Solvent Depend'enceof 9-Methyl Anthroate Fluorescence
1611
Relation between an Excited State Geometry Change and the Solvent Dependence of 9-Methyl Anthroate Fluorescence' T. C. Werner" and Ronald M. Hoffman Department of Chemistry, Union College, Schenectady, N e w York 12308 (Received November 7, 1972) Publication costs assisted by the Petroleum Research Fund
The interaction between the carboxyl group and the anthracene ring in the excited state results in solvent-dependent fluorescence properties for 9-methyl anthroate. From the increase in Stokes shift in aprotic solvents of increasing polarity, the excited singlet state dipole moment is about 4.5D larger than the ground-state value. The Stokes shifts in protic solvents are significantly greater than those in polar aprotic solvents. This is indicative of a large hydrogen bond contribution to solvation of the excited state. The fluorescence quantum yield ($e) of 9-methyl anthroate in nonpolar solvents is greater than that of anthracene. For the former compound, it is suggested that the excited singlet lies at a position which is intermediate between the first and second triplets. Intersystem crossing to either triplet is thereby inhibited and 4s increases. In solvents of increasing polarity, the 4f of 9-methyl anthroate decreases. This quenching effect is greatest in protic solvents. From fluorescence lifetime and qbs data, the protic solvent quenching is primarily due to increased rates of radiationless decay. A solvent-induced increase in the rate of internal conversion is the likely mode for nonradiative decay in protic solvents.
Introduction The absorption and fluorescence spectra of anthracene have considerable vibrational structure and show a classic mirror image relationship. When a carboxyl group is substituted a t the 9 position of anthracene, the absorption spectrum retains its anthracene-like structure while the fluorescence spectrum becomes diffuse and highly Stokes shifted.2a These observations are attributed to differences in the ground and excited singlet state molecular geometries .2a The carboxyl group substitution also effects other fluorescence propeirties of the parent hydrocarbon. In benzonitrile, the fluorescence quantum yield (4f) for 9-methyl anthroate (9-COOMe) is about twice as large as the relatively solvent independent +e of anthracene.28 This is in contrast to the observation that carboxyl substitution generally inhibits fluorescence.2b The values of 4s and the fluorescence lifetime ( ~ f )for 9-COOMe in ethanol are only about one-third of their respective values in benzonitrile.2a In addition, there is a significant difference in the wavelength maximum of fluorescence for 9-COOMe in the two solvents.2a These data suggest that the fluorescence behavior of 9-COOMe is strongly solvent dependent. We report here a n investigation of the solvent dependance of 9-COOMe fluorescence. From the solvent dependance of the fluorescence maximum, it can be shown that the excited singlet state is considerably more polar than the ground state. From the solvent dependance of and Tp, the radiative and the sum of the radiationless rate constants can lbe calculated. Using these data, it is possible to comment on the modes of solvent-induced quenching. Experimental ;Section Acetonitrile, benzonitrile, 2-propanol, p-dioxane, 2,2,2trifluoroethanoi, acetone, ether, methyl formate, and N,N-dimethylformamide were Matheson Coleman and Bell Spectroquality solvents. Benzene and methanol were
Fisher Scientific Spectroanalyzed solvents. Absolute ethanol from U. S. Industrial Chemicals was used. Deuterium oxide (99.8% minimum isotopic purity) was from Diaprep Incorporated, Atlanta, Ga. 9-Methyl anthroate was prepared from 9-anthroic acid as described p r e v i ~ u s l y . ~ Absorption spectra were recorded on a Cary Model 14 spectrophotometer. Fluorescence spectra were recorded on a Perkin-Elmer Hitachi-MPF-2A spectrofluorometer ,which was uncorrected for spectral response. Correction factors for the xenon source-excitation monochromator system were obtained with a rhodamine B quantum counter as described by Chem4 The response of the emission monochromator-detector side was corrected by comparing the uncorrected emission spectra of 9-methyl anthroate in cyclohexane and in ethanol with spectra obtained on a corrected spectral instrument at the Perkin-Elmer Corp., Norwalk, Conn. From the ratio of uncorrected to corrected response at a given wavelength, a sensitivity curve was generated from 380 to 580 nm. Uncorrected emission spectra were computer corrected by multiplication of the fluorescence intensity a t a given wavelength by the sensitivity factor at that wavelength, Fluorescence quantum yield ($f) determinations were made using the following equation where r and u stand for reference and unknown, respectively
