Solvent Dependence of the Fluorescence Lifetimes of Xanthene Dyes

Photochemistry and Photobiology, 1999, 70(5): 737-744

Solvent Dependence of the Fluorescence Lifetimes of Xanthene Dyes

Douglas Magde*l, Gail E. Rojas' and Paul G. Seybold2 'Department of Chemistry and Biochemistry, University of California at San Diego, La Jolla, CA, USA and *Departments of Chemistry and Biochemistry, Wright State University, Dayton, OH, USA

Received 9 March 1999; accepted 23 August 1999

ABSTRACT

Fluorescence lifetimes of five representative xanthene dye species-the rhodamine B zwitterion (RB'), the rhodamine B cation (RB+), the rhodamine 6 6 cation (R6G+), the rhodamine 101 zwitterion (R101') and the fluorescein dianion (F2-)-were measured in HzO, D20 and in a series of alcohol solvents ranging from methanol to octanol. The lietimes of both RB* and RB+ increased markedly as the solvent was varied from water to actanol. In contrast, the lifetimes of R6G+ and R101' decreased slightly over the alcohol series and that of F2- increased only slightly in the same series. For all the dyes studied the fluorescence lifetimes observed in DzO were slightly longer than those in H20. Possible causes for the variations observed are discussed.

INTRODUCTION

The xanthene dyes are probably the most intensely studied class of luminescent dyes. Interest has been spurred both by the special spectral characteristics of the dyes and by their wide range of applications as biological stains, sensitizers, tracing agents, photochromic and thermochromic agents and laser dyes, for example. Several members of this class have been recommended as luminescent standards (1-3), while others have found use as fluorescent probe indicators of mi- croscopic environments, especially in enzyme and mem- brane studies. For many of these applications, and especially for their roles as fluorescent probes, it is important that the factors determining their excited state behaviors be well understood. Despite the scrutiny the xanthenes have received for more than a century, a number of aspects of their photophysics remain unresolved or controversial.

In large part the same features that have made the xanthene dyes useful have led to difficulties in understanding their photophysical behaviors. Xanthene dyes typically exist in solution in a variety of neutral and ionic forms, and several can interact to form dimers. The equilibria among these forms that are sensitive to temperature, pH, solvent, concentration and other factors (4-9) are only partially understood.

*To whom correspondence should be addressed at: Department of Chemistry and Biochemistry, University of California at San Di- ego, La Jolla, CA 92093-0358, USA. Fax: 858-534-0130; e-mail: [email protected]

0 1999 American Society for Photobiology 003 1-8655/99 $5.00+0.00

A number of earlier reports have examined the influence of the solvent environment on xanthene dye photophysics, but there remains considerable scatter in the reported lifetimes and fluorescence efficiencies, and no broad-based systematic study appears to have been carried out. In this report we present a systematic examination of the influence of solvent environment on the fluorescence lifetimes of five xanthene dye species: the rhodamine B zwitterion (RB')?, the rhodamine dye cation (RB+), the rhodamine 6G cation (R6G+), the rhodamine 101 zwitterion (R101') and the fluorescein dianion (F-). These dyes were chosen because (1) they are well-known, important representatives of the main types of xanthene dyes, (2) they display key structural features that might be expected to be important in influencing emissive properties, and (3) they are widely used in practical applications such as laser dyes, fluorescent probes, etc., for which understanding variations in their emissive behaviors is ex- tremely useful. In addition, these compounds have been the subject of a large number of earlier individual examinations, so that accurate determinations of their lifetimes in common solvents would provide important certification and data for future reference.

MATERIALS AND METHODS Materials. The structures of the dye species examined are shown in Fig. 1. Laser-grade dyes for most were obtained from Acros Organ- ics (Fisher Co.) and used as received, except that RlOl was obtained from Exciton as the perchlorate salt, which they have named R640. Because of concern over preferential solvation by minor contami- nants (1&13), especially water, the solvents were carefully dried over molecular sieves. Prior to placement in the solvents the molecular sieves were washed in methanol and ethanol to remove inter- fering impurities and then activated by heating. Solutions in the alcohol solvents were prepared from octanol stock solutions. Except as noted, no attempt was made to exclude oxygen.

Several properties of the solvents used are summarized in Table 1. A trace of chloroacetic acid was added to the RB solutions to yield the cation form RB'. The zwitterion dominates in neutral solutions. The R6G+ cation and the R101' zwitterion dominate in neutral solutions. To obtain the fluorescein dianion, aqueous solutions of FL- were prepared as M NaOH; for the alcohols, a small amount of triethylamine M) was added to obtain the dianion. Except as noted, single exponential lifetimes provided ev-

?Abbreviations: 4-AF', 4-aminophthalimide; DMSO, dimethylsulfoxide; F, fluorescein; HCPF, 6-hydroxy-9-(3-carboxyphen- y1)fluorone; HPF, 6-hydroxy-9-phenylfluorone; IC, internal conversion; L, lactone; R101, rhodamine 101; R6G, rhodamine 6G; RB, rhodamine B; TFE, trifluoroethanol; TICT, twisted, intramolecular charge transfer; ULM, umbrella-like motion.

737

738 Douglas Magde et a/.

&"" RB+

n

8"""- Ft

&""- R101+

Figure 1. Structures of the dye species examined in the present study.

idence that a single, dominant, colored species was present for each condition, as expected from prior reports. Lactones coexist for some dyes, especially in the higher alcohols; but as they are colorless and do not absorb the excitation light, they are of no concern for the present purpose. Measurements were taken at 22 2 1°C.

