Anal. Chem. 1995, 67,4269-4275
Delayed Fluorescence Optical Thermometry Julius C. Fister, 111, Diana Rank, and Joel M. Had$*
Department of Chemistty, University of Utah, Salt Lake City, Utah 841 12
Acridine yellow dissolved in a rigid saccharide glass is proposed as a sensor material for optical thermometry. Following efficient excitation in the visible, triplet states of the dye are produced with a high quantum yield. Activated reverse intersystem crossing from the triplet to the singlet excited state, followed by delayed fluorescence, provides a temperature-dependent decay pathway that competeswith phosphorescence to depopulate the triplet state. Either the triplet-state lifetime or ratio of delayed fluorescence-to-phosphorescenceintensities may be used to monitor temperature. Lifetimes of >lo0 ms are observed at ambient temperatures which require modest instrumentation to measure and process. Since fluorescence and phosphorescence spectra are well separated, their intensity ratio cafl be determined using interference filters. The thermometer performance can be predicted from photophysical models for the temperature dependence of the triplet-state decay. The relative sensitivities of the triplet-state lifetime and of the ratio of delayed fluorescence-to-phosphorescenceintensities to temperature over the range of -50 to 50 "C are 2.0 and 4.5%/ "C,respectively, which are -10 times greater than typical optical thermometers. "he high sensitivities to temperature change result in temperature uncertainties of less than 1 "C over this range.
advantages and limitations. Methods relying on black-body radiation are limited to operation above -500 "C unless fluorescence-based correction techniques are emp10yed.l~Measurement of fluorescence or absorption spectral shifts requires expensive and often delicate dispersive optics. Fluorescence lifetime-based sensors do not require dispersive optics; however, the relative sensitivity of the excited-state lifetime to temperature can be smaller than desired,8 which can limit precision. The excitedstate lifetimes of some rare earth complexes are sensitive to temperature, but the lifetimes are generally short, -70 "C. Reverse intersystem crossing from the triplet to the excited singlet-state manifold occurs from thermally excited vibrational levels of the triplet state which have energies greater or equal to the SI - T1 energy splitting This small fraction of the triplet population undergoes reverse intersystem crossing at a limiting rate, ki,,. Thermal equilibration between vibrational levels of the triplet state is much faster than the rate at which reverse intersystem crossing depopulates the triplet state, kib >> k;,,. Therefore, the fraction of triplet states with sufficient energy for reverse intersystem crossing is at thermal equilibrium and may be described by a Boltzmann di~tribution.~~ This leads to an expression for the effectiveintersystem crossing rate for the triplet population as a whole, krisc:
where k is the Boltzmann constant and Tis the temperature in kelvin. The repopulated singlet states undergo the same decay processes as those formed directly following ground-state excitation. A fraction qhscreturns to the triplet state through forward intersystem crossing. The remaining fraction, &, decays by delayed fluorescence at a rate kf. Reverse intersystem crossing limits the rate at which delayed florescence depopulates the singlet state since the lifetime of the repopulated singlet state is short compared to the time scale of reverse intersystem crossing kf +
kfisc >> krisc kdf = 4&KsC exp(-AE/kT)
(2)
In the absence of bimolecular quenching, competition between direct decay to the ground state (phosphorescence and nonradiative intersystem crossing) at a combined rate kt, and delayed fluorescence determine the observed triplet state decay rate:
Since the rates of phosphorescence and nonradiative intersystem crossing to the ground state are not generally strong functions of temperature above 77 K,27,28variation in the rate of delayed fluorescence determines the temperature dependence of the triplet-state lifetime. EXPERIMENTAL SECTION
