J. Phys. Chem. 1993,97, 10458-10462
10458
Decay Associated Fluorescence Spectra of Coumarin 1 and Coumarin 102: Evidence for a Two-State Solvation Kinetics in Organic Solvents R. W. Yip’ and Y.-X.Wen Dkpartement de Chime, Universitk du Qukbec h Montrkal, C.P. 8888, Succ. A, Montrkal, Qukbec, Canada H3C 3P8
A. G. Szabo’ Institute for Biological Sciences, National Research Council, Ottawa, Ontario, Canada K I A OR6 Received: February 23, 1993; In Final Form: July 1 , 1993’
Bi- and triexpnential and emission-wavelength-independent fluorescence decays have been observed for the 7-aminocoumarins C1 and C102 in several polar solvents. An analysis of the lifetimes of the short-livedcomponents and of the decay-associated spectra of the decay components demonstrates that the solvation dynamics of the excited states fit an irreversible two-state kinetic model. Clear differences in the fluorescence decay kinetics were found between the coumarin dyes in 1-octanol on one hand and in methanol or 1-butanol, on the other, which may indicate subtle differences in the solvation process in the two cases.
Introduction The 7-aminocoumarin dyes C1 and C102 are important laser dyes1 with a charge-transfer type excited singlet state, whose fluorescence is sensitive to the conformation of the dialkylamino group and the. aromatic ring. Photophysical studies on these
c1
c102
dyes suggest that althoughthe lifetimes and fluorescencequantum yields of the conformationally mobile 7-(dialkylamino)coumarins are sensitive to solvent polarity,Z the excited singlet state of these dyes appears to form a twisted internal charge-transfer (TICT) state only in very polar solvents such as Dual fluorescence emission from the planar and twisted states of molecules with charge-transfer properties has not been observed for the aminocoumarin dyes. In the case of the structurally constrained C102, twisting of the dialkylaminogroup to form the TICT state cannot occur. The 7-aminocoumarin dyes display large differences in dipole moments between the ground and excited states and possess strongly emissive and structureless bands that show linear Stokes’ shifts with the Lippert-Mataga polarityfunction.5~6 Therefore, these dyes are prominent members of a class of photophysical probes used in the study of solvation dynami~s.~-lO The solvation process is conventionally analyzed in terms of multiexponential decay kinetics arising from the Bakhshiev modelll that assumes a dynamic shift of a single species. This single species undergoes a continuous shift in spectrum from the initially excited to the final equilibrium solvated excited state. The dynamic solvent shift is normally represented by the function C(t), the fractional change in the time-resolved emission maximum. Recently, we reported that C1 and C102 form weak but distinct complexes with both hydroxylic and nonhydroxylic solvents in both the ground and excited singlet state.12 These .Abstract published in Aduunce ACS Absrrucrs, September 1, 1993.
findings raise the possibility that the solvation process for these probes might be more appropriately represented by a two-state kinetic model in which the observed time-resolved emission spectra (TRES) can be represented by the emission spectra of two species: an initially populated Franck-Condon state and the final solvated state, which are kinetically coupled by a series reaction. In this paper, we report time-resolved fluorescence measurements at different emission wavelengths on C 1 and C 102in several alcohols and water and in aprotic solvents. The experimentshad two objectives: (1) to determine the validity of the two-state model in the solvation process for the two representative coumarins; (2) to search for possible kinetic differences that may arise from twisting of the dialkylamino group which is possible in C1 but not in C102. The data have been analyzed for wavelength-dependent correlations in terms of decay-associated spectra (DAS)” which can show a kinetic correlation between the initial excited state and the final solvent-equilibratedstate. Materials and Methods Chemicals. n-Butyl acetate (Fisher reagent) and dichloromethane (Anachemia spectrograde) were refluxed over CaH2 for 1 h and distilled prior to use. Methanol (Caledon distilled in glass), 1-butanol (Fisher certified), isooctane (MCB spectro) were used without further treatment. 1-Octanol (BDH reagent) was distilled under reduced pressure. All solvents were checked for fluorescenceimpurities. The coumarin dye 7-(diethylamino)4-methylcoumarin (Cl, Exciton) was recrystallized from methanol-water and sublimed. 