- Email: [email protected]
Anion and electron solvation in alcohols1
PERGAMON
Radiation Physics and Chemistry Radiation Physics and Chemistry 54 (1999) 433±440
Reviews on Special Topics
Anion and electron solvation in alcoholsp Xujia Zhang *, Yi Lin, Charles D. Jonah Chemistry Division, Argonne National Laboratory, Argonne, IL 60439, USA Received for publication 19 November 1997
Abstract Pulse radiolytic techniques have been used to measure the solvation of anions and electrons in alcohols. The results have shown the importance of the microscopic liquid structure in the solvation process. The activation energy for benzophenone solvation is equal to the hydrogen-bond energy for the liquids, which shows that the solvent reorganization requires the breaking of the hydrogen bonds between solvent molecules. For electron solvation, primary alcohols have a lower activation energy because the initial hydrogen-bonded structure of the liquid is amenable to solvation. The solvation of an electron in a secondary alcohol requires require hydrogen-bond breakage. # 1999 Published by Elsevier Science Ltd. All rights reserved.
1. Introduction The solvent has long been known to play an important role in chemical reactions (Reichardt, 1988). There are many mechanisms by which the solvent can aect chemical reactivity. These include trapping of reactants, limiting mobility and removing or adding energy to the chemical system. In polar ¯uids, one of the most important ways that the solvent can alter reactions is through solvation. In solvation, the solvent responds to a solute molecule in a way that lowers the energy of the system. Electrostatic forces are one of the strongest forces that will cause solvent reorganization. The
* Corresponding author. The submitted manuscript has been created by the University of Chicago as Operator of Argonne National Laboratory (`Argonne') under Contract No. W-31-109-ENG38 with the US Department of Energy. The US Government retains for itself, and others acting on its behalf, a paid-up, nonexclusive, irrevocable worldwide license in said article to reproduce, prepare derivative works, distribute copies to the public and perform publicly and display publicly, by or on behalf of the Government. Work performed under the auspices of the Oce of Basic Energy Sciences, Division of Chemical Science, US-DOE under contract number W-31109-ENG-38. p
charge distribution on a solute molecule will interact with the charge distribution on a solvent molecule and these interactions will reorient the solvent molecules to create a lower energy system. In general, this process will raise the energy of the solvent while lowering the energy of the system. For example in alcohols, the solvent molecules will be oriented by an anion so that the H of the O±H group will point towards the anion. To do this, the hydrogen bond that the hydrogen had formed with other solvent molecules must be broken. It is this balancing between solvent±solute interactions and solvent±solvent interactions that determine the solvation energetics. Both the energetics and the kinetics of solvation are important in modifying chemical reactivity. The amount by which the solvation process alters the energy of reacting states can determine whether the reaction pathway and velocity will be altered by solvation while the kinetics will de®ne whether the solvation will be fast enough to in¯uence the reaction. Early experiments probed solvation energetics by measuring spectral shifts in dierent solvents (Amos and Barrows, 1973; Nicol, 1974; Davis, 1975; Jauquet and Laszlo, 1975). The shift in the absorption spectrum re¯ects a change in the dierence of the energy of the ground state and the excited state in dierent
0969-806X/99/$ - see front matter # 1999 Published by Elsevier Science Ltd. All rights reserved. PII: S 0 9 6 9 - 8 0 6 X ( 9 7 ) 0 0 3 1 5 - 0
434
X. Zhang et al. / Radiation Physics and Chemistry 54 (1999) 433±440
solvents. These energy changes re¯ect the dierences in solvation. Early kinetic measurements of solvation include the measurement of electron solvation (Baxendale and Wardman, 1973, 1977; Chase and Hunt, 1975; Kenney-Wallace and Jonah, 1976, 1982; Gilles et al., 1977; Okazaki and Freeman, 1978; Kevan, 1980, 1981; Migus et al., 1987) and the measurements in the shift of an emission spectrum as a function of time (see recent reviews of Barbara and Jarzeba, 1990; Maroncelli, 1993). The studies of emission spectra determined the solvation of an excited state. Experiments have also been done where the solvation of an ionic excited state was measured (Tominaga and Walker, 1995). That experiment measured the solvation of the excited state of an ion. The solvent was already arranged around the ion and the perturbation of the solvent was done by the excited states rather than by the creation of charge. Pulse radiolytic techniques were applied to the measurement of anion solvation by Marignier and Hickel, who measured the solvation of the benzophenone anion in low temperature alcohols (Marignier and Hickel, 1982, 1984). Building on their results, we carried out experiments using the Argonne stroboscopic pulse radiolysis system and were able to make measurements for a wide variety of alcohols at room temperature and at temperatures to about ÿ608C (Lin and Jonah, 1992, 1993, 1994; Zhang and Jonah, 1995). Because we had developed techniques for measuring solvation over a range of temperatures, we applied these techniques to electron salvation in alcohols (Zhang and Jonah, 1996b). We found several dierences between solvation kinetics for the electron and the anions. In this paper, we will summarize these results and then explain the dierences and show how the dierences can be explained based on solvent structure. 2. Experimental All chemicals were purchased at the highest purity commercially available. Concentrations of the benzophenone were between 0.15 and 0.25 M. No signi®cant dierences were obtained between 0.05 and 0.25 M benzophenone. Such high concentrations were needed to ensure that all the electrons would react with the benzophenone before electron solvation. Experimental measurements were made using the Argonne stroboscopic pulse radiolysis system. The system makes use of the 30 ps pulse of the Argonne linac as a pump pulse to create the anion and uses Cerenkov radiation, which is generated from the same pulse, as an analyzing light. The time resolution is approximately 30 ps. Spectra and dynamics were
measured from 500±800 nm (Jonah, 1975). The data from 600±800 nm were used for data analysis. Below 600 nm, the ketyl radical has an absorption peak (Marignier and Hickel, 1982, 1984) and complicates the spectra and kinetics.
