the satellites were weak and difficult to measure except for a few orientations where they showed a splitting of ca.10 G. The magnitude of the splitting observed for A is certainly consistent with the proposed structure and the 13C tensors already established20 for Fe,(CO),@-PPh,). We are indebted to Drs E. Wasserman and D. L. Thorn for helpful discussion and to Mr Steven A. Hill and Mr Regent Dutrisac for technical assistance. References 1 J. R. Lyerla, C. S. Yannoni and A. A. Fyfe, Ace. Chem. Res., 1982, 15, 208. 2 B. E. Hanson, M. J. Sullivan and R. J. Davis, J. Am. Chem. Soc., 1984, 106, 251. 3 H. B. Chin, M. B. Smith, R. D. Wilson and R. Bau, J. Am. Chem. Soc., 1974, 96, 5285. 4 G. G. Sumner, H. P. Klug and L. E. Alexander, Acta Crystallogr., 1964, 17, 732. 5 K. Noack, Spectrochim.Acta, 1963, 19, 1925. 6 G. Bor, Spectrochim. Acta, 1963, 19, 2065. 7 M. Poliakoff and J. J. Turner, J. Chem. SOC. A, 1971, 2403. 8 R. L. Sweany and T. L. Brown, Inorg. Chem., 1977, 16, 415. 9 A. F. Hepp and M. S. Wrighton, J. Am. Chem. Soc., 1983, 105, 5934. 10 P. J. Krusic, J. San Philippo Jr, B. Hutchinson, R. L. Hance and L. M. Daniels, J. Am. Chem. Soc., 1981, 103, 2129. 11 T. Lionel, J. R. Morton and K. F. Preston, Inorg. Chem., 1983, 22, 145. 12 R. T. Baker, P. J. Krusic, J. C. Calabrese and D. C. Roe, Organometallics, 1986, 5, 1506; R. T. Baker, J. C. Calabrese, P. J. Krusic, M. J. Therien and W. C. Trogler, J. Am. Chem. SOC., 1988, 110, 8392. 13 J. K. Ruff, Znorg. Chem., 1968, 7, 1818. 14 (a) J. P. Collman, R. G. Finke, P. L. Matlock, R. Wahren, R. G. Komoto and J. I. Brauman, J. Am. Chem. Soc., 1978, 100, 11 19. (b) H. Strong, P. J. Krusic and J. San Filippo Jr, Znorg. Synth., 1986, 24, 157. 15 H. B. Chin and R. Bau, Znorg. Chem., 1978, 17, 2314. 16 J. S. Rollett, Computing Methods in Crystallography (Pergamon Press, London, 1965), chap. 3. 17 J. R. Morton and K. F. Preston, J. Magn. Reson., 1983, 52, 457. 18 J. E. Wertz and J. R. Bolton, Electron Spin Resonance: Elementary Theory and Applications (McGraw- 19 M. C. R. Symons, Chemical and Biochemical Aspects of Electron Spin Resonance Spectroscopy (Wiley, 20 P. J. Krusic, R. T. Baker, J. C. Calabrese, J. R. Morton, K. F. Preston and Y. Le Page, J. .4m. Chem. 21 D. A. Pensak and R. J. McKinney, Inorg. Chem., 1979, 18, 3407; R. J. McKinney and D. A. Pensak, 22 P. J. Krusic and R. Subra, J. Am. Chem. Soc., 1989, 111, in press. 23 P. A. Dawson, B. M. Peake, B. H. Robinson and J. Simpson, Inorg. Chem., 1980, 19, 465; B. M. Peake, M. C. R. Symons and J. L. Wyatt, J. Chem. Soc., Dalton Trans., 1983, 1171; C. Amatore, J-N. Verpeaux and P. J. Krusic, Organometallics, 1988, 7, 2426. Hill, New York, 1972), chap. 9. New York, 1978), chap. 12. Soc., 1989, 111, 1262. Znorg. Chem., 1979, 18, 3413. 24 D. L. Thorn and R. Hoffmann, Znorg. Chem., 1978, 17, 126. 25 J. R. Morton, K. F. Preston, A. J. Williams and P. J. Krusic, unpublished data. Paper 9/01581G; Received 17th April, 1989

ISSN:0300-9599    

DOI:10.1039/F19898504019

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. I, 1989, 85(12), 4031-4038 Electron Paramagnetic Resonance Study of Transition-metal- ion Impregnated Brookite Titanium Dioxide Powders Angelo Amorelli, Jeffrey C. Evans" and Christopher C. Rowlands School of Chemistry and Applied Chemistry. University of Wales College of Cardif, P.O. Box 912, Cardig CF1 3TB. Samples of polycrystalline brookite titanium dioxide (TiO,) powders have been impregnated with the transition-metal ions Fe3+, Cr3+ and Mn2+, and subjected to calcination at various temperatures up to 950 "C. The EPR spectra of samples heated to 650 "C indicated that the dopant ions had migrated from the surface into the bulk of the brookite. For samples heated above 650 "C, significant changes in the EPR spectra were observed. These were attributed to the structural transformation of the brookite crystal host first to the anatase and subsequently the rutile forms of TiO,.Titanium dioxide exists as three crystal polymorphs, anatase, brookite and rutile. For the rutile modification, extensive EPR work has been performed in both the single- crystal'? , and polycrystalline p o ~ d e r . ~ - ~ Similar studies have also been reported for the anatase form of Ti0,.5 In contrast, very little work has been carried out on the rarer brookite form. Rostworowski et aL6 have reported and analysed the EPR spectrum of substitutional Fe3+ within naturally occurring brookite TiO, single crystal, whilst Ohtani et al.' have considered the application of brookite powders as possible photocatalysts. In the present study we have impregnated samples of polycrystalline brookite with transition-metal ions and subsequently subjected the doped samples to high-temperature calcination for 2 h.The effect of this treatment was investigated using electron paramagnetic resonance, a sensitive technique for determining the environment of the paramagnetic species. For brookite, it has been established by Pauling and Sturdivant8 that it is a polymorph of TiO, with a DiE space group. Its crystallographic parameters are given in table 1, along with those of rutile and anatase for comparison. Brookite contains eight molecules per unit cell, i.e. eight equivalent Ti4+ ions surrounded by a distorted octahedron of oxide ions. Experimental Materials Samples of polycrystalline brookite TiO, powder were supplied by Tioxide UK Ltd.Transition-metal ions were obtained from BDH spectrosol solutions for atomic absorption, containing 1 mg of metal ion per cm3 of solution. Procedure Batches of brookite powder were impregnated to predetermined concentrations of dopant ions, i.e. Cr3+, Fe3+ and Mn2+. Loadings of 500 ppm wt/wt dopant ion to solid were prepared. The method for impregnation has been described previ~usly.~ All calcinations were performed in a double tube furnace controlled by a 'West ' temperature controller, employing a Pt 13 % Rh : Pt thermocouple and allowing direct temperature settings. 403 14032 Brookite Titanium Dioxide Powders Table 1. Crystallographic parameters of the three titanium dioxide polymorphs unit cell dimensions/nm density/g symmetry a b C - anatase 3.9 D4a 0.3758 - 0.9514 (tetragonal) brookite 4.0 D'2h 0.9166 0.5436 0.5135 (orthorhombic) - rutile 4.23 D4h 0.4584 - 0.2953 tetragonal EPR spectra were obtained on a Varian El09 spectrometer operating in the X band (9.5 GHz). Experimental g values were determined with reference to a standard marker : diphenyl picrylhydrazyl (DPPH) for which g = 2.0036.Results and Discussion The results of this study demonstrate that the transition-metal ions, impregnated onto the surface of the polycrystalline brookite TiO, powders, can migrate into the bulk of the crystal when heated to temperatures in excess of 600 "C. This behaviour is analogous to that reported for both the rutile and anatase polymorphs of TiO,. The transition-metal ions were impregnated as Fe3+, Cr3+ and Mn2+ onto the brookite; however, it was observed that each ion exhibited different oxidation state behaviour, and therefore each ion will be discussed separately. At the same time, it should be noted that the gross structural changes observed for the crystal host were the same in each case, i.e.brookite transforming to anatase and subsequently rutile in the range 650-850 "C. Iron(Ir1)-impregnated brookite TiO, The EPR spectrum of the Fe3+-doped polycrystalline brookite powder heated at 250 "C exhibited a weak broad resonance at g = 4.3 which is assigned to surface-bound Fe3+ ions with a high-spin electron configuration. Underlying the whole spectrum is a very broad (400 mT) but diffuse signal, which has previously been associated with iron-doped TiO, and ascribed to an iron species, possibly with a low-spin config~ration.~ Iron is well known for its ability to exist in a variety of spin states, i.e.S = 5 / 2 high-spin, S = 3/2 mid-spin and S = 1/2 low-spin.1° As the series of spectra demonstrate in fig. 1, this iron signal appears to remain even to relatively high temperatures, manifesting itself as a baseline anomaly. Upon 'rutilisation' at 850 "C this signal disappears [fig. 1 (41. The form of the signal prevents analysis, but it is possible that a relatively stable iron/TiO, surface phase in the form of an iron titanate could be responsible. The signal at g = 4.3 is analogous to that observed after iron impregnation of both rutile and anata~e.~ Upon calcination at 650 "C a new EPR spectrum is discovered, as shown in fig.1 (b). This spectrum is unlike any other previously reported for Fe/TiO, systems, and by analogy with anatase and rutile behaviour this result suggests that the Fe3+ ions have migrated from the surface into the bulk of the brookite crystals. The work by Rostworowski et aL6 on single crystals of brookite obtained from deposits in Switzerland concluded that Fe3+ was naturally incorporated at substitutional cation sites within the lattice. Using Q-band ESR spectroscopy at quite high temperatures (200 "C) they determined that the spectrum could be described by the spin-Hamiltonian parameters. g = 2.002, D = 0.1 17 cm-l and E = 0.033 cm-l.A . Amorelli, J . C. Evans and C. C . Rowlands 4 J. 339.0 mT 4033 Fig. 1. EPR spectra of iron-impregnated brookite calcination products.Calcination temperature/ "C: (a) 350, (b) 650, (c) 750, ( d ) 850. Table 2. Resonances observed in the X-band EPR spectrum attributed to Fe3+ in brookite resonant resonant field/mT g field/mT g 76.5 8.75 212.5 3.15 81 .O 8.26 326.0 2.05 129 .O 5.19 335.0 2.00 152.0 4.40 367.0 1.82 168.5 3.974034 Brookite Titanium Dioxide Powders In the present study, the field positions of the Fe3+/brookite resonance in the powder spectrum are given in table 2, corresponding to the spectrum in fig. 1 (b). The structure of iron(II1) EPR powder spectra has been treated theoretically both in Aasa's paper1' and previously by Dowsing and Gibson.', The latter gave a more general overview, listing several organic systems on models, but concentrating in the main on high E / D ratios.Brookite has E/D = 0.28, similar to that of ferrichrysin which gives a spectral line at 64 mT and several around 155 mT. These peaks correspond with the observed low-field side of the Fe3+ brookite spectrum, and are consistent with Dowsing's proposal that for a resonance to be detected below 70 mT a value of D > 0.1 cm-l is predicted, and with an expected E / D approaching 0.33. Since such a signal is detected, the values of D and E determined by Rostworowski et aZ.6 correlate with the powder spectrum detected in this study. Upon calcination of the Fe3+/brookite powders at the slightly higher temperature of 750 "C further EPR signals were observed to develop around the g = 2 region [fig. 1 (c)]. These peaks originate from Fe3+ ions within an anatase ~tructure.