Molecular Fluorescence

7.5 Examples of PCT fluorescent probes for polarity 215

Fig. 7.5. Uncorrected fluorescence spectra of PRODAN in cyclohexane (1), chlorobenzene (2), dimethylformamide (3), ethanol (4) and water (5) (redrawn from Weber and Farris, 1979).

ature is quite confusing in regard to the photoinduced processes occurring in this molecule (see Box 7.2).

Various derivatives of PRODAN have been used. Their formulae are shown in Figure 7.4. DANCA [4-20-(dimethylamino)-60-naphthoylcyclohexanecarboxylic acid] was used to determine the polarity of the myoglobin haem pocket (Macgregor and Weber, 1986). DANCA indeed has a higher a nity for apomyoglobin than PRODAN. It was concluded that the pocket is actually a polar environment, and the polarity can be accounted for by peptide amide dipoles. However, as explained in Box 7.2, specific interactions with NaH groups of the protein binding site may also account for the observations.

Another chemical variant of PRODAN is ACRYLODAN [6-acryloyl-2-(dimethyl- amino)naphthalene], which covalently binds to protein–SH groups.

Replacement of the ethyl group of PRODAN by a C11 para nic chain yields LAURDAN, which is well suited to the study of phospholipid vesicles (Parasassi et al., 1998). Experiments carried out with LAURDAN and PRODAN in bilayers with di erent polar head composition and charge, i.e. di erent phospholipids at pH values between 4 and 10, showed that the spectral shifts do not depend on the polar head residue and on its charge, but only depend on the phase state of the bilayer. Therefore, the presence of a few water molecules in the bilayer at the level of the glycerol backbone is likely to be responsible for the microenvironmental relaxation around the probe. But this relaxation occurs only when the bilayer is in the liquid crystalline phase, and increasing the temperature causes gradual red-shift of the emission spectrum as a result of the increased concentration of water in the bilayer and to the increased molecular mobility. LAURDAN and PRODAN have also been used in studies of polarity in natural membranes, and fluorescence microscopy allows spatial resolution of the polarity microheterogeneity of these natural membranes.

TICT probes can be used as polarity probes (Rettig and Lapouyade, 1994). The classical TICT compound exhibiting dual fluorescence, DMABN (see Section 3.4.4),

216 7 Effect of polarity on fluorescence emission. Polarity probes

Box 7.2 PRODAN – an ideal polarity probe?

The first point to be addressed is the increase in dipole moment, Dm, on excitation of PRODAN (Formula in Figure 7.4). The determinations of Dm reported in the literature, apart from one, are based on solvatochromic shifts analyzed with the Lippert–Mataga equation. In the original paper by Weber and Farrisa) Dm was estimated to be @20 D, but this value was later recognized to be overestimated and recalculation led to a value of 8 Db). Another study yielded a consistent value of 7 Dc). A completely di erent method based on transient dielectric loss measurement provided a somewhat lower value: 4.4–5.0 Dd). From all these results, it can be concluded that the increase in dipole moment on excitation is not responsible for the high sensitivity of PRODAN to solvent polarity.

The second point concerns the nature of the emitting state of PRODAN, and the question arises as to whether the emitting state of PRODAN is or is not a TICT (twisted intramolecular charge transfer) state involving rotation of the dimethylamino group and/or the propanoyl group. From semi-empirical calculations based on various methods, controversial conclusions have been drawn. Nowak et al.e) came to the conclusion that the TICT state could be the lowest state in polar media, whereas Ilich and Prendergastf ) found that specific electrostatic interactions were necessary for lowering the TICT state under the locally excited state. Parusel carried out calculations by several methods: (i) a gas phase ab initio calculation indicated a low-lying TICT stateg); (ii) other calculations suggested that the emission originates from a planar intramolecular charge transfer state and a TICT state resulting from the rotation of the dimethylamino grouph); (iii) the same author concluded that emission arises from a TICT state involving rotation of the propanoyl group for which the rotational barrier is significantly reducedi). These controversial conclusions exemplify the general di culty (as already outlined) in assigning a TICT character to an emitting state.

The moderate increase in dipole moment on excitation (5–7 D) is not in favor of emission from a highly polar TICT state. At any rate, whatever the exact nature of the emitting state, such a modest value of Dm cannot explain the outstanding solvatochromism of PRODAN. Specific interactions such as hydrogen bonding are then likely to play a role. In fact, PRODAN contains hydrogen bond acceptor groups (carbonyl and tertiary amine groups) and care should be taken when interpreting data from solvatochromic shifts of PRODAN in microenvironments containing hydrogen bond donor groupsd).

is of particular interest. The long-wavelength band (corresponding to emission from the TICT state) is red-shifted when the solvent polarity increases, with a concomitant increase in the ratio of the two bands, which o ers the possibility of ratiometric measurements. For instance, DMABN has been used for probing the polarity inside cyclodextrin cavities and micelles. PIPBN and DMANCN exhibiting dual fluorescence can also be used as polarity probes.

