8890
J. Phys. Chem. 1996, 100, 8890-8894
Volume Changes Associated with Intramolecular Exciplex Formation in a Semiflexible Donor-Bridge-Acceptor Compound Bas Wegewijs,† Jan W. Verhoeven,‡ and Silvia E. Braslavsky*,† Max-Planck-Institut fu¨ r Strahlenchemie, Postfach 10 13 65, D-45413 Mu¨ lheim an der Ruhr, Germany, and Laboratory of Organic Chemistry, UniVersity of Amsterdam, Nieuwe Achtergracht 129, 1018 WS Amsterdam, The Netherlands ReceiVed: December 12, 1995; In Final Form: March 1, 1996X
The photophysical processes taking place in the excited state of a donor-bridge-acceptor compound (1) in alkane solvents were studied with time-resolved laser-induced optoacoustic spectroscopy. 1 contains an aniline/ cyanonaphthalene D/A pair, separated by a semiflexible saturated hydrocarbon bridge. Excitation at 308 nm leads to efficient long-range charge separation, followed by rapid Coulomb-induced intramolecular exciplex formation. By monitoring the pressure waves generated by the decay of the excited species, three consecutive heat release processes could be discerned. In order to separate the contributions to the observed acoustic signals of structural volume changes (∆Vstr) and enthalpy changes, experiments were carried out in a series of normal alkanes, differing in their photothermal properties. A value of ∆Vstr ) -40 ( 5 mL/mol was obtained for the difference in reaction volume between the ground state and the exciplex state of 1. This large contraction should be attributed in part to electrostriction of the solvent around the dipolar species, according to classical electrostatic theory. Furthermore, there seems to be an additional contraction due to the internal volume change of 1, i.e., the conformational change involved in the intramolecular exciplex formation.
1. Introduction Many aspects of photoinduced electron transfer have been studied in model systems of the type donor-bridge-acceptor (D-br-A).1 One of the most challenging issues is that of the solvent reorganization around the developing and decaying charge centers, which plays a key role in Marcus theory.2 At present, there is not much experimental information regarding the actual movement of solvent molecules in response to large changes in dipole moment. In this respect it is of interest to investigate reaction volume changes accompanying charge separation processes, since these include local density fluctuations in the solvent, i.e., the transfer of solvent molecules between the bulk of the solvent and the solvation shells of the solute, in addition to possible intrinsic molecular volume changes. Traditionally, reaction and activation volumes have been studied with high-pressure techniques.3 However, several solvent parameters that might play a role in the electron transfer reaction are affected by the increase of the pressure, e.g., the dielectric constant, refractive index, viscosity, and density. Moreover, the properties of the D-A molecule itself can also be influenced by the elevated pressure, such as the electronic coupling between D and A.4 An alternative way to study volume changes involves laserinduced optoacoustic spectroscopy (LIOAS5), which does not interfere with the photophysical processes taking place in the excited molecules. By measuring the pressure waves generated by the relaxation of the absorbing species, information can be obtained on both enthalpy and volume changes of the system under study.6 Furthermore, consecutive decay processes with sufficiently distinct time constants can be monitored separately, depending on the time resolution of the experimental setup. * To whom correspondence should be addressed. † Max-Planck-Institut fu ¨ r Strahlenchemie. ‡ University of Amsterdam. X Abstract published in AdVance ACS Abstracts, May 1, 1996.
S0022-3654(95)03683-5 CCC: $12.00
Figure 1. Donor-bridge-acceptor compound 1.
