403
J. Phys. Chcm. 1993,97,403413
Anomalously High Efficiencies for Electronic Energy Transfer from Saturated to Aromatic Hydrocarbons at Low Aromatic Concentrations Yi-Mlag Wang,t David B. Johaston, and Sanford Lipsky' Department of Chemistry, University of Minnesota, Minneapolis, Minnesota 55455 Received: August I I , 1992; Zn Final Form October 26, 1992
The absolute efficiency of electronic energy transfer from cis-decalin excited at 161 nm to 2,5-diphenyloxazole (PPO) has been measured over a PPO concentration range from 1.0 X 1W2to 2.0 X 1W5M via measurements of both the cis-decalin and the PPO fluorescence. At concentrations above ca. lW3 M,the normal fluorescing state of cis-decalin plays the dominant role in the energy transfer. At lower concentrations, however, there appears to be an important contribution from some other nonfluorescing state of cis-decalin. The fraction of PPO fluorescence generated by this "dark" state rises from ca. 10% at 0.01 M to ca. 70% at 2 X lW5M. The effects of addition of 0 2 , of dilution with isooctane, and of cooling to-35 OC on the quantum yield of this process are reported. The results obtained here confirm earlier results with other saturated hydrocarbon donor aromatic acceptor systems that have suggested the existence of a "dark" donor state that dominates the transfer process at low acceptor concentrations via some anomalously efficient mechanism. For the system cis-decalin PPO at 21 OC,the transfer probability for this process at the lowest concentration studied of 2 X lW5 M is 2.5 X lW3per photon absorbed and 0.060 per "dark" state produced.
+
+
l C 5to M and at energies ecxc extending from the absorption onset of D to ca. 3-4 eV higher.'" However, under the same conditions and with the same A's but replacing D by a saturated hydrocarbon (e.g., cyclohexane, heptane, bicyclohexyl), sienificant deviations from eq 1have been reported in the direction to increase ZA above its value predicted from the quenching of the donor fluorescenceand increasingly so as c declines below ca. c = 10-2 M.3-7 At higher concentrations, the deviation does not appear to be particularly noticeable.3-10 If we represent this departure from eq 1 by an "excess" function,f(c), i.e.,
I. IntrodPCtion In studies of electronic energy transfer in liquids from a donor solvent, D, to an acceptor solute, A, it is usually observed that the intensity of acceptor fluorescence, ZA, is simply proportional to the decrease in intensity of donor fluorescence; i.e.,
=rND (1) where y is some constant independent of acceptor concentration, C.1.2
Equation 1 is expected to be always valid whenever the acceptor extracts its energy, CI, exclusively from the fluorescent state of thedonor, D1, regardless of whether this transfer occursvia binary A D1 collisionsand/or by energy migration within the D system. Thii follows quite generally from the following considerations. We define @A and @D to be the quantum yields of A and D fluorescence in the mixture per photon of energy, zac, that is absorbed and generates the state DI and @OA and @OD to be the "intrinsic" fluorescencequantum yields, respectively, of acceptor (i.e., when excited directly at energy 61) and of donor (is., also when excited directly at energy €1 but in the absence of acceptor). Accordingly, so long as DI is exclusively the precursor of the fluorescent state of A (Le., AI), the ratios @A/QPOA and @D/@OD can be simply interpreted as the fractions of excited donor that either do or do not transfer their energy to the acceptor and, therefore, must have a sum equal to unity; is.,
(3)
+
A simple rearrangement of eq 2 transforms it into eq 1. Indeed, in the not uncommon case that @OA is independent of z (at least over the range of z that exceeds to G ~ ~it )is, not difficult to see that eq 2 and, perforce, eq 1 would be generally valid even were not only DI but also any of its precursor excited states capable of transferring their energy to A. Equation 1 has been verified for a variety of systems involving D's and A's which are both aromatic (e& D = benzene, toluene, or pxylene and A = pterphenyl or diphenyloxazole) and over a wide concentration range of A that extends from at least c = Prsrsnt addrau: Radiation Laboratory. Universityof NotreDame,Notre Dame,IN 46556.
0022-3654/58/2097-0403$04.00/0
thenflc) appears to be well represented by the two-parameter form:
flc) =
U 1 + uc
(4)
with magnitudes of u and u (depending on the system studied) ranging from u ,=t 102 to lo3 and u ,=t lo3 to l(r M-I.5-' This form off(c) has suggested that its origin is based on the existence of another state of D that is nonfluorming but also capable of sensitizing the fluorescenceof A. However, from the magnitudes of u and u, it has been argued that unless this "dark" state has an abnormally large cross section for energy transfer to A, its lifetime must be at least several hundred nanoseconds-but no evidence for so long-lived a state has yet been reported. Since these very early investigations, there has been no further work on this problem, although the existence of an anomalously high sensitization efficiency has recently been invoked to explain, at least qualitatively, some very peculiar effects of energy transfer involving saturated hydrocarbon donors-but in much more complex, three-component systems? The present investigation was undertaken firstly to c o n f m the earlier work. To this end, we have used a more sensitive optical system and a measurement of the absolute quantum yield ratio, @A/@"A, rather than a parameter proportional to this. In this way, we are able to more directly study the anomaly via its effect to cause discrepancy from the predictions of eq 2 rather than eq 1. Secondly,we have extended the early work to examine (6
1993 American Chemical Society
Wang et al.
404 The Journal of Physfcal Chemistry, Vol. 97, No. 2, 1993
l'O
nr----I
200
150
250
300
WAVELENGTH, nm Figure 1. Intensity spectrumof exciting light as determinedwith sodium
salicylate. how the anomalous sensitization is affected by cooling to -35 OC, by diluting with "inert" solvent, and by quenching both D1 and AI with molecular oxygen. Finally, we also report here on negative results of a search to find evidence for a long-lived component in the time evolution of the intensity of acceptor fluorescence using two-photon pulsed laser excitation.
