SOLVENT EFFECTS IN THE ELECTRON TRANSFER
QUENCHING OF THE TRIPLET STATE OF ANTHRACENE
BY p-BENZOQUINONES

Graciela P. Zanini and Hernan A. Montejano*

Departamento de Química y Física
Universidad Nacional de Río Cuarto
5800 Río Cuarto - Argentina
E-mail: hmontejano@exa.unrc.edu.ar
(Received: December 15, 1998 - Accepted: November 29, 2000)

ABSTRACT

Here we present an investigation of the quenching of anthracene triplet by several p-benzoquinones, in two solvents of nearly equal macroscopic dielectric properties, methanol and acetonitrile. Bimolecular quenching rate constants were measured by laser flash photolysis. Quantum efficiencies of charge separation were also determined in the protic solvent. Quenching rate constants in acetonitrile follow a typical Rehm-Weller correlation for an electron transfer process. Although free radical ions are observed in methanol, the rate constants are nearly independent on the driving force, at variance of the expected behavior. In this solvent, the charge separation quantum yield decreases when the energy of the ion-pair decreases, in agreement with the energy gap law.

KEYWORDS: electron transfer, solvent effects, anthracene, quinones, quenching.

RESUMEN

Se presenta aquí una investigación de la desactivación del estado triplete de antraceno por p-benzoquinonas, en dos solventes de propiedades dieléctricas macroscópicas aproximadamente iguales, metanol y acetonitrilo. Las constantes de velocidad bimoleculares de desactivación fueron medidas por laser flash fotólisis. Las eficiencias cuánticas de separación de carga fueron determinadas en el solvente prótico. Las constantes de velocidad en acetonitrilo siguen una correlación de Rehm-Weller típica para un proceso de transferencia de electrones. Aunque se han observado iones radicales libres en metanol, las constantes de velocidad son aproximadamente independientes de la fuerza impulsora, contrariamente al comportamiento esperado. En este solvente, el rendimiento cuántico de separación de cargas disminuye cuando disminuye la energía del par iónico, en acuerdo con la ley del salto de energía.

PALABRAS CLAVES: transferencia de electrones, efectos de solvente, antraceno, quinonas, desactivación.

INTRODUCTION

The solvent effect on the photoinduced electron transfer (PET) reactions has been widely investigated¹⁾. However most of the works have been focused on intramolecular processes. For these cases, several experimental²⁾ and theoretical³⁾papers have dealt with the effects of solvent reorganization and dynamics. On the other hand, the intermolecular processes have received little consideration until recently. Particularly, several aspects related to the solvent effects in the intermolecular electron transfer reactions are not satisfactorily explained by the simple Rehm-Weller-Marcus model.

During the last years, in our work group has studied about the solvent effects on several intermolecular PET reactions. Same important thermodynamic and kinetic differences in PET reactions in protic and aprotic solvents were established in these papers. These differences are shown in bimolecular quenching rate constants, in their activation parameters and in the charge separation efficiency. The most remarkable finding related to quenching rate constants was that in two solvents of similar macroscopic dielectric properties such as methanol (e = 32.6, n_D = 1.331) and acetonitrile (e = 35.9, n_D = 1.342) the rate constants differ, in some cases, in more than one order of magnitude. An example of such effect was observed for the excited state quenching of Ru(bpy)₃⁺² by aromatic amines and nitrobenzenes⁴⁾. In all cases the rate constants are higher in methanol than in acetonitrile. Also, in the PET quenching of excited singlet states of polycyclic aromatic hydrocarbons (PAHs) by nitro and cyanobenzenes⁵⁾ the rate constants happen to be generally higher in methanol than in acetonitrile. Similar results are observed in the PET quenching of the triplet state of PAHs by nitrobenzenes⁶⁾. The observed solvent effects on the rate constants can not be adequately explained by Rehm-Weller-Marcus model if a continuum model of the solvent is considered.

In order to obtain a further insight on the solvent effect on bimolecular PET rate constants, we present here an investigation of the quenching of anthracene triplet by several p-benzoquinones, in two solvents of nearly equal macroscopic dielectric properties, methanol and acetonitrile.

