J . Phys. Chem. 1986,90, 5710-5715
5710
structures are given in Figure 1 of ref 12. It seems reasonable that our B complex for the t conformer corresponds to the linear hydrogen bond type a structure of Murto and Ovaska. In the trans complex labeled B, the OH stretch lies about 5 cm-I below the OH stretch for the uncomplexed alcohol. Since the N 2 B complex has a hydrogen-bond-like structure this lowering of the frequency is reasonable. In the G, conformer the OH group is already weakly hydrogen bonded to the C1 so the formation of the N 2 complex competes with the intramolecular hydrogen bond. Experimentally this causes a +ll-cm-’ shift in the OH stretching frequency. In the other complex (A) there is less perturbation of the O H stretch, especially in the Ggrform. In this case the structure is less clear. It could be the N, interacting either with the C1 atom or with the oxygen lone pair of electrons. The structure of the complex has an influence on the quantum yield. In the complex A where the O H stretch of the Ggtform is not perturbed very much, the quantum yield is only a factor of 5 below that of the uncomplexed form. In the B complex, where there is more evidence of a direct interaction between the OH stretch and the Nz, the quantum yield is lower by a factor of about 20. It had been reported previously that the torsion of 2-chloroethanol was perturbed when small amounts of N 2 were added to the In the spectra of the N 2 complexes of the 2-haloethanols that have been reported before, the OH torsion rises by more than 100 cm-’. While no absorptions specifically due to Nz-alcohol vibrations have been observed, they should appear in the range below 150 cm-’. Such low-frequency vibrations are near phonon frequencies and could increase the vibrations are near phonon frequencies and could increase the vibrational relaxation rate. A faster vibrational relaxation rate and a higher barrier for the reaction would account for the lower quantum yield of the N 2 complexes.
Conclusions It is clear that the single-photon IR photochemistry of 2chloroethanol in Ar and Xe is not the same process as heating the molecule. The photochemical process strongly favors the trans conformer whereas the thermal process favors the lower energy, lower symmetry gauche isomer. The initial quantum yield for the g t isomerization is near unity and is 50 times the quantum
-
-
-
yield for the reverse t g reaction. The g t conversion is both relatively and absolutely more efficient than would be possible if vibrational excitation resulted in a statistical distribution among the energetically available conformers. Within experimental uncertainty, the initial quantum yields imply that virtually every gauche OH molecule excited is converted to trans OH. The g t reaction is probably faster than vibrational relaxation. To achieve the high relative and absolute initial quantum yields t reaction, there must be (a) efficient energy flow for the g to excited torsional levels and (b) an efficient trap for molecules in the trans conformation. The trans isomer would be trapped if it had a rapid relaxation pathway which would quench the reverse t g reaction. Thus our results are consistent with the two conformers having different vibrational relaxation rates which bracket the reaction rates. The trans conformers could relax faster because they have larger amplitude torsional oscillations than the gauche conformers. We speculate that this leads to more efficient coupling of the torsion with the local librational and phonon modes and thus a faster relaxation of vibrational excitation to the lattice in the trans conformers. Evidence is presented for two different 1:l complexes for Nz-2-chloroethanol, in agreement with ab initio calculations. The 1:l complexes have quantum yields lower than that observed in pure Ar. Apparently the additional degrees of freedom and the possibility of photodissociating the complex increase the vibrational relaxation rate. Since relaxation competes with the reaction, the overall quantum yield is reduced. The quantum yield of the complexes is structure-dependent. The complex in which there is evidence for a stronger N2-.0H group interaction has a lower quantum yield, presumably because the N, is more efficient at relaxing the OH group in this complex. Higher complexes, e.g., (N2)z-2-chloroethano1, had quantum yields below our detection limit.
-
-
-
Acknowledgment. A.A. thanks the Magnus Ehrnrooth Foundation and the Finnish Cultural Foundation for grants. Illinois Institute of Technology provided support for part of the work conducted there. J.S.S.thanks the Naval Research Laboratory and Dr. C. A. Marquardt for their assistance and support. Registry No. 2-Chloroethanol, 107-07-3.