Recently, this technique was applied in our laboratory to study intermolecular electron transfer in bimolecular excited state quenching reactions,8 as well as intramolecular metal-to-ligand charge transfer (MLCT) in a series of ruthenium-cyanobipyridine complexes.9 From these studies, performed in aqueous solution, it was concluded that the observed volume changes were mainly due to reorganization of the solvent around the transient species, having different properties than the parent compound. Interestingly, it appeared that there was no correlation between the charge (or dipole moment) of the species and the observed expansion or contraction, and it was concluded that specific solute-solvent interactions between ligands and the surrounding water molecules were responsible for this behavior.9 Obviously, water is a rather special case among solvents, with its large dipole moment and ability to form hydrogen bonds, resulting in a highly organized structure. For comparison, we decided to investigate the effect of intramolecular charge separation in D-br-A systems in nonpolar, aprotic solvents, in order to analyze how such a system, lacking in specific interactions, reacts to the sudden creation of a large dipole moment. For this purpose compound 1 was selected (see Figure 1), since it has a relatively long-lived exciplex state in alkanes, enabling us to distinguish the signal generated by the decay of the exciplex state from that of the very fast relaxation processes of the initially excited S1 state. Another useful feature of 1 is that the exciplex is formed with virtually unity quantum yield at room temperature. The remarkable photophysical processes taking place in 1 have been studied in solution10 as well as in © 1996 American Chemical Society
Exciplex Formation in a Donor-Bridge-Acceptor Compound the gas phase (in a supersonic expansion11) and can be described in terms of a “harpooning” mechanism: the strong anilino/ cyanonaphthalene D/A couple allows for fast long-range charge separation across five σ-bonds in the extended conformation, irrespective of the dielectric properties of the surrounding medium; in nonpolar media this is followed by electrostatically induced folding of the semiflexible bridge to form an exciplex configuration, where the charge separation distance is strongly decreased. In order to interpret the acoustic signals, the thermal and structural volume changes have to be separated by varying the thermoelastic properties of the solvent (the thermal expansivity, β/cpF, where β is the cubic expansion coefficient, cp the heat capacity, and F the mass density), without significantly altering the thermodynamics and the structural volume changes of the system.5 Although this is most conveniently achieved in aqueous solutions by simply measuring the signals as a function of temperature (taking advantage of the strong variation of β), an alternative method can be employed using a homologous series of solvents, e.g., n-alkanes.12 The thermal expansivity of the solvent decreases by a factor of 2 in the range from n-pentane to n-hexadecane, while the energy of the exciplex state remains virtually constant, as inferred from the exciplex emission band maximum of 1. 2. Experimental Section 2.1. Materials. The synthesis of 1 and 2 has been published previously.13 The calorimetric reference compound 2-hydroxybenzophenone was obtained from Merck and recrystallized from ethanol. The alkane solvents (n-pentane, n-hexane, n-heptane, n-octane, n-nonane, n-decane, n-dodecane, and n-hexadecane; Merck) were purified before use in various ways, by distillation, column chromatography, or preparative gas chromatography. 2.2. Methods. Our LIOAS setup consists of an FL2000 Lamba Physik-EMG 101 MSC excimer laser (XeCl, 308 nm) operating at 1-2 Hz, irradiating a conventional 1 cm absorption cuvette equipped with a PVF2 film (24 or 40 µm thickness) as a piezoelectric detector. The signals were amplified 100 times (Comlinear E103) and fed into a transient recorder (Tektronix TDS 520A or TDS 684A, operating at 0.5 or 1 gigasample/s and averaging 200-400 signals). The absorbance of the reference solution at 308 nm (recorded with a Shimadzu UV-2102PC spectrophotometer) was always matched within 2% to that of the sample solution (measured before and after the LIOAS experiment), and the mean energy of the incident laser light (measured with a Laser Precision Corp. RJP735 head connected to a RJ7100 meter) was also kept similar (and below 10 µJ) for the sample and reference. In this way, the spatial distribution of the laser beam in the cuvette was similar, while small differences in laser intensity were corrected for during the data handling. By placing a 200 µm slit in front of the cuvette, the full width at half-maximum of the first wave of the reference signal amounts to ca. 120 ns. This allowed for a time resolution of 10-15 ns using deconvolution techniques, provided that the signals have sufficient quality. In this respect, an advantage of working in organic solvents (as compared to water) is the larger value of the expansion coefficient β, leading to much better signal-to-noise ratios. In order to avoid differences in arrival time due to the temperature dependence of the sound velocity, the cuvette was thermostated at 20.0 ( 0.1 °C. The solutions of 1 were deoxygenated by Ar- or N2-bubbling for 15-20 min. However, it appeared that this was not sufficient to remove the oxygen rigorously. By comparing the fluorescence lifetime under these degassing conditions (τ ≈ 40
J. Phys. Chem., Vol. 100, No. 21, 1996 8891
Figure 2. LIOAS signals for the reference (2-hydroxybenzophenone) and sample (1) in n-pentane at 20.0 °C following excitation at 308 nm, together with the fit and residuals after deconvolution. Note that the fitted curve completely coincides with the measured signal for 1.