~ This novel result indicates that the higher temperature of calcination has induced some degree of structural transformation of the crystal host from the brookite to the anatase form.As the final spectrum in fig. 1 illustrates, calcination at 850 "C completely converts the brookite polymorph into the rutile form of TiO,, thus preventing the familiar Fe"+/rutile spectrum. Chromium( 111)-impregnated Brookite TiO, Similar experiments were carried out for the corresponding chromium-doped brookite powder. The EPR spectrum for the sample heated to 250 "C exhibited an intense peak at g = 1.97 [fig. 2 (a)]. This signal is attributed to a surface-bound chromium species, and is the same as that found for the correspondingly doped rutile and anatase powders.' Heating of the samples to 650 "C produced samples exhibiting the ESR spectrum, shown in fig.2(b). This contains an intense g = 1.97 resonance and an additional weak scattering of peaks. Table 3 lists the various resonant field positions for this spectrum. Although weak, the latter set of peaks can be resolved and are found to occur at fields not previously detected in anatase or rutile doped with chromium, it is therefore proposed to be due to Cr3+ ions in brookite TiO,, i.e. located at substitutional sites. This Cr3+/brookite spectrum can be modelled13* l4 by the following spin-Hamiltonian parameters: g = 1.97, D = 0.12 cm-' and E = 0.03 cm-l for an S = 3/2 system, where Z = g/?HS+D[S:+ 1/3S(S+ l)]+E(S:-S2,). Using these values for g, D and E, the following values of geff were calculated (with experimentally derived values in parentheses) : gj = 1.54 (1.48); gj, = 5.15 (5.13); gi = 5.58(5.46); g: = 2.40 (2.52) gi not observed g: not observed where g j is the resonance corresponding to the transition within the +3/2 Kramer's doublet in the X-direction and g: is the transition within the 5 1/2 Kramer's doublet. The close fit of the calculated g values with the experimental data confirms that the weak resonances can be attributed to substitutionally incorporated Cr3+.However, the strong signal at g = 1.97 does not fit in with the weaker pattern of resonances. This peak is usually associated with a chromium surface-bound ion, and yet comparison with the iron(II1) system would suggest that diffusion into the brookite bulk is a relatively easy process.Therefore, it is tentatively proposed to be due to bulk incorporated Cr5+ ions; such species have previously been reported accommodated at substitutional sites within PbTiO3.l5A . Amorelli, J. C. Evans and C. C. 1 339.0 mT 3. Row lands 30 mT - \/I P v 1. 4035 x 10 Fig. 2. EPR spectra of chromium-impregnated brookite calcination products. Calcination temperature/ "C: (a) 350, (b) 650, (c) 750, ( d ) 850. Subsequent heating of the Cr/brookite powders to 750 "C induced the structural transformation to anatase, as indicated by the ESR spectrum [fig. 2(c)]. More severe calcination at temperatures in excess of 800 "C produces the final and irreversible transition of the crystal structure to the rutile modification [fig.2(d)]. Manganese(@-impregnated Brookite TiO, Impregnation of brookite with Mn2+ solution followed by air-drying at 100 "C produced a doped powder exhibiting a spectrum attributed to Mn2+ ions located upon the surface of the crystal. This is shown in fig. 3(a), and consists of two sextets possessing different4036 Brookite Titanium Dioxide Powders Table 3. Resonances observed in the X-band EPR spectrum attributed to Cr3+ in brookite resonant resonant field/mT g field/mT g 122.4 5.46 270.0 2.52 130.5 5.13 340.0 1.97 136.8 4.89 450.9 1.48 2 0 ' Fig. 3. EPR spectra of manganese-impregnated brookite calcination products. Calcination temperature/ "C: (a) 100, (b) 350, (c) 650, ( d ) 750, (e) 850.A . Amorelli, J. C. Evans and C. C. Rowlands 4037 Table 4.Resonances observed in the X-band EPR spectrum attributed to Mn4+ in anatase resonant resonant field/mT g field/mT 8 163.2 4.16 363.8 1.87 254.8 2.67 431.8 1.57 319.2 2.13 peak-to-peak linewidths (ppw), indicating that two different manganese species are involved. (I) g = 2.00; (11) g = 2.00; A possible explanation is that the species are (I) isolated Mn2+ and (II) clustered Mn2+ ions. In the latter case, broadening of the sextet would be expected. Clustering of surface ions has been proposed previously for chromium-doped rutile Ti0,.3 Subsequent calcination at higher temperatures, in the range 250-550 "C produced samples giving EPR spectra with rather diffuse structures and with barely resolvable hyperfine splittings. Fig. 3(b) shows such a result for a sample treated at 350 "C.Work by RelP6 on the corresponding manganese-impregnated rutile TiO, produced a series of ill-defined spectra in the temperature range 450-750 "C. These observations, a feature of Mn work, were explained by the presence of a variety of manganese species : (a) Mn2+ in the bulk at substitutional or interstitial sites, (b) Mn3+ ions and (c) the formation of an MnTiO, phase [cf. the iron(xI1) system discussed above]. In this study, the spectrum [fig. 3 (b)] of samples heated at 350 "C possesses a strong g = 2 resonance, which is more apparent in powders calcined at 650 "C [fig. 3(c)]. This signal is proposed to be due to Mn2+ ions within the brookite structure. This resonance at g = 2.0 hints at further unresolved structure, since the peak-to-peak widths and hyperfine splittings are not constant.Furthermore, the spectrum exhibits rather weak but broad 'shoulders' at the low-field side of the EPR scan. Such structureless wings have been discussed by Griscornl' for a variety of Mn2+ spectra in glasses and their origins were found to be caused by systems with a D/hv ratio of ca. 0.2-0.5, i.e. the region where brookite values of D might be expected. The migration of the Mn2+ ions indicated by this spectrum is in agreement with both the iron and the chromium experiments. After calcination at 750 "C, the EPR spectrum [fig. 3(4] of the products exhibited a wealth of resonances throughout the 400 mT scan. The symmetry of the spectral lines suggested that this system is attributable to Mn4+ ions substitutionally incorporated within an anatase structure (cf.Cr3+). This spectrum can therefore be analysed in a similar fashion, as has been done previously5* 18* 19, to determine the value of D, the zero- field splitting energy. Since Mn4+ has S = 3/2 and is present within anatase's axial crystal field, the spin-Hamiltonian is given byls a = 8.6 mT; ppw < 1 mT ppw 5 mT. a = 8.6 mT; .X = gpHS+D[Si- 1/3S(S+ l)] + ALS and the spectrum can be given to second-order by v = v, + ~ ( 3 cos2 e - 1) + 6 cos2 e (1 - cos2 e) ~ 2 / h v , . Therefore the splittings of the weaker fine-structure transition will give 2 0 and 4 0 . The field values of the transitions are given in table 4. From the field splitting of the - 3/2 ++ - 1 /2 and 1 /2 c-) 3/2 perpendicular transitions [the separation is 2 0 as shown in fig.3(d)] the value for D is 84.0 mT. If we then use this result to determine the second- order splitting for the 1/2 + - 1/2 transition, i.e. between the two strongest sextets,4038 Brookite Titanium Dioxide Powders then we obtain 42.3 mT from D2/hvo. The experimental value was found to be 44.6 mT, confirming the assignment of the spectrum to Mn4+ within an anatase structure. Calcination at 850 "C produces a sample with an EPR spectrum assigned to Mn4+ trapped at substitutional sites within rutile (cf. Cr3+ and Fe3+), as reported by Cordischi et aL20 Conclusions We have determined that the transition-metal ions impregnated upon the surface will migrate into the bulk of brookite when subjected to high-temperature calcination.The various oxidation states taken by these ions, as a function of temperature and environment, are analogous to those previously reported for the rutile and anatase crystal forms of TiO,. These results are summarized as follows: (i) Fe3+(soln) -+ Fe3+(surface) -, Fe3+(brookite) -, Fe3+(anatase/rutile) (ii) Cr3+(soln) --* Cr5+(surface) --+ Cr5+/Cr3+(brookite) -, Cr3+(anatase/rutile) (iii) Mn2+(soln) --+ Mn2+(surface) -+ Mn2+(brookite) -+ Mn4+(anatase/rutile) Using the resultant EPR spectra, we have been able to monitor the concomitant structural transformation of the crystal host as a function of temperature, i.e. brookite anatase 850 OC rutile. These experiments have demonstrated that brookite converts mainly to anatase around 750 "C, and that only at the higher temperature of 850 "C will it convert completely to the rutile form.This novel observation is made possible by the sensitivity of EPR to environmental conditions. We thank Dr N. Okey of Tioxide UK Ltd for supplying the samples of Brookite TiO,, and Mr Brian Barnard of Tioxide UK Ltd for generously permitting the use of the sample of Brookite which he made in 1967. A.A. thanks the S.E.R.C. for the award of a CASE studentship. References 1 H. J. Gerritsen, S. E. Harrison and J. P. Wittke, Phys. Rev. Lett., 1959, 24, 153. 2 T. Castner Jr, G. S. Newell, W. C. Holton and C. P. Slichter, J. Chem. Phys., 1960, 32, 668. 3 J. C. Evans, C. P. Relf, C. C. Rowlands, T. A. Egerton and A. J. Pearman, J. Muter. Sci. Lett., 1984, 4 J. S. Thorp and H. S . Eggleston, J. Muter. Sci. Lett., 1985, 4, 1146. 5 A. Amorelli, J. C. Evans, C. C. Rowlands and T. A. Egerton, J. Chem SOC., Faraday Trans. I, 1987,12, 6 J. A. Rostworowski, M. Horn and C. F. Schwerdtfeger, Chem. Phys. Lett., 1973, 34, 231. 7 B. Ohtani, J. Handa, S. Nishimoto and T. Kagiya, Chem. Phys. Lett., 1985, 120, 292. 8 L. Pauling and J. H. Sturdivant, 2. Kristallogr., 1928, 68, 239. 9 A. Amorelli, M.Sc. Thesis (Cardiff, 1986). 3, 695. 3541. 10 M. Maltempo, Chem. Phys. Lett., 1979, 60, 441. 11 R. Aasa, J. Chem. Phys., 1970, 52, 3919. 12 R. D. Dowsing and J. F. Gibson, J. Chem. Phys., 1969, 50, 1. 13 S. Doeuff, M. Henry, C. Sanchez and J. Livage, J. Non Cryst. Solids, 1987, 89, 84. 14 J. R. Pilbrow, J. Mugn. Reson., 1978, 31, 479. I5 R. Boettcher, W. Brunner, B. Milsch, G. Volkel, W. Windsch and S . T. Kirillov, Chem. Phys. Lett., 16 C. P. Relf, Ph.D. Thesis (Cardiff, 1986). 17 D. L. Griscom and R. E. Griscom, J. Chem. Phys., 1967,47, 8. 18 G. Burns, J. Appl. Phys., 1960, 32, 1074. 19 T. Ebert and J. Scheve, Magnetic Resonance and Relaxation. Proc. XIVth Colloque Ampere, ed. R. 20 D. Cordischi, M. Valigi, D. Gazzoli and V. Indovina, J. Solid State Chem., 1975, 15, 82. 1986, 129, 546. Blinc (North Holland, Amsterdam, 1967). Paper 9/0 159 1 D ; Received 1 7th April, 1989