II. ExperiumtdSection The apparatus used for the quantum yield experiments has been previously describcd9J0and consists essentially of a Spex F2 12 spectrophotometer suitably modified to permit excitation and emission analysis in the vacuum ultraviolet. Samples are illuminated at normal incidence with a 30-W Hamamatsu L879 D&mp(fittedwithaMgF2whdow) throughaSpex 1680doublegrating monochromator (operating at a band-pass of 1.8 nm), and the emission is collected in front face geometry through a second, similar monochromator (operating at a band-pass of 3.6 nm). The sample cell is a Pyrex cylinder, 0.6 cm in length, to which is attached (via Varian Torr Seal epoxy) a front l-mmthick LiF window of "active" diameter 1.1 cm and a similar, back, Suprasil quartz window. Samples were deaerated by purging with dry nitrogen and transferred to the fluorescence cell in a nitrogen-flushed drybox. For measurementsof OAand OD,excitation was at 161nm [which is the peak of our lamp spectrum at the position of the cell as determined with a sodium salicylate quantum counter (see Figure l)]. The intensities of the emission spectra of both donor and acceptor fluorescence were measured for each sample over a time interval during which lamp drift was less than 0.1%. Backgrounds for acceptor fluorescence were determined from scans of the acceptor emission spectral region with the neat (or diluted) solvent excited at 161 nm and the background for donor fluorescencedetermined from scans of the donor's spectral region with the acceptor excited directly at 161 nm in a solution of a nonfluorescent and nonsensitizing solvent, usually isooctane (fluorescence yield less than ca. lb6).I1 Corrections were also applied to accommodate contributions to A1 fluorescencein the A D system from direct absorption by A of 161-nm light. This correction (which was generally very small at low c) was made by measuring the emission of A in isooctane for Lc= 161 nm. Although this correction requires adjustment for the ca. 20% smaller absorption coefficient at 161 nm of isooctane vis-a-vis cis-dccalin (sea section IIIC), such adjustment was not applied in view of its relative uncertainty and its very small effect on the pertinent region of concentrations studied here (i.e., c S l e 3M). In any event, such adjustment would only have increased (albeit slightly) our energy-transfer efficiencies. To obtain the quantum yield ratio, @A/@A, measurements were required both at Lc= 161 nm (for @A) and at 225 nm (for @A). The ratio of absorbed light intensities, Jx, at these two
+
wavelengths was determined in two stages. First, the emission from neat cis-decalin at 231 nm was used as a quantum counter10 for the ratio fl6l/flSo, and then the emission at 400 nm from a saturated solution of 2,5-diphenyloxazole (PPO) in cyclohexane was used as a quantum counter12 for the ratio PB0/.P25.The emission at 410 nm from a deposit of sodium salicylate on a LiF window was found toagreeclosely with the PPOquantumcounter and was often substituted for it. The 161-to-180-nm ratio in J typically varied from experimentto experimentover a range from ca.4.8 to 5.1 depending on the quality of the cell window and on the extent to which the exciting monochromator and sample compartment were free of molecular oxygen. The 180-to-225nm ratio was typically 1.50 but also showed some small variation (ca.3%) on different days. A series of measurements on cis-decalin and PPO were performed at low temperatures. The sample cell was inserted into a copper block that was maintained at -35 2 OC with an Omega temperature controller that adjusted the rate of flow of cooled Nz through the block. Time scans made on the emission intensities of cis-dccalin and PPO (at the highest PPO concentration of 10-2 M)as the sample was temperature-cycled from 21 to -35 OC and back to 21 O C gave no evidence for any precipitation of PPO. Unless otherwise specified, all other measurements were made at 21 i 2 OC. For the time evolution experiments, liquid cyclohexane containiigpxylene as acceptor was excited in a two-photon transition using either the 308-nm output pulse of an XeCl excimer laser (Lumonia Hyper-Ex-400) or the 337-nm output of a nitrogen laser (NRG -0.7-5-200). In both cases, the laser was focusad to a point ca. 1-2 mm from the front window of a quartz cell, and both cyclohexane and pxylene emissions were collected at 90° and analyzed with either a 0.3-m monochromator (218 GCA/ McPherson) for the excimer pulse or a 0.5-m monochromator (Bausch & Lomb) for the Nz pulse. A chemical filter consisting of an aqueous solution of 2,7-dimethyl-3,6-diamyclohepta-1,6dieneper~hlorate~~ was additionally used to remove most of the scattered laser light. The dispersed fluorescence was detected with a photomultiplier (Hamamatsu R955), and the output of this (with 50 0 termination) was viewed by a Tekronix 2430A digital oscilliscope triggered by a pin diode that viewed a small fractim of the excitation pulse. Direct absorption of the laser pulse by pxylene was negligible as evidenced by the absence of any pxylene emission when cyclohexane was replaced with isooctane. Measurements were made at rep rates of 10 Hz for times of the order of several hundred seconds. Measurements made over longer time intenals showed no variation that could be attributed to the buildup of photochemical products. The Nz laser system was also used in some "steady-state" experimeats by operating at a higher rep rate and replacing the scope with a Keithley 602 electrometer. cis-Dccalin (Aldrich 99%), cyclohexane (Malhckrodt, spectrophotometric grade), and isooctane (American Burdick & Jackson, high purity) were additionally purified by percolation through activated silica gel. Some cls-decalin samples were also purified by shaking with HzS04/HNO3 (1:3), washing with alkaline water, and then drying over Na2S04without any effect on the observed emission quantum yield ratios. pxylene (Aldrich, 99%) was sometimes additionally purified by slow cooling to the freezing point and discarding ca.20% of the melt. This procedure, however, had no observable effect on our results. PPO (Aldrich, scintillation grade) was used without further purification.
*
III. RmdC A. Neat CibIkerlie + PPO. The system most extensively studied in this investigation was D = cfs-decalinand A PPO. Nine solutions were examined spanning a PPO concentration range from 2.00 X 10-5to 1.00 X 10-2M. Three sets of solutions were separately studied over a period of several months and the
The Journal of Physfcal ChemiJtry, Vol. 97, No. 2, 1993
High Efficiencies for Energy Transfer 1.o
1 .o
c
>
k
Ln Z
v,
z w
c
z
W
5-
0.5
0.5
m
m
DL Q
€K
a 0
280
180
380
1.o
480
WAVELENGTH, nm Fig" 2 Emission spectra (uncorrected for spectral response of the analyzer) of cisdecalin PPO excited at 161 nm, at PPO concentrations of2 x 1 r 5 M (A), 5 x icr5 M (B), 1 x io" M (c), 2 x io" M (D), 5 X lo" M (E). and 1 X lO-' M (F).
+
z v, Z
W
-
0.5
m
cv,
€K
a
z W
n 180
I-
z
280
380
480
WAVELENGTH, nm Figure 4. Emission spectra (uncorrected for spectral response of the analyzer) of (A) 1W2M solutionsof PPO in cls-decalin (1) and illooctane (2) and (8) lo" M solutions of PPO in cis-decalin (1) and isooctane (2) at & = 161 nm.
m
€K
a
WAVELENGTH, nm Figare 3. Emission spectra (uncorrected for spectral response of the analyzer) of cisdecalin + PPO excited at 161 nm, at PPO concentrations of 1 X lo-' M (A), 5 X lO-' M (B), and 1 X M (C).
resultsaveraged. The standard deviation of a singlemeasurement was ca. 1.7%. Figure 2 exhibits some typical scans over the emission spectral regions of both cis-decalin and PPO for excitation at 161 nm. It will be noted that over the concentration range exhibited here, from 2 X l W to 1 X lo-' M,there is a strong development in theemissionintensityof PPOwithvirtuallynoconcomitantchange in that from cis-decalin. However, for higher concentrations, from 1 X llP3 to 1 X 1W2M (see Figure 3), the cfs-decalin emission intensity does, noticeably, decline as the PPO emission intensity continues to develop. A comparison of the emission spectra of 1 X 10-4 and 1 X 1fY2M PPO solutions in isooctane and in cis-decalin for b,,, = 161 nm is shown in Figure 4. In isooctane, there is no donor emission and the PPO intensity is strictly proportional to PPO concentration. The observed emission intensities (usually measured at the maxima of the emission spectra and counted for sufficiently long times to make statistical counting errors negligible)were corrccted by subtracting from them both a background and a contribution from the direct absorption by the PPO of 161-nm light [as determined from measurements in the isooctane solution (see section II)]. The magnitudes of these corrections varied with PPO concentration. At the smallest concentration of 2 X M,the background correction had its highest value relative to the PPO signal of ca. 7% (dropping to ca. 0.07% at 1 X 1C2 M), whereas the correction for direct PPO absorption had its lowest value relative to the PPO signal at this concentration, of a.1% (rising to ca. 6% at 1 X 1W2M). For the cis-decalin emission intensity, only background corrections were required, and these always remained less than 0.5%.
As will be developed later, a useful parameter for discussion of the results is .
I
+D@"A
The use of this ratio avoids all refractive index and emission geometry corrections so long as the penetration depth of the exciting light is essentially the same for the determination of @A and @OA. The quantum yields in eq 5 are connected to the measured fluorescence intensities, Z, via
z=N @ ~ J ~ (6) where N is an instrumental parameter sensitive to the emission geometry, the spectral response of the analyzer, the refractive index of the liquid, etc.; 0 is the internal conversion efficiency which connects the definition of 0 in eq 2 as a yield per D1 formed (for @A, @D, and @D) or per A1 formed (for @A) to a yield per photon absorbed at Lc= 161 nm (for @A, @D, and @D) or at bxc = 225 nm (for @OA); and fiat is the intensity of absorbed exciting light. Accordingly, since @A and OD are both determined at the same kc= 161 nm, (7)
where / 3 ~is now the conversion efficiency from A' to AI, where ASisthestateofA that is populated bytheD-teAenergytransfcr. For the ratio @A/@D, which is determined at different excitation wavelengths (i.e., Lc= 225 and 161 nm),
where / 3 ~(or /PA)is the internal conversion efficiency for the transition from the donor's (or acceptor's) initial state at L =
Wang et al.
