Picosecond fluorescence lifetime measurements on dyes adsorbed at

Jan 6, 1983 - and D. K. Negus. Department of Chemistry, University of Pennsylvanle, phllad8lph&, Pennsylvanle 19104 (Received: August 9, 1982;. In Fin...
0 downloads 0 Views 378KB Size
The Journal of

Physical Chemistry

0 Copyright, 1983, by the American Chemical Society

VOLUME 87, NUMBER 1

JANUARY 6,1983

LETTERS Picosecond Fluorescence Lifetime Measurements on Dyes Adsorbed at Semiconductor and Insulator Surfaces Y. Llang, A. M. Ponie Ooncalves; Depatmmnt of Chemistty, Tempk University, Phlladelphle. Pennsyivanle 19 122

and D. K. Negus Department of Chemistry, University of Pennsylvanle, phllad8lph&, Pennsylvanle 19104 (Received: August 9, 1982; In Finel F m : October 1, 1982)

Fluorescence lifetimes in the 50-60-ps range were measured for rhodamine B and eosin adsorbed on tin oxide and indium oxide surfaces. Somewhat shorter lifetimes were obtained for rhodamine B adsorbed on plain glass. Implications these results have for models of the dye-sensitizedphotoinjection into semiconductorsare discussed.

Introduction In recent years there has been much renewed interest in the photocurrent of n-type semiconductor electrodes sensitized by dye molecules adsorbed on the surface.l The dyes sensitize stable semiconductors with large band gaps to the absorption of visible light, and thus make them potentially useful in solar energy conversionq2 The reported overall quantum efficiencies, $ = ((number of electrons detected)/ (number of photons absorbed)1, span almost two orders of magnitude.L4 Obstacles to high 6 values may either compete with the injection of an electron into the semiconductor (e.g., “concentration quenching” of the dye excited state by energy transfer) or, after in(1)H. Gerischer and F. Willig, Top. Current Chem., 61,31 (1976). (2)A. Heller, Acc. Chem. Res., 14,154 (1981). (3)M. Spitler and M. Calvin, J. Chem. Phys., 67, 5193 (1977). (4)W. Arden and P. Fromhen, Ber. Bumenges. Phys. Chem., 82,868 (1978). 0022-3654/83/2087-0001$01.50/0

jection, compete with its escape into the bulk of the semiconductor (e.g., either recombination or trapping in surface states).’ While much important information about $ has been gathered over the years for dye-sensitized semiconductor/electrolyte cells,ltb8 a few recent spectroscopic investigation^^-^*"'^ have placed the competing (5)K.Hauffe, H.J. Danzmann, H. Pusch, J. Range, and H. Volz, J . Electrochem. SOC.,117,993(1970). (6)R. Memming, Photochem. Photobiol., 16,325 (1972). (7) T. Miyasaka, T. Watanabe, A. Fujishima, and K. Honda, J . Am. Chem. SOC.,100,6657(1978). (8) M. Mataumura, Y. Nomura, and H. Tsubomura, Bull. Chem. SOC. Jpn., 50,2633 (1977). (9)M.Spitler and M. Calvin, J . Chem. Phys., 66,4294 (1977). (10)M.Spitler, M. Lubke, and H. Gerischer, Chem. Phys. Lett., 56, ,577 - . .(1978). ~

(11)T.Yamase, H. Gerischer, M. Lubke, and B. Pettinger, Ber. Bunsenges. Phys. Chem., 83,658 (1979). (12)M. Spitler, M.Lubke, and H. Gerischer, Ber. Bumenges. Phys. Chem., 83,663 (1979).

0 1983 American Chemical Soclety

2

The Journal of Physical Chemistry, Vol. 87, No. 1, 1983

Letters

processes into much sharper focus. This work has raised intriguing questions concerning, in particular, the mechanisms responsible for the decay of the excited states of dyes on semiconductor surfaces. This Letter describes preliminary measurements of the fluorescence lifetime of dyes (rhodamine B and eosin) adsorbed on semiconductor (tin oxide and indium oxide) and insulator (glass) surfaces. The experimental procedure used picosecond laser excitation and streak camera detection which, to our knowledge, have not been previously applied to this problem.