ISSN:0300-9599    

DOI:10.1039/F19898504031

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. I, 1989 85(12), 40394046 Electron Paramagnetic Resonance Study of Monoclinic Zirconium Dioxide Polycrystalline Powders doped with Paramagnetic Transi tion-metal Ions Jeffery C. Evans," C. Ruth Owen and Christopher C. Rowlands School of Chemistry and Applied Chemistry, University of Wales College of Cardif, P.O. Box 912, Cardif CFl 3TB Samples of monoclinic zirconium dioxide have been prepared with loadings of either 200ppm wt/wt Cr3+ or 200ppm Fe3+ ions, and the effect of calcination on these powders has been observed. For Fe3+, at temperatures below 1170 "C, a broad signal at g z 2 is observed due to low-spin Fe3+ ions and another signal at g FZ 4.3 due to high-spin Fe3+ ions, both on the surface. At temperatures above the transition temperature of 1 170 "C, migration into the ZrO, bulk is observed, i.e.loss of the peak at g = 2 along with the appearance of other absorptions at ca. g = 4. In the case of Cr3+ ions doped onto ZrO,, only absorptions due to Cr5+ ions situated on the surface are observed at g = 1.97. On heating, the intensity of this signal decreases. Calcination through the transition temperature (1 170 "C) does not cause migration of either the Cr3+ ions or the Cr5+ ions into the bulk. Treatment with ethane- 1,2-diol results in a nine- line EPR spectrum, demonstrating that the chromium ions undergo further oxidation to Cr6+ ions on the surface. The Cr3+ doped ZrO, powder undergoes a colour change from white to grey at the transition temperature. Spectra recorded on samples at liquid-nitrogen temperatures suggest that Zr4+ ions have been reduced to Zr3+ ions.ZrO, is used in the ceramics industry and the presence of transition-metal ions within its structure is of interest with regard to ceramic stains. Hence, in this present study, an investigation of the effect of calcination temperature upon polycrystalline samples of ZrO, doped with Cr3+ or Fe3+ ions and a comparison with similar doping in the TiO, system has been made. In the case of TiO,, transition-metal ions are initially observed on the surface and subsequent heating causes migration of the transition-metal ions into the bulk.'?' The EPR spectra observed depend upon the polymorph of TiO,. Experimental Materials Zirconia (ZrO,) was supplied by Aldrich (purity 99.99 %). Doping with transition-metal ions was effected using BDH spectrosol solutions for atomic absorption, which contained 1 mg of metal ion per 1 cm3 of solution.Method ZrO, was impregnated with a 200 ppm wt/wt loading of the individual metal ions. This was achieved by adding 2 cm3 of the spectrosol standard to 10 g of ZrO,. The sample was then diluted with distilled water, placed on a hot plate and heated to dryness. The mixture was rapidly stirred to allow for an even distribution of the metal ions on the 40394040 Monoclinic Zirconium Dioxide Polycrystalline Powders sample. The product was then air-dried overnight at 110 "C. Calcinations of the sample were then carried out using a carbolite furnace, the samples being placed in 10 cm long ceramic boats at the centre of the furnace.The furnace was purged with N, before introduction of the samples to remove any impurities. The doped samples were then calcined for 2 h. EPR spectra were obtained for the doped samples using a Varian El09 spectrometer operating at 9.5 GHz (X-band). The g values of the samples were obtained with reference to a standard marker; diphenyl picrylhydrazyl (DPPH) ( g value = 2.0036). All spectra were recorded on samples at room temperature unless stated otherwise. Results and Discussion Cr3+ Impregnation Fig. 1 shows the spectra of the Cr3+-doped ZrO, and the effect of increasing calcination temperature. Fig. l(a), the EPR spectrum prior to calcination, shows that there is a broad peak centred at g x 2 which is due to surface Cr3+ ions. Superimposed on it is a sharp peak at g = 1.97 which we attribute to surface Cr5+ ions, similar to those observed in the TiO, ~ystem.~ Therefore, this shows that the surface of ZrO, has oxidised the Cr3+ ions. Calcination at 600-950 "C increases the intensity of the Cr5+ absorption with the loss of the broad Cr3+ ion peak.On calcining to 1400 "C, the intensity of the Cr5+ peak decreases considerably (fig. 2). The transition temperature of ZrO, between the different crystal structures of monoclinic and tetragonal is 1170 "C. Calcining above this temperature does not appear to drive the Cr3+ or Cr5+ ions into the bulk of the ZrO, as there are no changes in the EPR spectra obtained. The chromium ions may have been further oxidised to oxidation state VI and then they either migrated into the ZrO, or remained on the surface (Cr6+ ions are not detectable by EPR).Reaction of Cr6+ ions with ethane- 1,2-diol results in the formation of the bis-ethane- 1,2-dioatochromate(v) anion which can be detected by EPR due to the reduction of the Cr6+ ions to form a Cr5+ anion complex with the ethane- 1,2- d i ~ l . ~ This can be used as a test for Cr6+ ions and produces a nine-line EPR spectrum, from eight equivalent protons in the two molecules of ethane- 1,2-diol. The g value of this complex is centred around g = 1.98 1. 1 cm3 of ethane- 1,2-diol was added at room temperature to ca. 0.1 g of 200 ppm Cr3+- doped ZrO, which had previously been heated to 1300 "C for 5 h. The tube was incubated for 5 min at 80 "C and the spectrum obtained is shown in fig. 3. Ethylenediamine tetra-acetic acid (EDTA) in water with Cr3+ ions forms a purple solution but does not react with Cr5+ or Cr6+ ions.Each sample of Cr3+-doped ZrO, heated at 600,900 and 1400 "C produced a pale purple solution, but the intensity of the colour decreased with increasing calcination temperature, suggesting that the chromium ions were being oxidised to higher oxidation states. No changes in the EPR spectra were observed with increasing calcination temperature (except for the appearance of absorptions due to Fe3+ ions present as impurities), showing that there appears to be no migration of the chromium ions into the ZrO, bulk. There is, however, a change in the colour from off-white to dull grey of the Cr3+-doped samples of ZrO, on calcining through the transition temperature (1 170 "C).This is similar to TiOZ5 where heating can result in the Ti4+ ions being reduced to Ti3+ ions, giving a grey colouration. Hence in this case it is possible that the Zr4+ ions are reduced to Zr3+ ions. Zr3+ ions are paramagnetic but, in order to observe them, the samples have to be run at liquid-nitrogen temperatures. The EPR spectra of the Cr3+-doped ZrO, heated at 900 and 1400 "C were run at 77 K and are shown in fig. 4. It seems that there is no change in the Cr3+-doped ZrO, sample heated to 900°C (except for greater resolution), therefore no Zr3+ ions are present. However, spectrum of the 1400 "C sampleJ. C. Evans, C. R. Owen and C. C. Rowlands 404 1 Fig. 1. EPR spectra of Cr3+-doped ZrO,: (a) at room temperature and calcined for 2 h ; (b) 600 "C; (c) 900 "C; ( d ) 1400 "C.clearly shows a second absorption at g = 1.86, characteristic of a Z P ion, situated next to an electron defect.6 These results suggest, therefore, that the Cr3+-doped ZrO, heated above the transition temperature of ZrO, causes reduction of the Zr4+ ions to Zr3+ ions. Hence, with the oxidation of the Cr3+ ions, it is possible to predict that a reduction/oxidation couple has been established. It can be concluded that Cr3+ ions are doped onto the surface of ZrO, and subsequent heating results in the oxidation of the Cr3+ ions to Cr6+ ions via Cr5+ ions. In addition to this oxidation, on heating above the transition temperature, the Zr4+ ions undergo reduction to Zr3+ ions, causing the grey colouration of the sample.There does not appear to be any migration of the Cr3+ ions into the bulk. If there was migration, the chromium ions would be in oxidation state VI. These results are quite different from4042 Monoclinic Zirconium Dioxide Polycrystalline Powders 100 1 X 0 2 00 400 600 800 1000 1200 1400 calcination temperature/" C Fig. 2. The effect of calcination temperature on the surface doped Cr5+ ion EPR intensity. P 1 I I Fig. 3. The EPR spectrum of the bis-ethane- 1,2-dioatochromate(v) anion together with its structure. The inset shows the third-derivative spectrum. those observed with TiO,, where migration of the Cr3+ ions does occur into each of its polymorphs, rutile, brookite and anatase. In the case of rutile, migration occurs at temperatures above 400 "C, whereas with anatase, migration occurs at temperatures above 500 "C.Fe3+ Impregnation Fig. 5 shows the effect of calcination on Fe3+-doped ZrO,. Fig. 5(a), the EPR spectrum prior to calcination, shows a broad signal centred at g w 2. This is due to low-spin Fe3+J. C. Evans, C. R. Owen and C. C. Rowlands 28.1 mT H A B 4043 Fig. 4. EPR spectra of Cr2+-doped ZrO, calcined at 900 "C (A) and 1300 "C (B), recorded at room temperature (a) and 77 K(b). ions. Also, there is another sharper peak at g = 4.27 due to high-spin Fe3+ ions. Both these peaks are surface Fe3+ peaks, similar to those observed in the TiO,.system.' The spectrum obtained from samples calcined at 400 "C is similar to that obtained from the non-calcined sample but has been run at an increased gain. From the samples calcined at 800, 1100 and 1400 "C, respectively, it can be seen that there is a decrease in the low- spin Fe3+ signal at g = 2 and an increase in the high-spin Fe3+ signal at g = 4.27.From samples calcined at 1100 "C, it can be seen that there is an increase in the number of signals observed around the g z 4 region. This, we believe, is due to the migration of Fe3+ ion into the ZrO, bulk, together with the signal of the remaining high- spin surface Fe3+ ion [peak marked e, fig. 5(c) and (e)]. Increasing the temperature of calcination to 1400 "C results in the loss of the low-spin Fe3+ absorption completely, and greater resolution of the peaks at g x 4 is obtained. This shows that the Fe3+ ions have undergone migration into the ZrO, bulk and that there could be more than one site.The measured g values for the various peaks are given in table 1.4044 Monoclinic Zirconium Dioxide Polycrystalline Powders 20.7 mT t+m .1 OPPH JJ a Fig. 5. EPR spectra of Fe3+-doped ZrO,: (a) at room temperature and calcined for 2 h; (b) 400 "C; (c) 800 "C; ( d ) 1100 "C; (e) 1400 "C. in is Heating of the Fe3+-doped ZrO, samples through the transition temperature did not volve a colour change to dull grey, i.e. Zr4+-+Zr3+ (unlike the Cr3+-doped ZrO,). This also confirmed by the absence of absorbances from Zr3+ ions in the EPR spectra; The positions of the lines in the EPR powder spectrum of high-spin Fe3+ (d5, therefore, the Zr4+ ions are not reduced to Zr3+ ions. S = 5/2) can be calculated by solving the spin Hamiltonian : J? = gpHS+ 1/3D[3S;-S(S+ 1)]+ E(Sz-5':).Two very similar treatments,*$ using the above Hamiltonian, have listed the allowed transitions along the three principal axes of the crystal field by plotting the reduced crystal field strength D/hv (or its inverse) against the reduced resonance energy gpH/hv (or its inverse) for various values of the asymmetry ratio E/D. Axial symmetry is represented by E / D = 0 and this is shown in the low-field EPR spectrum by theJ. C. Evans, C. R. Owen and C. C. Rowlands Table 1. The measured g values of the Fe'I'- doped ZrO,, on heating to 1400 "C resonancea g a 9.1 1 b 4.75 C 4.48 d 4.38 eb 4.27 f 4.19 g 3.89 h 3.61 "On fig. 5 Surface-bound species. 4045 Fig. 6. EPR spectrum of 57Fe3+-doped TiO, (rutile) on substitution into Ti4+ sites in the TiO, lattice after calcination at 950 "C for 2 h.appearance of an absorbance at g M 6. Increases in this ratio represent a departure towards rhombic symmetry and this is characterised by the appearance of three absorbances in the g = 4 region with the disappearance of the g = 6 absorbance that was observed previously. E / D = 1 /3 represents rhombic symmetry and this is reflected in the coincidence of the three EPR absorbances at g = 4.3. E / D > 1/3 represents a return to axial symmetry. From this the environment of the Fe3+ can be determined. Using the notation of A a ~ a , ~ if we consider the transition 3-4 when E / D = 1/3, then the x, y and z absorptions all coincide and are depicted by a nearly isotropic line at g x 4.3. This is usually found with low symmetry iron (111) complexes.This indicates that the peak observed in our results (g = 4.27) is that of an Fe3+ ion in a rhombic environment and we also attribute this to a high-spin surface Fe3+ ion peak. E / D = 0.25 for the 3-4 transition results in g values of 4.86,4.28 and 3.53. Our values of b = 4.75, f = 4.19 and h = 3.61 in fig. 5(e) are similar and suggest E / D = 0.25. The appearance of further lines [fig. 5(e)] suggests that the Fe3+ ions can also exist in more than one environment. The environment for ZrO, is monoclinic (< 1170 "C), which is less ordered than the tetragonal system (> 1170 "C). Our results show that Fe3+ ions doped onto a 21-0, surface have a rhombic environment [fig. 5 (a)] and do not enter the ZrO, bulk until near the monoclinic/tetragonal phase change [fig.5 (6) and 5 (e)].4046 Monoclinic Zirconium Dioxide Polycrystalline Powders At 1400 "C [fig. 5 (e)], no evidence for Fe3+ in an environment other than near rhombic is seen. Even though new absorbances at g x 4 are observed, we believe this to arise from Fe3+ ions at interstitial sites in a mixture of monoclinic and tetragonal ZrO, crystal lattices. For comparison, the EPR spectrum'' of 200 ppm 57Fe3+ (showing Fe hyperfine due to I = 1/2) doped into rutile TiO, (tetragonal crystal structure) is shown in fig. 6. We have shown previously that Fe3+ is substitutionally incorporated into rutile Ti02.7 We conclude that Fe3+ ions are doped onto the surface of ZrO, and subsequent heating results in their migration into the 21-0, bulk at the transition temperature of ZrO, (1 170 "C).This migration is such that the Fe3+ ions are substituted interstitially into both monoclinic and tetragonal environments. There is no colour change in the Fe3+-doped sample of ZrO, at its transition temperature and no change in the EPR spectrum when run at 77 K, hence there is no reduction of the Zr4+ ions to Zr3+. C.R.O.thanks the S.E.R.C. for the award of a CASE studentship, and Cookson Research plc for the provision of a bursary. The 57Fe EPR spectrum was recorded by Dr A. Amorelli. References 1 J. C. Evans, C. P. Relf, C. C. Rowlands, T. A. Egerton and A. J. Pearman, J. Mater. Sci. Lett., 1984, 2 J. S. Thorp and H. S. Eggleston, J. Mater. Sci Lett., 1985, 4, 1146. 3 J. C. Evans, C. P. Relf, C. C. Rowlands, T. A. Egerton and A. J. Pearman, J. Mater. Sci. Lett., 1985, 4 C. J. Winscorn, Mol. Phys., 1974, 28, 1579. 5 E. Serwicka and R. N. Schindler, Z . Naturforsch., Teil a, 1981, 36, 992. 6 M. M. Abraham, L. A. Boatner, J. 0. Ramey and M. Rappaz, J. Chem. Phys., 1984,81, 5362. 7 A. Amorelli, J. C. Evans, C. C. Rowlands and T. A. Egerton, J. Chem. SOC., Faraday Trans. I , 1987, 8 R. D. Dowsing and J. F. Gibson, J. Chem. Phys., 1969, 50, 294. 9 R. Aasa, J. Chem. Phys., 1970, 52, 3919. 3, 695. 4, 809. 83, 3541. 10 A. Amorelli, Ph.D. Thesis (University of Wales, 1988). Paper 9/01587F; Received 17th April, 1989