Formation of Disaccharide Glass. Approximately 5 g of D-(+)-trehalose dihydrate crystals (Sigma) was ground to a powder with a mortar and pestle. Anhydrous glucose (2.6 g; Mallinckrod) was added to the trehalose to make a 6535 mixture. This quantity produces -5 mL of saccharide glass. A 92 pL aliquot of a 10 mM aqueous solution of acridine yellow (Aldrich) was then stirred into the sugars to produce a slurry of -190 pM acridine yellow. A 1.5 g aliquot of this mixture was added to a small fused-silica test tube. The tube was then suspended in a glycerol bath heated to 140 "C. Heating was continued until the solution no longer boiled, -45 min. A second approach to glass formation utilized an aspirator to reduce pressure over the sample while the saccharide solution was heated to only 115 "C. This produces a more rigid glass while avoiding thermal decomposition of the saccharide host. Acquisition of Emission Lifetimes and Spectra. Fluorescence excitation and signal collection were performed in a 90" geometry. A battery-powered xenon flash (Kvitar, Model 2800) was used for sample excitation; the fwhm of the flash output was -10 ps, which is much shorter than the millisecond liietimes observed in the experiment. The flash was passed through a 450 nm short-pass dichroic filter (Reynard Enterprises, No. 950) and two blue-glass (Schott, BG23 3 mm) filters; the resulting optical energy was -100 mJ cm-2 at the sample in a 75 nm pass band centered at -410 nm. Three orange Schott glass (OG 550 2-3 mm thickness) filters were placed in front of the photomultiplier tube such that only light greater than -550 nm reached the detector. Data acquisition was initiated by triggering a LeCroy 9410 oscilloscope using the output of a photodiode arranged to collect scattered light from the excitation pulse. The temperature of the sample was regulated by immersing the tube for several minutes in slurry baths composed of ice and salts or dry ice and organic solvents having known nominal melting temperature^.^^ The actual bath temperature was monitored with a thermocouple. Lifetime measurements were performed over a sample temperature range of 23 to -70 "C. Ten measurements were averaged at each temperature. Fluorescence lifetimes were determined by nonlinear least-squares fitting of decay transients, performed using a SIMPLEX search algorithm in PC-MATLAB (Mathworks). (27) Tumo,N. J. Modern Molecular Photochemisty: Benjamin/Cummings: Menlo Park, CA, 1978. (28) Grycznski, I.; Kawski, A.: Nowaczyk, K.; Cherek, H. J. Photochem. 1985, 31, 265-272. (29) Pemn, D. D.; Armarego. W. L. F.; Pemn, D. R. Purification o f k b o r a t o y Chemicals, 2nd ed.; Pergamon Press: New York, 1980.
0.0
0.6
1.2
1.8
2.4
3.0
3.6
4.2
4.8
Time (sec) Figure 2. Decay of delayed fluorescence as a function of temperature. Smooth curves are biexponential fits to the data after omitting the first 25 ms from each curve. Aquisition temperatures, in order of decreasing lifetime, are -70, -19, -13, -4, 6, 16, and 23 "C.
Error estimates were determined by replicate trials; for nonlinear parameters, these were compared to uncertaintiespredicted from the least-squares analysis as determined by fitting the error surface to a parabola.30 Fluorescence and phosphorescence spectra were acquired using the same filtered flash excitation source in conjunction with a 0.25 m spectrograph (Chromex, Model 2501s) and TEcooled CCD detector (EG&G PAR). RESULTS
Photophysics of Acridine Yellow in a Saccharide Glass. In order to investigate the temperature-dependent triplet-state photophysics of acridine yellow, delayed fluorescence data from a 190 pM solution of this dye dissolved in a glass composed of a 1:l ratio of trehalose and glucose made without water aspiration (see the Experimental Section) were acquired at temperatures ranging from -70 to 23 "C as shown in Figure 2. The triplet lifetime decreases as the temperature is raised corresponding to the increased rate of delayed fluorescence as predicted by eqs 2 and 3. A shift in the color of the observed emission from orange to yellow-green is also observed since the rate of phosphorescence is independent of temperature above 200 IC Attempts to fit singleexponential functions to these data produced structured residuals. A double-exponential model produced a superior fit to the data based on the lack of structured residuals. Additionally, an F test of the ratio of the summed, squared residuals for the single- and double-exponential fits (F = 9.5 for 936 and 934 degrees of freedom for the -70 "C decay curve) was significant at x-99.99% confidence limit. The data in Figure 2 were fit to a doubleexponential model, and the temperature-dependentlifetimes of both components are listed in Table 1. Since the acridine dye is dissolved in a glassy (amorphous) host, it is not surprising that site-specific decay of the excited states would be observed. Lifetimes of both components decrease by nearly the same relative fraction with increasing temperature, which indicates a similar barrier to activated decay for each component. The amplitudes of the longer lived component were 3.3 (f0.9) times larger than the shorter lived component over the observed temperature range. Since emission from the longer lived com(30) Bevington, P. R. Data Reduction and ErrorAnalysis for the Physical Sciences; McGraw-Hill: New York, 1969.