The interaction of the excited states of aromatic molecules with benzoquinones is of great interest because the latter act as electron acceptors in a number of photobiological processes⁷⁾. A great emphasis in experimental work was placed on intramolecular electron transfer reactions involving a porphyrin as an electron donor covalently linked to a quinone⁸⁾. However, the intermolecular processes received comparatively less attention. In particular, the solvent effects on the rate constants and the charge separation efficiency seem to present special characteristics that merits a more profound investigation.

The quenching of the triplet states of aromatic hydrocarbons by quinones was previously investigated by Wilkinson and Schroeder⁹⁾ in benzene by laser flash photolysis. They were not able to detect transient species resulting from the quenching process. In a second paper10) the same authors studied the solvent effect on the triplet quenching of anthracene and 1,2-benzanthracene by p-quinones in several aprotic solvents. In low polarity solvents (e < ¹⁰⁾ they did not observe transients other than the triplet state of the arene. In aprotic polar solvents a weak long-lived transient species absorbing at 405 nm was observed and was ascribed to the quinone radical anion. While in low polarity solvents they observed a correlation between the rate constants and the Gibbs energy change DG_CT, in polar solvents there is not correlation of k_q neither with the solvent dielectric constant nor with DG_CT. They suggest that quenching may result in these solvents from only partial charge transfer.

Kuzmin et al.¹¹⁾ also investigated the quenching of triplet anthracene by quinones in acetonitrile. They found a linear correlation between the quenching rate constants and the reduction potential of the quinone.

Hilinski et al.¹²⁾ investigated the mechanism of PET between chloranil and arenes, naphthalene, 9,10 dihydrophenanthrene and indene in acetonitrile. With the use of picosecond absorption spectroscopy they were able to obtain direct information on the electron transfer process from ground state arene to excited singlet of chloranil.

Previtali et al.¹³⁾ previously investigated the singlet and triplet excited states quenching of the safranine by quinones in methanol and acetonitrile. It was concluded that the quenching process occurs by an electron transfer process with the excited dye acting as an electron donor. A Rehm-Weller type correlation between the rate constants and the overall free energy change for the electron transfer process was found in acetonitrile. However in methanol the rate constants for triplet quenching are substantially larger than those in acetonitrile.

EXPERIMENTAL

Anthracene, (Aldrich, Gold Label > 99.99 %) was used without further purification. The quinones (Scheme I), p-benzoquinone (BQ), methyl-p-benzoquinone (toluoquinone, TQ), 2,5-dimethyl-p-benzoquinone (DMBQ), duroquinone (DQ), chloro-p-benzoquinone (ClQ) and p-chloranil (Cl4Q) were obtained from various commercial sources. They were purified by recrystallization and/or sublimation when necessary, otherwise they were used as received. Methanol (MeOH) and acetonitrile (MeCN) were HPLC grade.

Transient absorption measurements were performed by excitation with a nitrogen laser (Laseroptics, 7 ns FWHM and 5 mJ per pulse at 337 nm). A Xe lamp, a monochromator and a red extended photomultiplier tube comprises the analyzing beam at right angles of the laser beam. The signal was acquired by a digitizing scope where it was averaged and then transferred to a computer.

Bimolecular quenching rate constants were determined from the slopes of the observed first order decay of anthracene triplet as a function of the quinone concentration. Replicate runs produced values that were within ± 10%. Quantum yields for free radical ion formation were calculated using the triplet-triplet absorption of zinc tetraphenylporphyrin Zn(TPP) in benzene as actinometer¹⁴⁾. All measurements were performed in deaerated solutions at 298 K.

RESULTS AND DISCUSSION

Quenching rate constants

In the presence of quinones in both solvents anthracene triplet decays by first order kinetics. In Fig. 1 the triplet decays at 425 nm in the presence of DMBQ are shown in both solvents. It can be seen that in acetonitrile the absorption decays to zero. However, in MeOH, after the triplet has totally decayed, a long lived absorption remains at the same wavelength of the triplet state. This new transient decays in the millisecond time scale with a non exponential kinetics. It is well known that benzoquinones radical anions (Q.-) and their protonated forms (QH.) absorb at this wavelength¹⁶⁾. Thus the long lived absorption can be ascribed to the Q^.- / QH^. pair produced in the protic solvents.