ISSN:0300-9599    

DOI:10.1039/F19898504039

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

J . Chern. SOC., Furuduy Trans. 1, 1989, 85(12), 40474052 Binuclear Radical Complexes of Heavy-metal Fragments containing Ruthenium, Osmium, Rhodium and Gold Stephan Kohlmann, Volker Kasack, Eberhard Roth and Wolfgang Kaim" Institut fur Anorganische Chemie der Universitat, Pfafenwaldring 55, D- 7000 Stuttgart 80, Federal Republic of Germany Homobinuclear chelate complexes of the 3,6-bis(2-pyridyl)- 1,2,4,5-tetra- zine anion radical ligand with the metal fragments [Os(bpy)12+ (d6), [Rh(norbonadiene)]+ (d8) and [Au(PPh,)J+ (d'O) have been studied by ESR spectroscopy. The unique suitability of this radical ligand for coordination of heavy metal fragments and for ESR investigations is discussed. Paramagnetic complexes of the precious metals are scarce mainly for two reasons : metals of the 4d and 5d series in the periodic table are notorious for avoiding configurations with an odd number of d electrons,' thus corresponding more to main- group element analogues.A second possibility, the coordination of such metals in their stable d2" configurations to the generally nucleophilic anion radical ligands,,, often fails because of the facile reducibility of ionized precious metal centres to the elements. Using the negatively charged ' spin label ligand '4 3,6-bis(2-pyridyl)- 1,2,4,5-tetrazine radical anion (bptz*-, 1)5-7 we have been able to generate and characterize by ESR the first binuclear radical complexes of osmium(n), rhodium(1) and gold(1). 1 . . . 1- The radical anion ligand (1) is perfectly suited for the stabilization and ESR characterization of metal It offers (i) two identical chelate coordination sites, each with one basic pyridyl 'anchor', (ii) a less negative reduction potential of -0.72 V us.SCE which can be further increased considerably by metal c~ordination,~ (iii) an extremely high disproportionation constant for the radical anion ir~termediate,~ (iv) a very simple ESR hyperfine splitting due to a nodal plane through the tetrazine 3, 6- positions in the singly occupied MO (2),536 (v) large spin densities on the coordinating tetrazine nitrogen atoms which cause relatively large metal (isotope) coupling of the two equivalent coordinated metal fragment^.^, Furthermore, the structure and electronic spectrum of one complex, a copper(1) dimer, have been determined and interpreted.8 Reported examples include complexes of bptz'- with [Ru (bpy),12+ (bpy = 2,2'- bipyridine),6 [(R,P),CU]+,~ (cod) Cu+ (cod = 1 ,5-cyclo-octadiene),s (OC),M (M = Cr, Mo, W),5 and (OC),XRe (X = Cl, Br).' We now describe the synthesis and ESR characteristics (fig.1-3, table 1) of {(bptz) [O~(bpy),],)'~+, {(bptz) [Rh(nbd)],}*+ (nbd = norbornadiene), and of {(bptz) [Au(PPh,)],)'+. 40474048 Binuclear Radical Complexes 31 0 320 330 340 350 mT Fig. 1. ESR spectrum of ((bptz)[O~(bpy),],}'~+ in frozen acetone at 3.8 K. x marks an impurity. Fig. 2. (a) ESR spectrum of ((bptz)[Rh(nbd)],)'+ in dichloromethane at 300 K; (b) computer simulation with the data from table 1. Experiment a1 ESR spectra were recorded in the X-band on Varian E9 and 109 spectrometers. g Factors and coupling constants were determined either by use of the double-cavity technique with perylene radical anion as reference59 or by directly measuring magnetic field and microwave frequency (Varian NMR gaussmeter, EIP 548A frequency counter).Liquid- helium temperatures were generated using an Oxford Instruments ESR 900 cryostat. The synthesis of the bptz ligand has been described previo~sly.~,~ The red binuclear osmium radical complex was obtained by reduction of {(bptz) [Os(bpy),],} (PF,),'O with tetrabutylammonium tetrahydridoborate in acetone solution ; however, an ESR spectrum was observed only at very low temperatures. The binuclear rhodium radical complex was generated by adding 1 mg of bptz and 10 mg of [Rh(nbd)Cl], in 3 cm3 dichloromethane, followed by reduction with zinc powder. The paramagnetic gold dimer was obtained in the following way: triphenyl- phosphanegold chloride (50 mg, 0.1 mmol) was dissolved in 30 cm3 methanol andTable 1.ESR dataa of binuclear complexes with the bptz anion radical and metal isotope characteristics metal metal fragment a N a N c AaN isotope A,,,(M)d a M lo3 aM/Ai.so g AMY ref. Cr"(CO), Mo"(CO), +Cu'(cod) +Cu'(PPh,), +Cu'(diphos) +Au1(PPh3) +Rhr(nbd) 2+Ru11(bpy), 2+0s11(bpy)2 Re'(CO),Cl W0(C0), - 0.459 0.420 0.415 0.415 0.463 0.442 0.37j 0.33k 0.5S g h 0.495 0.622 0.640 0.640 0.683 0.605 0.624 0.74' 0.66k 0.55f 8 h 0.495 0.163 0.220 0.225 0.267 0.142 0.182 0.37 0.33 f g h 0 - 22.5 - 47.1 53.6 183.0 183.0 183.0 31.3 - 28.6 - 39.6 118.3 320.0 0.1 1 0.230 0.41 5 0.960 0.758 0.936 <0.1 < 0.07 0.45 g 2.2 4.88 4.88 7.74 5.25 4.14 5.1 1 < 3.2 < 2.5 11.36 g 6.87 2.0033 2.0045 2.0068 2.0053 2.0055 2.0058 2.0073 2.0062 1.9980 1.986Y 2.005 1 2.0040 223 5 552 5 2089 5 828 9 828 9 828 9 5091 this work 1212 this work 990 6 ca.2500 this work 2200 7 5 - __ a Coupling constants in mT. Estimated error & 0.005 mT, except when noted otherwise. nitrogen centres. determined, possible error g , = 1.9609 and g,, = 2.0366. Non-coordinating nitrogen centres. Coordinating tetrazine Not accurately Isotropic value calculated from Isotropic hyperfine coupling constant of metal isotope M [from ref. (l)]. Spin-orbit coupling constant (in em-'). 0.15 mT. No hyperfine coupling detected in low-temperature spectrum. A Not resolved. Estimated error _+ 0.1 mT.Estimated error & 0.05 mT. S4050 Binuclear Radical Complexes Fig. 3. (a) ESR spectrum of ((bptz) [Au(PPh,)],)'+ in dichloromethane at 300 K; (b) computer simulation with the nitrogen coupling constants from table 1, a("P) = 0.37 mT (2P) and 0.38 mT linewid t h. treated with 20 mg (0.1 mmol) of silver tetrafluoroborate. After removal of precipitated AgCl, the filtrate was evaporated to dryness and redissolved in 50 cm3 dichlo- romethane/methanol (5 : 1) together with 10 mg (0.04 mmol) bptz. This mixture was reduced for 16 h with zinc powder and the paramagnetic solution used for ESR measurement. Result* The red paramagnetic osmium dimer was obtained by reduction of the blue-green diamagnetic tetracation'' with BH, in acetone. In contrast to the ruthenium analogue6 it does not exhibit an ESR signal at 300 K but only after cooling to very low temperatures (fig.1). This effect has previously been noted for complexes of [Ru(bpy),I2+ with the extremely z-accepting azodicarboxylic esters (adc),l' and it signifies very strong metal-ligand interactions and significant contributions from metal d orbitals to the singly occupied MO. The similar behaviour of the Ru/adc and Os/bptz systems is compatible with a more destabilized +II oxidation state for the heavier meta1.12 The rhombic signal found in a glassy matrix (fig. 1) shows a larger g anisotropy than that measured for the ruthenium analogue; this is a consequence of the higher spin-orbit coupling constant of osmium (table l).', l3 The dirhodium radical was obtained by treating bptz and [Rh(nbd)Cl], with reducing zinc metal in dichloromethane.A well resolved ESR spectrum is obtained (fig. 2); the increased g value and the ratio of the two 14N coupling constants of the tetrazine nitrogen centres are indicative of considerable perturbation by metal c~ordination.~? A lo3Rh splitting ( I = i, 100 %) was not observed and must be smaller than 0.07 mT, the isotropic hyperfine coupling constant Aiso of this isotope is rather small indeed (table 1)' Incidentally, a poorly resolved ESR signal of a mononuclear radical complex [(bpy) Rh(nbd)]' had been reported previo~s1y.l~ The paramagnetic gold dimer, the first radical complex of gold after a few silver(1) semiquinone species were reported,15 was obtained by treating bptz and ClAuPPh, first with AgBF, to remove the halide and then with zinc in dichloromethane.Tri- phenylphosphinegold cation is a rather inert and less reducible electrophile16 which can coordinate in a semi-chelating fashion to a-di-imines such as bpy.l' The relatively largeS. Kohlmann, V. Kasack, E. Roth and W. Kaim 6 4 1 '-1 C u ( 1 ) ARh(I) Mo(0) * 405 1 2.0020 2.0040 2.0060 2.0080 g Fig. 4. Correlation between g factors of binuclear bptz radical anion complexes and the spin-orbit coupling Constants < of coordinating metal centres. isotropic g indicates the presence of coordinated heavy-metal centres with large spin-orbit coupling constants (table l).l, 2 , 5, l8 Assuming reasonable magnitudes of the two nitrogen coupling constants from the tetrazine it is just a triplet from two 31P nuclei (Z = i, 100 YO) that is required to achieve a satisfactory computer-simulated spectrum (fig.3). Whereas the coupling a(,lP) is large for the PPh, co-ligands in the related bis(phosphane)copper(r) complex with tetrahedrally disposed metal,8, the small phosphorus coupling observed here is more compatible with the expected coplanar arrangement n ligand/Au/P.17 Similarly small coupling constants a("P) have been reported for phosphorus centres in the MO nodal planes of a phosphahetero~ycle~~ and of a P,-bridged triple-decker complex.2o The lg5Au metal coupling ( I = g, 100 %) could not be detected and the Aiso value of this isotope is similarly small, of the order of that of lo3Rh (table 1).l Discussion The data in table 1 illustrate that the sum of the tetrazine N splittings remains fairly constant with a,+a,, = 1.05 kO.08 mT; at least 90% of the total spin density is concentrated on these four centre^.^ The difference between the two coupling constants for coordinated and non-coordinated tetrazine centres varies with the extent of perturbation by the metal and shows a maximum at intermediate Apparently, this maximal difference occurs for the rhodium and gold complexes described here.Stronger perturbation, e.g. by the dipositive ruthenium centres, results in less disparate tetrazine N coupling constants. The free-anion radical ligand displays only one coupling constant for all four tetrazine nitrogen centres, which indicates free rotation of the pyridyl group; the bond order between the rings is not sufficiently increased upon reduction because of the n MO nodal plane through centres 3 and 6 of the tetra~ine.~ Metal coupling constants are dependent on the nature of the co-ligands as is illustrated by the copper(1) system^;^ nevertheless, the ratio a/Aiso does show some significant variation from very small values for the presumably planar rhodium(1) and gold(1) centres to the large a/Aiso for the ruthenium(I1) complex with its rather strong metal-ligand interaction.6 Osmium isotope coupling (lS9Os: Z = 4, 16.1 YO natural abundance) could not be observed in the low-temperature spectrum (fig.1); an expected a(lg90s) of ca. 1 mT should not be easily seen in view of the linewidth and of the low probability of 2.6 % for the combination with two lS9Os isotopes (the combination with one active isotope has 27 YO probability).2, l84052 Binuclear Radical Complexes The isotropic g factors of most of the radical complexes in table 1 show a correlation with the spin-orbit coupling factors of the coordinated metal centres (fig.4). This is expected2* l8 in the absence of interfering co-ligands, e.g. with unusually low-lying relevant molecular 21 The deviation of the rhenium chloride system7 (fig. 4) and the strong deviations of bis(bipyridine)-ruthenium and -osmium complexes are in fact due to low-lying unoccupied n* (bpy) or o* (Re-Cl) orbitals, and the g factors of these systems are unusually small, in agreement with theory.2* 6 * 21 Results from ESR spectroscopy on the molecular orbital situation in complexes can be useful in understanding electro- and photo-chemical reactivity.’? 22, 23 The unique properties of the bptz radical anion ligand should thus be further explored with other metal fragments in order to gain an understanding of the metal-ligand interaction. This work was generously supported by Deutsche Forschungsgemeinschaft and by Stiftung Volkswagenwerk.We thank Dr J. Jordanov and the C.E.N.G. (Grenoble, France) for the opportunity to perform low-temperature ESR studies. References 1 B. A. Goodman and J. B. Raynor, A&. Inorg. Chem. Radiochem., 1970, 13, 135. 2 W. Kaim, Coord. Chem. Rev., 1987, 76, 187. 3 A. von Zelewsky, C. Daul and C. W. Schlapfer in Landolt-Bornstein, New Series (Springer-Verlag, 4 D. R. Eaton, J. M. Watkins and R. J. Buist, J. Am. Chem.SOC., 1985, 107, 5604. 5 W. Kaim and S . Kohlmann, Inorg. Chem., 1987, 26, 68. 6 W. Kaim, S. Ernst, S. Kohlmann and P. Welkerling, Chem. Phys. Lett., 1985, 118, 431. 7 W. Kaim and S . Kohlmann, Chem. Phys. k t t . , 1987, 139, 365. 8 D. Fenske, W. Kaim and S. Kohlmann, manuscript in preparation. See also S . Kohlmann, Ph.D. 9 W. Kaim and S . Kohlmann, Inorg. Chem., 1987, 26, 1469. Berlin, 1986), vol. II/17a, p. 199. Thesis (Universitat Frankfurt, 1988). 10 V. Kasack, Ph.D. Thesis (Universitat Stuttgart, 1989). 11 W. Kaim, V. Kasack, H. Binder, E. Roth and J. Jordanov, Angew Chem., 1988, 100, 1229; Angew. 12 P. A. Lay, R. H. Magnuson and H. Taube, Inorg. Chem., 1988, 27, 2364; 2848. 13 D. E. Morris, K. W. Hanck and M. K. DeArmond, Inorg. Chem., 1985, 24, 977. 14 W. A. Fordyce, K. H. Pool and G. A. Crosby, Inorg. Chem., 1982, 21, 1027. 15 W. A. Muraev, W. K. Cherkasov, G. A. Abakumov and G. A. Razuvaev, Dokl. Akad. Nauk SSSR, 16 Cf. W. Beck and N. Kottmair, Chem. Ber., 1976, 109, 970. 17 W. Clegg, Acta Crystallogr., Sect. B, 1976, 32, 2712. 18 W. Kaim, Inorg. Chem., 1984, 23, 3365. 19 W. Kaim, P. Hanel and H. Bock, Z . Naturforsch., Teil b, 1982, 37, 1382. 20 0. J. Scherer, H. Swarowsky, G. Wolmershauser, W. Kaim and S. Kohlmann, Angew. Chern., 1987,99, 1178; Angew Chem. Int. Ed. Engl., 1987, 26, 1153. 21 (a) S . Ernst, V. Kasack, C. Bessenbacher and W. Kaim, Z . Naturforsch., Teil b, 1987, 42, 425; (b) S. Ernst, P. Hanel, J. Jordanov, W. Kaim, V. Kasack and E. Roth, J. Am. Chem. Soc., 1987, 111, 1733. 22 R. Gross and W. Kaim, Inorg. Chem., 1986, 25, 498. 23 S. Ernst, S. Kohlmann and W. Kaim, J. Organomet. Chem., 1988, 354, 177. Chem. Int. Ed. Engl., 1988, 27, 1174. 1977, 236, 620. Paper 9/0 1602C ; Received 17th April, 1989

