摘要:

J. Chem. Soc., Perkin Trans. 2, 1997 2461 Conjugated p-(tetrathiafulvalenylmethylideneamino)calix[4]arene Jean-Bernard Regnouf-de-Vains,*,a Marc Sallé b and Roger Lamartine a a Laboratoire de Chimie Industrielle, CNRS-ESA 5078, Université Claude Bernard Lyon I, 69622 Villeurbanne, France b Laboratoire d‘Ingéniérie Moléculaire et Matériaux Organiques, Université d‘Angers, 49045 Angers, France The tetrathiafulvalene has been coupled via an imine bond to the p-(tert-butyl)calix[4]arene upper rim.In the field of supramolecular chemistry, the calix[4]arene displays interesting behaviour as a pure entity 1 and as an organizing platform for various kinds of substituents such as acids, esters, amides and, more recently, transition metal heterocyclic complexing agents.2 As far as we know, no specific studies have been related in this sense to the organization, around a calixarenic macrocycle, of tetrathiafulvalenyl (TTF) species. This biheterocyclic electron-donor group, well known for forming electron-conducting charge transfer complexes with electron acceptors such as tetracyanoquinodimethane (TCNQ), has been widely studied since the early seventies but, as mentioned by Becher and co-workers,3 relatively little attention has been given to its implication in supramolecular chemistry as a building-block of supramolecular devices as well as a simply spatially organized electroactive substructure. For this last purpose, in relation with some of our recent works dedicated to the organization of biheterocyclic chelating systems at the lower and upper rims of calix[4]arene,2d,e,g we describe here the synthesis and some physico-chemical properties of the first conjugated electroactive TTF-calixarene in which conjugation is maintained via an imine bond.We took as a model platform the tri-tert-butylcalix[4]arene,4 which was nitrated according to the literature 5 then reduced into its mono-amino analogue 6 following the procedure of Hosseini and co-workers.7 The amine was reacted with TTF monocarboxaldehyde 8 in the presence of 3 Å molecular sieves in refluxing EtOH-free CH2Cl2 to give after chromatography (SiO2, CH2Cl2 :hexane) the desired imine 1 (53% yield).† Positive mode ESMS spectrum of 1 (CH3CN as medium) displayed the proto- and sodio-charged peaks at 822.4 and 844.4 amu respectively, but accompanied by an intense peak at 821.4 amu which was attributed to the corresponding radical cation. † 1: Mp: 336–337 8C; lmax(CH2Cl2)/nm 288 (e/dm3 mol21 cm21 24 000), 320 (18 500), 456 (5150); lmax(CH3CN)/nm 286.6 (e/dm3 mol21 cm21 25 350), 314.7 (19 225), 350.0 (sh); 437.5 (7125); dH(CDCl3 1 TMS, 300.133 MHz, J values given in Hz) 1.186 (s, 9H, But) 1.235 (s, 18H, But), 3.50 and 4.26 (AB, JAB 13, 8H, bridge CH2), 6.10 (d, J 2.5, 2H, TTF), 6.81 (s, 1H, TTF), 6.91 (s, 2H, Ar), 7.03 (s, 2H, Ar), 7.053 and 7.095 (AB, JAB 2.3, 4H, Ar), 7.97 (s, 1H, N]] CH), 10.25 (s, 4H, OH); dH(CD3CN) 1.178 (s, 9H, But), 1.185 (s, 18H, But), 3.59 (br s, 4H, bridge CH2), 4.14 (br s, 4H, bridge CH2), 6.46 (br s, 2H, TTF), 7.09 (br s 1 s, 1H TTF and 2H Ar), 7.18 and 7.26 (AB, JAB 2.2, 4H, Ar), 7.228 (s, 2H, Ar), 8.16 (s, 1H, N]] CH), 9.54 (s, 4H, OH); dC(CDCl3 1 TMS, 50.32 MHz) 31.51, 31.65 (Me, But), 32.53 and 32.60 (bridged CH2), 34.13 and 34.25 (C, But), 118.53 119.25, 121.91, 125.91, 126.04, 126.32 and 128.06 (4, 49, 59 of TTF; 3,5-Ar), 148.54 [N]] C(H)], 127.14, 127.73, 128.17, 129.23, 139.12, 144.63, 144.80, 144.85, 146.36, 146.82 and 148.23 (5, 2 and 29 of TTF; 2,6-Ar, 4-Ar, 1-Ar) (Found: C, 68.58; H, 6.35; N, 1.71; S, 15.40.Calc. for C47H51NO4S4 (822.19): C, 68.66; H, 6.25; N, 1.70; S, 15.60%). m/z (ESMS) negative mode 820.4 [M 2 H1]2; positive mode 821.4 [M?]1, 822.4 [M 1 H]1; 844.4 [M 1 Na]1. 1H NMR analysis of 1 in CDCl3 gave a well-resolved spectrum displaying notably a characteristic AB pattern for the Ar]CH2]Ar groups, a doublet (J = 15 Hz) and a singlet at 6.10 and 6.82 ppm, respectively, for the TTF unit.Changing CDCl3 for the more polar CD3CN resulted in the broadening of the Ar]CH2]Ar AB pattern and of the TTF signals; the latter and the imino proton supported a downfield shift of about 0.3 ppm. Titration of 1 by TCNQ in this solvent resulted in the rapid modification of the aromatic region (Fig. 1). After addition of 0.2 equiv. of TCNQ, the TTF resonance signals disappeared while those of the aromatic protons at 7.09 ppm and the imino group were broadened.At 0.6 equiv., the whole aromatic pattern gave a large multiplet, this was accompanied by the quasi-extinction of the imino-proton resonance signal. The tert-butyl and the Ar]CH2]Ar resonance signals were not affected by this addition, suggesting that the cavity was not implied in this process. These results suggest that the expected charge transfer complex is formed in these conditions. Attempts to isolate it failed until now.Additon of NEt3 to 1 in CH3CN, followed by UV–VIS spectroscopy (Fig. 2), resulted in an ipsochromic shift of the TTF transition from 437 to 412 nm with an increase of the molar absorptivity coefficient from 5700 to 11 800 dm3 mol21 cm21. The shoulder located at 350 nm, attributed to the imino-phenol group, disappeared while the band at 412 nm was formed. The latter, attributed to the imino-phenolate, joins the former via an isosbestic point at 358 nm and is gradually added to the close TTF band, thus explaining the high molar absorptivity coefficient at 412 nm.This titration involved exactly one equivalent of amine, leading to the conclusion that the conjugated phenolic group becomes the more acidic in the molecule, due to the increased mesomeric stabilization of the corresponding phenolate anion. Fig. 1 1H NMR spectrum of aromatic and heterocyclic protons of 1 (CD3CN, 300 MHz, room temp.): (a) alone (4.9 × 1023 mol dm23); in the presence of: (b) 0.2 equiv., (c) 0.6 equiv.of TCNQ2462 J. Chem. Soc., Perkin Trans. 2, 1997 The p-donor character of this TTF-grafted calixarene was evaluated by cyclic voltametry.‡ Compound 1 exhibits two fully reversible one-electron oxidation waves in acetonitrile (Fig. 3), Fig. 2 UV–VIS titration of 1 by NEt3 in acetonitrile solution: (a) alone (4.5 × 1025 mol dm23); in the presence of: (b) 0.07, (c) 0.21, (d ) 0.35, (e) 0.49, ( f ) 0.64, ( g) 0.75, (h) 0.90 and (i) 1.06 equiv.of NEt3 Fig. 3 Cyclic voltamogram of 1, (a) alone; in the presence of (b) 0.5, (c) 1 equiv. of Et3N ‡ 1023 mol dm23 in acetonitrile, Bu4NPF6 (0.1 mol dm23), 100 mV s21, 20 8C, argon atmosphere. with peak potentials (Epai) slightly more positive than for TTF itself, in respect of the withdrawing character of the imino functionality (1: Epa1 = 0.44; Epa2 = 0.83 V vs. SCE; TTF: 0.36 and 0.72 V vs. SCE). Interestingly, this good p-donating ability can be strengthened upon treatment with triethylamine. Indeed, a stepwise addition of Et3N (0.1 equiv.portions) results in the progressive appearance of a new reversible redox system located at Epa19 = 0.33 V, which corresponds to a 110 mV negative shift for the first oxidation potential (Fig. 3). The easier oxidizability of 1 in the presence of Et3N is attributed to the generated phenolate anion which acts as an electron donating substituent for the conjugated TTF fragment.Note, this new redox system remains reversible from 0 to 1 equiv. of added triethylamine, with a loss of reversibility appearing for larger amounts of base. Attempts to prepare electrogenerated cation radical salts derived from compound 1 are in progress, as well as the elaboration of its polysubstituted analogues. Acknowledgements The authors wish to thank Mrs Nicole Marshall for correcting the manuscript. References 1 (a) C. D. Gutsche, Calixarenes, Monographs in Supramolecular Chemistry, ed.J. F. Stoddart, The Royal Society of Chemistry, Cambridge, 1989; (b) Calixarenes, a Versatile Class of Macrocyclic Compounds, ed. V. Bohmer and J. Vicens, Kluwer, Dordrecht, 1990; (c) S. Shinkai, Calixarenes—The Third Generation of Supramolecules, Tetrahedron Report Number 340, Tetrahedron, 1993, 49, 8933. 2 (a) P. D. Beer, J. P. Martin and M. G. B. Drew, Tetrahedron, 1992, 48, 9917; (b) R. Grigg, J. M. Holmes, S. K. Jones and W. D. Amilaprasadh Norbert, J.Chem. Soc., Chem. Commun., 1994, 185; (c) G. Ulrich and R. Ziessel, Tetrahedron Lett., 1994, 34, 6292; (d ) J.-B. Ragnouf-de-Vains and R. Lamartine, Helv. Chim. Acta, 1994, 77, 1817; (e) J.-B. Regnouf-de-Vains, R. Lamartine, B. Fenet, C. Bavoux, A. Thozet and M. Perrin, Helv. Chim. Acta, 1995, 78, 1607; ( f ) P. D. Beer, Z. Chen, A. J. Goulden, A. Grieve, D. Hesek, F. Szemes and T. Wear, J. Chem. Soc., Chem. Commun., 1994, 1269; ( g) J.-B. Regnouf-de-Vains and R. Lamartine, Tetrahedron Lett., 1996, 37, 6311. 3 T. Jørgensen, T. Kruse Hansen and J. Becher, Chem. Soc. Rev., 1994, 41; (b) J. Becher, Z.-T. Li, P. Blanchard, N. Svenstrup, J. Lau, M. Brønsted Nielsen and P. Leriche, Pure Appl. Chem., 1997, 69, 465. 4 S. Berthalon, J.-B. Regnouf-de-Vains and R. Lamartine, Synth. Commun., 1996, 26, 3103. 5 K. C. Nam and D. S. Kim, Bull. Korean Chem. Soc., 1994, 15, 284. 6 J.-B. Regnouf-de-Vains, S. Berthalon and R. Lamartine, poster 67, 4th International Conference on Calixarenes, Parma, Italy, 31st August–4th September, 1997. 7 G. Mislin, E. Graf and M. W. Hosseini, Tetrahedron Lett., 1996, 37, 4503. 8 J. Garin, J. Orduna, S. Uriel, A. J. Moore, M. R. Bryce, S. Wegener, D. S. Yufit and J. A. K. Howard, Synthesis, 1994, 489. Paper 7/06640F Received 12th September 1997 Accepted 30th September 1997

ISSN:1472-779X    

DOI:10.1039/a706640f

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

J. Chem. Soc., Perkin Trans. 2, 1997 2463 Aryl-substituted derivatives of trimethylenemethane dianion: a dynamic NMR study Roy Shenhar and Mordecai Rabinovitz * Department of Organic Chemistry, The Hebrew University of Jerusalem, Givat-Ram, Jerusalem, Israel 91904 Evidence is given for the structure of aryl-substituted trimethylenemethane dianions from NMR data, and the importance of p-� conjugation is discussed. Cross-conjugated systems have been shown to possess a special stability compared to their linear analogues, and thus the parent trimethylenemethane dianion (TMM22) and its derivatives have become a focus of interest for the last several decades.Klein and Medlik 1 were the first to point out the facile formation of TMM22 as the dilithium salt using N,N,N9,N9- tetramethylethylenediamine (TMEDA) as a complexing agent. Extensive theoretical work has been carried out to establish the geometry of TMM22, and to understand the reasons for its remarkable stability.The first exciting suggestion was that TMM22 possesses a novel kind of aromaticity, coined ‘Y aromaticity’ 2 (that is feasible only when TMM22 has a planar D3h symmetry, making all four p-orbitals parallel to each other and thus enabling some kind of resonance through-the-centre), but this idea was doubted by Klein et al.,3 who claimed internal Coulombic stabilization to be the dominating factor, and later by Wiberg,4 who pointed out that TMM22 has to cope with 6pe2 over only three C]C bonds (compared to benzene, with only 1pe2 per bond), and suggested that the stability of TMM22 is due to its ability to distribute the extra charge to the three ‘corners’, thus minimizing the repulsive interactions.A support to Wiberg’s explanation, based on ab initio calculations,5 is found in the work of Frenking and co-workers,5a who reported TMM22 to exist in a non-planar geometry with strongly pyramidal methylene groups. They have shown that the planar geometry previously suggested is a high-order saddle point, and that the conformations found to be the true minima are characterized by a strong shift of s density towards the central carbon, with a concomitant counter-migration of p density to the terminal carbons, which is in accord with Wiberg’s notion.4 Here we report an NMR study † of tribenzylidenemethane dianion (1) 6 and dibenzylidene-(3,5-dimethylbenzylidene)- methane dianion (2) with the aim of giving further insight into the ideas mentioned above.Me Me H H H H H H 2– 2– 1 2 3 4 5 6 2¢ 3¢ 4¢ 5¢ 6¢ 1 2 3 4 5 6 1 2 † NMR experiments were carried out on a Bruker DRX-400 MHz spectrometer (400.13 MHz for proton, 100.61 MHz for carbon, THF, 295 K). Dianions 1 and 2 were prepared by a double deprotonation using n-butyllithium (without TMEDA) from the corresponding olefins. The olefins were prepared by a known procedure 7 ‡ (three isomers exist in the case of [H2]2 §). Both dianions 1 and 2 and the starting materials, i.e.olefins [H2]1 and [H2]2 were characterized by NMR spectroscopy. In order to simplify the discussion we shall denote all the 2- and 29-positions as ‘benzylic positions’ [and accordingly the ‘benzylic bonds’ will be all the bonds (C-2)–(C-3) and (C-29)–(C-39)], whereas whenever it will be needed to distinguish between the 2- and 29-positions it will be denoted specifically. Dynamic NMR experiments on the dianions in the temperature range 185 K–300 K revealed a dynamic behaviour for all the ortho and meta methine sites of 1 [DG‡ 241.7(H- 4) = 44.8 ± 1.3 kJ mol21 and DG‡ 213.6(H-5) = 44.8 ± 1.3 kJ mol21] and 2 [DG‡ 240.5(H-4) = 44.4 ± 1.3 kJ mol21, DG‡ 214.0(H- 5) = 45.2 ± 1.3 kJ mol21 and DG‡ 236.3(H-49) = 43.5 ± 1.3 kJ mol21], and also for the two methyl groups of 2 (DG‡ 214.0 = 47.3 ± 4.2 kJ mol21).This behaviour indicates the freezing of the rotation around the benzylic bonds on the NMR timescale. It is important to point out that the Y-frame protons in both dianions did not reveal any significant dynamic process.¶ Furthermore, C]H coupling constants measured for the benzylic positions indicate a hybridization (Table 1) of approximately sp2.5 (in both dianions) as compared with ca.sp2 hybridization for the ring carbons. Similarly, the charge densities (Table 1) at the Y-frame carbons indicate that only ca. 50% of the net charge (22) remained on the Y framework (in both Table 1 Charge distribution and hybridization for the Y skeleton carbons calculated from NMR data.Position r a Hybridization b dC dH 1JCH/Hz Dianion 1 12 10.08 20.35 — 2.51 145.9 80.4 — 4.52 s — 142.5 Dianion 2 122 9 10.08 20.35 20.35 — 2.51 2.50 146.0 80.5 80.1 — 4.53 s 4.51 s — 142.6 142.7 a Charge densities were calculated from the 13C chemical shifts, using O’Brien’s equation:8 r = (dC 2 134.1)/153.7, dC expressed in ppm. b Carbon hybridization was evaluated from the C]H coupling constants using the equation: n = 500/1JCH 2 1, where n is the power in spn, 1JCH is expressed in Hz units. ‡ For the preparation of [H2]2 ethyl (3,5-dimethylphenyl)acetate was used. § New compounds gave satisfactory mass spectra 2-benzyl-1-(3,5- dimethylphenyl)-3-phenylpropan-2-ol: white crystals, 21% yield, mp: 64 8C.Dehydration afforded three isomers ([H2]2): yellow oil, 80% yield, which were identified in the 1H NMR spectrum. ¶ The H-2 and H-29 protons of 2 were shifted at 245.8 K, but remained as singlets, retaining the same integration ratio of 2 : 1.2464 J.Chem. Soc., Perkin Trans. 2, 1997 dianions), and it follows that about half the charge was withdrawn from the Y framework (Fig. 1). All these data lead us to the conclusion that the benzylic bonds in both dianions had become closer in character to double bonds, due to an efficient p-p conjugation and delocalization of the extra charge into the rings. It is still not evident whether the rotation around the Y bonds slowed down with the lowering of the temperature. Slowing down the rotation about the Y bonds would have led to a fixed conformation, that corresponds to the global minimum of the potential energy surface. Bearing in mind that H-2 protons remained isochronous in both dianions 1 and 2 even at the lowest temperature measured, and that the two methyl groups of 2 became non-equivalent upon cooling, one may try to satisfy all the conditions by suggesting a conformation which has the two benzylidene groups in ‘exo,exo’ or ‘endo,endo’ positions (but not in ‘exo,endo’ positions), and the plane of the substituted ring perpendicular to the Y plane.In this conformation the plane of the substituted ring bisects the Y framework and forms a plane of reflection between the two benzylidene groups. Nevertheless, this conformation is not feasible, because placing the substituted ring perpendicular to the Y plane turns its benzyl p-orbital (C-29) also perpendicular to the other p-orbitals (C-2), thus breaking its conjugation with the rest of the p-system.Such an extreme situation requires that the substituted benzyl group should behave totally differently from the other benzyl groups. The similarity of the DG‡ calculated values as well as the even charge distribution negate this option. Another phenomenon to be taken into account is the hybridization of all the benzylic carbons (about sp2.5), which implies a strong pyramidalization of these carbons, thus making the two H-2 protons of dianion 2 non-equivalent in any possible conformation.MNDO calculations 6 on 1 and the X-ray structure of 1–2TMEDA (2Li1) 9 show that 1 should exist in a most stable propeller-like conformation (support for this conformation is found in the Fig. 1 Charge distribution and hybridization (in parentheses) of dianions 1 and 2 CH3 H3C –0.15 (2.24) (3.04) +0.01 –0.10 (2.30) +0.09 –0.35 (2.50) –0.17 (2.17) –0.09 (2.27) +0.10 –0.35 (2.51) +0.08 –0.17 (2.17) –0.09 (2.24) –0.35 (2.51) +0.10 +0.08 1 2 H H H H H H &ndas0.04 (2.31) –0.04 (2.46) 2– 2– dynamic behaviour of the methyl groups of 2).Assuming that 2 does not vary much (as is the case with the other properties), this conformation should distinguish one H-2 proton from the other. Since this has not been found even at the lowest temperature in spite of the hybridization measured, MNDO and X-ray findings, it is inevitable that the rotation about the Y bonds is fast in the NMR timescale.This reasoning leads us to the conclusion that the Y bonds are weaker than the benzylic bonds.|| This conclusion supports the notion that Y shaped dianions do not obtain some kind of through-the-center delocalization as a means of stabilization, but they rather tend to distribute the extra charge to the ‘corners’, thus minimizing the electrostatic repulsions between the three lone pairs. Although we are aware of the steric and electronic effects of the phenyl rings, we believe that the data reported here are experimental support to the theoretical considerations mentioned above, which claim the Coulombic interactions to be the governing factor, and they may also point to the ability of TMM22 and its derivatives to delocalize the extra charge in a most efficient way, which seems to be the origin of the remarkable stability of trimethylenemethane dianion (TMM22).References 1 J. Klein and A. Medlik, Chem. Commun., 1973, 275. 2 P. Gund, J. Chem. Educ., 1972, 49, 100. 3 J. Klein, A. Medlik-Balan, A. Y. Meyer and M. Chorev, Tetrahedron, 1976, 32, 1839; J. Klein, Tetrahedron, 1983, 39, 2733; 1988, 44, 503. 4 K. B. Wiberg, J. Am. Chem. Soc., 1990, 112, 4177. 5 (a) A. Gobbi, P. J. MacDougall and G. Frenking, Angew. Chem., Int. Ed. Engl., 1991, 30, 1001; (b) A. Skancke, J. Phys. Chem., 1994, 98, 5234, and references cited therein. 6 D. Wilhelm, T. Clark, P. v. R. Schleyer, K. Buckl and G. Boche, Chem. Ber., 1983, 116, 1669. 7 G. Boche, K. Buckl, D. Martens and D. R. Schneider, Liebigs Ann. Chem., 1980, 1135. 8 H. O’Brien, A. J. Hart and C. R. Russell, J. Am. Chem. Soc., 1975, 97, 4410. 9 D. Wilhelm, H. Dietrich, T. Clark, W. Mahdi, A. J. Kos and P. v. R. Schleyer, J. Am. Chem. Soc., 1984, 106, 7279. Paper 7/04020B Received 9th June 1997 Accepted 1st October 1997 || The X-ray data of 1–2TMEDA (2Li1) 8 show two Y bonds which are indeed longer than their corresponding benzylic bonds (1.443, 1.460 Å compared to 1.441, 1.434 Å respectively), and a shorter third Y bond.