Some bichromophoric systems, whose structure is based on the donor–bridge– acceptor principle, can undergo complete charge transfer, i.e. electron transfer. The resulting huge dipole moment in the excited state explains the very high sensitivity to solvent polarity of such molecules. An example is FP (1-phenyl-4-[(4-cyano-1- naphthyl)methylene]piperidine) (Hermant et al., 1990) in which photoinduced electron transfer occurs from the anilino group (donor) to the cyanonaphthalene moiety (acceptor).

The dipole moment in the excited state was estimated (by means of Eqs 7.8 and 7.9) to be 31.8 D. The fluorescence maximum is located at 407 nm in n-hexane and 697 nm in acetonitrile. Unfortunately, protic solvents cause complete quenching; therefore, this family of molecules cannot be used as polarity probes in protic microenvironments.

7.6

E ects of specific interactions

Several examples described above have shown that specific interactions such as hydrogen bonding interactions should be considered as one of the various aspects of polarity. This important point deserves further discussion because hydrogen bonding can lead in some cases to dramatic changes in absorption or fluorescence spectra.

2187 Effect of polarity on fluorescence emission. Polarity probes

7.6.1

E ects of hydrogen bonding on absorption and fluorescence spectra

In the case of n ! p transitions, the electronic density on a heteroatom like nitrogen decreases upon excitation. This results in a decrease in the capability of this heteroatom to form hydrogen bonds. The e ect on absorption should then be similar to that resulting from a decrease in dipole moment upon excitation, and a blue-shift of the absorption spectrum is expected; the higher the strength of hydrogen bonding, the larger the shift. This criterion is convenient for assigning a n–p band. The spectral shift can be used to determine the energy of the hydrogen bond.

It is easy to predict that the fluorescence emitted from a singlet state n–p will be always less sensitive to the ability of the solvent to form hydrogen bonds than absorption. In fact, if n ! p excitation of a heterocycle containing nitrogen (e.g. in solution in methanol) causes hydrogen bond breaking (e.g. N. . .HOCH3), the fluorescence spectrum will only be slightly a ected by the ability of the solvent to form hydrogen bonds because emission arises from an n–p state without hydrogen bonds.

In the case of p ! p transitions, it is often observed that the heteroatom of a heterocyle (e.g. N) is more basic in the excited state than in the ground state. The resulting excited molecule can thus be hydrogen bonded more strongly than the ground state. p ! p fluorescence is thus more sensitive to hydrogen bonding than p ! p absorption.

7.6.2

Examples of the e ects of specific interactions

A remarkable observation illustrating the dramatic e ect of specific interactions is the following. Addition of 0.2% of ethanol in cyclohexane causes a significant change in the fluorescence spectrum of 2-anilinonaphthalene, and for a 3% ethanol content, the spectrum is only slightly di erent from that observed in pure ethanol. Specific interaction via preferential solvation is undoubtedly responsible for these e ects (Brand et al., 1971). This observation shows that a very small amount of a hydrogen bonding solvent may be su cient to significantly change the fluorescence spectrum of a fluorophore, although the macroscopic properties of the solvent (dielectric constant, refractive index) are not significantly a ected.

Another interesting example is 4-aminophthalimide (4-AP) (whose time-resolved spectra during solvent relaxation are described in Section 7.2.1). As a result of the increase in the dipole moment upon excitation of about 4 D, we cannot expect a large shift in the fluorescence spectrum with polarity. In fact, when going from nonpolar diethylether (e ¼ 4:2) to acetonitrile (e ¼ 35:9), a shift of only 33 nm of the fluorescence spectrum is observed, whereas the shift is much larger in hydrogen bond-donating solvents such as alcohols and water (e.g. 93 nm in methanol). Specific interaction is thus undoubtedly responsible for such a dramatic shift (Saroja et al., 1998). The fluorescence quantum yield of 4-AP is high in nonpolar or polar aprotic solvents but very low in protic solvents (0.1 in methanol and 0.01 in water) because hydrogen bonding presumably enhances intersystem crossing. In accordance with these changes in fluorescence quantum yields, the excited-state lifetime is much longer in aprotic solvents (14–15 ns) than in protic solvents (A1 ns in water).