Merhods. Fluorescence lifetimes were measured using time-correlated photon counting. The excitation source at 400 nm was a frequency-doubled Ti-sapphire laser, self-mode-locked by the Kern lens effect. The photomultiplier was a Hamamatsu R1564U micro- channel plate device and the constant-fraction discriminator was a Tennelec TC454. A half-meter monochromator (bandpass 5 5 nm) was used for wavelength selection. Polarization was adjusted on input and output so that detection was at the magic angle, ensuring that rotational relaxation did not affect lifetime determinations. The instrument response function had a full width at half height of about 80 ps (that was more than adequate for the present purpose), largely due to dispersion in the monochromator. Much more important was long-term stability and freedom from satellite pulses. The Ti-sapphire laser offered significant improvements in these latter areas over earlier attempts. Additional details have been described elsewhere (14,15). (The added stability reduced the need for the internal timing reference described earlier, and it was not used.)

Although deconvolution is not really needed for the long lifetimes of xanthene dyes, it was used to verify the absence of fast processes. Typical measurements were accumulated over 8 min at 1500 counts per second. Goodness of fit was tested by calculating the reduced chi-square and the Durbin-Watson statistic. Measurements were considered satisfactory when the former was less than 1.3 and the latter was greater than 1.5. Each day multiple instrument response functions were recorded and at least two measurements were made on each sample. Different permutations of data and instrument response functions gave deconvoluted results that were usually identical within 10-20 ps. A standard solution of RB' in ethanol was remeasured frequently over several weeks and gave lifetimes identical within 10 ps, after averaging each day's multiple measurements.

RESULTS The results of the measurements are shown in Table 2 and Fig. 2. The emission wavelength detected was usually near 520 nm for FL-, 558 nm for R6G+, 568 nm for RB', 583 nm for RB' and 605 nm for RlOl'; however, measurements were recorded at times throughout the emission spectrum for each species to verify the absence of small fractions of un- desired forms. With the exception of the FL- solutions, the decay curves were single exponentials. The P- solutions persistently showed a second, shorter lifetime in the range 0.3 to about 1 ns and contributing from 1 to 4% of the total emission, with the shorter lifetimes in water and methanol and the longer lifetimes and larger yields in the higher alcohols. This component is ignored in the following discussion.

Preliminary measurements were carried out over a range of concentrations to assure that aggregation was not present. Most measurements were then made on solutions with concentrations near 2 pM. This concentration was high enough to overcome any slight interference from dark current or blank emission but did require attenuation of the laser beam. Even if aggregation is not a problem, higher concentrations produce longer apparent lifetimes for dyes with such high quantum yields, because of absorption and reemission. Mi- crochannel plate multipliers are known to have a slight sen- sitivity to count rate. Although this should be negligible for the long lifetimes of these dyes, we kept the count rate ap- proximately constant by varying laser attenuation and also verified that the results were independent of count rate for a range exceeding those used for measurements. Varying the

Table 1. Selected properties of the solvents employed; the Ostwald coefficient refers to dissolved O2

Boiling References Dielectric Ostwald point* Viscosity* index* constants* coefficients?

Solvent ("C) (mPa-s, 25°C) (20°C) (at 25°C) (at 25°C)

H2O 100.0 0.890 1.333 78.54 0.03 1 1

Methanol 64.70 0.544 1.3288 32.63 Ethanol 78.29 1.094 1.361 1 24.30 0.2412 n-Propanol 97.20 1.945 1.3850 20.1 0.2193 i-Propanol 82.26 2.038 1.3776 18.3 n-Butanol 117.66 2.544 1.3992 17.84$ 0.2119 n-Hexanol 157.0 4.578 1.4178 13.03$ 0.1798 n-Octanol 195.2 7.288 1.4293 10.30$ 0.1707

D2O 101.42 1.10 1.3384 77.94

*CRC Handbook (61). tWilhelm et al. (62) and Bo er al. (63). *At 20°C.

Photochemistry and Photobiology, 1999, 70(5) 739

Table 2. cohol solvents*

Fluorescence lifetimes (ns) of the dyes in water and al-

Solvent RB ' RB' R6G' P- R101'

H20 1.68 1.52 4.08 4.16 4.32 D20 1.84 1.64 4.36 4.36 4.80 MeOH 2.46 2.15 4.13 4.28 4.59 EtOH 2.93 2.45 3.99 4.25 4.46 i-PrOH 3.26 2.69 3.96 4.36 4.36 n-PrOH 3.27 2.79 3.95 4.42 4.30 BuOH 3.37 2.98 3.89 4.40 4.30 HxOH 3.41 3.09 3.88 4.48 4.27 OcOH 3.42 3.18 3.83 4.52 4.17

*Estimated precision -20.03 ns and absolute accuracy -20.1 ns (see text).

concentration of additives in the buffers might have an effect on lifetimes; in particular, excess triethylamine seemed to have a small quenching effect. We endeavored to keep conditions identical. Results are listed for air-saturated solutions; the effect of oxygen is discussed below. Overall the variations in the lifetimes listed, for conditions intended to be identical for all measurements, are believed reliable to at least 40 ps and probably to 20 ps. Absolute accuracy for the species claimed that might possibly be prepared and measured by several different protocols is believed to be better than 200 ps and probably better than 100 ps, for each condition listed.

It is apparent that different types of behaviors are occurring in these dyes. The fluorescence lifetimes of both RB species, RB' and RE%+, increase by more than a factor of two in going from the aqueous environment to the less polar alcohols. In contrast, R6G+, R101' and P- exhibit only relatively minor variations in their fluorescence lifetimes over the same range of solvents. Rhodamine 6G' and R101' show a small but significant decrease in fluorescence lifetimes in the alcohol series from methanol to octanol. Al- though the R101' cation is not included in the table, we did find that adding acid resulted in slightly reduced lifetimes for solutions of R101, similar to the results displayed in full for RB. Finally, the lifetime of P- increases slightly along the same series of alcohols.