Bimolecular quenching rate constants are collected in Table I. The results in MeCN are in agreement with those of Wilkinson, et al.¹⁰⁾. It can be seen that the quenching rate constant decreases as quinones reduction potential become more negative. Therefore, the quenching reaction may be ascribed to an electron transfer process:

According to Rehm and Weller¹⁷⁾ the Gibbs energy change DG^o in the electron transfer process can be described as:

Therefore, DGo can be calculated from the redox potentials of anthracene E_(An/An+) (1.09 V vs. SCE in MeCN)¹⁸⁾and the quinones (see Scheme I), and the energy of the excited triplet state of anthracene, E* (1.84 eV)¹⁹⁾. The last term in Eqn. (1) represent the coulombic energy necessary to form an ion pair with charges Z₁ and Z₂ in a medium of dielectric constant e at a distance r_12. The distance r₁₂ was assuming as 0.7 nm. The DGo values are also presented in the Table I.

The Rehm-Weller¹⁷⁾ mechanism may be used to discuss the observed trend in rate constants. This mechanism, in most cases, can be written as:

where ³Ant and Q represents the anthracene triplet and the p-benzoquinone ground states respectively, and the rate constant k_r includes all processes competing with the electron back-transfer to the precursor complex: electron back-transfer to the ground state, cage escape, etc. If k_ris assumed to be much larger than k_-et, the experimental steady state quenching rate constant k_q can be expressed as:

where k_d is the diffusional rate constant and K_D = k_d/k-_d is the equilibrium constant for the formation of precursor complex. Since k_d may be estimated with a high degree of confidence, the product K_Dk_et may be obtained from the experimental k_q using eqn. (2). For uncharged reactants K_D depends only on r₁₂, the donor-acceptor distance in the precursor complex. Usually K_D is a value²⁰, ²¹⁾ between 0.1 and 0.8 M-1.

The electron transfer rate constant k_et can be written²¹⁾ as:

where k is the transmission coefficient ^.(unity for an adiabatic reaction) and n_n is a nuclear frequency factor (usually between 10¹¹ and 10¹² s^-1). The activation Gibbs energy given by the Marcus classical theory for electron transfer²¹⁾ is defined by:

where l is the reorganization energy.

If eqns. (3) and (4) are included in eqn. (2), a functional dependence of k_q with DG^o is obtained:

In Fig. 2 the experimental rate constants k_q in both solvents are plotted vs. DG^o. It can be seen that the data form two separate sets of values. The fitting in MeCN was made assuming that the equilibrium constant for the formation of precursor complex with neutral reactants, the diffusional rate constant and the transmission coefficient are constant and correspondent to typical values²²⁾: 0.16 M^-1, 2.0x10¹⁰ M^-1s^-1 and 1 (adiabatic assumption) respectively, while n_n and l were assumed variables parameters.

The solid line is the best fitting in MeCN according to eqn. (5) with the following parameters: n_n = 7.1x10¹¹ s^-1 and l = 0.73 eV. These parameters are in good agreement with those that provide the best fit for the electron transfer quenching of triplet state of Ant and a,h-dibenzanthracene by nitrobenzenes as electron acceptors⁶⁾ and with the electron transfer quenching of triplet state of safranine-T by p-benzoquinones as electron acceptors¹³⁾, both in MeCN. This is another confirmation of the quenching mechanism.

On the other side, if the data for the protic solvent are treated in the same way there is not agreement with the theory. The experimental rate constants are practically independent on the driving force for DG^ovalues lower than - 0.1 eV, with values close to, but lower than, the diffusional limit. Moreover, for the two extreme values of DG^othe rate constants are particularly striking. For the more exergonic reaction, i.e. the quenching by chloranil with DG^o= -0.82 eV a diffusional rate constant is expected. However, it is clearly lower than the diffusion limit in MeOH (1,2x10¹⁰ M^-1s^-1)²³⁾. On the other extreme, DQ quench with a rate constant that is two orders of magnitude higher than the value in acetonitrile. The fact that methanol reactivity is greater than the analog value in acetonitrile was found previously in other systems⁵,⁶,¹³,²⁴,²⁵⁾.