ISSN:1472-779X    

DOI:10.1039/a704020b

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

BRUKER LECTURE J. Chem. Soc., Perkin Trans. 2, 1997 2465 Physical chemistry through electron spin polarisation. The Bruker lecture † K. A. McLauchlan Physical and Theoretical Chemistry Laboratory, South Parks Road, Oxford, UK OX1 3QZ New methods are introduced for studying novel physical chemistry through observations of spinpolarized radicals. This includes the photophysics of molecules in solution, including the anisotropy of the rates of inter-system crossing between excited singlet state molecules and excited triplet state sublevels, and measurement of the re-encounter probabilities of free radicals during the geminate period of their reaction.A simple and quite general method is described for measuring the absolute magnitudes of the spin polarization in both triplet states and radicals in solution. Introduction The first successful EPR observations of transient free radicals produced by flash photolysis 1,2 and pulse radiolysis 3 yielded spectra of quite unexpected appearance.The intensities, and sometimes the phases, of the lines were not those familiar from observations on stable radicals, although the line positions were unaffected and still allowed radical identification. Spectra from hydrogen and deuterium atoms observed in steady state concentrations under continuous radiolysis of solutions had shown similar anomalies some years before,4 but the real-time nature of the new experiments on transient radicals observed as they were formed and decayed enabled the true nature of the phenomenon to be deduced.In particular, the line intensities observed after instantaneous radical formation decayed in time at too fast a rate to be compatible with the re-combination of radicals in bimolecular processes under normal diffusion control. It was soon apparent that this was due to rapid population changes in an ensemble of radicals in which the initial populations of the hyperfine states were not those expected from systems at thermal equilibrium with their surroundings.This physical process would simply be spin-lattice relaxation did it not happen in the continuous presence of a microwave field which helps drive it.5 The ensemble was said to be ‘spin-polarized’ when first observed, and it appeared that the polarization arose in the chemistry of the system, a phenomenon called chemically induced dynamic electron polarization, CIDEP. Confusion was caused by the early flash photolysis experiments yielding spectra with all their lines in a single phase, either absorption or emission according to the molecule irradiated, whereas the radiolysis experiments displayed spectra with equal intensities of lines in absorption (A) and emission (E), with no overall spin polarization.In the former, the relative intensities of the lines were exactly as would be expected from the degeneracies of the hyperfine states, but in the latter unfamiliar intensities were observed in lines of opposite phase exactly distributed in total intensity between the low- and high- field parts of the spectrum.It was subsequently realised 6 that two separate and independent mechanisms generating polarization were in action, with different implications to the spectral appearance; more are now known, but are much less common.7 The spectra of mixed phases were observed from neutral radicals produced in pairs (‘radical pairs’), for example in simple bond-scission or atom-abstraction reactions.The polarization originated in magnetic interactions within the radical pair in what has become known as the radical pair mechanism (RPM, † Presented at the 30th International Meeting of the Electron Spin Resonance Group of the RSC, University of Lancaster, 6th–10th April 1997. see below). Since radicals are always created in pairs it was not obvious why this type of spectrum was not observed in all observations of transient radicals, for example in the flashphotolysis experiments. It happened that the early experiments of this type all involved radical ions, and not neutral radicals, and radical ions may undergo very rapid degenerate electron exchange reactions with their parent molecule.If these are suf- ficiently fast they average out any difference in phase or intensity between the low-and high-field halves of the spectrum, leaving only the net contribution from the second polarization mechanism.8 When flash photolysis was used to create neutral radicals, their spectra showed the combined effects of the two independent polarization processes, and could be reproduced by adding the biphasic contribution from the RPM to the single phase one with undistorted hyperfine intensities from the second mechanism.This caused the total signal intensity in one or the other phase to exceed that of the other. Spectra of this type were never, however, observed from the pulse radiolysis experiments, and it became apparent that the net signal arose in a process associated with the photophysics and photochemistry of the parent molecule.It is known as the triplet mechanism, TM.9 Analysis of CIDEP results usually involves interpretation of the spectra to identify the chemical natures of the radicals formed, to use the analysis outlined above to determine the origins of the spin polarization, and to use the phase of the signals to deduce the spin multiplicity of the precursor molecule which reacted to form radicals.It has provided a previously non-existent direct link between the photophysics of a molecule and the photochemistry which leads to an identified radical product. It has been used, for example, to demonstrate the predominant formation of a single geometric isomer of a radical from reaction of a singlet state, and another isomer from the triplet state of the same molecule, information which could be obtained in no other way.10 The time-dependence of CIDEP signals has provided values of reaction rate constants 11–14 and of electron15 and triplet 16 relaxation times in solution, but CIDEP spectra contain other information of fundamental interest within physical chemistry which has not yet been extracted.Although the RPM and TM have been understood for sometime, and detailed theory given, this theory has not been exploited to yield the properties of the triplet states or the intimate details of radical reactions.This we do here, but we must first summarise the polarization mechanisms themselves in sufficient detail for what follows. The triplet mechanism Those photochemical reactions which yield radicals whose spectra exhibit net spin polarization occur via reaction of the2466 J. Chem. Soc., Perkin Trans. 2, 1997 excited triplet states of molecules which are formed by intersystem crossing (ISC) from the excited singlet states which result from light absorption by the singlet ground state with conservation of spin multiplicity.This photophysical process commonly precedes the photochemistry of molecules. The triplet possesses three sub-levels whose energies are non-degenerate even in zero applied magnetic field, as a result of the dipolar (¡®zero-field¡�) interactions between the two unpaired electrons it contains. ISC occurs within the molecule at different rates into each, a fact known from solid state observations years before the first CIDEP experiments were performed17. This implies that at the instant following ISC the populations of the triplet sub-levels differ from those of a thermally-equilibrated triplet, and the ensemble of triplets is spin-polarized.In small molecules ISC is often dominated by spin-orbit coupling which allows change of the spin angular momentum of the molecule at the expense of orbital angular momentum, whilst conserving the whole. ISC links the singlet to a triplet sub-level with the same overall (orbital ¡¿ spin) symmetry, creating spin polarization because the spin symmetries of the three sub-levels differ.The ensemble of triplets is polarized in the molecular frame of reference, but the eventual EPR experiment is performed inside the external magnetic field of the EPR spectrometer in the laboratory frame. Rapid relaxation in the triplet normally causes line-broadening and precludes its direct obrvation by EPR (a few exceptions are known18,19), and we are reduced to detecting its polarization indirectly in the radicals produced on its reaction through the CIDEP phenomenon.We have to consider, therefore, whether polarization established in the triplet in the molecular frame can become apparent in the same triplet in the laboratory frame. This has been discussed explicitly,20,21 and is included in the theoretical treatments of the TM.22,23 At first sight it is not obvious that the polarization can be transferred between the reference frames, since this involves a correlation between the molecular energy levels and the Zeeman levels of the triplet at high field. It has long been known that in a fixed and isolated molecule in the solid state, the relative energies of the triplet sub-levels change as the magnetic field is applied along each of the major axes of the dipole coupling tensor of the molecule.A state which lies above the mean energy in one field-direction may lie below it in another.Although symmetryselective ISC continues to populate one specific level, this may lie above the barycentre of the system in one field direction, and below it in the other, leading to an emissively-polarized and an absorptively-polarized triplet in turn. In solution, the triplet molecule tumbles freely with respect to the field of the spectrometer, and this tumbling would be expected to average the population differences in the laboratory frame to zero. This is exactly true if the field is sufficiently high as to make the zero- field interaction negligible with respect to the Zeeman one, but at all normal operating fields of EPR spectrometers this is not the situation, and perturbation theory carried to second order shows that some polarization is transferred between the reference frames.20,21 Selective state population is, however, less specific than in the molecular frame (itself not wholly specific), each molecular frame state correlating to different extents with more than one Zeeman state.The sense of the polarization is preserved, at X-band for instance, but the magnitude depends on the competition between the rates of molecular tumbling and crossing between levels, which causes it to vary with the rotational correlation time of the triplet. The magnitude of the polarization in the triplet ensemble before it reacts to form radicals therefore depends both on the nature of the molecules and their environment.This aspect of the TM has received little attention. The effect of molecular rotation is summarised in the equation for the magnitude of the polarization in the triplet 22 (defined in an analogous way to that in the free radicals eventually formed, see below), taken to be cylindrically symmetric, [eqn. (1)]. PT = 24DKw0 15 I I O 4 4w2 0 1 [kT 1 tr 21]2 1 1 w2 0 1 [kT 1 tr 21]2¢© ¢� ¢­ (1) Here D is the zero field coupling constant, K the dimensionless anisotropy of the ISC rates into the states perpendicular and parallel to the major axis of the molecule (assumed parallel to the D-tensor axis) defined by eqn.(2), where k is the total K = (k^ 2 k|| )/k (2) value, w0 is the operating frequency of the spectrometer, kT is the reaction rate constant of the triplet molecule (see below), and tr is its rotational correlation time. For a given triplet system the size of the polarization therefore varies as the viscosity solution is changed, through its effect on tr, and sometimes kT, in a predictable way.It remains to consider the production of the polarized radicals which are the species observed. They are formed in pairs by reaction of the laboratory frame-polarized triplets with conservation of electron spin orientation. Thus, if the triplets are over-populated in their upper Zeeman (aa) state, the radicals are formed predominantly in their upper, a, state, and the EPR spectrum observed immediately after radical formation is in emission (Fig. 1). The inverse overpopulation would lead to absorptively polarized radicals; in either case the spectra of the two radicals produced exhibit identical polarizations. But the polarization of the triplets rapidly vanishes as spin-lattice relaxation occurs, usually in a few nanoseconds. In consequence the rate of reaction of the polarized triplet molecules must compete with their rate of spin-lattice relaxation if polarized radicals are to result, implying that the reactions must be very fast.TM polarization consequently only results from unimolecular bond-scission reactions and bimolecular electron and proton transfer ones at high substrate concentrations. Relaxation in the radicals is comparatively slow (with T1 ~ 1 ms typically), and can be neglected during the polarization transfer process. A simple consideration of the competitive kinetics then gives an expression for the spin polarization PR observed in the radicals which is defined (positively for a thermally-equilibrated system) as the ratio of the difference in populations of its electron spin states to their sum (the total number of radicals present), [eqn. (3)].PR = PTkd[M] 1 PT eqTT1 21 kd[M] 1 TT1 21 (3) Here PT eq is the polarization in the thermally equilibrated triplet, kd the diffusion-controlled rate constant of the triplet molecule with a substrate, M, and TT1 is the spin-lattice relaxation time of the triplet. This equation has been used to extract values of TT1 from independent measurement of the rate constant (see below), although it has not been directly verified.It has recently acquired a new significance. A third polarization mechanism, the radical.triplet pair mechanism (RTPM) can occur in solutions in which radicals and triplets co-exist.24 It yields polarized spectra which may exhibit both net and hyperfine-dependent biphasic contributions, the same qualitative characteristics as would radicals polarized along the more familiar TM 1 RPM route.The net contribution can be tested for TM origin by verifying whether the observed polarization varies with substrate in a manner consistent with eqn. (2). The radical pair mechanism Whereas the TM arises in the molecule which reacts to form radicals, the RPM results from interactions between the radicals once they are formed; it therefore originates later in time, although usually before observations commence even using fast detection methods (typically 30 ns post flash).This difference in time-origins is exploited below.J. Chem. Soc., Perkin Trans. 2, 1997 2467 Reaction or dissociation of a molecule normally occurs with conservation of spin multiplicity. Since the parent molecule exists in a pure spin state (usually singlet in ground-state chemistry and triplet in photochemistry) this leads at the instant of its formation to a radical pair in that same overall spin state. We shall discuss the spin state of the pair of radicals in the coupled representation in which we ask not what the electron spin state of an individual radical is at a given time after radical pair formation, but rather what the total spin of the pair is.This is even so when the radicals have separated by diffusion and do not interact. This is convenient for discussing radical pair effects in general, for re-combination of radicals, for example, is controlled by a strict spin selection rule: normally electrons must have antiparallel spins to form a bond, implying recombination through the singlet state of the pair. Reaction, however, plays no role in the generation of CIDEP in radicals produced together (‘geminate radicals’) although it does in that which arises in the encounters of freely-diffusing radicals (‘Fpairs’); we are not concerned with these here.When the pair is formed the radicals are not usually in contact, so that they do not immediately react (even if their electron spins are antiparallel) but diffuse apart (Fig. 2). The geminate pair persists for short periods (usually less than 30 ns, depending upon the solution) and it is diffusion on this timescale we are involved with, not the longer-time diffusion normally considered to control reactions of radicals in solution. Although the pair is created in a pure spin quantum state this is not an eigstate of the system and so the wave-function evolves in time under the influence of the spin hamiltonian. This can conveniently be separated into a magnetic part, HM, which contains the Zeeman and hyperfine terms familiar to the EPR spectroscopist, and an exchange part, HJ, representing the (electrostatic) exchange interaction, J(r), between the two spins.After radical formation, the latter is initially completely dominant and it causes the singlet (S) and triplet (T) radical pair states to differ so much in energy that their wave functions are not mixed by the magnetic interactions.However it falls rapidly in magnitude as the radicals drift apart, being of extremely short range, and allows an inexact but convenient simple model in which the two interactions are envisaged to operate over different periods of time.25 In neutral radical pairs J(r) is negative and S underlies the T states in energy. As the pair of radicals separates and J(r) tends towards zero, however, the Zeeman splittings between the triplet sub-levels, T0 and T±1 of the radical pair become significant.Fig. 1 The final stages of the production of spin-polarized radicals via the triplet mechanism. Anisotropic ISC within the molecule following light absorption leads to an ensemble of triplet molecules, here shown with excess population in the aa state, which rapidly relaxes to thermal equilibrium. However rapid reaction with spin conservation competes with the relaxation to form a spin-polarized radical sub-ensemble also with excess a spin.A further sub-ensemble of radicals, now with thermal equilibrium populations, is formed by reaction of the thermally equilibrated triplet. The EPR observation is of the whole ensemble, and therefore the spectra appear in emission, and calculation of the magnitude of the spin polarization in it in terms of the initial polarization in the triplet ensemble yields eqn. (3) of the text.32 Relaxation of the radicals is assumed sufficiently slow to be neglected when calculating the initial polarization in the radicals. These cause the T11 state to differ in energy from the singlet, S, state by an amount which prevents spin mixing through the magnetic interactions. The T21 state then must cross the S one as the radicals separate during diffusion, but it does this so rapidly that any spin mixing is negligible; exceptions occur if the diffusion is slowed in a restricted or unusually viscous medium, or if the hyperfine couplings are unusually large.The implication is that with neutral radicals in solutions of normal low viscosity it is only the T0 state of the radical pair which can mix with the S one, and this happens when the radicals are separated.The wave functions of each contain equal admixtures of the a and b spins on each radical [eqn. (4)] and |SÒ, |T0Ò = 1/÷2(ab 1 2 ba) (4) mixing cannot change this. ST0 mixing does not produce spin polarization in the system but rather causes the eigenstate of the pair to change continuously in time from its initial pure condition.Polarization of a sort does, however, arise if the radicals then re-encounter during their diffusive excursion. Its origin lies in the action of the exchange interaction, switched back on again as the distance between the radicals decreases. Its generation is a pure quantum phenomenon for which no simple physical model exists, the exchange interaction introducing equal and opposite phase shifts in the S and T0 contributions to the mixed wavefunctions, which leads in turn to non-zero spin polarizations in the sub-ensembles of each type of radical.26 If the two radicals differ in chemical type more a electron spin accrues to one, and more b to the other, with the overall polarization necessarily zero; the RPM ST0 mechanism sorts the spins into different hyperfine states on different radicals rather than produces an absolute polarization.If the two radicals are identical the spectrum exhibits two halves of opposite but equal inten- Fig. 2 The molecular dynamics associated with the generation of ST0 radical pair mechanism spin polarization. In this diagram we assume that the radicals are created as a pair following the reaction of a triplet state of a molecule. Reaction occurs with spin conservation so that when the radicals are first formed their spins are also triplet-correlated. They therefore cannot react and drift apart, some to leave the geminate cage for ever, never to re-encounter their geminate partner.Others do, however, re-encounter during their random diffusion within the cage, and when they do the radical pair exists in a mixed quantum state in which the radicals recombine with a probability proportional to the singlet character of the radical pair. Some therefore survive the collision and escape the geminate region. It is these radicals, denoted by asterisks, which constitute the sub-ensemble which is observed to exhibit RPM spin-polarization in an experiment. If the triplet state is polarized when it reacts, then the whole ensemble of radicals produced on its reaction may itself be polarized via the triplet mechanism before the RPM process commences.A certain fraction of the radicals only undergoes the later re-encounter process in which the RPM polarization is generated and the ratio of the contributions to the observed spectrum from the two mechanisms depends upon this fraction. This is the simple principle on which the estimation of the geminate re-encounter probability described later in the text depends.R1 R T 2 R2 R2 R1 R1 R1 R T/S 2 T Diffusion Geminate Cage * *2468 J. Chem. Soc., Perkin Trans. 2, 1997 sity, for neutral radicals with the low-field half in emission and the high-field half in absorption, an E/A pattern, if the molecular precursor was a triplet. With chemically-different radicals, one exhibits an E*/A pattern, and the other an E/A* one, where the asterisk denotes an excess of signal in that phase.The degree of spin-sorting depends upon the amount of spin mixing which has occurred prior to the encounter, and in the ensemble radicals encounter at different times after their creation during random diffusion. Each subset of encounters at a specific time contributes differently to the polarization than do those at different times, and the effect must be integrated over the distribution of encounter times. Independent of the model taken for diffusion (all predict the same long-term dependence relevant to polarization development) the result is given by eqns.(5) and (6), where radical (1) is in the overall nuclear spin PR µ |·S |HM|T0Ò|1/2 (5) ·S |HM|T0Ò = 1/2[g1 2 g2)mBB 1 on a1nm1n (a) 2 om a2mm(b) 2m] (6) state (a) and radical (2) in (b), and the other symbols have their usual meanings. The implication is that each different hyperfine line has an unique polarization which depends on its nuclear spin magnetic quantum number.The square root is the result of the diffusional averaging, and does not appear in a static model. These equations suffice to calculate the relative intensities of the lines due to the RPM mechanism for the purposes of this paper, although a fuller treatment 27 is needed to introduce the sign information which allows us to predict that lines corresponding to different signs of m (giving lines on opposite sides of the spectrum centre) exhibit opposite phases (assuming little difference in g-factors), as observed.Finally it must be remembered that in the whole ensemble radicals in a given hyperfine state encounter counter-radicals in all possible hyperfine states, and each encounter gives rise to a different contribution to the polarization, as is evident from the above equation. The polarization must consequently be summed over all possible encounters, with due attention paid to the degeneracies of the hyperfine states of both radicals.Although the origin of RPM polarization lies in the quantum world, we may summarise the physical picture of the processes necessary to its creation, from which novel physical chemistry can be extracted. The model we have is of spinmixing and of molecular diffusion to bring the radicals back together after an initial separation, within the geminate period of the reaction. The diffusion has a profound effect on the relative intensities of the lines, which are fully calculable, and CIDEP spectra are potential sources of information about molecular diffusion within the geminate cage.Absolute calculations of polarization magnitudes remain difficult due to uncertainties in precise modelling of both the diffusion and the exchange interaction processes, but the strategy introduced in this paper circumvents this to yield unique information. Experimental All experiments have been performed using the continuous wave flash photolysis EPR technique with digital field advance invented in this laboratory,28 but running in the mode in which all the information following each photolysis flash is stored and output as a three dimensional surface of signal plotted against field and time.Spectra over specific periods after the photolysis flash were extracted from the data off-line using the ‘time integration spectroscopy (TIS)’ technique.29 In the experiments reported here our interest was in the polarization patterns observed immediately after the flash and over a narrow timewindow, so as to be sure that the contributions to the observed polarization arose from interactions in the original pair of radicals created, with no contribution from later F-pair events.Inspection of the spectrum at a series of times post flash ensured this, the F-pair contributions becoming apparent at later times in experiments producing approximately 1025 M radical per flash from an excimer laser operating at 308 nm with feed-back control of the average pulse intensity. Having selected appropriate sample periods, the spectra were re-run using online TIS with sufficient field-sampling points to ensure correct and reproducible intensity information; using digital sampling systematic errors accrue if too few are used.All chemicals were used as supplied, with experiments conducted on flowing solutions at room temperature. Applications to physical chemistry Calculations of the relative intensities of the contributions from the TM and the RPM to a spectrum are straightforward, and spectra are usually reproduced by adding the two in empirically-adjusted proportions to reproduce the observations.This approach is necessitated by many of the parameters required to calculate the absolute magnitude of each being unknown, whilst the magnitude of the observed polarization can only be measured with difficulty.30,31 In this paper are described methods for overcoming these limitations, so as to abstract physical chemical information directly from the spectra.The first involves the verification and application of TM theory. Investigation of the photophysics of molecules in solution 32 The largely untested eqns. (1) and (2) for TM polarization offer the opportunity for determining the polarization in the triplet state itself before it reacts to form radicals, and through this to investigate the dynamics of the ISC process in solution.We introduce a new strategy for obtaining the absolute value of the polarization in the triplet state and in the radicals. Rather little firm evidence exists for the operation of the TM in fluid solution, since the observation of a single-phase contribution to polarization without hyperfine intensity distortion is, with the advent of the RTPM, only consistent with the mechanism rather than diagnostic for it. A second object of our studies is to Fig. 3 Irradiation of tetramethylpyrazine in the presence of 2,6-ditert- butylphenol produces the radical product of H-addition to the pyrazine, with a broad spectrum, and the phenoxyl radical, with a spectrum consisting of a doublet of triplets.The spectra shown are displayed with the amplitudes of the signals normalized, although their absolute intensities which are used in the calculations vary substantially. The series is obtained at various phenol concentrations in benzene solution (i) 4 × 1023 M, (ii) 0.024 M, (iii), 0.119 M, (iv) 0.34 M, (v) 0.485 M and (vi) 1.212 M.At low concentrations there is a strong biphasic contribution to the signal from the RPM mechanism, although one phase preponderates over the other, due to a TM contribution. As the concentration increases this becomes greater and the spectrum swings entirely into emission, although with some RPM distortion even at the highest concentration used. Since the RPM does not produce an absolute spin polarization, the integral across the whole spectrum yields a value which is proportional to the TM contribution alone.It is the variation of this with changing concentration which is used in the analysis, the integration process also improving the signal-to-noise ratio of the measurements.J. Chem. Soc., Perkin Trans. 2, 1997 2469 confirm its operation in a sample of historical importance. Some years ago we used group theoretical arguments to predict that the phase of signal produced through the TM should vary between radicals produced by reaction of two closely related molecules with identical chromophores and similar chemistry, pyrazine and quinoxaline; 33 the opposite phases of the signals in the two cases appeared to provide incontrovertible evidence for the TM, but the approach depended upon assumption of the orbital symmetries of the molecules in their excited states.We have consequently studied the reaction of excited tetramethylpyrazine (selected to yield strong signals, whilst not changing the symmetry 34) with 2,6-di-tert-butylphenol (DTBP) to yield the radical formed by H-addition to the pyrazine, and the phenoxyl radical.DTBP has the advantage of being extremely soluble in the benzene solvent used, whilst the phenoxyl radical spectrum consists of a sharp doublet of triplets which enable the polarization contributions to be identified simply. In this study, however, this is immaterial since, although TM and RPM polarization both occur (Fig. 3) we are interested only in the size of the net signal, which is obtained by integration of the whole spectrum, a procedure which eliminates any ST0 RPM contribution. As with thermally-equilibrated radicals, the signal is proportional through an apparatus constant (c) to the concentration of the radicals times the magnitude of the spin-polarization: S = cPR[R] (7) It follows that if the optical density of the solution is maintained constant (0.2 mol dm23 methyl pyrazine in benzene), that a constant intensity of laser light is used, and that the triplet concentration is so low (ca. 1025 M per flash) that all the triplets formed react to create radicals even at the lowest phenol concentration used, then the radical concentration remains unchanged and S ¥ì PR. The experiment then consists of varying the concentration of the phenol [M in eqn. (3)], and observing the change in S. In Fig. 4 is shown a plot of the integrated signal intensity as a Fig. 4 A plot of integrated signal intensity, that is the signal due to the TM contribution alone, versus phenol concentration in benzene solution. As predicted from eqn. (3), at low concentrations this exhibits a linear dependence, from the slope of which the parameter (kd/TT1 21) can be obtained, whilst at high concentrations a plateau is observed. Working at constant optical density, and constant laser intensity, the ratio of the plateau value to the intercept on the zero concentration axis gives the polarization in the triplet molecule in terms of the equilibrium polarization, and this can be made absolute by using the Maxwell.Boltzmann distribution to calculate the latter. The absolute polarization in the radical can also be deduced (see text). In this introductory paper a full fit of the observations to eqn. (3) is not attempted since several parameters within it are viscositydependent, and the viscosity of the solution increases as the phenol concentration is increased. function of phenol concentration in benzene solution.Its general shape is consistent with eqn. (3), which confirms the TM to be the origin of the net signal. So long as the phenol concentration is sufficiently low to make kd[M] ! TT1 21, then PR and S are linearly dependent on phenol concentration. As [M] is increased, however, PR eventually becomes independent of it and the curve approaches a plateau value. The concentration at which this is attained is controlled by the magnitudes of kd and TT1 21 which depend upon the viscosity of the solution through the translational diffusion coefficient, and the rotational correlation time, respectively.In consequence, the plateau is attained at lower concentrations in a low-viscosity solvent, such as benzene, than it is in a higher viscosity one, such as octan-2-ol, although the effect is not great. This, and other differences in behaviour in the two solvents will be reported in detail in a forthcoming paper.Extrapolation of the lower part of the curve to zero phenol concentration gives an intercept on the signal axis which is directly proportional to the thermal equilibrium polarization in the triplet state. In the high concentration region kd[M] @ TT1 21, and provided that PTkd[M] @ PT eq TT1 21, the asymptotic signal is directly proportional to PT. This is automatically the case under the first inequality, since PT > PT eq. The ratio of this signal to that at the low-concentration intercept, with [R] constant, therefore gives the polarization in the triplet when it reacts in terms of the polarization in the relaxed triplet.From the curve, we obtain a value of PT = 42 ¡¾ 1PT eq in the most viscous solution (PT varies with viscosity as discussed above). We stress that this is the polarization in the (unobservable) triplet, rather than that in the radicals themselves, although this (PR) can also be obtained at any concentration of phenol using the intercept as a calibration, and remembering to correct for the difference between the radical and triplet Boltzmann equilibrium values.Less directly, but interestingly from a theoretical point of view, eqn. (8) can be obtained from eqn. (2) with the aid of the equa- TT1 21 = 3 2 I I O ge 2m0m2B 4p"d3¢© ¢� ¢­ 2 tr (8) tions valid at low viscosity, where m0 is the permeability of free space, mB the Bohr magneton, d is the distance between the electrons and tr is the rotational correlation time given by eqn.(9), where a is the molecular radius and kB the Boltzman constant. tr = (4pha3)/(3kBT) (9) The experiment provides an accurate and rather simple method for measuring PR, independent of measurement of the concentration of radicals present, which can be made absolute (as can the PT value) by calculating the magnitude of the polarization in the thermally-equilibrated triplet in a given magnetic field, using the Maxwell.Boltzmann distribution to calculate the populations of the levels.By this route an absolute value for the electron spin polarization due to the TM is obtained. Since with neutral radicals formed from triplet precursors the observed polarization usually results from both TM and RPM mechanisms, and the observed spectrum may be reproduced by adding empirically-adjusted contributions from both, the absolute value of the RPM polarization can in turn be measured.This provides a general method of wide application, and provides an opportunity for studying the polarization processes, and the dynamic processes involved in them, in detail. Knowing the value of PT, fitting of eqn. (2) to the experimental curve appears to involve the single parameter (kd/TT1 21), from which a value of either kd or TT1 can be obtained if the other is measured independently, or calculated. In practice, a difficulty is encountered. The addition of large quantities of the phenol significantly affects the viscosity of the solution, and this implies, through eqn.(1), that the value of PT is not constant. This will be discussed elsewhere.2470 J. Chem. Soc., Perkin Trans. 2, 1997 Direct measurement of the polarization in the triplet precursor to the radicals allows this species itself to be investigated. Referring to eqn. (1), the absolute value of PT is now measured, whilst the motional term in brackets can be evaluated from estimates of the reaction rate constant of the triplet, kT, and of its rotational correlation time.We take kT to be the pseudo-first order rate constant kd[M], and calculate this using the Stokes– Einstein relation and a value of [M] in the plateau region of the curve. Here, as in calculating tr, it is assumed that the bulk viscosity may be used. Having calculated the motional term, and using the measured value of PT, the anisotropy in the rates of ISC, K, can be obtained from eqn.(1) if the zero-field coupling constant D is known, or conversely. This provides an unique method for measuring these quantities in solution, and it will be interesting to investigate how they vary with the phase of the sample. D is often known from solid state measurements, and K less frequently so. For tetramethylpyrazine D has been reported to be 0.0963 cm21 35 and 0.099 cm21 36 in different crystal hosts, yielding a value of K of ~0.63. In the solid state observations in host crystals of durene at low temperatures yielded a value of ~0.38,37 but this medium is known to cause the molecular axis system to differ from that of the zerofield coupling tensor, and it is not apparent whether the figures are directly comparable.Nevertheless this first solution-phase value seems not unreasonable. Radical re-encounter probabilities in solution 38 As explained above, radicals are created in pairs with conservation of spin angular momentum. If it is a triplet state which reacts to form them then the spins of the radicals are themselves triplet-correlated, and the radicals cannot react immediately after they are formed, but must separate.If a singlet radical pair is formed the energy of the reaction tends to separate the radicals, and they also do not recombine immediately. In both cases a ‘spin correlated radical pair (SCRP)’, has been formed, and in both cases the initial production of radicals is followed by molecular diffusion through the solution. During this period the spin wave function of the pair evolves and becomes mixed, and if the geminately-created radicals re-encounter at a later time then they react with a probability which depends upon the singlet character of the pair, reaction normally occurring through the singlet state according to the strict selection rule for bond formation.This short-term diffusive process, together with the spin mixing, therefore determines the probability of product formation in the geminate cage.It also controls the magnitudes of all the phenomena (not all of which involve reaction) now known collectively as spin chemistry, including chemically-induced dynamic electron and nuclear polarization (CIDEP and CIDNP), reaction yield detected magnetic resonance (RYDMR),39 stimulated nuclear polarization (SNP),40 and magnetic field effects in chemistry (MFE or magnetic effects on reaction yields, MARY).41,42 It is therefore of interest to measure the fraction of the free radicals created together that subsequently re-encounter within the geminate period of the reaction, and how this varies with experimental conditions.This has received theoretical attention 43,44 but has proved difficult to investigate experimentally. We demonstrate here how this may be done using CIDEP phenomena. The principle of the experiment lies in TM polarization existing in the ensemble when the radicals are created, whereas RPM polarization arises later in time as a result of magnetic interactions within the SCRP and radical re-encounter.Not all the radicals which are produced together do, however, re-encounter since during their random motion in solution some simply diffuse apart for ever. This means that a smaller sub-ensemble is involved in the RPM polarization-generation process than in the TM one. The size of this sub-ensemble, determined by the fraction of the original radicals which re-encounter, depends upon the viscosity of the solution, and we wish to monitor it as this is changed.This can be done without measuring the absolute value of the polarization (i.e. without using the method described above) by using the polarization from the TM as an internal standard. Our requirents for production of TM polarization in the radicals observed are now quite different from above. We need to produce radicals by direct, and very fast, bond scission so that to a good approximation the magnitude of the TM polarization is independent of the viscosity of the medium (although we correct for the small change that does occur).This requires an unimolecular decomposition which competes in timescale with the rotation of the triplet so as to trap a constant fraction of the molecular frame polarization in the laboratory frame. It is a simple matter to choose a suitable system experimentally because possible competing atom abstraction reactions of the triplet are bimolecular. These may, however, be very fast if the triplet can react with the solvent.A simple test is to observe the spectrum and see whether only those radicals formed by bond scission are present {in the case where reaction of the triplet with the solvent is possible, this immediately allows a lower limit on kT [eqn. (1)] to be deduced}. In this study we have used photolysis of 1,3-dihydroxypropanone which fulfils this condition in all the solvents investigated to produce ?CH2OH and ?COCH2OH radicals, and has been the subject of two detailed CIDEP studies so that its chemistry and its polarization and relaxation behaviour are well established.45,15 Fig. 5 (a) Observed (above) and calculated spectra from the ?CH2OH and ?COCH2OH radicals produced on photolysis of 1,3- dihydroxypropanone in ethandiol solution (i) 0.14–0.26 ms post flash and (ii) 0.26–0.37 ms post flash. The two spectra are reproduced using the same ratio of contributions from the TM and RPM mechanisms, where the RPM contribution is calculated for the geminate pair of radicals; this shows the absence of F-pair contributions over the two sampling periods and confirms the identities of the radicals which must be used to calculate the RPM contribution.The later time spectrum in particular shows the effects of Torrey oscillations, not wholly eliminated by the TIS method using the narrow sample window selected for these measurements,29 whilst the earlier one exhibits the familiar linebroadening inherent in the continuous wave technique.The relaxation times needed to calculate the lineshapes, and the relative sizes of the signals from the radicals as time evolves, were taken from our previous studies.15 (b) The intensity patterns calculated from TM and RPM polarization for the hydroxymethyl radical. Addition of the two in the ratio a/b reproduces the observed spectral intensities. In the case shown the ratio has been set equal to one.The experiment consists in observing the changes in the spectrum as the viscosity of the solution is varied, and analysing each for this ratio.J. Chem. Soc., Perkin Trans. 2, 1997 2471 Fig. 6 (i) Direct experimental data showing the decrease in the a/b ratio as the viscosity in the solution is increased. This is due to a greater contribution from the RPM and reflects an increasing fraction of those radicals which are created together which eventually re-encounter within the geminate cage. The ratio tends towards an asymptotic value at high viscosity.(ii) The fraction, f, of re-encounters, obtained from the corrected data according to eqn. (10) of the text, plotted against viscosity. The experimental points fit well to a square root dependence [eqn. (12)] expected from a simple model of the liquid. The intercept on the f-axis gives the fraction of re-encounters in an infinitely mobile fluid, and is consistent with expectations from a random-walk model.It is interesting that the probability of a re-encounter increases by only a factor of 4–5 in going to an infinitely viscous medium. The observations consist of observing the changes in the spectrum as the viscosity is varied, in the experiments reported here by altering the solvent; some sample spectra are shown, together with their theoretical simulations, in Fig. 5(a). There are large variations in appearance as the viscosity is changed, but they can be reproduced by adding the contributions from TM and ST0 RPM sources alone.Although the whole spectrum can be reproduced, the short relaxation time of the acyl radical makes it more convenient to use the spectrum of the hydroxymethyl radical only in our analysis; its spectrum is also more sensitive to the ratio than is that from the acyl which has a lower RPM contribution. The TM contribution is in the absorptive phase from this molecule, and the relative intensities of the three doublets are 1:2:1.The relative intensities of the doublets expected from ST0 RPM polarization were calculated using the measured parameters for the radical pair (for ?CH2OH, g = 2.003 23, ACH = 1.739, AOH = 0.114 mT; for ?COCH2OH, g = 2.000 75, ACH = 0.152 mT); the two low-field doublets are predicted in emission and the high field one in absorption, due to the radicals originating in a triplet reaction, and having quite different g-values. The TM and RPM patterns are shown in stick spectrum form in Fig.5 (b), where they are added in proportions of a and b to reproduce a typical observed spectrum. Spectra were obtained using identical photolysis flash energies over the same period after the flash from 0.5 M solutions of 1,3-dihydroxypropanone in solutions of 10 different viscosities measured using a calibrated pyknometer.These consisted of methanol, ethanol, propan-2-ol, ethanediol, propane-1,2-diol and mixtures of ethanol with ethanediol and with propane-1,2- diol in various proportions. They were fitted to give the ratio a/b for each, and this ratio was plotted against the viscosity [Fig.6 (i)].As expected from theory, the RPM contribution becomes progressively comparatively greater as the viscosity is increased and a/b falls, tending to an asymptotic value of approximately 3.1 at high viscosity. Although the optical density of the solution and the flash energy were maintained constant, so that the absolute value of the concentration of radicals was the same in each solution, this is unimportant to the analysis. In any solution, we assume that a concentration of radicals [R] is produced with TM polarization, but that only a fraction ‘f ’ of these radicals re-encounter within the geminate cage after an initial diffusive separation.The RPM polarization therefore arises in a concentration of f [R] radicals. With an obvious extension of the nomenclature in eqn. (6) it follows that the ratio of the contributions to the observed spectra from the two mechanisms is given by eqn. (10).STM/SRPM = a/b = PTM/( fPRPM) (10) If the polarization ratios are known in each solution, f can be extracted quite straightforwardly. However both PTM [through eqns. (1) and (2) as discussed above] and PRPM (µ h� �� through the diffusion process) are viscosity-dependent, and the equation should be written as eqn. (11). (STM/SRPM)h = 1/f(PTM/PRPM)h (11) We now assume that the value of 3.1 observed at the highest viscosity used represents the true asymptotic value which, in the limit, corresponds to all the radicals which are created together re-encountering inside the geminate cage, i.e.to f = 1. This implies that at this viscosity the ratio of polarizations is 3.1 and we can use this as a datum point to correct the polarizations, and their ratios, observed in all the other solvents. To do this we need a value of the unimolecular decay constant of the triplet which we expect to be in the range 1010–12 s21.Using the mean value of 1011 s21, the magnitude of PTM is predicted to increase by a factor of only 1.3 between the least viscous, and most viscous solutions: as expected with a rapidly-dissociating triplet the correction (although made) is a minor one. The correction to PRPM for change in viscosity is, however, more significant. The corrected values of a/b then allow the fraction of reencounters in all the other solutions to be assessed from eqn.(10). There is an interesting feature to this. In our experiments we observe the effects only of ST0 RPM polarization, which through spin-mixing involves only one half of the total number of radicals present, and it would seem that the asymptolue of f for these should be 0.5. But this would not give the total number of radicals which re-encounter since there are silent collisions which involve that half of the radical pairs which are in the T±1 states and produce no such polarization. We therefore correct for this by putting f = 1.The fraction of radicals which re-encounter is plotted against the viscosity of the solution in Fig.6(i), where it is seen that it varies by a factor of four between non-viscous and very viscous solutions. Over the viscosity range in which random-walk theory is applicable the probability of re-encounter of a pair of radicals created together at a later time varies as the square root of the time between diffusive steps, and through this as the square root of the viscosity.The solid line in the figure is a least means squares fit of a curve of the form eqn. (12) to the f = ch1/2 1 d (12) observed points, where c and d are constants, and d represents2472 J. Chem. Soc., Perkin Trans. 2, 1997 the fraction of re-encounters in an infinitely mobile liquid in which there is un-restricted random walk. The fit is very satisfactory, although at some viscosity the relationship must break down since f cannot exceed unity; it is possible that the highest viscosity points should not be included in the fit, which would affect the value of d extracted.Furthermore this initial analysis has used the value of the signal ratio observed in the highest viscosity solution as the asymptotic value, and this will need further experiment to confirm. Under these circumstances the value of 0.24 obtained for d, whilst approximate, seems reasonable since a simple random walk calculation for a hexagonally close-packed liquid gives a value of 0.2.This adds confidence to the interpretation. Fig. 6(i) has many implications to spin chemistry and radical recombination chemistry, and it will be interesting to compare the re-encounter probability obtained in this way with the reaction probability in the geminate cage. If it is assumed that the probability of reaction at a re-encounter depends simply on the singlet character of the SCRP, then the ratio of the two probabilities gives this character, integrated over the distribution of re-encounter times of the radicals in the geminate cage.This is calculable assuming a theory of the liquid, and provides a possible route for confirming the assumption or, alternatively, for measuring the reaction probability in an encounter involving two radicals forming a pure S state of the SCRP. Interpretation of the parameter ‘c’ in terms of the model of the liquid used is also possible, but will await a more sophisticated treatment than that provided here for purposes of illustration.Another polarization experiment might give similar information, one in which ST21 polarization contributes to the observations. This does not depend on diffusion in the same manner as does ST0 polarization, but its magnitude depends on the different number of radicals which are polarized through this mechanism as the radicals initially diffuse apart to the number which are polarized by the same mechanism when they reencounter at any later time.46 Conclusion Whereas in the past observations of spin-polarized radicals have yielded information on their reaction and relaxation rates, the specific opportunity they offer to investigate basic physical processes in the liquid phase has not been exploited.Here it has been shown that they can be used to illuminate the photophysics of molecules in solution, on the one hand, and the reencounter probabilities of radicals inside the geminate cage, on the other.These are both novel measurements, seemingly only possible through polarization studies. But further possibilities exist which depend upon the absolute measurements of the spin polarization which arises in chemical and photochemical processes. Such measurements have proved difficult in the past, and few have been made so that the opportunities have largely gone un-exploited. Here a new and rather general simple method has been introduced for determining the absolute polarization in triplets and radicals in solution, which opens the possibility for further studies of the intimate details of reaction processes and of photochemistry and photophysics in solution. In this Bruker Lecture a broad account has been provided to introduce these new methods in a transparent fashion so as to establish the principles involved, and to indicate their application.Fuller and extended descriptions will appear elsewhere.Acknowledgements The experimental results published here are the work of Oscar Ces, Robert Eveson and Tariq Qureshi. I am grateful to them and, on receiving this Prize, to all my past and present coworkers. I am similarly grateful to Dr Dieter Schmalbein for his help on instrumentation towards the start of my interest in the study of transient radicals using EPR methods. References 1 P. W. Atkins, K. A. McLauchlan and A. F. Simpson, Nature, 1968, 219, 927. 2 P. W. Atkins, I. C. Buchanon, R. C. Gurd, K. A. McLauchlan and A. F. Simpson, J. Chem. Soc., Chem. Commun., 1970, 513. 3 B. Smaller, J. R. Remko and E. C. Avery, J. Chem. Phys., 1968, 48, 5174. 4 R. W. Fessenden and R. H. Schuler, J. Chem. Phys., 1963, 39, 2147. 5 P. J. Hore and K. A. McLauchlan, Mol. Phys., 1981, 42, 533. 6 A. J. Dobbs, Mol. Phys., 1975, 30, 1073. 7 K. A. McLauchlan and M. T. Yeung, Electron Spin Resonance, ed. N. M. Atherton, M. J. Davies and B.C. Gilbert, (Specialist Periodical Report), RSC, Cambridge, 1994, vol. 14, p. 32. 8 K. A. McLauchlan and D. G. Stephens, J. Chem. Phys., 1987, 87, 4399. 9 J. K. S. Wan, S. K. Wong and D. A. Hutchison, J. Chem. Phys., 1973, 58, 985. 10 K. A. McLauchlan and C. D. Buckley, Chem. Phys. Lett., 1989, 164, 571. 11 H. Paul, Chem. Phys., 1976, 15, 115. 12 T. Prisner, O. Dobbert, K.-P. Dinse and H. van Willigen, J. Am. Chem. Soc., 1989, 110, 1622. 13 K. A. McLauchlan, in Modern Pulsed and Continuous-Wave Electron Spin Resonance, ed. L. Kevan and M. K. Bowman, Wiley Interscience, New York, 1990, 285. 14 R. W. Fessenden, J. Phys. Chem., 1973, 58, 2489. 15 K. A. McLauchlan and M. T. Yeung, Mol. Phys., 1996, 89, 1423. 16 P. W. Atkins, A. J. Dobbs and K. A. McLauchlan, Chem. Phys. Lett., 1974, 29, 616. 17 M. Schwoerer and H. C. Wolff, Magnetic Resonance and Related Phenomena, Proceedings of the 14th Congress Ampére, 1967, 544. 18 G.-H. Goudsmit and H. Paul, Chem. Phys. Lett., 1993, 208, 73. 19 K. A. McLauchlan, I. A. Shkrob and M. T. Yeung, Chem. Phys. Lett., 1994, 217, 157. 20 P. J. Hore, C. J. Joslin and K. A. McLauchlan, Electron Spin Resonance, ed. P. B. Ayscough, (Specialist Periodical Reports), RSC, London, 1979, vol. 5, p. 1. 21 P. W. Atkins and K. A. McLauchlan, in Chemically Induced Magnetic Polarization, ed. A. R. Lepley and G. L. Closs, Wiley, New York, 1973, 41. 22 P. W. Atkins and G. T. Evans, Mol. Phys., 1974, 27, 1633. 23 J. B. Pederson and J. H. Freed, J. Chem. Phys., 1975, 62, 1706. 24 C. Blättler, F. Jent and H. Paul, Chem. Phys. Lett., 1990, 166, 375. 25 F. J. Adrian, J. Chem. Phys., 1971, 54, 3918. 26 F. J. Adrian, in Chemically Induced Magnetic Polarization, ed. L. T. Muus, P. W. Atkins, K. A. McLauchlan and J. B. Pedersen, Reidel, Dordrecht, 1977, 77. 27 F. J. Adrian and L. Monchik, J. Chem. Phys., 1980, 72, 5786. 28 S. Basu, K. A. McLauchlan and R. C. Sealy, J. Phys. E, 1983, 16, 1767. 29 S. Basu, K. A. McLauchlan and R. C. Sealy, Mol. Phys., 1984, 52, 431. 30 G.-H. Goudsmit and H. Paul, Chem. Phys. Lett., 1993, 208, 73. 31 K. A. McLauchlan and D. G. Stephens, Mol. Phys., 1987, 60, 1159. 32 O. Ces. K. A. McLauchlan and T. Quershi, Appl. Magn. Reson., in the press. 33 S. Basu, K. A. McLauchlan and R. C. Sealy, Chem. Phys. Lett., 1982, 88, 84. 34 C. D. Buckley and K. A. McLauchlan, Chem. Phys., 1984, 86, 323. 35 M. S. de Groot, I A. M. Henselmann, F. J. Reinders and J. H. van der Waals, Mol. Phys., 1975, 29, 37. 36 J. S. Vincent, J. Chem. Phys., 1967, 47, 1830. 37 D. A. Antheunis, B. J. Batter, J. Schmidt and J. H. van der Waals, Mol. Phys., 1975, 29, 49. 38 R. W. Eveson, K. A. McLauchlan and E. Page-Croft, unpublished results. 39 N. I. Avdievich, E. G. Bagryanskaya, Yu. A. Grishin and R. Z. Dagdeev, Chem. Phys. Lett., 1989, 155, 141. 40 R. Z. Sagdeev, Yu. N. Molin and K. M. Salikhov, Bull. Magn. Reson., 1980, 2, 66. 41 U. E. Steiner and T. Ulrich, Chem. Rev., 1989, 89, 51. 42 Saburo Nagakura Festschrift, J. Phys. Chem. A, 1997, 101, many articles. 43 R. M. Noyes, J. Chem. Phys., 1954, 22, 1349. 44 K. Schulten and P. Wolynes, J. Chem. Phys., 1978, 68, 3292. 45 S. N. Batchelor, C. W. M. Kay, K. A. McLauchlan, P. D. Smith and M. T. Yeung, Mol. Phys., 1994, 82, 325. 46 T. Eykyn and K. A. McLauchlan, unpublished results. Paper 7/02507F Received 11th April 1997 Accepted 21st May 1997