406 The Journal of Physical Chemistry, Vol. 97, No.2, 1993
+,
TABLE I: Total Energy-Transfer Parameter, and Excess Transfer Parameter, 9 Q, for c b D e a b + PPO at 21 O C and Le = 161 11111
-
c. M
9'
*-eb-
2 x 10-5
0.003 53 0.008 40 0.01s 3 0.026 7 0.0448 0.052 7 0.086 0 0.298 0.585
0.002 47 0.005 75 0.0100 0.016 1 0.023 6 0.026 2 0.033 0 0.033 0.055
5 x lo-'
1x10-4 2x104 4x104 5x104 1 x lo-'
s x 10-3
1 x 10-2
air saturation. c,
2X 5X 1X 2X 4X 5X 1X SX
M
lW5 l0-'
104 10-4
104 104 lO-' lW3
1 X 1W2
161 nm (or 225 nm) to the donor's (or acceptor's) fluorescent state, DI(or AI). M The measurement of PA,was performed on a 1.0 X solution, which at bc= 225 nm has sufficient optical density to make negligible any difference in emission geometry between this solution and those used for the determination of IA at bXc = 161 nm. Accordingly, we can equate NAand P Ain eqs 7 and 8 and substitute both of these equations into eq 5 to obtain (9)
ID
+-
11111
Using K = 0.0857. Using Q = aw with U D = 53 M-I.
IA \k=-K
TABLE Ik TOW Eaergy-Transfer Parameter, 9, md Excess Trrprfer Parameter, Q, for &Dccrlin + PPO lo Ah-md oXygeU-&tW8tcd soldm 8t 21 'c and & 161
*b
9-9
9 d
9-0
0.002 96 0.00703 0.0128 0.023 4 0.038 6 0.0455 0.075 4 0.269 0.527
0.002 04 0.00473 0.00820 0.0142 0.0202 0.0225 0.0294 0.039 0.052
0.001 Elf 0.00477 0.00892 0.017 1 0.0306 0.0353 0.059 7 0.223 0.429
0.001 11f 0.00302 0.00542 0.010 1 0.016 6 0.017 8 0.024 7 0.048 0.079
footnote 14. Using K = 0.0778. e Using Q = a@ with a ~ =, 46 M-l for c = 1 X IF2M. Ushg K = 0.0657. *UsingQ = a@ with U D = 35 M-I. /Thew entried m a y be in mor by ca. 10% [see discussion in text and footnote 221. a Sce
M-I for c 5 5 X lW3M and 47.5
1.5c
B n
U
\ o
where
02 saturation.
1.2:
n
U
For PPO,BA, BA,O 1,I2 and for cis-decalin, BD = 1lo SO that K was simply determined with the term in parentheses in eq 10 set equal to unity. For the measurements described in this section, Kwas found to have thevalue 0.0857 f 0.0008.Table I presents values of 9 at the PPO concentrations listed. Their significance will be discussed in section IV. In our subsequent analysis, we will also require the limiting value, as c 0, of the Stern-Volmer quenching constant, QD, defined as
-
Unfortunately, for c I M, the ratio @OD/@D was too close to unity to permit any reliable evaluation of the difference. Accordingly, the value of CYDthat we report here of 53 f 2 M-' was determined as the average of measurements made at our two highestmcentrationsof5X l W a n d l X 10-2M. Inthefollowing section, a more careful study of CYDis reported with results that confirm that the value of 53 M-I used here is acceptably close to the limiting value. B. c&Deelllin + PPO + 02 (21 O C ) . The system cis-decalin + PPO was also studied in 1 atm air- and oxygen-saturated sol~tions.1~J5 Table I1 lists values of 9 for both sets of solutions. Although each solution was measured only once, our uncertainty in 9 is estimated not to exfeed ca. 2% in any solution except the one of lowest PPO intensity. For this solution (oxygen-saturated and with c = 2 X l(r5M), the background correction was ca. 30% of the PPO intensity, and accordingly, we estimate a possible error in Q of ca. 10%. The quenching of cis-decalin by PPO was carefully determined for air-saturated solutionsover a range of concentrationsextending from 5.00 X 1 V to 1.00 X 10-2 M. The results, exhibited in Figure SA, show a limiting slope as c 0 of 45.7 f 1.4 M-' with a slight upward concavity at the higher concentrations. The average value of QD for c between 5 X 10-3 and 1 X 10-2 M is
-
1S
O
A
/
i
0
/
/ 1
/
/
H
\
./
* /
r
1.25
o n H
in the absence of PPO (PD)to that in the presence of PPo (ID) a6 a function of PPO concentration: (A) at 21 OC in aerated solution, (B)at -35 OC in nitrogenated solution. The solid line ie a linear least-squares fit-ta-th~ pintrtc I5-x !e3 M-.
47.5 f 1.5 M-I. For oxygen-saturated solutions, CYDwas only determined from measurements above 5 X lO-' M to be 35 f 2 M-1 . Finally, we also report here 0 2 quenching factors for neat cisdecalin fluorescence (excited, at 161 nm, in the absence of PPO) of 1.13 and 1.52 for 1 atm air and oxygen saturation, respectively, and for PPO fluorescence (excited directly at 225 nm in neat cis-decalin) of 1.03 and 1.16. C. c&Deelllin + PPO + Isooctane. A series of measurements were made in which cis-decalin was diluted to 3% and 10% by volume with isooctane. In the absence of PPO,the ratios of the intensityofcisdecalinfluoresccncefromthe 3%and lO%solutions to that from neat cis-decalin were measured to be, respectively, 0.0459 and 0.149 for LXc 161 nm. Assuming that the CISdecalin fluorescence quantum yield is the same in the 3% and 10% solutions, we estimate from this that the ratio of the optical
-
The Journal of Physical Chemistry, Vol. 97, No. 2, 1993 4Q7
High Efficiencies for Energy Transfer
+,
TABLE IIk Total Energy-Tder Parameter, md Excess T d e r Parmeter, Q, for 3%8od 10% &Decalin in Irrooct.ae PPO at 21 "C and Lc= 161 MI
++
t
tv, Z
5 x-10-5 8 X l0-'
1x104 2x104 2.5 X 104 4x104
5x104 1
x 10-3
5 x 10-3 1x
10-2
0.008 38 0.011 7 0.0137 0.0222 0.023 5 0.031 0 0.0349 0.053 6 0.11 0.096
0.0141 0.0209 0.0252 0.0452 0.0522 0.0770 0.0924 0.1686 0.684 1.25
0.0138 0.0211 0.0256 0.0450 d 0.0749 0.0961 0.1555 0.661 1.27
0.008 10 0.011 9 0.014 1 0.0220 d 0.028 9 0.038 6
W + z
0.5
t
0.0405
0.086 0.12
Volume 46 cis-decalin in isooctane. Using K = 0.0885. Using Q with (llD 5 115 M-I. This solution was not studied for the 10% dilution.
WAVELENGTH, nm
@
5
UDC
TABLE Iv: Total Ekgy-Trmrler Parameter, 9, md Excess T d e r Puameter, e Q, for Air-hhnrted 3% &DedninIsooct.ae PPOrt 21 o C m d L . = 161 MI c, M *a *-Qb
+
-
__
~
5 x 10-5 8 X l0-'