6 u -
$--2 nu2 A, F r
nY2A, F,
where n is the solvent refractive index, A is the solution absorbance at the exciting wavelength ( 0.20 were deoxygenated by five-six freeze-pump-thaw cycles and sealed in cells under vacuum before fluorescence was measured. For smaller quantum yields, the short fluorescent lifetimes precluded significant oxygen quenching. All reported & values are the average of a t least three independent determinations. Fluorescence lifetimes were obtained with the Ortec nanosecond decay time fluorescence system. This instrument employs the monophoton technique.5 The decay curves were deconvoluted and analyzed using the method of moments6 in conjunction with an IBM computer. The best fit of the data was obtained by assumption of a single component decay. Solutions with lifetimes >4 nsec were deoxygenated and sealed as noted above,
Werner and Ronald M. Hoffman
0
0
a QQ
0 Q 0 A08
+ o
+
ia
+ Q
+
+ +
+
Results Table I summarizes the spectral data, Stokes shifts (ta (&), and some fluorescence lifetimes (7f)for 9-COOMe in 28 solvents and solvent mixtures. The absorption spectra of the methyl ester have anthracene-like structureza while the fluorescence spectra are broad, diffuse bands with slight structure being apparent only in cyclohexane. Due to the nature of the spectra, the 0-0 fluorescence band cannot be ascertained and the fluorescence maxima are reported in Table I. The Stokes - t f ) listed in Table I are from the 0-0 absorpshifts (I, tion band to the fluorescence maximum. As solvent polarity increases, as measured by Kosower 2 value^,^ the Stokes shifts also generally increased. This trend is clearly indicated by the plot of t , - ;f us. 2 in Figure 1. Since the positions of the absorption bands are relatively solvent independent (Table I), the increase in l a - tf with increasing 2 primarily reflects a shift of the fluorescence maximum toward lower energies with increasing solvent polarity. In nonpolar and in some of the polar aprotic solvents, the df of 9-COOMe is considerably greater than the relatively solvent independent $f of unsubstituted anthracene (0.28-0.36, ref 8 and 9). Like the fluorescence maximum, the df value is also solvent dependent. The plot of cpf us. 2 in Figure 2 shows that df generally decreases as solvent polarity increases. The relative effects of the various solhlx) vents on the radiative (kf) and nonradiative (& rate constants can be ascertained from the & and Tf data in Table I using the equations h f = 4dtf ( h , = fluorescence r a t e constant)
- t f ) , fluorescence quantum yields
+
= (kf/,4f) - h , ( h l c = i n t e r n a l conversion r a t e c o n s t a n t ) ( k l x = i n t e r s y s t e m crossing r a t e c o n s t a n t ) Values of k f and (h,,. + h l x )for ten representative solvents are listed in Table 11. The solvents in Table I1 can be grouped into three categories based on their ability to affect the rate constants. For the first three solvents, where df > 0.60, there is no significant variation in either k f or (& + k l x ) .In the second three solvents, @ f decreases from 0.55 to 0.32. This decrease in #f is the result of decreasing values of kf and increasing values of ( k l c k , x ) . Note, however, that the absolute change in either set of rate constants is less than twofold, except for acetonitrile. The last four solvents are (h1c
+ h,x)
+
The Journal of Physical Chemistry, Voi. 77, No. 73, 1973
Figure 1. The Stokes shift of 9-COOMe fluorescence as a function of Kosower Z value: 0, ethanol-water mixtures; A , dioxane-water mixtures; f , pure organic solvents. protic solvents for which df < 0.20. The rate constant data for these solvents are subject to greater error because . this fact, the of uncertainty in the very short T ~ ' sDespite trends in the rate constants are quite clear. The variation in kf for these solvents is less than a factor of 2. In addition, the hf values for the protic solvents are not markedly different than the kf values for the aprotic solvents. However, the (kj, + hi,) values for the protic solvents are greater than the values in the aprotic solvents by a factor of 2-25. Thus the greatly reduced &'s found in protic solvents are mostly the result of more efficient nonradiative processes in these solvents. The ratio of the intensities of 9-COOMe fluorescence in 90% water-10% dioxane and 90% deuterium oxide-10% dioxane was measured by comparing the areas of the fluoresence spectra in the two solvents. In this manner, the df value is found to be 14% greater in the latter solvent mixture.