3. Results and discussion 3.1. Solvation energetics 3.1.1. Energetics of anion solvation Fig. 1 shows the shift of the spectrum of the benzophenone anion in 1-octanol and 2-octanol as a function of time. The spectra clearly show the shift from the red to the blue as a function of time. Fig. 1 also shows that the maximum of the spectrum is dierent for primary and secondary alcohols. The solvated benzophenone anion spectrum has a maximum at about 630 nm in all primary alcohols from propanol to decanol. In secondary alcohols the maximum is shifted only to 650 nm. In acetonitrile, the maximum is at 730 nm while in neat n-hexane the spectral maximum is at 800 nm. It has been assumed that the amount of the spectral shift is correlated with the solvation energy (van der Zwan and Hynes, 1985). Simple models of solvation have suggested that the solvation energy should depend on the dipole density (Lin and Jonah, 1994). In a real ¯uid the dipole density will depend on the concentration and the strength of the dipoles. Because the dominant dipole in an alcohol is the OH bond, the dipole density in alcohols will depend on the concentration of OH bonds. The density will thus be larger for ethanol and smaller for decanol. Because the dipole moment is larger for acetonitrile than for methanol, the dipole density in acetonitrile will be larger than in methanol. However, our experimental data are not consistent with a model where the spectral shift depends on dipole density. As Fig. 1 shows, the spectra of 1-octanol and 2-octanol have dierent maxima, although they have virtually identical dipole densities. Also, the spectrum of the anion in 1-propanol is the same as in 1-decanol while the dipole density in 1-propanol is more than a factor of three greater than in 1decanol. The results in acetonitrile are also not consistent with a continuum model of solvation. In acetonitrile, the maximum of the absorption of the solvated benzophenone anion is considerably red-shifted from the maxima in the alcohols, although acetonitrile has the largest dipole density of all the solvents studied. The solvent aects the spectra of the solvated electron in similar ways. While the spectra are not the same for the electron in all the primary alcohols, they
X. Zhang et al. / Radiation Physics and Chemistry 54 (1999) 433±440
435
Fig. 1. The spectra of the benzophenone anion in 1-octanol and 2-octanol at dierent times.
are similar and are considerably blue shifted from the spectra of the electron in secondary alcohols. All of these results are consistent with the idea that the number and closeness of the solvent molecules to the solute molecule dominate the solvation energy rather than the average dielectric properties of the solution. For example, there will be interference between
the alcohol molecules if the hydroxyl group is on a secondary carbon, making it more dicult to cluster the alcohols around the anion. This will mean either the alcohols are farther away from the anion or there will be fewer alcohols in the inner shell; in either case the ®eld will be weaker at the benzophenone anion and thus the solvation energy will be less. However, if the
436
X. Zhang et al. / Radiation Physics and Chemistry 54 (1999) 433±440
Fig. 2. Kinetics (top) and spectral shifts (bottom) for the benzophenone in 2-methyl-1-propanol at room temperatures.
branching in the hydrocarbon is further back from hydroxyl group, such as for the case of 2-methyl-1-propanol, the spectrum is similar to the spectrum seen in simple primary alcohols as shown in Fig. 2. The data for acetonitrile also support the role of local structure in solvation. The exposed end of acetonitrile is negative so that the molecule can not be radially aligned to the anion. If the acetonitrile is tangential, with the car-
bon of the nitrile group nearest the anion, the interaction will be maximized. The tangential con®guration will then limit the number of acetonitrile molecules near the benzophenone, which lowers the solvation energy. We have also explored theoretically the importance in the solvation energy of the position and orientation of a dipole in a solvent molecule. To do this, we have
X. Zhang et al. / Radiation Physics and Chemistry 54 (1999) 433±440
carried out two kinds of simulations, one with molecular dipoles and one with point charges (Lin and Jonah, 1995). These results clearly showed that the local ®elds at the anion solute would be strongest for the primary alcohol surrogate, less strong for the secondary alcohol surrogate and weakest for the acetonitrile surrogate. Because the perturbation of the energy levels of the ground and excited states would be expected to depend on the strength of the ®eld at the anion, the calculations con®rm the experimental results. Similarly, studies were done on the ®eld as a function the length of the solvent molecule. The strength of the ®elds was independent of the length of the solute molecule, if the molecule was a surrogate of a primary alcohol. This is despite the fact that the dipole density was smaller for the longer solvent molecules. Thus our calculations have shown that the general model proposed above does make sense. It has been suggested that the important interactions that lead to the shift in the spectrum arise from hydrogen bonding with benzophenone, as has been suggested in this discussion (Ichikawa et al., 1988) rather than the interactions between charges. While hydrogen bonding clearly takes place via charge interactions, hydrogen bonds have a particular spatial structure. Those suggestions are based on low temperature spin resonance experiments and theoretical simulations. While optical experiments are incapable of directly establishing the origin of the states, several experimental results suggest to us that solvent polarization is a very important component. We have found similar spectral shifts in solutions of salts in acetonitrile as well as spectral shifts in propylene carbonate (Lin and Jonah, to be published). The concentration dependence of the rates of the spectral shifts in acetonitrile±salt solutions is similar to what has been observed previously for spectral shifts for excited states (Chapman and Maroncelli, 1991; Bart and Huppert, 1992). In these systems no hydrogen bonding is possible. 3.1.2. Energetics of electron solvation A similar explanation, which makes use of the importance of solvent structure, can explain the change in solvated electron spectra in dierent alcohols. The secondary alcohols will not pack as well around the electron as will a primary electron and thus the solvated electron is not solvated as much in secondary alcohols as in primary alcohols. This then leads to the solvated electron being red shifted in a secondary alcohol as compared to the electron in a primary alcohol. The dierences in the electron spectra are much bigger between primary and secondary alcohols than is found for the anion. This is to be expected because the molecular framework supplies most of the localization energy for an electron in the benzophenone alcohol and the solvent just slightly alters it, while for the elec-
437
tron, the solvent structure provides all the binding energy. 3.2. Solvation kinetics There are two characteristics of solvation kinetics: the rate at which the spectra change from the unsolvated to the fully solvated form and the wavelength dependence of the shift. This information can be a starting point to understanding how solvation takes place. The absorption spectra of the benzophenone anion and the electron in alcohols both shift from the red to the blue, re¯ecting the increase in solvation of the species. However, the way the spectra shift at room temperature is quite dierent. At room temperature, the electron spectrum shifts from an absorption in the infrared to an absorption in the red with little evidence of intermediate absorbing species. This picture is not rigorously correct because there is no clear isosbestic point in many solutions and while the kinetics are not the same at all wavelengths, they are similar at the wavelength extremes (Kenney-Wallace and Jonah, 1982). The solvation of the benzophenone anion on the other had, suggests many intermediate states and the kinetics are very clearly dierent at dierent wavelengths. When the electron is solvated at low temperatures in alcohols, the spectra show a continuous shift (Baxendale and Wardman, 1977). Solvation times for electron solvation and benzophenone solvation have been determined. The electron solvation times were determined by ®tting the growth of the electron absorption at 600 nm (Zhang and Jonah, 1996a). The rates at 500 and 600 nm are the same (and are very similar to the rates measured at 1300 nm). The determination of the rates for benzophenone solvation is more complicated and the rates were determined by ®tting decays at dierent wavelengths and spectra at dierent times to a common solvation time (Zhang and Jonah, 1996b). A single exponential lifetime was assumed to describe the change in width of the absorption band and the shift of the center of the absorption band. The dynamic range of the present signals would not allow us to observe a double exponential process where one of the processes is small. Fig. 3 shows a comparison of electron solvation, anion solvation and dipole solvation. To make it possible to display solvation in a variety of solvents meaningfully, the solvation times were plotted versus the solvent relaxation times. This was done by plotting the data against tl, the longitudinal relaxation time. The longitudinal relaxation time is de®ned by e1 tl td 1 e0 where td is the Debye time (relaxation time) for the
438
X. Zhang et al. / Radiation Physics and Chemistry 54 (1999) 433±440
Fig. 3. The solvation time ts as a function the longitudinal relaxation time tl (see text). The lines were drawn to guide the eye. The data for dipole solvation were for C102 and the results and solvent properties were taken from the tables collected by Maroncelli (1993). The data for electron solvation were taken from Kenney-Wallace and Jonah (1982). The data for the anion solvation came from Lin and Jonah, 1992, 1993.