Experimental Section Two types of semiconductor thin films were used: tindoped indium oxide (resistance < 250 ohm/square, coated on both sides of 0.5 mm thick glass substrates) was obtained from Optical Coating Laboratory, Inc.; antimonydoped tin oxide (resistance < 100 ohmlsquare, coated on one side of 1mm thick glass substrates) was obtained from Deutsche Uhrglassfabrik GMBH. Rhodamine B and eosin obtained from Aldrich were first M aqueous recrystallized and then used to make 4 X solutions. To prepare a dye monolayer the surface to be coated was first wetted with the dye solution for several seconds and then wiped dry by dragging a strip of lens tissue over it. The outcome of this procedure was monitored through the absorption spectrum of the dye on the surface, taken with a Hewlett-Packard 8450 spectrophotometer. The spectra were very similar to those recorded in dilute solution, but red-shifted (peak a t -560 nm for rhodamine B on tin oxide). The shape of the spectra of rhodamine B monolayers was just as reported by Spitler and Calvin13indicating in all cases that most of the dye molecules were on the surface as monomers, although some aggregation cannot be ruled out. While it is hard to estimate reliably the surface coverage, 6, the absorbances of the rhodamine B samples ued in the picosecond experiments were all identical within 10% (both on the semiconductors and on glass). Based on the measured absorbances, and by comparison with the work of Spitler and Calvin,3we estimate 6 0.5 for all rhodamine B samples. The laserlstreak camera apparatus previously described15was modified for in-line detection of fluorescence and operated basically as follows. Second harmonic (530 nm) pulses from an amplified mode-locked Nd:glass laser system were used for excitation. The pulse energy was always less than 1mJ. The fluorescence was detected by a GEAR Pic0 V streak camera triggered by first harmonic pulses. The exciting pulses were directed 30" away from the axis of the streak camera, and the photocathode was further protected with 560-nm cut-on color and interference filters. The streak camera output was digitized by an optical multichannel analyzer (Princeton Applied Research 1215116154)which was also used for data correction and simulation. Because of the low dye absorbance on one surface four to ten surfaces were stacked together for each experiment. The sample stack (overall thickness 2 mm or less) was placed against the entrance slit to the streak camera. Through the same slit passed weak 530-nm etalon pulses scattered by a diffuser plate, which were used to calibrate the sweep rate of the streak camera and to monitor the exciting pulse energy. The apparent width of the laser pulse profile detected with this arrangement was 17 ps (fwhm), a direct consequence of the comparatively large slit width required by the weak fluorescence.

-

(13)W.Arden and P. Fromherz, J.Electrochem. Soc., 127,370(1980). (14)W.Arden and P. Fromherz, J.Am. Chem. Soc., 102,6211 (1980). (15)Y.Liang, D. K. Negus, R. M. Hochstrasser, M. Gunner, and P. L. Dutton, Chem. Phys. Lett., 84,236 (1981).

-

Time

Figure 1. Fluorescence decay of rhodamine B on tin oxide (solid line), obtained with a single laser shot. The dashed line is the convolution of the apparent laser pulse profile with a 56-ps exponential decay.

TABLE I: Fluorescence Lifetimes (ps) of R h o d a m i n e B and Eosin o n Different Surfacesu sur face tin oxide indium oxide glass

rhodamine B

eosin

5 3 . 5 k 1.0 55.5 L 0 . 7 56.2 i 0.8 54.5 1.7 45.8 1: 0.8 47.3 i 0.9

5 1 . 3 2 2.0 6 1 . 6 k 1.6 6 7 . 6 t 1.5

+_

T h e errors given are one standard deviation.

Because excitation and detection were almost in-line the sample thickness had only a small (but detectable) effect on the time resolution. For each experiment on dye-coated surfaces a similar set of surfaces without the dye was also run to check for possible background interference.