Discussion I. Solvent Dependence of the Fluorescence Maximum. In the ground state of g-COOMe, the planes of the carboxyl group and anthracene ring are nearly perpendicular.lo This accounts for the lack of significant carboxyl group perturbation on the anthracene-like absorption spectrum of 9-COOMe. In the lowest excited singlet state, the carboxyl group and ring approach coplanarity. The diffuse and highly Stokes shifted fluorescence spectrum is suggested to be partly a result of charge transfer interaction between the carboxyl group and ring in this configu(5) W. R. Ware, "Creation and Detection of the Excited State," A, Lamola, Ed., Marcel Dekker, New York, N. Y., 1971. (6) I. lsenberg and R. Dyson, Biophys. J., 9,1337 (1969). (7) E. Kosower, d. Amer. Chem., SOC.,80,3253 (1958). (8) W. H. Melhuish, J. Phys. Chem., 65, 229 (1961). (9) R. S. Becker in "Theory and Interpretation of Fluorescence and Phosphorescence," Wlley-lnterscience, New York, N. Y., 1969, Chapter 1 1 . (IO) R. 0.C. Norman and P. D. Ralph,J. Chem. Soc., 2221 (1961).
1613
Solvent Dependence of 9-Methyl Anthroate Fluorescence TABLE I: Spectral Data for 9-COOMe in Various Organic Solvents and Solvent Mixtures
-
if x Solvent
Za
Ethanol (10%) Ethanol (20%) Ethanol (30%) Ethanol (40%) Ethanol (50%) Ethanol (6OYo) Ethanol (70%) Ethanol (SO'%) Ethanol (90%) Ethanol (1 00%) Dioxane (1 0%) Dioxane (20%) Dioxane (30%) Dioxane (40%) Dioxane (50%) Dioxane (60%) Dioxane (70%) Dioxane (8O0/) Dioxane (90Y0) Dioxane (1(30%) Methanol Trifluoroethanol 2-Propanol Acetonitrile Ben zon i tr i I f? Acetone Benzene Cyclohexarie
93.6 92.6 91.6 90.5 89.2 87.9 86.4 84.8 82.5. 79.6 93.0 91.4 89.9 88.4 86.7 85.0 82.8 80.2 76.7 55.4b 83.6 76.3 71.3 65.5 65.7 54 476
mr
Sa X 1 0 - 4 c m - 1
0.02 0.03 0.04 0.06
0.08 0.09 0.11 0.13 0.15 0.18 0.02 0.03 0.04 0.07 0.10 0.14 0.16 0.22 0.29 0.55 0.1 1 0.02 0.24 0.32 0.65 0.44 0.64 0.68
aSee ref 7. Estimated from ET values. See ref 7. 0-0 absorption band.
ration.za Henc'e the excited singlet is expected to be more polar than the ground state and should exhibit the greater solvent dependence. This is confirmed by the plot in Figure 1. The magnitude of the excited state charge transfer interaction can be evaluated from the solvent dependence of the Stokes shift. Several workers have proposed theoretical treatments for the solvent dependence of fluorescence.ll-l3 When the solute-solvent interaction is primarily of a dipole-dipole nature, eq 1 is obtained,14 where
the terms are defined as follows: b, - bf is the Stokes shift as defined previclusly (see Results), h is Plank's constant, co is the speed of light in a vacuum, t is the solvent dielectric constant, n is the solvent refractive index, p e is the dipole moment of the lowest excited singlet of g-COOMe, p a is the ground-state dipole moment of g-COOMe, and a is the Onsager cavity radius for 9-COOMe. If a known, p e - p a can be evaluated from the slope of a plot of ija - i j f us. the refractive index-dielectric constant term. The a parameter can be estimated from the size of the molecule. Froim X-ray diffraction studies on anthracene15 and Fisher-Hlershfield-Taylor models of 9-COOMe the a value is taken to be 5 A. A plot of eq 1 is shown in Figure 3 for 9-COOMe in several solvents. The line is drawn through the circled points
2.62 2.62 2.62 2.62 2.62 2.62 2.62 2.62 2.62 2.62 2.61 2.61 2.61 2.61 2.61 2.61 2.61 2.61 2.61 2.61 2.62 2.64 2.62 2.62 2.59 2.62 2.61 2.62
c
10-4 cm-1 d
2.01 2.02 2.04 2.04 2.06 2.07 2.08 2.09 2.08 2.1 1 2.03 2.04 2.04 2.06 2.08 2.08 2.08 2.1 1 2.14 2.18 2.10 2.05 2.1 1 2.13 2.14 2.15 2.1 7 2.21
(ia
- if) x
10-3 c m
6.1 6.0 5.8 5.8 5.6 5.5 5.4 5.3 5.4 5.1 5.8 5.7 5.7 5.5 5.3 5.3 5.3 5.0 4.7 4.3 5.2 5.9 5.1 4.9 4.5 4.7 4.4 4.1
rf,nsec
1.9 f 1
4.1 f 1
1.3f 1
14.4 f 0.5 1.7f 1
8.1 f 0.5 12.7 f 0.5 12.9 f 0.5 14.1 f 0.5 13.1 f 0.5
Fluorescence maximum. e f10% accuracy.
