J . Phys. Chem. 1990, 94, 4173-4182
4173
Forster Energy Transfer in Two-Dimensional Amphiphile Arrays at the Air/Water Interface Robert S. Urquhart, Robert A. Hall, Peter J . Thistlethwaite,* and Franz Grieser Department of Physical Chemistry, University of Melbourne, Parkville, 3052 Australia (Receiced: November 28, 1989)
Forster energy transfer between several donor and acceptor pairs has been examined in insoluble monolayers at the air/water interface. For the monolayer system consisting of N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)dipalmitoyl-~-a-phosphatidylethanolamine (donor molecule), N-( (lissamine rhodamine B)sulfonyl)dipalmitoyl-L-a-phosphatidylethanolamine,triethylammonium salt (RhB-DPPE) (acceptor molecule), in a dioleoyl-L-a-phosphatidylcholine(DOPC) matrix, the critical energy-transfer distance, Ro, was found to be 5.2 nm, at an average area per molecule of 1.15 nm2, and 4.7 nm, at an average area per molecule of 0.81 nm2. Both results are in good agreement with the theoretical value of 5.1 nm. Insoluble monolayers composed of 4-heptadecyl-7-hydroxycoumarin (donor) and RhB-DPPE (acceptor) diluted with DOPC yielded an Ro value of 4.2 nm, which is also in agreement with the theoretical value of 4.3 nm. However, for the monolayer systems 1,l’-dioctadecyl3,3,3’,3’-tetramethylindocarbocyanineperchlorate (donor) and l,l’-dioctadecyl-3,3,3’,3’-tetramethylindodicarbocyanine perchlorate (acceptor) with either eicosanol or dipalmitoyl-L-a-phosphatidylcholineas diluents, an Ro in the range of 2.3-3.7 nm was measured, which is considerably less than the theoretical value of 5.2 nm. The difference can be explained in terms of the acceptor molecules existing as clusters, with 2-5 monomer units per cluster, the number depending upon the diluent system and average area per molecule. Experiments were also performed to show that, in all three systems studied, quenching of the donors by ground-state complexation or dynamic diffusional processes did not interfere with the Forster energy-transfer mechanism being examined.
Introduction The burgeoning interest in Langmuir-Blodgett (LB) films, and thin organic films in general, arises not only from their potential usefulness in a wide range of optoelectronic devices and applications but also from the lack of fundamental knowledge of their physicochemical pr~perties.l-~In a recent review on molecular monolayers and films1 a range of areas of application of organic thin films was examined. The review committee indicated quite strongly that the current scientific knowledge of the physical and chemical properties of ordered molecular arrays was quite limited. They noted that in order to produce an organic film possessing a desired set of physical and chemical properties, an empirical approach is still needed. Basically, the same overall view was reached by a British review committee concentrating specifically on LB films some 4 years earlier.5 This LB working party pointed out that one of the areas that needed more attention was the monolayer at the air/water interface, the precursor of a LB film. Characterization of air/water monolayers has largely hinged on surface pressure-area ( P A ) isotherms, which are in most cases not unambiguously interpretable. However, recent advances in spectroscopic detection equipment, and the availability of highquality fiber optics, have enabled certain properties of air/water monolayers to be studied by using molecular spectroscopic probes.69 This provides an informative method to complement a-A information for characterizing insoluble monolayers. A case in point, and the topic of this report, is Forster energy transfer (FET) between molecules in a single air/water monolayer. FET is a well-established mechanism of energy transfer between an excited-state donor molecule and a ground-state acceptor molecule.’0 Although significant work has been done on FET between layers of donor and acceptor molecules in LB films,”%’2 far fewer studies have looked at FET in monolayers at the air/ water interface.I3-l5 As will be shown, FET is a highly sensitive method for detecting inhomogeneity of mixing of lipoidal components in multicomponent floating monolayer films. The experimental method to be described later is extremely useful in detecting quite small singlecomponent clusters in the film-cluster sizes which would not be resolved by fluorescence microscopic techniquesI6 or necessarily observed in *-A profiles of the monolayer.
* Author
to whom correspondence should be addressed
Experimental Details l,l’-Dioctadecyl-3,3,3’,3’-tetramethylindocarbocyanineperchlorate (DilC18(3)), l,l’-dioctadecyl-3,3,3’,3’-tetramethylindodicarbocyanine perchlorate (DilC18(5)),N-(7-nitrobenz-2-oxa1,3-diazol-4-yl)dipalmitoyl-~-a-phosphatidylethanolamine (NBD-DPPE), and N-((lissamine rhodamine B)sulfonyl)dipalmitoyl-L-a-phosphatidylethanolamine,triethylammonium salt (RhB-DPPE), were obtained from Molecular Probes. 4-Heptadecyl-7-hydroxycoumarin (HHC) was synthesized by the Institute of Drug Technology (Parkville, Australia) according to the method of Mobius et a1.I’ Eicosanol, dioleoyl-L-a-phosphatidylcholine (DOPC), and dipalmitoyl-L-a-phosphatidylcholine (DPPC) were obtained from Aldrich, Sigma, and Fluka, respectively. All monolayer components were used as supplied with the exception of eicosanol which was recrystallized once from acetone before use. All nonaqueous solvents were spectroscopic grade and were obtained from Ajax Chemicals. The subphase for all monolayer (1) Swalen, J . D.; Allara, D. L.; Andrade, J. D.; Chandross, E. A.; Garoff, S.; Israelachvilli, J.; McCarthy, T. J.; Murray, R.; Pease, F.; Rabolt, J. F.; Wynne, K. J.; Yu, H. Langmuir 1987, 3 , 932. (2) Thin Solid Films 1987, 152 (1/2). (3) Karube, I . Presented a t the Australian Academy of Science, Boden Conference on Membrane Science and Technology, Feb 1987. (4) Mann, J. A,, Jr. Langmuir 1985, 1, 10. (5) Langmuir-Blodgett Films: Current status and prospects for further development, Vol. 1. Report of the Langmuir-Blodgett Working Party, Science & Engineering Research Council, U.K., 1984. (6) Grieser, F.; Thistlethwaite, P. J.; Urquhart, R. S.; Patterson, L. K. J . Phvs. Chem. 1987. 91. 5286. ‘(7) Grieser, F.;’Thistlethwaite, P. J.; Urquhart, R . S. Chem. Phys. Left.
.
1987. . . -141. -. 108. ---
(8) Loughran, T.; Hatlee, M. D.; Patterson, L. K.; Kozak, J. J. J . Chem. Phys. 1980, 7 2 , 5791. (9) Subramanian, R.; Patterson, L. K. J . Phvs. Chem. 1985, 89, 1202. (10) Forster, Th. Discuss. Faraday SOC.1959, 27, 7. (11) Kuhn, H.; Mobius, D.; Bucher, H. In Techniques of Chemistry; Weissberger, A,, Rossiter, W. B., Eds.; Wiley: New York, 1972; Vol. 1, Part IIIB. D 577. (li)Leitner, A,; Lippitsch, M. E.; Draxler, S.; Riegler, M.; Aussenegg, F. R. Thin Solid Films 1985, 132, 5 5 . (13) Tweet, A. G.; Bellamy, W. D.; Gaines Jr., G. L. J . Chem. Phys. 1964, 41, 2068. (14) Agrawal, M. L.; Chauvet, J. P.; Patterson, L. K. J . Phys. Chem. 1985, 89, 2979. (15) Grieser, F.; Thistlethwaite, P.; Triandos, P. Langmuir 1987, 3, 1173. (16) Losche, M.; Mohwald, H. Reo. Sci. Insfrum. 1984, 55, 1968. (17) Mobius, D.; Bucher, H.; Kuhn, H.; Sonderman, J . Ber. Bunsen-Ges. Phys. Chem. 1969, 7 3 , 845.