There are not many records of this non Rehm-Weller’s correlation for reactions that have been precisely identified as photoinduced electron transfer processes. Therefore, we put special care in the characterization of the quenching process in methanol. The electron transfer nature of the reaction in methanol is further confirmed by the transient absorption spectra shown in Fig. 3. In the absence of the quinone, the T-T absorption of anthracene is the only feature of the spectrum. In the presence of DQ a new absorption band appears in the region of 700 nm. This spectrum is taken with a low quinone concentration and therefore a low fraction of triplet states quenched. The position of the maximum and band shape of the new absorption is typical of anthracene radical cation¹⁶,²⁶,²⁷⁾. Similar spectral characteristics are observed in the presence of the other quinones. After the initial growth the anthracene radical cation decays by a fast first order process. At higher quinone concentrations, and longer times, the only absorption remaining is the one shown in Fig. 4. The double peak at 425 - 445 nm corresponds to the characteristic absorption in this region reported for p-benzoquinones radical anions and their protonated forms¹⁶,²⁸,²⁹⁾. Thus, the electron transfer nature of the quenching process in MeOH is well established.

Charge separation efficiency and the reaction of anthracene radical cation

As it was pointed above, after an initial growth the radical cation decays by a first order rate law with a lifetime of a few microseconds. It was found that the radical cation lifetime decreases when anthracene or quinone concentrations increases. The second order rate constants for the reaction with ground state anthracene was measured as 6x10⁸ M^-1s^-1 (Fig. 5). Reactions of radical cations with their parent compounds, commonly termed radical cation dimerizations, are well documented³⁰⁾. On the other hand, the radical cation reacts faster with the quinones. Thus, a rate constant of 2.4x10⁹ M^-1s^-1was measured for DQ (Fig. 5) and nearly the same value was measured for Cl₄Q. This reaction may be understood as an addition of the radical cation to the quinone³¹⁾.

Quantum yields of free ions f_ion were determined in MeOH from the extrapolated initial absorbance of the anthracene radical cation and from the long lived absorption of the hidroquinone radical QH^. (which is formed by protonation of Q^.- in the solvent cage)²⁸⁾ using the molar extinction coefficients as given in the literature: e₇₂₀ = 10000 M^-1cm^-1 (for Ant.+)¹⁶,²⁶⁾, e₄₂₀ = 5500 M^-1cm^-1 (for DQH.)³²⁾ and e₄₅₀ = 6600 M^-1cm^-1 (for Cl₄QH.)³³⁾. Extrapolation to zero time of the radical cation absorbance used to calculate quantum yields of free ions, allows to disregard those reactions that consume the radical cation. The results obtained from Ant.+ and QH. absorbances were coincident within ±5%.

Charge separation efficiencies h from the quantum yields of free ions were obtained from:

where 3t_o is the lifetime of the excited triplet state in the absence of the quinone. The results are collected in Table II. The values of the corresponding driving force for the back electron transfer DG_b from the successor complex to ground state was calculated from eqn. (7) and are also included in Table II.

It can be seen that the charge separation efficiencies decrease as the back electron transfer reaction becomes less exergonic. This is the expected behavior according to the energy gap law for electron transfer reactions in the inverted region (DG_b < - l). This is an additional proof for the electron transfer nature of the quenching process in MeOH.

Having confirmed that the reactions under consideration are PET processes with total charge transfer, two remarkable facts arise from the experimental results and must be explained:

1) k_q values higher in MeOH than in MeCN in the region of low exergonicity are observed (almost two orders of magnitude for DQ)

2) Contrary to what was observed in MeCN, kq values in MeOH do not follow the behavior predicted by Rehm-Weller Model. This is, for k_q values to increase up to reach the diffusional rate regime when the reaction driving force is highly exergonic. k_q values reach a maximum value below -0.1 eV. This maximum value is remarkably lower to the diffusional limit in this solvent.

Referring to item 1, the classical Marcus theory can not explained the observed differences if a dielectric continuum model is used for the solvent. On the other hand, if a microscopic model that contemplates the molecular structure of the solvent and its interactions with reacting molecules is considered, it can be inferred that the protic solvents may alter both, the reorganization energy and the redox potentials as compared with those determined in MeCN.

Respecting the reorganization energy, the k_q values higher in MeOH can be interpreted by considering kinetic evidences showing that, in protic solvents the entropy activation change DS^# for electron transfer processes is lower than the corresponding value in aprotic solvent²⁵⁾. This can be understood through the entropy changes caused by the solvent reorganization on forming the transition state. In a solvent like MeOH, the molecules are in a more ordered structure in the initial state due to hydrogen bonding, therefore the entropy change for reaching the transition state will be lesser than in an aprotic solvent like MeCN. Summarizing, a minor change in the solvent reorganization is in agreement with a k_q increase.