Investigation of Irreversible Photochemical Reactions by Translent Grating Techniques F. W. Deeg, J. Pinsl, and Chr. Brauchle* Institut fur Physikalische Chemie der Universitat Miinchen, 0-8000 Miinchen 2, West Germany (Received: March 4 , 1986)
We have demonstrated that one can apply transient grating techniques to the investigation of irreversible photochemical reactions by performing single-shot experiments. Together with CW holographic techniques, this allows the evaluation of photochemical rate constants. The method is of great promise for the study of diffusion-controlled reactions in condensed phases. In the example chosen-the hydrogen abstraction of benzophenone in a PMMA matrix-the grating signal is a superposition of contributions from the triplet state TIof benzophenone, the formed radical pair and the thermal grating induced by the relaxation processes. Variation of the probe beam wavelength Xp and the fringe spacing A allows us to separate and distinguish the various contributions.
Within the last 15 years, laser-induced transient gratings have been used extensively for the investigation of molecular and solid-state dynamics. A great deal of this work has been thoroughly r e ~ i e w e d . ’ ~It~ encompasses the study of a variety of (1) Fayer, M.
D.Ann. Rev. Phys. Chem. 1982, 33, 63.
(2) Eichler, H., et al., unpublished results. (3) IEEE J . Quantum Electron., Special Issue, in press.
0022-3654/86/2090-5710$01.50/0
diffusion processes as, i.e., thermal diffusion in solid^,^ liquids,s and liquid crystals,6 the diffusion of polymers in a polymer blend: of free charge carriers in semiconductors8 and of excitons in (4) Eichler, H.; Knof, J. Appl. Phys. 1977, 13, 209. (5) Eichler, H.; Salje, G.; Stahl, H. J . Appl. Phys. 1973, 44, 5383. (6) Uibach, W.; Hervet, H.; Rondelez, F. J. Chem. Phys. 1983,78, 5 1 13. (7) Leger, L.; Hervet, H.; Rondelez, F. Macromolecules 1981, 14, 1132.
0 1986 American Chemical Society
Reactions by Transient Grating Techniques
The Journal of Physical Chemistry, Vol. 90, No. 22, 1986 5711
0%-'
a.0.65 kg"2.6 . l O L d
Figure 1. Schematic illustration of the transient grating experiment. The crossed excitation pulses Io and Z,generate an interferencepattern in the sample. A CW beam Ipis Bragg-diffracted and yields the grating signal Id.
molecular crystal^.^ Transient gratings have allowed the measurement of orientationalI0 and excited-state relaxation" in many environments. They have been applied to the generation of acoustic and optical phonons and to a thorough study of their propagation characteristic^.'^-'^ All these phenomena investigated so far have in common that they are based on reversible mechanisms. These processes can therefore be studied by signal-averaging or probe-pulse techniques which allow a drastic improvement of the signal/noise ratio. On the other hand C W holographic grating techniques have been developed recently as a versatile and powerful tool for the investigation of irreversible photoreactions in solid matrices,15J6 With this method one can evaluate reaction schemes," identify reactive intermediates,'* or measure total rate constants and quantum yield^.'^*^^ Having this background in mind we have addressed the question whether it is possible to follow irreversible photochemical reactions or, in general, any photoinduced irreversible process by transient gratings, Le., we have tested the feasibility of a single-shot transient grating experiment. At the same time we were interested in the additional results which could be obtained by this method compared to C W holography.