Res u1t s Figure 1 shows a fluorescence decay (solid line) for rhodamine B on tin oxide, obtained with a single laser shot. The shape of the fluorescence decay did not change when the laser pulse energy was attenuated by a factor of ten. The dashed line is the convolution of the apparent laser pulse profile with a 56-ps exponential decay. The somewhat slower rise of the experimental curve is a consequence of the loss in time resolution due to the sample thickness. Because this small effect arises only while the exciting pulse is still on (and thus affects only the apparent onset of the fluorescence), no effort was made to correct for it. A "best" fit to the experimental decay was derived by taking only the central 140-ps portion of the fluorescence decay: A linear least-squares fit to the log of the fluorescenceintensity yielded a fluorescence lifetime of 56.2 f 0.8 ps for this decay. (The error given is one standard deviation.) This result and those obtained with other samples of rhodamine B on tin oxide and on indium oxide, as well as with eosin on tin oxide, are listed in Table I. To put these values in perspective we may recall that the fluorescence lifetimes of rhodamine B and eosin in dilute aqueous solution are 1.5 and 1.4 ns, re~pectively.'~~'~ Since the fluorescence lifetimes for dyes on semiconductors were found to be much shorter than in solution, it was important to determine whether the reduction should be attributed to injection. Therefore, we examined the fluorescence of rhodamine B on plain glass and on synthetic quartz (Suprasil) at surface coverages,judged by (16)V. J. Koester and R. M. Dowben, Reu. Sci. Instrum., 49,1186 (1978). (17)G. R. Fleming, A. W. E. Knight, J. M. Morris, R. J. S. Morrison, and G.W.Robinson, J . Am. Chem. Soc., 99,1766 (1977).

The Journal of Physical Chemistry, Vol. 87, No. 1, 1983

Letters

2 0 ps CII

Time

-

Flgure 2. Fluorescence decay of rhodamine B on glass (solid line), obtained with a single laser shot. The dashed line is the convolution of the apparent laser pulse profile with a 46.5-ps exponential decay.

the dye absorbance, very close to those on the semiconductors. Figure 2 shows a fluorescence decay (solid line) for rhodamine B on plain glass, obtained with a single laser shot. A least-squares fit, as in the preceding paragraph, gave a fluorescence lifetime of 45.8 f 0.8 ps for this decay. This value and that obtained with another sample of rhodamine B on glass are shown in Table I. The average was used in Figure 2 to convolute the apparent laser pulse profile with a single exponential decay (dashed line). Experiments with rhodamine B on Suprasil led to lifetimes which were also somewhat shorter than on the semiconductors.

Discussion With the exception of one eosin value, the data in Table I do not show much scatter for a given dye/surface combination. Therefore, we are confident that the fluorescence lifetime is not, contrary to what we had anticipated, shorter on the semiconductors than on the insulators. Since the dyes cannot inject into the latter we conclude that a common process, other than injection, is responsible for the short lifetimes we have measured. Any injection taking place from the fluorescent state into the semiconductors is obscured by this other fluorescence quenching process and must therefore be slower than -50 ps. (Our results do not, of course, rule out injection into the semiconductors from some other state or species.) The short lifetimes on semiconductors as well as on glass are most likely due to energy transfer followed by trapping at defect sites, perhaps aggregates of some sort. Given the tight molecular packing on the surface for 8 0.5 this process could easily dominate the excited state decay. Recent measurements for rhodamine B on silica have shownls that the fluorescence is severely quenched at high surface coverages, presumably because of energy transfer. A dominant role for energy transfer was also suggested recently by Nakashima et al.,19who performed experiments similar to ours on rhodamine B adsorbed a t molecular crystal surfaces. Slightly shorter lifetimes were found on anthracene (35 f 7 ps) than on naphthalene and phenanthrene (-50 ps). Since excited state rhodamine B can inject holes only into anthracene, the difference was attributed to injection. Because hole injection into anthracene has nearly unity efficiencyz0 (in spite of energy

-

(18) S. Garoff, R. B. Stephens, C. D. Hanson, and G. K. Sorensen, J. Lumin., 24, 773 (1981). (19) N. Nakashima, K. Yoshihara, and F. Willig, J. Chem. Phys., 73, 3553 (1980).