ISSN:1472-779X    

DOI:10.1039/a702507f

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

J. Chem. Soc., Perkin Trans. 2, 1997 2473 EPR insights into aqueous solutions of gelatin and sodium dodecyl sulfate † Peter C. GriYths,a Christopher C. Rowlands,b Pascal GoyVon,c Andrew M. Howe d and Barney L. Bales e a Department of Chemistry, University of Wales Cardiff, PO Box 912, Cardiff, UK CF1 3TB b EPSRC National ENDOR Centre, Department of Chemistry, University of Wales Cardiff, PO Box 912, Cardiff, UK CF1 3TB c E.S.P.C.I., 10 rue Vauquelin, 75231 Paris Cedex 05, France d Kodak European R&D, Harrow, Middlesex, UK HA1 4TY e Department of Physics and Astronomy, California State University at Northridge, Northridge, CA 91330-8268 USA The characteristics of the EPR spectra of the spin-probe 16-doxyl-stearic acid‡ methyl ester (16-DSE) solubilised in micelles of the anionic surfactant sodium dodecyl sulfate (SDS) have been examined as functions of SDS and gelatin concentrations.For simple SDS solutions, the rotational correlation time increases slightly with surfactant concentration whilst the polarity decreases slightly.In contrast, however, in the presence of gelatin these properties vary markedly as a function of the stoichiometric ratio of the concentration of surfactant to gelatin; the correlation time decreasing and the hyperfine coupling constant increasing with increasing surfactant concentration. In the presence of gelatin therefore, 16-DSE reports a very different micellar environment compared with the simple SDS case.Furthermore, this environment differs significantly from that observed in solutions of synthetic, non-ionic homopolymers and SDS. These features arise due to the varied characteristics of the amino acids present in the protein. Introduction The interaction between certain polymer and surfactant pairs in aqueous solution has been studied extensively1,2 because of the useful properties that the polymer (rheological control, stability enhancement) and surfactant (surface tension lowering, wetting) impart to the system.Generally, interactions between nonionic polymers and anionic surfactants,3,4 polyelectrolytes and oppositely charged surfactants,5,6 or hydrophobically modified polymers and anionic surfactants 6,7 are significant. Any interaction between non-ionic polymers and non-ionic surfactants is considerably weaker and largely due to excluded volume effects. The interaction starts at a critical aggregation concentration denoted cac or cmc(1) and this concentration is substantially lower than the critical micelle concentration cmc, the concentration at which micelles would form in the absence of the polymer.The polymer must, therefore, stabilise the formation of the micelle. NMR studies of the interaction between synthetic homopolymers such as poly(ethylene oxide) (PEO) or poly(vinylpyrrolidinone) (PVP) and anionic surfactants, such as sodium dodecyl sulfate (SDS) have shown that only those surfactant carbon atoms closest to the headgroup interact with the polymer segments. Therefore, the polymer does not penetrate the micellar core.4,8 Neutron scattering and fluorescence studies have shown that these ‘adsorbed’ micelles are comparable in size to the micelles that would be formed in the absence of any polymer.9,10 The stabilising effect arises through the polymer segments adsorbing into the micelle palisade layer, thereby shielding part of the hydrophobic core of the micelle from the aqueous phase.There are thermodynamic penalties to this arrangement. Sections of the polymer coil are effectively constrained at the inter- † Presented at the 30th International Meeting of the Electron Spin Resonance Group of the RSC, University of Lancaster, 6–10th April 1997. ‡ 2-(14-Carboxytetradecyl)-2-ethyl-4,4-dimethyl-3-oxazolidinyloxy free radical. face, which results in a loss of both translational and configurational entropy. Furthermore, there are steric interactions between the micelle headgroups and polymer segments.A thermodynamic modelling of polymer–surfactant systems has recently been presented.11,12 The adsorbed micelles are charged and therefore intramicellar repulsion will be present between the micelles adsorbed on the same polymer molecule. This causes an expansion of the polymer molecule. Concomitantly, the intramicellar repulsion necessitates that the solution surfactant concentration required to place a second or any subsequent micelle onto the same polymer molecule is slightly higher than for the previous micelle.This has two effects: the micelle occupancy on the polymer molecules is smooth (no polymer molecule will have significantly more micelles than the next) and second, the solution surfactant unimer (a single surfactant molecule) concentration increases. The polymer is ‘saturated’ when the energy penalty due to this intramicellar repulsion is too great to force a further micelle onto the polymer; micelles then form in solution.This concentration is denoted cmc(2) and occurs when the unimer concentration exceeds the cmc of the surfactant under the prevailing conditions of ionic strength and pH.13 In this study, we are concerned with the proteinaceous material gelatin, which is significantly different from the more frequently studied synthetic homopolymers where all the segments are identical. Gelatin comprises both ionic and non-ionic amino acids and the non-ionic amino acids are both hydrophilic and hydrophobic.14 The principal constituents are glycine (32– 35%), proline (11–13%), alanine (10–11%), hydroxyproline (9– 19%), glutamic acid (7–8%), aspartic acid (4–5%) and arginine (5%). On addition of SDS, gelatin-bound micelles are formed around 1 mM.11,15 This interaction will contain both charge and hydrophobic character. 13C NMR studies at ambient pH have shown that the anionic surfactant interacts strongly with the cationic and non-ionic residues of gelatin, but not with the anionic residues.14 The onset of this interaction is accompanied by substantial increases in viscosity.11,15 Intriguingly, a local maximum in the viscosity occurs at a surfactant concentration2474 J.Chem. Soc., Perkin Trans. 2, 1997 corresponding to 1 micelle per gelatin molecule.13,15,16 The viscosity at this maximum is some ten or so times greater than the SDS-free value. A series of anionic alkyl sulfates ranging from octyl to tetradecyl have been studied 13 and with the exception of the octyl, all showed a maximum occurring at this stoichiometry. This is not observed in non-ionic polymer–anionic surfactant interactions. The basis of this manifestation is still unclear, but undoubtedly related to a connivance of the mechanisms of charge and hydrophobic interactions.The aim of this work, therefore, is to contrast the polyampholytic interactions occurring between SDS and gelatin with the corresponding interaction occurring between the equivalent segments of a non-ionic synthetic homopolymer (PEO) and SDS.Experimental Sample preparation Two samples of lime processed gelatin have been used: (a) a photographic grade alkali-processed polydisperse gelatin (Kodak Ltd., Harrow, UK), referred to as standard gelatin (nominal molecular weight 107 000 with Mw/Mn ª 3.0), with a bimodal size distribution, and (b) a fractionated sample, kindly supplied by Dr T. H. Whitesides, Eastman Kodak Co., derived from the standard gelatin by a fractional precipitation procedure from methanol with sodium nitrate in which the higher molecular weight fraction has been removed.Both gelatin samples have isoelectric points (iep) of 4.9–5.0. In the absence of acid, base or surfactant these gelatins form solutions with a pH of 5.8, at which they are slightly negatively charged. In the presence of SDS, solutions have pH values in the range 5.5–6.5. The two gelatin samples have very different molecular weight distributions.Other work17 has shown that molecular weight is not an important factor in the microenvironment as reported by the spin-probe. The spin probe, 16-doxyl-stearic acid ‡ methyl ester (Fluka), and sodium dodecyl sulfate (99%, Aldrich) were used as received. In order to minimise sample-to-sample variations, all samples were prepared from a stock gelatin solution which was prepared by warming the required amount of gelatin and distilled water to 45 8C.The solution was maintained at that temperature for 1–2 h. Due to its insolubility in water, the spin probe (concentration <5 × 1024 M) was first dissolved in a surfactant solution, c > cmc. To aliquots of stock gelatin and stock spin probe–surfactant solution were added varying amounts of surfactant solution (without spin probe) and distilled water. All samples were equilibrated at 45 8C for at least 1 h before being flame sealed in capillary tubes made of soda glass (Samco).The capillary tubes were placed in standard EPR tubes for the measurement. Throughout this paper, the surfactant concentration is expressed in mM units whilst the gelatin is given as percentage (w/w) in H2O. Electron paramagnetic resonance EPR spectra were recorded at 45 8C on a JEOL JES-RE-2X EPR spectrometer equipped with a variable temperature accessory controlled by a gas stream. 100 kHz field modulation of amplitude 1 G and 10 mW microwave power were used.The sweep-width of the magnetic field was set at 50 G, with a scan time of 60 s using a time constant of 0.1 s. This modulation amplitude broadens the Gaussian component of the EPR lines by 0.12 G, leaving the Lorentzian component unchanged.18 The effect of this Gaussian broadening, as well as that due to unresolved hyperfine structure, was corrected as described below. Each EPR spectrum is an average of five scans. Further measurements were performed using a Bruker ESP 300E, employing a 100 kHz field modulation of amplitude 1 G and 1 mW microwave power.The sweep-width of the magnetic field was 50 G, with a scan time of 83.9 s, using a time constant of 10.2 ms. On this machine, each EPR spectrum was averaged for at least three scans. Theoretical considerations 16-Doxyl-stearic acid methyl ester (16-DSE) was chosen as the spin-probe (a) due to its insolubility in water (no EPR signal could be detected from 16-DSE containing surfactant solutions below 8 mM—the critical micelle concentration), (b) because of its structural similarity to SDS and (c) because it, or the parent 16-doxyl-stearic acid, have been used previously to study PEO–SDS complexes 19,20 and hydrophobically modified PEO– SDS complexes.21,22 It is tacitly assumed here that the ester locates in a broadly similar position to the acid.Rotational correlation times The rotational correlation time, tc, and micropolarity can be determined from an analysis of the EPR spectra.For very fast motion of aminoxyl § radicals, i.e. tc < 10211 s, the EPR spectrum of the radical is insensitive to the rate of molecular motion and consists of three lines of equal intensities. For fast motion, 10211 < tc < 1029 s, the effective rotational correlation time to a good approximation can be calculated from eqn. (1), tc uncorrected = 6.6 × 10210 DH0F÷SV0 V21 D1÷SV0 V11 D2 2G (1) where DH0 represents the overall line-width of the central line and V21,0,11 represent the peak-to-peak intensity of the high-, middle- and low-field lines respectively.Alternatively, the approximation given in eqn. (2) can also be used. tb uncorrected = 6.6 × 10210 DH0F÷SV0 V11 D2÷SV0 V21 DG (2) The superscript ‘uncorrected’ in eqns. (1) and (2) refers to the fact that the lines are inhomogeneously broadened by unresolved hyperfine structure and modulation broadening. Therefore, the three lines in a spectrum are neither Lorentzian nor have the same shape.To correct these errors, Bales 23 adds corrections to the previous equations consisting of a Voigt approximation, such that the lines are a sum of Lorentzians with a Gaussian profile. The shape of the Voigt approximation curve depends only on the Voigt parameter c which is the ratio of the Gaussian and Lorentzian line-widths. To correct rotational correlation times calculated from eqns. (1) and (2), the value of the Voigt parameter for the central line is required.Thus, eqns. (3)–(6) follow where c is the Voigt parameter of the tb = S(c)Q0tb uncorrected (3) tc = S(c)Q0tc uncorrected (4) where Q0 = [21 1 ÷(1 1 4c2)] 2c2 (5) and S(c) = (1 1 1.78c 1 1.85c2) (1 1 2.08c) (6) central line. For isotropic motion, these two estimates of the correlation time are equal. For all the solutions presented in this paper, this was the case. However, to facilitate the comparison with the PEO–SDS data, it is the uncorrected data [eqn.(1)] that are presented in Fig. 3. Polarity determination Hyperfine coupling results from the magnetic interactions § Formerly referred to as nitroxide.J. Chem. Soc., Perkin Trans. 2, 1997 2475 between the electron and nuclear spins of atomic neighbours. In the case of aminoxyl radicals, hyperfine coupling to the 14N yields three possible spin states (m = 21, 0, 11), and thus three lines are observed in the spectrum. The hyperfine coupling constant is determined as half the separation of the two outermost lines.The hyperfine coupling constant varies with the local polarity in the vicinity of the aminoxyl group. This variation has been interpreted 21 to be due to a shift of the equilibrium illustrated in Scheme 1. It is well-known that polar solvents or those that can provide a hydrogen bond will stabilise form 1 thereby increasing Ao. In the case of micellar environments, the aminoxyl group can be engineered to remain in the micelle such that Ao reveals information regarding the polarity of the location of the spin probe. Results To facilitate comparisons between the gelatin–SDS and synthetic, non-ionic polymer–SDS systems, the results will be presented first and discussed in a later section.Hyperfine coupling constant Fig. 1 shows the behaviour of the hyperfine coupling constant of the spin-probe 16-DSE solubilised in SDS solutions, as a function of SDS concentration containing 0, 1.67 and 5 wt% gelatin. In the absence of gelatin, the hyperfine coupling constant decreases slightly with increasing SDS concentration, in good agreement with other work.24 In the two gelatin containing solutions, the hyperfine coupling constant increases signifi- cantly with increasing SDS concentration.Up to 150 mM SDS, the hyperfine coupling constant is lowest (i.e. the least polar) at any given SDS concentration for the system containing most gelatin and highest for the gelatin-free system.The micellar environment is therefore very different in the presence of gelatin. The hyperfine coupling constant is similar for all three systems at high SDS concentrations. Fig. 1 Hyperfine coupling constant for 16-DSE solubilised in SDS micelles as a function of SDS concentration: (d) no gelatin; (s) 1.67 wt% fractionated gelatin; and (j) 5 wt% fractionated gelatin Scheme 1 •N O:– :N O• •N O:– H O 1 2 Fig. 2 shows the same data as Fig. 1 normalised by dividing by the gelatin concentration.Since all the gelatin data overlay, the important factor in these systems would appear to be the composition of the gelatin–SDS micelle complex. In order to facilitate a comparison between the gelatin–SDS (45 8C) and literature PEO–SDS data (25 8C) (broken line), a further correction has been applied to remove the cac dependence. Thus, the abscissa is presented as [(Ctotal 2 Ccac)/Cpolymer] (NB for gelatin–SDS, the cac = 1 mM whereas PEO–SDS, the cac = 4 mM).Data for simple SDS (45 8C) (solid line) are also included. Approaching a surfactant–polymer concentration ratio of 30 (ª150 mM for 5 wt% gelatin) indicated by the vertical line in Fig. 2, the distinction between gelatin-bound SDS micelles and gelatin-free micelles is minimal. This concentration is signifi- cant as it corresponds to the saturation of the gelatin, c > cmc(2), and above this concentration, there are also SDS micelles present in solution. Our main interest is therefore the region up to this saturation concentration.Rotational correlation time Fig. 3 shows the calculated rotational correlation times for 16- DSE in the SDS micelles as a function of gelatin concentration for three concentrations of SDS, 10, 20 and 55 mM. In this case, the gelatin concentration has been varied yet the data can be normalised by the approach taken in Fig. 2. This strongly reinforces the hypothesis that it is the composition of the gelatin–SDS micelle complex that is the important feature in these systems.Due to the nature of this normalisation, the SDS-only data cannot be plotted,but tc increases smoothly with increasing SDS concentration from tc = 1.7 × 10210 s at 30 mM SDS passing through tc = 2.7 × 10210 s at 150 mM SDS. Discussion Deuterium electron spin echo modulation ESEM has been used 19,20 to probe the depth of penetration of a series of aminoxyl labelled doxyl-stearic acid spin-probes into the micelle.A maximum in the penetration depth was observed when the spinprobe was located at the 12-position. The depth for the 16- position labelled probe was estimated to be approximately 0.5 nm; just inside the hydrophobic core. Furthermore, according to studies involving the quenching of pyrene fluorescence by copper ions,25 these types of probe are readily accessible to the copper ions present in solution. The nature of the quenching Fig. 2 Hyperfine coupling constant for 16-DSE solubilised in SDS micelles as a function of normalised SDS–gelatin concentration ratio: (solid line) no gelatin (s) 1.67 wt% fractionated gelatin; and (j) 5 wt% fractionated gelatin.The dashed line corresponds to the PEO–SDS systems at 25 8C.2476 J. Chem. Soc., Perkin Trans. 2, 1997 was diffusion controlled at a rate approaching that of copper ions in bulk water. All of the data in liquid systems are consistent with a model of the spin-probe executing rapid motion near the micelle surface.Hence, the data presented here are concerned with structural perturbations occuring within this region. Other spin-probes would report on different regions of the micelle or the continuous phase.24 Hyperfine coupling constant The interpretation of the hyperfine coupling constant is nontrivial as several factors contribute to its behaviour including (a) the charged nature of the headgroup and amino acid residues as well as the presence of any bound counter-ions and (b) replacement of some water molecules previously in contact with the extremities of the micellar core 14 by the adsorbed polymer.It has been proposed that the most important of these for synthetic polymers is the charged nature of the headgroup and associated counter-ions. The hyperfine coupling constant of 16-DSE has been measured for a series of relevant surfactants in solution: SDS, Ao = 15.30 ± 0.05 G; C12E8, Ao = 14.50 ± 0.05 G; whereas for hydrophobically modified PEO, C12EO200C12, Ao = 14.85 ± 0.05 G.The interpretation of these values is informative as an insight into the gelatin–SDS system. The charges in the SDS headgroup result in the higher polarity for the spin-probe in micellar SDS. The smaller aggregation number (Nagg = 15) 26 of the polymeric surfactant C12EO200C12, compared to C12E8 (Nagg = 55–62) 27 results in a smaller micelle and thus, large surface area to volume ratio. Proportionally, the spin-probe spends more time at the surface of the micelle and thus in contact with the aqueous phase.Ao therefore is higher. These simple comparisons show, in particular, that charge has the largest effect on Ao. Consider the interaction between PEO and SDS. In the regions between the interdigitated charged headgroups, there are water molecules in contact with the extremity of the hydrophobic core. On adsorption of the homopolymer, the polymer segments will displace some of the water molecules rendering the micellar surface less polar.Thus Ao decreases. However, Ao for 16-DSE in PEO–SDS is largely insensitive to PEO concentration; 19 Ao = 15.41 ± 0.05 G, 100 mM SDS, 0 wt% PEO and Ao = 15.36 ± 0.01 G, 100 mM SDS, 10 wt% PEO. From these numbers, it may be assumed that this effect is relatively weak. The same cannot be said for the gelatin–SDS system. However, it should be noted that the PEO–SDS study 19 was performed in Fig. 3 Rotational correlation time for 16-DSE solubilised in SDS micelles as a function of normalised SDS–standard gelatin concentration ratio (gelatin concentration varied): (d) 10 mM SDS; (s) 20 mM SDS; and (h) 55 mM SDS D2O and whilst this may affect the value of Ao, any trend with increasing surfactant concentration should be comparable.On addition of SDS to both 1 wt% and 2.5 wt% hydrophobically modified PEO, C12EO200C12, solutions, Ao increases sharply (Ao = 15.05 ± 0.05 G at 40 mM SDS) before subsequently decreasing much more slowly to Ao = 14.90 ± 0.05 G at 180 mM SDS.When the data are corrected for the relative concentrations of polymer and surfactant, the aggregates formed over a range of SDS concentrations were found to be rather similar.21 Consider now the gelatin–SDS system. At very low SDS concentrations, the hyperfine coupling constant is close to the value of the C12 aggregates found in solutions of C12EO200C12. The hyperfine coupling constant increases with increasing SDS concentration, ultimately to a value identical to pure SDS micelles.In these solutions, gelatin as a whole is slightly negatively charged. 13C NMR studies at ambient pH have shown that the cationic and non-ionic residues of gelatin interact strongly with the anionic surfactant.14 If the interaction is predominantly between the non-ionic residues and the surfactant, one would expect the polarity at the micelle surface to decrease due to the displacement of water molecules. With increasing SDS concentration, the relative proportions of non-ionic residues and anionic surfactant in the gelatin–SDS micelle complex will change in favour of the surfactant.The complex will become more negatively charged. Hence, the hyperfine coupling constant would be expected to increase towards the pure SDS value. This scenario agrees with the experimental observations. However, a similar mechanism would be present in the PEO– SDS system yet the Ao data show little change across the same relative concentration range.19 To account for the observed changes in magnitude of the hyperfine coupling constant in the gelatin–SDS system, we suggest there is some charge neutralisation.Furthermore, addition of a little SDS to the uncharged ‘pure’ C12 end-groups present in the C12EO200C12 solutions 21,22 results in a substantial increase in Ao (Ao = 14.88 G no SDS to Ao = 15.05 G at 20 mM SDS, 2.5 wt% polymer). This is due to the increase in charge of the mixed aggregate. This magnitude of change in Ao is comparable to that observed in the gelatin– SDS case.In the C12EO200C12–SDS system, a subsequent gradual decrease in Ao with increasing SDS concentration is observed due to the interaction between the SDS and the nonionic segments of the PEO backbone (Ao = 14.95 G at 180 mM SDS, 2.5 wt% polymer). Hence, the displacement of water is a much weaker contribution to the changes seen in Ao than the charge effects. The behaviour of the hyperfine coupling constant observed in the gelatin case involves a change in the charge on the gelatin–SDS micelle complex rather than the displacement of water molecules.Cationic residues of gelatin interacting with the anionic surfactant lead to some charge neutralisation. A less polar environment results and Ao decreases. At low SDS concentrations, it is probable that the charged interactions between cationic residues and the anionic micelle dominate. Ao is therefore at its lowest value.With increasing SDS concentration, the gelatin–SDS micelle complexes are diluted with non-ionic residues as well as the anionic surfactant. Thus, the charge increases 11 and hence, the polarity of the micelle. Ultimately, the micelle takes on the character of a pure SDS micelle. Rotational correlation time The correlation time data of Fig. 3 show that the rotation of the spin-probe is much more restricted in the gelatin-bound micelles compared with the simple SDS micelle.This motion is however, still isotropic. The gelatin residues adsorbed around the headgroups and within the outer regions of the hydrophobic core restrict the motion of the spin-probe. With increasing SDS–gelatin concentration ratio, the proportion of gelatin residues present in a single micelle decreases. The dynamics of the spin-probe tend towards that of the gelatin-free SDSJ. Chem. Soc., Perkin Trans. 2, 1997 2477 micelle. Above the saturation concentration, the distinction between those SDS micelles formed in the presence of gelatin to those in simple SDS solutions is minimal.Conclusions The interaction between gelatin and SDS shows many similarities with the frequently studied synthetic polymer–anionic surfactant systems. In this paper, some of its unique character is presented. The spin-probe 16-DSE solubilised in SDS micelles shows that at low SDS concentrations, the gelatin adsorbs onto the micelle surface and greatly restricts the motion of the spinprobe.The considerably less polar environment suggests that a very high proportion of the cationic residues are present at the micelle surface. The cationic residues can be regarded as ‘pinning’ the gelatin to the micelle surface and are saturated at very low SDS concentrations. The strength of this binding results in a much more rigid environment for the spin-probe. tc is therefore high. At higher SDS concentrations, the composition of the micelle is much more rich in SDS and non-ionic residues and thus the polarity increases.The gelatin can take on a much more extended conformation and the rotation of the spin-probe becomes more fluid. tc decreases. Approaching the saturation level, the distinction between the gelatin-bound SDS micelle and the gelatin-free SDS micelle is negligible. Acknowledgements This work has been supported by the EPSRC, Ministère de l’Agriculture, de la Pêche et de l’Alimentation and Institut National Agronomique Paris-Grignon.We are grateful to Tom Whitesides and Bonnie Howell (Eastman Kodak Co.) for supplying and characterising the fractionated gelatin sample, and to EPSRC for provision of an ENDOR/EPR spectrometer. References 1 E. D. Goddard, J. Am. Oil Chem. Soc., 1994, 71, 1. 2 E. D. Goddard, Interaction of Surfactants with Polymers and Proteins, ed. E. D. Goddard and K. P. Anathapadmanabhan, CRC Press, Boca Raton, FL, 1993, p. 395. 3 K. Chari, B. Antalek, M.Y. Lin and S. K. Sinha, J. Chem. Phys., 1994, 100, 5294. 4 K. Chari, B. Antalek and J. Minter, Phys. Rev. Lett., 1995, 74, 3624. 5 Y. J. Li, J. L. Xia and P. L. Dubin, Macromolecules, 1994, 27, 7049. 6 F. Guillemet, L. Piculell, S. Nilsson, M. Djabourov and B. Lindman, Prog. Colloid Polym. Sci., 1995, 98, 47. 7 B. Nyström, H. Walderhaug, F. K. Hansen and B. Lindman, Langmuir, 1995, 11, 750. 8 R. Ramachandran and G. J. Kennedy, Colloids Surf., 1991, 54, 261. 9 B. Cabane, J. Phys. Chem., 1977, 81, 1639. 10 B. Cabane and R. Duplessix, J. Phys. (Paris), 1987, 48, 651. 11 T. H. Whitesides and D. D. Miller, Langmuir, 1994, 10, 2899. 12 Y. J. Nikas and D. Blankschtein, Langmuir, 1994, 10, 3512. 13 P. C. Griffiths, P. Stilbs, A. M. Howe and T. H. Whitesides, Langmuir, 1996, 12, 5302. 14 D. D. Miller, W. Lenhart, B. J. Antalek, A. J. Williams and J. M. Hewitt, Langmuir, 1994, 10, 68. 15 J. Greener, B. A. Contestable and M. D. Bale, Macromolecules, 1987, 20, 2490. 16 P. C. Griffiths, P. Stilbs, T. Cosgrove and A. M. Howe, Langmuir, 1996, 12, 2884. 17 P. C. Griffiths, A. M. Howe, T. G. Whitesides, C. C. Rowlands, B. L. Bales and P. Goyffon, submitted for publication in Langmuir. 18 M. Peric and H. J. Halpern, J. Magn. Reson., Ser. A, 1994, 109, 198. 19 Y. S. Kang and L. Kevan, J. Phys. Chem., 1994, 98, 7624. 20 P. Baglioni and L. Kevan, Heterogeneous Chem. Rev., 1995, 2, 1. 21 K. Persson and B. L. Bales, J. Chem. Soc., Faraday Trans. 1, 1995, 91, 2863. 22 K. Persson, Associative Polymers. Their Self-Association and Interaction with Surfactants, Royal Institute of Technology, Stockholm, 1995. 23 B. L. Bales, Biological Magnetic Resonance, ed. L. J. Berliner and J. Reuben, Plenum Press, New York, 1989, Vol. 8, p. 77. 24 F. M. Witte, P. L. Buwalda and J. B. F. N. Engberts, Colloid Polymer Sci., 1987, 265, 42. 25 B. L. Bales and M. Almgren, J. Phys. Chem., 1995, 99, 15 153. 26 A. Yekta, J. Duhamel, P. Adiwidjaja, P. Brochard and M. A. Winnik, Langmuir, 1993, 9, 881. 27 L. J. Magid, R. Triolo and J. S. Johnson, Jr., J. Phys. Chem., 1984, 88, 5730. Paper 7/02945D Received 29th April 1997 Accepted 16th June 1997