1x104 2x104 2.5 X 104 4x104 5Xle 1 x 10-3 5 X 1V3 1 x 10-2 a
0.008 12 0.0129 0.0157 0.0298 0.0344 0.0494 0.0578 0.106 0.427 0.786
Using K = 0.0596. Using Q = a@ with a D
0.004 47 0.007 08 0.008 37 0.0152 0.016 2 0.0202 0.021 3 0.0329 0.062 0.056 5
73 M-I.
absorptivity of neat isooctane to that of cis-decalin in isooctane is ca. 0.78 at 161 nm and that the fluorescence quantum yield of cis-decalin in isooctane is 12% higher than in the neat liquid. Using this 12% correction, we obtained a value for K,averaged over all 3% and 10% solutions, of 0.0885 f 0.001. Values of 9 calculated with this K are presented in Table 111for both the 3% and 10% solutions. The 3% measurements were performed on three separate sets of solutions over a period of several months. For the entries presented in Table 111, the standard deviation of a single measurement for c < 10-3M was ca. 5%. The large error here is due mainly to the significantly lower emission intensities from these solutions and the larger relative corrections for background and PPO direct absorptions that this mandated. For M, background and direct absorption example, at 5 X corrections in one set of experiments were ca. 16% and 12% of the total signal, respectively, and changed to 2% and 28% at 1 M. For the 10% X 10-3 M and to 0.3% and 48% at 1 X solutions, these corrections were typically reduced to 7% and 5% M, and 0.1% and 20% at 5 X lW5 M, 0.7% and 10% at 1 X at 1 X M. The Stem-Volmer constant, U D (see eq 1l), was found to be the same for 3% and 10% solutions, and an average over all and 1 X M gave solutions of concentrations of c = 5 X a value of U D = 115 f 3 M-l. The 3% solutions were also studied at 1 atm of air saturation. Quenching factors of the diluted cis-decalin fluorescence (in the absence of PPO) at Lc= 161 nm and of PPO fluorescence at Lc= 225 nm were found to be 1.58 and 1.19, respectively.Again, using the estimated 12% reduction in the emission intensity of cis-decalin in iwoctane, the value of K was determined in these aerated 3% solutions to be 0.0596 f 0.009. Values of 9 were averaged over three sets of experiments (standard deviation ca. 10%). The results are displayed in Table IV. The Stern-Volmer constant, UD, was found to be 73 & 2 M-1. D. CiSDeclrlln + PPO (-35 "C). On cooling from 21 to -35 "C, the cis-decalin and PPO emission quantum yields both
200
175
150
Figwe 6. Intensity of cis-dccalin fluortscence in the neat liquid at 21 OC (A) and at -35 OC (B) as a function of excitation wavelength. The spectrum is not corrected for variation in lamp intensity with & (see Figure 1).
TABLE V Total Ihrgy-Tranafer Parameter, e, md Excess -fer Parameter, Y Q, for CLbDedn + PPO at -35 "C md L c 161 IUU
-
c, M 2 x 10-5 5 X 1V5 1XlCr 2x104 4X1e
sx1e 1 x 10-3 5 X 1W3 1 x 10-2
*-@ 0.00364 0.00848 0.0157 0.025 4 0.0396 0.046 8 0.067 8 0.183 0.336
0.00302 0.006 93 0.0126 0.019 2 0.027 2 0.031 3 0.036 8 0.028 0.026
Using K = 0.0896. Using Q = a@ with QD = 31 M-l.
increased only slightly by factors of 1.06 and 1.02, respectively. The emission spectra sharpened somewhat and shifted by ca.1S nm to the red. The onset of the cis-decalin absorption spectrum, as determined from its fluorescenceyaction' spectrum (see Figure 6), shifted blue by ca. 2.0 nm. Quenchingof the cis-decalin fluorescence by PPO was studied over a range of concentrations from 2.00 X 10-3 to 1.00 X 10-2 M. The results are shown in Figure 5B. Again, only a slight upward concavity was noted, with a limiting slope of 31.4 f 3.2 M-l, whereas the average value of U D from 5 X l t 3 to 1 X 1W2 M is 32 f 2 M-'. Table V lists values of the parameter 9 as a function of PPO concentration. Corrections for the direct absorption by PPO at 161 nm were made as previously described by using the emission from PPO in neat isooctane at the same temperature of -35 OC. Again, no adjustment was made for the change in the fraction of light absorbed by PPO due to different absorptivities of isooctane and cis-decalin at 161 nm. However, as in the previous analysis of the 21 OC data of section IIIA, such adjustment is not significant for results obtained at PPO concentrations leas than c = 1 X 1C3M. E. Strpy Light,TWO-PhotOnStaddy-Strte,and T~Iw-RHoIv~ MeuwemenCS. A few experiments that were designed to seek evidence for particular effects, but which gave negative results, are briefly summarized in this section. With regard to stray light, there was initially some conccrn that the anomalouslylargeemissionintensitiesthat wereobserved in very dilute solutions excited at 161nm had a contribution from direct absorption by the PPO of longer wavelength stray light that contaminated the 161-nm beam. This concem developed from observation of the emission intensity from PPO as a function of &. A typical such spectrum is shown in Figure 7A for 1 X 10-2 M PPO in both cis-decalin and isooctane. For cfs-decalin, one sees that although there is a well-resolved peak at 161 m,
ec#, The Journal of Physical Chemistry, 1.01
,
'
'
'
,
'
Wang et al.
Vol. 97, No.2, 1993 '
1'
'
E3 L
W
5-
0.5
m E
Q
'
1 .o
1.0 -50
I
I
50
100
I
150
TIME, ns
>
!= m Z
W 5-
0
0.5
m CY
Q
0
WAVELENGTH, nm F ~ ~ I7.w(A) Intensity of PPO fluorescencefrom a 1C2M PPO solution in cisdecalin (1) and isooctane (2) as a function of &. (B)Same as A but corrected for spectralvariation of lamp intensity using the emission from neat cisdecalin as a quantum counter.
where our lamp spectrum maximizes (see Figures 1 and a), neverthela, there is increasingly strong absorption by PPO of exciting light as LC increases p s t ca. 175nm. This is illustrated more clearly in Figure 7B,which corrects the data in Figure 7A for spectral variation of the lamp intensity. In more dilute solutions,the direct absorptionby PPObegins at somewhat longer wavelengths (e.g., at 182 nm in 1 X 10-4 M PPO). It is simple to show that any stray light contamination of the 161-nm beam that is relevant to our studies must be, at least, confined to a wavelength region below 182 nm. This is derived from the observation that the admission of air into the optical eicitation system (thereby eliminating all exciting light below ca. 182 nm as determined with sodium salicylate) eliminated all PPO fluorescence @e., to background levels wen at c = 1 X 1C2 M)when the exciting monochromator was set at 161 nm. In order to demonstrate the absence of significant stray light below 182 nm, we compared the ratio of the intensity of cis-decalin fluorescence to that of PPO for Lc= 161 nm using sample cell front windows of LiF and of Suprasil quartz (by reversing the cell). Sincethe absorptioncoefficientof Suprasil increasessharply below 182 nm, if there were significant stray light below this wavelength, the use of this window should have c a d a much higher ratio of PPO fluorescence intensity to that of cisdecalin. But in nosolution did we observeanychangein this ratio, although both cisadecalinand PPO had their intensities separately reduced by a factor of ca. 8 on replacing LiF with Suprasil quartz. Another potential problem that we considered was the papsibility that somehow the anomalously large PPO fluorescence that was observed at low PPO concentrations (sce section IV) was due to a surface effect connected to the rather short penetration depth of the 161-nm light. Although a number of arguments can be advanced that make this possibility seem unlikely, wenevertheless decided tocompare the ratioof acceptorto-donor fluorescenceintensities using 161-nmlight and using a more penetrating two-photon (308-or 337-nm)excitation of the donor. pXylene was chosen as acceptor to avoid any one-photon
F@e 8. Decay curves of cyclohexane +pxylme for twephoton lmcr excitation at 337 nm. (1) Emiuion from pxylene at c = 1 X 10-1 M. (2) Emiasion frompxylene at c = 1 X lo-' M. (3) Emission from neat cyclohexane. The curmare normalizedto unity at their "a(which definca t = 0) but are othemice uncorrected. The dirtortion from t PI 120 to 170 ns ir due to photomultiplier after pulre. contribution to its fluorescence,and cyclohexanewas then chosen as the appropriate donor, since its emission is sufficiently blueshifted from that of cis-decalin to avoid overlap with thepxylene fluorescence. Although the laser experiments had considerable scatter compared to the 161-nm experiments, there was nevertheless no indication of any change in the mode of excitation for any solution studied from 1 X 10-4 to 1 X 10-* M. The laser experiments were also used to seek evidence for a long-lived (ca. 100 ns) transferring state of the donor by comparing, at long times after a two-photon pulse, the decay curves ofpxylenefluorescencefrom both dilute and concentrated SO~U~~O inMcyclohexane. From published steady-state data on this system: it can be derived that for pxylene concentrations c 1 1C2M,the overwhelmingmajority of pxylene fluorescence derives from energy transfer from the normal fluorescing state of cyclohexane (lifetime, T Y 1 ns).I6 Accordingly, for the ca. 15-nspulse delivered by our lasers, the decay curve of a 1 X 10-2 M solution should not (and did not) differ significantly from the decay curve of directly excitedpxylene (T Y 30 ns).17 On the other hand, for a 1 X 10-4 M solution ofpxyleae in cyclohexane, where there appears to be a ca. 5096 contribution to pxylene fluorescence from the postulated dark state,' the decay should be significantly slower. Calculations indicated that at a time t 125ns after the maximum fluorescence intensity WM achieved, the ratio of pxylene intensity from the 1 X 10-4 M solution to that from the 1 X 1C2M solution (both normalized to unity at t = 0) should equal 1.4,2.1,and 2.7 for dark states of lifetimes 20.50, and 100 ns, respectively, and at a time f = 175 ns should equal 1.4, 3.1, and 7.6. However, as can be seen from representative traces in Figure 8, these ratios stayed essentially unity. A more detailed analysis indicated that it was unlikely for the system to have any transferring state that lived longer than ca. 10 ns.