ISSN:0300-9599    

DOI:10.1039/F19898504047

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

J . Chern. Soc., Faraday Trans. I , 1989, 85(12), 40534062 Quantitative Analysis of the Electron Paramagnetic Resonance Spectrum of a Uranium(II1) Compound Edgar J. SouliC*T and Pierre C. Lesieur CEA-IRDI-DESICP-DLPC-SCM, CEN Saclay, F91191 Gif sur Yvette, France The fundamental formula of the EPR lineshape of field-swept spectra has been modified for the case of a broad homogeneous line. As a consequence, a new formula has been obtained for the derivative of the absorption. For the case of an inhomogeneous line, the original formula of Pilbrow holds. these formulae lay the basis for the determination of the number of spins by integration of the experimental EPR spectrum. The Pilbrow lineshape theory is then combined with the standard results for a Kramers ion having an effective spin 1/2, an anisotropic g tensor and no hyperfine interaction.The assumption is made that the linewidth in frequency units 0” is independent of the orientation. This assumption completes the model specifying the line position, intensity, shape and width for any given orientation. The model thus enables one to simulate a powder EPR spectrum and its derivative knowing the principal values of the g tensor and the linewidth 0”. The simulation entails a double integration over the spherical coordinates of the magnetic field in the frame of the principal axis of the g tensor. The number of sampling points for numerical integration is determined as a function of g,, gV, gz and 0”. Experimentally the EPR spectrum of a coordination compound of trivalent uranium in powder form, U,(BH,),(dcc),, in which there are two kinds of uranium sites, has been recorded at 10.5 K in X-band.The least-squares fit between the observed and simulated spectra has been achieved using the optimisation software BSOLVE, and excellent agreement was reached with a Laplace-Gauss lineshape. The adjusted parameters for the two uraniums were confirmed by an independent fit based on the least absolute values criterion. Based on the final parameters, the EPR spectrum predicted at higher frequencies is better resolved, whereas the actual resolution of the spectrum at 16 GHz is no better. Finally, the effective magnetic moment is predicted to be slightly above 2. Uranium is encountered in insulating compounds with a valence ranging from 3 to 6 . However, it presents an EPR signal only when it possesses an odd number off electrons, that is with valence 5’ or 3.2,3 The EPR spectra (derivative of the absorption signal) of uranium(II1) obtained previously with powders and frozen solutions at very low temperature display a very large and asymmetrical line owing to the large anisotropy of the g tensor.The degeneracy of the P configuration, relevant to the uranium(m) and neodymium(m) free ions, is lifted by the repulsion between the electrons which give rise to the spectroscopic terms and by spin4rbit coupling which decompose the terms into multiplets. From Hund’s rule, the fundamental multiplet of the P configuration is 419,2.4 In the solid state this multiplet is decomposed into five doublets by the crystal field of the surrounding ligands and EPR is usually concerned with the lowest doublet from this multiplet : uranium(II1) has an effective spin of S = 1 /2.In the present work, we analyse a powder spectrum in the simple case of an anisotropic Zeeman interaction without hyperfine structure. While the classical results t Also at URA 331 du CNRS. 40534054 EPR of a Uraniurn(rI1) Compound about the shape and width of the EPR signal deal with the ideal experience (at constant magnetic field and variable frequency) a theory due to Pilbrow5 states how to simulate a real powder spectrum (recorded at a fixed frequency and variable magnetic field). However, the fundamental formula given by Pilbrow has to be modified due to a small error. We give here the modified formulae for the lineshape of the real experiment when the lineshape of the ideal experiment is either a Cauchy-Lorentz or a Laplace-Gauss function.In the compound under study, uranium(rr1) occupies two sites. Each site has a corresponding powder spectrum and the observed spectrum is thus the sum of the contributions of the two sites. Least-squares fitting of the principal components of the g tensor and of the linewidth for each of the two sites leads to excellent agreement between the observed and calculated spectra. Experiment a1 The compound U,(BH,), (dicyclohexyl- 18-crown-6), was synthesized by Anne Dejean de la Biitie.' In a quartz tube sealed under vacuum the compound is stable and the same EPR signal is observed after several years. The actual spectra were recorded with an X- band (9.45 GHz) Bruker spectrometer equipped with a TE-102 rectangular cavity and an Oxford Instruments cryostat cooled by a helium flow.The temperature was 10.5 K. The EPR signal was digitized by an analogue to-digital converter card embedded into an Apple I1 microcomputer. The 1024 data sampled were transferred to an IMB-PC microcomputer and eventually to a CRAY-XMP/28 computer for comparison with the simulated spectrum. Theoretical Before discussing the specific case of a powder, we shall consider the simpler problem of the lineshape of a single crystal. Aasa and Vanngard' noticed that the lineshape of the usual (at fixed frequency and varying magnetic field) and the ideal (at fixed magnetic field and varying frequency) experiments differ not only in their linewidths but also by an intensity factor l/g.This l/g factor that they introduced led to a much better agreement of the spectra, but their analysis of the relation between the field- and frequency-dependent lineshapes and linewidths remained obscure. Important progress was made by Pilbrow5 who developed a general theory including the two types of experiments. In this theory, Pilbrow uses as a starting point the expression for the absorption of the microwave radiation detected by EPR due to the transition between two levels i and j . In eqn (1) Cis a constant which depends on the detection system and on the microwave bridge, q is the filling factor of the sample and Qo is the quality factor of the empty cavity. I &I2 is the resonant transition probability, andfis the lineshape of the constant- field experiment.v,, vii and 0, represent, respectively, the resonance frequency of the cavity, the resonance frequency of the paramagnetic species and the linewidth in frequency units. The factor NhvJZk, T in eqn (l), where N is the number of spins and 2 the partition function, is an approximation of the Boltmann population factor when hv, is small compared to k, T. However, an error has been introduced with this factor since v, should not appear here and must be superseded by the resonance frequency vii of the transition between the two effective spin levels considered. It is important to note here that EPR relies on the superposition of two resonance phenomena: the cavity resonance is obtained by tuning the frequency of the microwave generator and makesE. J .SouliP and P. C. Lesieur 4055 it equal to the frequency of one mode of the cavity. The quality factor of the previous resonance is then modified by the energy absorption due to the resonance of the system of spins. This quantum phenomenon of magnetic resonance occurs at a frequency vii generally near to, but different from the previous frequency v,. The lineshape depends precisely on how the absorption varies as a function of the separation between these two frequencies. Eqn (1) after correction can thus be rewritten as: Since the frequency vij depends on the magnetic field B, the expression of the derivative of the absorption relative to B given by Pilbrow's eqn (4) should be modified as follows : (4) C = CyQ,- Nh zkB T' where In the particular case of a pure Zeeman interaction, that we shall assume from now on, only one resonance frequency vij = v, is considered.This frequency is proportional to the magnetic field and is given by ( 5 ) Vij = V, = g& B/h. The Case of a Homogeneous Line Let us consider here the case of a homogeneous line which corresponds most of the time to a Cauchy-Lorentz lineshapef,[(v, - v,)~, a,] = (a,/n)/[a: + (v, - v,)~]. With the overall assumption that the linewidth av does not depend on the magnetic field B, the use of eqn (2) allows us to obtain the following expression for the derivative of the absorption: where This expression differs from the original one only if the linewidth av is comparable to or larger than the resonance frequency v,, i.e.if the resolution of the spectrum is poor. On the contrary, when the line is narrow, the two frequencies are so close that v, can be replaced by v,. More precisely, the lineshape function (as a function of v,) is different from zero only in a small vicinity of v, having a width equal to several ov. The Case of an Inhomogeneous Line This case is often encountered, for example when an electron magnetic moment interacts through the magnetic dipolar interaction with the spins of many nuclei. One value of the magnetic field corresponds to a distribution of the resonance frequencies v (near the central resonance frequency v,) characterized by a probability density function P(v - v,) normalized by the condition P(v-vv,)dv = 1. (7)4056 Eqn (2) leads here to the expression EPR of a Uraniurn(m) Compound When the width of the homogeneous line is narrow relative to the inhomogeneous broadening, the form function may be taken as equal to a Dirac distribution.In that case, eqn (8) takes the simpler form: We recover eqn (1) given by Pilbrow where the lineshape is now the probability distribution function P , and not the transition probability function f. Thus, only those spins whose frequency coincides with that of the cavity contribute to the EPR signal. To summarize, eqn (2) must be substituted for eqn (1) only when the width of the homogeneous line is so large that it becomes comparable to the resonance frequency. Normalization of an Experimental Spectrum The procedure usually followed to determine the number of paramagnetic ions contained in a sample consists in making a comparison of the integral of the absorption signal of the sample studied with that of a reference sample.To that end the left-hand integral of eqn (10) is performed numerically. The right-hand integral of eqn (10) gives the theoretical value of this integral that we shall discuss in this paragraph. In the case of a homogeneous line and when the lineshape is a function, the integral over the magnetic field of the EPR absorption becomes infinite : vrI fL(v,, B)dB = 00. I Cauchy-Lorentz given by eqn (2) This is due to the presence of the hvr factor which is proportional to the magnetic field. One may argue that for very large fields the approximation hv,/ZkB T of the population has to be replaced by a more general expression, which for a Kramers doublet is equal to tanh(hvr/2kB T).This is equivalent to truncating the integral (1 1) at a field of the order of k, T/gpB. However, this field is so high that, while theoretically true, this remark is of no help in practice (except maybe at high frequencies, i.e. Q-band, and very low temperatures, i.e. 4 K). One may introduce a convergent integral which allows the determination of the number of spins : However, the above divergence is due only to the choice of a Cauchy-Lorentz function as lineshape function. In the case of a homogeneously broadened line of a pure Zeeman interaction one may use the Bloch equations in order to find the response function of the paramagnetic species. If the microwave is linearly polarized, the system of spins possesses two resonance lines for two opposite values of the magnetic field: gpB B = khv,.Furthermore, owing to interference effects, the lineshape function is not the sum of two Cauchy-Lorentz functions centred on these values of the field. TheE. J. Soulie' and P. C. Lesieur 4057 calculation of the linear response of the Bloch equations leads to the following -. - lineshape : where fB is normalized by the condition : rIfB(Vc, Vr)dvc = 1- With this lineshape function, integral (10) is now finite, and is equal to so that, when the line is narrow, the number of spins is given by thanks to which the number of spins may also be evaluated. The divergence in eqn (1 1) for a Cauchy-Lorentz lineshape is only a formal one in that it is due to the model.An other advantage of integral (lo), and thus of eqn (16), is that for an inhomogeneously broadened line (and assuming as previously that the matrix element y , j is constant) integration of the EPR absorption signal again leads to eqn (16). Powder Spectrum from a Uranium Compound An EPR spectrum from a powdered compound, either the absorption or its derivative, is the superposition of the spectra of the numerous crystallites constituting the powder. In order to analyse a powder spectrum one needs a theory indicating how the position, the intensity, the width and eventually the shape of the resonance signal depend on the magnetic field. We examine here these characteristics of the signal for a uranium(n1) compound. Position of the Resonance The 419,2 multiplet is decomposed into levels (Kramers doublets), the energy differences of which are much larger than the microwave quantum.The fundamental level is thus described by a fictitious 1/2 spin. The uranium 238 nucleus has zero spin, and superhyperfine interactions with neighbouring nuclei of the ligands lead only to an inhomogeneous broadening of the line which we shall consider as isotropic.The effective hamiltonian is thus reduced to the unique term describing an anisotropic Zeeman interaction : H = pBB*g*S. (17) Following Van Veeq8 we shall use for each crystallite the frame of the principal axis of the g tensor and define the direction of the static magnetic field by its spherical coordinates 8 and #J. The g factor corresponding to the (8,#J) direction is given by the relationg g2 = (gi cos2 #J + gi sin2 #J) sin2 9 + g," cos2 8.(18) Intensity of the Resonance Line The transition probability I K, J2 between the two sublevels of the fundamental Kramers doublet varies with orientation in proportion to the g: factor given'' by the following (19) expression : g: = (g: g;(ll m - lmJ2 + g; g,"(m, n - mn1)2 + g," gi(n1 I - nll)2)/g2. 135 FAR 14058 EPR of a Uranium(II1) Compound Here, 1, m, n are the director cosines of the static magnetic field H, and I,, m,, 12, are the director cosines of the microwave magnetic field H,. The latter is usually perpendicular to the static magnetic field so that only a third angle x is needed to determine its orientation. The gX factor is usually the only one to depend on x, so that integration upon that angle can be analytically performed.The factor g; is thus replaced by its average [ref. (8), Appendix 31: 3 = kig:(cos2 8 cos2 q5 + sin2 q5) +g: gz(cos2 8 sin2 q5 + cos2 q5) + gf g: sin2 8]/g2. (20) Width and Shape of the Resonance Line Many EPR spectra display a broader linewidth for high fields (equivalent to small g values) than for low fields (or large g values). The corresponding linewidth in frequency units generally displays much smaller variations. The two linewidths are related by the equation : It has previously been observed that for the Nd3+ ion, which is isoelectronic with uranium(w), in the double nitrate of lanthanum and magnesium, the linewidth in frequency units was ~onstant.~ We have thus assumed the width av to be constant.When the paramagnetic ion is surrounded by several ligands, each carrying nuclear spins, and when the inhomogeneous linewidth is larger than the homogeneous linewidth, the lineshape function follows the Laplace-Gauss law.ll The lineshape function introduced in our model for the derivative of the absorption is thus Dejean de la Bitie69l2 deduced from the analysis of NMR spectra in solution of U(BH,), and of two complexes of the type U,(BH,),L, that in U,(BH,),dcc, uranium is present simultaneously within an anionic form U(BH,)i- and a cationic form U(BH,), (dicyclohexyl- 18-crown-6)+, in the proportions 1 /3 and 2/3, respectively. No single crystal of the compound U,(BH,),dcc, was obtained, although it is moderately soluble in CH,Cl,,' and the crystalline structure is unknown.However, an EXAFS study at the L(m) edge of uranium of the related compound U3(BH,), (18-cr0wn-6)~'~~ l4 has shown that uranium is present in the two forms U(BH,):- and U(BH,), (18-crown-6)+. Furthermore, the cation U( BH,), (dicyclohexyl- 18-crown-6)+ appears in a binuclear compound of uranium(r1r) and uranium(1v) : [U'I'(BH,), (dicyclohexyl- 18-crown- 6)],UTVCl5(BH,) the structure of which has been determined by X-ray crystallography. l4 Dejean's structural hypothesis has thus been confirmed and the EPR spectrum must be the sum of the spectra of the anionic and cationic species, with weights 1/3 and 2/3. Simulation of Spectra and Adjustment of the Parameters For each value of the magnetic field, the derivative of the absorption is calculated as a double integral over angles 8 and q5.The analytical integration over one of the two angles 0 or q5 using Risch's algorithm15 did not succeed owing to insufficient memory in the computer used.'' Since a double integration has to be performed for each value of the magnetic field, we have looked for a good compromise between speed and precision of the calculations. The sphere has been decomposed into zones (regularly spaced along angle O), and the alternative integration variable cos el7 has not been introduced. In order to split the surface of the sphere into parts of nearly equal areas, the number of points within each zone (angle q5) has been taken roughly proportional to the area of the zone. A better approximation scheme would be to sample smaller areas in those directions where the integrand varies significantly, and larger areas in those directions where it varies smoothly; we have not attempted to implement an integration algorithmE.J . Soulie' and P. C. Lesieur 4059 along these lines. When the number of sampled points is too low, the calculated spectrum presents some oscillations increasing in amplitude as the number of points is lowered. Furthermore, the smaller the linewidth ov, the larger this number must be. More precisely, to the linewidth in frequency units corresponds an angular width: The number of points sampled for angle 8 has been chosen inversely proportional to the minimum of the angular width, relative to both 8 and q5: The factor K was arbitrarily taken to be equal to 3, i.e. at least 3 values of 8 (at constant angle qb) lie within the half width at half maximum of the line.A similar choice was made for the sampling of q5 values. The physical model described above and the algorithm allow the simulation of a spectrum for each set of parameters, namely the principal components of the g tensor and the inhomogeneous linewidth 0" for each site plus one scale factor. The usual least- squares minimization was performed, and the nine parameters were adjusted with the help of the optimization program BSOLVE'~ in order to best-fit the calculated and experimental spectra. An approximate fit of the experimental spectrum by a simulation including only one site and an inhomogeneously broadened Laplace-Gauss lineshape is presented on fig. 1. The value of the statistic estimator x2 = 31.4 when compared to that of a simulation performed with an homogeneously broadened Cauchy-Lorentz lineshape k2 = 542) clearly shows that broadening is inhomogeneous, while relaxation prevents the spectrum being recorded at high temperatures.When the simulation is performed including two sites, excellent agreement is obtained, as shown in fig. 2. Table 1 shows a low value of the statistic estimator x2 = 8.1. This confirms the hypothesis of a Laplace-Gauss lineshape and of a constant linewidth in frequency units.5 Fairly close values of the principal g values and linewidths were obtained as a result of another fitting in which the sum of the absolute values of the parameters was minirni~ed'~ (table 1). The model presented above allows the prediction of the spectrum at other frequencies : at higher frequencies the resolution should be better.However, the resolution of the spectrum obtained at 16 GHz is in fact no better than that of fig. 1 ; the linewidth thus seems to increase with microwave frequency. The model also allows the calculation of the molar susceptibility of this uranium(rrr) compound at temperatures such that the fundamental Kramers doublet is the only populated For a single site the susceptibility is expressed by The effective magnetic moment is thus4060 EPR of a Uranium(rr1) Compound Fig. 1. Observed ( x ) and calculated (-) spectra for the powder compound [U(BH,),], (dicyclohexyl-l8-crown-6), at a microwave frequency of 9.457 GHz. The temperature is 10 K. Calculations are performed including one site, a Laplace-Gauss inhomogeneous lineshape and a least-squares optimization. The components of the g tensor and the linewidth are given in Table 1.When two sites are present with proportions p1 and 1 -pl, the effective magnetic moment can be written as With the parameters given in table 1, one finds peff = 2.019. Conclusion We have specified in detail the validity of the original equation given by Pilbrow [eqn (l)] and replaced it by a closely related equation in the case of a broad homogeneous line. The theory developed by Pilbrow allowed the first quantitative interpretation of a powder compound of trivalent uranium. The principal components of the g tensor have been determined for the two sites as well as the linewidths. The simulation of the spectrum at a higher frequency and its comparison with the experimental spectrum obtained at 16 GHz seems to indicate that the linewidth increases with microwave frequency.E. J.Soulie' and P. C . Lesieur 406 1 2000 h Y .- C 1000 ta v 0 -1000 0 4000 *Oo0 HIG 6000 Fig. 2. Same as fig. 1, but with two sites. Table 1. Principal components of the g tensor, linewidth ov (in GHz) and estimator x2 for different adjustments of the parameters initial values 2.77 2.37 1.90 1.5 adjusted valuesa 2.837 2.303 1.756 1.8 31.4 adjusted valuesb anionic site 2.820 2.097 1.617 2.0 8.1 cationic site 2.842 2.349 1.839 1.3 anionic site 2.769 2.106 1.602 2.2 cationic site 2.847 2.351 1.832 1.4 adjusted valuesc Least-squares fittings with, respectively, 1 and 2 sites performed with BSOLVE optimizer ; obtained by minimization of the sum of the absolute values with the help of DYNEPS optimizer .We thank James Hodges for the 16 GHz spectra, Claude Chachaty for fruitful discussions, Maurice Goldman for reading the manuscript and James Davenport for attempting an analytical integration on the double integral of the problem.4062 EPR of a Uranium(m) Compound References 1 L. A. Boatner and M. M. Abraham, Rep. Prog. Phys., 1978, 41, 87. 2 E. J. Soulii, G. Folcher and H. Marquet-Ellis, Can. J. Chem., 1982, 60, 1751. 3 E. J. Soulie, G. Folcher and B,Kanellakopulos, Can. J. Chem., 1980, 58, 2377. 4 A. Abragam and B. Bleaney, Electronique des Ions de Transition (Presses Universitaires de France, 5 J. R. Pilbrow, J. Magn. Reson., 1984, 58, 186. 6 A. Dejean-Meyer, G. Folcher and H. Marquet-Ellis, J. Chim. Phys., 1983, 80, 579. 7 R. Aasa and T. Vanngard, J . Magn. Reson., 1975, 19, 308. 8 G. Van Veen, J. Magn. Reson. 1978, 30, 91. 9 Ref. (4), eqn (3.5). 10 J. R. Pilbrow, Mol. Phys., 1969, 16, 307; ref. (4), eqn (3.10b). 11 D. W. Marquardt, R. G. Bennett, E. J. Burrell, J . Mol. Spectrosc., 1961, 7 , 269. 12 A. Dejean de la Bltie, Thesis (Universite de Paris Sud, 1986). 13 A. Dejean, P. Charpin, G. Folcher, P. Rigny, A. Navaza and G. Tsoucaris, J. Phys. (Paris) Colloq. C8 14 A. Dejean, P. Charpin, G. Folcher, P. Rigny, A. Navaza and G. Tsoucaris, Polyhedron, 1987, 6, 189. 15 R. H. Risch, Bull. Am. Math. SOC., 1967, 76, 605. 16 J. Davenport, personal communication. 17 P. C. Taylor and P. J. Bray, J . Magn. Reson., 1970, 2, 305. 18 E. Soulii, communication at Cinquiimes Journies Franco- Allemandes d’optimisation, Varetz, 4-8 October, 1988. 19 W. E. Ball, BSOLVE, in Optimization Techniques with Fortran, ed. J. L. Kuester and J. H. Mize (McGraw-Hill, New-York, 1973), pp. 240-250; W. E. Ball, in Material and Energy Balance Computations, ed. E. J. Henley and E. M. Rosen (John Wiley, Chichester, 1969), Appendix E, pp. 560-566. 20 C. Lemarechal, communication WP-80-36, International Inst. for the Analysis of Applied Systems, A- 236 I , Laxenburg, Austria. 21 E. Soulik, Thesis (CEA-4849, 1977); E. Soulii, Inst. Phys. Conf. Ser., 1978, 37, 166. Paris, 1972), Chap. 3. 1986, 47, 645. Paper 9/01 598A ; Received 17th April, 1989