A second observation is that the lifetimes of all the dyes are noticeably longer in D,O than in H20. The differences range from 100 to 500 ps and are certainly significant. There are also small, barely significant differences in the dye lifetimes measured in the two propanol solvents.

DISCUSSION Despite long-standing examination, the mechanisms govern- ing many of the photophysical properties of the xanthene dyes remain controversial and even enigmatic. This is not for lack of proposals, but rather because the observed variations in wavelength, fluorescence yield and fluorescence lifetime in these dyes apparently result in many cases from a complex interplay of intrinsic and external influences. Be- cause intersystem crossing in rhodamine dyes is generally regarded as unimportant (& < 0.01) (16) and the radiative decay is thought to be relatively insensitive to environmental influences, the nonradiative internal conversion rate is pre-

5.0 d

4.0 m

1.5

1.0 I I I I 1 I I I I H 2 0 D20 MeOH BOH i-PtOHn-ROH BuOH HxOH OcOH

Solvent Figure 2. Variations of the dye lifetimes with changes of solvent: *, R101'; 0, H2-; +, R6G'; W, RB'; A, RBI.

sumably the critical factor controlling excited-state emission behavior, at least for any large variations. Regarding this last process, attention has focused primarily on the possible roles of the xanthene-ring substituents, the pendant carboxyphenyl group and postulated twisted, intramolecular charge-transfer (TICT) (17) intermediate forms, although additional factors have sometimes been discussed. Many of these proposed mechanisms have been addressed by L6pez Arbeloa and co- workers (18,19).

RB Several features have complicated understanding of the spectroscopy of RB. For one thing, this compound exists in solution in a number of neutral and ionic forms (20). The observed absorption spectrum of the neutral form is complicated by the existence of a temperature- and solvent-sensitive equilibrium between the deeply colored zwitterion and a colorless lactone form in protic solvents (5.7-9). T h i s equilibrium causes the apparent zwitterion absorption strength to be less than its intrinsic absorption strength, and can thereby lead to deceptively low estimates for the inherent radiative rate of this form. The scatter among early estimates of the fluorescence lifetimes and quantum yields of these forms, perhaps partially caused by incomplete understanding of the relevant equilibria, has added to the confusion. In this regard we note that the present results for the lifetimes of the RB zwitterion and cation in both aqueous solution and ethanol (Table 2) are identical within -70 ps to those measured fair- ly recently by L6pez Arbeloa et al. (21), lending confidence to these values and suggesting also that our estimates of relatively small experimental uncertainties are reasonable.

Before considering environmental influences on the emission characteristics of RB, it will be helpful to review the influences of inherent molecular structural features on the fluorescence of this compound. Two structural features are observed to influence RB emission: (1) the condition of the pendant carboxyphenyl group and (2) the nature of the ami-

740 Douglas Magde et a/.

no group substituents. In the case of the former, it is found that the fluorescence quantum yield in water decreases as the carboxylic acid group is progressively protonated and ester- ified, from +f = 0.31 for RB' (COO-) to +f = 0.24 for the cation RB+ (COOH) and +f = 0.19 for the ethyl ester (rhodamine 3B, COOEt) (21). The fluorescence lifetimes decrease in the same order. As for the alkylamino substituents (-NR,R2), it is found that the fluorescence yield in water falls progressively as these groups are increasingly alkylated, from +f = 0.93 for rhodamine 110' (R, = R2 = H), to 0.71 for rhodamine 19' (R, = Et, R2 = H), to 0.31 for RB' (21), and the fluorescence lifetimes decrease in this same order. A similar trend, but with consistently smaller +f, was determined for the analogous cationic forms (21). Solvent influences at these two positions therefore represent a good start- ing point for understanding environmental effects on RB emission.

Theoretical understanding of the variations in dye excited- state lifetimes has evolved from the early loose bolt concept of Lewis and Calvin (22), which recognized that flexible groups attached to chromophores tend to enhance nonradiative deactivation. Drexhage was among the first to stress the importance of the flexibility of the amino moieties of rhodamine dyes in promoting nonradiative deactivation (23,24). Tredwell and Osborne showed that the fluorescence lifetime of the rhodamine dye fast acid violet 2R in a series of solvents could be related to the solvent viscosity q according to an equation of the form Tf = Cq*, presumably due to restraint on torsional motion about the C-N bonds (25-27). In contrast, R101, in which the amino groups are held rigid by bridging carbon structures, displays a fluorescence lifetime and fluorescence efficiency nearly independent of changes in solvent and temperature (2,24). The trends in the lifetimes of RB' and RB+ appear roughly, but not exactly, to parallel the variations in the viscosities of the solvents employed (c j Tables 1 and 2). This suggests that some viscosity-dependent process, possibly motion of the diethylamino groups, may be operative. Another possibility for a viscosity-dependent process is that the zwitterion t) lactone equilibrium, familiar in the ground state, is occurring in the excited singlet state as well. The similarity of the lifetime trends displayed by the cation to those of the zwitterion would thereby imply that the cation can somehow engage in an analogous process. However, the relationship between the observed lifetimes and viscosities is at best approximate, and in some instances not at all followed (28-30), so that one must look beyond this simple picture.

Vogel et al. (31.32) and Casey and Quitevis (33) have proposed a model in which nonradiative deactivation of the lowest singlet state in the rhodamines proceeds via a non- fluorescent TICT state. Accessibility of the TICT state in this model is governed both by the electron-donating power of the substituted amino groups and by steric factors that might impede the transition from the originally excited planar fluorescent state. The electron-donating ability of the amino groups is expected to increase with alkylation, consistent with the decrease in fluorescence yield and lifetime observed as this occurs. The role of the solvent is seen as arising from its viscosity (resistance to twisting) and its polarity (that af- fects stabilization of the polar TICT state). Casey and Qui- tevis argue that solvent polarity is the dominant influence,

and that viscosity plays a weak or negligible role. Thus, when a dye possesses the structural features required for TICT state formation it should exhibit a shorter lifetime in more polar solvents such as water that stabilize the TICT state, than in less polar solvents such as the higher alcohols.