ns) with that obtained with an evacuated sample (three freezepump-thaw cycles, τ ) 75 ns), it became clear that quenching of the exciplex state was still rather efficient. Without degassing, this lifetime even drops to ca. 15-20 ns, which can be explained by taking into account a quenching rate constant of about 3 × 1010 M-1 s-1 and an oxygen concentration of ca. 1.4 × 10-3 M. Thus, the heat release of the exciplex decay was strongly influenced by the complicated photophysics of oxygen quenching, obscuring the information that might be obtained from its amplitude. Unfortunately, it is difficult to work under evacuated conditions with LIOAS, due to the requirement that the sample and reference have to be measured in the same cuvette, without disturbing the alignment of the laser beam, slit, and cuvette. Some tests with a specially designed vacuum cuvette yielded indeed longer lifetimes (>60 ns) for the exciplex decay, but also rather large differences in arrival time caused by unavoidable disalignment of the system. 3. Results and Discussion 3.1. Results of the LIOAS Measurements. Figure 2 shows the LIOAS signal obtained for 1 in n-pentane at 20 °C, together with that of the calorimetric reference 2-hydroxybenzophenone (2HOBP). The signal of 1 is shifted compared to that of 2HOBP, and it is also broader, which is an indication of the presence of a transient species with a liftime on the order of the experimental time window. In such a case, the signals have to be analyzed using a deconvolution procedure, in order to obtain the amplitudes and lifetimes of the subsequent decay processes. The procedure of time-resolved LIOAS has been described previously.5,7,9 In short, the acoustic signal of the sample can be regarded as a convolution of the instrument response function (which is given by a calorimetric reference solution that releases all of the excitation energy promptly as heat) and a time-dependent multiexponential decay curve describing the pressure evolution in the sample after excitation. The reference signal depends only on the heat release multiplied by the thermal expansivity of the solvent, while the sample signal might have an extra contribution from structural volume changes (∆Vstr), which does not depend on the expansivity. Thus, it can easily be derived that the recovered sample amplitudes φi, normalized with respect to the amplitude of the reference compound (φ ) 1), are given by
φi )
( )
qi ∆Vstr,i cpF + Eλ Eλ β
(1)
where Eλ is the excitation energy per absorbed mole of photons, qi stands for the heat released in the particular relaxation process,
8892 J. Phys. Chem., Vol. 100, No. 21, 1996
Figure 3. Schematic representation of the photophysical behavior of 1 in alkane solvents. S0 ) ground-state D-br-A, S1 ) locally excited cyanonaphthalene state (D-br-1A*), CT ) extended charge transfer state (D+-br-A-), T1 ) triplet state of cyanonaphthalene (D-br3A*), and isc ) intersystem crossing. The amplitudes φ are assigned i to the various relaxation processes according to their associated LIOAS lifetimes. The energy levels of the excited states are drawn to scale.
and the subscript i refers to the ith decay of the sample molecule. From this simple formula (which refers to a process with unit quantum yield) it is clear that a plot of φi against cpF/β will yield qi from the intercept and ∆Vstr from the slope. As explained in the Introduction, the variation of cpF/β was achieved by measuring the acoustic signal of 1 in a series of n-alkanes ranging from pentane to hexadecane. The underlying assumption of this procedure is that the molecular volume changes have the same time behavior as the heat release processes with regard to the launching of the acoustic waves. It appeared that satisfactory fits of the sample signals were only obtained with three sequential exponential decays included in the fitting procedure. In all cases, the program found a very fast decay time (τ1 < 1 ns), a τ2 of about 40 ns, and a slow decay (τ3) of a few microseconds to a few milliseconds, allowing all six parameters to be varied freely.14 The value of τ1 is more or less arbitrary; it only means that this decay is faster than the time resolution of the experiment. Fixing this parameter at any value between 1 fs and 1 ns always resulted in a similar value for the associated amplitude of the process, so this amplitude φ1 can be considered to be a reliable measure for the “prompt” processes.15 The lifetime τ2 could be determined rather accurately, and agreed well with the fluorescence lifetime measured under the same degassing conditions, which is ca. 40 ns. The third decay time is far too long to be determined reliably with our present setup, and here the recovered amplitude may contain slight differences in base-line between sample and reference signals. Nevertheless, it has to be included into the fitting parameters; otherwise the program will give interpolated values for the other lifetimes and amplitudes. Apart from these limitations, the photoacoustic decay times are in line with the previous information on the photophysical behavior of 1, obtained with luminescence techniques10 (see Figure 3): (1) After local excitation at 308 nm of the cyanonaphthalene chromophore and vibrational relaxation to the zero level of the S1 state, fast charge separation takes place, virtually quenching all of the local cyanonaphthalene fluorescence (τ , 1 ns, quantum yield Φ ) ca. 1). (2) The so-called “extended CT state”, which has a conformation similar to the ground state, folds rapidly to the exciplex state, under influence of the Coulomb attraction between D+ and A-. The folding rate depends strongly on the viscosity of the solvent,10b but even in n-hexadecane this process still takes place within 1 ns at room
Wegewijs et al.
Figure 4. First amplitude of the fitting function for the LIOAS signal (φ1) Vs the thermoelastic parameters (cpF/β) of 1 in n-pentane, n-hexane, n-heptane, n-octane, n-nonane, n-decane, n-dodecane, and n-hexadecane (from left to right). φ1 corresponds to the formation of the exciplex state of 1.
temperature, in turn completely quenching the CT fluorescence of the extended state (τ < 1 ns, Φ ≈ 1). (3) The exciplex decays via various pathways, radiatively (λmax ) 454 nm, Φ ≈ 0.2) and nonradiatively to the ground state, and via intersystem crossing to the triplet state of cyanonaphthalene. These three processes together result in a lifetime of ca. 75 ns. (4) The triplet state of cyanonaphthalene has a rather long lifetime, typical of spin-forbidden transitions, of 4-6 µs under optimal degassing conditions. In Figure 3 the amplitudes φ1-φ3 have been assigned to the processes 1-4 described above. With the time resolution of our LIOAS experiment it is not possible to resolve processes 1 and 2. These are detected as one fast decay, having a time constant of