Experimental Section The grating was generated by the split beam of a frequencytripled Nd:YAG laser (Quantel YG 481) with a pulse duration T = 10 ns. Pulse energies used were in the range 3-40 mJ/cm2. The grating was probed by either the 633-nm line of a HeNe laser (NEC GLG 5700/S25 mW) or the 514-nm line of an Ar+ laser (Coherent Innova 90). the diffracted grating signal was detected by a RCA 4840 and 1P28 photomultiplier, respectively, and recorded with a 40-MHz digital oscilloscope (Hitachi VC 441). Benzophenone (purum) was obtained from Fluka and recrystallized twice from methanol. Poly(methy1 methacrylate) (PMMA) was purchased from EGA-Chemie and used without further purification. The polymer was dissolved in acetone to form Eichler, H.J.; Massmann, F. J. Appl. Phys. 1982, 53, 3237. (9) Rose, T.; Righini, R.; Fayer, M. D. Chem. Phys. Letr. 1984, 106, 13. (10) Phillion, D. W.; Kuizenga, D. J.; Siegman, A. E. AppL Phys. Letr. 1975, 27, 85. (11) Eichler, H. J. Opr. Acra 1977, 24, 631. (12) Nelson, K. A.; Fayer, M. D. J . Chem. Phys. 1980, 72, 5202. (13) Nelson, K. A.; Casalegno, R.; Miller, R. J. D.; Fayer, M. D. J. Chem. Phys. 1982, 77, 1144. (14) De Silvestri, S.; Fujimoto, J. G.; Ippen, E. P.; Gamble, E. B., Jr.; Williams, L. R.; Nelson, K. A. Chem. Phys. Left. 1985, 226, 146. (15) Briuchle, Chr.; Burland, D. M. Angew. Chem. 1983,95,612; Angew. Chem. Int. Ed. Engl. 1983, 22, 582. (16) BrBuchle, Chr. Mol. Crysr. Liq. Cryst. 1983, 96, 83. (17) Burland, D. M.; Briuchle, Chr. J . Chem. Phys. 1982, 76, 4502. (18) Briuchle, Chr.; Burland, D. M.; Bjorklund, G. C. J. Am. Chem. Soc. 1981, 103, 2515. (19) Deeg, F. W.; Pinsl, J.; Briuchle, Chr.; Voitlinder, J. J. Chem. Phys. (8)
1983, 79, 1229. (20) Grygier, R.K.; Brugger, P.-A.; Burland, D. M. J. Phys. Chem. 1985, 89, 112.
0
'
20
'
io
'
60
'
do
'
100 t [ p l
Figure 2. Transient grating data for benzophenone/PMMA excited with X, = 355 nm and 8, 4.90° and probed with a HeNe laser at Xp = 633 nm.
1.shot indicates the signal of a fresh sample. nthshot indicates the single-shot signal after n - 1 shots have excited the sample. The solid lines correspond to a theoretical fit as described in the text. a viscous solution, and 5 wt % of benzophenone was added to this solution. Thin films (50-200 Nm) were produced by casting the mixture in a mold and allowing the solvent to evaporate slowly.
Grating Theory A schematic illustration of a transient grating experiment is given in Figure 1. Two mutually coherent laser beams are crossed inside the sample to form an optical interference pattern
[
Z(x) = 2ZO 1
1
+ vcos 2ax A
The fringe spacing A is given by
where & is the excitation wavelength and Be the angle of incidence as shown in Figure 1. Vis the fringe contrast which for Io = Z, is equal to 1. This optical interference pattern triggers photoprocesses in the sample which in turn give rise to a spatially periodic modulation of the complex index of refraction of the sample
iz=n+ik
(3)
A probe beam Zp which strikes the sample under its appropriate Bragg angle is partly diffracted into the collimated beam Z,. The diffraction efficiency 17, Le., the intensity ratio of the diffracted and probing beams, is given by
v = -zd= Ill
Xp is the probing wavelength, Op the angle of incidence of the probe beam, d the thickness of the sample, and OD,, its average optical density after excitation at the probe wavelength. The two terms in parentheses give the separate phase and amplitude grating contributions to the grating efficiency. For small amplitudes of modulation An and AK,i.e., for small diffraction efficiencies 7 C 0.01, as realized in the experiments described here, the functions in parentheses of q . 4 can be replaced by their arguments and the grating signal S turns out to be proportional to the sum of the squares of the changes in n and K, S (An)' (AK)' (5)
-
+
In the experiments presented here contributions to the grating signal can arise from excited states, photochemically new species, and thermal gratings. Whereas the thermal grating is a pure phase
The Journal of Physical Chemistry, Vol. 90, No. 22, 1986
5112
Deeg et al.
I vl
1 ,
m
-e C
01
m
in
c
+ 0
L
W
Figure 3. Transient grating data for benzophenone/PMMA excited with X, = 355 nm and Be = 4.90° and probed with an Art laser at Xp = 514 nm. l.shot indicates the signal of a fresh sample. nthshot indicates the single-shot signal after n - 1 shots have excited the sample. The solid lines correspond to a theoretical fit as described in the text.
grating, excited states and new species can give rise to phase as well as amplitude gratings depending on the wavelength of the probe beam. Therefore, the grating signal is given by S(t)
-
+
(Anth(t) Ani(t) + Anj(t) + . . . ) 2 + (AK,(t) + AK,(t)
+ ...)2
(6)
where i and j denote the various excited states and species present in the sample and the index th stands for the thermal grating.