Casey and Quitevis (33) studied the RB+ cation in normal alcohols up to n-hexanol and Vogel et al. (31) examined the photophysics of a series of rhodamine ethyl esters. The results of the present study are generally consistent with these studies, as well as results of Snare et al. (34) but show also that the RB zwitterion follows the same pattern of behavior as the cation. It is plausible, then, that the solvent-dependent trends found for these species result from the relative abili- ties of the solvents to stabilize a TICT state in equilibrium with the planar fluorescent excited singlet state. Results from extended Hiickel and Pariser-Parr-Pople molecular orbital calculations, however, appear not to support the TICT model (35). No direct experimental detection of the TICT state has so far been reported (18), but because this state is nonfluo- rescent such evidence would be difficult to obtain.

L6pez Arbeloa et al. (19,21) have proposed a third mechanism in which internal conversion is governed by umbrella- like motion (ULM) of the amino groups from a planar to a pyramidal structure. This change is proposed to allow free rotation of the amino substituents (which in this case does not necessarily increase internal conversion) and to displace the positive charge from the amino group to the xanthene ring system, especially to the critical 9-carbon position. They suggest that increases in the nonradiative decay rate are related to decreases in the interaction between the xanthene ring and the pendent phenyl-COOR group, so that nonradiative decay is in the order COO- < COOH C COOEt. Unfortunately, the TICT and the ULM hypotheses predict the same behavior after postulating different mechanisms.

Specific solventaye interactions can alter and interfere with these proposed mechanisms. Water strongly solvates the COO- group, so that the interaction between the COO- group and the xanthene ring is decreased and the nonradiative decay rate increased. This causes the fluorescence lifetimes of both RB species in water to be relatively short. The shifts to longer lifetimes in the higher alcohols coincide with the weakening solvation of the phenyl-COOR group. An additional possible solvent-dependent factor arises from the observation in solvatochromic and thermochromic studies that the zwitterion RB' is stabilized by hydrogen bonding (presumably at the COO- group) in protic solvents, whereas only the lactone (L) form of RB is present in aprotic solvents (5,7-9). Ability to stabilize the zwitterion decreases in alcohols with longer hydrocarbon chain lengths or greater steric hindrance about the -OH group. Although the lifetime of RB' is slightly longer in n-propanol (with an unhindered -OH) than in i-propanol (with a hindered -OH), the general trend toward longer lifetimes in the higher alcohols makes unlikely a hydrogen-bonding effect on the RB' t) L equilibrium as an important feature influencing nonradiative decay and determining the lifetimes.

In addition, the solvent can participate in specific interactions with the diethylamino groups (18,28,29). Water does not interact strongly with the diethylamino groups, whereas the alcohols, because of their combined hydrophobic and polar characters, can interact to a greater extent with these


substituents. Hydrogen bond donation from the alcohol solvent to the amino nitrogen produces a slight blue shift in the absorption spectrum and stabilizes the positive charge on the amino substituent. This effect, plus the increase in solvent viscosity in the higher alcohols (which inhibits their motion), decreases internal conversion and leads to a higher fluorescence yield and a longer emission lifetime.

The results for RB are thus consistent with both the TICT and the ULM pictures of nonradiative decay in RB. Within the TICT model the decreased polarity of the higher alcohols leads to diminished stabilization of the polar, deactivating, TICT state, and hence to longer fluorescence lifetimes. In the ULM model specific solvent-solute interactions stabiliz- ing the planar welectron system lead to decreased internal conversion and longer lifetimes.

R6G and RlOl

As seen in Table 2 and Fig. 2, the lifetime of R6G' is longer thpn that of the two RB species, and after a modest increase in going from H 2 0 to D20, falls just slightly in the alcohol series from methanol to octanol. The latter lifetime trends are quite distinct from those shown by the two RB forms and draw attention to the differences in the structures of these two dyes. The R6G' differs in several crucial ways from RB: (1) R6G' has monoethylamino groups in place of the diethylamino groups in RB, (2) R6G+ has methyl substituents adjacent to the two amino substituents, and (3) the carboxyl group on the pendent 9-phenyl substituent is ester- ified.

The R6G' results permit a test of the TICT-state model. According to this model R6G' does not exhibit the variations seen for RB' and RB' because the ionization potential of its monoethylamino groups is too high to favor TICT state formation. However, if TICT state formation is the only important nonradiative decay channel from the planar excited state, and this channel is absent in R6G+, the fluorescence quantum yield +f of R6G' should be near unity. Because quantum yields are difficult to measure, it is not surprising that they exhibit even more scatter than lifetime measurements; we believe that Drexhage's value of = 0.95 in H 2 0 (36) is the most reliable of the available values, even though L6pez Arbeloa et al. prefer 0.59 (21). The integrated absorption spectrum of R6G' implies an oscillator strength close to unity in all solvents, and the calculated natural radiative lifetime is about 4 ns, close to the values measured. Several other groups have obtained fluorescence yields close to +f = 0.95 for this compound (37-39) using a variety of methods. Consequently, mechanisms should be sought that would lead to relatively small differences among the solvents examined. The uniform behavior through the series of alcohols does suggest that there may be a simple explanation for the trend observed among the alcohols.