0022-3654/90/2094-4173$02.50/00 1990 American Chemical Society
4174
The Journal of Physical Chemistry, Vol. 94, No. 10, 1990
experiments was Milli-Q filtered water which has a pH of about 5.6 when in equilibrium with atmospheric C 0 2 . The spreading solvent for all monolayer solutions was chloroform. All fluorescence measurements, the surface pressure measurements associated with them, and the isotherms of H H C , RhB-DPPE, and DOL were performed on a 16.5 cm X 59.7 cm Teflon Langmuir trough which had a barrier rate of 1.5 cm/min. The isotherms for the other monolayer components were performed on a 14.0 cm X 57.0 cm Teflon Langmuir trough which had a barrier rate of 2.9 cm/min. When a monolayer of the same compound was compressed on both troughs for comparison, the surface pressure-area isotherms were very similar. Both Langmuir troughs were contained within metal cabinets, and a laminar-flow cabinet was present in the room to minimize the presence of airborne dust. All experiments were carried out at room temperature, 24 f 4 OC, but the temperature only varied with f 3 “C within a particular series of fluorescence experiments. As FET is a process that is independent of temperature, this temperature range does not affect the results obtained. The temperatures are given in the text. All surface pressure measurements were made by the Wilhelmy plate method, using a 4.38-cm-wide mica plate suspended from a calibrated Shinkoh 2-g weight capacity strain gauge feeding a chart recorder. For both surface pressure and fluorescence measurements, the trough and barrier were cleaned, between experiments, with the solvent used to spread the previous monolayer, and periodically cleaned with hexane, chloroform, methanol, and water to remove any impurities that may have accumulated over time. The mica plate was cleaned in nitric acid and rinsed with Milli-Q water between experiments. The efficacy of these procedures was checked by recording the surface pressure-area isotherm of stearic acid. Donor and acceptor concentrations were calculated from the known mole percent of the compound and the average area per molecule. Fluorescence measurements were made on a Perkin-Elmer LS-5 spectrofluorimeter linked to the Langmuir trough by two silica fiber optic bundles. The use of silica optical fibers instead of glass greatly reduces the secondary emission from the fibers.I5 The fibers were placed 6 mm above the water surface in all fluorescence experiments. This distance was kept constant by connecting a Teflon tip to the assembly which held the optical fibers above the water surface. The fibers were lowered until the tip just touched the water surface. As the fluorescence intensity from a monolayer is small, a blank measurement had to be taken of the water surface for all fluorescence experiments. After addition of the monolayer material to the water surface and compression to the desired area per molecule, the fluorescence from the monolayer system was measured and the blank subtracted. Measurement of the fluorescence properties of the monolayers was performed by two different methods: one gave the fluorescence intensity of the monolayer at a particular wavelength, and the other gave the fluorescence spectrum of the monolayer. In a typical fluorescence intensity measurement, the intensity of the scattered light from the water surface was monitored over a period of time, taking readings every second at the desired wavelengths. Typically, measurements were taken in the region where donor fluorescence could be observed as well as measurements where acceptor fluorescence could be observed. All donor scattered light measurements were taken over 20 min to increase the signal/noise ratio (1200 readings) while acceptor scatter readings were taken over I O min. The monolayer was then spread on the surface at an area per molecule at which the monolayer was in a gaseous state. (For each particular quenching curve this initial area per molecule and the volume of solution delivered on the surface (around 100 pL) were constant.) The solvent was allowed to evaporate for 10 min, and the monolayer was then compressed to the desired area per molecule. The monolayer was then left for I O min in order for it to attain equilibrium. Intensity measurements of the monolayer system were then taken. Donor fluorescence intensities which showed the quenching of donor fluorescence by acceptor molecules were typically monitored
Urquhart et al. over a period of between 1 and 2 h to ensure there were no variations in the fluorescence intensity with time. Variations in fluorescence intensity can be indicative of the monolayers being inhomogeneous. Intensity measurements of acceptor fluorescence were taken, after the donor fluorescence, over a period of 10 min (as there was already evidence for little variation in the fluorescence intensity of the monolayer). The scattered light from the water surface was then subtracted from the fluorescence intensities observed with the monolayer present, to obtain fluorescence intensities of the monolayer itself at the desired wavelengths. Fluorescence and excitation spectra of the monolayers were obtained in a similar way with the exception that the blank and the monolayer spectrum were obtained by scanning the spectrum of both the water surface and the water surface with a monolayer 25 times (to increase the signal/noise ratio), and taking the difference between the average of the spectra with and without the monolayer. For all systems, each pair of donor and acceptor intensities shown in the quenching curves were derived from separate monolayer experiments. Each particular quenching curve was also derived from a separate series of monolayer experiments. The exception to this was the NBD-DPPE/RhB-DPPE/DOPC system where experiments at average areas of 1.15 and 0.8 1 nm2/molecule were performed on the same monolayer. Fluorescence measurements were initially taken at l . 15 nm*/moiecule, and then the monolayer was compressed to 0.81 nm2, where further fluorescence measurements were taken, thus yielding two quenching curves. Corrected excitation and emission spectra were used to obtain theoretical values of Ro (see later). Excitation spectra were corrected by comparing the excitation spectrum of a dilute solution of the acceptor molecule with its absorption spectrum, to produce a correction curve. For use in computation, emission spectra were corrected by using the method and computer program supplied with the Perkin-Elmer LS-5 spectrofluorimeter.’* However, the spectra shown in the figures have not been corrected. Solution absorption spectra were measured on a Hitachi 150-20 spectrophotometer. The quantum yields of H H C and NBD-DPPE in solution were determined with quinine bisulfate in 0.05 M H2S04as reference.lg The quantum yield of DilCl,(3) was determined by using rhodamine B in M HCI in methanol as reference.20
Theory It has been shown by Wolber and Hudson2’ that for a twodimensional array of donor and acceptor molecules, the steadystate fluorescence intensity of donor molecules quenched by acceptor molecules by the Forster mechanism is
where I and Io are the steady-state fluorescence intensities of the donor molecules in the presence and absence of acceptor molecules, respectively, and X is the ratio ? / T where t is time and T is the lifetime of the donor molecules in absence of acceptor molecules. r(2/3) is the gamma function which is listed in standard mathematical tables, and c is the two-dimensional concentration of acceptor molecules. Ro is the critical distance between donor and acceptor molecules where the rate of FET equals the rate of all the other deactivation processes occurring in the donor molecules. Tweet et al.13 have shown that the Forster equation holds for a monolayer system and that Ro (in cm) is given by the expression
( 18) Pecls-II Luminescence Software Condensed Operating Instruciions f o r Models LS-3, 12-4,LS-5; Perkin-Elmer: Oak Brook, IL, 1983; p 5-29. (19) Parker, C. A.; R e s , W. T. Analyst (London) 1960, 5 8 7 . (20) Snare, M. J.; Treloar, F. E.; Ghiggino, K . P.; Thistlethwaite, P. J . J . Photochem. 1982, 18, 335.