Results and Discussion Figure 2 shows typical oscilloscope traces obtained by recording the diffracted intensity at a probe beam wavelength Xp = 633 nm. "1 .shot" designates the signal from the fresh sample; the other traces represent the single-shot signals from the same sample after the indicated number of shots. One can perceive a rather complicated time dependence of the grating signal with a slow-decaying component whose intensity decreases with the number of shots, i.e., time of illumination, and a fast-decaying part which seems to increase with the number of shots. To disentangle a grating signal and to separate and distinguish the underlying processes, one can change two parameters: the probe wavelength Xp and fringe spacing A. Variation of A allows the separation of diffusive and nondiffusive processes; variation of Xp gives access to the various species and states involved in the photoprocess due to their different absorption and refraction spectra. Figure 3 shows the grating signals obtained under the same experimental conditions as in Figure 2 but read at a probe beam wavelength Xp = 514 nm. At first view the signals look totally different to those obtained with hp = 633 nm: instead of the minimum and maximum as found in Figure 2, we detect only a smooth decay. A closer look, however, tells us the opposite: again there is a slow-decaying component decreasing with the number of shots and a remaining fast-decaying part virtually identical with the one found in an old sample (i.e., after a large number of shots) a t Xp = 633 nm. Variation of A shows that the fast-decaying component depends on the fringe spacing, whereas the slow-decaying part does not (at least on the length scale of the experiments, = 1 km). The grating signals from an old sample (where only the fast component can be found) for various angles of incidence, i.e., fringe spacings (seeeq 2), are shown in Figure 4a. The smaller the fringe spacing the faster the signals decay. Figure 4b shows that the decay of these signals is purely exponential. If we extract the relaxation time T, from these graphs and plot it vs. A2, we can fit the data by a straight line as is shown in Figure 5. Such a behavior is characteristic of a diffusive process for which theory predicts that the grating signal decays as2s5
When eq 7 is applied, the plot in Figure 5 yields a diffusion
d
0
4
' k ' b ' r b ' Ih ' 1 ' 4 ' Ik 'tI;sl Figure 4. Fringe spacing dependence of the transient grating data of an old sample ( n L 100). These data have been recorded with a probe beam wavelength Xp = 633 nm. Essentially the same results are obtained for Xp = 514 nm.
'
1 ' 1
x1 ,2
,
3
I
4 5 Fringe Spacing
6 AIw]
I
-
Figure 5. Fringe spacing dependence of the relaxation time T, of the data in Figure 4.
constant D = 1 X cm2/s. The relation between diffusion constant D and heat conductivity A,
D = XW/PCV gives X, = 0.174 W m-l K-' close to the value X, = 0.193 W m-l K-' found for pure PMMA.21 The small discrepancy has to be attributed to the addition of benzophenone to the PMMA matrix which reduces the heat conduction in the sample. We have also performed transient grating experiments on an azulene-doped PMMA matrix. After excitation azulene dumps heat very efficiently into the sample but does not undergo a photochemical transformation. The value of A, = 0.185 W m-I K-I obtained by the transient grating experiments in these samples is in good agreement with the other data reported here. Altogether we can safely associate the fast-decaying component in Figures 2 and 3 with the thermal grating induced in the sample by radiationless relaxation following excitation. The slow component in Figures 2 and 3 must therefore be due to the photophysical and photochemical transformations. If we investigate this slow decay in more detail, we recognize two features. Firstly, the signal does not decay to zero but has a small offset on the time scale of the experiment. Secondly, the signal (21) Calvet, E.; Bros, J. P.; Pinelle, H. C. R. Hebd. Seances Acud. Sci. 1965, 260, 1164.
The Journal of Physical Chemistry, Vol. 90, No. 22, 1986 5713
Reactions by Transient Grating Techniques
t-.
t
t
m
Wavelength [nml
Loo
*
L-60
5b0'660'
Wavelength [nml
Absorption spectrum of the triplet state TI of benzophenone/i~opentane~~ and concomitant molar refraction spectrum calculated from these data by assuming homogeneous broadening. Figure 6.