ISSN:1472-779X    

DOI:10.1039/a702945d

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

J. Chem. Soc., Perkin Trans. 2, 1997 2479 EPR study of spin-trapped free radical intermediates formed in the heterogeneously-assisted photodecomposition of acetaldehyde † Charles A. Jenkins,a Damien M. Murphy,a Christopher C. Rowlands a and Terry A. Egerton b a EPSRC National ENDOR Centre, Department of Chemistry, University of Wales Cardiff, PO Box 912, Cardiff, UK CF1 3TB b Tioxide UK Ltd., Haverton Hill Road, Billingham, Cleveland, UK TS23 1PS Electron paramagnetic resonance spectroscopy is used to detect radical adducts of PBN (a-phenyl N-tertbutyl nitrone) generated by exposure of solutions and suspensions to ambient or high power UV at 300 K.Exposure of acetaldehyde to direct sunlight generates a different PBN radical adduct to high power UV irradiation. Direct sunlight irradiation of deoxygenated acetaldehyde generates PBN–acetyl adducts whereas direct sunlight exposure of oxygenated acetaldehyde produces PBN–acetoxyl adducts. High power UV irradiation of TiO2/acetaldehyde suspensions yields the same radical adduct generated when no TiO2 is present—this adduct (assigned to trapped formyl radicals or PBN degradation products) is produced irrespective of the state of oxygenation of solution.Direct sunlight irradiation of deoxygenated TiO2/acetaldehyde suspension results in the production of PBN–acetyl adducts as the primary species. In oxygenated TiO2/acetaldehyde suspension, PBN–acetyl adducts are again produced as the primary species, together with a weakly adducted secondary species—assigned to PBN–acetoxyl adducts.TiO2 band gap transitions are observed to play no part in the production of radical intermediates in sunlight irradiated acetaldehyde/TiO2 suspension. The extent of non-band gap dependent processes is shown to be sensitive to the surface basicity of the metal oxide. Band gap mediated radical production is demonstrated to arise when acetaldehyde photoreduction is coupled to the concomitant photooxidation of ethanol.Ethanol derived PBN–ethoxy adducts are detected as the primary species arising from sunlight irradiation of both oxygenated and deoxygenated TiO2/acetaldehyde/ethanol suspensions. 1. Introduction Over the period since 1971, when Fujishima and Honda 1 first reported the use of TiO2 in splitting water for solar energy conversion, the use of irradiated semiconductor suspensions for the purpose of photomineralization of organic pollutants has received continual and growing scientific interest.Chlorinated hydrocarbons were the earliest pollutant substrates to be studied in heterogeneous photocatalytic systems,2 most probably because of their high toxicity and common occurrence as industrial effluents, and many detailed mechanisms for their photocatalytic destruction have been reported.3 Although there have now been many studies conducted for a wide range of organic contaminants—from simple alkanes to pesticides and dyes 4—there have been relatively few studies conducted concerning the detailed photodegradative mechanism for aliphatic aldehydes.Sources of aldehyde pollution range from their use in industry as synthetic precursors, where effluent discharges may cause soil and groundwater contamination, to atmospheric emissions arising from vehicle exhausts (ppm range) and slow release from synthetic furnishing materials (sub-ppm range). In photochemical smogs they are readily converted into respiratory irritants such as peroxyacyl nitrate and the possible mutagenic effects of long term, low level indoor exposure are still largely unknown.5 It is the destruction of such airborne organic pollutants that perhaps still poses one of the biggest challenges for environmental scientists. The recent interest in photocatalytic decontamination systems is driven by the search for clean effluent disposal technologies.The photo-mineralization reaction (1) has the advantage Organic substrate 1 O2 photocatalyst CO2 1 H2O (1) † Presented at the 30th International Meeting of the Electron Spin Resonance Group of the RSC, University of Lancaster, 6–10th April 1997.over many other clean-up technologies of producing relatively non toxic waste materials, even though the treated effluent cannot be considered as totally innocuous within the living environment. The TiO2 catalysed photomineralization of organic species is a complex process which has been shown, in many cases, to involve surface bound hydroxyl radical intermediates.6 However, the mechanism of aldehyde photoreaction at TiO2 surfaces is known to proceed via a non-conventional route.7 In this study, spin trapping is used to detect radical intermediates formed during the room temperature photoreaction of acetaldehyde in a TiO2 suspension.The spin trapping technique affords the indirect detection of short-lived radicals which, when present at low steady state concentrations, cannot be detected easily by other spectroscopic methods.The spin trap a-phenyl N-tert-butyl nitrone (PBN) has been employed successfully to detect radical intermediates in the similar phenol/TiO2 photoredox system.8 The coupling of acetaldehyde photoreduction to the concomitant photooxidation of ethanol is observed to result in a large increase in the number of radical intermediates trapped. We also report that the initial photoexcitation of substrate—as opposed to the classical mechanism of initial excitation of semiconductor—may result in an enhancement of the rate of photoredox processes, particularly at high photon fluxes. 2. Experimental 2.1 Chemicals Rutile titanium dioxide and AlPO4-coated TiO2 were supplied by Tioxide UK Ltd. (surface areas 93 and 7.7 m2 g21, respectively). MgO was supplied by Johnson Matthey (surface area ª150 m2 g21). Surface area measurements were made by single point N2 adsorption. The spin trapping reagent a-phenyl N-tert-butyl nitrone (PBN) was supplied by Aldrich Chemicals, and stored at 273 K in the dark.Acetaldehyde and ethanol (supplied by Aldrich Chemicals) were analytically pure, and2480 J. Chem. Soc., Perkin Trans. 2, 1997 stored at 295 K in the dark. All powders were pre-treated by heating to 600 K at 1024 mbar for 2 h to remove surface carbon impurities. 2.2 Experimental procedures The spin trap concentration was 0.05 mol dm23 in all experiments. The mode of action of the spin trap is shown in Scheme 1.The extent of the hyperfine interactions between the unpaired electron and the nitrogen and a-hydrogen nuclei are dependent upon both the nature of the radical adducted and the solvent employed. Deoxygenation of solutions was carried out by repeated freeze–pump–thaw cycles. Deoxygenated solutions were maintained under a nitrogen atmosphere. Oxygenation of solutions was carried out by bubbling air through the solution for 20 min at 293 K.Oxygenated solutions were maintained under an atmosphere of air. All oxygenation/deoxygenation procedures were performed in a quartz cell of 1 cm internal diameter, in the dark. In the case of all powder suspension studies, the powder : solution ratio was 0.01 g per gram. A 600 W high pressure Hg/Xe lamp, of broad band l output (Oriel Instruments), fitted with a water filter, was used for high power irradiation of solutions. Ambient irradiations were carried out by standing solutions in direct unfocused sunlight, in quartz cells (1 cm inside diameter) which were sealed with rubber septa.EPR spectra were recorded at 293 K on a Varian E109 spectrometer operating at 100 kHz modulating frequency and connected to a Stellar DS/EPR data acquisition system. 3. Results and discussion Titanium dioxide has been the most widely studied semiconductor photocatalyst for the destruction of organic wastes. It occurs naturally in three crystalline forms, anatase, rutile and brookite—the latter form having no practical use in photocatalytic systems.Anatase is the most active form, with a band gap energy equal to 3.2 eV compared to 3.0 eV for rutile.9 Consequently, anatase absorbs electromagnetic radiation less than 380 nm whilst rutile absorbs wavelengths below 405 nm. Fig. 1 shows the main photoprocesses which occur during the TiO2- sensitised heterogeneous photocatalytic mineralization of organics. The primary photochemical event is the absorption of a UV photon by the semiconductor, and this absorption excites an electron from the filled valence band to the conduction band [reaction (1)]. The negatively charged surface hydroxy groups, which terminate the TiO2 lattice, cause upward bending of the electron bands.There is, therefore, a tendency for the electrons to be repelled and the positive holes to be attracted to the surface. Consequently, the electrons and holes tend to move apart and this reduces their recombination rate [reaction (2)].Despite this, most of the excited electrons and valence band holes either recombine at recombination centres or are trapped at electron or hole traps [steps (5) and (6)]. At the surface, holes are trapped by hydroxy groups and form hydroxyl radicals which can participate in subsequent chemical reactions. Similarly, electrons can be trapped by adsorbed oxygen to form the adsorbed O2~2. The excited electrons can reduce the oxidised forms of redox couples with a redox energy beneath that of the conduction band, and holes can similarly oxidise the reduced form of redox couples with a redox energy above the valence band.Both processes—reduction by ecb and hvb—may occur competitively when more than one type of reactant is available for adsorption. A general photoreaction sequence is given in Scheme 2. Scheme 1 Mode of action of the spin trap PBN C H N CMe3 O– + + X• C H N CMe3 O• X (a) Production and initial reaction of electron–hole pairs TiO2 1 hn (l < 400 nm) æÆ hvb 1 ecb (2) hvb 1 OH(ADS) æÆ OH? (3) ecb 1 O2(ADS) æÆ O2~2 (4) (b) Primary photodissociation events CH3CHO æÆ CH3? 1 CHO? (l < 340 nm) (5) CH3CHO æÆ ?CH2CHO 1 H? (l < 220 nm) (6) CH3CHO æÆ CH3CO? 1 H? (l < 310 nm) (7) (c) Oxidation of primary photolysis products CH3CO? 1 O2 æÆ CH3CO3? (8) CH3CO3? 1 CH3CHO æÆ CH3CO3H 1 CH3CO? (9) CH3CO3H 1 CH3CHO æÆ 2 CH3CO2H (10) Scheme 2 The steady state concentration of surface electrons and holes is controlled by the electron–hole recombination in the bulk of the solid—a bimolecular reaction.Because of this, the surface concentration of electrons, or holes, varies not as the intensity of the UV radiation but as its square root. Photocatalytic mineralization processes have been shown by many authors 11,12 to be oxidatively coupled to the trapping of valence band holes by surface hydroxyls—the trapping process occurring primarily at the most coordinatively saturated, and therefore most basic, OH groups.13 Joyce-Pruden et al. 14 have shown that the addition of NaOH, in amounts not exceeding the number of TiO2 surface sites available for OH adsorption, leads to a significant increase in the quantum efficiency of TiO2 catalysed photo-processes. Surface bound OH? has been shown to be a ‘transient species’ 15 on TiO2, and there is a wealth of EPR evidence available, using the spin trapping method, to demonstrate that OH? radicals participate in the photocatalytic oxidation of many organic molecules.6 The spin adduct parameters for OH? trapped in irradiated colloidal TiO2 systems were reported as long ago as 1979 by Bard.16 Aliphatic compounds may react directly with photogenerated holes (R 1 hvb æÆ R? 1 H1), or with the photogenerated OH? radical (R 1 OH? Fig. 1 Photoinduced charge-transfer processes in TiO2 suspensions (CB = conduction band, VB = valence band). 1, Absorption of a UV photon to produce valence band holes (hvb) and conduction band electrons (ecb); 2, electron–hole recombination; 3, reduction of adsorbed substrate by ecb at the conduction band edge; 4, oxidation of adsorbed substrate by hvb at the valence band edge; 5, ecb trapping at a lattice defect; 6, hvb trapping at a lattice defect.J.Chem. Soc., Perkin Trans. 2, 1997 2481 æÆ R? 1 H2O).6,16 Hydrogen abstraction by OH? is very fast 17 and either mechanism results in the production of the same aliphatic radical, which is then capable of reacting with O2 to produce an organoperoxy radical, RO2?.The fates of photogenerated electrons have been recently investigated 7 and it has been shown that, in aerated suspensions, ecb can be reductively coupled through reaction with adsorbed O2 (ecb 1 O2(ADS) æÆ O2~2). The superoxide radical anion thus produced is postulated to further react to form peroxide (O2~2 1 H1 æÆ ?OOH) or to combine with an organoperoxy radical to yield an unstable tetraoxide intermediate (RO2? 1 O2~2 æÆ ROOOO2).7 The bimolecular combination of two superoxide anion radicals is known to be very slow in comparison with the rate of combination of an organoperoxy species.Anpo et al.18 have demonstrated clearly the formation of superoxide radicals on irradiated TiO2 surfaces. The aliphatic aldehydes are electron rich and have reduction potentials of 21.48 to 21.90 eV vs. SCE (standard calomel electrode) compared to 20.85 eV vs. SCE for the electron deficient aromatic aldehydes, e.g., benzaldehyde.19 In contrast to an electron deficient aldehyde, the reduction potential of an aliphatic aldehyde does not lie close to the conduction band edge of TiO2.As a consequence of this, the photocatalytic reduction of aliphatic aldehydes occurs at a much slower rate— even when coupled to the oxidation of a highly efficient electron donor such as ethanol. Schwitzgebel et al.7 have also shown that the photocatalytic air oxidation of aliphatic aldehydes does not follow the mechanistic scheme outlined above—Scheme 2 shows the main aldehyde photoprocesses.Due to the high reactivity of the H atom of the terminal carbonyl function, aliphatic aldehydes are efficient reducing agents. Thus, hydrogen abstraction by hvb or OH?(SURF) is readily accomplished, and serves as the initiation step in the photocatalytic air oxid- Fig. 2 EPR spectra of PBN spin adducts. (A) 600 W UV irradiation (6 s) of neat acetaldehyde; (B) 20 min direct sunlight irradiation of deoxygenated acetaldehyde; (C) 20 min direct sunlight irradiation of oxygenated acetaldehyde.ation of aliphatic aldehydes. Following the initiation step, the process continues as a conventional alkylcarbonyl radical propagated chain reaction, in which neither conduction band electrons or O2~2 participate. Recent studies have shown that both photoreduction and photooxidation by TiO2 can be effected in non aqueous media.20 In the case of aliphatic aldehydes, the heterogeneous photocatalytic process has been observed to halt when the aldehyde becomes predominant in light absorption,14 i.e., there is a critical aldehyde concentration above which the semiconductor initiated photocatalysis is shut down due to the inefficient recycling of the photogenerated electron.We are able to show, however, that radical intermediate formation still occurs even when the critical aldehyde concentration is exceeded. 3.1 Radicals generated in the absence of oxide Irradiation of CH3CHO/PBN solutions. Prior to the irradiation of powder suspensions, a series of neat aldehyde irradiations were performed in order to aid in the identification of radical adducts formed in the heterogeneous process. No PBN adducts were generated in the dark, irrespective of the state of oxygenation of the solution. Fig. 2(A) shows the PBN adduct spectrum obtained from a 600 W irradiation of acetaldehyde (aN = 14.32 G, aH = 2.17 G), displaying the triplet of doublets structure typical of PBN adducts of aliphatic and aromatic radicals.The same adduct spectrum is produced in both oxygenated and deoxygenated solution, indicating that this adduct is not an oxidative product. The signal intensity reaches a maximum following only 8–10 s of irradiation. Scheme 2 illustrates the possible photoscission products arising from a broad band UV irradiation of acetaldehyde 21 by (a) primary photodissociation and (b) oxidation of primary products.Table 1 lists the hyperfine parameters for PBN adducts obtained in this study and studies by other authors. Comparison of our values with the list of PBN adduct parameters given in Table 1 indicates that this adduct is best attributed to trapped formyl radicals [eqn. (5)] or some Fig. 3 EPR spectra of PBN spin adducts. (A) 20 min direct sunlight irradiation of TiO2/acetaldehyde suspension (coated and uncoated TiO2 produce the same adducts); (B) 20 min direct sunlight irradiation of MgO/acetaldehyde suspension.2482 J.Chem. Soc., Perkin Trans. 2, 1997 Table 1 Hyperfine parameters for PBN adducts obtained in this and other studies Species adducted H? (hydrogen) HO? (hydroxyl) ?CH2CHO (C-centered) CH3CO? (acetyl) CH3? (methyl) CH3CH2O? (ethoxy) CH3(?CH)OH (C-centered) CH3CO2? (acetoxyl) ?CH2CH2OH (C-centered) PBN–Ox (paramagnetic) PBN–O? (PBN–oxy radical) PBN? (PBN radical) CHO? (formyl or PBN?) CH3CO2? (acetoxyl) CH3CO? (acetyl) PBN–Ox (paramagnetic) CH3CH2O? (ethoxy) CH3CH2O? (ethoxy) aN/G 15.30 15.30–15.60 14.80 14.0 14.91 14.40 15.40 13.40 14.66 7.95 15.80 16.20 14.32 13.39 14.28 7.96 14.38 14.37 aH/G 8.20 2.60–2.70 3.40 3.0 3.66 2.60 3.60 1.40 3.58 N/A 2.0 3.7 2.17 1.88 3.35 N/A 3.00 3.00 Solvent CH2Cl2 H2O Acetaldehyde 2-Methylpropane Toluene Ethanol Ethanol CH2Cl2 Ethanol CCl4 H2O H2O Acetaldehyde Acetaldehyde Acetaldehyde Acetaldehyde Ethanol 50 : 50 v/v EtOH–acetaldehyde Ref. 32 33 31 22 34 30 35 25 34 26 33 33 a a a a a a a This study. photodegradative product of the spin trap itself. Since neither has been reported previously in acetaldehyde solvent, a con- fident assignment of this species remains somewhat elusive. However, this result does aid in the identification of other adducts reported in this study by ruling out the adduction of the formyl radical in ambient light experiments. This result also demonstrates the uncertainties associated with the assignment of radical species formed during the high energy irradiation of spin trap solutions: UV irradiation (l < 350 nm) of solutions of the spin trap DMPO have been previously reported to result in the formation of paramagnetic dimers of the photoexcited spin trap.8 The spectrum of the adduct produced during a 20 min ambient light irradiation of deoxygenated acetaldehyde is shown in Fig. 2(B) (aN = 14.28, aH = 3.35). The signal intensity reaches a maximum following 25–30 min of irradiation.Because the solution is deoxygenated in this case, no peroxyacyl radicals are trapped [eqn. (8)]. It is also highly unlikely that photodissociations leading to the formation of ?CH2CHO [eqns. (5) and (6)] are occurring at any appreciable rate, since there is insufficient solar output at the high energies required. The spectrum in Fig. 2(B) can be assigned to the presence of trapped acetyl radicals generated in the process represented by eqn. (7). The spin adduct parameters of PBN–acetyl adducts have been previously reported, in 2-methylpropane solvent, as aN = 14.0 G and aH = 3.0 G.22 Signal intensities in all ambient photolyses of neat acetaldehyde are approximately two orders of magnitude lower than in the case of high power UV photolysis, probably because of the higher photon intensities used in the lamp photolysis.The EPR spectrum of the adducts formed during the ambient light photolysis of oxygenated acetaldehyde is shown in Fig. 2(C) (aN = 13.39 G, aH = 1.88 G). Arguments previously articulated above rule out the adduction of adducts arising as a result of high energy photoscissions. The adduct parameters in Fig. 2(C) differ markedly from those measured following the irradiation of deoxygenated solutions, indicating that the adducted species in this instance is an oxidative product. Inspection of eqns. (7) and (8) indicates that the initial radical product of the reaction between ambient light irradiated acetaldehyde and O2 is a peroxyacyl species (CH3CO3?), which has long been proposed by many authors as an intermediate in the autoxidation of aldehydes.23,24 In view of the inherent instability of peroxyacetyl radicals, the total absence of any reported PBN– peroxyacetyl adduct parameters is unsurprising.We conclude that Fig. 2(C) represents the PBN adduct spectrum of acetoxyl radicals (CH3CO2?) generated via decomposition of the peroxyacetyl species. Table 1 shows the close agreement with PBN– acetoxyl adduct parameters reported by Pryor et al.25 in CH2Cl2 solvent. 3.2 Radicals generated in the presence of oxide 3.2.1 Irradiation of uncoated TiO2 powder suspensions. No PBN radical adducts were detected from acetaldehyde/TiO2 powder suspensions kept in the dark. Irradiation of suspensions at 600 W UV power resulted once again in the spectrum shown in Fig. 2(A); the assignment of this adduct is discussed in section 3.1. The extent of oxygenation of the solution has no effect upon the spectrum generated by 600 W irradiation of acetaldehyde/TiO2 suspensions.The spectrum generated during a 20 min ambient light irradiation of oxygenated acetaldehyde/TiO2 suspension is shown in Fig. 3(A). Three paramagnetic species are evident, labelled (a), (b) and (c). The weak signal exhibiting a single coupling to nitrogen (aN = 7.96 G) is the well known paramagnetic form of oxidised PBN (PBN–Ox),26 and—in this study—is seen only in the ambient irradiation of oxygenated powder suspensions.A minor species of aN ª 14 G is also present. The concentration of this secondary adduct is too low for an accurate determination of its hyperfine parameters. However, the observation that this signal disappears when the irradiation is performed under anaerobic conditions suggests that the adduction of acetoxyl radicals is responsible for this adduct. The primary adduct in Fig. 3(A) has almost identical hyper- fine parameters to those reported for the ambient irradiation of neat deoxygenated acetaldehyde/PBN solution [Fig. 2(B)]. This signal is assigned to the trapping of acetyl radicals arising as a result of some heterogeneously assisted photoprocess occurring at the solid–liquid interface. The trapping of acetyl radicals as the primary species in oxygenated solution is most likely due to two independent effects. The initial products of acetyl radical oxidation are peroxyacetyl radicals, which have never been previously trapped due to their intrinsic instability.For example, k(fragmentation) at 300 K for peroxyacetyl radicals is >1010 s21 compared to ca.7 s21 for acetyl radicals.27 Inspection of eqn. (9) also reveals that the reaction of peroxyacetyl radicals with acetaldehyde yields acetyl radicals as the only paramagnetic products. It is then likely that acetyl radicals trapped in oxygenated suspension are generated via the two different reaction routes given in eqns.(7) and (9), resulting in the oxidative product appearing as the secondary adduct in the EPR spectrum. Separation of solution from powder by centrifugation, and subsequent measurement of spin adduct concentration in the separated phases, gives a solution signal only four times as intense as the powder signal, indicating that there is some tendency for the PBN spinJ. Chem. Soc., Perkin Trans. 2, 1997 2483 adducts to accumulate at the powder surface. This observation suggests that the lower signal intensity, for oxidative products in the mixed adduct spectrum, is due to the concentration of spin trap at the solid–liquid interface, where acetyl radicals would be preferentially trapped.Soria et al.8 have also reported the aggregation of spin adducts of DMPO at TiO2 surfaces. The primary radical adduct signal in Fig. 3(A) is found consistently to be some thirty times more intense than that assigned to the trapping of acetyl radicals in ambient irradiations of neat acetaldehyde.This result is considered unusual in view of the reported discontinuation of the heterogeneous photocatalytic process in TiO2/aliphatic aldehyde systems, when the aldehyde becomes predominant in light absorption.14 Clearly, given that band gap transitions play no part in aliphatic aldehyde photocatalysis at high aldehyde concentrations, the process leading to formation of radical intermediates in this instance must involve a surface assisted decomposition of photoexcited substrate molecules located near the surface region.These observations serve to reaffirm the concerns of Turchi and Ollis,28 who reported the difficulties associated with attempting to distinguish between the reaction of radicals formed following substrate adsorption and the reaction of radicals located close to the TiO2 surface. 3.2.2 Irradiation of TiO2 (AlPO4 coated) and MgO powder suspensions. In order to test the hypothesis that no semiconductor mediated electron transfer occurs when the upper limit for aldehyde concentration is exceeded, powder suspensions in which the band gap activated process would be effectively quenched were studied; AlPO4 coated TiO2 and uncoated MgO.AlPO4 coating of TiO2 prevents the photoexcited charge carriers of the semiconductor from reaching the surface of the material, thus preventing the transfer of electrons across the solid–liquid interface. Ambient light irradiation of AlPO4 coated TiO2 also produces qualitatively the spectrum shown in Fig. 3(A). The signal intensity of the primary acyl adduct arising from the irradiation of the coated material is 0.6 times less intense than that observed for irradiation of uncoated TiO2. This difference may be, in part, due to a higher degree of aggregation taking place in the coated powder as a result of differences in surface charge between coated and uncoated TiO2 samples. However, this explanation cannot account for the whole of the 40% loss in signal intensity observed.The degree of PBN oxidation is also reduced significantly when the AlPO4 coated material is used to sensitise the photoreaction of acetaldehyde. Ambient light irradiation of acetaldehyde/MgO suspension also results in the production of acetyl radicals [Fig. 3(B)]. The primary adduct signal is now three times as intense as that recorded during the ambient irradiation of uncoated TiO2 suspension. The signal intensities of both the secondary adduct and PBN–Ox are also significantly increased when compared to Fig. 3(A), indicating that the rate of both oxidative processes is enhanced. The observation that the production of all radical species is greatest in irradiated MgO suspensions demonstrates unequivocally that band gap transitions are not responsible for the observed photochemistry; the promotion of an electron across the MgO band gap (7.8 eV) requiring the absorption of a photon with l < 160 nm.29 A clear trend emerges in the relationship between extent of radical production/spin trap oxidation and surface basicity of the oxide studied; the production of both acetaldehyde derived radicals and PBN–Ox increasing along the series representing an increase in surface basicity; TiO2 (AlPO4 coat), TiO2, MgO.The relative changes in primary radical adduct signal intensity are 0.6:1:3 respectively. Such an increase in photocatalytic efficiency with increasing surface basicity of TiO2 has been reported for aldehyde photoredox reactions mediated by an initial excitation of the semiconductor.7 However, we are unaware of any previous reports of heterogeneous photoreactions—resulting from an initial excitation of the aldehyde—behaving in a similar manner.The trapping of the same primary adduct species in all three heterogeneous systems studied suggests that there is a common process involved in the formation of acetyl radicals in illuminated acetaldehyde/hydroxylated powder suspensions.We propose that abstraction of the H atom from the carbonyl carrying carbon in a photoexcited acetaldehyde molecule is responsible for acetyl radical formation, even when the band gap mediated photoprocess is shut down. The increase in power of hydrogen abstraction by surface bound OH groups then explains the observed correlation between primary adduct signal intensity and surface basicity. 3.3 Irradiation of CH3CHO/CH3CH2OH/PBN/TiO2 suspensions In order to investigate the comparative significance of nonband gap mediated aldehyde photochemistry at TiO2 surfaces, it was necessary to devise spin trapping experiments in which the semiconductor sensitised photocatalytic process is invoked.The TiO2 photocatalysed reduction of aliphatic aldehydes has been demonstrated as highly efficient when coupled to the concomitant photooxidation of ethanol.14 However, great care must be taken in controlling the experimental conditions since, under aerobic conditions, some degree of aldehyde oxidation may also occur. Ambient light photolysis of ethanol/PBN solution did not result in the production of radical adducts.However, ambient light exposure of a TiO2/ethanol/PBN suspension results in the production of the very weak PBN–adduct spectrum shown in Fig. 4(A) (aN = 14.38 G, aH = 3.00 G). By comparison with the list of adduct values given in Table 1, this adduct is assigned to the trapping of ethoxy radicals.30 The very low signal intensity is attributed to the absence of a suitable electron acceptor (such as acetaldehyde) which would otherwise retard the rate of electron–hole recombination within the TiO2.Once again, the same primary adduct is formed irrespective of the amount of dissolved O2 present. The total signal intensity of ethoxy adducts formed in neat ethanol/TiO2 suspension is ca. 30% of that observed for acetyl adduct formation in neat acetaldehyde/ TiO2 suspension, despite the higher rate of PBN spin adduction for ethoxy radicals reported in this study.This lower reactivity is due partly to the absence of any chromophore within the ethanol molecule, as well as the known lower reactivity of the terminal H atom of ethanol compared with the H atom of the acetaldehyde carbonyl function. Evidently, H abstraction from ethanol then requires a more powerful surface basic site, Fig. 4 EPR spectra of PBN spin adducts. (A) 20 min direct sunlight irradiation of TiO2/ethanol suspension; (B) 20 min direct sunlight irradiation of TiO2/ethanol suspension.2484 J.Chem. Soc., Perkin Trans. 2, 1997 so that the production of an ethoxy radical—in neat substrate/ TiO2 suspension—requires a more coordinatively unsaturated surface OH group than acetyl radical formation. The failure to detect any radical adducts from the ambient irradiation of neat PBN–ethanol solution is attributed to the absence of a suitable chromophore within the ethanol molecule, indicating that the heterogeneously unaided photodecomposition of ethanol does not occur at any appreciable rate under ambient light conditions.Ambient irradiation of a suspension of deoxygenated TiO2 in 1 : 1 v/v ethanol–acetaldehyde solution with PBN produces the spectrum shown in Fig. 4(B) which is qualitatively the same PBN–adduct spectrum shown in Fig. 4(A) and assigned to the trapping of ethoxy radicals. The signal intensity of PBN–ethoxy radicals produced from ambient irradiation of acetaldehyde/ethanol/TiO2 mixture is 80 times greater than that recorded for the ambient irradiation of ethanol/TiO2.The absence of any acetaldehyde derived PBN–acetyl adducts is due to the higher PBN spin adduction rates reported for alkoxy radicals compared to acyl radicals (ª108 and 106 dm3 mol21 s21, respectively).35 Joyce-Pruden et al.14 have reported the production of ethoxy radicals from the UV irradiation of TiO2/aldehyde/ethanol suspensions, suggesting that the ethoxy radical is one of the most efficient species for preventing electron–hole recombination at the TiO2 surface.The observed 80 fold increase for the trapping of the same radical adduct, upon addition of acetaldehyde to a suspension of TiO2 in ethanol, indicates that efficient retardation of electron–hole recombination is taking place. Due to the difference in PBN spin adduction rates for acetyl and ethoxy radicals, a direct comparison between the signal intensities of the acetaldehyde/TiO2 and acetaldehyde/ethanol/TiO2 systems is not possible.However, the signal intensity ratio of 1 : 80 observed for ethoxy radicals trapped in ethanol/TiO2 and ethanol/acetaldehyde/TiO2 systems suggests that nonband gap mediated processes account for a maximum of only ca. 1.25% of radical forming processes in acetaldehyde/ ethanol/TiO2 suspensions irradiated with ambient sunlight. However, neglecting the rate of photon flux and any molecular geometric requirements for the reaction between a photoexcited substrate molecule and OH(surf), it is likely that the substrate initiated process actually accounts for <1.25% of all radical forming processes when interfacial electron transfer is taking place.The rate of substrate initiated process is then directly proportional to the number of photo-activated substrate molecules located close enough to the TiO2 surface to undergo hydrogen abstraction by OH groups not acting as hole traps.H abstraction by such OH? groups would then be considered a semiconductor sensitised process, occurring with an enhanced rate when the substrate molecule is in the photoexcited state prior to surface interaction. At higher photon fluxes, however, it is conceivable that the pre-adsorption excitation of substrate molecules may make a significant contribution to the overall rate of the heterogeneous photodegradative process. 4. Conclusions The generation of organic substrate derived radical intermediates in the direct sunlight exposure of TiO2 suspensions containing acetaldehyde, ethanol and 50: 50 v/v acetaldehyde– ethanol is confirmed using the spin trapping technique. By carefully controlling the experimental conditions, it is possible to distinguish between the paramagnetic intermediates of oxidative and non-oxidative photolysis. Direct sunlight irradiation of neat deoxygenated acetaldehyde is shown to yield trapped acetyl radicals, whereas irradiation of oxygenated acetaldehyde is believed to result in the trapping of acetoxyl radicals; secondary species formed from the degradation of unstable peroxyacetyl radicals, initial products of the reaction between acetyl radicals and O2.This information then affords the identification of radical species arising in both types of heterogeneous process reported here. In both processes, the primary intermediates trapped are shown to be acetyl radicals, irrespective of the state of oxygenation of the organic/metal–oxide suspension. The inherent instability of peroxyacetyl species generated in oxygenated suspensions, together with the tendency of spin adducts to accumulate at the solid-solution interface, is held to be responsible for this effect.Heterogeneous radical forming processes are demonstrated to be greatly enhanced by coupling the photooxidation of ethanol with the concomitant photoreduction of acetaldehyde, thereby facilitating the complete recycling of an electron within the photoexcited semiconductor.Radical formation in the absence of semiconductor mediated electron transfer is believed to arise as a result of the interaction between basic surface OH groups and photoexcited substrate molecules located at the metal oxide surface. The generation of higher concentrations of acetyl radicals in sunlight irradiated MgO/neat acetaldehyde suspensions, compared to irradiation of TiO2/neat acetaldehyde suspensions, demonstrates unequivocally that no band gap mediated electron transfer is occurring.The enhanced rate of radical production in MgO suspension is most likely due to easier abstraction of aldehydic H atoms, from directly photoexcited aldehyde molecules, by the stronger basic OH groups of the MgO surface. The direct excitation of substrate molecules, vs. initial photoexcitation of the semiconductor, is shown to account for the production of only 1–1.25% of all radical intermediates when direct sunlight is used as the photon source.At higher photon fluxes, however, it is possible that such pre-adsorption excitation of substrate molecules may make a significant contribution to the overall process rate. Acknowledgements C. A. Jenkins would like to thank EPSRC for the provision of Ph.D. funding, and Tioxide UK Ltd. for further financial support. Financial support from EPSRC for funding the National ENDOR Centre (GR/K39554) is gratefully acknowledged. References 1 A.Fujishima and K. Honda, Nature, 1972, 37, 238. 2 A. Wold, Chem. Mater., 1993, 5, 280. 3 W. Choi and M. Hoffmann, Environ. Sci. Technol., 1995, 29, 1646. 4 A. Mills, R. H. Davies and D. W. Worsley, Chem. Soc. Rev., 1993, 417. 5 P. B. Shepson, T. E. Kleindienst, E. O. Edney, C. M. Nero, L. T. Cupitt and L. D. Claxton, Environ. Sci. Technol., 1986, 20, 1008. 6 M. Anpo, T. Shima and K. Kubokawa, Chem. Lett., 1985, 1799. 7 J. Schwitzgebel, J. Ekerdt, H. Gerischer and A. Heller, J. Phys. Chem., 1995, 99, 5633. 8 J. Soria, M. J. Lopez-Munoz, V. Augugliaro and J. C. Conesa, Colloids and Surfaces – A, Physicochemical and Engineering Aspects, 1993, 78, 73. 9 M. X. Tang, P. E. Labinis, S. T. Nguyen, J. M. Kesselman, C. E. Stanton and N. S. Lewis, Prog. Inorg. Chem., 1994, 41, 21. 10 M. Grätzel, Heterogeneous Photochemical Electron Transfer, CRC Press, Florida, 1994, pp. 87–149. 11 R. F. Howe and M. Grätzel, J.Phys. Chem., 1987, 91, 3906. 12 H. Noda, K. Oikawa, H. Ohya-Nishiguchi and H. Kamada, Bull. Chem. Soc. Jpn., 1994, 67, 2031. 13 P. Kamat, Prog. Reaction Kinetics, 1994, 19, 277. 14 C. Joyce-Pruden, K. Li and J. K. Pross, J. Org. Chem., 1992, 57, 5887. 15 E. Coresa, L. Burlamacchi and M. Visca, J. Mater. Sci., 1983, 18, 289. 16 C. Jaeger and A. Bard, J. Phys. Chem., 1979, 83, 3146. 17 C. Von Sonntag, The Chemical Basis of Radiation Biology, Taylor & Francis, London, 1987, pp. 37–38. 18 M. Anpo, N. Aikawa, Y. Kubokawa, M. Che, C. Louis and E. Giamello, J. Phys. Chem., 1985, 89, 5689.J. Chem. Soc., Perkin Trans. 2, 1997 2485 19 Handbook of Organic Chemistry, ed. J. A. Dean, McGraw-Hill, New York, 1987, p. 261. 20 G. Stewart and M. A. Fox, Res. Chem. Intermed., 1995, 21, 933. 21 N. A. Clinton, R. A. Kenley and T. G. Traylor, J. Am. Chem. Soc., 1975, 97, 3746. 22 W. A. Pryor, D. G. Prier and D. F. Church, J. Am. Chem. Soc., 1983, 105, 2883. 23 J. F. Griffiths and G. Shirrow, Oxidation and Combustion Reviews, vol. 3, ed. C. F. H. Tipper, Elsevier, Amsterdam, 1968, p. 47. 24 J. A. Howard, Advances in Free Radical Chemistry, Vol. IV, ed. G. A. Williams, Academic Press, New York, 1972, p. 49. 25 W. A. Pryor, G. Govindan and D. F. Church, J. Am. Chem. Soc., 1982, 104, 7563. 26 A. Halpern, J. Chem. Soc., Faraday Trans. 1, 1987, 83, 219. 27 J. Fossey, D. Lefert and J. Sobra, Free Radicals in Organic Chemistry, Wiley, New York, 1995, p. 96. 28 C. S. Turchi and D. F. Ollis, J. Catal., 1990, 122, 178. 29 X. L. Zhou and J. P. Cowin, J. Phys. Chem., 1996, 100, 1055. 30 A. Ledwith, P. J. Russel and L. H. Sutcliffe, Proc. R. Soc. Lond. A, 1973, 332, 151. 31 A. G. Fadnis, J. Ind. Chem. Soc., 1990, 67, 682. 32 H. A. Edwards, C. C. Rowlands, A. F. Carley, M. W. Roberts, B. Mile, F. E. Hancock and S. D. Jackson, J. Chem. Soc., Faraday. Trans. 1, 1994, 90, 3341. 33 K. M. Schaich and D. C. Borg, Autoxidation in Food and Biological Systems, eds. M. G. Simic and M. Karel, Plenum Press, New York, 1980, pp. 71–88. 34 P. Maillard, J. C. Masset and C. Giannoti, J. Organomet. Chem., 1978, 159, 219. 35 A. N. Saprin and L. H. Piette, Arch. Biochem. Biophys. Res. Commun., 1977, 180, 480. 36 M. J. Davies and G. S. Timmins, Biomedical Applications of Spectroscopy, eds. R. J. H. Clark and R. E. Hester, John Wiley, 1996, pp. 237–241. Paper 7/02944F Received 29th April 1997 Accepted 23rd June 1997