The Journal of Physical Chemistry, Vol. 94, No. 10, 1990 4175
Forster Energy Transfer in 2-D Amphiphile Arrays where ( K ~ is ) an orientation factor (assumed to be 2/3 for the systems studied), OD the quantum yield of the donor molecules in absence of acceptor molecules, n the refractive index of the medium surrounding donor and acceptor molecules, and N Avogadro’s number. The quantityfD(ij) is the monolayer donor emission profile subject to the condition
cASOln(ij)is the solution molar absorptivity of the acceptor molecule (in M-’ cm-I), and ij is the wavenumber (in cm-l). Wolber and Hudson2’ have numerically evaluated the integral in eq I and derived an approximate form by fitting the solution to a double-exponential function. Thus, eq 1 can be approximated as I / I o = 0,6463e-4.7497R02c+ 0.3537e-2sIf~1 8R02c
(4)
By fitting the experimental quenching data to eq 4 using a nonlinear least-squares method, one can derive an experimental value of Ro for the monolayer system. This can be compared with the theoretical value of Ro calculated from eq 2. For each quenching curve, the value of Io was determined by taking the average value of donor fluorescence intensity of a number of monolayers with no acceptor molecules being present. In order to calculate the theoretical value of Ro, the quantum yield of the donor molecule was determined in solution as monolayer quantum yields are not readily available. The refractive index of the monolayer system was considered to be that of ethanol as previous studies6 have shown that the effective polarity in a monolayer is similar to that of ethanol. The fluorescence spectrum of the donor molecule could be measured directly and the extinction coefficient spectrum of the monolayer was calculated by using the maximum value of the extinction coefficient in solution as the value for the monolayer at nmaxin the corrected excitation spectrum. This procedure had to be followed as in some cases the excitation spectra of acceptor molecules were different in the monolayer and solution environments. Finally, the integral in eq 2 was calculated by the trapezoidal method. The approximations mentioned above do not have a substantial effect on the theoretical value of Roas Ro has a sixth-root dependence on the terms on the right-hand side of eq 2. When a donor molecule transfers energy to an acceptor molecule, the acceptor may also fluoresce in some cases. If the donor molecule is excited in a region of the spectrum where the acceptor does not absorb, the energy-transferred fluorescence of the acceptor molecule can be measured directly. Mobius and Debuch22 have shown that the ratio of the absolute quantum yield of the acceptor is molecule OA‘ to the quantum yield of the donor molecule given by the expression
where nA is the number of photons emitted by the acceptor molecules per second and flJ,quenched is the number of photons per second that have been lost from the donor molecules due to the presence of acceptor molecules in the system. In terms of the spectral properties of the system eq 5 can be rewritten as
wherefA(ij) is the corrected fluorescence intensity of the acceptor, and fD0(v) andfD(iJ) are the corrected fluorescence intensities of the donor in the absence and presence of acceptor molecules, respectively. (21) Wolber, P. K.; Hudson, B. S. Biophys. J . 1979, 28, 197. (22) Mobius, D.: Debuch, G. Chem. Phys. Lett. 1974, 28, 17.
1
43,
I
I
-=I I\ 30
03
0 4
08
2
If
2c
Area per molecule 1 nm*
Figure 1. Surface pressure-area isotherms of DilC,,(3), DilC,,(S), eicosanol, and DPPC: (a) eicosanol, (b) DPPC, (c) DilCI8(5), (d) DilCI8(3). All isotherms recorded at 21 f 1 OC.
Substituting eq 4 into eq 6 and assuming the presence of acceptor molecules does not change the shape of donor fluorescence gives
Thus
When the above donor and acceptor fluorescence is monitored at a single wavelength instead of the whole spectrum being taken, this expression becomes IA = AIDO(1 - 0.6463e-4.7497R02C- 0,3537e-2.0618R0~c ) (9)
where I A and I D o are the fluorescence intensities of the acceptor molecule and the unquenched fluorescence intensity of the donor molecule at their respective wavelengths, and A is a constant, depending on the quantum yields of donor and acceptor molecules and the wavelengths used in the experiment. A can be determined by dividing the gain in fluorescence intensity of the acceptor molecule by the loss of donor fluorescence intensity for each acceptor concentration. A can be then obtained by averaging these ratios for all the sets of experimental points in a given system. The ratio of quantum yields for donor and acceptor molecules can be found in an analogous way, except that the total corrected wavenumber spectrum of both donor and acceptor needs to be used instead of the intensities at particular wavelengths.
Results Di/c18(3)/Dilc18(5)Energy-Transfer Systems. The a-A isotherms of DilCI8(3),DiICl8(5), DPPC, and eicosanol are shown in Figure 1. As can be seen from the figure, all the isotherms contain either plateaux or inflections which are indicative of either phase transitions or some form of aggregation behavior occurring in the monolayers as they are compressed. The fluorescence and excitation spectra of the donor [DilCI8(3)] and acceptor [Dilc18(5)]diluted individually in a DPPC monolayer at average areas per molecule of 0.795 and 0.455 nm2 are shown in Figure 2, a and b, respectively. As the mole fractions of both donor and acceptor are small, Le., 2% at 0.795 nm2/ molecule and 1.15% at 0.455 nm2/molecule, the surface pressures corresponding to these average areas per molecule are very close
The Journal of Physical Chemistry, Vol. 94, No. 10, 1990
4176
10
-
Lrquhart et al.
,/+
0 4
0 8
0.6
1 0
AIerage area per molecule
1.2 1
1 4
nm'
Figure 3. Surface pressure-area isotherm and fluorescence behavior of a 1.15 mol % DilCl,(3)/99.85 mol 5% DPPC monolayer: (a) surface pressure-area isotherm; (b) fluorescence intensity per unit DilC,,(3) concentration as the monolayer was compressed. A,, = 505 nm, A,, = 565 n m . Temperature = 22 f 1 OC. d
I
TABLE I: Experimental R o Values for the DilC18(3)/DilC18(5) Couple Diluted into a Variety of Monolayer Systems I
diluent eicosanol eicosanol eicosanol DPPC DPPC \+a\elengtli
rim
Figure 2. Excitation and emission spectra of DilCl,(3) and DilCl,(5) both diluted individually into DPPC ([probe] = 0.025 molecule/nm2): (a) DilC18(3)excitation spectrum (A,, = 625 nm), (b) DilC,,(3) emission spectrum (Aex = 490 nm),(c) DiiCl,(5) excitation spectrum (Ae,,, = 720 nm), (d) DilC,,(5) emission spectrum (Aex = 595 nm). Spectra were taken at 23 f 1 "C. (a) Spectra at an average area of 0.455 nm2/ molecule. All maximum spectral intensities are normalized to 50. (b)
Spectra at an average area of 0.795 nm2/molecule. Spectral heights are relative to those seen for the same spectra at an average area of 0.455 nm*/molecule
to that seen in the pure DPPC isotherm, shown in Figure 1. The concentration of donor and acceptor molecules at both average areas per molecule is 0.025 molecule/nm2. As can be readily seen from both figures, there is a large degree of overlap between the fluorescence spectrum of the donor molecule and the excitation spectrum of the acceptor molecule, conditions ideal for FET between these two molecules in a monolayer. The spectra shown in Figure 2 are almost identical with those obtained when DiiCl,(3) and DilCl,(5) are diluted in eicosanol monolayers. The spectra in Figure 2a have all been normalized to maximum intensities of 50. The spectral heights in Figure 2b show the relative intensities of each of the spectra a t an average area of 0.795 nm2/molecule compared with 0.455 nm2/molecule. The concentration of probe molecules is the same in both cases. Thus, DilC18(3)has a larger fluorescence intensity at an average area of 0.455 nm2/molecule than at an average area of 0.795 nm2/ molecule in DPPC. As the short-chain derivative DilC,(3) has been shown to be a useful molecule for probing interfacial viscosity,23 the increase in fluorescence intensity at lower average area per molecule most likely reflects an increase in the "effective viscosity" of the DPPC monolayer as it is compressed. The fluorescence intensity per unit concentration of DilC18(3)versus average area per molecule of a I . 15 mol O/o DiICl8(3)/98.85 mol % DPPC monolayer as it is compressed is shown in Figure 3, along with the x-A isotherm of the monolayer. It can be readily seen that there is a dramatic (23) Grieser, F.; Lay, M.; Thistlethwaite, P. J. J . Phys. Chem. 1985, 89. 2065.