solvent and e1 and e0 are the dielectric constant at in®nite frequency and zero frequency (static dielectric constant). In practice what it does is quantify how fast the solvent would respond to a charge (or if slightly altered, for a dipole) in the solution assuming a continuum solvent. This allows us to plot meaningfully many dierent solutions on the same axis. This graph of solvation times at room temperatures provides several interesting insights. First, we see that the solvation of an anion and the solvation of the electron are similar to the electron occurring slightly faster. Second we see that the electron solvation and anion solvation appear to be considerably faster than the solvation of an excited state. The correspondence between the anion and the electron could be interpreted as `to be expected', if one assumes that it is simply the charge on the species that is important. However, the electron is a quantum particle that is de®ned solely by the solvent structure. The dierence in solvation kinetics between the solvation of dipoles and the solvation of the benzophenone is not yet understood. Experiments have also been done where an excited state of an ion was created and the solvation time of the ionic excited state was measured (Tominaga and Walker, 1995). These results showed similar solvation times to the solvation of neutrals, suggesting that it is the response to the dipolar ®eld as compared to the coulombic ®eld rather than the initial ®eld that exists that is important. Information about the microscopic motions that occur in solvation can be obtained from measurements of the activation energy for solvation. Lower temperatures can decrease the number of solvation motions that can take place. If the solvation activation energies can be correlated with activation energies for motions
such as rotation or translational diusion in the solvent, one might suggest that such motions are important in the solvation process. Of course, this assumes that the same molecular motions are occurring throughout the temperature range where the activation energy is measured. This does not always appear to be the case for electron solvation, where the entire character of the spectral shift will change at suciently low temperatures. For example, in ethanol at room temperature the spectrum shifts as we discussed above for our alcohol spectra; however at temperatures near ÿ808C, the spectra shift continuously as we saw for the spectra of the benzophenone anion (Gilles et al., 1977). The same behavior occurs in 1-propanol; however the change in behavior occurs between 110 and 120 K (Baxendale and Wardman, 1973). Fig. 4 shows that the initial and ®nal spectra of the benzophenone anion in n-propanol are independent of temperature. This same behavior was true for all the alcohols studied. As the temperature changed, the form of the kinetics remained the same (decay in the red, growth in the blue portion of the spectrum) but just were slower. No shift in the spectral behavior was observed for the electron solvation over the temperature range we studied. From these results we deduced that solvation of the benzophenone anion occurs by the same mechanism over the temperature range studied. Similar conclusions were drawn for the electron solvation process.
Fig. 4. The spectra of the benzophenone anion in 1-propanol (top) directly after the pulse and (bottom) 3 ns after the pulse.
X. Zhang et al. / Radiation Physics and Chemistry 54 (1999) 433±440
Fig. 5. The rates for the solvation of the electron and benzophenone anion in 1-propanol (top) and 2-propanol (bottom) as a function of temperature. The lines are Arrhenius ®ts to the data and the values given are the slopes of the lines.
The rates for electron solvation and benzophenone solvation as a function of temperature are shown in Fig. 5. There are several interesting facets to these results. We see that the activation energy for solvation is very clearly dierent for electron solvation and anion solvation in 1-propanol. In 2-propanol, the dierence is not as great, but it is still greater than the error limits (standard errors for ®tting linear leastsquare equations). In addition, the activation energy for electron solvation is lower than the activation energy for anion solvation in 1-propanol but larger in 2-propanol (Zhang and Jonah, 1996a). The activation energies for anion solvation are very well correlated with the hydrogen-bond energy in all the alcohols. This suggests that the important process in solvation is the breaking of the hydrogen bonds between solvent molecules to allow the solvent molecules to reorganize around the solute (Levin, 1954; Zhang and Jonah, 1996a). This result suggests that simulations that consider only a single solvent molecule and a single solute molecule are unlikely to be a good representation of the physical process in a liquid (Tachikawa, 1996). One can suggest several reasons for the dierences in activation energy for electron and anion solvation. The most important reason may involve the role of solvent structure in the formation of the species to be solvated. It seems likely that the electron will react
439
with any benzophenone molecule, independent of the solvent structure around it. Thus the benzophenone solvation will re¯ect the average of the solvent around the benzophenone. However, it is well known that electron solvation takes place through particular structures in the solvent (Kenney-Wallace and Jonah, 1982). Thus, the electron solvation will occur at specialized locations within the solvent. There has been considerable eort in studying how alcohols cluster in solution and how the molecular clusters are altered in dierent alcohol solutions (see references cited in Zhang and Jonah, 1996b). Many of these experiments have focused on the structures that occur in solutions of alcohols in various diluents. These studies have shown that alcohol clustering is greatly dierent in dierent alcohols. The present results are consistent with these results. In anion solvation, the alcohol molecules near the anion are clustered with the solvent molecules While the alcohol molecule can hydrogen bond with a benzophenone molecule, hydrogen bonding to other alcohol molecules is more favorable. Thus for the alcohol molecule to reorient towards an anion, it must break its hydrogen bond with the other solvent molecules before it can rotate. The activation