Analytical Chemistry, Vol. 67, No. 23, December 1, 7995
4271
Table 1. Acridine Yellow Triplet Lifetimes in Saccharide Glass
temp ("C)
long-lived componenta
-70 -19 - 13 -4 6 16 23
1.21 0.82 0.76 0.55 0.47 0.36 0.29
lifetime (s) short-lived componenta 0.26 0.14 0.14 0.06 0.07 0.06 0.05
single-exponential fitb
3
-0.5
-2.0
1.137 0.789 0.732 0.565 0.466 0.354 0.286
E
3
-3.5
-5.0
Lifetimes determined from biexponential fit to decay curves in Figure 3; uncertainty in the reported lifetimes is 2u = 0.012 and 0.01 s for the long and short-lived components, respectively. Lifetimes from a single-exponential fit after omitting a length of data equal to 2.8times the short-lived componentliietime from the beginning of each trace. Uncertainty in the reported lifetimes is 2u = 0.007 s. a
ponent dominates the data in Figure 2 at all temperatures, the decay curves could be fit to a single-exponential model if intensity at early times were neglected. Lifetimes were determined from single-exponentialfits to the data after omitting, from the beginning of each curve, a length of data equal to 2.× the shortlived component lifetime (retaining -90% of the measured transient in each case). The quality of fit of the shortened transients to a single-exponential model is within the noise in the data, and lifetimes from the single-exponential fit are nearly indistinguishable from those of the longer lived component, as listed in Table 1. For thermometry applications, the precision of determining the excited-state lifetime is crucial; from analysis of a series of replicate transients, the lifetimes determined with a single-exponentialfit were found to be more precise (40%smaller standard deviation) than the longer lived component found by fitting the entire transient to a two-exponential model. For this application, therefore, we discard the initial -10% of each transient and fit the remaining data to a simple single-exponential model. To determine whether the change in delayed fluorescence lifetime with temperature follows a simple Arrhenius relationship, eq 3 can be linearized by subtracting the direct decay rate of the triplet state, k,,from both sides and taking the log: ln(kobs- kJ = ln[kris,(l - q5fis,>1 - AE/kT
(4)
A plot of the left side of eq 4 vs 1lkTallows the activation energy of reverse intersystem crossing, AE,and the limiting reverse intersystem crossing rate, ki,,, of the dye to be determined." This analysis requires an estimate of the triplet-state decay rate, kt, in the absence of delayed fluorescence. Since thermally activated delayed fluorescence contributes to the triplet-state decay even at -70 OC, k, was determined by a nonlinear fit of the temperature dependence of the single-exponential lifetimes in Table 1 to eq 3, in which the parameters to be optimized were kt, AE,and the product kKs& (see below). The triplet lifetime in the absence of delayed fluorescence thus estimated was llk, = 1.155 (f0.02) s. Using the inverse of this value to correct the observed decay rate of delayed fluorescence, the results were plotted according to eq 4 in Figure 3, along with the weighted linear least-squares fit. The resulting slope and its standard deviation were AE = 2370 (h186) cm-I, while the intercept was 4272
Analytical Chemistty, Vol. 67, No. 23, December 1, 1995
l/kT x 103 (cm) Figure 3. Arrhenius plot of the delayed fluorescence decay rates vs 1/T. Straight lines are fits of eq 4 to the long-lived lifetimes shown in Figure 1, Error bars represent f l u uncertainties.