One possible explanation for slight variations in lifetimes might be differences in oxygen quenching. As emphasized in the classic compilation of Berlman (40), dioxygen quenches the fluorescence of dyes at a rate that is not far from diffusion controlled. For molar concentrations near 1 mM, this leads to a deactivation channel with a rate close to lo7 s-*. This deactivation rate will increase with O2 concentration but decrease with solvent viscosity. In air-saturated

water [O,] = 0.2 mM. The solubility of O2 in the alcohols, as measured by their Ostwald coefficients (Table l), is higher than in water. However, O2 solubility decreases in the sequence from MeOH to OcOH and, in addition, the viscosity increases. Consequently, the oxygen-quenching hy- pothesis is not attractive for explaining the trend in R6G'. Nevertheless, we measured the effect of O2 directly. Ethanol solutions degassed by bubbling argon had lifetimes about 20 ps longer than air-equilibrated solutions (0.2 atm 02), and ethanol solutions equilibrated with a full atmosphere of oxygen showed a lifetime shortening of about 60 ps compared to air-equilibrated solutions. The changes in hexanol were barely discernible, less than half as much as in ethanol, as expected for lower O2 solubility and increased viscosity. We conclude that variations in oxygen quenching among the solvents are very small, because the oxygen effect itself is very small under all conditions employed in this study. Moreover, the changes are in the wrong direction to explain the trend for R6G'.

One small effect is not only in the proper direction, but even has the correct magnitude to explain the variation in the R6G' lifetimes from D20 to octanol. The radiative rate constant is proportional to the square of the refractive index of the medium surrounding an emissive species. If nonradiative processes are in fact negligible and quantum yields very close to unity, then the observed lifetimes should approximate the radiative lifetimes and should vary inversely with the refractive index squared (absent any solvent-dependent variation in osciIlator strength). This mechanism predicts a 14% difference between octanol and D20, close to what is measured. The mechanism also predicts a smooth variation along the series, as is observed, and even the correct ordering for the two isomers of propanol. The solvent- dependent variation in the radiative rate is necessarily present, and a mechanism based on its variation is consistent with the notions presented above that nonradiative processes are almost absent in R6G' and that R6G' has a fluorescence yield close to unity. The only significant nonradiative influence in R6G' appears to be a process that shortens the lifetime by a small amount in water.

So far as we know, a near-unity fluorescence quantum yield for R101' is not disputed (2,24); and R101' exhibits behavior very similar to that of R6G'. The fact that R6G' and RlOl' are so similar suggests that neither rotation of the amino groups nor any behavior of the carboxyphenyl group necessarily results in enhanced internal conversion. Either those considerations enter only when there is a particular relation among electronic state energies or other features are important. The only process affecting lifetimes discernible in Rl01' is probably the same one active in R6G' that we have tentatively assigned to variation in radiative rate constant, combined with slight, specific quenching in water.

Any such solvent-dependent radiative rate variation must also apply to the other dyes studied here. In the two forms of RB it would be a relatively minor effect. In FI-. however, the implication is that we will need to identify a nonradiative variation sufficient both to offset the change in the radiative rate and to produce observed lifetimes that vary in the op- posite direction.

742 Douglas Magde eta/.

Fluorescein

Like RB, fluorescein exists in solution in a number of ionic and neutral forms (41-43), but we confine our attention to the dianion form. The solvent dependence of Tf for the p- is much smaller than, but in the same direction as, the results obtained for the halogenated derivatives of this compound by Fleming er al. (44). Those workers found that the fluorescence lifetimes of eosin, erythrosin and rose bengal increased substantially in passing from water to the higher alcohols. They attributed the variation to differential solvation of the S1 and TI states resulting in a decreased S, - TI energy gap hEsT and hence more efficient intersystem crossing in the more polar solvents. If this explanation is correct, it is likely that in fluorescein that lacks the strong spin-orbit coupling of its halogenated derivatives intersystem crossing is so small that even significant changes in hEsT have only a small effect on the overall quantum yield and lifetime, as is observed. Spin-orbit coupling is known to be weak in unhalogenated xanthene dyes (45), and triplet yields are cor- respondingly low (46). Bowers and Porter (47) found the triplet yield of fluorescein to be h = 0.05; and Soep et al. (48) found & = 0.03, values that were said not to be affected greatly by the nature of the solvent. (This compares, e.g. with & = 0.76 for eosin, which is tetrabromo-fluorescein (44).) The fluorescence yield of P- in water is high, +f

= 0.93-0.95 (49,50), but it is not unity and increases slightly in going to the less polar alcohols in agreement with the present observed increases in lifetime.

Martin and Lindqvist have argued that hydrogen bonding plays a central role in the nonradiative deactivation of fluorescein and that an equilibrium is established in the excited state between fluorescent H-bonded dyes and nearly non- fluorescent free dyes (51,52). The internal conversion rate is said to depend exponentially on the SI - So energy gap. Studies of the fluorescein analog 6-hydroxy-9-phenylfluorone (HF'F), which lacks the carboxylic acid substituent, showed that this alteration considerably reduces the fluorescence yield while leaving the absorption and emission spectra nearly unchanged (53). Moreover, displacing the carboxylic acid group to the para position of the pendent phenyl moiety, as in 6-hydroxy-9-(3-carboxyphenyl)fluorone (HCPF), has a very similar effect (54). (In acid solution the fluorescence efficiencies of P-, HPF and HCPF are nearly identical.) Supporting an important role for hydrogen bonding is the fact that the fluorescence yield of fluorescein itself increases by 70% in going from dimethylsulfoxide (DMSO) to trifluoroethanol (TFEi); the fluorescence yield of HPF increases seven-fold from DMSO to TFE. However, this effect is in the wrong direction to explain the slightly increased lifetime of P- in the higher alcohols in the present study, because they are likely to have the weaker hydrogen bonding.