which represent data from polar aprotic solvents. From the slope, the value of p e - pLzis 4.5D. The accuracy of this value is uncertain since it depends critically on the Onsager cavity radius. However, even with a possible error of &50%, as previously estimated,14 the p e - pg. value is indicative of a rather polar excited state. By comparison, Mataga, et al., have found pe - pa to be 0.5-1.OD for pnaphthyl methyl ether and a,P-naphtholsll while Seliskar and Brand found pe - p a to be 9D for 2-amino-6-naphthalenesulfonate.14 The increased dipole moment of the excited state is consistent with two other observations. Werner has studied the temperature dependence for the fluorescence spectra of the tert-butyl ester of 9-anthroic acid in ethanolmethanol ( l : 5 ) . I 6 He has estimated the solvent relaxation energy to be ca. 1500 cm-l which is characteristic of excited states with a charge-transfer component. Recently, Schulman and Pace have shown that the 9-anthroic acid acidium cation becomes more basic in the excited state.17 The lowest anthracene excited state (ILa) is polarized parallel to the short axis. A migration of electron density on excitation from the ring to the carboxyl group (11) N. Mataga, Y. Kaifu, and M. Koizumi, Bull. Chem. SOC.Jap., 29, 465 (1956). (12) E. Lippert, Z.Naturforsc,',. A, 10,541 (1959). (13) E. McRae, J. Phys. Chem., 61,252 (1957). (14) C.Seliskar and L. Brand, J. Amer. Chem. Soc., 93,5414 (1971). (15) J. M. Robertson in "Determination of Organic Structures by Physical Methods," E. Braude and F. Nachod, Ed., Academic Press, New York, N. Y . , 1955. (16) T. C. Werner, Ph.D. Thesis, M.I.T., 1969. (17) S. Shulman and I . Pace, J. Phys. Chem., 76,1946 (1972).
The Journal of Physical Chemistry, Vol. 77,No. 13, 1973
1614
T. C. Werner and Ronald M. Hoffman
0.70-
,
-
0.60
0,
0.60-
+
-
40-
0.40
-
0.30
36 0 I5
a
+
% : 60
70
z
Figure 3. The Stokes shift of 9-COOMe fluorescence as a function of Solvent polarity parameters. The numbers refer to the following solvents: 1 = ethyl ether; 2 = benzonitrile; 3 = methyl formate: 4 = N,N-dimethylformamide; 5 = acetone; 6 = acetonitrile; 7 = 80% dioxane-20% water, 8 = 60% d1oxane-40% water; 9 = 40% dioxane-60% water; l o ’ = 20% dioxane-80% water.