-
lv. Ins" In the case that eq 2 is valid, the substitution of cq 5 into eq 2 would predict that
'PD
where both 9 and the right-hand side of eq 12,hereafter referred to as Q,Le., @OD Q=--l
@D
(13)
can be independently measured. However, whereas 9 is capable
The Journal of Physical Chemistry. Vol. 97, No. 2, 1993 409
High Efficiencies for Energy Transfer 0'1°
% -
400
-
300
-
200
-
I
n
0 I
s
W
t/'
0
0
,0005
loo 0 1 0
,001
"
of reliable determination (to within a few percent) at the lowest concentrations studied, the ratio VD/&, as noted earlier, becomes too close to unity for c < 10-3 M to permit equally reliable determination of Q. To circumvent this problem, we utilize the observationthat for the quenching of the fluorescence of saturated hydrocarbons, Q, although superlinear on cat high concentrations,usually achieves a limiting low concentration slope (Le,, Q/c = LTD) at c 5 X le3M (see Figure 5 ) . This behavior is not unexpected and, in fact, is predicted reasonably well by standard diffusion models of the quenching e n c o ~ n t e r Accordingly, . ~ ~ ~ ~ ~ ~from ~ a measurement of Q at c Y 5 X lC3M, we can deduce a value for the limiting slope, QD (or at least a value only slightly larger than this limit), and then use it to extrapolate Q into regions of c where Q is too small for reliable measurement. Thus, the determination of the validity of eq 12,at some PPO concentration, c, is based on a comparison of the experimental value of 9 with a value of Q computed from the product QN. Because of the superlinear character of Q, any difference between 9 and Q is likely to be slightly underestimated by this procedure. In column 3 of Table I, we show at various PPO concentrations values for 9 - Q for the system cis-decalin + PPO (deaerated at 21 "C). These values clearly depart significantly from the zero value predicted from eq 12and, indeed, grow proportionately larger as c decreases (reaching ca.70% of 9 at c = 2 X M). This deviant behavior from the prediction of eq 12 is also manifested in Figure 9, which shows the dependence of 9 on c below 1 X 1W M to exhibit neither linear nor even superlinear dependence on c (expected were Q equal to Q) but rather to be distinctly concave downward. At low concentrations (Le., I lo-) M),where @D = @OD, the parameters 9 and Q become, respectively, the probabilities that A has been somehow excited and DIhas been somehow quenched. The implication of 9 exceeding Q is, therefore, that there must exist an alternative channel for producing excited A. Stated otherwise, the violation of eq 12 and, perforce, eq 2 has the immediate consequence (as already noted in section I) that the fluorescing state of cis-decalin cannot be the exclusive precursor of excited PPO. Additionally, we note that since (9 - Q)/9 becomes larger as c decreases, this second channel must be responsible for an increasing fraction of PPO fluorescence as c 0. In previous studies, it has also been suggested that there exist two energy-transferring states of saturated hydrocarbons-the fluorescing state, DI, and an unidentified, nonfluorescent state, hereafter referred to ST.^ Additionally, however, it was assumed that T has DIas its immediate precunor. The consequences of this assumption are simple to rederive, and we will attempt, in what follows, todetermine to what extent these consequences are supported by our data.
-
'
'
'
25000
c-' Figure 9. Energy-transfer parameter. 9, as a function of PPO concentration for cis-dccalin + PPO excited at X, = 161 nm.
"
"
I
50000
(M-')
energy-transfer parameter, (P- Q)-l, plotted vs the inverse PPO concentration,c-l, for nitrogenated cfAccalin at 21 O C . The solid line is a linear least-squares fit to the experimental points. Figure 10. Inverse of the ex-
We define cps and (pr to be the probabilities that D1and T, respectively, transfer their energy to A and define x to be the intrinsic probability (i.e., in the absence of A) that DIconverts into T. Accordingly,
and @A = q + (1 - *)X*
(15)
@OA
+
where in eq 15 we have assumed, as is true for the cis-decalin PPO system, that all pertinent internal conversion efficiencies within the D and A manifolds are unity. Dividing eq 15 by eq 14 and substituting eqs 5 and 13 gives a two-state generalization of eq 12;i.e.,
*=Q+xM.
(16) Since (pr, by definition, cannot exceed unity, eq 16 predicts that 9 - Q should approach a limit, Le., x, as c increases. From knowledge of this limit, the absolute value of (pr can then be determined at any c from evaluation of (Q - Q)/x. A qualitative examination of the results in Table I appears to confirm the approach to a limit, at least up to concentrations of c = 5 X lo-' M. However, at 1 X 1C2 M, 9 - Qseems too large relative to the lower concentration values to support the existence of a limit, and this may, as we will explore later, imply a problem with eq 16. On the other hand, at high concentrations (Le., c L 5X M), as approaches Q, their difference becomes much too sensitive to small experimental to take too seriously any deviations from expected behavior. Accordingly, we will,at least for the moment, ignore this discrepancy. To more quantitatively determine the limit, x , we have plotted in Figure 10 the reciprocal of 9 - Qvs the reciprocal of c. Clearly, if eq 16 is valid, the extrapolation of (9- Q)-l to c-l = 0 should provide the value of x . As is clear from Figure 10,a linear regression of (9 Q)-I on c*fits our data from 2 X 10-s to 1 X 10-3 M aurprisin@ywell (correlationcocfficient,r= 0.999 98) withleast-squaresestimates of the slope and intercept of (7.63& 0.02)X le3M and 23.1 & 0.4, rcspcctively.20 Accordingly, x = 0.043 f 0.001 and M. is well representable by a Stern-Volmer form: Le.,
*
-
QTC
with UT = (3.03f 0.06)X
l o 3 M-l.