ISSN:1472-779X    

DOI:10.1039/a702944f

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

J. Chem. Soc., Perkin Trans. 2, 1997 2487 Free radicals from cyclic enones: an electron paramagnetic resonance investigation. Part 1. Radicals formed by hydrogen abstraction † Andrew Hudson,*,a Maria Antonietta Della Bona,a Daniel Waterman,a Angelo Alberti,*,b Massimo Benaglia b and Dante Macciantelli b a School of Chemistry, Physics and Environmental Science, University of Sussex, Brighton, UK BN1 9QJ b I.Co.C.E.A. – CNR, Area della Ricerca, Via P. Gobetti 101, I-40129 Bologna, Italy Electron paramagnetic resonance has been used to characterise free radicals formed by the reaction of photochemically generated tert-butoxyl radicals with a range of methyl substituted cyclopentenones and cyclohexenones. The spectra have been interpreted using computer analysis and simulation and assigned to cyclic allylic and alkyl radicals formed by abstraction of a hydrogen atom.The 1-oxocyclohept-2-en-4-yl radical derived by abstraction of bromine from 4-bromocyclohept-2-enone exhibits a temperature dependent EPR spectrum attributed to ring inversion.Introduction The role of anthropogenic and natural biogenic hydrocarbons in tropospheric chemistry 1 has been the subject of much recent research. It is our aim to characterise by EPR spectroscopy radicals formed by hydrogen abstraction from and radical addition to biogenic VOCs (volatile organic compounds) implicated in atmospheric processes; we have already described some radicals derived from monoterpenes.2 The present paper describes some radicals formed by hydrogen abstraction from cyclic enones designed to provide a set of reference data for cyclic allyl radicals.These data should be useful in interpreting the complicated, and often weak, spectra obtained from natural products. Experimental Materials Cyclopent-2-enone 1, 3-methylcyclopent-2-enone 2, 2,3- dimethylcyclopent-2-enone 3, 2,3,4,5-tetramethylcyclopent-2- enone 4, 4,4-dimethylcyclopent-2-enone 5, cyclohex-2-enone 6, 3-methylcyclohex-2-enone 7, 3,5-dimethylcyclohex-2-enone 8, 3,5,5-trimethylcyclohex-2-enone 9 and 4,4-dimethylcyclohex- O O H3 C H3 C O H3 C O CH3 H3C H3C CH3 O CH3 CH3 O H3C O H3C CH3 O H3 C CH3 CH3 O O CH3 H3C O Br 1 2 3 4 5 6 7 8 9 10 11 † Presented at the 30th International Meeting of the Electron Spin Resonance Group of the RSC, University of Lancaster, 6–10th April 1997. 2-enone 10 were purchased from Aldrich, as was di-tertbutyl peroxide. 4-Bromocyclohept-2-enone 11 was prepared as follows. 4-Bromocyclohept-2-enone 11. Cyclohept-2-enone (Aldrich) (0.506 g, 4.53 mmol), N-bromosuccinimide (0.858 g, 4.82 mmol) and dibenzoyl peroxide (Fluka) (0.014 g, 0.057 mmol) were dissolved in 6 ml of carbon tetrachloride and refluxed with stirring for 90 min. The mixture was then cooled, filtered and the filtrate was evaporated under reduced pressure. The residue was chromatographed over silica gel using methylene chloride as eluent. After removing the solvent, pure 4-bromocyclohept- 2-enone 11 was obtained (0.7 g, 3.7 mmol, 81%), dH(CDCl3, internal ref.TMS, J/Hz) 1.7 to 2.9 (6H, m), 4.97 (1H, q, H-4, J4–3 5.1, J4–5 5.0), 5.88 (1H, d, H-2, J2–3 12.6), 6.48 (1H, dd, H-3, J3–2 12.6, J3–4 5.1); m/z 190, 188 (M1), 109 (M1 2 Br), 81 (M1 2 Br 2 CO). EPR experiments The spectra were recorded on Bruker ER-200D and Varian E104 spectrometers using previously described procedures.3 The enones were mixed with di-tert-butyl peroxide in a ratio of 1 : 3 by volume, in quartz EPR tubes. Oxygen was removed by purging the solutions with nitrogen for between 5 and 20 min and then cooled to between 245 and 250 8C.The solutions were photolysed in situ by focusing ultraviolet light from a 1 kW high pressure mercury–xenon lamp into the cavity. Infrared radiation was removed by passing the light through a water- filled quartz window. The EPR spectrometers were coupled to computers and the results were digitised and recorded using a spectral manipulation program.This allowed the spectra to be Fourier filtered and smoothed, if necessary, before interpretations were carried out. Results and discussion Radicals from cyclopentenones Only abstraction from the 4-position in cyclopent-2-enone 1 will produce a radical that is allylic in character. Fig. 1 shows the spectrum obtained along with a simulation. The assignment is straightforward. The coupling constants of the allylic moiety are much as expected; 4 the rather small coupling to the methylene group is noteworthy and will be the subject of later discussion.An alternative description would be to regard the system as an oxapentadienyl radical but the coupling constants suggest that the radical is best regarded as carbon-centred and2488 J. Chem. Soc., Perkin Trans. 2, 1997 essentially allylic with some delocalisation onto the carbonyl group. The radical formed by hydrogen abstraction from 3-methylcyclopent- 2-enone 2 supports the above interpretation of the cyclopent-2-enone radical.The presence of a well resolved 2.4 G quartet in the EPR spectrum clearly indicates that abstraction has occurred from the 4-position and not from the 3- methyl group. In the case of 2,3-dimethylcyclopent-2-enone, the introduction of a further methyl group produces some variation in the couplings. The methyl group in the 2-position has a coupling of 11.1 G which is less than is observed for an a-hydrogen in the same position.Similarly the a-hydrogen coupling in the 4-position is slightly larger than the previous couplings. This indicates some shifting of the electron density away from the 2-position. We also note the lower (5.45 G) splitting from the methylene hydrogens. However, the couplings in the radical from 2,3,4,5- tetramethylcyclopent-2-enone are much as expected. The substrate in this case was actually a mixture of isomers in which the 4- and 5-methyl groups are cis and trans to each other but both will yield the same radical on abstraction of hydrogen from the 4-position.Fig. 1 The experimental EPR spectrum of the radical obtained by hydrogen abstraction from cyclopent-2-enone (a) and the computersimulated spectrum of the 1-oxocyclopent-2-en-3-yl radical (b) H H O H • H H (13.5 G) (6.7 G) (12.5 G) (2.5 G) H H3C O H • H H (13.8 G) (7.45 G) (12.3 G) (2.4 G) H H3C O H3C • H H (14.1 G) (5.45 G) (11.1 G) (2.48 G) The final example containing a five-membered ring is 4,4- dimethylcyclopent-2-enone. Hydrogen abstraction in this case is distinctive since the system has no abstractable hydrogens which will form an allylic system.Fig. 2 shows the spectrum and simulation. The following parameters were assigned to the radicals as shown. The a-proton coupling of 19.4 G is comparable with that observed in the acetonyl radical.5 Radicals from cyclohexenones As expected, cyclohex-2-enone forms an allylic radical by abstraction of hydrogen from the 4-position. The allylic couplings are comparable with those found in five-membered rings but the methylene coupling is much larger than the 6–8 G observed in the cyclopentenone radicals.This difference is an example of the effect of orbital symmetry on methylene hyper- fine couplings as was first explained by Whiffen.6 He showed that the proton hyperfine splitting for a CH2 group bridging two p-centres depends on (c1 1 c2)2 where c1 and c2 are the linear combination of atomic orbital (LCAO) coefficients of the neighbouring p-centres in the singly occupied molecular orbital (SOMO).In the cyclopentenone radicals we have a p-system over five centres with the methylene group spanning C1 and C4. Fig. 2 Experimental EPR spectrum of the radical formed by hydrogen abstraction from 4,4-dimethylcyclopent-2-enone (a) and the computersimulated spectrum of the 4,4-dimethyl-1-oxocyclopent-2-en-5-yl radical (b) CH3 H3C O H3C • H CH3 (13.9 G) (1.0 G) (11.9 G) (3.0 G) O H • (1.6 G) (19.4 G) (1.6 G) H CH3 CH3 (1.6 G) H H H O H H H H • (13.2 G) (15.9 G) (0.8 G) (12.9 G) (3.2 G)J.Chem. Soc., Perkin Trans. 2, 1997 2489 In a pentadienyl radical C1 would be at a nodal position in the SOMO. However, simple Huckel calculations show that replacing the terminal carbon in pentadienyl by oxygen leads to a SOMO in which the coefficients at positions 1 and 4 are of opposite sign, thus explaining the small coupling.An analogous situation has recently been noted in the spectra of some 1,2-indansemidiones and we refer to Strom’s paper 7 for a more detailed discussion. In the six-membered ring the methylene coupling depends only on the spin density on C4 and a normal value is observed. A similar pattern of coupling constants is found for the radical derived from 3-methylcyclohex-2-enone. In 3,5-dimethylcyclohex-2-enone the hydrogens in the 4- position are still abstractable to form a delocalised radical.The splittings suggest that spin density has shifted away from the 2-position and towards the 4-position as illustrated by the imbalance of the couplings of the two hydrogens of 11.1 G and 16.45 G. The 2.7 G coupling from the 3-methyl group is unexceptional. The hydrogen coupling of 19.9 G in the 5- position indicates a conformation in which the hydrogen is in an out-of-plane position. The triplet of 1.1 G is comparable with that attributed to other hydrogens in the 6-position.Isophorone, 3,5,5-trimethylcyclohex-2-enone, introduces a further methyl group in the 5-position. Computer simulation leads to the assignment shown. Enlarging the detail of the spectrum reveals the presence of a further septet splitting of 0.47 G. This splitting is assigned to the magnetically equivalent methyl groups on the 5-position carbon. The hydrogens in the 6- position do not appear to give a resolvable coupling although they might have been expected to exhibit an interaction of comparable size to the methyl groups.We report no coupling for the two hydrogens as the seven line group of intensities 1:6:15:20:15:6:1 fits considerably better than a group of nine lines with ratios 1:8:28:56:70:56:28:8:1 that would be produced by eight protons. Finally we consider the six-membered ring analogue of 4,4-dimethylcyclopent-2-enone. 4,4-Dimethylcyclohex-2-enone gives a simple six line spectrum consisting of a doublet of triplets.As expected the b-hydrogens are producing the largest coupling. H H3C H O H H H H • (13.2 G) (15.4 G) (0.8 G) (11.9 G) (2.3 G) H H3C H O H H CH3 H • (16.45 G) (19.9 G) (1.1 G) (11.1 G) (2.7 G) H H3C H O H CH3 CH3 H • (12.69 G) (0.47 G) (12.9 G) (2.5 G) (0.47 G) H H O H H • (29.5 G) (18.7 G) H CH3 H3C Radical from cycloheptenone Photolysis of a deoxygenated di-tert-butyl peroxide solution of cyclohept-2-enone in the temperature range 250 to 0 8C failed to afford any detectable EPR signal. On the other hand, photolysis of argon purged tert-butylbenzene solutions of 4- bromocyclohept-2-enone containing some hexabutylditin led to spectra whose pattern varied drastically with temperature (see Fig. 3) and which, although very weak, could be attributed to the cyclic allyl radical resulting from bromine abstraction. The room temperature spectrum results from coupling of the unpaired electron with the protons in positions 2, 3 and 4 and with the two protons in position 5 which appear to be equivalent, owing to inversion of the cycloheptenyl ring at a rate which is fast compared with the difference in hyperfine coupling constants.At lower temperatures the inversion process slows down and the ca. 18 G triplet is replaced by a ca. 36 G doublet. This new doublet is in fact the sum of the couplings of the two protons in positions 5 when the rate of inversion is not fast enough to make them equivalent and yet not slow enough to make them completeley inequivalent.The computer simulation of the spectra at the different temperatures would allow the determination of the energy barrier to inversion, provided the couplings of the two exchanging atoms are known for the slow exchange limit. Unfortunately such determination was impossible. In fact, further lowering of the temperature resulted in the disappearance of the spectrum, and it proved impossible to reach a situation when the inversion is completely frozen out thus allowing the determination of the couplings of the two individual hydrogens.However, we were able to obtain spectra at 263, 273, 300 and 313 K and simulate them, assuming that the coupling constants exchange between values of 0 and 36 G. An Arrhenius plot then leads to an activation energy of ca. 7 kcal mol21. This procedure yields the correct barrier but not the preexponential factor provided the spectra are in the fast exchange region where the broadening is proportional to Da2.There do not seem to be any comparable systems containing seven-membered rings in which ring inversion has been investigated; it may be worth noting that the cycloheptyl radical is much more flexible,8 owing to the absence of the C]] C double bond. The closest analogy is with the cyclohex-3-enyl radical where studies of line-width alternation have yielded a barrier Fig. 3 Experimental EPR spectra obtained by bromine abstraction from 4-bromocyclohept-2-enone at 240 8C (a) and at room temperature (c) and computer-simulated spectra of the 1-oxocyclohept-2-en- 4-yl radical [(b) and (d )].The line marked with an asterix in spectrum (c) is due to an impurity of the sample tube. H H H O H H H H H H H O H H H H H H H H (12.45 G) (4.23 G) (13.25 G) (17.98 G) • •2490 J. Chem. Soc., Perkin Trans. 2, 1997 for ring inversion of about 30 kJ mol21.9,10 In that case the allylic system is essentially rigid and the hydrogen atoms in positions 4 and 6 are rendered magnetically equivalent by inversion of the methylene group in the 5-position through the C1]C2]C3 plane.Conclusions We have shown that it is possible to observe and analyse EPR spectra from primary radicals formed by hydrogen abstraction from a series of cyclic enones. In general the single species observed is an allylic radical stabilised by electron delocalisation. In the absence of allylic hydrogen atoms acetonyl radicals are formed. A subsequent paper will describe a study of radical addition reactions to cyclopentenones and cyclohexenones. Acknowledgements We thank Dr R. A. Jackson and Dr Marco Lucarini for the use of their computer programmes, the EPSRC for the award of a research studentship to D. W., the University of Bologna for the award of a training grant to M. A. D. B., and CNR for the award to A. H. of a short-term mobility fellowship. References 1 Volatile Organic Compounds in the Atmosphere, ed. R. H. Hester and R. M. Harrison, Royal Society of Chemistry, Cambridge 1995. 2 A. Hudson D. Waterman and A. Alberti, J. Chem. Soc., Perkin Trans. 2, 1995, 2091. 3 J. Levillain, S. Masson, A. Hudson and A. Alberti, J. Am. Chem. Soc., 1993, 115, 8444. 4 J. K. Kochi and P. J. Krusic, J. Am. Chem. Soc., 1968, 90, 7157. 5 H. Zeldes and R. A. Livingston, J. Chem. Phys., 1966, 45, 1946. 6 D. H. Whiffen, Mol. Phys., 1963, 6, 223. 7 E. T. Strom, J. Org. Chem., 1995, 60, 5686. 8 C. J. Burkey, D. Griller and R. Sutcliffe, J. Org. Chem., 1985, 50, 1138. 9 N. M. de Tannoux and D. W. Pratt, J.Chem. Soc., Chem. Commun., 1978, 394. 10 J. A. Berson, D. Griller, K. Owens and D. D. M. Wayner, J. Org. Chem., 1987, 52, 3316. Paper 7/02510F Received 11th April 1997 Accepted 9th July 1997

ISSN:1472-779X    

DOI:10.1039/a702510f

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

J. Chem. Soc., Perkin Trans. 2, 1997 2491 Uranium reactions with hydrogen peroxide studied by EPR–spin trapping with DMPO† Matthew M. Hamilton, John W. Ejnik and Alasdair J. Carmichael Applied Cellular Radiobiology Department, Armed Forces Radiobiology Research Institute, Bethesda, Maryland 20889–56033, USA Uranyl nitrate [UO2(NO3)2] (1.0 × 1023 M) is reacted with hydrogen peroxide (H2O2, 5 × 1023 M) in the presence of the spin trap 5,5-dimethyl-4,5-dihydro-3H-pyrrole N-oxide (DMPO, 5 × 1022 M) in acidic solutions.The reaction generates a 1:2:2:1 quartet with hyperfine coupling constants, aN = aH b = 1.50 mT. These values are consistent with reported values for the DMPO]OH spin adduct. It is possible that the uranous ion (UO21), which corresponds to UIV, generates hydroxyl radicals (?OH) to give the observed DMPO]OH. This is suggested by two observations: (1) the intensity of the EPR spectrum is dependent on pH, reaching a maximum at pH = 0.6, which is consistent with the formation of UO21 from UO2 21 at lower pH values; and (2) the uranyl ion (UO2 21) corresponds to UVI and cannot be further oxidized.To determine whether ?OH radicals generate DMPO]OH, the reaction is carried out in the presence of varying concentrations of ethanol (EtOH). The results indicate that at a low ratio of [EtOH]/[DMPO] the EPR signal corresponds to DMPO]OH, while at a [EtOH]/[DMPO] ratio ca. 1 the signal is mixed and equally intense for DMPO]OH and the hydroxyethyl adduct to DMPO (DMPO–EtOH).When the [EtOH]/[DMPO] ratio is ca. 10 the EPR signal is mainly that of the DMPO–EtOH adduct. This confirms that uranium reacts with H2O2 generating hydroxyl radicals. Introduction The isotopic composition of naturally occurring uranium is 238U (99.28%), 235U (0.718%) and 234U (0.0056%). Depleted uranium is obtained as a byproduct in the enrichment process during the production of nuclear fuel. It is also obtained in the recycling of spent fuel.The word depleted refers to depletion of the 235U isotope, thus raising the purity of 238U from 99.28 to 99.8%. Although containing low radioactivity (0.4 mCi g21), depleted uranium is continuously emitting a-particles, bparticles and g-rays. In the decay process from 238U to 234U depleted uranium emits two a-particles (ª4.2 MeV). The b (0.076–2.8 MeV) and g (0.0633–1.001 MeV) components are less significant.1 Therefore, the toxicity of uranium may originate from two sources, one having a chemical component and the other a radiological component. The toxicity of uranium has been studied since the discovery of nuclear fission.2–6 These studies investigated the effects of uranium either by inhalation, ingestion or parenteral administrations of uranium compounds ranging from insoluble oxides to soluble salts.There is little or no information of the toxicity of embedded depleted uranium in its metallic form.7 It is generally assumed that chemical toxicity originates from dissolved uranium.However, radiological toxicity can originate from metallic and dissolved uranium. There are several questions that need to be addressed when uranium is ingested or finds its way into the body in a metallic form: (1) Will this uranium dissolve? (2) Does the oxidation state of uranium play a role in the uranium toxicity? (3) What are the mechanisms of transport to the target tissues? (4) Will the toxicological effects be chemical or radiological in nature? Uranium’s most stable and soluble state is in its sixth and highest oxidation state, UVI.This form of uranium is usually associated with oxygen as the uranyl cation, UO2 21. However, in solution UO2 21 is usually in equilibrium with small quantities of UIV in the form of the uranous cation, UO21. The equilibrium can readily be shifted toward UO21 in the presence of acid or a mild reducing agent (sodium hydrosulfite).8 Due to the † Presented at the 30th International Meeting of the Electron Spin Resonance Group of the RSC, University of Lancaster, 6–10th April 1997. reducing environment inside cells, it is possible that UIV as the UO21 cation is the predominant form of ingested or embedded dissolved uranium.Furthermore, the cycling between UVI and UIV may lead to various reactions involving free radical intermediates which may be detrimental to biological systems. One such reaction is the reaction with hydrogen peroxide (H2O2) to yield hydroxyl radicals (?OH).Several other metal ions (e.g. Fe21 and VO21) react in this way with H2O2 generating ?OH in what is known as a Fenton-type reaction.9 It was of interest to investigate whether uranium reacts in a similar fashion with H2O2. The reactions were studied by EPR–spin trapping. Spin trapping is a technique which uses a spin trap, usually a nitrone or nitroso compound, to trap short lived free radicals generating a longer lived aminoxyl ‡ spin adduct which can be measured and identified by EPR.In this work the reaction between uranium and H2O2 was studied in the presence of the nitrone 5,5-dimethyl-4,5-dihydro-3H-pyrrole N-oxide (DMPO). Materials and methods Uranyl nitrate [UO(NO3)2], vanadyl sulfate [VOSO4], ferrous ammonium sulfate [FeSO4?(NH4)2SO4?6H2O] and potassium permanganate (KMnO4) were obtained from Fisher Scientific Co. (Fair Lawn, NJ). Sodium oxalate (Na2C2O4) was obtained from J. T. Baker Chemical Co. (Phillipsburg, NJ).The concentration of UO(NO3)2 was determined by phosphorescence using a Kinetic Phosphorescence Analyzer (KPA) (Chemchek, WA).10 The concentration of VO21 in the VOSO4 solution was determined spectrophotometrically (l = 750 nm, e = 18 M21 cm21).11 KMnO4 was standardized against measured quantities of dried Na2C2O4. FeSO4?(NH4)2SO4?6H2O and hydrogen peroxide (H2O2) were titrated with the standard KMnO4 solution to determine their concentration.12 H2O2, sodium hydrosulfite (Na2S2O4) and Sephadex DEAE were obtained from Sigma (St.Louis, MO). The spin trap DMPO and nitric acid (HNO3) were obtained from Aldrich Chemical Co. (Milwaukee, WI). The concentration of DMPO was determined spectro- ‡ Formerly known as nitroxide.2492 J. Chem. Soc., Perkin Trans. 2, 1997 photometrically (l = 227 nm, e = 8 × 1023 M21 cm21).13 The DMPO was determined free of radical impurities by EPR spectroscopy. Solutions of UO(NO3)2 (UIV) were prepared by reducing UO2(NO3)2 with Na2S2O4.The excess S2O4 22 was removed by addition of the anion exchange resin Sephadex DEAE. Experiments were carried out under air-saturated conditions, mixing metal ion solutions (UO2 21/UO21, VO21 or Fe21), DMPO and ethanol (when required) immediately prior to addition of the appropriate amount of hydrogen peroxide. Controls containing uranium were also carried out under air-free nitrogen saturated solutions following the procedure described by Russell et al.14 and Evans.15 The pH of the metal ion solutions were adjusted with HNO3 prior to addition of other reactants to the mixture.Solutions of UO(NO3)2 are usually pH 3.6. After addition of the H2O2, the samples were rapidly mixed and transferred to an EPR quartz flat cell (60 × 10 × 0.25 nm) and their EPR spectra recorded. The EPR quartz flat cells were soaked in a concentrated mixture of HNO3–H2SO4 (1 : 1), to eliminate all metal ion contamination. These were rinsed with deionized water from a Sybron Barnstead NANO pure system.This water was also used to make the various solutions required for the experiments. All other labware used was plastic and required no further treatment. All EPR spectra were recorded on a Varian E-9 X-band spectrometer at 100 kHz magnetic field modulation. The magnetic field was set at 338.50 mT; microwave power, 10 mW; modulation amplitude, 0.1 mT; microwave frequency, 9.510 GHz; time constant, 0.5 s; scan time, 4 min. The hyperfine coupling constants of the spin adducts (DMPO]OH and DMPO–EtOH) were obtained by computer simulation, generating a theoretical EPR spectrum that matched the experimental spectrum.The chemical structures of DMPO, DMPO]OH and DMPO– EtOH are shown. Results and discussion When H2O2 is mixed with a UO2(NO3)2 solution containing spin trap DMPO, an EPR spectrum consisting of 1:2:2:1 quartet is obtained (Fig. 1). Figs. 1(a)–1(c) are the control experiments yielding EPR spectra of DMPO in 250 mM HNO3 [Fig. 1(a)], UO2(NO3)2 mixed with 250 mM HNO3 [Fig. 1(b)] and H2O2 mixed with 250 mM HNO3 [Fig. 1(c)]. The pH of these solutions was 0.6. Fig. 1(d) shows the results obtained at pH 0.6 after mixing H2O2 (5 mM) with a UO2(NO3)2 (1 mM) solution containing DMPO (50 mM). The 1:2:2:1 quartet with hyperfine coupling constants, aN = aH b = 1.50 mT, corresponds to the DMPO]OH spin adduct.16 The intensity of the EPR spectra from the solutions [Figs. 1(a)–1(c)] resulted in less than 10% of the intensity observed in Fig. 1(d). The formation of ?OH in the reaction of uranium ions and hydrogen peroxide can be by three reaction mechanisms: (1) the radiation emitted by uranium (a-particles, g-rays) reacting with water to yield ?OH [reaction (1)]; (2) the reaction of uranium UO2 21V a,g V H2O æÆ ?OH, ?H, e2 aq (1) ions with H2O2 to yield ?OH in a Fenton-type reaction [reactions (2a) and (2b)]; and (3) the reaction of UVI in UO2(NO3)2 UO2 21 1 2H1 1 2e2 æÆ UO21 1 H2O (2a) N H3C H3C H O N H3C H3C O• N H3C H3C O• OH CHOHCH3 H OH DMPO DMPO–OH DMPO–EtOH UO21 1 H2O2 æÆ UO2 21 1 ?OH 1 H1 (2b) with dissolved oxygen to yield superoxide anion radicals (O2~2) [reaction (3)].Superoxide reacts directly with DMPO to form an unstable adduct (DMPO]O2 2) which decomposes forming DMPO]OH. UO2 21 1 4H1 æÆ U51 1 2H2O (3a) U51 1 O2 1 2H2O æÆ UO2 21 1 O2~2 1 4H1 (3b) Several experiments were conducted to obtain insight on the actual reaction mechanism. Since it is known that the formation of UIV is facilitated at lower pH, the DMPO]OH EPR signal intensity as a function of pH was plotted (Fig. 2). Fig. 2 shows the peak-to-peak heights of the low field EPR line plotted as a function of pH. Provided that the EPR line shape does not change from sample to sample, the peak-to-peak height is proportional to the actual EPR signal intensity. The EPR signal intensity is obtained by double integration of the complete spectrum and is directly proportional to the concentration of the species being measured. Therefore, the peak-to-peak height of an EPR spectrum is also related to the species concentration, providing that the EPR line shapes remain constant from spectrum to spectrum.In addition, it must be noted that there are two competing reactions occurring over the pH range plotted, the formation of DMPO]OH and the decay of the DMPO]OH due to its instability at lower pH values.When the rate of form- Fig. 1 (a)–(c) Controls. (d) Addition of H2O2 to a solution of UO2(NO3)2 in the presence of DMPO (50 mM). Receiver gain 1.25 × 104. Fig. 2 DMPO]OH EPR signal intensity plotted as a function of pHJ. Chem. Soc., Perkin Trans. 2, 1997 2493 ation of DMPO]OH is faster than the rate of decay, the DMPO]OH EPR signal intensity will increase, however, when the rate of decay is faster than the rate of formation the EPR signal intensity will decrease. For these reasons, the profile (Fig. 2) increases reaching a maximum at pH 0.6, then rapidly decreases due to the short lifetime (instability) of the DMPO]OH adduct in extreme acidic solutions. As the equilibrium between UVI and UIV is shifted toward the formation of UIV in acidic environments, the results suggest that the observed DMPO]OH originates from the reaction of UO21 (UIV) with H2O2 to form ?OH [reaction (2)]. Furthermore, the oxygen exchange between oxyions and metal ions is faster in acid because of the enhanced reactivity of protonated oxyions.17 Thus, the rate of formation of hydroxyl radicals should increase with increasing acidity of the solution.This is consistent with the observed increase in the DMPO]OH signal as the pH is lowered (Fig. 2). The pH profile shown in Fig. 2 virtually eliminates the radiological mechanism shown in reaction (1), which is independent of pH. It would be expected that if the formation of DMPO]OH originates from the ?OH generated in the radiation of water, the EPR of DMPO]OH would have equal intensity under any conditions.The emission of a-particles, g-rays and bparticles is continuously occurring, as part of the natural decay of uranium, and therefore the production of DMPO]OH would remain equal and would not increase as the pH is lowered (Fig. 2). A similar radiological mechanism was also discarded for plutonium. The oxidative damage caused by plutonium was found not to be due to a-particle emission by this element.18 The mechanism in reaction (3) is also unlikely for two reasons: (1) the uranyl ion corresponds to the highest oxidation state or uranium (UVI) and cannot transfer an electron directly to oxygen unless the uranium is in a lower oxidation state [reactions (3a) and (3b)]; and (2) the formation of the DMPO]OH adduct occurs significantly only in the presence of H2O2.Furthermore, in the control experiments shown [Fig. 1(b)] the intensity of the DMPO]OH EPR signal is identical in the presence and absence of oxygen (nitrogen saturated). This leaves the mechanism in reactions (2a) and (2b) as the likely explanation for the observed formation of the DMPO]OH adduct.Since UO2(NO3)2 corresponds to UVI, the highest valence state of uranium, it is not possible for UO2 21 to react directly with H2O2 and form ?OH. For this reason in addition to the pH profile [Fig. 2], the mechanism in reaction (2) was further investigated by reducing the UVI to UIV using Na2S2O4 and then reacting the resulting UIV with H2O2.Na2S2O4 is known to specifically reduce UVI to UIV. The solution first takes on a pink color due to UV, then turns green due to the formation of UIV, at which point there is no further reduction to the lower oxidation states.8 The purpose of the Fig. 3 DMPO]OH EPR at pH 0.6 as a function of varying concentrations of Na2S2O4. UO2(NO3)2 concentration was 1.0 mM, DMPO concentration was 50 mM, H2O2 concentration was 5 mM.Receiver gain 2.5 × 104. experiment was to reduce all uranyl ions UO2 21 (UVI) to uranous ions UO21 (UIV) prior to reacting it with hydrogen peroxide in the presence of DMPO. It would be expected that if the formation of DMPO]OH originated from ?OH produced in the reaction of UO21 (UIV) with H2O2 [reaction (2)], the DMPO]OH EPR signal intensity would significantly increase after addition of Na2S2O4. Fig. 3 shows the results after adding Na2S2O4 (various concentrations) to solutions containing 1 mM UO2(NO3)2 and 50 mM DMPO and reacting these mixtures at pH 0.6 with 5 mM H2O2.These results suggest that the Na2S2O4 also reduced the aminoxyl to the hydroxylamine, therefore eliminating the EPR signal altogether at higher concentrations of Na2S2O4. To verify this, the experiment was repeated in the presence of an anion exchange resin (Sephadex-DEAE) in order to chelate all excess of S2O4 22. The results in Fig. 4 show that after Sephadex-DEAE treatment of a solution at pH 0.6 containing 1 mM UO2(NO3)2, 1 mM Na2S2O4 and 50 mM DMPO, and subsequently reacting the supernatant mixture with 5 mM H2O2, a significant increase (approximately two-fold) in the DMPO]OH EPR signal intensity was observed [Fig. 4(b)].Fig. 4(a) is the control experiment carried out in the absence of Na2S2O4. The results in Fig. 4 strongly suggest that the direct reaction of UO21 (UIV) with H2O2 is occurring, as shown in reaction (2).A free radical scavenging experiment using ethanol (EtOH) was carried out to verify that the DMPO]OH spin adduct originates from the direct formation of ?OH and its reaction with DMPO. EtOH and DMPO react with ?OH at approximately equal rates (k > 109 M21 s21).19 The reaction of EtOH with ?OH yields the hydroxyethyl radical (EtOH?). DMPO competes for the ?OH and the EtOH? formed when the reaction between UO21 and H2O2 is carried out in the presence of EtOH. Furthermore, the relative EPR intensities of the DMPO]OH and DMPO]hydroxyethyl (DMPO–EtOH) adducts vary depending on the [DMPO]/[EtOH] ratio in the solution.Fig. 5 shows the results obtained after adding 5 mM H2O2 to a solution containing 1 mM UO2(NO3)2, 50 mM DMPO and varying concentrations of EtOH. The EPR spectra from these solutions change from the DMPO]OH spectrum [Fig. 5(a)] to a mixed DMPO]OH/DMPO–EtOH [Fig. 5(b)] to predominantly the DMPO–EtOH spectrum [Fig. 5(c)]. In Fig. 5(a), [EtOH]/ [DMPO] = 0.The ratio [EtOH]/[DMPO] ª 1 [Fig. 5(b)] and [EtOH]/[DMPO] ª 10 [Fig. 5(c)], respectively. The EPR spectra, consisting of a triplet of doublets [Fig. 5(b) and 5(c)], correspond to the DMPO–EtOH adduct. This was verified by com- Fig. 4 Sephadex-DEAE treatment of the UO2(NO3)2 solutions (1.0 mM, pH 0.6) (a) control, without Na2S2O4 and (b) after addition of Na2S2O4 (1.0 mM) and subsequently adding DMPO (50 mM) and H2O2 (5 mM) to the supernatant mixture. Receiver gain 2.5 × 104.2494 J.Chem. Soc., Perkin Trans. 2, 1997 puter simulation using hyperfine coupling constants aN = 1.58 and aH b = 2.28 mT, which correspond to the reported values for the DMPO–EtOH adduct.16 The results obtained in Fig. 5 con- firm that ?OH is formed in the reaction between uranium ions and H2O2 and this reaction occurs via the mechanism described in reactions (2a) and (2b). It is of interest to determine how the uranium ions compare with other known metal ions that participate in Fenton-type Fig. 5 ?OH scavenging experiments at pH 0.6 using varying concentrations of ethanol (EtOH). [DMPO] = 50 mM; (a) [EtOH]/[DMPO] = 0; (b) [EtOH]/[DMPO] = 1; (c) [EtOH]/[DMPO] = 10. UO2(NO3)2 and H2O2 concentrations were 1.0 mM and 5 mM respectively. R. Gain 5.0 x 104 1.0 mT R. Gain 2.5 x 104 (a) (b) (c) Fig. 6 Capacity of various metal ions to generate ?OH at pH 0.6. (a) UO2 21/UO21; (b) Fe21; (c) VO21. The concentration of metal ions in each solution was 1.0 mM.The final concentration of H2O2 added to each solution was 5 mM. 1 mM Uranyl Nitrate R. Gain 2.5 x 104 1 mM Ferrous ammonium sulfate R. Gain 1.0 x 104 1 mM Vanadyl sulfate R. Gain 5.0 x 104 1.0 mT (a) (b) (c) reactions. For this reason uranium was compared with Fe21 and VO21. The results are shown in Fig. 6. The reactions were carried out at pH 0.6 under the same conditions with regard to metal ion (1 mM) concentration, DMPO (50 mM) and H2O2 (5 mM) concentrations.The results show that the relative order of the capacity to generate ?OH for the three metal ions is VO21 > Fe21 > UO21. In addition, from the EPR intensities of the DMPO]OH spectra (Fig. 6) VO21 is approximately twice as efficient as Fe21, which in turn is approximately twice as efficient as UO21 in their capacity to generate ?OH in Fentontype reactions. In a previous report comparing the capacity of VO21 and Fe21 to generate ?OH in Fenton-type reactions at pH 5.5, the order was reversed and Fe21 > VO21 by a factor of two.9 For biological and toxicological purposes, the knowledge that uranium participates in a Fenton-type reaction is important.Furthermore, the equilibrium between UO2 21 (UVI) and UO21 (UIV) is interesting because it suggests that cycling between UVI and UIV is possible and requires only a reactant such as H2O2 to drive the reaction and cycle. Although the valence state of uranium in biological systems remains to be determined, uranium( IV) is the likely species in the cells reducing environment.It must be kept in mind that the uranium reactions described in this work were carried out using solutions of plain inorganic salts in nitric acid. It remains to be determined if this property is true when these metal ions are chelated by complex biological systems and at a higher pH. References 1 B. Shleien, The Health and Radiological Health Handbook, Scinta, Silver Spring, MD, 1992, p. 283. 2 (a) G. L. Diamond, Radiant. Prot. Dosim., 1989, 26, 23; (b) G. L. Diamond, P. E. Morrow, B. J. Panner, R. M. Gelein and R. B. Baggs, Fundam. Appl. Toxicol., 1989, 13, 65. 3 Y. D. La Touche, D. L. Willis and O. I. Dawydiak, Health Physics, 1987, 53, 147. 4 P. Morrow, R. Gelein, H. Beiter, J. Scott, J. Picano and C. Yuile, Health Physics, 1982, 43, 859. 5 (a) A. Ortega, J. L. Domingo, M. Gomez and J. Corbella, Pharmacol. Toxicol., 1989, 64, 247; (b) A. Ortega, J. L. Domingo, J. M. Llobet, J. M. Thomas and J. L. Paternain, Bull. Environ. Contam. Toxicol., 1989, 42, 935. 6 M. E. Wrenn, J. Lipszten and L. Betrtelli, Radiat. Prot. Dosim., 1989, 26, 243. 7 C. A. Castro, K. A. Benson, E. G. Daxon, J. B. Hogan, H. M. Jacocks, M. R. Laudauer, S. A. McBride and C. W. Shehata, Armed Forces Radiobiology Research Institute, Technical Report, #96-3, Bethesda, MD, 1996. 8 C. Voegtlin and H. C. Hodge, Pharmacology and Toxicology of Uranium Compounds, 1949, vol. 1, part 1, p. 91. 9 A. J. Carmichael, Free Radical Res. Commun., 1990, 10, 37. 10 B. Brina and A. G. Miller, Anal. Chem., 1992, 64, 1413. 11 J. J. Fitzgerald and N. D. Chasteen, Anal. Biochem., 1974, 60, 170. 12 I. M. Kolthoff, E. B. Sandell, E. J. Meehan and S. Bruckenstein, Quantitative Chemical Analysis, The Macmillan Company, New York, 1969, 4th edn., pp. 828 and 834. 13 B. Kalyanaraman, C. C. Felix and R. C. Sealy, Photochem. Photobiol., 1982, 36, 5. 14 G. A. Russell, E. G. Janzen and E. T. Strom, J. Am. Chem. Soc., 1964, 86, 1807. 15 C. A. Evans, Aldrichimica Acta, 1979, 12, 23. 16 G. R. Buettner, Free Radical Biol. Med., 1987, 3, 259. 17 R. G. Wilkins, Kinetics and Mechanisms of Reactions of Transition Metal Complexes, VCH, New York, 1991, p. 41. 18 H. G. Claycamp and D. Luo, Radiat. Res., 1994, 137, 114. 19 Farhataziz and A. B. Ross, NSRDS-NBS59. U.S. Govt. Printing Office, Washington, DC, 1997. Paper 7/02509B Received 11th April 1997 Accepted 29th July 1997