Figure 7. Absorption spectrum of the ketyl radical in isopentane30and concomitant molar refraction spectrum calculated from these data by assuming homogeneous broadening.
cannot be fitted by the assumption of a monoexponential decay. To interpret these data, we have to identify the relevant intermediates in the photochemical process. The general features of the reaction scheme are w e l l - k n o ~ n . ~ After ~ - ~ ~excitation to the lowest singlet
shown in Figures 6 and 7 are not quantitatively reliable. The absorption spectra in Figures 6 and 7 from Godfrey et aL30 refer to the triplet state TI of benzophenone and the free ketyl radical in an isopentane solution; Le., they have been taken in a different solvent and there is some debate about the absolute value of the extinction c ~ e f f i c i e n t . ~Moreover, ~,~~ they refer to the spectrum of the free ketyl radical rather than the radical pair which presumably is the intermediate in solid PMMA.33 Nevertheless we have calculated the wavelength dependence of the molar refraction R which is concomitant with these absorption spectral9 by assuming homogeneous broadening and have depicted them in the same figures. The two probe-beam wavelengths are indicated by small arrows. We recognize immediately that at Xp = 633 nm there is very little absorption by the benzophenone triplet but a sizable molar refraction, the same being valid for the ketyl radical. On the other hand, at Xp = 514 nm we find considerable absorption in Figures 6 and 7, whereas the molar refraction and, concomitantly, An should be negative and go through zero near Xp. These qualitative features, i.e. large An and zero AK at Xp = 633 nm and large AK and small negative An at Xp = 514 nm, have been used as a first guess for the modulation parameters in the fit procedure. The fit procedure has been as follows. The grating data at the probe beam wavelength Xp = 633 nm have been chosen. From the data of an old sample the decay constant kthof the thermal grating has been extracted. Assuming that the triplet decays according to [TI] exp[-(k$)"] and in consequence the radical pair concentration follows [RP] exp[-(kot')"] dt', we have subsequently varied ko and a as well as An and AK associated with the benzophenone triplet and the radical pair in a simplex algorithm34 until the best least-squares fit was obtained. The starting values of Ani and AKi were chosen according to the absorption and refraction spectra in Figures 6 and 7 as mentioned in the last section. The final fit parameters obtained for the first experiment (1 .shot, fringe spacing A,) were then used as initial values for fitting the remaining data after taking into account the different boundary conditions. That is in the case of a (1 .shot, fringe spacing Ak # A I ) experiment, the appropriate thermal decay constant kthwas put in, and in the case of a (nthshot, fringe spacing AI) experiment the scaled values of Ani and AKi were used. This scaling is possible as we can calculate from the laser intensity, the absorption of the sample, and the photochemical quantum yield 4 that after each shot 3% of the existing benzophenone molecules are transformed into LAT. As the changes of the optical constants
lBP li, lBP* ps 3Bp*
(BPH'
+
R')
-
LAT
L kder
state SI,benzophenone relaxes within picoseconds and virtually to unit quantum yield to the lowest triplet state From there it either deactivates to the ground state or-if possible-abstracts a hydrogen from the solvent or host to form a radical pair. This radical pair can react further to form the so-called LAT (light absorbing transient) and in a second photoinduced step finally stable photoproducts. Recently we have shown with a C W holographic technique that the overall quantum yield for the reaction BP LAT is 4 = 0.2.26 On the time scale of our experiment ( T = ~10 ns), the triplet state TI is the first intermediate which can be resolved. It is responsible for the signal at t = 0 (neglecting the thermal grating for the moment) and then decays to the ground state and the radical pair, respectively. We can therefore associate the mentioned offset of the grating signals with the radical pair which decays on a much longer time scale. Unfortunately in our experiment the radical pair signal is too weak for extracting a decay time. The nonexponential signal characterizes the decay of the triplet state TI of benzophenone which-as has recently been pointed out by other groupsZ7~ -exhibits a marked deviation from exponential behavior for temperatures between Ts and Tg,the 8-transition and the glass-transition temperatures, respectively. Several reasons have been put forward to explain this behavior, and we shall come back to this point later. We have been able to fit the data in Figures 2 and 3 (as well as grating signals recorded for other fringe spacings and not depicted here) with the same consistent set of decay parameters assuming a stretched exponential or Williams-Watts law c ( t ) exp[-(kat)"] .29 The mean rate constant ko and the dispersion parameter a are indicated in Figures 2 and 3. W e have not tried to calculate absolute numbers for the phase and amplitude gratings in our experiment as the accessible data
-
-
(22) Schenck, G. 0.; Cziesla, M.; Eppinger, K.; Matthias, G.; Pape, M. Tetrahedron Lett. 1967, 193. Schenck, G. 0.;Matthias, G. Ibid. 1967, 699. (23) Filipescu, N.; Minn, F. L. J . Am. Chem. SOC.1968, 90, 1544. (24) Chilton, J.; Giering, L.; Steel, C . J . Am. Chem. SOC.1976, 98, 1865. (25) Hochstrasser, R. M.; Lutz, H.; Scott, G. W. Chem. Phys. Lett. 1974, 24, 162. (26) Deeg, F. W.; Pinsl, J.; Briiuchle, Chr. J . Phys. Chem., in press. (27) Horie, K.; Morishita, K.; Mita, I. Macromolecules 1984, 17, 1746. (28) Salmasski, A.; Schnabel, W. Polym. Phorochem. 1984, 5, 215. (29) Williams, G.; Watts, D. C. Trans. Faraday SOC.1970, 66, 80.