SO
,
SO
100
(KC%OLI)
Figure 2. The quantum yield of 9-COOMe fluorescence as a function of Kosower 2 value: 0, ethanol-water mixtures: A , dioxane-water mixtures; +, pure organic solvents. TABLE II: Values of
the Radiative (SI- S O ) and the Sum of the Radiationless Rate Constants (S,W T j , S1 -SO) and Estimated Singlet-Triplet &-TI) Split for 9-COOMe in Several Solvents
Solvent
Cyclohexane Benzonitrile Benzene Dioxane Acetone Aceton itrile Ethanol Methanol Dioxane (50%) Ethanol (500/,)
$f
0 30
28+1 2n*+1
B
000 50
0 25
e-1 ne1
--I
010-
40
0 20
k f x 10-7 sec-’
(kix
x
+
kic) 70-7
sec-’
0.68
5.2a
0.65 0.64 0.55 0.44
5.1a 4.5a
3.8a 3.4a
3.1a
0.32
3.9a
0.18 0.1 1 0.10 0.08
4.4Ib 6.5c
8.4a 18.7b
7.7c 4.2c
2.4a 2.7a
2.5a
4.3a 52.6c 69.3c 48.3c
&(s1Tl),dcm-l
7400 6700 7000 71 00 6800
6600 6400 6300 6100 5900
azk15%. bzk25%. ck50%. d S , is fluorescence maximum, T I is assumed to be 14,700 c m - l which IS the T1 energy for anthracene.
produces an excited-state resonance contribution shown as
t
I
I. Hence protonation of the carbonyl type oxygen should be facilitated in the excited state. Besides the data from polar aprotic solvents, several points are included in Figure 3 from dioxane-water mixtures, It is readily apparent that these points (triangular in shape) significantly deviate from the line through the polar aprotic solvent data. Similar deviations are obThe Journalof PhysicalChemistry, Vol. 77, No. 13, 1973
served for all the data taken in protic solvents. This means, that in protic solvents besides the dipolar effects, significant hydrogen bonding interaction must contribute to the Stokes shift. Referring again to structure I, it is apparent that a stronger hydrogen bond would be formed between solvent and the coplanar excited state than between solvent and the perpendicular ground state. II. Solvent Dependence of $ f . The large increase in $f for 9-COOMe in nonpolar s o h e n t s relative to $f for anthracene is an anomaly. Generally, meta-directing substituents tend to inhibit the fluorescence of the parent hydrocarbon,2b One possible source for the enhanced 4f of 9-COOMe could arise from the relative energies in the singlet and triplet manifolds. For anthracene, the first (TI) and second (T2) triplet levels are located at 14,700 and 26,050 cm-l, re~pectively.~ Since the lowest excited singlet (SI) is only 650 cm-l above T2 favorable Franck-Condon factors enable the intersystem crossing process (s1-T~) to compete quite well with fluorescence.9 Internal conversion (S1-630) is negligible for anthracene.9 Attempts to locate the triplet levels of 9-COOMe by phosphorescence and triplet-triplet absorption have been unsuccessful.18 However, since triplet states are generally less affected by charge-transfer effects than excited singlet states,lg it is expected that the shift in the energies of TI and Tz for 9-COOMe relative to those of anthracene would be smaller than the shift of SI.From the fluorescence maxima, SI of 9-COOMe is ca. 3000 cm-I lower in energy than S1 of anthracene. Assuming the shift in Tz to be considerably less than this, SI may now be as much as 2000 cm-1 below Tz. Thus crossing to T2 from SI might now require an appreciable activation energy. At the same time, SI would still be quite a bit higher than TI SO that the &-+TI process would still be inefficient due to unfavorable Franck-Condon factors. Hence, for 9-COOMe, intersystem crossing might compete less effectively with flu(18) W. Hardy and D. M. Hercules, private communication. (19) E. Vander Donckt and G . Porter, Trans. Faraday Soc., 64, 3218 (1 968).