ISSN:1472-779X    

DOI:10.1039/a702509b

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

J. Chem. Soc., Perkin Trans. 2, 1997 2495 Role of cytokine and nitric oxide in the inflammatory response produced by sulfur mustard (HD)† Carmen M. Arroyo *,a and Alasdair J. Carmichael b a Drug Assessment Division, U.S. Army Medical Research Institute of Chemical Defense, 3100 Ricketts Point Road, Aberdeen Proving Ground, MD 21010-5425 b Applied Cellular Radiobiology Department, Armed Forces Radiobiology Research Institute, Bethesda, Maryland 20889-5603 We have determined by immunocytochemistry the levels of interleukin-1 beta; (IL-1‚) in cultured human epidermal keratinocytes (NHEK) following exposure to HD.Human skin keratinocytes release significant numbers of IL-1 cytokine as determined by the QuantikineTM Interleukin-1‚ kit, an enzyme-linked immunosorbent assay (ELISA) procedure. Exposure of NHEK [~106–107 cells, to HD (2 mM) and preincubation for 3 h at 37 8C] results in significant changes in IL-1 activation. In neonatal NHEK exposed to HD, IL-1‚ is decreased.Conversely, in adult breast NHEK exposed to HD, IL-1‚ is increased. Nitric oxide (?NO) has been implicated as the effector molecule that mediates IL-1‚. To confirm the involvement of ?NO in the expression of the IL-1‚, electron paramagnetic resonance (EPR) spectroscopy was employed. EPR detectable iron–nitrosyl complex in NHEK exposed to HD (18 h post exposure to 1 mM HD) were measured, and the generation of ?NO and this induced complex was blocked by N¢-nitro-L-arginine (L-NOARG), a competitive inhibitor of nitric oxide synthase (NOS).Our results show the release of nitric oxide during IL-1 cytokine expression when keratinocytes are exposed to HD. Based upon this work, it appears possible that IL-1 could be used as a specific marker for epidermal cytoxicity in mechanistic studies of the toxicity of HD and in defining interventive and therapeutic regimens against HD vesication. The experiments described in this work deal with the effects of sulfur mustard (HD) on normal human epidermal keratinocytes (NHEK).Keratinocytes are the cells found in the outer layer of the skin and are a primary target of mustard exposure. Analysis of cultured normal human epidermis using specific enzyme-linked immunosorbent assays (ELISA) and bioassays for the cytokine interleukin 1 (IL-1) have led to the conclusion that the average adult harbors 20–60 mg (0.6–1.9 nmol) of IL-1 in his or her epidermis.1 Taking the epidermis as a threedimensional space with a thickness of 0.1 mm and an area of 1.5 m2, the concentration of IL-1 in this space is 4–12 nmol l21.This concentration exceeds the concentration of IL-1 required to activate certain cells by three orders of magnitude (<10 pmol l21). Most keratinocyte interleukin (in vivo and in vitro) is cell associated; relatively little is released from the cell. Keratinocytes can express higher numbers of IL-1 receptors than can any other cell type studied.1,2 A variety of stimuli can enhance IL-1 receptor expression in vitro. These include phorbol esters, calcium and UV irradiation.3 In this report, we have determined by immunocytochemistry levels of interleukin-1 beta (IL-1b) in cultured NHEK following exposure to sulfur mustard (HD).HD is a known alkylating agent causing, among other things, DNA mutations which contribute to cell injury or death. However, the exact mechanism of alkylation remains unclear. What is known is that the action of alkylating agents, such as HD, proceeds through a mechanism involving, as a first step, formation of a cyclic alkylating agent (sulfonium ion) and the release of chloride [eqn.(1)]. The hydrolysis of HD in aqueous media Cl S Cl Cl S + Cl KC H2O (1) Sulfur mustard Cyclic ethylene sulfonium ion – + † Presented at the 30th International Meeting of the Electron Spin Resonance Group of the RSC, University of Lancaster, 6–10th April 1997. at 37 8C is rapid, with t2— 1 a 2 min.4,5 Furthermore, the HD may affect or alter various cell pathways prior to reaching the DNA located in the nucleus.Therefore, the action of alkylation agents on target organ cells may proceed via three principal mechanisms: (i) the physical interaction of the alkylating agent with cellular receptors; (ii) the chemical reaction of the alkylating agent with these receptors; or (iii) either type of metabolite– receptor interaction following metabolism of the alkylating agent within the target organ or elsewhere.Although direct alkylation of DNA and RNA has been widely described, we hypothesize additional alkylation events of potential importance in skin injury. We have determined by immunocytochemistry levels of interleukin-1 beta (IL-1b) in cultures of normal human epidermal keratinocytes (NHEK) following exposure to HD (3 h post exposure to 2 mM HD). This exposure time is sufficient to allow release of a significant quantity of IL-1b.However, it is not long enough for the cells (adults and neonatals) to go through a complete cycle and divide, thus eliminating possible errors due to differences in cell turnover time (~21 h for neonatal-NHEK and ~24 h for adult- NHEK). The HD concentration range was chosen because it is generally observed that exposure to approximately 1 mM HD causes the formation of microblisters, while full blister formation occurs after exposure to approximately 2 mM HD. The expression of IL-1b in NHEK was found to be related to cell culture donor age.EPR spectroscopy was used to show the formation of an EPR detectable, g = 2.04, feature characteristic of iron–nitrosyl complex formation, and the generation of this induced complex by NHEK exposed to HD (18 h post exposure to 1 mM HD) was blocked by Nw-nitro-L-arginine (L-NOARG), a competitive inhibitor of nitric oxide synthase (NOS). Nitric oxide (?NO) has been implicated as the effector molecule that mediates IL-1b.6 Our results show the release of nitric oxide during cytokine expression, IL-1b, when keratinocytes are exposed to HD.The combination of the nitric oxide with the chloride (Cl2) in the plasma which is approximately 0.10 M,72496 J. Chem. Soc., Perkin Trans. 2, 1997 and the one released from sulfur mustard ([ClCH2CH2]2S) upon cyclization to the sulfonium ion may lead to the formation of nitrosyl chloride (NOCl), a known potent nitrosylating agent. If NOCl is formed, it may play a role in the skin injury.In addition, EPR and ELISA results show the generation of nitric oxide (?NO), therefore suggesting that nitric oxide synthase (NOS) plays a role during IL-1 expression when keratinocytes are exposed to HD. Inducible nitric oxide synthase (iNOS) has been associated with inflammatory and autoimmune tissue injury.8 Their findings show the presence of iNOS in human skin that can be localized to keratinocytes in the epidermal layer.This paper shows the possible involvement of active nitrogen species in the pathways of cell injury/repair following exposure to sulfur mustards. Active nitrogen species are those derived from nitric oxide generated in cells. The possible involvement of the known alkylating agent nitrosyl chloride (NOCl) is discussed. The thermodynamic considerations and possible in vivo production of NOCl by macrophages and neutrophils have been reported.9 Experimental NHEK (adult and neonatal, Clonetics, San Diego, CA) were cultured to confluency and harvested for experiments.The culture medium used was keratinocyte basal medium, modified Molecular Cellular Developmental Biology 153 (MCDB 153), which was supplemented with bovine pituitary extract (7.5 mg ml21), human recombinant epidermal growth factor (0.1 mg ml21), hydrocortisone (0.5 mg ml21), bovine insulin (5 mg ml21), gentamicin sulfate (50 mg ml21) and amphotericin-B (50 mg ml21). The medium was changed after 2 days of culture. After 3 and 6 days, the cells were harvested.After harvesting, the NHEK were resuspended in phosphate medium (5 × 106– 5 × 107 cells per 0.5 ml). IL-1b and EPR spectra were measured after exposure of the suspension to 1–2 mM HD, 0.1 mM Nwnitro- L-arginine (L-NOARG; C6H13N5O4, Sigma Chemical Co., St. Louis, MO), or a combination of these agents (see Fig. legends). IL-1b release was measured after 3 h exposure and the heme-NO EPR spectrum was recorded after 18 h of exposure.The difference in post exposure time between the ELISA and the EPR experiments is due to the sensitivity of the two techniques. The IL-1b release was measured using a commercially available kit for the ELISA technique (QuantikineTM). This kit is specific for IL-1b and shows no cross reactivity with other cytokines (e.g. IL-1a) as stated in the QuantikineTM brochure for IL-1b. The QuantikineTM human IL-1b Immunoassay (Catalog Number DLB50, R&D Systems, Inc., Minneapolis, MN) was used for the quantitative determination of human IL-1b concentration in the cell cultures.This assay employs the quantitative ‘sandwich’ enzyme immunoassay technique. A monoclonal antibody specific for IL-1b is coated onto the microtiter plate provided in the kit. Standards and homogenous cell suspensions (100 ml) were pipetted into the wells. Following a wash to remove any unbound antibody–enzyme reagent, a substrate solution was added to the wells and color was developed in proportion to the amount of IL-1b bound in the initial step.The color development was stopped and the intensity of the color was measured. The absorbance of each well was read at 450 nm. By comparing the optical density of the samples to the standard curve, the concentration of the IL-1b in the unknown samples was then determined. The assays were run in triplicate, and statistical evaluation was carried out using the paired sample t test with significance defined as p < 0.05.Representative data are shown in Fig. 2. EPR spectroscopy experiments were performed using cell suspensions containing at least 5 × 106 keratinocyte cells per 0.5 ml suspended in 1 ml of complete MCDB 153 media. The cells were incubated at 37 8C for an additional 18 h in the presence or absence of 1 mM HD or 50–100 mM of L-NOARG at which time the cells were isolated and frozen at 270 8C. EPR spectroscopy was performed at 77 K on the cell suspensions using a Varian E-109 spectrometer equipped with an X-band (9.5 GHz) microwave bridge.The instrumental parameters at which the EPR spectra were recorded are given in the legend to Fig. 1. Results and discussion Adult-NHEK cells exposed to HD and incubated for 18 h at 37 8C induced the formation of EPR-detectable signals (Fig. 1). The EPR spectra in Fig. 1 were obtained by freezing the samples to 77 K in liquid nitrogen after incubation at 37 8C for 18 h. Fig. 1(A) is the control and represents the EPR spectrum obtained when adult-NHEK were incubated in the absence of HD and then frozen.However, adult-NHEK exposed to HD and incubated for 18 h generate the EPR spectrum shown in Fig. 1(B). This EPR spectrum has an approximate g-value of g = 2.04, which is similar to the g-values of known reported iron–nitrosyl complexes, suggesting the formation of nitric oxide (?NO).10 The observed EPR spectrum may originate directly from the formation of ?NO or one of the reactive nitrogen by-products (NOx) generated in the biological decomposition of ?NO.Neonatal-NHEK incubated in the presence and absence of HD generate the EPR spectrum as shown in Fig. 1(C). It is possible that if ?NO is produced by neonatal-NHEK either in the unexposed controls or in the HD-exposed cells, this ?NO may originate from another source and have a different function, thus rendering it unavailable to interact with iron to form an iron–nitrosyl type complex as observed in Fig. 1(B). It is also conceivable that low concentrations of ?NO are always present in neonatal-NHEK. This concentration may be too low to detect by EPR spectroscopy, but high enough to cause other types of biological effects (e.g. cytokine release). Since the spectrum in Fig. 1(B) suggests that adult-NHEK generates nitric oxide (?NO) during sulfur mustard (HD) exposure, the cells were incubated with HD in the presence of the specific nitric oxide synthase (NOS) inhibitor Nw-nitro-Larginine (L-NOARG)11 to confirm the production of ?NO.The result from this experiment is shown in Fig. 1(D). It can be seen that the EPR spectrum shown in Fig. 1(B) is not generated when the cells are incubated with HD in the presence of LNOARG. This strongly suggests that ?NO is produced when adult-NHEK are exposed to HD. Fig. 1 Low temperature (liquid nitrogen, 77 K) EPR spectra recorded from NHEK suspensions (3.8 × 106 cells ml21). (A) Control, nonexposed adult-NHEK; (B) adult-NHEK exposed to 1 mM sulfur mustard (HD) collected 18 h post-HD exposure; (C) neonatal-NHEK exposed to 1 mM HD EPR spectrum obtained 18 h post exposure; (D) sample treated with Nw-nitro-L-arginine (L-NOARG) to give a final concentration of 100 mM.EPR conditions: magnetic field, 334.5 mT; modulation frequency, 100 kHz; microwave frequency, 9.475 GHz; microwave power, 10 mW; receiver gain, 5 × 105; modulation amplitude, 0.5 mT and scan rate 0.62 mT s21.J.Chem. Soc., Perkin Trans. 2, 1997 2497 Fig. 2 IL-1b activation of adult- and neonatal-NHEK exposed to 2 mM HD. The IL-1b levels in cell suspensions (5 × 106–5 × 107 cells ml21) were measured using IL-1b QuantikineTM ELISA kit. (A) 100 ml aliquots of adult-NHEK (2.6 × 107 cells ml21) as a function of IL-1b (pg ml21) produced in HD-exposed and non-exposed controls. (B) 100 ml aliquots of neonatal-NHEK (7.5 × 106 cells ml21) as a function of IL-1b (pg ml21) produced in HD-exposed and non-exposed controls.The effect of Nw-nitro-L-arginine (L-NOARG) alone and the combined effects of sulfur mustard and L-NOARG on IL-1b release by NHEK are also included in panel (A) and panel (B). 700 600 500 400 300 200 100 0 250.66 ± 0.17 653.56 ± 0.17 247.00 ± 0.08 187.11 ± 0.03 (A) IL-1 Beta Adult NHEK Quantikine TM 2.6E7 Cell Density Control HD exposed Pretreated N-nitro-Larginine Pretreated N-nitro-Larginine; HD exposed IL-1 Beta concentration (pg ml–1) 700 800 600 500 400 300 200 100 0 746.10 ± 0.01 609.43 ± 0.01 472.11 ± 0.01 297.85 ± 0.01 (B) IL-1 Beta Neonatal NHEK Quantikine TM 2.6E6 Cell Density Control HD exposed Pretreated N-nitro-Larginine Pretreated N-nitro-Larginine; HD exposed IL-1 Beta concentration (pg ml–1) It is important to establish what role the production of ?NO plays during exposure of NHEK to HD.Previous reports have shown that ?NO is involved in immune/cytokine regulation.12 Therefore, to establish the role ?NO plays during HD exposure, the effect that HD exposure has on cytokine regulation must be determined.Initially the effect of HD on the production of interleukin 1b (IL-1b) was chosen because human skin keratinocytes express significant activation of IL-1 determined by the ELISA technique. For this reason, IL-1b was assayed in adult- NHEK and neonatal-NHEK (5 × 106–5 × 107 cells) exposed to HD (2 mM) and incubated for 3 h at 37 8C, using the ELISA technique. This exposure time is sufficient to allow release of a significant quantity of IL-1b.However, it is not long enough to allow the cells (adult and neonatal) to go through a complete cycle and divide, thus eliminating possible errors due to differences in cell turnover time (~21 h for neonatal-NHEK and ~24 h for adult-NHEK). The HD concentration range was chosen because it is generally observed that exposure to approximately 1 mM HD causes formation of microblisters while full blister formation occurs after exposure to approximately 2 mM HD (data not shown).The results are shown in Fig. 2. Adult-NHEK exposed to HD (2 mM) show a significant increase in the production of IL-1b [Fig. 2(A)]. When these cell suspensions are incubated with the specific NOS inhibitor L-NOARG, the production of IL-1b is decreased to the level obtained for the cells (controls) not exposed to HD. In addition, the combined effects of incubation with L-NOARG followed by exposure to HD slightly lowers the production of IL-1b when compared to the cell suspensions incubated with L-NOARG alone or to controls.For neonatal-NHEK exposed to HD (2 mM) the production of IL-1b is decreased when compared to the cells (controls) not exposed to HD [Fig. 2(B)]. Furthermore, incubation of the cell suspensions with L-NOARG alone also decreases the production of IL-1b. The combined effects of incubation with L-NOARG followed by HD exposure are also shown in Fig. 2(B).These show a further decrease in the production of IL-1b when compared to the controls and to the cells exposed to HD alone or cells incubated with L-NOARG alone. The concentrations of IL-1b per cell before and after exposure to HD or treatment with L-NOARG are given in2498 J. Chem. Soc., Perkin Trans. 2, 1997 Table 1 Concentration of interleukin-1b per cell a Cell type Adult Neonatal n b 33 [Control]/ (pg/cell) × 106 9.64 99.48 [HD]/ (pg/cell) × 106 25.14 81.26 [L-NOARG]/ (pg/cell) × 106 9.50 62.95 [L-NOARG 1 HD]/ (pg/cell) × 106 7.20 39.91 a All columns are the average of a number of experiments.b n = number of experiments. Table 1. The data show that the neonatal-NHEK contain a significantly large amount of IL-1b per cell when compared to the adult-NHEK. The approximately 10-fold difference in IL-1b concentration per cell in the untreated control adult- and neonatal-NHEK can be attributed to two factors: (i) a continual higher level of IL-1b in neonatal-NHEK is required for the induction of growth factors for proliferation and other structural proteins required for the normal keratinocyte development; and (ii) the fully developed ?NO-related immune pathways in adult NHEK differ from the developing ?NO-related immune pathways in the neonatal-NHEK.13 For instance, the data in Fig. 2 show that the production of IL-1b is directly linked to the production of ?NO. Therefore, it is possible that the adult-NHEK contain a low concentration of the constitutive form of NOS (cNOS), but are capable of rapidly producing the inducible form of NOS (iNOS), which is activated immediately upon the presence of a foreign toxic substance.11 This would mean that adult-NHEK produce iNOS on demand in the presence of a foreign toxic substance (HD in this case) generating the ?NO which triggers the observed increase in IL-1b production.Alternatively, the neonatal-NHEK contain mainly the cNOS and thus require only small concentrations of the iNOS.However, neonatal- NHEK contain a steady concentration of cNOS which is continuously producing low levels of ?NO which, in turn, triggers the observed higher level production of IL-1b. Although the adult- and neonatal-NHEK appear to contain different forms of the NOS enzyme, both types of cells are affected in the same manner by the specific NOS inhibitor L-NOARG even though each type of cell (adult and neonatal) reacts in an opposite manner when exposed to HD.The lower yield of IL-1b [Fig. 2(B)] when neonatal-NHEK are exposed to HD as compared to unexposed controls is possibly due to the interaction of HD with the cell surface. It is possible that activation of cNOS and iNOS originates from different receptors on the cell surface. Therefore, since neonatal-NHEK appear to contain only the cNOS it is likely that the interaction of HD with the cell surface would interfere with the continuous production of ?NO and IL-1b [Fig. 2(B)]. This is consistent with the fact that no EPR signal is observed for HD-exposed neonatal-NHEK when suspensions of these cells were run in the same manner as the adult-NHEK (Fig. 1). These observations are consistent with previous studies 14 that suggest that the surface characteristics of NHEK are continuously changing. These modulations reflect the stage of differentiation and activation of the NHEK. Thus, the NHEK in various stages of differentiation have distinct sets of surface moieties that are expressed in different manners (Fig. 2). One thing is clear from the results in Fig. 2: the dependence of the production of IL-1b in adult- and neonatal-NHEK on the generation of ?NO. This established ?NO as an effector molecule in cytokine regulation, at least for IL-1b. This fact is supported by the observation that when the production of ?NO is blocked by the specific NOS inhibitor (L-NOARG) the production of IL-1b is also decreased. The results suggest that the production of ?NO by the NHEK serves as a direct interleukin- 1 converting enzyme activator or that the ?NO formed triggers a signal at the cell membrane which activates the interleukin-1 converting enzyme.The fate of the ?NO generated by NHEK after activating the IL-1b system remains to be addressed. Since chloride (Cl2) is released upon dissolution of HD in aqueous environments, it is possible that the reactive nitrogen species formed is NOCl. NOCl is a nitrosylating agent that is consistent with the known biological action of HD.In addition, NOCl and ?NO react with hemoglobin yielding the EPR spectra shown in Fig. 3. Fig. 3(A) represents the reaction of NOCl with hemoglobin and Fig. 3(B) represents the reaction of a solution of dissolved ?NO with hemoglobin (1 × 1024 M). Both these EPR spectra have similar g-values to the one observed in Fig. 1(B). However, in Fig. 3(A) the NOCl was generated chemically in a reaction vessel and then carried over with an inert gas (N2) and bubbled through hemoglobin (1 × 1024 M) solution.The hemoglobin solution containing the NOCl was rapidly frozen at 77 K in liquid nitrogen prior to obtaining its EPR spectrum. The chloride ion concentration (1–2 mM) in HD is small compared to the Cl2 concentration in plasma and cytosol (100 mM). However, HD interacts with the cells, ultimately generating nitric oxide and IL-1b. It is this increase of ?NO over the base level in cells that is available for the production of NOCl.The Gibbs energy of formation of NOCl is energetically feasible.9,15 Nitrosyl chloride can nitrosylate organic compounds directly, and therefore its presence poses two dangers: (i) it is a strong oxidant and (ii) nitrosylation leads to compounds that are often mutagenic or promutagenic (alkylation of DNA). Therefore, the EPR spectrum in Fig. 1(B) could originate from either the direct interaction of ?NO with some type of porphyrin-containing molecule or the interaction of NOCl with the porphyrincontaining molecule.However, because of the suggested difference in the NOStype (inducible or constitutive) in these cells, their reaction to HD exposure is quite different. In terms of IL-1b production, keratinocytes in normal skin do not express significant numbers of IL-1 receptors. The normal keratinocyte cells can be in an ‘activated’ or ‘refractory’ state. The IL-1b receptor expression is associated with the active state of the keratinocyte and requires Fig. 3 Low temperature (77 K) EPR spectrum of hemoglobin (~1024 M) after reaction with (A) chemically generated NOCl [nitrosyl chloride is generated in situ from alkylnitrite and titanium tetrachloride in DMF at 0 8C: TiCl4 1 4RONOÆTi(OR)4 1 4ClNO]; and (B) dissolved ?NO. EPR conditions: magnetic field, 334.5 mT; modulation frequency, 100 kHz; microwave frequency, 9.475 GHz; microwave power, 10 mW; receiver gain, 1.25 × 104; modulation amplitude, 1.0 mT and scan rate 0.833 mT s21J. Chem.Soc., Perkin Trans. 2, 1997 2499 an outside stimulus to express the receptor and generate IL-1b. Exposure to HD causes adult-NHEK to go into the activated state generating ?NO, which in turn activates the IL-1b receptors causing the production of IL-1b. The ?NO generated by the exposure of human keratinocytes to HD activates the interleukin-1 converting enzyme producing the observed increase in interleukin-1b.16 The production of IL-1b triggers a series of other events in the keratinocyte (‘effector’ state).These observations are consistent with the results shown in Fig. 2(A). The neonatal-NHEK are continuously producing higher levels of ?NO and IL-1b. However, when neonatal-NHEK are exposed to HD the cells go directly into the ‘refractory’ state either from the active state or the effector state.17,18 In the refractory state the IL-1 receptors are downregulated consistent with the results in Fig. 2(B). In both cases NOCl could be produced leading to cell damage and DNA alkylation. These findings may provide guidance to the understanding of sulfur mustard toxicity and also suggest that L-NOARG has significant novel pharmacological effects other than the inhibition of NOS. References 1 T. S. Kupper, F. Lee, N. Bichall, S. Clark and S. K. Dower, J. Clin. Invest., 1989, 82, 1787. 2 B. Blanton, T. S. Kupper, J. McDougall and S. Dower, Proc. Natl. Acad. Sci. USA, 1989, 86, 1273. 3 C. A. Dinarello, Blood, 1991, 77, 1627. 4 P. D. Lawley, Carcinogenesis by alkylating agents. ACS Monograph 173, American Chemical Society, Washington, D.C., 1976. 5 C. N. Lieske, R. S. Klopcic, C. L. Gross, J. H. Clark, T. W. Dolzine, T. P. Logan and H. G. Meyer, Immunol. Letters, 1992, 31, 117. 6 J. A. Corbett, J. R. Lancaster, M. A. Sweetland and M. L. McDaniel, FASEB J., 1993, 7, 369. 7 W. R. Faulkner, J. W. King and H. Damm, Handbook of Clinical Laboratory Data, Chemical Rubber Co., Cleveland, Ohio, 1968. 8 H. Kolb and V. Kolb-Bachofen, Immunol. Today, 1992, 13, 157. 9 W. H. Koppenol, FEBS Lett., 1994, 347, 5. 10 J. A. Corbett, J. R. Lancaster, Jr., M. A. Sweetland and M. L. McDaniel, J. Biol. Chem., 1991, 266, 21 351. 11 J. M. Fukuto and G. Chaudhuri, Annu. Rev. Pharmacol. Toxicol., 1995, 35, 165. 12 C. Nathan and Q. Xie, J. Biol. Chem., 1994, 269, 13 725. 13 P.-A. Becherel, L. LeGoff, S. Ktorza, F. Quaaz, J. M. Mencia- Huerta, B. Dugas, P. Debres, M. D. Mossalayi and M. Arock, Eur. J. Immunol., 1995, 25, 2992. 14 J. Hunyadi, M. Simon, Jr. and A. Dobozy, Immunol. Lett., 1992, 31, 209. 15 D. L. H. Williams, Nitrosation, Cambridge University Press, New York, 1988. 16 C. M. Arroyo, A. J. Carmichael and C. A. Broomfield, In Vitro Toxicology, 1997, 10, 253. 17 T. S. Kupper, Arch. Dermatol., 1989, 125, 1406. 18 T. S. Kupper, Immunophysiology. The role of cells and cytokines in immunity and inflammation, University Press, Oxford, 1990. Paper 7/02508D Received 11th April 1997 Accepted 8th September 1997