The D,O/H,O effect

The persistent, albeit small, variations in lifetimes found between H,O and D20 solutions logically relate to differences in the nonradiative deactivation rates of the dye excited singlet states in these solvents, because radiative rates are not appreciably affected by solvent deuteration. At least three possible explanations present themselves. In the first, one

can argue that H20 is more effective in deactivating the excited states than is D20 because its higher-energy 0-H vibrations are more efficient at removing energy than are the lower-frequency 0-D vibrations of D20. An extreme example of this type of influence appears in the case of deactivation of the forbidden IAS + 38, emission of singlet oxygen, where the emission lifetime is 55 p.s in D20 but just 4 ps in H 2 0 (55)! An analogous intramolecular effect, based on Franck-Condon factors, occurs in the phosphorescence lifetime of benzene, where the lifetime of the deuterated compound is two to three times that of the protonated form at 4.2 K (56). For the present highly allowed transitions of the xanthene dyes the effect is expected to be much smaller but still detectable. A second general possibility is that some crucial dye proton position is deuterated in D,O, such that a high-frequency (e.g. N-H or 0-H stretching) vibrational accepting mode is rendered less effective, thereby decreasing the nonradiative SI + So transition (36). A recent example of this process is found in 4-aminophthalimide (4-AP), where both the fluorescence efficiency and lifetime increase more than four-fold in D20 compared to H 2 0 (57). In 4-AP this increase has been attributed to deuteration of the h i d e proton. Other examples of this type of effect can be found (58). However, there do not appear to be logical possibilities for this effect in RB, and furthermore, even in rhodamines with N-H groups, there is no evidence for such an effect. Finally, we note that the viscosity of D20 is about 20% greater than that of H20 at 300 K (59), so that any viscosity- dependent process can be expected to be slowed according- ly. For example, the diffusion coefficient of 0, in D20 is roughly 17% below that in H20 (60). However, O2 quenching alone does not appear sufficient to explain the changes found for the dyes examined in the present study. Some other viscosity-dependent mechanism might account for the difference between H20 and D20. Trying to invoke a single viscosity-dependent process for both the aqueous media and the alcohols (for RB=, RB+ and possibly P-) does not appear attractive, as methanol and ethanol would be seriously out of place.

CONCLUSIONS Fluorescence lifetimes for five xanthene dye species have been determined in nine solvents. The values determined can be employed with some confidence in a variety of applications and may be useful as calibration standards. Most com- parative studies of xanthenes include at least one of the five species in one of the solvents. These results should facilitate intercomparisons among different studies. They also provide a caution, if more are needed, against relying on any nu- merical value when conditions, such as concentration, have been incompletely specified. The internal consistency of the measurements suggests that the individual lifetimes presented are accurate to about 2-3% (95% probability), while the trends themselves are more precise than that. Three of the five compounds have very high quantum yields for emission. They allow one to exclude a number of factors as necessarily leading to substantial nonradiative decay. Because any effects in those thre species are small, they are not the best examples for elucidating what factors may cause large nonradiative decays in other dyes. Nevertheless, the trends ob-


served in the solvent series allow some inferences concern- ing the nonradiative decay processes in the five compounds. It seems clear that different processes must be dominant under different circumstances. For the two RE3 species, the zwitterion and the cation, an internal conversion process in- volving the substituent groups appears to govern the observed lifetime variations, this lifetime being shortest in water and increasing along the series of higher alcohols. The trend observed in the lifetimes of R6G' for this same series of solvents imply that this compound has a fluorescence quantum yield near unity in all the solvents examined (but slightly less in water) and that the small changes found in T~ may be due simply to effects of the refractive index of the medium on the radiative decay rate. For RlOl' the evidence for near-unity fluorescence yield is even stronger, and the variation of radiative rate is again a plausible explanation for the trend observed, with the same special situation in water. Variations observed in the fluorescence lifetime of fluorescein appear tied to relatively small solvent-dependent influences on the intersystem crossing rate from the fluorescent singlet state to the triplet state. Small but consistent differences found in the fluorescence lifetimes of all the dyes in H,O and D20 (the lifetimes being shorter in H,O) may be associated with more effective vibrational accepting modes for nonradiative decay in H20.

REFERENCES 1. Demas, J. N. and G. A. Crosby (1971) The measurement of

photoluminescence quantum yields. A review. J. Phys. Chem.

2. Karstens, T., and K. Kobs (1980) Rhodamine B and rhodamine 101 as reference substances for fluorescence quantum yield measurements. J. Phys. Chem. 84, 1871-1872.

3. Eaton, D. F. (1988) Reference materials for fluorescence measurement. Pure Appl. Chem. 60, l 107-1 114.

4. Huth, B. G., G. I. Farmer and M. R. Kagan (1969) Temperature- dependent measurements of a flashlamp-pumped dye laser. J.

5. Rosenthal, I., P. Peretz and K. Muszkat (1979) Thermochromic and hyperchromic effects in rhodamine B solutions. J. Phys. Chem. 83, 350-353.

6. Barigelletti, F. (1987) Effect of temperature on the photophysics of rhodamine 101 in a polar solvent. Chem. Phys. Letr. 140, 603-606.

7. Faraggi, M.. P. Peretz, I. Rosenthal and D. Weinreb (1984) So- lution properties of dye lasers. Rhodamine B in alcohols. Chem. Phys. Lett. 103, 3 10-3 14.

8. Hinckley, D. A., P. G. Seybold and D. P. Borris (1986) Sol- vatochromism and thermochromism of rhodamine solutions. Spectrochim. Acra 42A, 747-754.

9. Hinckley, D. A. and P. G. Seybold (1988) A spectroscopic/ thermodynamic study of the rhodamine B lactone zwitterion equilibrium. Specrrochim. Acta 44A. 1053-1059.