3

transfer) the authors suggested that nonfluorescent dimers may inject efficiently. The elegant work of Arden and Fromherz4J3J4has yielded the best overall efficienty ($ = 0.8) of all dye/ semiconductor systems reported to date. Monolayers of long-chain cyanine dyes were preassembled a t water-air interfaces and then transferred to indium oxide surfaces. The relatively high fluorescence quantum yield (0.2) in these monolayers13was reduced fivefold by contact with the semiconductor, presumably because of i n j e ~ t i o n . ~ Thus, while concentration quenching of the fluorescence without the semiconductor is not negligible, it is much less severe than that encountered in our experiments. We note that low 4 values (50.03) have been reported for systems such as ours, in which the dye is adsorbed directly from s o l ~ t i o n . ~While J ~ the low 4 might result from processes which follow injection, our results suggest that efficient energy transfer quenching is likely to be the dominant factor. The importance of energy transfer quenching is well documented for preassembled chlorophyll monolayers placed on and its potential relevance to dyesensitized photoinjection has been disc~ssed.~ We feel that the key to the success of the Arden and Fromherz systems is simply that energy transfer does not severely quench the dye excited state before injection can occur. However, it is not clear why fluorescence quenching is only limited in spite of the relatively high 8 in their monolayers. This is in sharp contrast to what happens in preassembled chlorophyll monolayers, for which the fluorescence is quenched by nearly two orders of magnitude at similar 0 values.21 Although the high molecular organization (J-aggregates) of appropriately assembled dye monolayers suppresses fluorescence q u e n ~ h i n g this , ~ ~is? apparently ~~ not the case in the work of Arden and Fromherz. Finally, it should be pointed out that comparison of the absorption and photocurrent action spectra shown in ref 4 suggests some involvement of dimers in the injection step. Because of the relative difficulty in using picosecond techniques to study monolayers, this method cannot routinely replace fluorescence quantum yield measurements. However, we feel that it provides an important and much less incumbered look at the decay dynamics of the dye excited states. This is particularly true here since the fluorescence quantum yield is low because of the fast radiationless decay of the excited state. In addition, the significance of low fluorescence quantum yields is rendered somewhat uncertain by the danger that dyes at a relatively few sites not subject to energy transfer might dominate the fluorescence quantum yield. In order to prove that energy transfer is indeed the cause of the short fluorescence lifetimes obtained in our work, it will be necessary to repeat these experiments at progressively lower 8. Only in the limit of negligible energy transfer will it be possible to obtain injection rate constants from the fluorescence lifetimes.

Acknowledgment. This research was supported by the Office of Basic Energy Sciences, U S . Department of Energy, under Contract No. DE-AC02-81ER10881. The pi~~~

(20) B. Nickel, Mol. Cryst. Liq. C r y s t . , 18, 227 (1972). (21) A. G. Tweet, G. L. Gaines, Jr., and W. D. Bellamy, J. Chem. Phys., 40, 2596 (1964); 41, 2068 (1964). (22) R. S. Knox in “Bioenergetics of Photosynthesis”, Govindjee, Ed., Academic Press, New York, 1975, p 183. (23) G. R. Seely in ‘Primary Processes of Photosynthesis”, J. Barber, Ed., Academic Press, New York, 1977, p 1. (24) D. Mobius, Ber. Bunsenges. Phys. Chem., 82, 848 (1978). (25) H. Kuhn in “Light-Induced Charge Separation in Biology and

Chemistry”, H. Gerischer and J. J. Katz, Ed., Verlag Chemie, New York, 1979, p 151.

4

The Journal of Physical Chemistry, Vol. 87, No. 1, 1983

cosecond experiments were performed at the Regional Laser Laboratory of the University of Pennsylvania, a facility supported by the National Science Foundation. D. K. N. was supported by the National Science Foundation under Grant CHE-8000016. The authors thank Professor

Letters

R. M. Hochstrasser for making the picosecond apparatus available for this work. Registry No. Rhodamine B, 81-88-9; eosin, 17372-87-1; tin oxide, 1332-29-2; indium oxide, 12672-71-8.