ISSN:1472-779X    

DOI:10.1039/a702508d

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

J. Chem. Soc., Perkin Trans. 2, 1997 2501 Reduction of ‚-phosphorylated cyclic aminoxyl radicals by flavins: an EPR kinetic study † Corinne Mathieu, Béatrice Tuccio,* Robert Lauricella, Anne Mercier and Paul Tordo UMR 6517, Chimie, Biologie et radicaux libres, CNRS et Universités d’Aix-Marseille 1 et 3, Av. Normandie-Niemen 13397, Marseille Cedex 20, France A series of stable ‚-phosphorylated aminoxyl radicals has been tested towards their resistance to photoreduced flavin reduction in a pH 7.4 phosphate buffer.Their decay kinetics, monitored by EPR spectroscopy, have been compared to those of three non-phosphorylated reference compounds. In each case, both a second-order reduction and a first-order aminoxyl decay were found to contribute to the EPR signal loss. On the basis of only the second-order process, a classification of the aminoxyl radicals has been established as a function of their ease of reduction by the two flavins tested. The potential applications of stable aminoxyl free radicals became increasingly important with the observation that these compounds, which have been widely used in vitro, could also be employed successfully for various in vivo studies.1 For example, they have been used as EPR spin probes to study biomembrane properties,2 or to determine oxygen level in various cellular microenvironments.3 Moreover, they were found to be valuable contrast-enhancing agents for magnetic resonance imaging,4 and they also play a key-role in EPR imaging when used in intact organs or in whole animals.5 These aminoxyls have been shown to be transformed into EPR-silent hydroxylamine by a large range of reducing agents, such as ascorbate or thiols,6 and this reduction should be regarded as a serious limitation in a lot of in vivo applications.This is the reason why both their in vivo and in vitro reduction by various models of bioreducing agents have been widely studied.7 The variety of stable aminoxyl applications resulted in the necessity to dispose of a large range of compounds showing various properties, particularly concerning their lipophilicity, their EPR parameters and their decay rates in various environments. 8 In this field, a new series of stable b-phosphorylated cyclic aminoxyls has been elaborated in our laboratory a few years ago.9 All these pyrrolidinoxyl radicals exhibited EPR spectra showing a large hyperfine splitting constant (hfsc) with the phosphorus nucleus ranging from 3.6 to 5.5 mT, which was found to be very sensitive to the five-membered ring conformation.Because of this strong coupling, these b-phosphorylated aminoxyls could be valuable tools for example in the study of biomembrane properties. Furthermore, the presence of a dialkoxyphosphoryl group on the ring could result in interesting modifications of the metabolism of these compounds or in their biodistribution. Anyway, before using these new spin labels in vivo, it is of prime importance to evaluate their resistance to the action of various bioreducing agent models, in order to have a notion of their lifetime in biological systems.Thus, their reduction rate by ascorbate has been previously evaluated in pH 7.4 phosphate buffer.10 In order to proceed with this study, we have examined the resistance of these bphosphorylated compounds towards the action of photoreduced flavins. Results When irradiated with visible light in the presence of an electron donor, oxidised flavins are known to undergo a two-step reduc- † Presented at the 30th International Meeting of the Electron Spin Resonance Group of the RSC, University of Lancaster, 6–10th April 1997.tion, yielding a dihydro form of the flavins.11,12 These photoreduced compounds can then react efficiently with many chemical or biochemical molecules, such as triplet oxygen,13 and this reaction has been frequently employed to generate superoxide in aqueous media.14,15 A few studies have also been devoted to the transformation of aminoxyls into EPR-silent hydroxylamines by photoreduced flavins.12,16 In our study, two flavins, riboflavin and flavin mononucleotide (FMN) considered as bioreducing agent models, were employed to reduce a series of aminoxyls.All the experiments were conducted in a pH 7.4 phosphate buffer and in the absence of oxygen to avoid reoxidation of the hydroxylamine formed. The seven aminoxyls, the structures of which are shown below, were studied.They can be divided into two groups: four b-phosphorylated compounds, i.e. 2-diethoxyphosphoryl- 2,5,5-trimethylpyrrolidin-1-oxyl 1 (TOMER-Et), 2-diisopropyloxyphosphoryl-2,5,5-trimethylpyrrolidin-1-oxyl 2 (TOMER-Pri), r-2-diethoxyphosphoryl-c-4-phenyl-2,5,5- trimethylpyrrolidin-1-oxyl 3 (TOBER-36) and r-2-diethoxyphosphoryl- t-4-phenyl-2,5,5-trimethylpyrrolidin-1-oxyl 4 (TOBER-53), and three non-phosphorylated probes, i.e. 2,2,5,5-tetramethylpyrrolidin-1-oxyl 5 (Proxyl), 3-carboxy- N P(O)(OEt)2 O• 1: TOMER-Et N P(O)(OPri)2 O• 2: TOMER-Pri N P(O)(OEt)2 O• 3: TOBER-36 N P(O)(OEt)2 4: TOBER-53 Ph H H Ph N O• N O• 5: PROXYL 6: PCA N O• 7: TEMPO CO2H2502 J.Chem. Soc., Perkin Trans. 2, 1997 2,2,5,5-tetramethylpyrrolidin-1-oxyl 6 (PCA) and 2,2,6,6- tetramethylpiperidin-1-oxyl 7 (TEMPO). The EPR parameters of the aminoxyl spectra recorded in phosphate buffer, i.e. the hfscs for the nitrogen and the b-phosphorus, have been listed in Table 1.In each kinetic experiment, the medium containing the aminoxyl, an oxidised flavin and diethylenetriaminetetraacetic acid (DTPA) was first irradiated with visible light to generate photoreduced flavin, as described in the Experimental section. After termination of the illumination, the aminoxyl decay was followed by recording the EPR signal at a fixed field corresponding to the first low-field peak of the spectrum. The 14 experimental kinetic curves, corresponding to the signal loss of aminoxyls 1–7 in the presence of each one of the two flavins, were first modelled considering only a simple secondorder reduction.The rate of aminoxyl decay was then given by eqn. (1) in which [RR9NO?] and [Flred] represent the aminoxyl 2d[RR9NO?]/dt = kred[RR9NO?][Flred] (1) and the photoreduced flavin concentrations, respectively, and kred the second-order rate constant. However both the standard deviation between experimental and calculated curves, and the uncertainty in the determination of kred were found much too high (e.g.a 30% error was evaluated for kred in the case of the reduction of 5 by riboflavin). Furthermore, a second-order reduction process was not appropriate to model experimental TEMPO decay curves. Fig. 1(a), in which both the experimental curve of TOMER-Pri decay in the presence of reduced ribo- flavin and the curve calculated using eqn. (1) have been plotted, clearly shows that a single second-order reduction does not correctly reproduce the experimental decay kinetics.Another model, described in Scheme 1, was then examined. Fig. 1 Decay kinetic of TOMER-Pri 2 recorded in the presence of photoreduced riboflavin in a 0.1 mol dm23 pH 7.4 phosphate buffer. Modelling of the decay (- - - -) has been achieved using (a) eqn. (1), with kred = 107 dm3 mol21 s21, or (b) eqn. (2), with kdes = 4 × 1024 s21 and kred = 137 dm3 mol21 s21. Table 1 EPR hyperfine splitting constants of aminoxyls 1–7 in 0.1 mol dm23 phosphate buffer, pH 7.4 Aminoxyl Proxyl 5 PCA 6 TOMER-Pri 2 TOMER-Et 1 TOBER-36 3 TOBER-53 4 TEMPO 7 aN/mT 1.63 1.62 1.54 1.53 1.52 1.52 1.72 aP/mT — — 4.76 4.79 3.64 5.50 — RR9NO? k1 X(diamagnetic) RR9NO? 1 Flred kred RR9NOH 1 Flox Scheme 1 Model proposed for the description of the aminoxyl 1–7 experimental decay observed in the presence of photoreduced flavins in phosphate buffer.RR9NO? represents the aminoxyl, RR9NOH the corresponding hydroxylamine, X an unidentified diamagnetic product, Flred and Flox the reduced and oxidised flavin, respectively; k1 and kred correspond to the first- and second-order rate constants, respectively.In this case, a first-order decay was considered to contribute to the aminoxyl decay together with its reduction by the flavin. The corresponding rate is thus given by eqn. (2) in which 2d[RR9NO?]/dt = k1[RR9NO?] 1 kred[RR9NO?][Flred] (2) [RR9NO?] and [Flred] represent the aminoxyl and the photoreduced flavin concentrations, respectively, k1 being the firstorder rate constant of the aminoxyl decay and kred the second-order rate constant of the aminoxyl reduction by flavin.Using this second model, simulation of each one of the 14 experimental curves could be successfully achieved, yielding to standard deviation values always in the range of the signal-tonoise ratio. The values thus determined for the two rate constants k1 and kred have been listed in Tables 2 and 3, respectively.Fig. 1(b) shows the experimental curve of TOMER-Pri decay in the presence of riboflavin which has been modelled using eqn. (2); it appears that experimental and calculated curves are perfectly superimposed. This excellent fit between the two curves clearly illustrates the amelioration obtained in the modelling by taking into account a first-order decay process in addition to the second-order reduction, and supports the validity of the model proposed.Discussion Considering only the values indicated for kred, that is, taking into account only the second-order reduction process itself, the seven aminoxyls tested can be compared as a function of their resistance to the reduction by flavins. The classification thus obtained appears in Table 3, in which aminoxyls 1–7 have been listed from the more to the less resistant to the reduction. This Table 2 Rate constants determined for the first-order decay of the aminoxyls 1–7 in the presence of either riboflavin (k1,rib) or FMN (k1,FMN).The values have been determined by computer simulation of experimental decay curves using eqn. (2). Aminoxyl Proxyl 5 PCA 6 TOMER-Pri 2 TOMER-Et 1 TOBER-36 3 TOBER-53 4 TEMPO 7 k1,rib/1024 s21 0–8 0–8 × 1023 4.0 ± 0.1 0.3 ± 0.08 2.14 ± 0.04 0.92 ± 0.4 3.0 ± 0.1 k1,FMN/1024 s21 0–0.5 0–0.4 1.0 ± 0.1 1.2 ± 0.1 1.06 ± 0.03 1.3 ± 0.03 3.4 ± 0.1 Table 3 Second-order rate constants determined for the reduction of aminoxyls 1–7 by either riboflavin (kred,rib) or FMN (kred,FMN).The values have been determined by computer simulation of experimental decay curves using eqn. (2). Aminoxyl Proxyl 5 PCA 6 TOMER-Pri 2 TOMER-Et 1 TOBER-36 3 TOBER-53 4 TEMPO 7 kred,rib/ dm3 mol21 s21 29 ± 3 63 ± 3 137 ± 2.5 251 ± 3 323 ± 10 422 ± 16 503 ± 29 kred,FMN/ dm3 mol21 s21 52 ± 4 77 ± 1 233 ± 10 426 ± 8 608 ± 10 718 ± 14 811 ± 10J. Chem. Soc., Perkin Trans. 2, 1997 2503 classification was found to be independent of the flavin, although FMN was a ca. 1.5 times stronger reducing agent than riboflavin in our experimental conditions. Thus, TEMPO is clearly the most easily reduced aminoxyl, and this is in accordance with all the results given in the literature which specify that piperidinoxyl compounds are more easily reduced than are pyrrolidinoxyl compounds.17 It also appeared that the introduction of a b-phosphoryl group on the pyrrolidinyl ring resulted in an increase in the reduction rate of the aminoxyl.Thus, TOMER-Pri 2 and TOMER-Et 1 were reduced ca. 4.5 and 8.5 times more rapidly, respectively, than their nonphosphorylated analogue, Proxyl 5. The same classification of these aminoxyls was previously observed when they were tested for their resistance to ascorbate reduction,10 but ascorbate was found to reduce compounds 1–6 from 4 to 10 times slower than riboflavin. This result is in accordance with the literature data,16 which indicate that flavins are much more powerful reducing agents than ascorbate. In addition, the reduction of compounds 1–6 by ascorbate has been shown to be reversible in pH 7.4 phosphate buffer,10 thereby enhancing the persistence of these aminoxyls in the presence of this reducing agent.So we thought that it was important to verify that the aminoxyl reduction by flavins was not reversible. Thus, various aminoxyl reduction experiments have been performed using the conditions described in the Experimental section and followed by EPR spectroscopy. A large amount of oxidised flavin was then added after the loss of ca. 66% of the initial EPR signal, and the medium was kept in the dark. No significant increase in the EPR signal was observed, indicating that the oxidised flavins were unable to convert hydroxylamine into aminoxyl. If the relative reduction rate in the pyrrolidinoxyl series can be considered independent of the reducing agent, it seems to increase while the hfsc aN decreases (see Table 1).Aminoxyl compounds can be represented by the two mesomeric forms A and B, as indicated in Scheme 2, and the aN value is expected to increase when the B form is favoured. As previously observed,10 the presence of an electron-withdrawing group, such as a dialkoxyphosphoryl, in the b-position to the aminoxyl function results in favouring the mesomeric form A, in which the unpaired electron is located on the oxygen. This aminoxyl is then a better hydrogen abstracting agent and is more easily reducible to the corresponding hydroxylamine.However, beside electronic factors, the aminoxyl stereochemistry seems to have a small influence on the second-order decay mechanism. Different reduction rates have been determined for the two diastereoisomeric aminoxyls 3 and 4, although these two compounds show the same aN. Thus, TOBER-36 3 appeared to be ca. 1.2 times more resistant than TOBER-53 4 to the flavin reduction. Considering now only the first-order decay process, the interpretation of the results obtained in this kinetic study seems to be much more tricky.As can be seen from Table 2, k1 was most often found in the range of 1024 s21. However, in the case of Proxyl, for example, the estimation of k1 was tedious and values ranging from 0 to 8 × 1024 s21 were found. In general, we observed that the determination of k1 was all the more difficult as this rate constant was found to depend greatly on experimental conditions, such as the power of the lamp, the irradiation time, or the sample volume irradiated.This could explain the importance of the error in the determination of k1 (see Table 2). Since the meaning of the first-order decay Scheme 2 Representation of the two limit mesomeric forms of an aminoxyl N O• R R¢ •• N O R R¢ •• +• – A B remained unclear, we thought that it was necessary to verify that as aminoxyl destruction really occurred during the spin loss observed.Using potassium ferricyanide, a selective oxidating agent for hydroxylamines, a reoxidation of the hydroxylamine formed during the aminoxyl decay resulted in the recovering of only 65–85% of the initial EPR signal. When H2O2 was employed as the oxidating agent, the percentage of the aminoxyl obtained was generally found higher than 85% and could sometimes reach 95%. This clearly indicates on one hand that an irreversible destruction of at least 5% of the aminoxyl occurred, and on the other hand that the photoreduced flavins tested were also able to reduce hydroxylamines in amines.By computer integration of eqn. (2), the proportion of diamagnetic product X formed after 1800 s of aminoxyl decay was found to vary between 6 and 25%, depending on the aminoxyl and the flavin tested. The comparison between these two results indicated that in each case, the product X concentration calculated was slightly higher than the effective proportion of aminoxyl decay.In addition, an EPR signal decrease was observed when an aminoxyl solution containing oxidised flavin was kept in the dark. This aminoxyl decay was found to be pure first-order, and the rate constants thus evaluated were in the range of 1025 s21, i.e. significantly lower than k1 values indicated in Table 2. The same kind of phenomenon was also observed when a solution containing DTPA and an aminoxyl was irradiated in the absence of flavin. This corroborates the hypothesis that the reduction by flavins was not the only process involved in the aminoxyl decay.However, all these results seem also to indicate that the first-order decay experimentally observed possibly corresponds to several pseudo-first-order processes occurring simultaneously. Actually, it is quite clear that the use of eqn. (2) to model the aminoxyl decay sometimes resulted in a slight over-estimation of the aminoxyl destruction. A likely hypothesis to explain this result is the following.The reaction of an aminoxyl with a photoreduced dihydroflavin (i.e. FlH2) has been reported to give hydroxylamine and a monohydroflavin (i.e. FlH?),11,12 the two forms FlH2 and FlH? being able to reduce an aminoxyl. In this case, the kinetic model indicated in Scheme 3 would describe the aminoxyl decay. The rate RR9NO? kdes X(diamagnetic) RR9NO? 1 FlH2 kred RR9NOH 1 FlH? RR9NO? 1 FlH? k9red RR9NOH 1 Flox Scheme 3 Hypothesis for the mechanism of the aminoxyl decay observed in the presence of photoreduced flavins in phosphate buffer. RR9NO? represents the aminoxyl, RR9NOH the corresponding hydroxylamine, X a diamagnetic product, FlH2, FlH? the dihydro and monohydro reduced forms of the flavin, respectively, Flox corresponding to oxidised flavin; kdes, k9red and kred correspond to first- and secondorder rate constants. 2d[RR9NO?]/dt = (kdes 1 k9red[FlH?])[RR9NO?] 1 kred[RR9NO?][FlH2] (3) of aminoxyl decay would then be given by eqn.(3) in which kdes corresponds to the first-order rate constant of the aminoxyl destruction, kred and k9red being the second-order rate constants of the aminoxyl reduction by the dihydro- and the monohydro- flavin, respectively. On the basis of UV–kinetic studies of aminoxyl reduction by flavins, Chan and Bruice12 have clearly shown that under our experimental conditions (pH < 9), the monohydroflavin FlH? accumulates in the medium as a socalled ‘disproportionation dimer’ complex, which can also be obtained from the forms FlH2 and Flox, as indicated in Scheme 4.Because of this phenomenon, no simple relation2504 J. Chem. Soc., Perkin Trans. 2, 1997 2FlH? [complex] Flox 1 FlH2 Scheme 4 can be established between the FlH? and FlH2 concentrations. Thus, a steady-state concentration can be assumed for the monohydroflavin FlH?, the aminoxyl reduction by FlH? became pseudo-first-order, and substituting eqn. (2) into eqn. (3), with [Flred] = [FlH2], gives eqn.(4). k1 = kdes 1 k9red[FlH?] (4) In addition, this could explain why the constant k1 was found to depend greatly on the experimental conditions. Since under these conditions it was impossible to determine the FlH? concentration, we were unable to evaluate separately k9red and kdes. In addition, note that the mechanism responsible for the irreversible destruction of the aminoxyl remained unclear. However, this last reaction was much less important than the reduction, which appeared to be the major process responsible for the aminoxyl decay in the presence of photoreduced flavin.Conclusions The EPR kinetic study presented in this paper indicates that the aminoxyl signal loss in the presence of photoreduced flavins could be correctly simulated by considering both a secondorder reduction and a minor first-order decay. Although effective aminoxyl destruction was unambiguously shown to occur, the observed rate constant k1 was probably the sum of several first- or pseudo-first-order constants.In particular, an aminoxyl reduction by the monohydro form of the flavins tested might intervene in the first-order behaviour observed. Since the second-order reduction appeared to be the major mechanism responsible for the aminoxyl decay studied, the determination of the corresponding rate constant allowed us to classify the different aminoxyls in terms of their resistance to flavin reduction. In the pyrrolidinoxyl series, the reduction rate was found to be higher for the b-phosphorylated compounds, and this is probably because of the electron-withdrawing effect of the dialkoxyphosphoryl group.Nevertheless, these phosphorylated compounds show the considerable interest to present a strong coupling with the phosphorus. Since the hfsc aP is very sensitive to the pyrrolidinyl ring conformation, aminoxyls 1–4 could be considered as valuable tools in all the studies realised with partially immobilised aminoxyls.TOMER-Et 1 and more especially TOMER-Pri 2 were shown to present a reasonable resistance to the flavin reduction, although they have been found to be more easily reduced than Proxyl or PCA, and these two b-phosphorylated probes could thus present interesting in vivo applications. In order to verify this hypothesis and to continue this study, further kinetic experiments are now in progress in our laboratory to evaluate the reduction rate of aminoxyls 1–7 in blood and in the presence of different blood constituents. Experimental Compounds 1–4 were synthesised and purified in our laboratory as previously described.9,10 Proxyl 5 was prepared according to the method of Keana.18 Aminoxyls 6 and 7 have been purchased from Aldrich Chemicals and used without further purification.Riboflavin, FMN, DTPA and potassium ferricyanide were provided by Sigma Chemical Company. All the buffers were stirred for 6 h in the presence of a chelating iminodiacetic acid resin, provided by Sigma Chemical Company, in order to remove trace metal impurities, and the solutions were prepared just before use in 0.4 mmol dm23 phosphate buffer at pH 7.4. EPR measurements were carried out at 20 8C in an EPR capillary tube by using a computer controlled Varian E-9 EPR spectrometer, operating at X-band with 100 kHz modulation frequency.The instrument settings were as follows: nonsaturating microwave power, 10 mW; modulation amplitude, 0.2 mT; receiver gain ranging from 12 500 to 32 000; scan time, 1800 s; time constant, 1 s.A standard sample contained 0.1 mmol dm23 aminoxyl, 1 mmol dm23 oxidised flavin and 1 mmol dm23 DTPA. The mixture was transferred to an EPR capillary tube and argon was bubbled through the solution for at least 10 min in order to remove molecular oxygen before closing the tube. The reaction mixture was then irradiated directly in the spectrometer cavity using a slide projector lamp as the visible light source.The decrease in the low field peak of the aminoxyl was followed and the light was shut off after a loss of ca. half of the signal intensity. In these conditions, the concentration in photoreduced flavin was always found to be lower than 0.5 × 1024 mol dm23. Then the aminoxyl decay was followed in the dark by recording the EPR signal at the fixed field corresponding to the first low field peak. A complete spectrum was also systematically recorded at the end of the kinetic study in order to verify the stability of the magnetic field. Computer modelling of the kinetic curves was achieved using the program DAPHNIS elaborated in our laboratory.10,15 Using this program, the signal at time tn was calculated from the signal amplitude at time tn-1 and using a rate equation, such as eqn.(1) or (2). The standard least-squares method was then applied to fit the experimental curves, leading to the kinetic parameters and the errors in these values indicated in Tables 2 and 3.References 1 R. I. Zhdanov, Bioactive Spin Labels, Springer-Verlag, Heidelberg, 1992; A. Iannone, A. Tomasi, V. Quaresima and M. Ferrari, Res. Chem. Intermed., 1993, 19, 715; A. Iannone and A. Tomasi, Acta Pharm. Jugosl., 1991, 41, 277. 2 M. Burr and D. E. Koshland, Proc. Nat. Acad. Sci. USA, 1964, 52, 1017; E. G. Janzen and R. A. Towner in Bioactive Spin Labels, ed. R. I. Zhdanov, Springer-Verlag, Heidelberg, 1992, p. 573. 3 J. M. Backer, V.G. Budker, S. I. Eremenko and Y. N. Molin, Biochim. Biophys. Acta, 1977, 460, 152; J. S. Hyde, J. J. Yin, J. B. Feix and W. L. Hubbell, Pure Appl. Chem., 1990, 62, 255; P. D. Morse and A. I. Smirnov, Magn. Reson. Chem., 1995, 33, S46; J. E. Baker, W. Froncisz, J. Joseph and B. Kalyanaraman, Free Rad. Biol. Med., 1997, 22, 109. 4 R. C. Brasch, Radiology, 1983, 147, 781; J. F. W. Keana and F. L. V. Van Nice, Physiol. Chem. Phys. Med. NMR, 1984, 16, 477; M. G. Wikstrom, D. L.White, M. E. Moseley, J. W. Dupont and R. C. Brasch, Invest. Radiol., 1989, 24, 692; C. Corot, A. M. Hentsch and L. Curtelin, Invest. Radiol., 1994, 29, S164; S. Pou, P. L. Davis, G. L. Wolf and G. M. Rosen, Free Rad. Res., 1995, 23, 353. 5 G. M. Rosen, H. J. Halpern, L. A. Brunstig, D. P. Spencer, K. E. Strauss, M. K. Bowman and A. Wechsler, Proc. Natl. Acad. Sci. USA, 1988, 85, 7772; S. I. Ishida, S. Matsumoto, H. Yokoyama, N. Mori, H. Kumashiro, N. Tsuchihashi, T. Ogata, M.Yamada, M. Ono, T. Kitajima, H. Kamada and E. Yoshida, Magn. Res. Imaging, 1992, 10, 109; S. Colacicchi, M. Alecci, G. Gualtieri, V. Quaresima, C. L. Ursini, M. Ferrari and A. Sotgiu, J. Chem. Soc., Perkin Trans. 2, 1993, 2077. 6 W. R. Couet, R. C. Brasch, G. Sosnovsky, J. Lukszo, I. Prakash, C. T. Gnewuch and T. N. Tozer, Tetrahedron, 1985, 41, 1165; W. R. Couet, R. C. Brasch, G. Sosnovsky and T. N. Tozer, Magn. Reson. Imaging, 1985, 3, 83; T. Prelesnik, F. Demsar, M. Nemec, S.Pecar and M. Schara, Periodicum Biologorum, 1986, 88, 185; R. J. Mehlhorn, J. Biol. Chem., 1991, 266, 2724; S. Morris, G. Sosnovsky, B. Hui, C. O. Huber, N. U. M. Rao and H. M. Swartz, J. Pharm. Sci., 1991, 80, 149; Z. Yu, Y. Kotake and E. G. Janzen, Redox Report, 1996, 2, 133. 7 E. J. Rauckman, G. M. Rosen and L. K. Griffeth, in Spin Labeling in Pharmacology, ed. J. L. Holtzman, 1984; H. M. Swartz, M. Sentjurc and P. D. Morse, Biochim. Biophys. Acta, 1986, 888, 82; H.C. Chan, R. L. Magin and H. M. Swartz, Magn. Reson. Med., 1988, 8, 160; M. Kveder, M. Sentjurc and M. Schara, Magn. Reson. Med., 1988, 8, 241. 8 K. Chen and H. M. Swartz, Biochim. Biophys. Acta, 1988, 970, 270; M. Senjurc, S. Pecar, K. Chen and H. M. Swartz, Biochim. Biophys. Acta, 1991, 1073, 329.J. Chem. Soc., Perkin Trans. 2, 1997 2505 9 A. Mercier, Y. Berchadsky, Badrudin, S. Pietri and P. Tordo, Tetrahedron Lett., 1991, 32, 2115; F. Le Moigne, A. Mercier and P. Tordo, Tetrahedron Lett., 1991, 32, 3841; L.Dembkowski, J. P. Finet, C. Fréjaville, F. Le Moigne, R. Maurin, A. Mercier, P. Pages and P. Tordo, Free Rad. Res. Commun., 1993, 2, S23; V. Roubaud, F. Le Moigne, A. Mercier and P. Tordo, Phosphorus Sulfur, 1994, 86, 39. 10 C. Mathieu, A. Mercier, D. Witt, L. Dembkowski and P. Tordo, Free Rad. Biol. Med., 1997, 22, 803. 11 H. R. Merkel and W. J. Nickerson, Biochim. Biophys. Acta, 1954, 14, 185; L. P. Vernon, Biochim. Biophys. Acta, 1959, 36, 177; G. R. Penzer and G. K. Radda, Biochem. J., 1968, 109, 259; G. R. Penzer, Biochem. J., 1970, 116, 733. 12 T. W. Chan and T. C. Bruice, J. Am. Chem. Soc., 1977, 99, 7287. 13 H. P. Misra and I. Fridovich, J. Biol. Chem., 1972, 247, 1888; C. Kemal and T. C. Bruice, Proc. Natl. Acad. Sci. USA, 1976, 73, 995. 14 E. Finkelstein, G. M. Rosen and E. J. Rauckman, J. Am. Chem. Soc., 1980, 102, 4994; G. R. Buettner and L. W. Oberley, Biochem. Biophys. Res. Commun., 1978, 83, 69; B. Tuccio, A. Zeghdaoui, J. P. Finet, V. Cerri and P. Tordo, Res. Chem. Intermed., 1996, 22, 393. 15 B. Tuccio, R. Lauricella, C. Fréjaville, J. C. Bouteiller and P. Tordo, J. Chem. Soc., Perkin Trans. 2, 1995, 295; V. Roubaud, R. Lauricella, B. Tuccio, J. C. Bouteiller and P. Tordo, Res. Chem. Intermed., 1996, 22, 405. 16 R. J. Mehlhorn and L. Packer, Can. J. Chem., 1982, 60, 1452; S. Belkin, R. J. Melhorn, K. Hideg, O. Hankovsky and L. Packer, Arch. Biochem. Biophys., 1987, 256, 232; P. D. Morse, E. K. Ruuge, M. J. Petro and H. M. Swartz, Biochim. Biophys. Acta, 1990, 1034, 298; P. D. Morse and J. M. Yuann, Appl. Radiat. Isot., 1993, 44, 455. 17 C. T. Craescu, I. Baracu, N. Grecu, L. Busca and I. Niculescu- Duvaz, Rev. Roum. Biochim., 1982, 19, 15; J. F. W. Keana, S. Pou and G. M. Rosen, Magn. Reson. Med., 1987, 5, 525. 18 J. F. W. Keana in Spin labelling: Theory and Applications, ed. B. Horecker, N. O. Kaplan J. Marmur and H. A. Scheraga, 1979, p. 159. Paper 7/02506H Received 11th April 1997 Accepted 11th September 1997