-
-
(30) Godfrey, T. S.; Hilpern, J. W.; Porter, G. Chem. Phys. Lett. 1967, 1, 490. (31) Porter, G.; Topp, M. R. Proc. R. SOC.London,Ser. A 1970,325, 163. (32) Land, E. J. Proc. R. SOC.London., Ser. A 1968, 305, 457. (33) Karpukhin, 0. N.; Kutsenova, A. V. Vysokomol. Soedin. Ser. B. 1977, 19, 344. (34) Daniels, R. W. An Introduction to Numerical Methods and Optimization Techniques; North-Holland: Amsterdam, 1978.
5714 The Journal of Physical Chemistry, Vol. 90, No. 22, 1986
Deeg et al.
AK
t
@ I \ \ \
,I,----'0 ' 20 '
I
io
60
'
80
iI pT
Figure 8. Phase and amplitude grating contributions to the grating probed at Xp = 633 nm as given by the fit procedure described in the text. AK and An are proportional to the concentrations of the intermediate species we can predict how the phase and amplitude grating contributions due to triplet state T1 and radical pair must change going from shot 1 to shot n. Within these restrictions it is possible to obtain good fits for all data a t the probe beam wavelength Xp = 633 nm with a consistent set of decay parameters as is shown in Figure 2. The dispersion parameter a = 0.65 is independent of the illumination of the sample, whereas the rate constants ko increase with the number of shots. What gives credibility to these data is the fact that when the same procedure is performed for the experimental data a t the probe beam wavelength Xp = 514 nm (starting with initial values taken out of Figure 7) we obtain-within the accuracy of measurement-the same set of decay parameters (ko, CY)although the optical grating amplitudes Ani and AKi are different (see fitted data in Figure
3). The different contributions to the grating as they come out from the fit procedure are shown in Figure 8 (for probe beam wavelength Xp = 633 nm) and Figure 9 (for probe beam wavelength Xp = 514 nm). The upper parts of the figures (denoted a ) show the various grating amplitudes An and hK, the lower parts (denoted b) show the total phase grating (An)2 and total amplitude grating ( w z contributions which add up to give the grating signal S (see eq 5 and 6). The thermal grating only gives rise to a change of the index of refraction Anth. Anth is negative as the density of the sample and, therefore, the index of refraction decrease with increasing temperature. At Xp = 633 nm, AnT, and Anw > 0 (see Figure 8a), and the population of the triplet state increases the refractive index of the sample. Because of the opposite sign of Anth and AnT1and because of their different time dependence, x i A n i (the solid line in figure 8a) goes through 0 for t = t l and has a maximum for t = tz, so that (An)* has a minimum at t = t l and a maximum at t = tZ. As the phase grating dominates the amplitude grating, this behavior is reflected in the total grating signal (see Figure 2). On the other hand, at Xp = 514 nm, An,, and AnRp < 0; i.e., they have the same sign as Anthand x i A n i < 0 at all times. Further, the grating a t this probe wavelength is dominated by the change of the absorption AK,and the phase and amplitude grating add up to a montonically decaying grating signal as shows up in the experimental data a t Xp = 514 nm (see Figure 3). If one compares the fit parameters in Figures 8 and 9 with the data in Figures 6 and I , one finds qualitative agreement. At Xp = 633 nm, An,, 5 AK,, and the radical pair only shows up via
-
Figure 9. Phase and amplitude grating contributions to the grating probed at Xp = 514 nm as given by the fit procedure described in the text. a phase contribution as one should expect from the spectra. Similarly, at Xp = 514 nm a large AK and a negative An for the BP triplet and only an amplitude contribution by the radical pair is found as suggested by Figures 6 and 7. The values of Anw and AKR~in Figures 8 and 9 implicitly contain the quantum yield = 0.2 for the process BP(Tl) R P and are therefore much smaller than the corresponding values found for the triplet state An?l and UT,. However, taking this into account we should still arnve at considerable larger values of A n ~ pand AKW than found in the experiments. We think that this discrepancy is due to the fact that the ketyl radical spectrum in Figure 7 does not quantitatively match with the spectrum