energy for this process will be the hydrogen bond energy in the solvent. The electron will solvate in a potential minimum in the solution. If the structure in the alcohol where the electron solvates is similar to the structure around the fully solvated electron, it may not be necessary for hydrogen bonds to be broken. This is what appears to occur in primary alcohols, where the activation energy is less than the hydrogen bond energy in the alcohol. However, in the secondary alcohol, the activation energy for solvation is similar to the hydrogen bond energy, suggesting that the structure is not as similar in the secondary alcohols. Experimental measurements of alcohol clustering have suggested that cyclic structures are quite prevalent in primary alcohols but that linear chains are more prevalent in secondary alcohols. (see references cited in Zhang and Jonah, 1996b) To convert a linear chain to a structure that surrounds the electron requires the breaking of hydrogen bonds; however if the structure is cyclic, hydrogen bonds need not be broken if the electron solvates in the center of the alcohol. This is consistent with the model that we had suggested several years ago. Recently there has been a simulation of the dynamics of benzophenone anion solvation using as a model, one benzophenone molecule and one methanol molecule (Tachikawa, 1996). The discussion that has been given above suggests that because the model does not include information about the liquid, it is unlikely to explain solvation in liquids. As discussed above, the activation energy suggests that the kinetically important processes (at least on the time scale measured)
440
X. Zhang et al. / Radiation Physics and Chemistry 54 (1999) 433±440
include the breaking of solvent±solvent hydrogen bonds. These processes do not occur in the a single solvent molecule model. Simulations that we have carried out also show the importance of the initial solvent con®gurations and solvent±solvent interactions in the solvation of benzophenone. The simulation that we made uses much less sophisticated potentials but uses many solvent molecules. These two types of calculations provide dierent windows on the physical reality. The future will require a fusing of the techniques. 4. Summary In this report we have tried to summarize our research on solvation of anions and electrons in alcohols. The similarity of kinetics between the electron and the anion at room temperature appears to be accidental. Both the energetics and the temperature dependence of the kinetics depend on the microscopic solvent structure. This structure both determines the energetics, i.e. what the solvation energy is, and the activation energy. 5. Unlinked references Castner et al., 1987; Chapman et al., 1990; DecleÂmy and RullieÁre, 1988; DecleÂmy et al., 1987; Huppert et al., 1981; Maroncelli and Fleming, 1987; Nagarajan et al., 1987; Simon and Su, 1987; Su and Simon, 1987; Wang et al., 1981 Acknowledgements We gratefully acknowledge the assistance of Donald Ficht for linac operations. The work was performed under the auspices of the Oce of Basic Energy Sciences, Division of Chemical Science, US-DOE under contract number W-31-109-ENG-38. References Amos, A.T., Barrows, B.L., 1973 Adv. Quantum Chem. 7, 289. Barbara, P.F., Jarzeba, W., 1990. In: Volman, D.H., Hammond, G.S., Gollnick, K. (Eds.), Advances in Photochemistry, vol. 15. Wiley Interscience, New York, pp. 1±68. Bart, E., Huppert, D., 1992 Chem. Phys. Lett. 195, 37. Baxendale, J.H., Wardman, P., 1977 Can. J. Chem. 55, 1996. Baxendale, J.H., Wardman, P., 1973 J. Chem. Soc. Faraday Trans. I69, 584. Castner, E.W., Jr., Maroncelli, M., Fleming, G.R., 1987 J. Chem. Phys. 86, 1090.
Chapman, C.F., Maroncelli, M., 1991 J. Phys Chem. 95, 9095. Chapman, C.F., Fee, R.S., Maroncelli, M., 1990 J. Phys. Chem. 94, 4929. Chase, W.J., Hunt, J.W., 1975 J. Phys. Chem. 79, 2835. Davis, K.M.C., 1975. Solvent eects on charge±transfer complexes. In: Forster, R. (Ed.), Molecular Association, Vol. 1. Academic Press, London, New York. DecleÂmy, A., RullieÁre, C., 1988 Chem. Phys. Lett. 146, 1. DecleÂmy, A., RullieÁre, C., Kottis, Ph., 1987 Chem. Phys. Lett. 133, 448. Gilles, L., Bono, M.R., Schmidt, M., 1977 Can. J. Chem. 55, 2003. Huppert, D., Rand, S.D., Rentzepis, P.M., Barbara, P.F., Struve, W.S., Grabowski, Z.R., 1981 J. Chem. Phys. 75, 5714. Ichikawa, T., Ishikawa, Y., Yoshida, H., 1988 J. Phys. Chem. 92, 508. Jauquet, M., Laszlo, P., 1975. In¯uence of solvents on spectroscopy. In: Dack, M.R.J. (Ed.), Solutions and Solubilities, Vol. VIII, Part I of Weissberger, A. (Ed.), Techniques of Chemistry. Wiley-Interscience New York. Jonah, C.D., 1975 Rev. Sci. Instrum. 46, 62. Kenney-Wallace, G.A., Jonah, C.D., 1976 Chem. Phys. Lett. 39, 596. Kenney-Wallace, G.A., Jonah, C.D., 1982. J. Phys. Chem., 2572. Kevan, L., 1980 J. Phys. Chem. 84, 1232. Kevan, L., 1981 Acct. Chem. Res. 14, 138. Levin, B.Y., 1954 Zh. Fiz. Khim. 28, 1399. Lin, Y., Jonah, C.D. To be published. Lin, Y., Jonah, C.D., 1995 Chem. Phys. Lett. 233, 138. Lin, Y., Jonah, C.D., 1994. In: Simon, J.D. (Ed.), Ultrafast Dynamics of Chemical Systems. Kluwer Academic Publishers, The Netherlands, pp. 137±162. Lin, Y., Jonah, C.D., 1992 J. Phys. Chem. 96, 10119. Lin, Y., Jonah, C.D., 1993 J. Phys. Chem. 97, 295. Marignier, J.L., Hickel, B., 1982 Chem. Phys. Lett. 86, 95. Marignier, J.L., Hickel, B., 1984 J. Phys. Chem. 88, 5375. Maroncelli, M., Fleming, G.R., 1987 J. Chem. Phys. 86, 6221. Maroncelli, M., 1993 J. Mol. Liq. 57, 1. Migus, A., Gauduel, Y., Martin, J.L., Antonetti, A., 1987 Phys. Rev. Lett. 58, 1559. Nagarajan, V., Brearley, A.M., Kang, T.-J., Barbara, P.F., 1987 J. Chem. Phys. 86, 3183. Nicol, M.F., 1974 Appl. Spectrosc. Rev. 8, 183. Okazaki, K., Freeman, G.R., 1978 Can. J. Chem. 56, 2305. Reichardt, C., 1988. Solvents and Solvent Eects in Organic Chemistry. VCH, Weinheim, FRG. Simon, J.D., Su, S.G., 1987 J. Chem. Phys. 87, 7016. Su, S.G., Simon, J.D., 1987 J. Phys. Chem. 91, 2693. Tachikawa, H., 1996 J. Phys. Chem 100, 17090. Tominaga, K., Walker, G.C., 1995 J. Photochem. Photobio. A: 87, 127. van der Zwan, Hynes, J.T., 1985 J. Phys. Chem. 89, 4181. Wang, Y., McAulie, M., Novak, F., Eisenthal, K.B., 1981 J. Phys. Chem. 85, 3736. Zhang, X., Jonah, C.D., 1996a J. Phys. Chem. 100, 7042. Zhang, X., Jonah, C.D., 1996b Chem. Phys. Lett. 262, 649. Zhang, X., Jonah, C.D., 1995 Chem. Phys. Lett. 245, 421.