2,2,5,6- lH,4H-Tetrahydro-8-methylquinolazino[9,9a,l-gh]coumarin(C102, Exciton) was recrystallized from methanol. Steady-State Measurements. The steady-state fluorescence spectra were measured with a SLM 8000C spectrofluorimeter. The emission spectra were corrected using correction factors derived from a calibrated tungsten lamp. Time-Resolved Measurements. The fluorescence decays were obtained with 310-nm excitation pulses (15-ps fwhm) generated at a repetition rate of 825 kHz by a Spectra Physics mode-locked synchronously pumped and cavity-dumped laser system whose output was frequency doubled with a KDP crystal. The fluorescence was detected by a Hamamatsu 1564 U microchannel plate after passing through a polarizer set at the “magic angle” to exclude errors from Brownian rotation and a Jobin Yvon H10 monochromator (4-nm band-pass) and measured by usual time-
0022-3654/93/2097- 10458%04.00/0 @ 1993 American Chemical Society
The Journal of Physical Chemistry, Vol. 97,No. 40, 1993 10459
Two-State Solvation Kinetics
TABLE I: Fluorescence Decay Parameters for C1 and C102 in Different Solvents fluorescence max, nm solvent water
C1 C102 472.0 492.2
TI
1-butanol 445.5 466.0 440.5 457.5
5.96 f 0.01C SW (500 nm)
7L.b
-1
Ps
-3
J
0.53d
= 4.33 f 0.07
1
uc*
i
_. . b
GF(12) 71 =
0.016 f 0.001 2.10 f 0.001 GF (9) 71 = 0.107 f 0.009 72 3.75 f 0.01 GF (14) i1 = 0.190 f 0.003 ~j 0.413 f 0.003 T J 3.80 f 0.01 GF (14) T = 3.17 f 0.01 SW (460 nm) T I = 0.37-0.86r 72 = 3.28 f 0.01 T = 2.89 f 0.01h 72
1-octanol
I0.0005
T Z = 0.416 f 0.001
456.5 476.5
ClO2
c1
73
methanol
1
fluorescence decay parameters,‘ ns
71 72
= 0.022 f 0.004 5.50 h 0.01
9.3e
SW (450 nm) 71 = 0.109 f 0.001 72
= 4.67 f 0.01
GF (11)
118e
rl = 0.139 f 0.001 72 0.439 f 0.002 471f
4.48 h 0.01 GF (15) 73
dichloro- 424.0 441.0 methane n-butyl 407.5 432.0 71 = 0.62-0.8Or acetate 72 3.83 f 0.01 isooctane OGF = global fit (no. of data sets); SW = single wavelength (wavelength). The majority of data is taken with 10-ps channel width except for the data in 1-butanol and 1-wtanol where a 20.5-ps channel width was used. In the case of water solutions, where a higher time resolution was required, 1- and 10-ps channel widths were used. b Longitudinal relaxation time at 293 K, TL = C*/QTD, where tc and g the high-frequencyand zero-frequencydielectric constants, respectively; TD B i the Debye relaxation time. x2 = 1.045 SVR = 1.98. Singleexponentialdecay was obsmed from46540 nm. d Mason, P.R.; Hasted, J. B.; Moore, L. Ado. Mol. Relaxation Processes 1974,6,217. e Castner, E. W. Jr.; Bagchi, M.; Maroncelli, S.P.; Ruggiero, A. J.; Fleming, G. R. Ber. Bunsenges. Phys. Chem. 1988, 92, 363. f Calculated using e.. estimated from the arc plot of the first dispersion region from: Garg, S. K.; Smyth, C. P. J . Phys. Chem. 1965,69, 1294. The lifetime did not appear to be independentof the emission wavelength. * Singlewavelength at 420 nm ( x 2 = 1.045, SVR = 1.98). correlated single-photon in~trumentati0n.l~ The instrumental response function was 60-ps fwhm measured at a resolution of 10or 1ps/channel and was recorded before and after each sample decay measurement. The fluorescence decay parameters were obtained after convolution analysis using the Marquardt algorithm, a background signal being always subtracted prior to deconvolution. The adequacy of fit to the decay data was determined by the inspection of the weighted residuals plot, serial variance ratio (SVR), and root mean sum of weighted squares of residuals (RMSR).IS Decay-Associated Spectra (DAS).The “global” analysis of a multiple set of timeresolved data for the DAS have been previously described.16 The decay associated spectra, DAS,13J6J7which is defined here, as the time-integrated fractional contribution to the total fluorescence by the ith exponential term ai(k)e-t/71 in the generalized fluorescence decay function Z(h,t) = Zy=lai (k)e-‘’rd to the total or steady-state fluorescence spectra Fss(X) is given by
Results
The fluorescence decay kinetics of C 1and C102 in the different solvents were fitted to a sum of exponentials, no physical significance a priori being attributed to the recovered lifetimes or preexponential terms. The “best-fit” parameters are presented in Table I. Where there is more than one kinetic component (bior triexponential decays), the shortest component is labeled as 71, the next longest component as 7 2 , etc. The following trends were observed: (1) The number of components necessary to describe the decay kinetics was the same for both C1 and C102
C
B
p%
J
-1 l
-3
J 1
E3
cl&Q.