of the intermediate radical pair in our sample. At the end of this paper we want to come back to the nonexponential decay of the T1 state of benzophenone and the application of the stretched exponential exp[-(kot)*] to explain our data. The stretched exponential has recently been successfully applied to the explanation of a number of photochemical events in the underlying assumption always being the presence of inequivalent reactive sites in the amorphous host leading to a distribution of first-order rate constants. We think that the same concept is applicable to the deactivation of the triplet state T, of benzophenone and explains the time dependence of our grating signal. We assume that the deactivation of TI is dominated by an intermolecular mechanism, e.g., energy transfer to matrix molecules, a concept recently put forward by Horie and cow o r k e r ~ The . ~ ~meaning ~ ~ ~ of ~ ~ko is t h t of a mean rate constant; whereas the dispersion parameter a is a measure for the width of the site distribution, a gets smaller as the distribution becomes broader. The question arises how can one interpret the fact that k,,incream with the number of shots, i.e., the time of illumination? As we have pointed out, a molecule in the T1 state can either deactivate to the ground state or abstract a hydrogen from the matrix. If we assume to a first approximation that the rate for
+
-+
(35) Doba, T.; Ingold, K. U.; Siebrand, W.; Wildman, T. A. J . Phys. Chem. 1984,88, 3165. (36) Doba, T.; Ingold, K. U.; Siebrand, W. Chem. Phys. Lett. 1984, 103, 339. (37) Vyazovkin, V. L.;Bolshakov, B. V.; Tolkatchev, V. A. Chem. Phys. 1983, 75, 1 1 . (38) Richert, R.; BHssler, H. Chem. Phys. &ti. 1985, 116, 302. (39) Horie, K.; Mita, I. Chem. Phys. Lett. 1982, 93, 61. (40) Hone, K.; Morishita, K.;Mita, I. Kobunshi Ronbunshu 1983,40,217.
J . Phys. Chem. 1986,90, 5715-5719 hydrogen abstraction is constant for all sites (in reality we expect that there exists also a certain distribution for the photochemical rate constant kchwhich may or may not be correlated with the distribution for the deactivation constant kdes),then after the first shot those molecules which deactivate slowly to the ground state have a larger chance to abstract a hydrogen than those which are quenched very easily. As those benzophenone molecules which abstract a hydrogen are irreversibly transformed into LAT, they can no more be excited by the next laser shot. So with each shot the site distribution shifts to molecules which have a larger deactivation constant, and the mean rate constant ko increases as seen in our experiments. Horie and co-workersz7have recently investigated the nonexponential phosphorescence decay of benzophenone in acrylic acid methacrylic polymers. They have explained their data by the introduction of a the-dependent transition term for the dynamic quenching of the T1 state by ester groups of the matrix polymers. At this stage we cannot discriminate between the two models. For a direct observation of the diffusion processes postulated by Horie et al., the fringe spacing in a grating experiment would have to be of the order A = (DT) V2 = [(10-13)(4 X 10-5)]1/2 cm = 0.02 nm. The C W holographic determination of the quantum yield 4 = 0.2 and the mean rate constant ko = 2.6 X lo4 s-l measured in this transient grating experiment allow us to evaluate the pure photochemical rate constant kch. Assuming that the dominant branching takes place in the triplet state, this rate constant for hydrogen abstraction kch = 4ko = 5 X lo3 s-',
Conclusions and Prospects We have demonstrated in this paper that it is indeed possible to perform single-shot transient grating experiments. So in principle every photoinduced irreversible process can be investigated through transient grating techniques. We have additionally shown that photochemical intermediates can be detected and their kinetics be evaluated. Together with the total rate constant and quantum yield obtained through a C W holographic study, photochemical rate constants can be extracted. The transient grating method offers several advantages for the investigation of photoprocesses compared to traditional absorption
5715