On the other hand, respecting the redox potentials, when studying the solvent dependence on the electron transfer reaction from triplet N,N-dialkyl-1-naphthylamines to benzophenones, Shizuka et al³⁴⁾ observed that the driving force for intraexciplex electron transfer is the negatively enlarged reduction potential of benzophenone in the exciplex due to the hydrogen bonding to the carbonyl group. This effect can be valid in the present system since the quinone anion radicals and alcohol molecules can form a complex through hydrogen bonding³⁵⁾.

In the present work we used standard redox potentials measured in MeCN for both solvents, due to the fact the quinone redox products exhibit protonation reactions in protic solvents, this redox potential values do not represent simple redox equilibria as in the case of MeCN and can be used only as an approximation of the thermodynamic values. In this sense Mäntele et al³⁶⁾ have measured the standard redox potentials of nine p-benzoquinones by cyclic voltammetry in protic as well as aprotic solvents. In protic solvents only one standard redox potential value could be determined and they found differences up to 0.3 V less negative in MeOH than MeCN. If these differences are assumed in the calculus of DG^o for our results a shift to more exergonic region of k_q values is produced. This might be another possible justification for the great difference observed in k_q for DQ since this displacement would produce an approximation of the results corresponding to MeOH with the correlation type Rehm-Weller for MeCN. This hydrogen bonding interactions of the quenchers with the hydroxylic solvents seem to be a distinctiveness of the oxidative quenching by quinones. A similar departure from the normal behavior was observed for the PET quenching of the triplet state of safranine by benzoquinones in MeOH¹³⁾.

Respecting to the item 2, no explanation has been proposed for the fact that the rate constants are nearly independent on the driving force at values remarkably lower than those expected for a diffusional regime in methanol. Even when references to this type of behavior are scarce in literature, it has been observed in the electron transfer quenching of the exited states of thionine by amines in MeOH³⁷⁾. Though no explanation has been proposed for this unusual behavior, the remarkable fact is that this comportment has been observed in MeOH.

In summary, the quenching of the triplet state of anthracene by quinones is strongly solvent dependent. While the solvent effects on PET rate constants may be reasonably explained by a continuum dielectric model for the aprotic solvent, a microscopic model that contemplates the molecular structure of the solvent and its interaction with the reacting molecules is required to justify the results obtained in the hydroxylic solvent.

ACKNOWLEDGMENTS

Thanks are given to the Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET), Consejo de Investigaciones Científicas y Tecnológicas de la Provincia de Córdoba (CONICOR), Agencia Nacional de Promoción Científica y Tecnológica (ANPCyT) and Secretaría de Ciencia y Técnica de la Universidad Nacional de Río Cuarto for financial support.

REFERENCES

1. J. R. Bolton, N. Mataga, G. McLendon (eds.), Electon Transfer in Inorganic, Organic and Biological Systems, American Chemical Society, Washington, DC, 1991.

2. E. M. Kosover, D. Huppert, Annu. Rev. Phys. Chem., 37, 127 (1986). J. A. Schmidt, J. Y. Liu, J. R. Bolton, M. D. Archer, V. P. Y. Gadzekpo, J. Chem. Soc. Faraday Trans., 1 85, 1027 (1989). G. P. Wiederrecht, S. Watanabe, M. R. Wasielewski, Chem. Phys., 176, 601 (1993).

3. R. Y. Cukier, D. G. Nocera, J. Chem. Phys., 97, 7371 (1992); M. Dutkiewicz, J. Chem. Soc. Faraday Trans., 92, 637 (1996).

4. H. A. Garrera, J . J. Cosa and C. M. Previtali., J. Photochem. Photobiol. A.: Chem., 47, 147 (1989); H. A. Garrera, J . J. Cosa and C. M. Previtali., J. Photochem. Photobiol. A.: Chem., 56, 267 (1991). 0

5. G. P. Zanini, H. A. Montejano and C. M. Previtali, J. Chem. Soc. Faraday Trans., 91, 1197 (1995).

6. H. A. Montejano, V. Avila, H. A. Garrera and C. M. Previtali., J. Photochem. Photobiol. A.: Chem., 72, 117 (1993).

7. M. Gouterman and D. Holten, Photochem. Photobiol., 25, 85 (1977).

8. J.A. Schmidt, A.R. McIntosh, A.C. Weedon, J. R. Bolton, J.S. Connolly, J.K. Hurley and M.R. Wasielewski, J. Am. Chem. Soc., 110, 1733 (1988); D. Gust and T. A. Moore, Adv. Photochem., 16, 1 (1991); V. V. Borokov, A. Ishida, S. Takamuku and Y. Sakata, Chem. Lett., 737 (1993); M. Fuchs, J. Von Gersdorff, H. Dicks, H. Kurreck, K. Möbius and T. Prisner, J. Chem. Soc. Faraday Trans., 92, 949 (1996).