Radiation Physics and Chemistry Radiation Physics and Chemistry 54 (1999) 433±440
Reviews on Special Topics
Anion and electron solvation in alcoholsp Xujia Zhang *, Yi Lin, Charles D. Jonah Chemistry Division, Argonne National Laboratory, Argonne, IL 60439, USA Received for publication 19 November 1997
Abstract Pulse radiolytic techniques have been used to measure the solvation of anions and electrons in alcohols. The results have shown the importance of the microscopic liquid structure in the solvation process. The activation energy for benzophenone solvation is equal to the hydrogen-bond energy for the liquids, which shows that the solvent reorganization requires the breaking of the hydrogen bonds between solvent molecules. For electron solvation, primary alcohols have a lower activation energy because the initial hydrogen-bonded structure of the liquid is amenable to solvation. The solvation of an electron in a secondary alcohol requires require hydrogen-bond breakage. # 1999 Published by Elsevier Science Ltd. All rights reserved.
1. Introduction The solvent has long been known to play an important role in chemical reactions (Reichardt, 1988). There are many mechanisms by which the solvent can aect chemical reactivity. These include trapping of reactants, limiting mobility and removing or adding energy to the chemical system. In polar ¯uids, one of the most important ways that the solvent can alter reactions is through solvation. In solvation, the solvent responds to a solute molecule in a way that lowers the energy of the system. Electrostatic forces are one of the strongest forces that will cause solvent reorganization. The
* Corresponding author. The submitted manuscript has been created by the University of Chicago as Operator of Argonne National Laboratory (`Argonne') under Contract No. W-31-109-ENG38 with the US Department of Energy. The US Government retains for itself, and others acting on its behalf, a paid-up, nonexclusive, irrevocable worldwide license in said article to reproduce, prepare derivative works, distribute copies to the public and perform publicly and display publicly, by or on behalf of the Government. Work performed under the auspices of the Oce of Basic Energy Sciences, Division of Chemical Science, US-DOE under contract number W-31109-ENG-38. p
charge distribution on a solute molecule will interact with the charge distribution on a solvent molecule and these interactions will reorient the solvent molecules to create a lower energy system. In general, this process will raise the energy of the solvent while lowering the energy of the system. For example in alcohols, the solvent molecules will be oriented by an anion so that the H of the O±H group will point towards the anion. To do this, the hydrogen bond that the hydrogen had formed with other solvent molecules must be broken. It is this balancing between solvent±solute interactions and solvent±solvent interactions that determine the solvation energetics. Both the energetics and the kinetics of solvation are important in modifying chemical reactivity. The amount by which the solvation process alters the energy of reacting states can determine whether the reaction pathway and velocity will be altered by solvation while the kinetics will de®ne whether the solvation will be fast enough to in¯uence the reaction. Early experiments probed solvation energetics by measuring spectral shifts in dierent solvents (Amos and Barrows, 1973; Nicol, 1974; Davis, 1975; Jauquet and Laszlo, 1975). The shift in the absorption spectrum re¯ects a change in the dierence of the energy of the ground state and the excited state in dierent
0969-806X/99/$ - see front matter # 1999 Published by Elsevier Science Ltd. All rights reserved. PII: S 0 9 6 9 - 8 0 6 X ( 9 7 ) 0 0 3 1 5 - 0
434
X. Zhang et al. / Radiation Physics and Chemistry 54 (1999) 433±440
solvents. These energy changes re¯ect the dierences in solvation. Early kinetic measurements of solvation include the measurement of electron solvation (Baxendale and Wardman, 1973, 1977; Chase and Hunt, 1975; Kenney-Wallace and Jonah, 1976, 1982; Gilles et al., 1977; Okazaki and Freeman, 1978; Kevan, 1980, 1981; Migus et al., 1987) and the measurements in the shift of an emission spectrum as a function of time (see recent reviews of Barbara and Jarzeba, 1990; Maroncelli, 1993). The studies of emission spectra determined the solvation of an excited state. Experiments have also been done where the solvation of an ionic excited state was measured (Tominaga and Walker, 1995). That experiment measured the solvation of the excited state of an ion. The solvent was already arranged around the ion and the perturbation of the solvent was done by the excited states rather than by the creation of charge. Pulse radiolytic techniques were applied to the measurement of anion solvation by Marignier and Hickel, who measured the solvation of the benzophenone anion in low temperature alcohols (Marignier and Hickel, 1982, 1984). Building on their results, we carried out experiments using the Argonne stroboscopic pulse radiolysis system and were able to make measurements for a wide variety of alcohols at room temperature and at temperatures to about ÿ608C (Lin and Jonah, 1992, 1993, 1994; Zhang and Jonah, 1995). Because we had developed techniques for measuring solvation over a range of temperatures, we applied these techniques to electron salvation in alcohols (Zhang and Jonah, 1996b). We found several dierences between solvation kinetics for the electron and the anions. In this paper, we will summarize these results and then explain the dierences and show how the dierences can be explained based on solvent structure. 2. Experimental All chemicals were purchased at the highest purity commercially available. Concentrations of the benzophenone were between 0.15 and 0.25 M. No signi®cant dierences were obtained between 0.05 and 0.25 M benzophenone. Such high concentrations were needed to ensure that all the electrons would react with the benzophenone before electron solvation. Experimental measurements were made using the Argonne stroboscopic pulse radiolysis system. The system makes use of the 30 ps pulse of the Argonne linac as a pump pulse to create the anion and uses Cerenkov radiation, which is generated from the same pulse, as an analyzing light. The time resolution is approximately 30 ps. Spectra and dynamics were
measured from 500±800 nm (Jonah, 1975). The data from 600±800 nm were used for data analysis. Below 600 nm, the ketyl radical has an absorption peak (Marignier and Hickel, 1982, 1984) and complicates the spectra and kinetics.