ISSN:0300-9599    

DOI:10.1039/F19898504053

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. I , 1989, 85(12), 40634074 Electron Addition to Xanthine Oxidase An Electron Spin Resonance Study of the Effects of Ionizing Radiation Martyn C. R. Symons* and Fatai A. Taiwo Department of Chemistry, The University, Leicester, LEI 7RH Richard L. Petersen Department of Chemistry, Memphis State University, Memphis, Tennessee 38152, U. S. A . Exposure of aqueous xanthine oxidase to 6oCo y-rays at 77 K resulted in electron addition to an iron-sulphur centre at low doses. At higher doses, addition to the Mo”’ unit was also observed, using ESR detection. The results are interpreted in terms of rapid electron transfer from MoV to the nearest Fe/S cluster, with permanent trapping at molybdenum only when a second electron is generated in a given molecule. The ESR spectrum for the primary Mo” centre closely resembles that for a centre previously only detected in the presence of a substrate molecule, such as xanthine, and known as the ‘very rapid’ centre.Hence we conclude that there is no major covalent bonding between Mo“ and the substrate in this complex. On annealing, the first change involved conversion of the primary Mo” centre to a secondary Mo” centre, exhibiting a 13 G proton hyperfine coupling. The ESR parameters closely resemble those for a centre previously described as the ‘rapid’ centre. This is the first detectable species in rapid- freeze experiments in the absence of specific substrates. When D,O was used, the proton coupling was lost. The ESR parameters are compatible with the postulate that the initial species has a sulphide ligand (Mo-S)- and the second species is protonated on sulphur (Mo-SH).Further annealing results in irreversible loss of the MoV signal and concomitant growth of a second Fe/S cluster signal. These results are discussed in terms of the remarkable selectivity of electrons ejected within the protein by y-rays, and the very different rates of electron transfers between the different centres. There was also evidence for electron capture at an RS-SR unit giving, initially, an RS-SR- radical anion, followed by an irreversible conversion into a characteristic centre, probably formed from the anion by protonation. This centre is of interest since it is also formed from solvated electrons in pulse-radiolysis studies. Introduction Xanthine Oxidase Xanthine oxidase is a member of a group of enzymes known as molybdenum iron-sulphur flavin hydroxylases.It catalyses the hydroxylation of purines, and in particular, xanthine to uric acid.’ This reaction is thought to occur in a pocket at the molybdenum site, with concomitant formation of MoIV. This is accompanied by electron transfer to the iron-sulphur centres and to an FAD unit. The latter completes the catalytic cycle by reducing dioxygen. OH OH xanthine 4063 uric acid4064 Electron Addition to Xanthine Oxidase The enzyme used in the present study, bovine-milk xanthine oxidase, is a dimeric unit, comprising two identical monomers containing one molybdenum centre, one flavin, and two different Fe,/S, clusters.' In its resting form the enzyme is fully oxidized (MoVT, 4Fer1', FAD), and has a capacity for accepting at least six extra elections. The crystal structure is not yet known, but EXAFS studies have revealed some details of the MeV' coordination2y3 and ESR studies of various reduced forms have been interpreted in terms of magnetic coupling between unpaired electrons of different centres to give some measure of their ~eparation.~-~ It was originally suggested that electron transfer followed the sequence xanthine + molybdenum + Fe/S + FAD + 02,' but later studies8y have indicated a non-linear scheme in which transfer from Mo directly to FAD is possible (scheme 1).Fe/S Scheme 1. Mode of electron transfer from xanthine to oxygen Rates of electron movements through the enzyme are unknown, but it has been clearly established that they are fast relative to the rate of enzyme turnover.8 Effects of Ionizing Radiation on Proteins Despite the reasonable expectation that the large energies available in y-rays should be able to generate a wide variety of chemical damage to proteins, recent studies seem to suggest that damage is generally remarkably selective.lo Ejected electrons are able to avoid reaction with amide groups by rapid migration, and tend to add to sites of high electron affinity. For example, addition seems to be largely confined to the Fe-0, unit in oxy-haemoglobin,1'?12 and to S-S bonds in proteins containing one or more such units. l3 In marked contrast, electron-loss centres do not seem to be mobile, and are thought to be rapidly trapped at amide sites by loss of the N-H proton.1° This remarkably simple mechanism can be exploited in the study of intra-protein electron transfer in a manner14 which is perhaps more direct than other procedures so far designed.l5-I8 One aim of the present study was to establish relative rates of electron transfer between various centres, and to explore the possibility of kinetic studies of intra-electron transfers.ESR Studies of Xanthine Oxidase Extensive ESR studies using a rapid mixing-rapid freezing technique developed and thoroughly exploited by Bray and co-workers have been recently reviewed. l9 The results of these studies, which are of vital importance in the present work, are outlined in the Discussion section. Preliminary studies of our work were reported some time ago,,' but these have been almost totally overlooked in subsequent reports.Therefore, another of our aims has been to confirm and extend those studies, and to stress their relevance to the identification of paramagnetic molybdenum centres. Experiment a1 Xanthine oxidase isolated from fresh bovine milk was kindly provided by Dr Russ Hille (Ohio State University). The enzyme was supplied in 100 mmol dm-3 pyrophosphate,M . C. R. Symons, F, A . Taiwo and R. L. Petersen 4065 pH 8.5. All experiments were performed with the stock solution of the enzyme without further purification. The enzyme was deuterated by three cycles of concentration by vacuum dialysis and dilution with D,O at 277 K. y-Irradiation of frozen aqueous solutions of xanthine oxidase was performed at 77 K in a 6oCo Vickrad y-ray source, at a dose rate of ca.0.6 Mrad h-l for up to 14 Mrad. X-Band ESR spectra were measured at 77 K or ca. 4.2 K on a Varian El09 spectrometer using 100 kHz modulation. All g values were calculated using a Bruker B- H12E field calibrant. Well separated ESR features for the MoV signal (8, = 1.956) and the [Fe/S I]- (gz = 1.925) recorded with non-saturating microwave powers were monitored as a function of radiation dose. These selected signals were shown to have a constant lineshape and linewidth at 77 K, allowing peak intensities to be normalized relative to ferric acetylacetonate and DPPH standards and utilized as a measure of relative concentrations. At the end of the dose-dependence measurements the samples were annealed while continuously monitoring the ESR spectrum for changes. As soon as any significant change was observed, the sample was quickly recooled to 77 or 4.2 K and the ESR spectrum recorded again.Results and Discussion ESR Signals following Low-temperature Radiolysis Xanthine oxidase has no ESR spectrum in its resting state because molybdenum is in the 6+ oxidation state (do), the FeIII ions in both two-iron-two-sulphur centres are magnetically coupled, and the oxidized flavin unit, FAD, has no unpaired electrons. On exposing the xanthine oxidase solutions to ionizing radiation, electrons are ejected statistically from protein components. These liberated electrons may migrate over considerable distances even at 4.2 K before they are selectively trapped within the protein.10 The ESR spectra of samples following 6oCo y-irradiation at 77 K is dominated by signals from the 'OH radical formed in ice crystals, and organic radicals formed in the protein during irradiation.Unfortunately, these intense signals near free spin obscure any ESR evidence for production of FAD- (or FADH) by electron trapping at the flavin site. Components of the ESR signals for [Fe/S I]- and MoV centres extending to the high-field side of the central line region are clearly observed at 77 or 4.2 K with g , and g , values in close agreement to those previously obtained.20 At low doses, only an (Fe/S)- signal was observed [fig. 1 (a)] but with increasing dose, narrower features which we assign to a MoV centre grew in [fig. l(b)], and the rate of growth of the Fe/S signal fell.On annealing to ca. 130 K, features of the *OH radicals trapped in ice crystallites were lost irreversibly. (These radicals combine within their host crystals to give H,O,, and are not involved in any protein reactions.) In the range 130-160 K the MoV features changed, quantitatively," to a new set of features also assigned to a MoV centre (fig. 2). A characteristic 13 G hyperfine coupling for this centre was lost when H,O was replaced by D,O, showing that this splitting is from a strongly coupled exchangeable proton. On further annealing, the MoV signals were lost irreversibly, and a set of features assignable to a second iron-sulphur unit were detected at ca. 4 K. Sulphur Centres The form of the central signal after annealing to remove 'OH radicals suggests the presence of RS--SR- radicals as well as other protein centres. The shoulder at g x 2.02 was lost on further annealing, and weak, but well defined doublet features grew in at g x 2.06 and g z 2.025 (fig.3), which are characteristic of a radical derived from (RSSR)'- centres on protonation.21* 224066 Electron Addition to Xanthine Oxidase Fig. 1. First-derivative X-band ESR spectra for xanthine oxidase after exposure to 6oCo y-rays at 77 K showing (a) g, and g , features for the (Fe/S I)- centre (low dose), (b) gain of the g , and g, features for the (VR) Mo" centre (high dose) and (c) the same sample as in (b) after annealing to convert the (VR) into the (R) centre, showing the 'H doublet splitting. [These spectra were measured using high power in order to obtain a good (Fe/S I)- signal.] Temperature (K): (a) 77, (b) 77, (c) 185.Table 1. ESR parameters for various Mo(V) centres in reduced xanthine oxidase 'H hyperfine g values coupling/G" g1 8 2 g3 g,, 4 4 4 reduc tan t xanthineb e- (prays)" e- (y-rays) xanthine (1) xanthine (2) salicyl (A)d salicyl (B) e- (y-rays)" e- (y-rays) 2.0252 (2.025) (2.025) 1.989 1.994 1.991 1.991 1.990 1.992 (VR)Mo(V) 1.9550 1.9494 1.9765 1.956 1.951 (1.977) 1.955 1.950 (1.977) 1.969 1.964 1.974 1.968 1.961 1.974 1.968 1.963 1.974 1.966 1.963 1.973 1.969 1.965 1.975 1.968 1.964 1.975 (R) M 0 (V) 12.4 12.0 12.0 12.1 12.1 12.1 14.3 14.0 14.0 14.3 14.0 14.0 12.0 12.0 12.0 -13 -13 - a 1 G = text). different forms of the (R) species.] T. Ref. (25). Ref. (20) (features in parenthesis were estimated as described in the R.C . Bray and J. C . Swann, Struct. Bonding, 1972,11, 107. [Types 1 and 2 are two slightlyM. C. R. Symons, F. A . Taiwo and R. L. Petersen 4067 Fig. 2. ESR features for the Mo” (R centre at low microwave power, (a) using H,O and (6) using D,O. Note the loss of doublet splitting for the g,, g, and g, features, and the presence of 9 5 M ~ hyperfine features. The features marked a in (6) are assigned, in part, to (RSSR)’- radicals. [The features for the (Fe/S I)- centres make only a minor contribution at low powers.] ESR parameters for these centres are given in tables 1 and 2, together with pertinent literature data. Identijication of the Iron-Sulphur Centres The three g values for the primary centre are characteristic of an (Fe/S)- unit.They are much closer to those normally observed for the (Fe/S I)- centre obtained on rapid- freezing experirnent~’~*~~ (table 2) rather than those for the (Fe/S 11)- unit, and we accept this identification in the following discussion. We were unable to detect features resembling those for the (Fe/S 11) centre also obtained in rapid-freeze studies, even at 4.2 K and after high exposures (ca. 14 Mrad). However, these features grew in during the final annealing studies, as other signals were lost. In this case our g values are identical with those reported by ~ t h e r s , l ~ * ~ ~ and there can be no doubt that this is the normal (Fe/S 11)- unit.4068 Electron Addition to Xanthine Oxidase Table 2. ESR parameters of reduced Fe/S(I) and Fe/S(II) centres in xanthine oxidase g values reductant g1 g2 g3 gav Fe/S(I) s,o;- a 2.045 1.925 1.890 1.953 e- (prays)* 2.022 1.935 1.899 1.952 e- (prays)" 2.044 1.935 1.896 1.958 Fe/ S(I1) s p - a 2.120 2.007 1.910 2.0 12 e- 2.120 2.007 1.910 2.0 12 e- (y-rays)" 2.120 2.007 1.905 2.010 a Ref.(6). Ref. (20). " This study. 31926 I g2 Fig. 3. ESR features (g, and g2) assigned to a radical derived from an RS-SR centre in xanthine oxidase. The g3 features, expected to occur close to g = 2.003, are obscured by more intense signals. Identification of the Molybdenum Centres The g , and g, parameters are unambiguously derived for the primary MoV centre, but the low-field g , feature is obscured by the intense central lines. We have previously identified this centre as a 'very rapid' (VR) MoV centre obtained in rapid-freeze studies in the presence of xanthine.20 This is an important suggestion, since, if it is correct, it shows that the (VR) centre is formed essentially by electron transfer from the substrate rather than via covalent bonding, as has been suggested.We must therefore consider just how similar the present primary electron capture centre really is to the (VR) centre. First, we stress that this centre is generated after electron addition to the (Fe/S I) unit is largely complete and hence there is some overlap and broadening of the MoV features (see below). In a careful search for a g, feature at 2.025, we have been able to detect a shoulder in this region after carefully controlled removal of the 'OH radical features on annealing.20 However, other features are also present, some conversion of the primary MoV to the secondary MoV species has occurred, and the probable presence of some RSSR'- centres all make this identification difficult.Nevertheless, two pieces of information strongly suggest that g, is in this region. One is that an integration of the signals for the primary and secondary MoV units showsM. C. R. Symons, F. A . Taiwo and R. L. Petersen 4069 complete conversion, provided g, is close to 2.025. The other is that there is a very low- field feature seen at high gain which correlates with the primary centre and comes close to the field region expected for the MI = +4 line for 9 5 M ~ provided A , ( 7 5 M ~ ) x 47.5 G and g , x 2.025 which are the VR parameters.Finally, there is complete conversion of the primary into the secondary MoV centre. This is important since the secondary centre can be monitored more readily, and has enough features for unambiguous identification as one of the closely similar centres described as 'rapid' (R) (table 1). In particular, there is a characteristic proton splitting of ca. 13 G. The only difference from the R spectra previously published is that the features are broader. This centre is the first MoV species detected in simple chemical reduction of xanthine oxidase in the absence of substrates using the rapid-freeze technique. Thus its formation is to be expected on electron addition, and there are no problems in postulating its formation. The key point is that the MoV (R) centre is not formed directly on electron addition to MoV1 : it is clearly a secondary species.However, it is formed from an MoV precursor which has magnetic parameters closely resembling those for the (VR) centre. The simplest interpretation of these results is that electron capture at 4 or 77 K gives an MoV centre having effectively the unrelaxed structure of the parent MeV' centre. The formation of such 'hot' complexes at low temperatures is common experience in radiation chemistry.24 On annealing, the centre relaxes rapidly to give an MoV unit in its preferred geometry. There are two important facets of this relaxation. One is that the high g value feature, which is quite exceptional for a d' complex, shifts into the normal low g value region (from 2.025 to ca.1.99) and the other is the appearance of a proton hyperfine coupling from an exchangeable proton, in the (R) complex. We originally suggested2' that these characteristic changes can be most simply explained in terms of the partial reaction \ / \ / / \ /I -Mo'S- + AH -+ -Mo'SH which involves proton transfer from a neighbouring proton donor, AH, which may well be water. The high g shift is explicable in terms of n-delocalisation onto sulphur, which is considerably reduced on protonation. This postulate now appears to be the accepted explanation for the ESR parameter changes. 