9. F. Wilkinson and J. Schroeder, J. Chem. Soc. Faraday Trans. II, 75, 441 (1979).

10. J. Schroeder and F. Wilkinson, J. Chem. Soc. Faraday Trans. II, 75, 896 (1979).

11. V. A. Kuzmin, I.V. Renge and Yu. E. Borisevich, Chem. Phys. Lett., 70, 257 (1980).

12. E. F. Hilinski, S.V. Milton and P. M. Rentzepis, J. Am. Chem. Soc., 105, 5193 (1983).

13. S. G. Bertolotti and C. M. Previtali, J. Photochem. Photobiol. A: Chem., 103, 115 (1997).

14. I. Carmichael and G. L. Hug, J. Phys. Chem. Ref. Data, 1, 15 (1986).

15. S. Fukuzumi, S. Koumitshu, K. Hironaka, and T. Tanaka, J. Am. Chem. Soc., 109, 305 (1987); H. Kim, N. Kitamura, Y. Kawanishi and S. Tazuke, J. Phys. Chem., 93, 5757 (1989).

16. T. Shida, Electronic Absortion Spectra of Radical Ions, Elsevier, Amsterdan, 1988.

17. D. Rehm and A. Weller, Isr. J. Chem., 8, 259 (1970).

18. A. Baird and H. Lund (eds.), Encyclopedia of the Electrochemistry of the Elements. Organic Section, Vol XI, Marcel Dekker, New York, 1984.

19. S. Murov, Handbook of Photochemistry, Marcel Dekker, New York, 1973.

20. F. Meeus, M. Van der Auweraer and F.C. De Schryver, J.Am.Chem.Soc., 102, 4017 (1980).

21. R.A. Marcus and N. Sutin, Biochim.Biophys.Acta, 811, 265 (1985).

22. L. Eberson, Adv. Phys. Org. Chem., 18, 79 (1982).

23. S.L. Murov, I. Carmichael and G. L. Hug, Handbook of photochemistry, 2nd. ed. New York, 1993.

24. E. O’Driscoll, J. Simon and K. Peters, J. Am. Chem. Soc., 112, 7091 (1990).

25. C. M. Previtali, Pure Appl. Chem., 67, 127 (1995).

26. P. K. Das, J. Chem. Soc., Faraday Trans. I, 79, 1135 (1983).

27. A. Trifunac, J. Photochem. Photobiol. A: Chem., 67, 197 (1992).

28. P. Neta in S. Patai and Z. Rappoport (eds.), The Chemistry of Quinonoid Compounds Vol II, 1988, Wiley & Sons Ltd.

29. J. Mayer and R. Krasiuklanis, J. Chem. Soc., Faraday Trans., 87, 2941 (1991).

30. N.P. Schepp and L. J. Johnston, J. Am. Chem. Soc., 118, 2872 (1996), and references therein.

31. P. Nebois, Z. Bouaziz, H. Fillion, L. Moeini, M. Aurell Piquer, J. Luche, A. Riera, A. Moyano and M. Pericàs, Ultrasonics Sonochem., 3, 7 (1996).

32. E. Amouyal and R. Bensasson, J. Chem. Soc. Faraday Trans. I, 72, 1274 (1976)

33. G. Jones and N. Mouli, J. Phys. Chem., 92, 7174 (1988).

34. T. Kiyota, M. Yamaji and H. Shizuka, J. Phys. Chem., 100, 672 (1996).

35. P. O’Malley, J. Phys. Chem. A., 102, 248 (1998).

36. M. Bauscher and W. Mäntele, J. Phys. Chem., 96, 11108 (1992).

37. M. Neumann and M. Rodrigues, Polymer, 39, 1657 (1998)