3. Results and discussion 3.1. Solvation energetics 3.1.1. Energetics of anion solvation Fig. 1 shows the shift of the spectrum of the benzophenone anion in 1-octanol and 2-octanol as a function of time. The spectra clearly show the shift from the red to the blue as a function of time. Fig. 1 also shows that the maximum of the spectrum is dierent for primary and secondary alcohols. The solvated benzophenone anion spectrum has a maximum at about 630 nm in all primary alcohols from propanol to decanol. In secondary alcohols the maximum is shifted only to 650 nm. In acetonitrile, the maximum is at 730 nm while in neat n-hexane the spectral maximum is at 800 nm. It has been assumed that the amount of the spectral shift is correlated with the solvation energy (van der Zwan and Hynes, 1985). Simple models of solvation have suggested that the solvation energy should depend on the dipole density (Lin and Jonah, 1994). In a real ¯uid the dipole density will depend on the concentration and the strength of the dipoles. Because the dominant dipole in an alcohol is the OH bond, the dipole density in alcohols will depend on the concentration of OH bonds. The density will thus be larger for ethanol and smaller for decanol. Because the dipole moment is larger for acetonitrile than for methanol, the dipole density in acetonitrile will be larger than in methanol. However, our experimental data are not consistent with a model where the spectral shift depends on dipole density. As Fig. 1 shows, the spectra of 1-octanol and 2-octanol have dierent maxima, although they have virtually identical dipole densities. Also, the spectrum of the anion in 1-propanol is the same as in 1-decanol while the dipole density in 1-propanol is more than a factor of three greater than in 1decanol. The results in acetonitrile are also not consistent with a continuum model of solvation. In acetonitrile, the maximum of the absorption of the solvated benzophenone anion is considerably red-shifted from the maxima in the alcohols, although acetonitrile has the largest dipole density of all the solvents studied. The solvent aects the spectra of the solvated electron in similar ways. While the spectra are not the same for the electron in all the primary alcohols, they
X. Zhang et al. / Radiation Physics and Chemistry 54 (1999) 433±440
435
Fig. 1. The spectra of the benzophenone anion in 1-octanol and 2-octanol at dierent times.
are similar and are considerably blue shifted from the spectra of the electron in secondary alcohols. All of these results are consistent with the idea that the number and closeness of the solvent molecules to the solute molecule dominate the solvation energy rather than the average dielectric properties of the solution. For example, there will be interference between
the alcohol molecules if the hydroxyl group is on a secondary carbon, making it more dicult to cluster the alcohols around the anion. This will mean either the alcohols are farther away from the anion or there will be fewer alcohols in the inner shell; in either case the ®eld will be weaker at the benzophenone anion and thus the solvation energy will be less. However, if the
436
X. Zhang et al. / Radiation Physics and Chemistry 54 (1999) 433±440
Fig. 2. Kinetics (top) and spectral shifts (bottom) for the benzophenone in 2-methyl-1-propanol at room temperatures.
branching in the hydrocarbon is further back from hydroxyl group, such as for the case of 2-methyl-1-propanol, the spectrum is similar to the spectrum seen in simple primary alcohols as shown in Fig. 2. The data for acetonitrile also support the role of local structure in solvation. The exposed end of acetonitrile is negative so that the molecule can not be radially aligned to the anion. If the acetonitrile is tangential, with the car-
bon of the nitrile group nearest the anion, the interaction will be maximized. The tangential con®guration will then limit the number of acetonitrile molecules near the benzophenone, which lowers the solvation energy. We have also explored theoretically the importance in the solvation energy of the position and orientation of a dipole in a solvent molecule. To do this, we have
X. Zhang et al. / Radiation Physics and Chemistry 54 (1999) 433±440
carried out two kinds of simulations, one with molecular dipoles and one with point charges (Lin and Jonah, 1995). These results clearly showed that the local ®elds at the anion solute would be strongest for the primary alcohol surrogate, less strong for the secondary alcohol surrogate and weakest for the acetonitrile surrogate. Because the perturbation of the energy levels of the ground and excited states would be expected to depend on the strength of the ®eld at the anion, the calculations con®rm the experimental results. Similarly, studies were done on the ®eld as a function the length of the solvent molecule. The strength of the ®elds was independent of the length of the solute molecule, if the molecule was a surrogate of a primary alcohol. This is despite the fact that the dipole density was smaller for the longer solvent molecules. Thus our calculations have shown that the general model proposed above does make sense. It has been suggested that the important interactions that lead to the shift in the spectrum arise from hydrogen bonding with benzophenone, as has been suggested in this discussion (Ichikawa et al., 1988) rather than the interactions between charges. While hydrogen bonding clearly takes place via charge interactions, hydrogen bonds have a particular spatial structure. Those suggestions are based on low temperature spin resonance experiments and theoretical simulations. While optical experiments are incapable of directly establishing the origin of the states, several experimental results suggest to us that solvent polarization is a very important component. We have found similar spectral shifts in solutions of salts in acetonitrile as well as spectral shifts in propylene carbonate (Lin and Jonah, to be published). The concentration dependence of the rates of the spectral shifts in acetonitrile±salt solutions is similar to what has been observed previously for spectral shifts for excited states (Chapman and Maroncelli, 1991; Bart and Huppert, 1992). In these systems no hydrogen bonding is possible. 3.1.2. Energetics of electron solvation A similar explanation, which makes use of the importance of solvent structure, can explain the change in solvated electron spectra in dierent alcohols. The secondary alcohols will not pack as well around the electron as will a primary electron and thus the solvated electron is not solvated as much in secondary alcohols as in primary alcohols. This then leads to the solvated electron being red shifted in a secondary alcohol as compared to the electron in a primary alcohol. The dierences in the electron spectra are much bigger between primary and secondary alcohols than is found for the anion. This is to be expected because the molecular framework supplies most of the localization energy for an electron in the benzophenone alcohol and the solvent just slightly alters it, while for the elec-
437
tron, the solvent structure provides all the binding energy. 3.2. Solvation kinetics There are two characteristics of solvation kinetics: the rate at which the spectra change from the unsolvated to the fully solvated form and the wavelength dependence of the shift. This information can be a starting point to understanding how solvation takes place. The absorption spectra of the benzophenone anion and the electron in alcohols both shift from the red to the blue, re¯ecting the increase in solvation of the species. However, the way the spectra shift at room temperature is quite dierent. At room temperature, the electron spectrum shifts from an absorption in the infrared to an absorption in the red with little evidence of intermediate absorbing species. This picture is not rigorously correct because there is no clear isosbestic point in many solutions and while the kinetics are not the same at all wavelengths, they are similar at the wavelength extremes (Kenney-Wallace and Jonah, 1982). The solvation of the benzophenone anion on the other had, suggests many intermediate states and the kinetics are very clearly dierent at dierent wavelengths. When the electron is solvated at low temperatures in alcohols, the spectra show a continuous shift (Baxendale and Wardman, 1977). Solvation times for electron solvation and benzophenone solvation have been determined. The electron solvation times were determined by ®tting the growth of the electron absorption at 600 nm (Zhang and Jonah, 1996a). The rates at 500 and 600 nm are the same (and are very similar to the rates measured at 1300 nm). The determination of the rates for benzophenone solvation is more complicated and the rates were determined by ®tting decays at dierent wavelengths and spectra at dierent times to a common solvation time (Zhang and Jonah, 1996b). A single exponential lifetime was assumed to describe the change in width of the absorption band and the shift of the center of the absorption band. The dynamic range of the present signals would not allow us to observe a double exponential process where one of the processes is small. Fig. 3 shows a comparison of electron solvation, anion solvation and dipole solvation. To make it possible to display solvation in a variety of solvents meaningfully, the solvation times were plotted versus the solvent relaxation times. This was done by plotting the data against tl, the longitudinal relaxation time. The longitudinal relaxation time is de®ned by e1 tl td 1 e0 where td is the Debye time (relaxation time) for the
438
X. Zhang et al. / Radiation Physics and Chemistry 54 (1999) 433±440
Fig. 3. The solvation time ts as a function the longitudinal relaxation time tl (see text). The lines were drawn to guide the eye. The data for dipole solvation were for C102 and the results and solvent properties were taken from the tables collected by Maroncelli (1993). The data for electron solvation were taken from Kenney-Wallace and Jonah (1982). The data for the anion solvation came from Lin and Jonah, 1992, 1993.