25 If this is correct, then an alternative explanation to that previously proposed is required for the fact that, in rapid-freeze experiments, the (VR) centre is only detected when xanthine or similar substrates are u ~ e d .' ~ , ~ ~ We suggest that the presence of the substrate either locks the protein into its original shape, preventing geometrical relaxation, or, by filling a 'pocket' in the protein, prevents the protonation step (2). Our results show that reaction (2) is extremely rapid even at very low temperatures, and, in the absence of such a blocking agent, rapid-freeze techniques are far too slow to prevent this change. Is Xanthine bound to Mo? Two factors, in addition to the special ESR parameters, have led to the suggestion that xanthine is bonded to Mo in the (VR) ~ e n t r e . ' ~ ' ~ ~ One is that when the 12C(,, carbon is replaced by 13C in xanthine, a doublet hyperfine splitting is observed.26 The other is that the proton acquired in the change from (VR) to (R) is the C,-H proton of xanthine.Structures (I), (11) and (111) have been proposed in order to accommodate these observations. The observed 13C splitting, which was poorly defined, is ca. 3 G, and seems to be almost isotropic. If it is indeed isotropic, this corresponds to ca. 0.27% 2s character. This small effect may well simply indicate a small degree of charge transfer in4070 Electron Addition to Xanthine Oxidase S 0 ‘“II ,s (11) Mo’ N- 1 N- ‘ tl ’ (I) ,,‘I \ ,N- I 0-c,- (xanthine) N- the complex (however, see below). Alternatively, the coupling may be larger dipolar, with two of the components having negative signs. It is, of course, unlikely that the detected splittings represent principle values. There seems to be no requirement that there be covalent bonding.In particular, bonding to carbon, as in (11) seems to be unlikely, in view of the magnetic similarity between the present centre and that formed from xanthine. Also, in all these structures, the C,-H hydrogen is missing. It is not clear why it should use a pathway in which it moves from carbon to an unspecified base prior to being transferred to sulphur. If our proposed structure for the (VR) centre is correct, many mechanistic problems remain. Since xanthine and related substrates have low ionization potentials, the first step could be an electron transfer to MeV', giving (xanthine)’++MoV. This could be followed by OH- (H,O) addition at c8 to give (IV), together with movement of the MoV electron to other acceptor units. A second electron transfer could then give MoV, still having the original MoV1 coordination, with the complex (IV)+ in close proximity.This can then transfer the C , proton to sulphur giving the (R) complex. This proposal is intended simply to show that the (VR) complex could be structurally similar to the parent MoV1 centre and still fulfil the experimental requirements. Direct attack on xanthine via the 0x0 ligand thought to be present2T3 is an attractive mechanism which would avoid the OH- addition stage, but an -OR ligand is expected to be quite different from an 0x0 ligand in its influence on the magnetic properties of the MoV complex. It is difficult to see how this type of change could occur in the present studies unless, even at ca. 4 K, protonation at oxygen occurs to give an -OH ligand.This proton could subsequently transfer from oxygen to sulphur to give the (R) centre. Dose Dependence of [Fe/S 11- and Mo” Signal Intensities [Fe/S I]- and MoV g values were shown to be independent of radiation dose. The relative intensities of these two ESR signals, however, were discovered to have a distinct dependence on time of irradiation. The [Fe/S I]- centre was first to appear regardless of whether the irradiation involved y-radiolysis at 77 K or X-radiolysis at 4 K. The plot of signal intensity for the feature at g = 1.925 (i.e. g, for [Fe/S I ] - ) versus dose [fig. 4(a)] shows the electron trapping at this site to be a typical first-order process, resembling the normal hyperbolic saturation curve for ligand binding to identical non-interacting enzyme sites.In contrast, growth in the MoV ESR signal (g = 1.956) exhibits a lag phase. This feature begins to increase in relative intensity compared to the [Fe/S I]- signal at higher radiation doses, and at very high y-doses the MoV signal also begins to saturate [fig. 4(b)]. The sigmoidal dependence of the intensity of the MoV signal on the time of irradiation [fig. 4(b)] is in accord with a sequential reduction pathway wherein trapping a single electron at the [Fe/S I] centre must precede electron capture at MoV1. The consecutive reaction scheme used in our analysis of electron trapping is based onM. C. R. Symons, I;. A . Taiwo and R. L. Petersen 407 1 0 10 20 30 irradiation time/h "0 10 20 30 irradiation time/ h Fig. 4. Trends in the relative intensities of the (Fe/S I)- and Mo" (VR) centres with increasing radiation dose (a) the (Fe/S I)- centre and (b) the MoV centre.The points are experimental intensities and the lines were derived using eqn (1) and (2) in the text. the assumptions that (I) the [Fe/S I] and the MoV1 centres can be treated as a linked unit able to exchange electrons at all temperatures, (2) the [Fe/S I]- represents a 'deeper' trap for capturing an electron than does MeV', (3) the [Fe/S I]- ESR signal grows in with the addition of the first electron and does not change on addition of the second electron to the molybdenum centre, and (4) the entry of electrons into the [Fe/S I] and MeV' system takes place at a constant rate throughout the course of irradiation. The rate for electron addition to the coupled [Fe/S I] and MoVr unit can be described by a pseudo- rate constant, k, and matched to the data for the first-order growth of the [Fe/S I]- ESR signal using eqn (l), where the [Fe/S 11- signal height is represented by A,, A, = A,-A,exp(-kt) (1) A , is the extrapolated signal height when this centre is fully reduced and t is the time of y-irradiation.The constant k for the first electron capture by [Fe/S I] is thought to be equal to the rate constant for trapping the second electron at the MoV1 centre under these assumptions and can be used to calculate the dependence of the MoV ESR signal on radiation dose using the equation Bt = BJ1 -exp(-kt)-ktexp(-kt)] (2) where B, and Bo relate to the corresponding MoV signal intensities.The calculated dose dependence of B, is in good accord with the experimental data as shown in fig. 2. From4072 Electron Addition to Xanthine Oxidase these results, we estimate that at the highest doses used, ca. 80% of the Mo”‘ was converted into MoV. Electron-Electron Coupling between Centres Extensive ESR studies have been conducted in which spin-spin interactions between various paramagnetic centres in xanthine oxidase have been measured. Most of this work has involved saturation studies at high power levels, but in one case what appears to be a substantial splitting of a MoV signal has been reported.26v27 This splitting, observed only at low temperatures (30 K) is apparently almost isotropic (ca. 11 G). The normal spectrum decayed simultaneously, so these new features are not satellites.They can be caused by normal dipolar coupling between spins provided the two electrons are quite close, and with a precise xparation. Coffman and Buettner28 calculate a mean separation of between 8 and 14 A, the uncertainty arising because the principle values of the zero-field splitting are not directly revealed by the spectra. Using controlled electrolysis it has been shown that this splitting of the MoV signal is due to a nearby [Fe/S I]- 29 We see no such splitting for our (VR) MoV centre at ca. 4 K. It is interesting that Bray and co-workers also failed to detect any splitting from their (VR) centre^.^' However, we also failed to detect splitting for the MoV (R) signals, despite the presence of well defined [Fe/S I]- signals.The features are broader than those produced on rapid freezing (in the absence of zero-field splitting), but this is only in the region of 3 G and is far less than the 11 G zero-field splitting reported. It seems possible that electron addition to [Fe/S I] and MoV1 at room temperature causes the protein to change shape such that the Mo and Fe sites get significantly closer. Low temperatures prevent this relaxation. On annealing, electron transfer to give [Fe/S I]- centres occurs prior to any shape change, so we have not been able to detect the expected zero-field splitting. Electron-transfer Rates The present results show that electron transfer from MoV to [Fe/S I] centres is fast even at ca. 4 K. However, transfer to the [Fe/S 111 centre both from MoV and [Fe/S I]- is ‘infinitely’ slow at 77 K but sets in on annealing. These rates should be readily measured in the range 170-200 K.Unfortunately, since the flavin semiquinone intermediate cannot be detected accurately by ESR spectroscopy under our conditions, rates involving these centres are inaccessible. Massey and c o - ~ o r k e r s ~ ~ ~ 31 have shown that at room temperature, the semiquinone transfers its electron to one of the Fe/S centres with a rate constant of ca. 290 s-l. They conclude3’ that a that electrons equilibrate between Mo and [Fe/S I]- centres relatively slowly (ca. 77 s-l) and between other centres even more slowly must be in error. Our results suggest that there should be large rate differences for transfer between these centres at room temperature.Sub h ur Cen t res Although there is general agreement that the centre with g, = 2.06, g , = 2.025 and g, = 2.002 (fig. 3) is derived from RSSR’- radicals by protonation, there is argument as to ,33 or the break-down product, whether it is the simple proton adduct RS‘S RSS*.34* 35 Its formation confirms that electron addition also occurred at the RSSR units, which are apparently not able to pass the electron on to one of the other centres, suggesting that the RSSR unit is well removed from the normal redox centres. This /H ‘RM. C. R. Syrnons, F. A . Taiwo and R. L. Petersen 4073 suggests that it is not involved in the normal reactions of xanthine oxidase, probably having a structural rather than a mechanistic r6le. Addition of electrons to RS-SR units has also been observed for aqueous solutions of xanthine oxidase at room temperature during pulse radiolysis.31 It seems that, for powerful electron donors such as CO,, this step is by far the most facile, even though it is the least favoured, thermodynamically. These results are of some importance since they show that electron reactions not involving specific substrates may proceed via unimportant side reactions which must be taken into account in kinetic studies. Conclusions Since electron-capture within the enzyme must initially be random, and since the Fe/S(II) centre is not formed on exposure at 77 K, there must either be a major barrier to electron capture by this centre or it must transfer captured electrons to other centres. Clearly, no major barrier exists for the Fe/S(I) centre, and Fe/S(I)-l is readily formed.Since MeV* and Fe/S(I) are thought to be close together, it is probable that e- capture by MeV' is also facile, but rapid electron-transfer to Fe/S(I) occurs. Hence MoV centres accumulate only in molecules containing Fe/S(I)- units. On further annealing, Fe/S(II) can accept electrons and Fe/S(II)- is formed by transfer from MoV units, and, in part, from Fe/S(I) units. The presence of a thermal barrier to e- capture by an iron-sulphur cluster would be surprising. It seems to be generally supposed that there is little change in shape for the cluster on e- addition. If this is true, it is difficult to understand the nature of any barrier. Alternatively it could be that there is no major barrier to e- addition to the Fe/S(II) unit, but, under our unrelaxed conditions at 77 K, e- transfer to Fe/S(I) is rapid.On warming the reverse transfer becomes favoured because relaxed stabilisation by hydrogen bonding or even proton transfer is greater for Fe/S(II) units. We tentatively suggest that the problem may be connected with the tendency for bridging sulphide ligands to form hydrogen Such H-bonds will increase the effective electron affinity of a cluster, and will be more important in the reduced form of the cluster. One way of explaining the curious behaviour outlined above would be that the Fe/S(I) cluster has several hydrogen bonds prior to e- addition, thereby favouring addition prior to any relaxation. Thus at 77 K e- addition to Fe/S(II) is rapidly followed by e- transfer to Fe/S(I). However, if we postulate that strong hydrogen bonding or perhaps proton transfer can grow in relatively slowly for Fe/S(II)-, this can reverse the relative e-- affinities, thus tipping the equilibrium back to favour Fe/S(II)- at higher temperatures.Once this has occurred, the Fe/S(II)- unit is permanently stabilised, so there is no temperature-induced reversible behaviour. We thank the British Council and European Development Fund for a grant to F. A. T. ; Memphis State University for a Faculty Development Leave to R. L. P. ; Dr Russ Hille for a generous gift of purified xanthine oxidase and both Dr Hille and Dr R. F. Anderson for very helpful discussions. References I V. Massey, P. E. Brumby, G. Palmer and H. Komai, J. Biol. Chem., 1969, 244, 1682.2 J. Bordas, R. C. Bray, C. D. Garner, S. Gutteridge and S. S. Hasnain, Biochem. J., 1980, 191, 499. 3 S. P. Cramer, R. Wahl and K. V. Rajagopalen, J . Am. Chem. Soc., 1981, 103, 7721. 4 R. C. Bray, Enzymes, 1975, 12, 299. 5 D. J. Lowe and R. C. Bray, Biochem. J., 1978, 169, 471. 6 R. C. Bray, J. Less Common Metals, 1974, 36, 413. 7 R. C. Bray, Biochem. J . , 1961, 81, 196. 8 D. Edmondson, D. Ballou, A. Van Heuvelen, G. Palmer and V. Massey, J . Biol. Chem., 1973, 248, 6135.4074 Electron Addition to Xanthine Oxidase 9 H. Komai, V. Massey and G. Palmer, J . Biol. Chem., 1969, 244, 1692. 10 G. D. D. Jones, J. S. Lea, M. C. R. Symons and F. A. Taiwo, Nature (London), 1987, 772. 11 M. C. R. Symons and R. L. Petersen, Proc. R. SOC. London, Ser. B, 1978, 201, 285.12 M. C. R. Symons and R. L. Petersen, Biochim. Biophys. Acta, 1978, 537, 70. 13 D. N. R. Rao, M. C. R. Symons and J. M. Stephenson, J. Chem. SOC., Perkin Trans. 2, 1983, 727. 14 M. C. R. Symons and F. A. Taiwo, J. Chem. SOC., Faraday Trans. 1, 1987, 83, 3653. 15 S. E. Peterson-Kennedy, J. L. McGourty, J. A. Kalweit and B. M. Hoffman, J. Am. Chem. SOC., 1986, 16 A. G. Sykes, Chem. SOC. Rev., 1985, 14, 283. 17 H. B. Gray, Chem. SOC. Rev., 1986, 15, 17. 18 G. McLendon, Ace. Chem. Res., 1988, 21, 160. 19 R. Hille and V. Massey, Molybdenum Enzymes, ed. T. G. Spiro (Wiley, New York, 1985), p. 443. 20 M. C. R. Symons and R. L. Petersen, J. Chem. Res. (S), 1978, 382. 21 M. C. R. Symons, J. Chem. SOC., Perkin Trans. 2, 1974, 1618. 22 M. C. R. Symons in Oxygen and Sulphur Radicals in Chemistry and Medicine, ed. A. Breccia, M. A. J. Rogers and G. Semeran (Lo Scarabeo, Bologna, 1986), p. 71. 23 R. C. Bray, D. J. Lowe, C. Capeillere-Blandin and E. M. Fielden, Biochem. SOC. Trans., 1973, 1, 1067. 24 M. C. R. Symons, Pure Appl. Chem., 1981,53, 223. 25 G. N. George and R. C. Bray, Biochem., 1988, 27, 3603. 26 S. J. Tanner, R. C. Brau and F. Bergmann, Biochem. SOC. Trans., 1978, 6, 1327. 27 D. J. Lowe, R. M. Lynden-Bell and R. C. Bray, Biochem. J., 1972, 130, 239. 28 R. E. Coffman and G. R. Buettner, J . Phys. Chem., 1979, 83, 2392. 29 M. J. Barber, J. C. Salerno and L. M. Siegel, Biochem., 1982, 21, 1648. 30 J. S. Olson, D. P. Palmer and V. Massey, J . Biol. Chem., 1974, 249, 4363. 31 R. F. Anderson, R. Hille and V. Massey, J. Biol. Chem., 1986, 261, 15870. 32 A. Battacharyya, G. Tollin, M. Davis and D. F. Edmondon, Biochem., 1983, 22, 5270. 33 D. J. Nelson, R. L. Petersen and M. C. R. Symons, J. Chem. Soc., Perkin Trans. 2, 1978, 225. 34 J. H. Hadley and W. Gordy, Proc. Natl Acad. Sci., 1975, 72, 3486; 1974, 3106. 35 F. C. Adam and A. J. Elliot, Can. J. Chem., 1977, 55, 1546. 36 C. W. Carter, Jr, J. Biol. Chem., 1977, 252, 7802. 108, 1739. Paper 9/01 579E ; Received 17th April, 1989