z
-Lo
8
e Q
0
12000
10000
8000
U
2P
6000
X
4000
W Y
.r( v1
E
Y
2000
C
3
n
I
I 17
18
19
20
21
Wavenumber x 10-3 (cm-1) Figure 4. Spectra of delayed fluorescence (right) and phosphorescence (left).
12.5 (f1.0) and the correlation coefficient += 0.99. The parameters estimated from this analysis can be related to photophysics of the triplet state of the dye. Arrhenius activation energies for reverse intersystem crossing can be compared with the enthalpy difference between excited triplet and singlet states, estimated from the phosphorescence and delayed fluorescence spectra. Figure 4 shows the delayed fluorescence spectrum acquired at room temperature and the phosphorescence spectrum acquired at -70 "C of acridine yellow in the saccharide glass. Acquisition of both spectra was delayed with respect to the excitation pulse to assure that emission from the long-lived component behavior dominated the observed spectra. Extrapolating the high-frequency edge of each spectrum to the baseline allows the energy of the 0-0 transitions to be estimated. The enthalpy difference between singlet and triplet states thus determined was AHs = 2850 h 70 cm-l. To derive an activation enthalpy from the slope of an Arrhenius plot, one must correct the slope for error introduced by the slight temperature dependence of the preexponential factor.31 Correcting the Arrhenius slope by subtracting the value of kT at the average temperature over which the data were acquired (175 cm-9 results in a activation enthalpy, @ = 2200 f 186 cm-l. The 655 i.200 cm" difference between the spectroscopic and kinetic enthalpies could derive from a large conformational difference between the (31) Connors. K. A. Chemical Kinetics: The Study ofReaction Rates in Solution; VCH Publishers: Kew York. 1990.
excited triplet state and the ground ~ t a t e . ~If~ the B ~ triplet state radiates to a ground state of different geometry, fluorescencephosphorescence measurements can overestimate the true enthalpy difference between the first excited singlet and triplet states.33 For example, Rosenberg and Shombert reported that the spectroscopic singlet-triplet energy difference for acriflavine (3,6 diaminel0-methylacridinium chloride) adsorbed on silica gel was -700 cm-I larger than the activation enthalpy determined from the temperature dependence of delayed fluorescen~e.~~ Information regarding conformational differences between the triplet and singlet states can also be obtained by comparing the rates of forward and reverse intersystem crossing. The limiting rate of reverse intersystem crossing, k&,, can be estimated from the intercept of the Arrhenius plot in Figure 3. Equation 4 indicates that the intercept is the log of the product of the limiting reverse intersystem crossing rate and the fluorescence quantum yield. Using the fluorescence quantum yield of acridine yellow in this mixed saccharide glass, q% = 0.33 f 0.03,26the limiting rate of reverse intersystem crossing is estimated to be k:,, = 8.1 f 3.8 x lo5 s-l. The forward intersystem crossing rate can be estimated from the lifetime of the first excited singlet state and the fluorescence quantum yield. The radiative liietime of the singlet state has not been determined in the saccharide glass. However, Grycznski reported a lifetime of 7.7 ns in rigid poly(vinyl alcohol) film,28which is similar to the lifetime in ethanol, 5.1 nsZ4Taking an average of these values and the fluorescence quantum yield and assuming that the contribution to the decay of the singlet state from internal conversion is small, the forward intersystem crossing rate is estimated to be kfisc x 1 x lo8 s-'. This value is -2 orders of magnitude larger than the limiting rate of reverse intersystem crossing, k;,,. On the basis of the spinstatistics of the two states, one would expect a 3-fold smaller reverse intersystem crossing rate as compared to the forward rate. The larger difference between the forward and reverse rates is consistent with the discrepancy between the activation and spectroscopic enthalpies if the conformational freedom and/or nuclear coordinates of the triplet state at equilibrium are sign% cantly different from the singlet state. Differences in conformationalfreedom between the two excited states would be reflected in the activation entropy of reverse intersystem crossing.33 The activation entropy, A.9, can be determined from the limiting rate of reversed intersystem crossing (determined from the intercept of the Arrhenius plot above) using