av area per molecule, nm2 0.199 0.246 2.053 0.795 0.455
[DilC,,(3)], molecule/nm2 0.050 0.08 1
0.049 0.025 0.025
exptl R0, nm
2.3 2.6 3.7 2.9
3.1
increase in the fluorescence intensity per unit concentration, at average areas of 1 .OO and 0.53 nm2/molecule. These abrupt jumps in the fluorescence intensity are probably due to abrupt changes in the "effective" viscosity felt by Di1Ct8(3)as the monolayer is compressed. The trend seen in Figure 3 occurred when both DilC18(3) and DilC18(5) were present in the system, the abrupt changes in fluorescence intensity per unit concentration of DilC18(3)occurring at about the same average areas per molecule. However, the actual intensities were lower due to energy transfer between DilC18(3) and DilC18(5). Figures 4 and 5 show the fluorescence intensity of DilC18(3) as DilC18(5)is added in a variety of monolayer systems. As can be seen from the figures, there is a large degree of scatter in the experimental points. The energy-transferred fluorescence intensity of DilCl8(5) is also shown for a number of these systems. As DilC18(5)does not absorb at 505 nm to any significant extent, then any fluorescence from this species, when the monolayer is excited at 505 nm, must be due to energy transfer from donor to acceptor. The acceptor fluorescence intensity should increase as the donor fluorescence is quenched, but as Figures 4 and 5 show, even though the energy-transferred fluorescence intensity of DilC18(5)does generally increase with concentration of DilC18(5) in the monolayer, there is a very large degree of scatter in the experimental points. The experimental values of Ro for the different systems, calculated by fitting the values of I / I o to eq 4, are shown in Table 1. The fits of eq 4 to the experimental data are shown in Figures 4 and 5. The theoretical value of Ro was calculated by using eq 2 assuming @(DilC18(3))= 0.131 and c(Dilc18(5),vmax) = 2.1 15 X IO5 1M-I cm-I. (The values are for the donor and acceptor molecules in ethanol.) Ro was calculated to be 5.2 nm for all the DilC18(3)/DilC18(5) systems studied. NBD-DPPEIRhB-DPPE and HHCIRhB-DPPE Systems in DOPC. The *-A isotherms of NBD-DPPE, HHC, RhB-DPPE, and DOPC are shown in Figure 6. All of these isotherms except NBD-DPPE show no inflections or plateau regions which implies that there are no phase transitions- in these monolayers. The fluorescence spectrum of NBD-DPPE (donor) as well as the fluorescence and excitation spectra of H H C (donor) and
The Journal of Physical Chemistry, Vol. 94, NO. 10, 1990 4177
Forster Energy Transfer in 2-D Amphiphile Arrays 1.2
-
0.8
0
.
6
1
-
,
-
0.6
0.4
0.2
0.0 0
0.02
0.06
0.04
0.06
0.10
-0.2 0
0.02
[acceptor] / molecules nm.*
0.05
0.04
0.08
[acceptor] / molecules nm,'
1.2
1
1.2
1 .o
I
i
(b)
i
-
0.6
0.6 -
0.4
1
n
4i
0.2
0.2
0.0 0
0.02
0.06
0.04
0.08
I
I1
0 O
1."
0.02
0
0.10
0.04
0.06
0.08
[acceptor] i molecules nm *
[acceptor] / molecules nm.'
Figure 5. Fluorescence behavior of a variety of DilC18(3)/DiiC,8(5)/ DPPC systems: (U) relative donor fluorescence intensity (f/I& A,, = 505 nm, A, = 565 nm; (0)relative energy-transferred acceptor fluorescence intensity (I(acceptor)/Io), kX= 505 nm, k,,,= 669 nm; (-)
fit of data to eq 4. (a) Average area per molecule = 0.455 nm2; temperature = 22 f 1 OC. (b) Average area per molecule = 0.795 nm2; temperature = 24 f 1 OC.
0.0 0
0.02
0.01
[acceptor]
0.03
0.04
molecules nm.*
Figure 4. Fluorescence behavior of a variety of DilC,,(3)/DiICl8(5)/ eicosanol systems: (m) relative donor fluorescence intensity (I/fo), A, = 505 nm, A,, = 565 nm; (0)relative energy-transferred acceptor fluorescence intensity (f(acceptor)/f,, where f, is referring to the donor molecule; division by Io allows both donor and acceptor fluorescence to be observed on the same scale), A,, = 505 nm, A,, = 669 nm; (-) fit of data to eq 4. (a) Average area per molecule = 0.199 nm2. (b) Average area per molecule = 0.246 nm2. (c) Average area per molecule = 2.053 nm2.
RhB-DPPE (acceptor) diluted individually into DOPC are shown in Figure 7 . The excitation spectrum of NBD-DPPE was not measured as it is not necessary for the theoretical calculations of Ro. As can be seen from this figure, the fluorescence spectra of both NBD-DPPE and H H C have substantial overlap with the excitation spectrum of RhB-DPPE. Thus, both these donor molecules are suitable to participate in FET to RhB-DPPE. Figure 8 shows the fluorescence intensity of NBD-DPPE as RhB-DPPE is added to the monolayer at average areas of 1.15 and 0.81 nm*/molecule. The experimental values show much less scatter than those observed for the DilC18(3)/DilC18(5 ) systems. There also seems to be slightly different R, values at the two different average areas per molecule. The experimental values
0.0
0.5
1.0
1.5
2.0
2.5
3.0
Area per molecule / nm'
Figure 6. Surface pressure-area isotherms of HHC, NBD-DPPE, DOPC, and RhB-DPPE: (a) HHC, (b) NBD-DPPE, (c) DOPC (d) RhB-DPPE. All isotherms recorded at 25 f 1 OC. TABLE I 1 Experimental R , Values for NBD-DPPE/RhB-DPPE/DOPC and HHC/RhB-DPPE/DOPC Svstems
av area per [donor], exptl svstem molecule, nm2 molecule/nm2 Rn,nm NBD-DPPE/RhB-DPPE 1.15 0.043 5.2 NBD-DPPE/RhB-DPPE 0.8 1 0.062 4.7 HHC/RhB-DPPE 0.808 0.062 4.2 of Ro for the two different average areas per molecule are shown in Table 11. The fitted quenching curves using the R,values given in Table I1 are shown in Figure 8. The theoretical R, value was
The Journal of Physical Chemistry, Vol. 94, No. 10, 1990
41178
-00
300
600
500
700
___
L--
0.1 200
0
0.01
\ ~ a ~ ~ e l e n g r nni h
0 03
0.02
j R h B DPPE]
Figure 7. Excitation and emission spectra of RhB-DPPE and HHC, and emission spectrum of NBD-DPPE all diluted individually into DOPC. All maximum spectral intensities normalized to 50. (a) Excitation (A, = 452 n m ) and (b) emission (Aex = 320 nm) spectra of HHC. [HHC] = 0.062 molecule/nm2, average area per molecule = 0.808 nm2. (c) Emission spectrum of NBD-DPPE (A, = 457 nm). [NBD-DPPE] = 0.062 molecule/nm*, average area per molecule = 0.81 nm2. (d) Excitation (A, = 650 n m ) and (e) emission (A,, = 495 nm) spectra of RhB-DPPE. [RhB-DPPE] = 0.025 molecule/nm2, average area per molecule = 0.808 n m 2 . Spectra recorded at 25 f 2 O C .
&
0 04
0 c j
niolecule, n m
Figure 9. Fluorescence behavior of the HHC/RhB-DPPE/DOPC system at average area per molecule = 0.808 nm2(temperature = 25 f 2 "C): (m) relative donor fluorescence intensity ([/Io), A, = 320 nm, ,A, = 452 nm; (0)relative energy-transferred acceptor fluorescence intensity ( I (acceptor)/210), A,, = 320 nm, ,A, = 590 nm; (-) fit of data to eqs 4 and 9. (a) and (b) are curves given by eqs 4 and 9, respectively.
I
t 0.00
/
,
0.01
0 02
0.03
0 04
[RhB DPPE] / molecules nm.'