This suggests an unusually
410 The Journal of Physical Chemistry, Vol. 97, No. 2, 1993
efficient transport mechanism with a transfer probability, m, of 0.057 wen at our lowest concentration of 2 X lW5 M. The Stem-Volmer form for m has previously been surmised in earlier work on cyclohexane and n-heptane as donors and PPO andpxylene as acceptors ( 10-5-10-3M) using bzc = 147 nm and fitting ID and 1~(each proportional to @D and @A, respectively) to the right-hand sides of eqs 14 and 15.5 For cyclohexane + PPO, x = 0.012 and UT = 5600 M-I were reported; for cyclohexane pxylene, x = 0.01 1 and aT 5 5200 M-I; and for n-heptane + PPO, x = 0.0035 and aT = 1.2 X 104 M-I. Again, in unpublished work but using an analysis similar to that employed here, Walter' obtained for bicyclohexane + pxylene, at Lc= 174 nm, also a Stem-Volmer form for m with x = 0.05 and UT = 2550 M-I. As predicted by the model, x appears from the published results to depend only on the nature of the donor and a T seems to exhibit a possible viscosity dependence. Thus, for PPO as acceptor, we have, in order of decreasing viscosity (at 20 0C)21for cis-decalin ( q = 3.38 cP), cyclohexane ( q = 0.980 cP), and n-heptane (g = 0.428 cP), values for aT of 3.2 X 103, 5.6 X 103, and 1.2 X 104 M-1, respectively, whereas for pxylene as an acceptor in bicyclohexyl ( q = 3.75 cP) and in cyclohexane ( q = 0.980 cP), a? = 2.6 X lo3 and 5.2 X lo3 M-l. The effect of 0 2 on ?t - Q is presented in Table I1 for both airand oxygen-saturated sol~tions.~~Js The values for U D of 47 M-l (air) and 35 M-' (oxygen) that were utilized to compute Q are consistent with their predicted values using a D = 53 M-I for 02-free solutions and the quenching factors of 1.13 and 1.52 for quenching of neat cis-decalin fluorescence in air-saturated and oxygen-saturated solutions, respectively. Thus, if one assumes, as is usually done, that 0 2 and PPO are independently interacting with thesameD1, thenit issimpletoderivethat theaforementioned quenching factors should equal the ratio of aD's without and with 02 present. For air and oxygen saturation, respectively, these ratios of 1.13 and 1.51 indeed differ from the fluorescence quenching factors by insignificant amounts. Figure 11 shows plots of (?t - Q)-l vs c-l for air- and oxygensaturated solutions. In both cases, again, excellent linear regressions are obtainedz2with x = 0.039 f 0.001 and aT = (2.78 f 0.1 1) X lo3 M-' for air saturation and x = 0.041 f 0.003 and UT = (1.58 f 0.16) X lo3 M-' for 0 2 saturation. Comparing N2 with air- and oxygen-saturated solutions, it is clear that 0 2 has essentially no effect on x but has reduced UT by factors of ca. 1.1 and 1.9. In the presence of 0 2 . q s 14 and 15 require no changes in form but only changes in the interpretations of (ps, m, and x as probabilities in oxygenated rather than nitrogenated solutions. Accordingly,since x is the probability for a particular exit channel from D1 (Le., that leading to T), it would have been expected, in the presence of 0 2 , to be reduced to the same extent as is any other exit channel of D1. But, whereas the fluorescence intensity of D1 is reduced by factors of 1.13 and 1.52 for air and oxygen saturation, respectively, clearly x remains essentially unaffected. From this, we can only conclude that Dl cannot be a precursor of T. Similarly, we can conclude that T also cannot be the precursor of DI,since otherwisethe reduction in 9 - Q on addition of air or oxygen could not either have exceeded the quenching factors of D1 fluorescence. Yet clearly, at low concentrations, these quenching factors are significantly exceeded (e.g., at 2 X lW5 M, - Q is reduced by a factor of 2.2 in oxygen-saturated solutions, whereas the oxygen quenching factor of cis-decalin fluorescence is only 1.52). Thus, our results seem to force us to the conclusion that T and D1 are separately generated from a common precursor state and that we must reinterpret x as the branching ratio from this precursor state to T. Equation 15, accordingly, becomes modified to
+
*
-- -cPs+xPr *A @OA
Wang et al. 500
I
"
"
3
w
\
n
0
+
200
7
W
100
0 ' 0
'
'
'
'
'
'
25000
c-'
"
"
50000
(M-')
Figwe 11. Inverse ex-
energy-transfer parameter, (9,Q)-I, plotted vs the inverse PPO concentration, c-l, for (1) nitrogenated, (2) 1 atm of air-equilibrated. and (3) 1 atm of oxygen-equilibratedcis-dccalin at 21 O C . (B)The same as A except to replace (9 - Q)-I with (1 + Q)/(S - Q). The solid lines are linear least-squares fits to the experimental data.
and q 16 is replaced by
*=Q+xP# +Q) (19) whereas q s 13 and 14 remain unchanged. For PPO concentrations of less than ca. 1 X lW3 M, the predictions of q 19 for the product x- should differ only very slightly from the predictions of eq 16. This follows simply from the fact that at low concentrations, Q is negligible in comparison to 1. Thus,alinearregressionof(l +Q)/(?t-Q)onclcontinues to have a large correlation coefficient with least-squares parameters not differingsignificantlyfrom their previousvalues. Figure 11B exhibits these regressions for nitrogenated, aerated, and oxygenatedsolutions,and TableVI lists theleast-squarcsestimastimates of x and UT and their standard deviations. With regard to the two higher concentration measurements at c = 5 X lP3and 1 X l e 2 M, their values of ?t - Q are significantly altered when divided by 1 + Q and (in support of eq 19) in the direction to bring their predicted values of x m more nearly in line with expectations from the lower concentration points. Nevertheless, the uncertainties at these concentrations remain high for the reasons previously cited,'g and we will continue not to include these points in the linear regressions. However,as can be verified from the data in the Tables, when included, they leave the values of x and UT essentially unaltered. There is one last conclusion that can be extracted from a comparison of the data in Tables I and 11. The effect of 0 2 on UD, as pointed out earlier, is consistent with a competition between 0 2 and PPO for reaction with D1. If this competition also exists between PPO and 0 2 for reaction with T,then we could conclude that the ratio of ~1.r 3.17 X lo3 M-l in oxygen-free solutions to UT = 1.60 X lo3M-l in oxygen-saturated solutions,12is., 1.98, should equal the 0 2 quenching factor of T. Comparing this to the oxygen quenching factor of D1 (Le., 1-52), it follows that if the rate constants for 0 2 quenching of D1 and T are not too
-
The Journal of Physical Chemistry, Vol. 97, No.2, 1993 411
High Efficiencies for Energy Transfer
TABLE YI: Effects of % Dihltion with koctane, .ad Temperature on --Transfer system &.r)cccrlia + PPO
Parameters, UD, x, .ad a~ for the
a T x~ 10-3 M M-I Xb 3.17 0.07 neat, deaerated, 21 O C 53i2 0.041 0.001 2.89 & 0.10 neat, aerated," 21 O C 46 1 0.037 f 0.001 1.60 0.16 neat, oxygcnated,U 21 O C 35 2 0.039 f 0.003 3.98 0.41 3%.deaerated/21 O C 115 i3 0.049 f 0.004 4.04 0.32 lo%, deaerated/21 O C 11513 0.048 & 0.003 2.38 f 0.30 3%.aerated:* 21 O C 13 2 0.043 f 0.004 3.30 0.15 neat. deaerated, -35 O C 31+3 0.049 0.002 M. Determined from linear least-squares r e p d o n Determined from PPO quenching of cfs-decalin fluorescence at c = 5 X lW3 to 1 X of (1 + Q)/(* - Q) vs c-' from c = 2 X 1O-' M to 1 .O X M. e See footnote 20. Air quenching factor of cfs-decalin fluorescence is 1.1 3. e Oxygen quenching factor of cis-decalin fluoresance is 132.1%by volume of cis-decalin in isooctane. 8 Air quenching factor of cis-decalin fluorescencek 1.58.
condition
OD,'
*
* *
2oo
* ** *
*
*
1
B
100
f 0.004 and (3.98 0.41) X lo3 M-' for the 3% solutions and 0.048 f 0.003 and (4.04 f 0.32) X lo3M-l for the 10%solutions. Thus, there appears to be some increase in both x and a~ on dilution and to the same extent for 3% and 10% solutions. Comparing the effect of dilution on UT with that on U D (see Table VI), it would appear that if the T-state sensitization of PPO is of a diffusive nature, it is less influenced by a change in visCosity than is the, normal, D1-state sensitization. The effect of air on the 3% solutions is presented in section IIIC and Table IV. The increase in the air quenching factor of cis-decalin fluorescence from 1.13 in neat cis-decalin to 1.58 in the 3% solution is consistent with the diffusive nature of the D1 0 2 reaction and the large viscosity change on dilution (Le., 3.384.504 CPat 20 0C).21 This is seen, too, in the air quenching factor of directly excited PPO, which changes from 1.03 for PPO in neat cis-decalin to 1.19 in the 3% solution. Also,we note that the reduction of UD by a factor of 1.57 (from 115 to 73 M-l) on aeration of the 3% solution agrees well with the 1.58 quenching factor of the cis-decalin fluorescence, confirming again (but now for the 3% solution) the competition between PPO and 0 2 for the same D1 state. With regard to the effect of air on the T state, we retum to Figure 12A, which compares the regressions of (1 + Q)/(q - Q) on c-l for both aerated and deaerated solutions. The least-squares parameters are shown in Table VI. Comparing the values of x , we conclude that although there may be some difference, it is certainly not large (and definitely in a ratio much smaller than the 1.58 value obtained as the air quenching factor of cis-decalin fluorescence in the 3% solution). So we are supported in our conclusion that D1 is not a precursor of T. Similarly, from a comparison of the 1-58quenching factor with the larger effect of air on q - Q in the solutions of lowest PPO concentration (see Table IV), we also codirm that T is not the precursor of D1. Finally, wecompare thevalue of UT = (2.38 f 0.30) X lo3M-l in the 3% aerated solution with its analogue value of UT = (2.89 f 0.10) X lo3 M-' in aerated neat cis-decalin. As we had noted earlier in a similar comparison for nonaeratedsolutions, if T-state transfer involvesa diffusive process, UTShOUld increase on dilution to about the same extent as does U D (Le., from 46 M-' in aerated neat cis-decalin to 73 M-l in the 3% aerated solution). Clearly, this is not the case, and so we conclude again that in T-state sensitization,there is littlemanifestation of a diffusivecomponent. This is also confirmed by the results of our measurementsat -35 OC. At -35 OC, the viscosity of cis-decalin is ca. 19.7 cP.2) This represents an increase by a factor of ca. 5.7 from the viscosity at 20 OC, and as expected, the quenching parameter for DI fluorescence, i.e., UD, decreases severely from 53 to 31 M-l. However, comparison of Tables I and V indicates virtually no change in 9 and very little in q - Q, at least for concentrations below c = 2 X 10-4 M. Using eq 19, we again attempt to separate q - Q into its x and cpr components and find, as is illustrated in Figure 13, a good linear regression of (1 + Q)/(q - Q) vs c1with x 0.049 i 0.002 and UT = (3.30 f 0.15) X l@ M-l. Thus, lowering the
+
200
100
+
v
7
W
0 I 0
I
I
10000
20000
c-'
(M-')
Elsw 12. Exctas energy-transfer parameter, (1 + &)/(e- Q),plotted
vs the inverse PPO concentration,c-l. for (A) 3%cfs-decalinin isooctane at 21 O C nitrogenated (1) and aerated (2) and for (B) 10% cfs-decalin in isooctane at 21 O C nitrogenated. The solid lines are linear lcast-
squares fits to the experimental data.