where k is the Boltzmann constant.31 The activation entropy thus determined is A.9 = -11.5 f 0.7 cm-l K-I. If spin statistics alone contributed to the entropy term, a value of A 9 = k ln('/3) = -0.75 cm-I K-I would be expected for reverse intersystem crossing. Although entrppic contributions to triplet-state photophysics have not been reported frequently,the entropy change accompanying triplet excitation of several ground-state sensitizers has been estimated from energy transfer reactions. For example, Gessner and Scaiano found that the rate of energy transfer from tripletstate benzophenone to biphenyl was anomalously slower than that (32) Rosenberg, J. L.; Shombert, D. J.J. Am. Chem. SOC.1960,82,3252-3257. (33) Wagner, P. J. j,Am. Chem. SOC.1 9 6 7 , 8 9 , 2820-2825.
- 5
~
0.00.: 12
-
; :
-.-.- - _ . _ . _ . _
: : : ; : : : : :
-;
: ; :
c 0
9 -
-80
-65
-50
-35
-20
-5
10
25
40
Temperature (oC) Figure 5. Temperature response of acridine yellow thermometer. (a) Temperature-dependenttriplet-state lifetimes. Smooth curve is a nonlinear fit of the reciprocal of eq 3 to the data. (b) Temperaturedependent delayed fluorescence-to-phosphorescenceintensity ratio. Smooth curve is nonlinear fit of the reciprocal of eq 8 to the data. Dashed lines in both panels represent f l u temperature uncertainties predicted by eqs 7 and 9, respectively.
predicted from the spectroscopic enthalpy They attributed the slow rate to a loss of conformational freedom accompanying excitation of the biphenyl triplet state and estimated the entropy change upon triplet excitation to be AS = -2.1 cm-I K-l. Due to the rigid excited-state structure, the sign of the entropy change is opposite that predicted by spin statistical contribution^,^^ lending even more weight to the impact of conformational differences on triplet-state photophysics. Delayed FluorescenceThermometry. The photophysics of delayed fluorescence provide a foundation for predicting the performance of an acridine yellow optical thermometer. Figure 5a shows the temperature dependence of the acridine yellow triplet-state lifetimes determined using a single-exponential model for the delayed fluorescence decay curves in Figure 2. As described above, a length of data equivalent to 2.gtimes the shortlived component lifetime (which is equivalent to 0.5-times the longer component lifetime) was deleted from the beginning of each decay curve to allow the use of a simple, single-component model. A key advantage of a delayed fluorescence thermometer is that an Arrhenius model is sufEcient to describe the temperature dependence of its response. The smooth curve in Figure 5a represents the best fit of eq 3 to the data; note that uncertainty in the bath temperature contributes to deviations of the calibration model. The fit of the results in Figure 5a to eq 3 illustrates where the measured lifetime is most sensitive to changes in temperature. More complex models are required to account for the fluorescence lifetime dependence on temperature of materials such as which limits the utility of those models for calibrating a thermometer. Additionally,given a model for the temperature dependence of the excited-state lifetime, the temperature uncertainty due to variance in the measured lifetime, robs, can be predicted. Solving (34) Gessner, F.; Scaiano, J. C. J. Am. Ckem. SOC. 1989, 107, 7206-7207.