"
"Y
6
0 005
0 010
0.015
0 020
0.025
0 030
' Figure 8. Fluorescence behavior of NBD-DPPE/RhB-DPPE/DOPC systems, ( 0 ,W) relative donor fluorescence intensity ([/Io),A,, = 457 nm, ,A, = 525 nm; (0, 0)relative energy-transferred acceptor fluorescence intensity (f(acceptor)/2fo),A, = 457 nm, ,A, = 590 nm; (-) fit of data to eqs 4 and 9. Circles: fluorescence behavior at average area per molecule = 1.15 nm2. Squares: fluorescence behavior at average area per molecule = 0.81 nm2. (a) and (b) are fits of data to eq 4 for 0.81 and l , 1 5 nm2, respectively. (c) and (d) are fits of data to eq 9 for 0.8 1 and I . 15 nm2, respectively. Temperature = 25 f 3 O C . [KhH DPPE] molecules n m
calculated to be 5.1 nm for both areas per molecule assuming that +(NBD-DPPE) = 0.350 (value in ethanol) and t(RhB-DPPE, D,,,) = 80 140 M-l cm-I (value in chloroform). Thus, in this system there is good agreement between experimental and theoretical values of Ro. The energy-transferred acceptor fluorescence intensity is also shown in Figure 8. Even though NBD-DPPE does fluoresce at 590 nm, this contribution to the fluorescence intensity has been subtracted out and only RhB-DPPE fluorescence is shown in the figure. The growth in energy-transferred fluorescence intensity of RhB-DPPE predicted by eq 9 with Ro = 5.2 nm, A = 2.096 and Ro = 4.7 nm, A = 2.033 for average areas of 1.15 and 0.8 1 nm*/molecule, respectively, is also shown in the figure. The values of A were calculated by taking the average of the A values for each of the separate pairs of points. If the calculation of A is carried out with corrected integrated spectra, this leads to +,'/@D = 1.1 f 0.4. Figure 9 shows the fluorescence intensity of H H C as RhBDPPE is added to the DOPC monolayer as well as the intensity of RhB-DPPE energy-transferred fluorescence at an average area of 0.808 nm2/ molecule. The intensity of energy-transferred
Figure 10. Fluorescence intensity of RhB-DPPE versus concentration of RhB-DPPE in a variety of mixed HHC/RhB-DPPE/DOPC systems: (D) RhB-DPPE fluorescence intensity, A, = 550 nm, ,A, = 590 nm; average area per molecule = 0.808 nm2. Temperature = 25 f 2 OC.
fluorescence is obtained by exciting the monolayer at 320 nm and monitoring the fluorescence intensity at 590 nm. The contributions of HHC and directly excited RhB-DPPE to the fluorescence a t 590 nm have been subtracted. The experimental points for energy-transferred fluorescence in Figure 9 are solely due to the fluorescence of RhB-DPPE obtained by quenching H H C fluorescence. The experimental value of Ro for this system is shown in Table 11. The theoretical value of Ro was calculated to be 4.3 nm as= 80 140 suming that @(HHC-) = 0.7 and e(RhB-DPPE,?,,,) M-I cm-I (value in CHC13). Hence, for the HHC/RhBDPPE/DOPC system the theoretical value of Ro agrees with the experimentally determined one within experimental error. As can be seen from Figure 9, there is also very little scatter in the experimental points. The quenching behavior of the system substituting the experimental value Ro in Table I1 into eq 4 can also be seen in Figure 9. The growth in energy-transferred fluorescence intensity of RhB-DPPE predicted by eq 9 with R, = 4.2 nm and A = 0.699 is also shown in the figure. This corresponds to @A'/@D = 0.4 i 0.1 if the entire corrected fluorescence spectra are used. Figure 10 shows the intensity of RhB-DPPE fluorescence versus RhB-DPPE concentration for the HHC/ RhB-DPPE/DOPC system obtained by exciting the monolayer at 630 nm and monitoring the fluorescence at 669 nm. As HHC does not absorb at 630 nm, the intensities given in Figure 10 are due to simple excitation of RhB-DPPE and its resulting fluorescence. As can be seen from the figure, the fluorescence intensity of RhB-DPPE increases almost linearly with concentration, and this suggests that the RhB-DPPE molecules are not
Forster Energy Transfer in 2-D Amphiphile Arrays interacting with each other as concentration quenching24does not seem to be occurring.
Discussion Establishment of Forster Energy Transfer as the Mechanism f o r Quenching. I n order to be certain that FET is occurring in the systems examined, it is necessary to establish that no other energy-transfer processes are occurring. The other major mechanisms that can cause quenching of donor fluorescence are (i) radiative energy transfer (where a donor molecule emits a photon and this photon is absorbed by acceptor molecule), (ii) static quenching (in which donor and acceptor molecules form a ground-state complex which causes quenching to occur), and (iii) dynamic quenching (in which the donor and an acceptor molecule have to diffuse together in order for quenching to occur). For all systems studied, radiative energy transfer can be dismissed largely due to geometric constraints. For radiative energy transfer to occur, a donor molecule has to emit a photon in the plane of the monolayer which is absorbed by an acceptor molecule. Even in the absence of acceptar molecules, photons (emitted in the plane of the monolayer) will not be detected by the fiber optics used to measure the fluorescence intensity. These photons will not be detected in measurements of either I or I,, and so the observed quenching is unlikely to be due to radiative energy transfer in any of the systems studied. As static quenching involves a donor/acceptor complex, it would be expected that the quenching curve would depend on the concentration of donor molecules in the monolayer if this was a major mechanism for energy transfer. It can be seen in Tables I and I 1 that for the Di~C,,(3)/Di~C,,(~) and NBD-DPPE/RhB-DPPE systems the experimental R, values are approximately constant even though the donor concentration is varied considerably. Thus, it seems in these systems static quenching does not make a significant contribution to the observed quenching behavior. At the same time this argues against any role for donor-donor excitation transport by resonant Forster transfer. Even though the HHC/RhB-DPPE experiments were only performed at one donor concentration, it seems unlikely that static quenching is occurring. If static quenching was occurring in this system and every donor molecule was complexed by an acceptor molecule, then 50% quenching of donor fluorescence would occur when the concentration of RhB-DPPE was 0.031 molecule/nm2 (i.e., half the donor concentration). Figure 9 shows that 50% quenching of donor fluorescence occurs at a RhB-DPPE concentration of about 0.01 1 molecule/nm2. Thus, the observed quenching behavior is far more pronounced than that of the most extreme case of static quenching. The other possible mechanism of energy transfer is dynamic quenching. If dynamic quenching occurs between two molecules, then the observed quenching should be dependent upon the lateral diffusion coefficient or the surface viscosity of the monolayer. The surface pressures at which fluorescence was monitored for systems at average areas of 0.795 the DilC,,(3)/DilC,,(5)/DPPC and 0.455 nm2/molecule ranged between 0.1-8 and 24-33 mN/m, respectively, the surface pressure depending on the acceptor concentration in the monolayer. Peters and Beck,2s using fluorescence recovery after photobleaching and the probe 4nitrobenz-2-oxa-1,3-diazole-egg phosphatidylethanolamine, have shown that the lateral diffusion coefficient of DPPC is between 0.020 and 0.025 pm2/s at the higher surface pressures mentioned above, and between 20 and 50 pm2/s at the lower surface pressures. Since the experimental values of R, given in Table I, and the quenching curves given in Figure 4, for both DilC,,(3)/ DilCI8(5)systems, are similar, it implies that the quenching behavior is not dependent on the lateral diffusion coefficient, and hence dynamic quenching cannot account for the observed quenching. Even though there is a large amount of scatter in the curves of Figures 4 and 5 , the change expected in the DPPC case, (24) Tweet, A. G.; Gaines Jr.. G. L.; Bellamy, W. D. J . Chem. Phys. 1964, 49, 2596. (25) Peters, R.; Beck, K. Proc. Natl. Acad. Sci. U.S.A. 1983, 80, 7183.