solvent and e1 and e0 are the dielectric constant at in®nite frequency and zero frequency (static dielectric constant). In practice what it does is quantify how fast the solvent would respond to a charge (or if slightly altered, for a dipole) in the solution assuming a continuum solvent. This allows us to plot meaningfully many dierent solutions on the same axis. This graph of solvation times at room temperatures provides several interesting insights. First, we see that the solvation of an anion and the solvation of the electron are similar to the electron occurring slightly faster. Second we see that the electron solvation and anion solvation appear to be considerably faster than the solvation of an excited state. The correspondence between the anion and the electron could be interpreted as `to be expected', if one assumes that it is simply the charge on the species that is important. However, the electron is a quantum particle that is de®ned solely by the solvent structure. The dierence in solvation kinetics between the solvation of dipoles and the solvation of the benzophenone is not yet understood. Experiments have also been done where an excited state of an ion was created and the solvation time of the ionic excited state was measured (Tominaga and Walker, 1995). These results showed similar solvation times to the solvation of neutrals, suggesting that it is the response to the dipolar ®eld as compared to the coulombic ®eld rather than the initial ®eld that exists that is important. Information about the microscopic motions that occur in solvation can be obtained from measurements of the activation energy for solvation. Lower temperatures can decrease the number of solvation motions that can take place. If the solvation activation energies can be correlated with activation energies for motions
such as rotation or translational diusion in the solvent, one might suggest that such motions are important in the solvation process. Of course, this assumes that the same molecular motions are occurring throughout the temperature range where the activation energy is measured. This does not always appear to be the case for electron solvation, where the entire character of the spectral shift will change at suciently low temperatures. For example, in ethanol at room temperature the spectrum shifts as we discussed above for our alcohol spectra; however at temperatures near ÿ808C, the spectra shift continuously as we saw for the spectra of the benzophenone anion (Gilles et al., 1977). The same behavior occurs in 1-propanol; however the change in behavior occurs between 110 and 120 K (Baxendale and Wardman, 1973). Fig. 4 shows that the initial and ®nal spectra of the benzophenone anion in n-propanol are independent of temperature. This same behavior was true for all the alcohols studied. As the temperature changed, the form of the kinetics remained the same (decay in the red, growth in the blue portion of the spectrum) but just were slower. No shift in the spectral behavior was observed for the electron solvation over the temperature range we studied. From these results we deduced that solvation of the benzophenone anion occurs by the same mechanism over the temperature range studied. Similar conclusions were drawn for the electron solvation process.
Fig. 4. The spectra of the benzophenone anion in 1-propanol (top) directly after the pulse and (bottom) 3 ns after the pulse.
X. Zhang et al. / Radiation Physics and Chemistry 54 (1999) 433±440
Fig. 5. The rates for the solvation of the electron and benzophenone anion in 1-propanol (top) and 2-propanol (bottom) as a function of temperature. The lines are Arrhenius ®ts to the data and the values given are the slopes of the lines.
The rates for electron solvation and benzophenone solvation as a function of temperature are shown in Fig. 5. There are several interesting facets to these results. We see that the activation energy for solvation is very clearly dierent for electron solvation and anion solvation in 1-propanol. In 2-propanol, the dierence is not as great, but it is still greater than the error limits (standard errors for ®tting linear leastsquare equations). In addition, the activation energy for electron solvation is lower than the activation energy for anion solvation in 1-propanol but larger in 2-propanol (Zhang and Jonah, 1996a). The activation energies for anion solvation are very well correlated with the hydrogen-bond energy in all the alcohols. This suggests that the important process in solvation is the breaking of the hydrogen bonds between solvent molecules to allow the solvent molecules to reorganize around the solute (Levin, 1954; Zhang and Jonah, 1996a). This result suggests that simulations that consider only a single solvent molecule and a single solute molecule are unlikely to be a good representation of the physical process in a liquid (Tachikawa, 1996). One can suggest several reasons for the dierences in activation energy for electron and anion solvation. The most important reason may involve the role of solvent structure in the formation of the species to be solvated. It seems likely that the electron will react
439
with any benzophenone molecule, independent of the solvent structure around it. Thus the benzophenone solvation will re¯ect the average of the solvent around the benzophenone. However, it is well known that electron solvation takes place through particular structures in the solvent (Kenney-Wallace and Jonah, 1982). Thus, the electron solvation will occur at specialized locations within the solvent. There has been considerable eort in studying how alcohols cluster in solution and how the molecular clusters are altered in dierent alcohol solutions (see references cited in Zhang and Jonah, 1996b). Many of these experiments have focused on the structures that occur in solutions of alcohols in various diluents. These studies have shown that alcohol clustering is greatly dierent in dierent alcohols. The present results are consistent with these results. In anion solvation, the alcohol molecules near the anion are clustered with the solvent molecules While the alcohol molecule can hydrogen bond with a benzophenone molecule, hydrogen bonding to other alcohol molecules is more favorable. Thus for the alcohol molecule to reorient towards an anion, it must break its hydrogen bond with the other solvent molecules before it can rotate. The activation