10. Midwinter, J. and P. Suppan (1969) Spectral shifts in solvent mixtures. Spectrochim. Acta 25A, 953-958.

11. Chatterjee, P. and S. Bagchi (1991) Preferential solvation of a dipolar solute in mixed binary solvent. A study by UV-visible spectroscopy. J. Phys. Chem. 95, 331 1-3314.

12. Acree, W. E., S. A. Tucker and D. C. Wilkins (1993) Spectro- chemical investigations of preferential solvation. Fluorescence emission behavior of select polycyclic aromatic hydrocarbon solute probes dissolved in mixed solvents. J. Phys. Chem. 97, I 1 199-1 1203.

13. Gratias, R. and H. Kessler (1989) Molecular dynamics study on microheterogeneity and preferential solvation in methanolkhlo- roform mixtures. J. Phys. Chem. 102, 2027-2031.

14. Magde, D. and B. F. Campbell (1989) Improved methodology for time-correlated single photon counting. In Proc. SPIE-

75, 991-1024.

Appl. Phys. 40, 5145-5147.

Fluorescence Detection I l l , vol. 1054 (Edited by E. R. Menzel), pp. 61-68. SPIE, Bellingham, WA.

15. Magde, D. and M. D. Magde (1993) Time-correlated photon counting: Differential methods applied to anisotropy decay. In Proc. SPIE-Advances in Fluorescence Sensing Technology, vol. 1885 (Edited by J. R. Lakowicz, R. B. Thompson and A. Katzir), pp. 413422. SPIE, Bellingham, WA.

16. Korobov, V. E., V. V. Shubin and A. K. Chibisov (1977) Triplet state of rhodamine dyes and its role in production of interme- diates. Chem. Phys. Lett. 45, 498-501.

17. Rettig, W. (1986) Charge separation in excited states of decou- pled systems-TICT compounds and implications regarding the development of new laser dyes and the primary processes of vision and photosynthesis. Angew. Chem. Inr. Ed. Engl. 25,

18. Ldpez Arbeloa, F., T. Ldpez Arbeloa, M. J. Tapia EstCvez and I. Ldpez Arbeloa (1991) Photophysics of rhodamines. Molecular structure and solvent effects. J. Phys. Chem. 95, 2203-2208.

19. L6pez Arbeloa, F., T. Ldpez Arbeloa, P. Hernandez Bartolomb, M. J. Tapia Esttvez and I. Ldpez Arbeloa (1992) TICT and ULM models for the radiationless deactivation of rhodamines. Proc. Indian Acad. Sci. (Chem. Sci.) 104, 165-171.

20. Ramette, R. W. and E. B. Sandell (1956) Rhodamine B equilibria. J. Am. Chem. Soc. 78, 48724878.

21. Ldpez Arbeloa, T., F. Mpez Arbeloa, P. Hernandez BartolomC and I. L6pez Arbeloa (1992) On the mechanism of radiationless deactivation of rhodamines. Chem. Phys. 160, 123-130.

22. Lewis, G. N. and M. Calvin (1939) The color of organic substances. Chem. Rev. 25, 320-328.

23. Drexhage, K. H. (1973) What's ahead in laser dyes? Laser Fo- cus 9, 35-39.

24. Drexhage, K. H. (1977) Structure and properties of laser dyes. In Dye Lasers. 2nd ed. (Edited by F. P. Schafer), pp. 144-179, 269-287. Springer-Verlag, New York.

25. Tredwell, C. J. and A. D. Osborne (1980) Viscosity dependent internal conversion in the rhodamine dye fast acid violet 2R. J. Chem. SOC. Faraday I1 76, 1627-1637.

26. Osborne, A. D. (1980) Internal conversion in the rhodamine dye fast acid violet 2R. J. Chem. Soc. Faraday II 76, 1638-1643.

27. Osborne, A. D. and A. C. Winkworth (1982) Viscosity-dependent internal conversion in an aryl-substituted rhodamine dye. Chem. Phys. Lett. 85, 513-517.

28. Mpez Arbeloa, I. and K. K. Rohatgi-Mukhejee (1986) Solvent effects on photophysics of the molecular forms of rhodamine B. Solvation models and spectroscopic parameters. Chem. Phys. Lett. 128, 474479.

29. Ldpez Arbeloa, I. and K. K. Rohatgi-Mukherjee (1986) Solvent effects on photophysics of the molecular forms of rhodamine. Internal conversion mechanism. Chem. Phys. Len. 129, 607- 614.

30. Chang, T.-L. and W. L. Borst (1990) Effect of solvent polarity on a rotational isomerization mechanism of rhodamine-B in normal alcohols. J. Chem. Phys. 93, 47244729.

31. Vogel, M., W. Rettig, R. Sens and K. H. Drexhage (1988) Struc- tural relaxation of rhodamine dyes with different N-substitution patterns: a study of fluorescence decay times and quantum yields. Chem. Phys. Lett. 147, 452460.

32. Vogel, M., W. Rettig, R. Sens and K. H. Drexhage (1988) Ev- idence for the formation of biradicaloid charge-transfer (BCT) states in xanthene and related dyes. Chem. Phys. Lett. 147,461- 465.

33. Casey, K. G. and E. L. Quitevis (1988) Effect of solvent polarity on nonradiative processes in xanthene dyes: rhodamine B in normal alcohols. J. Phys. Chem. 92, 6590-6594.

34. Snare, M. J., F. E. Treloar, K. P. Ghiggino and P. J. Thistle- thwaite (1982) The photophysics of rhodamine B. J. Photochem. 18, 335-346.

35. Bergamasco, S., G. Calzafem and K. Hadener (1990) Solute- solvent relaxation of electronically excited xanthene dyes. J. Photochem. Photobiol. A53, 109- 125.