The Journal of Physical Chemistry, Vol. 94, No. 10, 1990 4179
if quenching was due to a diffusion-controlled reaction, should still be evident, because the diffusion coefficient changes by a factor of 1000 between 0.455 and 0.795 nm2/molecule. The changes in the fluidity as a DPPC monolayer is compressed can also be seen in this study. Figure 2 shows that the fluorescence intensity of DilCl,(3) in DPPC is about twice as large at an average area of 0.455 nm2/molecule than at 0.795 nm*/molecule even though the dye concentration is the same in both experiments. Figure 3 shows that the fluorescence intensity per unit fluorophore concentration of a 1.15 mol % DilC18(3)/98.85 mol % DPPC monolayer increases dramatically at average areas of 1 .OO and 0.54 nm2/molecule. This behavior can be explained when one considers the properties of these types of molecules in different solvent environments. Studies of DilCI8(3) in alcoholic solvents of varying viscosity, and the shorter chain derivative DilC6(3)23in a range of solvents, indicate that the quantum yield of the carbocyanine chromophore increases as the viscosity of the solvent increases. This is due to inhibition of a nonradiative process involving torsion around the carbocyanine linkage. There is also another nonradiative process which is independent of viscosity.23 The extinction coefficient of DilC18(3) was found to be generally insensitive to the viscosity of the solvent. Thus, increases in the fluorescence intensity per unit fluorophore concentration between average areas of 0.795 and 0.455 nm2/molecule in the monolayer system must be due to increases in monolayer viscosity as reported by Peters and Beck.25 However, the growth of fluorescence per unit concentration seen in Figure 3 does not seem to correlate exactly with that observed by Peters and Beck.2s These workers observed a dramatic decrease in the lateral diffusion coefficient (Le., increase in monolayer viscosity) between about 7 and 14 mN/m, whereas in Figure 3, the fluorescence intensity per unit concentration is fairly constant in this range. Peters and Beck25also observed that the monolayer was inhomogeneous in this range, and that the dye molecule they used was more soluble in the fluid than the crystalline domains. If it is assumed in this study both that the monolayer is inhomogeneous in the plateau region of the isotherm and that DilC,,(3) is more soluble in the fluid phase, the discrepancies between the two studies can be explained. Above an average area of 1 .OO nm2/molecule there is equilibrium between gaseous and fluidlike phases, the Di1C18(3) being present in the fluid patches. In fact, if DilClg(3) fluorescence is monitored above 1.00 nm2/molecule over a period of hours, the fluorescence intensity occasionally increases above the small value shown in Figure 3, indicating the presence of large mobile patches. The small fluorescence intensity observed above an average area of 1.00 nm2/molecule in Figure 3 suggests that in this particular experiment no large patches were present under the fibers at these areas per molecule. At an average area of about 1.00 nm2/molecule, there is only a fluid phase and the dye molecules are evenly distributed under the optical fibers. As there are dye molecules under the fibers where previously there were very few, a dramatic increase in fluorescence intensity per unit concentration occurs. The fluorescence intensity per unit concentration stays constant throughout the plateau region of the isotherm. In the plateau region of the isotherm the monolayer consists of patches of crystalline and fluid film which are probably of the order of 10 pm.25 If DilC18(3) is more soluble in the fluid phase, then most of the dye molecules would partition into this phase in the plateau region of the isotherm. The quantum yield would not appear to increase, even though patches of crystalline phase were present. As the patches are much smaller than the area illuminated by the optical fibers, no gross variations in fluorescence intensity with compression would be observed. Hence, the fluorescence intensity per unit concentration would appear to be constant in this region of the isotherm. By about an average area of 0.53 nm2/molecule the monolayer is mostly in a crystalline phase and so the DilC,,(3) starts to partition into this phase. This increases the quantum yield of the
4180
The Journal of Physical Chemistry, Vol. 94, No. 10, 1990
dye molecules and causes an increase in the fluorescence intensity per unit concentration. The fluorescence intensity of DilC,,(5) shown in Figure 2 does not show an increase between average areas of 0.455 and 0.795 nm2/molecule. Solution studies of this probe in alcoholic solvents of different viscosity showed that for DilCls(5) the quantum yield is virtually independent of viscosity. Thus, apparently for this molecule, the viscosity-independent nonradiative rate must be more important than the viscosity-dependent nonradiative rate. N o literature values could be found for either the lateral diffusion coefficient or surface viscosity of eicosanol monolayers to see whether the lateral diffusion coefficient changed in the DilCls(3)/Di~C18(~)/eicosano~ experiments. However, shorter chain alcohols26-28do show variations in surface viscosity with surface pressure. It would be expected that the diffusion coefficients at 0.1 99 (condensed region), 0.246 (expanded region), and 2.053 nm*/molecule (gaseous region) would be markedly different and also different from the diffusion coefficient seen in DPPC monolayers. As the experimental values of Ro in Table I and the quenching curves in Figure 4 are similar for all the DilCl,(3)/DilC,,(5)/eicosanol systems, it again seems unlikely that dynamic quenching is a major contributing factor in the observed quenching results. The two quenching curves for the NBD-DPPE/RhB-DPPE system shown in Figure 8 are also very similar even though the surface pressures recorded while measurements were being taken were between 0.5-6 and 7.3-25.2 m N / m at 1.15 and 0.81 nm2/molecule, respectively. Peters and BeckZShave measured the lateral diffusion coefficient of 4-nitrobenz-2-oxa- 1,3-diazole-egg phosphatidylethanolamine in L-cu-dilaurylphosphatidylcholine (DLPC) monolayers and found that in the range of surface pressures used in the NBD-DPPE/RhB-DPPE system the lateral diffusion coefficient of their probe molecule was between 65-1 10 pm2/s (a = 0.5-6 mN/m) and 30-60 pm2/s (a = 7.3-25.2 m N / m ) . As DLPC has a very similar structure to DOPC and both monolayers have expanded type isotherms, it would be expected that the lateral diffusion coefficients in DLPC and DOPC would be very similar. Thus from Figure 8, changes in the lateral diffusion coefficient of the monolayer do not substantially change the quenching curves, and thus it seems unlikely that dynamic quenching is causing the observed quenching in this system. As quenching experiments for the HHC/RhB-DPPE system were not performed at two different areas per molecule, the above argument excluding dynamic quenching cannot be used for this system. An alternative argument using a relation due to Adam and D e l b r i i ~ kcan ~ ~ be applied. These workers have derived an equation relating the distance a molecule diffuses in a given time, in a two-dimensional system, to its two-dimensional diffusion coefficient. Surface pressures while taking fluorescence measurements for the HHC/RhB-DPPE system ranged from 4 to 11.5 mN/m. Using the data of Peters and BeckZ5for DLPC and assuming that a = 4 mN/m (i.e., the most fluid case studied) gives a diffusion coefficient of about 70 pm2/s. Taking t = 6.6 ns (the lifetime of the anion form of 7-hydroxycoumarin in water3,) and the distance of closest approach between donor and acceptor molecules to be 1.07 nm (derived from the limiting areas per molecule of both H H C and RhB-DPPE and assuming both molecules have circular cross sections), the distance traveled by a RhB-DPPE molecule in one lifetime of a H H C molecule can be calculated. These distances are shown in Table 111. The distance between donor and acceptor molecules was calculated via the equation d = 1/2N'I2
(10)
(26) Fourt, L.; Harkins, W. D. J . Phys. Chem. 1938, 42, 897. (27) Jarvis, N. L. J . Phys. Chem. 1965, 69, 1789. (28) Joly, M . Kolloid Z.1951, 126, 3 5 . (29) Adam, G.; Delbriick, M. Structural Chemistry and Molecular Biology; Rich, A., Davidson, N., Eds.; W. H. Freeman and Co.: San Francisco, 1968; p 198. (30) Chen, R . F. Anal. Lett. 1968, I , 423.