4-AP has been used to probe micellar media (Saroja et al., 1998). The probe is located at the micellar interface and is well suited to monitoring micellar aggregation. In fact, the sharp change in the fluorescence intensity versus surfactant concentration allows the critical micellar concentration (CMC) to be determined. Excellent agreement with the literature values was found for anionic, cationic and nonionic surfactants. The electroneutrality of 4-AP and its small size are distinct advantages over ionic probes like ANS or TNS.

A very marked e ect of specific interactions can also be observed with anthroyl derivatives and in particular with methyl 8-(2-anthroyl)-octanoate. The fluorescence spectrum of this compound in hexane exhibits a clear vibrational structure, whereas in N,N-dimethylformamide and ethanol, the loss of vibrational structure is accompanied by a pronounced red-shift (Figure 7.6) (Pe´rochon et al., 1991).

When Stokes shifts are plotted as a function of the orientation polarizability Df (Lippert’s plot, see Section 7.2.2), solvents are distributed in a rather complex manner. A linear relationship is found only in the case of aprotic solvents of relatively low polarity. The very large Stokes shifts observed in protic solvents (methanol, ethanol, water) are related to their ability to form hydrogen bonds.

The 2-anthroyl fluorophore can be incorporated synthetically in phosphatidylcholine vesicles (anthroyl-PC), which provides an elegant tool for investigating the bilayers of egg phosphatidylcholine vesicles (Pe´rochon et al., 1992).

222 7 Effect of polarity on fluorescence emission. Polarity probes

(0.15 in methanol). In nonpolar solvents, fluorescence is emitted from an n–p state. When the solvent polarity increases, the p–p state that lies slightly above the n–p state is brought below the n–p state by solvent relaxation during the lifetime of the excited state, and thus becomes the emitting state.

The fluorescence maximum of pyrenecarboxaldehyde varies linearly with the solvent dielectric constant over a broad range (10 to 80) (Kalyanasundaran and Thomas, 1977a). This molecule has been used for probing polarity at the micelle– water interface where it is located owing to the presence of polar carbonyl groups. For a given hydrocarbon chain length, the surface polarity depends on the nature of the head group, as expected: it increases on going from anionic to cationic to nonionic micelles. Estimates of the polarity at the micelle–water interfaces were found to be in excellent agreement with z potential values for the micellar Stern layer, derived from the double layer theory.

7.7

Polarity-induced changes in vibronic bands. The Py scale of polarity

In some aromatic molecules that have a high degree of symmetry, i.e. with a minimum D2h symmetry (e.g. benzene, triphenylene, naphthalene, pyrene, coronene), the first singlet absorption (S0 ! S1) may be symmetry forbidden6) and the corresponding oscillator strength is weak. The intensities of the various forbidden vibronic bands are highly sensitive to solvent polarity (Ham e ect). In polar solvents, the intensity of the 0–0 band increases at the expense of the others.

The relative changes in intensity of the vibronic bands in the pyrene fluorescence spectrum has its origin in the extent of vibronic coupling between the weakly allowed first excited state and the strongly allowed second excited state. Dipoleinduced dipole interactions between the solvent and pyrene play a major role. The polarity of the solvent determines the extent to which an induced dipole moment is formed by vibrational distortions of the nuclear coordinates of pyrene (Karpovich and Blanchard, 1995).

The changes in the fluorescence spectrum of pyrene in solvents of di erent polarities (Figure 7.8) show that the polarity of an environment can be estimated by measuring the ratio of the fluorescence intensities of the third and first vibronic bands, II=IIII (Kalyanasundaram and Thomas, 1977b; Dong and Winnik, 1984). This ratio ranges from A0:6 in hydrocarbon media to A2 in dimethylsulfoxide (see Table 7.4). These values provide a polarity scale called the Py scale. When dividing solvents by class (aprotic aliphatics, protic aliphatics, aprotic aromatics), each class gives an excellent correlation between the Py scale and the p scale. The Py scale appears to be relatively insensitive to the hydrogen bonding ability of protic solvents.

6)This rule is not general and, in particular, it does not apply to anthracene and perylene whose symmetry is also D2h.

7.7 Polarity-induced changes in vibronic bands. The Py scale of polarity 223

Fig. 7.8. Fluorescence spectra of pyrene in hexane, n-butanol, methanol and acetonitrile showing the polarity dependence of vibronic band intensities (excitation wavelength 310 nm) (reproduced with permission from Kalyanasundaran and Thomas, 1977b).

Tab. 7.4. Solvent dependence of the ratio II=IIII of the fluorescence intensities of the first and third vibronic bands in the fluorescence spectrum of pyrene.

a) Data from Dong and Winnik, 1984.