ISSN:0300-9599    

DOI:10.1039/F19898504063

出版商:RSC    

年代:1989

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. 1, 1989, 85(12), 4075-4082 Stabilization of Biochemically Interesting Intermediates by Metal Coordination Part 5.-Complexes of Zn", Cu', Re' and Ru" with Singly Reduced 2,5-Diace tylpyrazine 7 Christian Bessenbacher, Sylvia Ernst, Stephan Kohlmann, Wolfgang Kaim," Volker Kasack, Eberhard Roth and Jeanne Jordanov Institut fur Anorganische Chemie der Universitat, Pfaflenwaldring 55, 0-7000 Stuttgart 80, Federal Republic of Germany and De'partement de Recherche Fondamentale, Centre d'Etudes Nucle'aires, 85X, F-38041 Grenoble Cedex, France Homobinuclear chelate complexes of the 2,5-diacetylpyrazine anion radical with the d'O and d6 metal fragments Zn2+, [(PPh,),Cu]+, (OC),ClRe and [(bpy),RuI2+ have been studied by ESR and cyclic voltammetry. The radical state of this centrosymmetric ligand with two biochemically relevant a- carbonylpyrazine chelate sites is considerably stabilized by coordination. The redox-active a-iminoketone (or p-azaenolate) chelate arrangement is a common occurrence in biochemical molecules methoxatin5 or antibiotics of the streptonigrin type :6' R '.--O flavin 0 such as flavins,l.pterin~,~. methoxatin streptonigrin Flavins and pterins contain a chelating a-carbonylpyrazine group and N,O- coordination in different oxidation states of these two-step redox systems has been t Part 4, ref. (10). 407 54076 Stabilization of Biochemically Interesting Intermediates discussed especially for the flavins.' Metal coordination was found to stabilize the flavosemiquinone intermediate (and vice versa).' In an attempt to understand this mechanism we used the centrosymmetric bis-chelating model ligand 2,5-diacetyl- pyrazine' (dapz), for coordination with the metal fragments Zn2+, (PPh,),Cu+, (CO),ClRe and [(bpy),Ru]'+ (bpy : 2,2'-bipyridine).The latter three fragments contain fairly electron- rich metals which are capable of n-back-bonding to n-acceptor ligand~,~-'l complexes of monochelating 2-acetylpyridine9 e.g. with CU' salts have been reported. l2 The compounds have been studied by absorption spectroscopy and cyclic vol- tammetry, while ESR was used to characterize the persistent singly reduced forms, the radical complexes. Results from Huckel MO perturbation calculation^'^^ l4 were correlated with the spectroscopic data. Experiment a1 ESR spectra were recorded in the X-band on Varian E9 and 109 spectrometers.g Factors and coupling constants were determined either by use of the double-cavity technique with perylene radical anion as reference or by directly measuring magnetic field and microwave frequency (Varian NMR gaussmeter, EIP 548A frequency counter). Liquid-helium temperatures were generated using an Oxford Instruments ESR 900 cryostat. 2,5-Diacetylpyrazine was prepared via the radical reaction described by Minisci and co-workers.' All subsequent reactions were carried out under an atmosphere of dry argon. The free-radical anion dapz'- was generated electrolytically by reducing a 0.001 mol dmP3 solution of dapz in dry N,N-dimethylformamide/O. 1 mol dmP3 tetrabutylam- monium perchlorate at ca.- 1.5 V. The zinc triple ion [Zn(dap~)Zn]'~+ was obtained by treating dapz with zinc powder in dichloromethane containing 0.1 mol dm-, Bu,N+ClO,. The synthesis of the binuclear rhenium complex has been described previously. lo The very labile binuclear complex { (dapz) [Cu(PPh,),],) (BF,), was prepared from 100 mg (0.61 mmol) dapz and 1680 mg (1.4 mmol) [Cu(PPh,),](BF,) in 100 cm3 di- chloromethane. After 2 h at room temperature the purple solution was concentrated and the product precipitated by addition of hexane [yield 412 mg (45 %)I. The compound was too labile for elemental analysis. The ruthenium dimer was directly obtained as the radical anion complex {(dapz)[Ru(bpy),],)(PF,),. 100 mg (0.61 mmol) of dapz and 650 mg (1.25 mmol) of Ru(bpy),Cl, - 2H,O were heated to reflux for 3 h in water-ethanol (10 : 1).The product was obtained by adding a saturated aqueous ammonium hexafluorophosphate solution to the cooled reaction mixture ; the precipitate was collected and chromatographed on a column (alumina Woelm A, super I, W 200, acetone as primary eluent). The green-brown zone eluted with acetonitrile was collected, the solution was concentrated and the paramagnetic product was precipitated with diethyl ether [yield 90 mg (9.4 %)I. Found : C, 41.01 ; H, 2.72; N, 9.98 ; C,,H,,F,,Nl,O,P,Ru, (1425.95) requires : C, 40.42; H, 2.74; N, 9.83.C. Bessenbacher et al. 4077 Table 1. Oxidation and reduction potentialsa of the 2'5- diacetylpyrazine radical anion and of its binuclear complexes (LJ) (dapz'-) (ML,) MLn E o x Ered AE solventb - -0.95 - 1.59 0.64 DMF [(pph3)2cU]' +0.04" -0.73 ca.0.7 DCE DPY) 2Rul 2+ +0.23 -0.23 0.46 AN Cl(CO),Re +0.48 -0.17 0.65 AN a From cyclic voltammetry (100 mV s-l) at a glassy carbon electrode, saturated calomel reference electrode (SCE). 0.1 mol dmT3 solutions of tetrabutylammonium perchlorate, DMF = N,N-dimethylformamide, DCE = lY2-dichloroethane, AN = acetonitrile. Anodic peak potential (scheme 1). Results and Discussion HMO Calculations Neglecting the methyl group, using a common Coulomb integral parameter h for the N and 0 coordination centres and a resonance integral parameter kc-, = 1.3,15 we calculated the energy levels and the electron distributions of the lowest (LUMO) and second-lowest unoccupied 71 molecular orbital (SLUMO) of the 10-centre n system (3).For the whole range of 0.0 < h, = h, < 2.0 of the Coulomb integral parameters there is a fairly large LUMO/SLUMO gap. While the SLUMO has most of its electron density on the 3,6-positions and a little on the nitrogen, the charge is more evenly distributed in the LUMO, corresponding to a 'resonance' situation: The LUMO electron density is rather small on the ring CH centres but fairly large on the (acety1)C atoms. # @ LUMO SLUMO (5) depicts the LUMO and SLUMO electron densities (the former corresponding to the HMO spin populations of the anion radica1)16,17 for h, = h, = 0.75. Electrochemistry Cyclic voltammetry reveals (table 1) that the ligand 2,5-diacetylpyrazine is already rather easily reduced in two steps which are separated by 0.64 V.Coordination by two strong4078 Stabilization of Biochemically Interesting Intermediates I I 1 +0.6 0.0 EIV -1 .o Fig. 1. Cyclic voltammogram of {(dapz) [Cu(PPh,),],) (BF4)2 in 1,2-dichloroethane/O. 1 mol dm-3 Bu,N+ClO;. For interpretation see text and scheme 1. Fig. 2. ESR spectrum of the electrochemically generated dapz radical anion in DMF/O. 1 mol dm-3 Bu,N+ClO; at 300 K. metal electrophiles to four centres of high LUMO electron density causes sizeable positive shiftsls of both potentials. These effects are strongest with the neutral Re(CO),Cl for the first electron uptake''? 11, l9 and with the dipositive bis(bipyri- dine)ruthenium(II) fragment for the second reduction. The marked decrease in the potential difference AE for the binuclear ruthenium system can be attributed to charge effects and to significant interactions with the metal d orbitals.'l Considering the rather positive potentials, it is not surprising that the ruthenium complex was isolated in the radical ( 3 + ) form'' and that the dirhenium system also revealed the presence of paramagnetic (reduced) fractions upon isolation.lo The radical stabilization through coordination as established in flavin chemistry1 is apparently also found for model systems as described in this work. The electrochemistry of the dicopper(1) system (fig. 1) indicates why it is difficult to isolate this complex: owing to the low basicity of the dapz ligand and the relatively weak1' back-donation from the metal centres there is partial dissociation of the non- reduced binuclear complex even in 1,2-dichloroethane solution.Reduction of the binuclear form (A) and of the mononuclear form (B) is clearly visible (fig. 1).20 TheC. Bessenbacher et al. 4079 strong increase in ligand basicity after the first reductionl3' 21 causes coordinative saturation,22 i.e. formation of only the binuclear radical complex, as is evident from the cathodic peak current for the second reduction. In the return scan, the anodic peak currents for the first and second reoxidation are similar and there is no signal from a reduced mononuclear complex. Dissociation sets in only after second oxidation to the dication. Scheme 1 describes this ' ECEC ' sequence,23 which has similarly been observed for other binuclear complexes. 13, 20, 24 Again, it illustrates the superior coordinating properties of reduced paramagnetic forms, as has been established for flavins.l A /-e- +e- B I L = dapz [CU'] = (Ph3 P)2Cut Scheme 1.ESR Spectroscopy The ESR spectra of reduced acyl-substituted 7z systems show the effect of hindered rotation around the C(O)-C(n) bond because the bond order of this 'single' bond increases considerably after electron addition to the LUM0.25 While more symmetrical disubstituted systems such as the 1,4-diacetylbenzene radical anion thus show the presence of 'cis' and 'trans' isomers,25 the free dapz'- should be able to adopt three coplanar conformations with maximal 7~ overlap : "i' CH3 $LH. @o T .Jy0 I CH3 H3C'1 0 "'"'1 0 So far, it has not been possible to analyse the room-temperature ESR spectrum of the electrolytically generated dapz radical anion which shows signs of alternating linewidth effects (fig.2).26 Variable-temperature measurements and double-resonance experiments are necessary to establish the coupling constants and contributions from each of the three conformers. Only the total splitting Ca(N, H) and the g value could be determined. Addition of Zn2+ dramatically simplifies the spectrum: the metal centres can coordinate to the two chelate sites of the ligand to give a 'triple ion ',27 thus reducing the number of degrees of freedom and leaving only one (6a) of the three conformations possible. The ESR spectrum shows coupling from the acetyl(6H) and pyrazine protons (2H) as a septet of triplets (fig, 3); the I4N coupling is apparently too small to be detected even at fairly high resolution.4080 Stabilization of Biochemically Interesting Iaterrnediates Fig.3. ESR spectrum of the bis-chelate triple-ion formed from dapz radical anion and two zinc cations in dichloromethane at room temperature. Fig. 4. ESR spectrum of the electrochemically produced cation radical ((dapz) [Cu(PPh,),],}'+ in dichloromethane at 300 K [low-field section (a)] and computer simulation with the data from table 2 and a linewidth of 0.05 mT (6).C. Bessenbacher et al. 408 1 I I I I 31 0 320 330 340 H/mT Fig. 5. ESR spectrum of ((dap~)[Ru(bpy),],)'~+ in frozen acetone (3.8 K). Table 2. ESR dataa of the 2,Sdiacetylpyrazine radical anion and of its binuclear complexes (L,M) (dapz'-) (ML,) C C C 1.52 2.0043 DMF Zn2+ <0.01 0.064 0.303 1.946 2.0038 CH,Cl, [(Ph3P) 2CU1+d 0.076 0.076 0.305 2.286d 2.0039 CH,Cl, [(bPY),W2+ e e e e 1 .9977f (CH,),CO Cl(CO),Re e e e e 2.O0Sg CH,CN a Coupling constants a in mT.(see text). 1.9645, g, = 2.0027, g, = 2.0252 (at 3.8 K in acetone). g , = 1.972, g, = 2.0043, g, = 2.040 measured at 4 K in acetonitrile [ref. (lo)]. Total spectral width from 'H and 14N coupling. Not analysed Not resolved. g, = Isotropic value calculated from a(63,65Cu) = 1.06 mT (2 Cu); a(,lP) = 0.835 mT (4 P). A similar coordinative (chelate) fixation of the ligand conformation should occur in the binuclear copper(1) complex ; however, the presence of two equivalent 6 3 9 65Cu nuclei ( I = $) and four equivalent 31P centres ( I = i) results in a large number (3675) of theoretical ESR' lines.Considerable overlap reduces the number of lines actually observed (fig. 4) ; however, a computer-simulated spectrum was still needed to analyse the spectrum. ESR spectra of radical complexes with the Re(CO),Cl and [Ru(bpy),I2+ fragments are not usually resolved because of small line-broadening contributions from 35* 37Cl and 14N(bpy) coupling constants,2s* 29 therefore we studied the two dapz radical intermediates at low temperatures in order to determine the g anisotropy (fig. 5, table 2). The ESR data from table 2 allow several interesting comparisons to be made : deviations of g from the value for the free ligand are most notable for the complexes of heavy metals Re and Ru with their large spin-orbit coupling constants.30* 31 As was argued previo~sly,~~ the rather modest increase of (g) for the rhenium halide complex can be attributed to the presence of relatively low-lying unoccupied MOs associated with the (axial) metal- halogen bond.Close-lying unoccupied MOs are certainly present in the ruthenium dimer because of the bpy co-ligands with their low-lying n* levels;ll* 28 accordingly, the (g) factor lies below the value of the free ligand or, for that matter, of the free electron.28 While the extent of deviation of g from g (electron) = 2.0023 is determined by the spin-orbit coupling constants of the atoms according to their individual share of electron spin density in the singly occupied molecular orbital, the sign of Ag reflects the4082 Stabilization of Biochemically Interesting Intermediates positioning of the next fully occupied and completely empty molecular orbitals relative to the singly occupied M0.l'. 28 A g value below g (electron) indicates that there are low- lying unoccupied MOs rather close to the singly occupied molecular orbital.'* The g anisotropy is a little larger for the rhenium dimer than for the ruthenium complex, in agreement with the larger spin-orbit coupling constant30 of the heavier metal centre.This anisotropy is larger than that found for dimeric a-di-imine radical com~lexes,~~ but smaller than that of azodicarboxylic ester-bridged systems, which show stronger contributions from a mixed-valence situation. l1 It should be finally noted that no ESR spectroscopic or electrochemical complications because of possible isomerism of the ruthenium and rhenium systemslO.11, 2 9 9 32 were found. Both dimers contain two equivalent chiral metal centres and may thus form pairs of enantiomers and meso forms. Generous support for this work has come from Deutsche Forschungsgemeinschaft and Stiftung Volkswagenwerk. We also thank the DAAD for a grant within the French-German scientific exchange program ' PROCOPE ' and the Degussa AG for donation of RuC1,. References 1 P. Hemmerich and J. Lauterwein, in Inorganic Biochemistry, ed. G. L. Eichhorn (Elsevier, Amsterdam, 2 M. J. Clarke, Comm. Znorg. Chem., 1984, 3, 133. 3 S. J. N. Burgmayer and E. Stiefel, J. Am. Chem. Soc., 1986, 108, 8310. 4 T. Kohzuma, A. Odani, Y. Morita, M. Takani and 0. Yamauchi, Znorg. Chem., 1988, 27, 3854. 5 S. Suzuki, T. Sakura, S. Itoh and Y.Oshiro, Znorg. Chem., 1988, 27, 392. 6 J. Hajdu and E. C. Armstrong, J. Am. Chem. SOC., 1981, 103, 232. 7 A. Garnier-Suillerot and M. M. L. Fiallo, Reel. Trav. Chim. Pays-Bas, 1987, 106, 391. 8 T. Caronna, G. Fronza, F. Minisci and 0. Porta:J. Chem. SOC., Perkin Trans. 2, 1972, 2035. 9 W. Kaim and S. Kohlmann, Znorg. Chem., 1987, 26, 1469. 1975), pp. 1168-1 190. 10 C. Bessenbacher and W. Kaim, Z. Anorg. Allg. Chem., in press. 11 (a) W. Kaim, V. Kasack, H. Binder, E. Roth and J. Jordanov, Angew Chem., 1988, 100, 1229; Angew. Chem. Znt. Ed. Engl., 1988, 27, 1174; (b) S. Ernst, P. Hanel, J. Jordanov, W. Kaim, V. Kasack and E. Roth, J. Am. Chem. SOC., 1989,111, 1733; (c) C. Bessenbacher, C. Vogler and W. Kaim, Znorg. Chem., in press. 12 M. A. S. Goher, Bull.SOC. Chim. Fr., 1981, 1-329. 13 W. Kaim, Coord. Chem. Rev., 1987, 76, 187. 14 W. Kaim, Chem. Ber., 1981, 114, 3789. 15 E. Heilbronner and H. Bock, The HMO Model and its Applications (Verlag ChemieIWiley, 16 F. Gerson, High Resolution E.S.R. Spectroscopy (Wiley/Verlag Chemie, New YorkIWeinheim, 1970). 17 M. Symons, Electron-Spin Resonance Spectroscopy (Van Nostrand Reinhold, New York, 1978). 18 S. Ernst and W. Kaim, Znorg. Chem., 1989,28, 1520. 19 W. Kaim, H. E, A. Kramer, C. Vogler and J. Rieker, J. Organomet. Chem., 1989, 367, 107. 20 W. Kaim, S. Kohlmann, A. J. Lees and M. M. Zuln, 2. Anorg. Allg. Chem., in press. 21 S. Ernst and W. Kaim, J. Am. Chem. SOC., 1986, 108, 3578. 22 W. Kaim, B. Olbrich-Deussner, R. Gross, S. Ernst, S. Kohlmann and C. Bessenbacher in Importance of Paramagnetic Organometallic Species in Activation, Selectivity and Catalysis, ed. M. Chanon et al. (Reidel, Dordrecht, 1989), pp. 283-294. Weinheim/London, 1976). 23 J. Heinze, Angew Chem., 1984, 96, 823; Angew. Chem. Znt. Ed. Engl., 1984, 23, 831. 24 R. Gross and W. Kaim, J. Organomet. Chem., 1987, 333, 347. 25 P. H. Rieger and G. K. Fraenkel, J. Chem. Phys., 1962, 37, 281 1. 26 A. Hudson and G. R. Luckhurst, Chem. Rev., 1969,69, 191. 27 H. Sharp and M. C. R. Symons in Ions and Zon Pairs in Organic Chemistry, ed. M. Szwarc (Wiley, New 28 W. Kaim, S. Ernst, S. Kohlmann and P. Welkerling, Chem. Phys. Lett., 1985, 118, 431. 29 W. Kaim and S. Kohlmann, Chem. Phys. Lett., 1987, 139, 365. 30 B. A. Goodman and J. B. Raynor, Adv. Znorg. Chem. Radiochem., 1970, 13, 135. 31 S. Kohlmann, V. Kasack, E. Roth and W. Kaim, J. Chem. Soc., Faraday Trans. I, 1989, 85, 0000. 32 S. Ernst, V. Kasack and W. Kaim, Znorg. Chem., 1988, 27, 1146. York, 1972). Paper 9/01597C; Received 17th April, 1989

ISSN:0300-9599    

DOI:10.1039/F19898504075

出版商:RSC    

年代:1989

数据来源: RSC