Urquhart et al. TABLE 111: Comparison of the Calculated Distance Traveled during a Lifetime of HHC with the Distance between Donor and Acceptor Molecules [RhB DPPE], calcd distance distance between molecule/nm2 4.96 x 10-3 8.68 X IOF3 0.0124 0.0186 0.0248 0.0372
traveled, nm 0.9 1 .o 1.1 I .3 1.4 1.6
molecules, nm
7.1 5.4 4.5 3.7 3.2 2.6
where d is the average distance between donor and acceptor molecules and N is the number of acceptor molecules per unit area. This equation was derived by using the analysis of Chandrasekhar3' but applying it to a two-dimensional system. As can be seen from Table 111, the calculated distances a RhB-DPPE molecule can diffuse in one lifetime of H H C are much shorter than the actual distances between the molecules. Thus, it seems that dynamic quenching is not the process causing the observed quenching in the HHC/RhB-DPPE system. On the basis of the above arguments, it is reasonable to conclude that FET is the major mechanism responsible for the observed fluorescence quenching in all the systems studied in this work. Comparison of Theoretical and Experimental Ro Values. It can be seen from Table I and the discussion following it that the calculated value of R, (eq 2) and the experimentally determined value of R, (eq 4) for the DilCl8(3)/DilCIs(5)energy-transfer systems, in both eicosanol and DPPC, do not coincide. This discrepancy cannot be simply dismissed by saying that some of the values used in calculating Ro for these systems were taken from bulk solution rather than from a monolayer. As Ro has a sixth-root dependence on the terms on the right-hand of eq 2, even a difference of a factor of 3 in the quantum yield between the monolayer and solution experiments would only change R, by a factor of about 1.2. Similarly, differences in most of the other terms in eq 2 between solution and the values in a monolayer would have little effect on the value of R,. The only term in eq 2 that could possibly be significantly different in a monolayer situation , in solution has a value of 2/3 and which is the value of ( K ~ ) which was used in the calculation of R,. Tweet et al.I3 have shown for a monolayer system in which the transition dipole moments of both donor and acceptor molecules are constrained to make a constant angle, denoted as (3, for the donor molecule and p2 for the acceptor molecule, with respect to the water surface, but are free to rotate through the azimuthal ) be expressed as angle y about a vertical axis, that ( K ~ can ( K ~ = )
sin2 PI sin2 p2 + y4 cos2 PI cos2 P2
(1 1)
However, in order to get the experimental monolayer Ro for the Di~C,,(3)/Di~C18(5) systems to agree with the calculated, it is necessary to have the two angles PI and p2 differ by close to 90'. It seems unlikely that the donor and acceptor molecules would have vastly different orientations at the air/water interface as they are similar in structure. Furthermore, as the theoretical value of Ro calculated by eq 2 coincides with the experimental one for the NBD-DPPE/RhB-DPPE and HHC/RhB-DPPE systems, it seems probable that the approximations made in calculating R, for DilC,,(3)/DilCl,(5) systems are equally valid. Wolber and Hudson,21in their derivation of eq 4, assumed that the donor and acceptor molecules can be infinitely close. They also developed a series of equations that describe the shape of the quenching curve when the molecules must be at least a distance Re from each other. The discrepancy between the theoretical and experimental quenching curves could be due to the finite sizes of both DilC,,(3) and DilCls(5). By using the equations of Wolber and Hudson, it was found that a Re value of between 3.6 and 5.2 nm (depending on the diluent and average area per molecule) was ~
(31) Chandrasekhar, S. Rev. Mod. Phys. 1945, 1 5 , 1.
Forster Energy Transfer in 2-D Amphiphile Arrays necessary in order for the theoretical Ro to describe the experimental quenching curves. However, the distance of closest approach for DilC,,(3) and Dil18(5)was found to be 1.46 nm (derived from the limiting areas per molecule of both molecules in the expanded region of the isotherm and assuming both molecules have circular cross sections), which is much less than that needed for the calculated and experimental curves to coincide. In fact, if an Re of 1.46 nm is substituted into Wolber and Hudson’s equations, there is hardly any change in the quenching curves at all from when Re = 0. Thus, it seems unlikely that the discrepancy between calculated and experimental Ro values can be due to finite molecular size. Wolber and Hudson’s equations show that there is very little change in the predicted quenching curves until R,/Ro > 0.3. As RJRo < 0.3 in all the systems studied, eq 4 seems appropriate for describing these types of systems. The most plausible explanation for the discrepancy between the calculated and experimental values of Ro for the DiIC,,(3)/DilCI8(5)systems is that the monolayers themselves are inhomogeneous; Le., some of the components are not homogeneously distributed in the monolayer but are present as patches which could contain from a few to many molecules. Aggregation of cyanine dyes is ~ell-documented.”,~~ Thus, it is not unreasonable to expect that the donor or acceptor may be present in patches in the monolayer, rather than being homogeneously distributed. This inhomogeneity can be seen in Figures 4 and 5 as there is a large degree of scatter in the donor fluorescence and an even larger degree of scatter in the energy-transferred acceptor fluorescence for the DilC,,(3)/DilCl,(5) systems. Also, the energy-transferred fluorescence, even though it generally does increase as the [DilC,,(5)] is increased, does not show the functional dependence of eq 9 in most of the situations shown in the figure. In some cases it even decreases at large DilC,,(S) concentrations. The experimental points in Figures 4 and 5 were obtained by measuring the fluorescence intensity of the monolayer at a particular wavelength over a period of time (up to 2 h). Thus, if the inhomogeneities in the monolayer were large, it would be expected that the fluorescence intensity would vary with time. For the DilC,,(3)/DilCI,(5) systems the fluorescence intensity with time was fairly constant (within 10%). This suggests that the domains of differing fluorophore density are fairly small compared to the area of the surface illuminated by the optical fibers (1 cm2) and that the scatter is indicative of a different series of small domains or patches of molecules being under the fibers during individual experiments. As FET gives a guide to the relative distances between excited-state donor molecules and acceptor molecules, quenching experiments such as described in this study should show whether or not acceptor molecules are homogeneously distributed in a monolayer. These types of experiments will tell much less about the state of the donor molecules. Single-component Langmuir-Blodgett films often show nonexponential fluorescence d e ~ a y , ~ ~raising - ~ ’ the question of whether such systems are dominated by excitation trapping by nonfluorescent dimers or aggregates. The Forster limit, where the (reduced) concentration of such Forster traps greatly exceeds the (reduced) donor concentration, represents a limiting situation that would provide one reason for nonexponential decay.35 Despite this expectation, it has not proved possible to demonstrate the occurrence of the Forster limit by fitting observed decays to the theoretical Forster expression for a two-dimensional array,35and this has led to more exotic proposals involving fractal distribut i o n ~ . I~f ~the donor molecules in this study were subject to (32) Vaidyanathan, S.; Patterson, L. K.; Mobius, D.; Gruniger, H. R. J . Phys. Chem. 1985, 89, 491. (33) Takami, A.; Mataga, N. J . Phys. Chem. 1987, 91, 618. (34) Alivisatos, A. P.; Arndt, M. F.; Efrima, S.; Waldeck, D. H.; Harris, C . B. J. Chem. Phys. 1987, 86, 6540. (35) Anfinrud, P.; Crackel, R. L.; Struve, W. S. J . Phys. Chem. 1984,88, 5873. (36) Tamai, N.; Yamazaki, T.: Yamazaki, I. J . Phys. Chem. 1987,91,841. (37) Tamai, N.; Yamazaki, T.; Yamazaki, I. Chem. Phys. Lett. 1988,147, 25
The Journal of Physical Chemistry, Vol. 94, No. 10, 1990 4181 quenching by donor aggregates, eq 1 would no longer be valid, and disagreement between experimental and theoretical Ro values would be expected. Earlier time-resolved studies on LangmuirBlodgett film^^^,^' have shown appreciable, though not large, deviations from exponential decay, suggesting that the interference with a steady-state quenching experiment will be small. The question of homogeneity of air-water films is an important one. The above-mentioned data on Langmuir-Blodgett films raise the question of whether aggregation exists in the air-water precursor film, or whether it is a property of the cast film. As only relatively small mole fractions of donor molecules were used in this study, the area that the acceptor molecules were excluded from would be relatively small. If the acceptor molecules were homogeneously distributed, there would be little change in their concentration from that calculated assuming the whole monolayer was homogeneously distributed. The small effect of donor aggregation on the observed quenching behavior in systems of the present type has also been observed by Florsheimer et aL3, These workers, using fluorescence microscope techniques, observed that the donor molecule in their system was present as patches in the monolayer. However, FET quenching experiments, like the ones described in this study, still showed a correlation between theoretical and experimental Ro values. The aggregation of acceptor molecules will have a more marked effect on the experimental value of Ro. Following the method of F o r ~ t e for r ~ ~three-dimensional systems, one can define a critical concentration co such that CO
=
1
r( 2 1
Substituting this expression for co into eq 1 yields
When c = co eq (4) can be evaluated and the value of I / I o when c = co is 0.429. If it is assumed that an acceptor aggregate behaves as though it is a single acceptor molecule and the aggregates themselves are homogeneously distributed, the average number of acceptor molecules per aggregate, N , can be described by using eq 12 from the ratio of calculated and experimental Ro values:
n=co(experiment) co(theorY)
Ro2(theory) RO2(experiment)
(14)
Substituting the experimental values of Ro for the DilC,,(3)/ DilC,,(S) systems given in Table I and using the theoretical value of Ro of 5.2 nm gives acceptor aggregates between two and five molecules depending on the particular diluent system. Tamai et aL40 have studied the quenching of rhodamine 6G and rhodamine B by malachite green in dihexadecyl phosphate vesicles using time-resolved and steady-state fluorescence techniques and have attributed the discrepancies between FET theory and experiment to the presence of fractals. Their steady-state data show a correlation between the quenching curves predicted by Forster theory for I / I o values greater than 0.1, but the theoretical Ro value does not coincide with the experimental one. The presence of fractals could be the reason for the discrepancies between the theoretical and experimental Ro values observed in the DilC,,(3)/DilC,,(S), but time-resolved measurements would be necessary to elucidate this further and provide further insight into the actual structure of the DilC,,(S) aggregates. The simple model given above, however, gives a general insight into the ~~
~
~
~
(38) Florsheimer, M.; Mohwald, H. Thin Solid Films 1988, 159, 115. (39) Forster, Th. Z. Nafurforsch. 1949, 4A, 321. (40) Tamai, N.; Yamazaki, T.; Yamazaki, I.; Mataga, N. In Ulfrafast Phenomena V; Fleming, G. R., Siegman, A. E., Eds.; Springer-Verlag: Berlin, 1986; p 449.