d ’
3 1 -1
-3 E3 C1/F120 I
I
I
I
I
400
800
1
0
1200
1600
2000
channels Figure 1. Plot of the weighed residual vs channel number for the
fluorescence decay of C1 in water at 430 nm. (a) Biexponentialfit; 1-ps channel width. (b) Biexponential fit; 10-ps channel width. (c) Triexponential fit; 1-ps channel width. (d) Triexponential fit; 10-pschannel width.
3 5 P
$00
5;O
d0
1i10
16b0
2d5C,
channels
Figure 2. Plots of the weighted residuals vs channel number for a monoexponential fit of the fluorescence decay of Cl02 in water at 460 nm; 10-ps channel width.
in the various solvents except for water. (2) The short-lifetime components were similar for both C1 and C102. (3) A singleexponential decay function was observed for the dyes in dichloromethane and isooctanesolutions. (4) Double exponential decays were observed in 1-butanol, methanol, and n-butyl acetate. In 1-octanol, three exponentialswere required to describe the decay. In the alcohols where multiexponential decays were found, the short-lifetime components had negative preexponential factors at long wavelengths. The residual plots for the two dyes C1 and C102 in water are shown in Figure 1 and 2, respectively. The single and biexponential residual plots for C102 are virtually unchanged (Figure 2), and therefore the decay of C102 in water is best described by a single exponential. In contrast, the residual plots for a biexponential decay a t channel widths a t 10 or 1 ps (Figure la,b) for C1 in water show a poor fit a t low channel numbers. Inclusion of a third, short component of 0.5 ps in the decay function (Figure lc,d), gave an excellent fit. In the data analysis, the value of the 0.5-ps component was held at this constant value and represents an upper limit to a very short decay time component. The 0.5-ps
Yip et al.
10460 The Journal of Physical Chemistry, Vol. 97, No. 40, 1993
TABLE II: "Best-Fit" Fluorescence Decay Parameters for C1 in 1-Butanol at Selected Emission Wavelengths lifetime, ns preexponential factoP a2 L, nm TI 72 415 420 425 430 435 440
0.10 0.13 0.12 0.14 0.18 0.52
443 0.91 445 450 455 460 465 470 475 480 490 500 a
0.03 0.05 0.08 0.08 0.09 0.12 0.10 0.11 0.12 0.12
3.81 3.80 3.78 3.77 3.77 3.756 3.77 3.766 3.78 3.76b 3.75 3.76 3.75 3.75 3.75 3.75 3.74 3.72 3.72 3.72
0.72 0.57 0.47 0.33 0.18 0.04 0.02 -0.68 -0.84 -0.60 -1.12 -1.13 -0.49 -0.191 -1.39 -1.31 -1.61
0.28 0.42 0.53 0.67 0.82 1.oo 0.96 1.oo 0.98 1.oo 1.68 1.84 1.60 2.12 2.13 1.49 2.91 2.39 2.3 1 2.61
Normalized to a1 + a2 = 1. Single-exponentialfit was satisfactory.
decay component is not an experimental or instrumental artifact. The fluorescence results of the same C1 in other solvents such as dichloromethane and isooctane gave excellent fits to singleexponential decay. In methanol, the fluorescence decay of C1 was satisfactorily fit to a double-exponential decay model with no evidence of a very short (