1615
Solvent Dependence of 9-Methyl Anthroate Fluorescence orescence than it does in the case of anthracene. The result would be a larger df for the former. Cowan and Schmiegel h a w suggested a similar ordering of the electronic energy levels for 9-anthroic acid.20 Another factor which might affect the intersystem crossing rate is; possible geometry differences between SI and the triplet states, If charge-transfer effects are smaller in TI and T2 than in SI, it is possible that the triplet levels maintain the ground-state geometry. The vibrational overlap factors which control the radiationless transition S -T might be unusually small even if the energy of SI is closer to that of TI or T2 than anticipated above. Lim, e t a1.,21 have previously suggested the existence of such “geometry factors.” It is apparent from Table I1 that the 2 dependent solvent quenching of 9-COOMe fluorescence is greatest in solvents or solvent mixtures which have hydrogen bond donor capabilities. The quenching in these protic solvents is primarily a iresult of increased nonradiative decay rates. In addition, the magnitude of ( h l x h,,) is greater for methanol and the water-containing solvents than for ethanol which is consistent with the expected hydrogen bond donor abilities. In the absence of data on the 9-COOMe triplet, it is impossible to determine the relative contributions of h,, and k,, to the quenching process. However, it is possible to make some comment on possible quenching pathways. The shift in the fluorescence maximum of 9-COOMe over the range of solvents used (see Table I) is ca. 2000 cm-l. If the shift in TI is considerably less than this, AE(S1-TI) should decrease as the energy of the fluorescence maximum decreases. Seliskar and Brand showed that such a situation leads to fluorescence quenching through enhanced rates of intersystem crossing (&-TI) for amino-6-naphthalenesulfonate derivatives.14 Table I1 lists estimates of AE(SI-T1) for 9-COOMe in several solvents assuming the TI energy to be the same as that of anthracene ( 14,700 cm-1). Although AE(S1-TI) does decrease as 4f decreases in protic solvents, the minimum value of AE(S1-TI) is far too large to produce significant Franck-Condon integrals between SI and TI. Also the difference in AE( &-TI) between protic and aprotic solvents is too small to account for the differences in & between these solvents. If this mechanism is operative for g-COOMe, there should be a triplet level at ca. 20,000 cm-I. This would mean either a very large blue shift of TI relative to anthracene or a highly red-shifted T2 which is less solvent dependent than SI. Either case is unlikely and thus an enhanced k i , uia this mechanism is improbable.
+
From the variation in ( t , - t f ) between protic and aprotic solvents, it was implied that the 9-COOMe excited state possesses some of the resonance character of structure I. It seems likely that the hydrogen bond formed between the carbonyl-t ype oxygen in the ester excited state and a protic solvent would be intimately involved in the solvent-induced queiiching. In support of this, the intensity of 9-COOMe fluorescence is 14% greater in 90% Dz0-10% dioxane than i n 90% H20-10% dioxane. The hydrogen bond may provide strong coupling between the excited solute and solvent with the result being an enhanced internal conversion rate a n d subsequent loss of excitation energy to the vibrational modes of the environment. Forster has previously suggested such a mechanism to account for a deuterium effect on N,N-dimethyl-1-amino-5naphthalenesulfonate fluorescence.22 In addition, enhanced internal conversion rates could account for the failure to observe the 9-COOMe triplets by flash or phosphorescence in protic solveniLs.18 I n aprotic solvents (Ta bile 11), a decreasing radiative rate constant also contributes to the observed quenching. The variation in kf for these solvents is attributed to the dependence of hf on the square of the refractive index.Z3 The variation in (hlX kIc) for these solvents is most likely due to a n enhanced hi, resulting from the greater solute-solvent interaction as so lvent polarity increases. It should be pointed out t h a t the benzonitrile data represent an exception to the 4f polarity dependence. Based on 2 values, the +f in benzonitrile should be closer to that of acetone. Apparently the coupling between benzonitrile and excited 9-COOMe which leads to increased k,, values is smaller than solvent polarity would reflect.
+
Acknowledgments. Acknowledgment is made to the donors of the Petroleum Research Fund, administered by the American Chemical Society, for support of this research. The authors would also like to thank the following: Drs. Renata Cathou and James Bunting of Tufts Medical School for the use of their Ortec fluorescence lifetime equipment and for their aid in the analysis of the data, Mr. Tom Porro of the Perkin-.Elmer Corporation for supplying the corrected spectral data necessary for the generation of our spectral sensitivity curve, and Professor Peter Frosch of Union College for helpful discussions. (20) D. Cowan and W. Schmiegel, J . A m e r . Chem. SOC., 94, 6779 (1972). (21) E. Lim, J. Laposa, and J. Yu, J. Mol. Spectrosc., 19,412 (1966). (22) Th. Forster and K. Rokos, Chem. Phys. Lett., 1, 279 (1967). (23) S. J. Strickler and R. A. Berg, J . Chem. Phys., 37, 814 (1962).
The Journalof Physical Chemistry, Vol. 77, No. 13, 1973