different, T must have a lifetime within ca. 30% of that of DI. Of course, if T reacts much faster, its lifetime could be significantly shorter. Next, we tumour attention toTable I11andconsidertheeffects of diluting cis-decalin with isooctane. As noted in section IIIC, U D = 115 M-l in both 3% and 10% solutions. This represents a substantial increase over the value aD = 53 M-l obtained in neat cisdecalin, due, most plausibly, to the much reduced viscosity of isooctane?' Because U D (and, therefore, Q) is now so much larger, the difference between the predictions of eqs 16 and 19 becomes more apparent. Thus, from Table 111, we note that the values of q - Q for the two points at c = 5 X 10-3and 1 X 10-2 are very obviously too large relative to the limit that the lower concentration points seem to be approaching, whereas when divided by 1 Q, they become much closer to what would have been expected for the approach to this limit. In Figure 12, we showplotsof(l+ Q)/(q-Q)vscl forthe3%and lO%dilutions. Although, as previously commented on, the values of q for dilute solutions are somewhat less reliable than for the neat liquid, nevertheless the data continue to fit reasonably well a linear regression even when including all concentrations from 5 X to 1 X M. The least-squares estimates of x and UT are 0.049
+
-
412 The Journal of Physical Chemistry, Vol. 97, No.2, 1993 I
300
'
"
"
"
1
-
n
0 I
n
'0 0
0
25000
°
50000 1
c-' (M-') Flgm 13. Ex- energy-transfer parameter, (1 + Q)/(q- Q), plotted invene PPO concentration, c-1, for neat cis-decalin at -35 O C nitrogenated.
va the
temperature seems to have some small effect on the branching ratio, x , but to leave essentially unaffected the T-state transfer constant, aT (see Table VI). We conclude, therefore, that aT must certainly be determined most importantly by some nondiffusive process, and if there exists a diffusional component, it must only make a contribution to aT (albeit small), at viscosities lower than that of cis-decalin at 21 OC. There are essentially only three possible candidates for a nondiffusional effect. These include (a) direct absorptionof 161nm light by PPO (is., to a larger extent than we have corrected for based on our results in isooctane), (b) a radiative transfer from DI directly to PPO, or (c) a nonradiative but otherwise unspecified transfer of energy from T to PPO. The arguments against direct absorption and radiative transfer appear to us to be unassailable. Considering first direct absorption, we note that at low PPO concentrations, e.g , 2 X l W M in neat cis-decalin at 21 OC, in order to remove the difference between 8 and Q (i.e., to make 8 = 1.06 X l t 3 ) ,it would be required for PPO to absorb ca. 50 times more 161-nm light in cis-decalin than in isooctane. Certainly this would be much too large an effect to attribute to a change in absorptivity of the PPO (for such optically similar solvents). But, also, were it due instead to a ca. 50 times larger absorptivity of the isooctane at 161-nm vis-8-vis cis-decalin, it would require a similarly unreasonable increase in the emission quantum yield of cis-decalin in isooctane solution as compared to that of neat cisdccalin (by a factor of ca. 60).24 Also, were we observing only directly excited PPO, we would have expected absolutely no effect of 0 2 on 8 - Q since, in our Kvalue (through which we compute q),we have already taken into account the 0 2 quenching factor of PPO fluorescence (Le., 1.03 for air and 1.16 for oxygen saturation). With regard to radiative transfer, we note first that were this the exclusive origin for 8 # Q,then the effect of 0 2 on 8 - Q could not have exceeded (as it does at low concentrations) the effect of 0 2 on the fluorescence quantum yield of cis-decalin. Indeed it was already on this basis that we argued that T could not be the precursor of D1. Additional to this, however, there are many other arguments that negate the radiative explanation. For example, we have observed no change in the emission spectrum of cis-decalin even at PPO concentrations as high as 0.1 M.This implies no differential absorption of the cis-decalin spectrum by PPO in a spectral region, where PPO's absorptivitiy is changing strongly (from ca. 190 to 260 nm). Also, the excess quantum yields that we arc attempting to rationalize are not particularly small numbers. For example, at 2 X lW5M,8 - Q = 2.5 X (see Table I), and were this attributable to radiative transfer, it would require an effective absorption coefficient for cis-decalin fluorescence of ca. 5300 M-l. Such a large reabsorption effect would be impossibleto observe in the very short path length available to the emitted light in the forward direction (Le., the
Wang et al. penetrationdepthat 161nm) andevenin thebackwarddirection, which our collection geometry is anyhow very strongly b i d againat, would be difficult to rationalize. Finally, it is to be noted that at c Z 2 X lo-' M,\k - Q even exceeds the cfs-decalin emission quantum yield of 0.021.11 We return, therefore, to the remaining porsibility, namely that there indeed exists some nonradiative, nondiffusional transfer of energy from cis-decalin to PP0.25J6 In this caw, the form of e given by cq 17 (which all of our data appear to support) suggests that ~ T be C replaced by (&R)3 where R (-c1/9 can be taken to represent some average distance between T and PPO and & is some effective range parameter which clearly must be quite large (ca. 100 A) to accommodate an ar of 3.2 X 103 M-1. The precise significance of k depends, of course, on the mechanism of the transport procc8s, and it is not nectsrarily to be construed as the average distance between T and PPO at which cpr = 0.5. Nevertheless, to explain a transfer probability, cpr, of 0.060 at 2 X M will certainly require a very large distance over which the energy can be effectively moved. For example, in the case of a Fbrster transfer (which utilizes an inverse sixth power dependence on the separation distance)?' we calculate for the sensitization of PPO fluorescence by the DI state of cisdecalin a Fbrster &value of 13.4A, which translates to a transfer probability of only 2 X 10-4 at c = 2 X l C 5 M.28,29 A possible explanation for the large effective range is that the T-state energy can migrate through the donor solvent. Although we have observed cpr to increase on dilution with i8ooctBne, it remains possible that this increase was due to the development of a diffusionaltransport mechanism (due to the severelyreduced viscosity of isooctane) which effectively masked the effect of dilution to diminish the energy migration. In any event, we have not, thus far, been able to make reliable measurements at dilutions less than 3%. At this concentration, the average decalin4ccalin distance is only ca. 13 A, which is not too unphysical a distance for some residual energy migration. Measurements at substantially higher dilutions would be required to confirm this point. The nature of the T state remains unspecified. Clearly, it must not be unique to decalin since, as we have already noted, similar results to ours have been earlier reported for bicyclohexyl, cyclohexane, and heptane.c7 In the caw of bicyclohexyl: the T state has been generated at Lc= 173 nm, which suggests its energetic position to be close to the allowed electronic absorption threshold (at ca. 185 nm). Also in the present investigation, there is some qualitative evidence that the intensity of PPO fluorescence in cis-decalin, per photon absorbed, is independent of exciting