Analytical Chemistty, Vol. 67, No. 23, December 1, 1995
4273
eq 3 for the temperature,
and propagating the error in the lifetime yields an expression for the temperature uncertainty, UT:
The lietime uncertainty was determined from single-exponential fits to 11delayed fluorescence decay transients acquired at 0 "C, where the bath temperature was relatively stable. Over a 2 h period at 0 "C, the delayed fluorescence liietime was found to be Uiobs = 3.4 x s. Substituting the lifetime uncertainty into eq 7 predicts the temperature uncertainty over the range of the thermometer, which is plotted in Figure 5a as a dashed line. The uncertainty in temperature, UT, minimizes near -10 "C at a value of 0.25 "C and is better than 1 "C over the range of -50 to 55 "C; this precision compares very favorably with optical thermometers based on other materials as discussed below.l,ll The temperature accuracy was also checked at 0.0 "C by comparing the temperature predicted from the 11 replicate delayed fluorescence curves interpreted with the model developed in Figure 5a. Note that the calibration model was developed on the day prior to its application in the precision experiment. The sample temperature determined from the 11 replicate runs averaged 0.16 "C with a standard deviation UT = 0.22 "C; the model predicts the temperature of the bath with no detectable error beyond the uncertainty in the bath temperature. Since delayed fluorescence competes with phosphorescence for depopulation of the triplet state, the ratio of delayed fluorescenceto-phosphorescence intensities, Z , I ~ ~ ~ / Z also ~ I , ~exhibits ~, Arrhenius behavior:
In order to investigate delayed fluorescence thermometry based on an intensity ratio approach, time- and wavelength-resolved delayed fluorescence spectra of a 25 pm acridine yellow in a rigid glass composed of a 3:l ratio of trehalose to glucose were acquired using CCD detection at 21.3, 0.3, -9, and -79 "C. Integration times were 10 ms at 21.3 and 0.3 "C, 15 ms at -9 "C, and 25 ms at -79 "C. The triplet lietime, reverse intersystem crossing rate, and activation energy estimated from the temperature dependence of the delayed fluorescence decay rate of the long-lived component in this sample were l/kt = 2.018 f 0.05 s, k,: = 7.2 f 3.0 x lo5 s-l, and hE = 2446 f 233 cm-'. The reverse intersystem crossing rate and activation energy are indistinguishable from those discussed above. The longer triplet liietime in this sample is consistent with a more rigid glass prepared under evacuation by aspiration as described in the Experimental Section. In order to ensure that delayed fluorescence from the long-lived component dominated the observed emission, acquisition of the spectra was initiated at a time equal to 2 . 8 ~the short-lived component lifetime following the excitation pulse. Prior to calculating the ratio, the 4274
Analytical Chemistry, Vol. 67, No. 23, December 7, 7995
phosphorescence intensity must be corrected for overlap with the red tail of the fluorescence emission. The contribution of fluorescence (which maximizes at -500 nm) to the phosphorescence intensity at 580 nm can be estimated from the prompt fluorescence spectrum in ethanolz4as -12.5% of the fluorescence maximum. The ratios of corrected intensities at 500 and 580 nm were determined from eight spectra at each temperature to be 4.053 f 0.028, 1.605 f 0.024, 0.953 f 0.024, and 0.0 f 0.002 at 21.3, 0.3, -9, and -79 "C, respectively. In order to generate a calibration curve, the corrected intensity ratios were fit to eq 8 by using the triplet lietime and activation energy determined from the decay kinetics and optimizing the preexponential, q5k&,,kt-', since the quality of fit was sensitive to this parameter. The reverse intersystem crossing rate determined from the best fit was koisc = 1.0 x lo6 ssl, which is indistinguishable from the value determined from analysis of the decay rates. The intensity ratios and best fit calibration curve are shown in Figure 5b. The temperature uncertainty from this method can be estimated by propagation of errors from eq 8:
The uncertainty in the ratio of delayed fluorescence-to-phosphorescence intensities arose from a constant contribution from detector dark current and read noise and a proportional component equal to -1% of the signal intensity. Based on this model and eq 9, the predicted performance of an intensity ratio-based delayed fluorescence thermometer is comparable to that achieved by measuring the delayed fluorescence lifetime; compare the dashed lines in panels a and b of Figure 5. Even better precision could be achieved by integrating the emission intensities for > 15 ms. An advantage of the intensity ratio-based thermometer would be more straightforward electronic processing based on calculating a direct ratio of the two signals. Emission maxima are well resolved (Figure 4) so that the intensity ratio could be determined using a single interference filter. DISCUSSION The excellent temperature precision of delayed-fluorescence thermometry derives from a large relative change in the rate of delayed fluorescence with respect to temperature. The average relative sensitivities of the triplet-state lifetime and intensity ratio to temperature over the range -50 to 50 "C of 2.0%and 4.5%/"C are -10 times larger than the relative sensitivities of either neodymium glass or Additionally, the relative sensitivity to temperature is comparable to the sensitivity of the fluorescence lifetime of a rare earth complex BaClF-Sm2+ over the same range, -1.5%/"C4 An additional advantage of the delayed fluorescence thermometer, however, is the modest excitation source and detection electronics required to excite and process the > 100 ms lietimes of the dye compared to the 100 ps liietimes of the rare earth complex. Rigid hosts, more robust at higher temperatures than the saccharide matrix described here, can be produced from leadtin-fluorophosphate g l a ~ s e s .Arrhenius ~ ~ . ~ ~ behavior of long-lived ~~~
~~
~~~
~
~
(35) Tick, P. A,; Hall. D. W. Difis. Defect Data 1987.53-54, 179-188. (36) Tompkin, W. R.; Boyd. R. W.: Hall, D. W.; Tick, P. A. J. Opt SOC.Am. B 1987,4,1030-1034.
acridine yellow triplet states has been demonstrated in glasses made from these materials35 Such hosts would extend the temperature range of optical thermometers made from organic dyes to both higher and lower temperature regimes. Furthermore, dyes solvated in the lead-tin-fluorophosphate glasses appear very stable with respect to photodegradation. Tick and Hall35reported that fluorescence from acridine yellow, acridine orange, or rhodamine 6G solvated in lead-tin-fluorophosphate glasses showed no indication of photobleaching after exposure to illumination at -1 W cm-2. Fluorescence from rhodamine 6G photobleached only slowly with a time constant of I 103 min when exposed to 5 x 105W cm-2 laser radiation at 568 nm. The xenon flash used in the current work delivers 10 kW cm-2 in a Gaussianlike pulse of 10 ps fwhm. Given the long excited-state lifetimes of the acridine yellow triplet state, the same excited-state population and subsequent delayed fluorescence intensity could be achieved using a blue-green source delivering 1W cm-2 in a 100 ms pulse; delivering this power density into a 100 ym optical fiber
would require 100 pW of power. Furthermore, an excitation source better matched to the ground-stateabsorption of dye than the flash/filter combination used here would require even lower power density for the same excitation efficiency. -ACKNOWLEWMENT This work was supported in part by National Science Foundation under Grants CHEW10319 and CHE95-10312. Fellowship support for J.C.F. was provided by the ACS Division of Analytical Chemistry and Dupont. Summer support for D.R was provided by a National Science Foundation REU Grant CHE-9300381 and the University of Utah Department of Chemistry. Received for review June 5 , 1995. Accepted September 18, 1995.@ AC9505435 @
Abstract published in Adoance ACS Abstracts, November 1, 1995
Analytical Chemistry, Vol. 67, No. 23, December 1 , 1995
4275