ISSN:1472-779X    

DOI:10.1039/a702506h

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

J. Chem. Soc., Perkin Trans. 2, 1997 2507 Spin-trapping of free radicals by PBN-type ‚-phosphorylated nitrones in the presence of SDS micelles † Cécile Rizzi, Robert Lauricella,* Béatrice Tuccio, Jean-Claude Bouteiller, Viviane Cerri and Paul Tordo UMR-CNRS 6517 ‘Chimie, Biologie et Radicaux Libres’, Universités d’Aix-Marseille 1 et 3, Av. Escadrille Normandie-Niemen, 13397 Marseille Cedex 20, France Three PBN-type ‚-phosphorylated nitrones, PPN 1, 4-PyOPN 2, and 4-ClPPN 3, have been used to trap two carbon centred free radicals, CH3 ? and CO2~2, in a water–SDS biphasic system.The location of the traps and their adducts has been found by EPR spectroscopy. 4-PyOPN is found to remain in water, while 4-ClPPN is preferentially sequestered into the micellar structure, and PPN partitioned between the two phases. With the three nitrones, the CO2~2 spin adduct always remains in the bulk aqueous phase. In contrast, the three CH3 ? spin adducts seem to be present in both phases.However, 4-PyOPN-CH3 is essentially located in water, and 4-ClPPN–CH3 in the micelles, while PPN–CH3 is clearly shown to partition between the aqueous environment and the micellar interior. For all these spin adducts, the phosporus hyperfine splitting constant, aP, is found to be a good indicator of aminoxyl behaviour in the presence of micelles. For the various sodium dodecyl sulfate concentrations, the PPN–CH3 affinity for the micellar phase was evaluated from the average aP value.The use of nitrones for the in vivo detection of free radicals is of growing importance, since radicals have been shown to be involved in the development of many pathological conditions.1–3 To be an efficient tool in the study of radical processes occurring in biological cells, a spin trap must fulfil three main conditions. First, it has to react rapidly with free radicals, giving rise to rather persistent spin adducts in polar environments. Secondly, the EPR spectra of the various spin adducts must be characteristic of the radical trapped.Finally, it has to be hydrophilic enough to be used in biological media but also lipophilic enough to cross biomembranes to enter the cells. Among all the commercially available nitrones, DMPO (5,5- dimethyl-3,4-dihydropyrrole N-oxide) and PBN (N-tert-butylbenzylideneamine N-oxide) remain the most popular,4,5 but their use in the in vivo identification of free radicals is not without its limitations. For example, DMPO has been shown to trap efficiently hydroxyl radical (HO?) in every kind of media, but its use in the detection of superoxide in water is dramatically limited by the short life-time of the corresponding spin adduct.6,7 A few years ago, the synthesis of a new b-phosphorylated cyclic nitrone, DEPMPO (5-diethoxyphosphoryl-5-methyl-3,4-dihydropyrrole N-oxide), was described, and this compound was found to trap very efficiently superoxide in aqueous media, leading to an exceptionally persistent spin adduct.8–10 But this trap was found to be as hydrophilic as its non-phosphorylated analogue, i.e.DMPO,9,11 and this limits in vivo applications to extracellular media. In contrast, PBN and structurally related nitrones show various degrees of lipophilicity, depending on the nature of the substituents on the aromatic ring.11–13 However, their in vivo spin trapping applications are limited to carboncentred radical detection, since the corresponding oxyl-radical adducts decompose too rapidly in polar media.14–18 One should also consider that the various adducts of PBN or its analogues often show similar EPR spectra, which could be a source of dramatic misinterpretations in spin trapping experiments.16–21 So new and more lipophilic spin traps are still needed for the in vivo identification of various free radicals, including oxylradicals, and for the study of intracellular processes involving radical intermediates.† Presented at the 30th International Meeting of the Electron Spin Resonance Group of the RSC, University of Lancaster, 6–10th April 1997. Recently, we described the synthesis of three b-phosphorylated nitrones derived from PBN:22,23 PPN 1 (N-benzylidene-1- diethoxyphosphoryl-1-methylethylamine N-oxide), 4-PyOPN 2 {1-diethoxyphosphoryl-1-methyl-N-[(1-oxidopyridin-1-ium-4- yl)methylidene]ethylamine N-oxide}, and 4-ClPPN 3 {N-[(4- chlorophenyl)methylidene]-1-diethoxyphosphoryl-1-methyl ethylamine N-oxide}.These new compounds have been shown to trap efficiently not only carbon-centred radicals, but also superoxide even in polar environments, yielding reasonably persistent superoxide spin adducts in phosphate buffers.24 In addition, these three compounds represent a relatively large scale of lipophilicity: the PPN octanol–water partition coefficient Kp has been evaluated at 10.1, indicating the same lipophilicity as PBN, while 4- PyOPN (Kp = 0.18) was found to be more hydrophilic and 4-ClPPN (Kp = 195) poorly water-soluble.24 However, these traps have never been used in vivo and at the moment no information is available concerning their behaviour in the presence of biological cells.The purpose of this work was then to predict the possible capacity of PPN, 4-PyOPN and 4-ClPPN to cross the phospholipid bilayer of biomembranes. We have undertaken a spintrapping study with these three nitrones in the presence of sodium dodecyl sulfate (SDS) micelles, considered here as a very simple membrane model. Two carbon-centred radicals, CO2~2 and CH3 ?, have been generated in the bulk aqueous phase at various SDS concentrations in the presence of a nitrone, by carrying out a standard Fenton reaction in the pres- CH N O – P(O)(OEt)2 1 + N+ CH N O – P(O)(OEt)2 2 + – O CH N O – P(O)(OEt)2 3 + Cl2508 J.Chem. Soc., Perkin Trans. 2, 1997 ence of sodium formate or dimethyl sulfoxide, respectively. The location of the three nitrones and of their spin-adducts has been determined from the detailed analysis of the various spin adduct EPR spectra recorded in the presence of SDS micelles.Results and discussion Whatever the environment polarity the three nitrones employed, i.e. PPN, 4-PyOPN and 4-ClPPN, have been shown to give persistent spin adducts with various carbon-centred free radicals,23 and were used in this study to trap CH3 ? and CO2~2 at different SDS concentrations.CO2~2 trapping First, we examined the EPR signals obtained by trapping CO2~2 with each one of the three nitrones. In the absence of SDS, the corresponding spin adducts, i.e. PPN-CO2 2, 4- PyOPN-CO2 2 and 4-ClPPN-CO2 2, presented the hyperfine splitting constants (hfscs) given in Table 1. All these spectra remained unchanged when the SDS concentration was kept lower than the critical micellar concentration (CMC), which is ca. 8.2 mmol dm23.25 At higher SDS concentration, in the presence of micelles, different behaviour was observed with the three traps.Using 4-PyOPN as spin trap, both the intensity and the hfsc values determined on the 4- PyOPN-CO2 2 spectra remained the same, whatever the SDS concentration. This clearly shows that, in the presence of micelles, both 4-PyOPN and its CO2~2 spin adduct were located in the bulk aqueous phase. In contrast, when 4-ClPPN was used as spin trapping agent, an increase in the SDS concentration resulted in a strong decrease in the 4-ClPPN–CO2 2 EPR signal intensity and, at 50 mmol dm23 SDS, this spectrum almost vanished (see Fig. 1). These results indicate that the nitrone 4-ClPPN shows a strong affinity for the micellar phase, although the corresponding spin adduct was always located in the aqueous environment. The nitrone PPN represented an intermediate between these two extreme cases. When the SDS concentration was raised above the CMC, a significant decrease in the PPN–CO2 2 EPR signal was detected, while the hfsc values remained unchanged, but this signal was always clearly observed at high SDS concentrations, such as 100 mmol dm23.This indicates that the nitrone Fig. 1 EPR signal of 4-ClPPN–CO2 2 recorded at various SDS concentrations: (a) [SDS] = 0 mol dm23 (receiver gain, 12 000), (b) [SDS] = 10 mmol dm23 (receiver gain, 20 000), and (c) [SDS] = 50 mmol dm23 (receiver gain, 63 000) Table 1 EPR parameters of the various CO2~2 radical spin adducts in pure water Spin adduct PPN–CO2 2 4-PyOPN–CO2 2 4-ClPPN–CO2 2 aN/mT 1.44 1.41 1.45 aH/mT 0.45 0.29 0.42 aP/mT 4.97 4.84 5.00 PPN was partitioned between the micellar environment and the bulk aqueous phase, although the corresponding spin adduct was unable to penetrate the micelles.It should be mentioned here that, at high SDS concentration (ca. 100 mmol dm23), a second paramagnetic species was always observed when trapping CO2~2 with both 4-PyOPN and PPN (see Fig. 2). This radical EPR spectrum exhibited six broad lines, and showed the following hfsc values: aN = 1.40 and aP = 3.83 with 4-PyOPN, aN = 1.46 and aP = 3.92 mT with PPN. The same signals were recorded when HO? was generated by a standard Fenton system in the presence of 100 mmol dm23 SDS and in the absence of sodium formate, i.e. in the absence of CO2~2. The same kind of phenomenon has previously been observed by Bakalic and Thomas 26 who noticed the formation of a carbon-centred radical adduct of a nitroso spin trap when HO? was generated in the presence of SDS at high concentration.According to their results and to our observations, the signals recorded in our experiments with both 4-PyOPN and PPN could be reasonably assigned to the spin adducts resulting from the addition to these nitrones of a radical produced by HO? attack of the SDS monomers solubilised in the bulk aqueous phase. Because of the large linewidth, which was always 0.3–0.5 mT, it was impossible to resolve the coupling with the b-hydrogen.CH3 ? trapping The trapping of the methyl radical by 4-PyOPN in the absence of SDS yielded the corresponding spin adduct, 4-PyOPN–CH3 (aN = 1.45, aH = 0.24 and aP = 4.67 mT). No change in the spectrum was observed while the SDS concentration was raised to ca. 20 mmol dm23. Above this value, a second minor paramagnetic species was detected that showed the following hfsc values: aN = 1.43, aH = 0.16, and aP = 4.15 mT (see Table 2).Since this signal was found to grow while the SDS concen- Fig. 2 EPR signal recorded by carrying out a standard Fenton reaction in the presence of 0.2 mol dm23 sodium formate, 0.05 mol dm23 4- PyOPN and 0.1 mol dm23 SDS. The signal obtained was found to correspond to the superposition of two nitroxide spectra, one being the CO2~2 radical spin adduct of 4-PyOPN (aN = 1.41, aH = 0.29 and aP = 4.84 mT), the other being probably the 4-PyOPN spin adduct of a carbon-centred radical derived from SDS monomers (aN = 1.40 and aP = 3.83 mT).Table 2 EPR parameters of the various CH3 ? spin adducts in pure water and in micelles, determined by EPR signal simulation hfsc values in water/mT hfsc values in micelles/mT Nitroxide 4-PyOPN–CH3 PPN–CH3 4-ClPPN–CH3 aN 1.45 1.49 1.49 aH 0.24 0.35 0.32 aP 4.67 4.66 4.64 aNm 1.43 1.39 1.30 aHm 0.16 0.40 0.30 aPm 4.15 4.39 4.45J. Chem. Soc., Perkin Trans. 2, 1997 2509 tration was increased, we thought that it might correspond to the 4-PyOPN–CH3 spin adduct in the micellar phase.Whatever the SDS concentration, two other radical species were also observed by EPR spectroscopy. The first one showed the following hfsc values: aN = 1.31, aH = 1.26 and aP = 5.20 mT, and has previously been identified as the aminoxyl 5.22 The second one (aN = 1.57, aH = 1.33 (3H) and aP = 5.10 mT) corresponds to nitroxide 6. According to the mechanism proposed for the decay of hydroxyl radical spin adducts of PBN-type nitrones 16 and to previous results,22 both compounds 5 and 6 were believed to be formed from the decomposition of the hydroxyl radical spin adduct of 4-PyOPN (4-PyOPN-OH, 4), as shown in Scheme 1.When the methyl radical was trapped by PPN in water in the absence of SDS, the EPR spectrum of the corresponding spin adduct (PPN–CH3) showed the hfsc values given in Table 2. This signal remained unchanged as long as the SDS concentration was kept below the CMC. From ca. 8 mmol dm23 SDS, significant variations were observed in aN and aP values, while only a slight increase was detected for aH. Thus, aN and aP lost 0.05 and 0.15 mT, respectively, while the SDS concentration was increased from 1 to 100 mmol dm23 (see Fig. 3). The aN decrease indicated a diminution in the polarity of the spin adduct environment, and the aP decrease could also indicate a modification in the nitroxide conformation. These two results were both in favour of a partial penetration of PPN–CH3 into the micellar phase.The aN and aP values obtained for different SDS concentrations correspond to average values of the hfsc of the aminoxyl in water and in the micelle. From this average value found for the coupling with the phosphorus, denoted ·aPÒ, the affinity of the aminoxyl PPN– CH3 for the micellar phase could be estimated. An expression for ·aPÒ is given by eqn. (1), in which aPm and aPw represent aP ·aPÒ = (aPm 2 aPw)xm 1 aPw (1) values in the micellar phase and in water, respectively, xm being the molar ratio of the aminoxyl present in the micelles.The aminoxyl affinity for the micellar phase could be evaluated by a distribution coefficient Kd, defined as the ratio of aminoxyl moles in micelles divided by the number of SDS moles associ- Scheme 1 Decomposition pathway of the hydroxyl radical spin adduct of 4-PyOPN (4-PyOPN–OH, 4) in aqueous medium in the presence of methyl radical N+ CH N O • P(O)(OEt)2 4 – O N+ CH – O OH • N P(O)(OEt)2 O + N P(O)(OEt)2 O N P(O)(OEt)2 • O 6 H3C N+ CH – O OH + N P(O)(OEt)2 – O + • N+ CHO – O N P(O)(OEt)2 • O + H 5 OH •CH3 ated in micelles, to the aminoxyl moles in water divided by the water mole number, as indicated by eqn.(2), in which nNm and Kd = nNm/nSDS nNw/nw (2) nNw represent the number of aminoxyl moles solubilised in micelles and in water, respectively, nSDS being the number of SDS moles associated in micelles and nw the number of water moles in the medium.Eqn. (3) was then derived from eqn. (2). Kd = xmnw (1 2 xm)nSDS (3) Eqn. (1) and (3) led to eqn. (4). ·aPÒ = (aPm 2 aPw)Kd Kd 1 nw/nSDS 1 aPw (4) This distribution coefficient Kd should be considered as a good indicator of the aminoxyl affinity for the micellar phase. In addition, Kd was directly related to the micelle–water partition coefficient KP of the aminoxyl considered, which can be defined as the ratio of the aminoxyl concentration in micelles to that in water.Thus, the relation between Kd and KP is given by eqn. (5), in which Vw and Vm represent the total volume of KP = Kd nSDSVw nwVm (5) water and micellar phases, respectively. Since the mean number of monomers in a micelle and the average radius of an SDS micelle have been reported to be 64 and ca. 25 Å, respectively,25 Vm can be calculated approximately from the total number of SDS moles associated in micelles. Thus, eqn. (5) can be transformed into eqn. (6), which permitted us to evaluate KP.In Fig. 3 Experimental variation of average values of (a) aN and (b) aP of PPN–CH3 (j), 4-PyOPN–CH3(r) and of 4-ClPPN–CH3 (m) versus SDS concentration (logarithmic scale)2510 J. Chem. Soc., Perkin Trans. 2, 1997 KP ª 0.03 Kd (6) Fig. 4, the ·aPÒ diminution observed in our experiments has been plotted against the concentration of SDS monomers associated in micelles, and computer modelling of this decrease has been achieved using eqn. (4). The good fit between experimental points and the calculated curve confirmed the validity of the method used to evaluate the affinity of the aminoxyl PPN–CH3 for the micellar phase.This computer modelling led to the determination not only of Kd, but also of the coupling constant with the phosphorus in a micellar environment, i.e. aPm. We thought that Kd was estimated much more precisely than KP, since the evaluation of the total micellar volume was rather approximate. However, KP could be considered as a more traditional indicator of a compound affinity for a lipophilic phase, and this is the reason why the values obtained for both Kd and KP have been reported in Table 3, together with the hfsc values of PPN–CH3 in micelle determined by two methods.The values thus determined for these two parameters (Kd = 1270 and Kp = 38.1) indicate that PPN–CH3 shows a rather high affinity for the micellar phase. When the same method was applied to evaluate Kd from the variation of the average value of aN, using eqn.(4) in which aP has to be replaced by aN, this led to an evaluation of aN in the micellar environment (i.e. aNm, see Table 3). But in this case, the experimental decrease measured in ·aNÒ was much weaker, thereby making the evaluation of Kd less accurate. The existence of a strong coupling with the phosphorus, which permitted us to correctly evaluate the affinity of PPN–CH3, for the micellar phase, should thus be regarded as an important advantage of the b-phosphorylated spin adducts.In other respects, one should also notice that it was impossible to correctly simulate the PPN–CH3 EPR signals recorded in the presence of micelles without taking into account an exchange of this adduct between the micellar and the water phases. For example, the PPN–CH3 EPR signal shown in Fig. 5 Fig. 4 Modelling of the ·aPÒ variation versus the concentration of SDS monomers associated in micelles (i.e. [SDS] 2 CMC), using eqn.(4). The hfsc aP for the nitroxide in water has been determined on an EPR spectrum recorded in a pure water environment. The modelling led to the following parameters: aPm = 4.40 mT and Kd = 1270. Table 3 Determination of the hfsc values with the phosphorus and the nitrogen nuclei for PPN–CH3 in micelle, and of their distribution (Kd) and partition coefficients (KP) between water and micelle phases. These evaluations have been carried out using computer modelling of the variation observed in the mean hfsc values versus SDS concentration, using eqn.(4). aNm/mT 1.44 ± 0.1 aPm/mT 4.40 ± 0.015 Kd 1270 ± 220 KP 38.1 was recorded in the presence of 25 mmol dm23 SDS. The calculated spectrum, which has been superimposed on the experimental one, was obtained by considering an equilibrium between two adduct forms, one present at 69% corresponding to the aminoxyl solubilised in the bulk aqueous phase, and the other present at 31% (aN = 1.39, aH = 0.40 and aP = 4.39 mT) corresponding to the same aminoxyl located into the micelles.The correlation time of the exchange was found to be ca. 0.5 × 1027 s, which is in the range of the life-time of an SDS monomer in the micellar structure.25 It is important to notice here that almost the same values have been determined for the hfsc with the phosphorus nucleus for the aminoxyl in micelles by computer modelling of the average value of these coupling constants using eqn. (4) and by computer simulation of the EPR signal.The good agreement between these two methods confirms that PPN–CH3 partitioned between water and micellar phases, and corroborates the validity of the evaluation of the spin adduct affinity for the micellar phase using Kd or KP, determined from the variation observed in ·aPÒ. In contrast, these two methods gave rather different results for aN in micelles, and this is because the variation in the average constant ·aNÒ observed at the various SDS concentrations is too weak to allow a valid determination of this parameter in micelles.By trapping CH3 ? with 4-ClPPN in the absence of SDS, the corresponding spin adduct (4-ClPPN–CH3) was observed by EPR in pure aqueous media and its spectrum showed the hfsc values given in Table 2. This spectrum remained unchanged when the SDS concentration was kept below the CMC. However, in the presence of micelles, a significant diminution was observed in aN and aP values while SDS concentration was raised.Thus, aN and aP were found to diminish by 0.5 and 2 mT, respectively, while SDS concentration was increased from 1 to 150 mmol dm23 (see Fig. 3). In addition, an important broadening of the lines was observed in the EPR spectra recorded in the presence of micelles, and this could be due to an immobilisation of 4-ClPPN–CH3 in the micellar structure. But other causes, such as a second exchange of the aminoxyl between the micellar interface and the micellar interior, or the modification of the micelle shape at high SDS concentration, resulting in a diminution of the aminoxyl environment isotropy, could also induce the same kind of phenomenon.Nevertheless, all these results indicated that 4-ClPPN–CH3 was preferentially located in the micellar phase, where its EPR spectrum exhibited the hfsc values indicated in Table 2. Using the method previously described for PPN–CH3, we tried to evaluate the 4-ClPPN–CH3 affinity for the micellar phase by calculating first its Kd from eqn.(4), and then its KP from eqn. (5). However in this case, the determination of these two parameters was too approximate, since the measurement of the average hfsc values was rather difficult because of the Fig. 5 Experimental EPR spectrum of PPN–CH3 (——) recorded in the presence of 25 mmol dm23 SDS and simulation of this signal (· · · ·) obtained by considering two paramagnetic species in rapid equilibrium, the first one (69%) being PPN–CH3 in water (aN = 1.49, aH = 0.35, and aP = 4.66 mT), the second one (31%) being the same spin adduct in micelles (aN = 1.39, aH = 0.40, and aP = 4.39 mT).The exchange correlation time has been evaluated as 0.5 × 1027 s. The lines which are not topped by a cross correspond to paramagnetic byproducts 5 and 6.J. Chem. Soc., Perkin Trans. 2, 1997 2511 broadening of the EPR lines. Thus, the method used to evaluate the two coefficients Kd and KP indicated above cannot be applied when the aminoxyl considered is immobilised in the micellar structure.Last, it is also important to notice here that the 4-ClPPN– CH3 EPR signal was still observed in the presence of 150 mmol dm23 SDS, while 4-ClPPN–CO2 2 was difficult to detect at 50 mmol dm23 SDS. The difference between these two results could be explained by the following hypothesis. 4-ClPPN could be sequestered within the micellar structure by locating its polar group, i.e. the nitrone function, between the polar head groups of the SDS molecules, and by burying its aromatic moiety in the hydrophobic micelle interior.Thus, the methyl radical could be trapped at the water–micelle interface, while CO2~2 could not approach the nitrone function because of electrostatic repulsions with the negatively charged micelle surface. Although this hypothesis is consistent with the results described by either Janzen et al.27 or by Struhl et al.28 in their studies concerning the location of hydrophobic traps in model membranes, it should also be considered that the dimethyl sulfoxide present in the reaction medium could enhance either the micelle permeability to the methyl radical or the nitrone solubility in the water phase.Conclusion This EPR spin trapping study allowed us to determine the localisation of the three nitrones PPN, 4-PyOPN and 4-ClPPN, and of their CO2~2 and CH3 ? spin adducts in water–SDS micelle heterogeneous media.The hydrophilic 4-PyOPN was located in the aqueous phase, while the lipophilic 4-ClPPN was preferentially sequestered into the micellar structure. The nitrone PPN represents an intermediate between these two extreme cases, since it was found to partition between the micelles and the bulk aqueous phase. When CO2~2 was trapped with each one of the three nitrones, the corresponding spin adduct could never enter the micelles, and this is certainly due to the repulsive electrostatic forces between the negatively charged aminoxyls and the sulfate groups of SDS molecules.In contrast, the methyl radical adducts of the nitrones always seemed to be present in both phases. However, in the case of 4- PyOPN–CH3, the penetration of a small amount of this adduct into the micelles was uncertain, since all the spectra recorded for the various SDS concentrations have been correctly simulated considering the presence of two aminoxyls, but without taking into account a chemical exchange.In the case of PPN– CH3 and of 4-ClPPN–CH3, the existence of a rapid exchange of the aminoxyls between the two phases has been clearly and unambiguously demonstrated by the decrease observed in the hfsc values while SDS concentration was increased. As far as we know, regular variations of hfsc values in spin trapping studies performed in the presence of micelles 26,29–31 have never previously been observed. In the present case, such an observation was possible and easy because of the existence in these spin adducts of a strong coupling constant with the phosphorus nucleus, and this should be considered as an important advantage of our new spin trapping agents when comparing them to commercially available PBN-type nitrones.In particular, the aP variation permitted us to evaluate for the first time the partition coefficient Kp of PPN–CH3 between the bulk aqueous phase and the micelles. Since the evaluation of KP depended on an approximation of the total volume of the micellar phase, we proposed the use of another coefficient Kd, the determination of which is much more precise, to quantify the aminoxyl affinity for the micellar phase.Thus, PPN–CH3 (Kd = 1270) was found to partition significantly between the two phases. This determination of Kd, and of KP, based on EPR measurements should be regarded as a new method for evaluating spin adduct lipophilicity, and experiments are planned in our laboratory to extend this technique to various stable b-phosphorylated aminoxyls.Nevertheless, this method is not usable when the aminoxyl considered shows too high an affinity for either the bulk aqueous phase (e.g. 4-PyOPN–CH3) or for the micellar structure (e.g. 4-ClPPN–CH3). On the other hand, it should be mentioned here that, using simulation software elaborated by Rockenbauer,32 which permitted us to fit an experimental spectrum to a rapid exchange of the aminoxyls between the two phases, we were able to determine not only the EPR parameters of the methyl radical spin adducts in both water and micelles, but also the rate of the aminoxyl exchange between the two phases.However, the SDS micelles represent a rather simplified biomembrane model, since the micelle core is strongly hydrophobic while intracellular media are aqueous. Thus, considering eventual in vivo applications of our traps, 4-PyOPN seems to be too hydrophilic and would remain in the extracellular environment, while the too lipophilic 4-ClPPN would be located in the phospholipid bilayer of cellular membranes.Nevertheless, a likely hypothesis is that PPN could partition first between the extracellular environment and the biomembrane, and then between the biomembrane and the intracellular medium. In this case, PPN could be a useful tool in the study of radical processes occurring in cells. Work is in progress in our laboratory to verify this hypothesis. Experimental The three spin trapping agents PPN, 4-PyOPN and 4-ClPPN were synthesised and purified in our laboratory as described previously.24 SDS, iron(II) sulfate, sodium formate, hydrogen peroxide and EDTA (ethylenediaminetetraacetic acid) were purchased from Sigma Chemical Co.and used without further purification. Water and dimethyl sulfoxide (DMSO) were distilled twice before use. The CO2~2 and the CH3 ? radical spin adducts were obtained by carrying out a standard Fenton reaction in the presence of a nitrone and of sodium formate or of DMSO, respectively.To an aqueous solution of 0.05 mol dm23 PPN or 4-PyOPN, or to a saturated solution of 4-ClPPN, containing EDTA (2 mmol dm23), FeSO4 (5 mmol dm23), SDS (0–150 mmol dm23), and either sodium formate (0.2 mol dm23) or DMSO (10%), was added H2O2 (0.2%). The two carbon-centred radicals were then always generated in the bulk aqueous phase. In addition, CO2~2 was unable to enter the micelles because of electrostatic repulsive forces with the sulfate groups of the micelle surface.Samples were then transferred in capillary tubes and EPR spectra were recorded at 25 8C using a computer-controlled Bruker EMX spectrometer operating at X-band with 100 kHz modulation frequency, and equipped with an NMR gaussmeter for magnetic field calibration. The instrument settings were as follows: non-saturating microwave power, 10 mW; modulation amplitude, 0.08 mT; receiver gain ranging from 12 × 103 to 63 × 103; scan time, 180 s; time constant, 0.128 s.The standard EPR signal simulations were achieved using the computer program elaborated by Dulling.33 When the spin adducts were found to partition between the two phases, the EPR spectrum simulations were realised using the computer program elaborated by Rockenbauer,32 which allows spectra resulting from two paramagnetic species in rapid equilibrium to be calculated. Acknowledgements We are very grateful to Professor Antal Rockenbauer, from the Technical University of Budapest, who kindly provided us with his EPR simulation computer program.References 1 B. Halliwell and J. M. C. Gutteridge, Free Radicals in Biology and Medicine, 2nd edn., Clarendon Press, Oxford, 1989.2512 J. Chem. Soc., Perkin Trans. 2, 1997 2 H. Sies, Am. J. Med., 1991, 91, 31S. 3 M. Martinez-Cayuela, Biochim., 1995, 77, 147. 4 G. R. Buettner and R. P. Mason, Methods Enzymol., 1990, 186, 127. 5 G. M. Rosen and E.Finkelstein, Free Rad. Biol. Med., 1985, 1, 345. 6 G. R. Buettner and L. W. Oberley, Biochem. Biophys. Res. Commun., 1978, 83, 69. 7 I. Yamazaki, L. H. Piette and T. A. Grover, J. Biol. Chem., 190, 265, 652. 8 C. Fréjaville, H. Karoui, B. Tuccio, F. Le Moigne, M. Culcasi, S. Pietri, R. Lauricella and P. Tordo, J. Chem. Soc., Chem. Commun., 1994, 1793. 9 C. Fréjaville, H. Karoui, B. Tuccio, F. Le Moigne, M. Culcasi, S. Pietri, R. Lauricella and P. Tordo, J. Med. Chem., 1995, 38, 258. 10 B. Tuccio, R. Lauricella, C. Fréjaville, J. C. Bouteiller and P. Tordo, J. Chem. Soc., Perkin Trans. 2, 1995, 295. 11 E. G. Janzen, M. S. West, Y. Kotake and C. Du Bose, J. Biochem. Biophys. Methods, 1996, 32, 183. 12 E. A. Konorev, J. E. Baker, J. Joseph and B. Kalyanaraman, Free Rad. Biol. Med., 1993, 14, 127. 13 S. Pou, C. L. Ramos, T. Gladwell, E. Renks, M. Centra, D. Young, M. S. Cohen and G. M. Rosen, Anal. Biol., 1994, 217, 76. 14 J. R. Harbour, V. Chow and J. Bolton, Can. J. Chem., 1974, 52, 3549. 15 E. Finkelstein, G. Rosen and E. Rauckman, Arch. Biochem. Biophys., 1980, 200, 1. 16 Y. Kotake and E. Janzen, J. Am. Chem. Soc., 1991, 113, 9503. 17 E. G. Janzen, Y. Kotake and R. D. Hinton, Free Rad. Biol. Med., 1992, 12, 169. 18 E. G. Janzen, R. D. Hinton and Y. Kotake, Tetrahedron Lett., 1992, 33, 1257. 19 K. Pan, C. Lin and T. I. Ho, Magn. Reson. Chem., 1993, 31, 632. 20 E. G. Janzen, C. M. Du Bose and Y. Kotake, Tetrahedron Lett., 1990, 31, 7395. 21 Y. Abe, S. Seno, K. Sakakibara and M. Hirota, J. Chem. Soc., Perkin Trans. 2, 1991, 897. 22 A. Zeghdaoui, B. Tuccio, J. P. Finet, V. Cerri and P. Tordo, J. Chem. Soc., Perkin Trans. 2, 1995, 2087. 23 B. Tuccio, A. Zeghdaoui, J. P. Finet, V. Cerri and P. Tordo, Res. Chem. Intermed., 1996, 22, 393. 24 V. Roubaud, R. Lauricella, B. Tuccio, J. C. Bouteiller and P. Tordo, Res. Chem. Intermed., 1996, 22, 405. 25 A. M. Wasserman, Russ. Chem. Rev., 1994, 63, 373. 26 D. P. Bakalic and J. K. Thomas, J. Phys. Chem., 1977, 81, 1905. 27 E. G. Janzen, D. L. Haire, G. H. Coulter, H. J. Stronks, P. H. Krygsman, R. A. Towner and J. W. Hilborn, J. Org. Chem., 1989, 54, 2915. 28 G. Struhl, H. E. Gottlieb, A. A. Frimer and L. Weiner, J. Chem. Soc., Perkin Trans. 2, 1994, 1229. 29 J. R. Harbour and J. R. Bolton, Photochem. Photobiol., 1978, 28, 231. 30 T. H. Walter, E. E. Bancroft, G. L. McIntire, E. R. Davis, L. M. Gierasch, H. N. Blount, H. J. Stronks and E. G. Janzen, Can. J. Chem., 1982, 60, 1621. 31 E. G. Janzen and G. A. Coulter, J. Am. Chem. Soc., 1984, 106, 1962. 32 A. Rockenbauer, Appl. Magn. Reson., 1996, 10, 29. 33 D. R. Dulling, J. Magn. Reson., 1994, 104, 105. Paper 7/02474F Received 10th April 1997 Accepted 24th July 1997

ISSN:1472-779X    

DOI:10.1039/a702474f

出版商:RSC    

年代:1997

数据来源: RSC