energy for this process will be the hydrogen bond energy in the solvent. The electron will solvate in a potential minimum in the solution. If the structure in the alcohol where the electron solvates is similar to the structure around the fully solvated electron, it may not be necessary for hydrogen bonds to be broken. This is what appears to occur in primary alcohols, where the activation energy is less than the hydrogen bond energy in the alcohol. However, in the secondary alcohol, the activation energy for solvation is similar to the hydrogen bond energy, suggesting that the structure is not as similar in the secondary alcohols. Experimental measurements of alcohol clustering have suggested that cyclic structures are quite prevalent in primary alcohols but that linear chains are more prevalent in secondary alcohols. (see references cited in Zhang and Jonah, 1996b) To convert a linear chain to a structure that surrounds the electron requires the breaking of hydrogen bonds; however if the structure is cyclic, hydrogen bonds need not be broken if the electron solvates in the center of the alcohol. This is consistent with the model that we had suggested several years ago. Recently there has been a simulation of the dynamics of benzophenone anion solvation using as a model, one benzophenone molecule and one methanol molecule (Tachikawa, 1996). The discussion that has been given above suggests that because the model does not include information about the liquid, it is unlikely to explain solvation in liquids. As discussed above, the activation energy suggests that the kinetically important processes (at least on the time scale measured)
440
X. Zhang et al. / Radiation Physics and Chemistry 54 (1999) 433±440
include the breaking of solvent±solvent hydrogen bonds. These processes do not occur in the a single solvent molecule model. Simulations that we have carried out also show the importance of the initial solvent con®gurations and solvent±solvent interactions in the solvation of benzophenone. The simulation that we made uses much less sophisticated potentials but uses many solvent molecules. These two types of calculations provide dierent windows on the physical reality. The future will require a fusing of the techniques. 4. Summary In this report we have tried to summarize our research on solvation of anions and electrons in alcohols. The similarity of kinetics between the electron and the anion at room temperature appears to be accidental. Both the energetics and the temperature dependence of the kinetics depend on the microscopic solvent structure. This structure both determines the energetics, i.e. what the solvation energy is, and the activation energy. 5. Unlinked references Castner et al., 1987; Chapman et al., 1990; DecleÂmy and RullieÁre, 1988; DecleÂmy et al., 1987; Huppert et al., 1981; Maroncelli and Fleming, 1987; Nagarajan et al., 1987; Simon and Su, 1987; Su and Simon, 1987; Wang et al., 1981 Acknowledgements We gratefully acknowledge the assistance of Donald Ficht for linac operations. The work was performed under the auspices of the Oce of Basic Energy Sciences, Division of Chemical Science, US-DOE under contract number W-31-109-ENG-38. References Amos, A.T., Barrows, B.L., 1973 Adv. Quantum Chem. 7, 289. Barbara, P.F., Jarzeba, W., 1990. In: Volman, D.H., Hammond, G.S., Gollnick, K. (Eds.), Advances in Photochemistry, vol. 15. Wiley Interscience, New York, pp. 1±68. Bart, E., Huppert, D., 1992 Chem. Phys. Lett. 195, 37. Baxendale, J.H., Wardman, P., 1977 Can. J. Chem. 55, 1996. Baxendale, J.H., Wardman, P., 1973 J. Chem. Soc. Faraday Trans. I69, 584. Castner, E.W., Jr., Maroncelli, M., Fleming, G.R., 1987 J. Chem. Phys. 86, 1090.
Chapman, C.F., Maroncelli, M., 1991 J. Phys Chem. 95, 9095. Chapman, C.F., Fee, R.S., Maroncelli, M., 1990 J. Phys. Chem. 94, 4929. Chase, W.J., Hunt, J.W., 1975 J. Phys. Chem. 79, 2835. Davis, K.M.C., 1975. Solvent eects on charge±transfer complexes. In: Forster, R. (Ed.), Molecular Association, Vol. 1. Academic Press, London, New York. DecleÂmy, A., RullieÁre, C., 1988 Chem. Phys. Lett. 146, 1. DecleÂmy, A., RullieÁre, C., Kottis, Ph., 1987 Chem. Phys. Lett. 133, 448. Gilles, L., Bono, M.R., Schmidt, M., 1977 Can. J. Chem. 55, 2003. Huppert, D., Rand, S.D., Rentzepis, P.M., Barbara, P.F., Struve, W.S., Grabowski, Z.R., 1981 J. Chem. Phys. 75, 5714. Ichikawa, T., Ishikawa, Y., Yoshida, H., 1988 J. Phys. Chem. 92, 508. Jauquet, M., Laszlo, P., 1975. In¯uence of solvents on spectroscopy. In: Dack, M.R.J. (Ed.), Solutions and Solubilities, Vol. VIII, Part I of Weissberger, A. (Ed.), Techniques of Chemistry. Wiley-Interscience New York. Jonah, C.D., 1975 Rev. Sci. Instrum. 46, 62. Kenney-Wallace, G.A., Jonah, C.D., 1976 Chem. Phys. Lett. 39, 596. Kenney-Wallace, G.A., Jonah, C.D., 1982. J. Phys. Chem., 2572. Kevan, L., 1980 J. Phys. Chem. 84, 1232. Kevan, L., 1981 Acct. Chem. Res. 14, 138. Levin, B.Y., 1954 Zh. Fiz. Khim. 28, 1399. Lin, Y., Jonah, C.D. To be published. Lin, Y., Jonah, C.D., 1995 Chem. Phys. Lett. 233, 138. Lin, Y., Jonah, C.D., 1994. In: Simon, J.D. (Ed.), Ultrafast Dynamics of Chemical Systems. Kluwer Academic Publishers, The Netherlands, pp. 137±162. Lin, Y., Jonah, C.D., 1992 J. Phys. Chem. 96, 10119. Lin, Y., Jonah, C.D., 1993 J. Phys. Chem. 97, 295. Marignier, J.L., Hickel, B., 1982 Chem. Phys. Lett. 86, 95. Marignier, J.L., Hickel, B., 1984 J. Phys. Chem. 88, 5375. Maroncelli, M., Fleming, G.R., 1987 J. Chem. Phys. 86, 6221. Maroncelli, M., 1993 J. Mol. Liq. 57, 1. Migus, A., Gauduel, Y., Martin, J.L., Antonetti, A., 1987 Phys. Rev. Lett. 58, 1559. Nagarajan, V., Brearley, A.M., Kang, T.-J., Barbara, P.F., 1987 J. Chem. Phys. 86, 3183. Nicol, M.F., 1974 Appl. Spectrosc. Rev. 8, 183. Okazaki, K., Freeman, G.R., 1978 Can. J. Chem. 56, 2305. Reichardt, C., 1988. Solvents and Solvent Eects in Organic Chemistry. VCH, Weinheim, FRG. Simon, J.D., Su, S.G., 1987 J. Chem. Phys. 87, 7016. Su, S.G., Simon, J.D., 1987 J. Phys. Chem. 91, 2693. Tachikawa, H., 1996 J. Phys. Chem 100, 17090. Tominaga, K., Walker, G.C., 1995 J. Photochem. Photobio. A: 87, 127. van der Zwan, Hynes, J.T., 1985 J. Phys. Chem. 89, 4181. Wang, Y., McAulie, M., Novak, F., Eisenthal, K.B., 1981 J. Phys. Chem. 85, 3736. Zhang, X., Jonah, C.D., 1996a J. Phys. Chem. 100, 7042. Zhang, X., Jonah, C.D., 1996b Chem. Phys. Lett. 262, 649. Zhang, X., Jonah, C.D., 1995 Chem. Phys. Lett. 245, 421.