techniques. It is inherently a method of high sensitivity as the signal due to the intermediates is recorded on a zero background, whereas absorption techniques often have to rely on the evaluation of a small difference between two large quantities. The grating method is much more versatile as the wavelength of the probe beam used can be chosen outside of the absorption bands of the photochemical species and one can still follow the process in the sample through the pure phase grating. In this case one can additionally increase the sensitivity of the experiment by increasing the intensity of the probe beam. We do not run the risk of damaging the sample as-in contrast to the conventional methods-the probe beam is not absorbed by the sample. What we think is the most promising prospect of the experiment described here is the application to diffusion-controlled chemical reactions. It is possible to probe the diffusion of reaction species over small distances (micrometer scale), and one can separate the contributions of the diffusion constant D and the reaction rate constant k by varying the fringe spacing A. The limitations of this method are naturally given by the minimum accessible fringe spacing Aminwhich determines the minimum diffusion constant D and the maximum rate constant K which can be evaluated. As a consequence it seems unrealistic to investigate such a reaction in the solid state where the grating decay constant will always be dominated by the reactive part of the process. The situation is different in liquids: here an appropriate choice of concentrations and fringe spacings should allow us to follow diffusive and reactive processes simultaneously. A general problem of the grating experiment remains: it is impossible to identify reactive intermediates by a pure grating technique. One can expand the technique described here in a straightforward manner to the study of more complex reaction schemes as, e.g., a two-photon process by performing two-pulse experiments, etce41
Acknowledgment. For support of this work we thank the VW-Stiftung and the Fonds der Chemischen Industrie. Registry No. PMMA, 901 1-14-7; benzophenone, 119-61-9; azulene, 275-51-4. (41) Deeg, F. W.
Ph. D. Thesis, Universitat Miinchen,
1985.
Hydrogen Abstraction of Benzophenone from Polymer Matrices: Evaluation of Quantum Yields and Photomechanical Effects F. W. k e g , J. Pinsl, and Chr. Brauchle* Institut fur Physikalische Chemie der Universitat Miinchen, 0-8000Miinchen 2, West Germany (Received: March 20, 1986; In Final Form: June 24, 1986)
We have investigated the hydrogen abstraction of benzophenone in polymer matrices by holographic grating and absorption spectroscopic procedures. The four polymers used, poly(methy1 methacrylate), poly(isobuty1 methacrylate), poly(buty1 methacrylate), and poly(viny1 acetate), are characterized by different glass transition temperatures ( T J . We find a strong dependence of the photochemical quantum yield I#J on Tg,Le., the rigidity of the polymer matrix. The holographic grating experiments are very sensitive to changes of the overall density of the sample, and a comparison to the absorption experiments allows an accurate determination of photomechanical effects in the sample. The different magnitude of these photomechanical effects in the matrices investigated supports the assumption that the rigidity and microviscosity of the matrix are the rate-determining factors for the photochemial reaction.
1. Introduction The influence of the physical properties of polymeric media on the reactions of incorporated photoactive molecules has been an area of great interest in the past years.' The investigators have focused on reversible processes as the cis-trans isomerization of (1) Smets, G. Adu. Polym. Sci. 1983, 50, 17 and references therein.
aromatic azo compoundsZand the ring opening/closure reaction Of spirobenzopyran derivative^.^ It has been shown for these (2) (a) Lovrien, R. Proc. Natl. Acad. Sci. U.S.A. 1967,57, 236. (b) Blair, H. S.; Pogue, H.I.; Riordan, E. Polymer 1980,21, 1195. (c) Irie, M. et al. Macromolecules 1981.14, 262. (d) Irie, M.; Schnabel, W. Macromolecules 1981, 14, 1246. (e) Matejka, L.; Dusek, K. Makromol. Chem. 1981, 182, 3223.
0022-3654/86/2090-5715$01SO/O 0 1986 American Chemical Society