36. Drexhage, K. H. (1976) Fluorescence efficiency of laser dyes. J. Res. Nar. Bur. Stand. 80A. 421428.

37. Kubin, R. F. and A. N. Fletcher (1982) Fluorescence quantum yields of some rhodamine dyes. J. Luminesc. 27, 455462.

971-988.

744 Douglas Magde et al.

38.

39.

40.

41.

42.

43.

44.

45.

46.

Vieira Ferreira, L. F., M. Rosario Freixo, A. R. Garcia and F. Wilkinson (1992) Photochemistry on surfaces: fluorescence emission quantum yield evaluation of dyes adsorbed on micro- crystalline cellulose. J. Chem. SOC. Faraday Trans. 88, 15-22. Lee, W. K., A. Giingor, P.-T. Ho and C. C. Davis (1985) Direct measurement of dilute dye solution quantum yield by photo- thermal laser heterodyne interferometry. Appl. Phys. Lett. 47, 916-918. Berlman, I. B. (1971) Handbook of Fluorescence Spectra of Aromatic Molecules, 2nd ed. Academic Press, New York. Lindqvist, L. (1960) A flash photolysis study of fluorescein. Arkiv Kemi 16, 79-138. Martin, M. and L. Lindqvist (1975) The pH dependence of fluorescein fluorescence. J. Luminesc. 10, 38 1-390. Sjoback, R., J. Nygren and M. Kubista (1995) Absorption and fluorescence properties of fluorescein. Spectrochim. Acta A51,

Fleming, G. R., A. W. E. Knight, J. Moms, R. J. S. Momson and G. W. Robinson (1977) Picosecond fluorescence studies of xanthene dyes. J. Am. Chem. SOC. 99, 4306-431 1. Pedash, Y. F., E. N. Babich and 0. A. Ponomarev (1990) Spin- orbit interaction of T-T* states in molecules of aromatic compounds. Opt. Spektrosk. 69. 3 13-3 16 [Transl. Opt. Spectrosc. 69, 190-1911 Butenin, A. V., B. Y. Kogan and V. I. Gavrilyuk (1985) Inter- conversion and oxveen ouenchine of fluorescence of rhodamine

L7-L2 1 .

6G in porous glass.-Opt: Spektrosk. 57, 736-737. [Transl. Opt. Specrrosc. 57, 447-4481

47. Bowers, P. G. and G. Porter (1967) Triplet quantum yields for some aromatic hydrocarbons and xanthene dyes in dilute solution. Proc. R. SOC. (Lond.) A299, 348-352.

48. Soep, B., A. Kellman, M. Martin and L. Lindqvist (1972) Study of triplet quantum yields using a tunable dye laser. Chem. Phys.

49. Seybold, P. G., M. Gouterman and J. B. Callis (1969) Calori- metric, photometric, and lifetime determinations of fluorescence yields of fluorescein dyes. Photochem. Photobiol. 9, 229-242.

50. Brannon, J. H. and D. Magde (1978) Absolute quantum yield

Lett. 13, 241-244.

determination by thermal blooming. Fluorescein. J. Phys. Chem.

51. Martin, M. and L. Lindqvist (1969) Transitions internes dans les colorants xanthtniques Ctudites par photolyse-Cclairs. 2oC Rkunion Socittt Chemie Phys., Paris. J. Chemie Phys., pp. 144- 148.

52. Martin, M. M. (1975) Hydrogen bond effects on radiationless electronic transitions in xanthene dyes. Chem. Phys. Lett. 35,

53. Lindqvist, L. and G. Lundeen (1966) Radiationless transitions in xanthene dyes. J. Chem. Phys. 44, 1711-1712.

54. Fink, D. W. and C. R. Willis (1970) Structure and internal mix- ing in xanthene dyes. J. Chem. Phys. 53, 47204722.

55. Rodgers, M. A. J. (1983) Solvent-induced deactivation of singlet oxygen: additivity relationships in nonaromatic solvents. J. Am. Chem. SOC. 105, 6201-6205.

56. Wright, M. R., R. P. Frosch and G. W. Robinson (1960) Phos- phorescence lifetime of benzene. An intermolecular heavy-atom effect, a deuterium effect, and a temperature effect. J. Chem. Phys. 33, 934-935.

57. Das. S., A. Datta and K. Bhattacharyya (1997) Deuterium iso- tope effect on 4-aminophthaliminde in neat water and reverse micelles. J. Phys. Chem. A 101, 3200-3304.

58. Beck, W. F., M. D. Schuh, M. P. Thomas and T. J. Trout (1984) Intermediate-sized deuterium effect on TI-&, intersystem crossing in acetaldehyde. J. Phys Chem. 84, 3431-3435.

59. Matsunaga, N. and A. Nagashima (1983) Transport properties of liquid and gaseous D20 over a wide range of temperature and pressure. J. Phys. Chem. Re$ Data 12, 933-966.

60. Han, P. and D. M. Bartels (1996) Temperature dependence of oxygen diffusion in H,O and D20. J. Phys. Chem. 100, 5597- 5602.

61. CRC Handbook of Chemistry and Physics, 77th ed. ( 1995) CRC Press, Boca Raton, FL.

62. Wilhelm, E., R. Battino and R. J. Wilcock (1977) Low-pressure solubility of gases in liquid water. Chem. Rev. 77, 219-262.

63. Bo, S., R. Battino and E. Wilhelm (1993) Solubility of gases in liquids. 19. Solubility of He, Ne, Ar, Kr, Xe, N,, 02, CH,, and SF, in normal 1-alkanols n-C,H,,,,OH (1 5 1 5 11) at 298.15 K. J. Chem. Eng. Data 38, 611-616.

82, 705-709.

105-111.

Solvent Dependence of the Fluorescence Lifetimes of Xanthene Dyes

Documents