4182
J . Phys. Chem. 1990, 94, 4182-4188
reduction of quenching ability due to the presence of the aggregates. For both the NBD-DPPE/RhB-DPPE and HHC/RhB-DPPE systems the experimental values of Ro given in Table I1 agree favorably with that calculated theoretically via eq 2. The energy-transferred fluorescence intensity also seems to agree favorably with that given by eq 9. It is also important to note that there is far less scatter in these systems than in the DilCls(3)/DilCls(5) systems. All of these points seem to indicate that these monolayers, at least with respect to the acceptor molecules, are homogeneous. The linearity of RhB-DPPE fluorescence shown in Figure 10 also suggests that no self-quenching is occurring between RhB-DPPE molecules at the concentrations used in these experiments. This in turn is further evidence for the homogeneity of RhB-DPPE in DOPC monolayers. The intensity measurements over time needed to calculate the quenching curves for both these systems shown in Figures 8 and 9 were also constant and showed less (about 5%) variation than that of the DilC,,(3)/DilCI,(5) systems. This suggests that if either NBD-DPPE or HHC is aggregated in the systems studied, the aggregates must be much smaller than the area illuminated by the optical fibers. In light of the above discussion, it is interesting to look at the a-A isotherms given in Figures 1 and 6. For the DilC18(3)/ DilC,,(5) systems all of the components, including the diluent molecules, exhibit isotherms with either plateau regions or some type of phase transition. The NBD-DPPE/RhB-DPPE/DOPC and HHC/RhB-DPPE/DOPC systems, on the other hand, with the exception of NBD-DPPE, all show continuous isotherms with no plateaux or inflections. Since at least DilC,,(5) is inhomogeneouslq mixed in eicosanol and DPPC, this suggests that if the isotherms of individual components in a monolayer of more than one component show inflections or plateaux, the components may not mix together. On the other hand, as RhB-DPPE at least seems to be homogeneously mixed in DOPC, then this suggests that multicomponent monolayers, whose components exhibit smooth expanded type isotherms, may mix. This behavior was also ob-
served by Tweet et al.,24who looked at the concentration quenching of chlorophyll a in a variety of monolayer diluents. They found that chlorophyll a (which exhibits an expanded type isotherm) was concentration quenched in stearyl alcohol (a condensed film) to the extent that no fluorescence could be observed from the monolayer. However, when chlorophyll a was diluted in either phytol, oleyl alcohol, or triolein (all of which are expanded films), fluorescence could be observed from chlorophyll a in the monolayer. These workers attributed this behavior to the miscibility of chlorophyll a with diluents which possessed, like itself, expanded isotherms and its immiscibility in monolayers in which the diluents possessed condensed-type isotherms. These statements are in line with our observations described above. Conclusion
In summary, the observed quenching seen in this study can be attributed to FET in all the systems studied. FET also seems to be a method by which aggregation or immiscibility can be detected in monolayer systems, down to an aggregate size which may not be accessible by other methods. This technique is far more sensitive than that of F A isotherms as systems with only a few mole percent of molecules can be tested for homogeneity. As one of the prerequisites for the use of Langmuir-Blodgett films in many industrial applications is that the films must be homogeneous, this method may be a way of determining whether or not the precursor air/water monolayers are homogeneous, and thus provide a monitor by which homogeneous Langmuir-Blodgett films may be attained.
Acknowledgment. R.S.U. acknowledges support from a Commonwealth Postgraduate Research Award. This project was supported by an Australian Research Council grant. Registry No. NBD-DPPE, 92605-64-6; RhB-DPPE, 1261 11-99-7; DOPC, 4235-95-4; H H C , 26038-83-5; DilC,,, 41085-99-8; DPPC, 6389-8; eicosanol, 629-96-9.
Long-Lived Nonmetallic Silver Clusters in Aqueous Solution: A Pulse Radiolysis Study of Their Formation Paul Mulvaney and Arnim Henglein* Hahn- Meitner- Institut Berlin GmbH, Bereich Strahlenchemie, IO00 Berlin 39, FRG (Receiued: December 5, 1989)
Deaerated solutions of AgCIO4 containing (1-5) X IO4 M sodium polyphosphate and 0.1-1 M alcohol are irradiated with single electron pulses or trains of pulses and the intermediates of silver ion reduction detected by optical absorption measurements. Polyphosphate is found to exert a drastic effect on the reduction as small nonmetallic silver clusters are stabilized. In fact, when single pulses are applied, in which only a few percent of the Ag' ions are reduced, the cluster Ag,2+ is the final product which lives for hours. The elementary steps leading to this cluster are investigated. They include ( I ) reduction of Ag' ions by hydrated electrons, (2) complexation of the Ago atoms formed to yield Ag2', and (3) dimerization of the Ag2+complexes. Reaction 1 occurs rather slowly when Ag+ ions adsorbed on polyphosphate chains are involved. Under the experimental conditions chosen, reactions 1 and 2 occur mainly in bulk solution. Reaction 3 is accelerated by a factor of 10 in the presence of polyphosphate. The effect is understood in terms of the reduction in the dimensionality of reaction space as the Ag2+ complexes are attracted by polyphosphate chains. A small fraction of the Ag2+ escapes dimerization by stabilization on the polyphosphate chains. When pulse trains are applied, up to 100% Ag+ reduction can be achieved. The first clusters, i.e., Ag2+ and Ag4*+,are further reduced to yield larger clusters Ag, of nonmetallic silver having absorption bands at 300 ( n = 3) and 330 nm ( n > 3) and in the 345-360-nm range ( n >> 3). A small amount of colloidal metallic silver is also formed.
Introduction
When silver ions are reduced radiolytically in aqueous solution, silver atoms are generated which subsequently agglomerate to yield colloidal silver It has been found recently that small ( I ) Henglein, A . Ber. Bunsen-Ges. Phys. Chem. 1977, 81, 556.
0022-3654/90/2094-4 182$02.50/0
clusters, which do not yet possess the properties of metallic silver, can be stabilized for a while when sodium polyphosphate is present during the r e d u c t i ~ n . These ~ ~ ~ clusters live for a few hours until (2) Tausch-Treml, R.; Henglein, A,; Lilie, J. Ber. Bunsen-Ges. Phys. Chem. 1978, 82, 1335. (3) Henglein. A,; Tausch-Treml. R. J . Colloid Interface Sci. 1981,80, 84.
0 1990 American Chemical Society