wavelength from 150 toca. 175-180 nm. Thisresult is illustrated in Figure 7B for 1 X 1W2 M PPO in cis-decalii at 21 OC, and similar results (extending to ca. 180 nm) have been obtained at lower concentrations (where the direct absorption by PPO is less severe).30 Certainly the location of the T state very near the allowed electronicabsorptiononset, rather than at higher energies, would also be consistent with the observation that, at least for cis-decalin and for bicyclohexyl, their fluoreacence quantum yields remain constant, or very nearly so, from 120 to 180 m.10J1*32 Finally, in this regard, it is perhaps pertinent to note that almost coincident with the absorption onsets of t h m saturated hydrocarbons lies the most intense subionization transition of the aromatics, with an oscillator strength of about unity. Indeed, this transition can be seen even in our action spectrum of PPOgmission in isooctane (see Figure 7B) where its intense absorption competes so favorably with that of isooctane (but not cis-decalin) that it generates a shoulder at ca. 183 nm (e PL 6 X 104 M-1). We mention this merely to point out that there exists at least an acceptor state at the requisite energy for a very effective coupling with an appropriate T state and, a h , one that would be unavailable for energy transfer from aromatic donors (in whichcase, it should be recalled, exclusiveecnaitization by DI is capable of explaining all of the data).'Jm5 Beyond this,
High Efficiencies for Energy Transfer
The Journal of Physical Chemistry, Vol. 97, No. 2, 1993 413
(10) Johnston, D. B.; Wang, Y.-M.; Lipsky, S.J. Phys. Chem. 1991.95, there is little else that can be deduced at this time. Although the 5524. lowest triplet state of a saturated hydrocarbon is considered to (11) Rothman, W.; Hirayama, F.; Lipsky, S.J. Chcm. Phys. 1973,58, lie very close to the allowed electronic absorption t h r e s h ~ l d , ~ ~ * ~ '1300. (12) Wu, K.C.; Lipsky, S . J. Chem. Phys. 1977.56, 5614. it is difficult to envision a triplet with such special property as (13) Murov, S.L. Handbook of Phorochemisrry; M. Dekker: New York, to permit it to so effectively sensitize an aromatic singlet. On the 1973; p 99. other hand, solvent charge transfer states, which may also lie (14) The Ostwald coefficient for 0 2 in cis-decalin at 25 OC har been close to the absorption threshold, present perhaps a more viable determined to be 0.18 (see ref IS). This translates to an 0 2 concentration of 7.4 X le3M at a pressure of 1 atm. alternative. Certainly our knowledgeof the electronic properties (1 5) Horsman-van den Dool, L. E. W.; Warman, J. M.The Solubilities. of the saturated hydrocarbons remains incomplete. Indeed, even in a Variety of Organic Liquids, of Some Ga~eausCompounds commonly the origin of the relatively high stability of the DIstate that Used as Scavengers in Radiation Chemistry. IRI Report 134-81-01. Interuniversitair Reactor Institut, Mehelweg, IS Delft, 1989. permits it to live ca. 1-3 ns16 must still be considered somewhat (16) Hermann, R.; Mehnert, R.; Wojnarovits, L. J. Luminesc. 1985,33, of an enigma. 69;Mehnert, R.;Brede, 0.;Naumann, W.; Hermann,R. Radiat. Phys. Chem.
v.
ConchrPiolrs
The electronic state of cis-decalin that is generated by the absorptionof 161-nmlight ultimatelydecayaintotwoindependent stata, one being the normal fluorescing state, DI, and the other being some as yet unspecified state, T. The branching ratio for T-state production is ca. 0.04, and the branching appears to occur somewhere near the allowed electronic absorption threshold of cis-decalin. Both DIand T sensitize PPO fluorescence. The DI sensitization has a large diffusive component, whereas T-state sensitization at high viscosities (i.e., 1 3.5 cP) occurs almost entirely via some nonradiative, nondiffusive process. The efficiency of this process, measured in terms of the number of PPO fluorescingstates generated per T state produced, is anomalously large and of the order 0.060 at 2 X lW5M and 0.97 at 1W2M. A mechanism of energy migration of the T state through the decalin solvent is suggested. Theseresultsconfirmearlier surmises of the existence of a dark state in other saturated hydrocarbons with anomalously high efficiencyfor energy transport to aromatic acceptors. AcLaowledgment. This research was supported in part by the
U.S.Department of Energy, Division of ChemicalSciences,Office of Basic Energy Science. We acknowledge Ms.Agnes E. Ostafin for her technical assistance.
References d Notes (1) Birks, J. B. Photophysics of Aromatic Molecules; Wiley-Interscience: New York, 1970; Chapter 11. (2) Alwattar, A. H.; Lumb, M.D.; Birb, J. B. In Organic Molecular Photophysics; Birks, J. B., Ed.; Wiley: New York, 1973;Chapter 8. (3) Laor, U.; Weinreb, A. 1. Chem. Phys. 1969, 50, 94. (4) Weinreb, A. In Organic Scintillators and Liquid Scintillation Counting.Horrocks, D. L., Peng, C. T., Eds.;Academic Press: London, 1971; p 45. (5) Hirayama, F.; Lipsky, S . In Organic Scintillators and Liquid Scintillation Counting; Horrocks, D. L., Peng, C. T., Ed.; Academic Press: London, 1971;p 205. (6) Walter, L.; Hirayama, F.; Lipsky, S . Int. J. Radiat. Phys. Chem. 1976,8, 237. (7) Walter, L. Ph.D. Dissertation, University of Minnesota, 1980. (8) Yahida. Y.;Lipsky, S . J. Phys. Chem. 1988,92, 2240. (9) Johnston, D. B.; Lipsky, S . J. Phys. Chem. 1991, 95, 1896.
1988,32, 325 and rcferenca cited therein. (17) Berlman, I. B. Handbook of Fluorescence Spectra of Aromatic Molecules, 2nd ed.; Academic Press: New York, 1971;p 149. (18) Choi, H. T.;Lipsky, S . J. Phys. Chem. 1981,85,4089. (19) At high concentrations, 8 - Q becomes too sensitive to small errors in ~YDor in K or even in the small corrections that were applied to remove the contributions from directly excited PPO, for any significant reliability. For example. using the data in Table I, it can be seen that a ca. 2% error in a which causa only a ca. 3% error in 8 Qat 1V3M, causa 8 Q tochange by 8% at 5 X lV3M and by 24% at 1 X le2M. (20) Had the point at 5 X l O also been included, the fit to a linear regreacion would have decreased with a correlation coefficient of 0.9995. However, the slope and intercept would change only slightly to 7.59 X 10-3 M and 24.2, respectively. (21) Riddick, J. A.; Bunger, W. B. Organic Solvents. Techniques of Chemisrry, 3rd ed.; Wiley-Interscience: New York, 1970; Vol. 11, Chapter 111. (22) In the plot for 0 2 saturation, we have left out the point at c = 2 X le' M since this entry had large uncertainty in 8 (see section B). The regression without thispoint, however, predicts its value to be 0.001 25,which lies within the expected uncertainty of ca. 10%. (23) Viwanath, D.S.;Natarajan, G .DataBookon the WsmityofLiquf&, Hemisphere: New York, 1989; p 207. (24) This follows from our measurements on the intensity of cisdealin fluorescence in 3% and 10% solutions in is0octane (see section IIIC). (25) Having observed some diffuse structure in the action spectrum of PPO fluoracmce in cyclohexane at X, = 159 nm and at 128 nm,Kimura and HormesX have suggested the existence of a very fast transfor of electronic energy to PPO from cyclohexane excited at theee high energica. However. we have been unable to observe the structure at 159 nm in the sytem cisdecalin PPO at le2M.Indeed, from 150 to 165 nm,our action spectra appear to be flat within ca. 3% (see Figure 7B). A mcMurcment made with cyclohexane saturated with PPO also failed to see any dhtinct structure in this region. (26) Kimura, K.;Hormes. J. J. Chem. Phys. 1983, 79, 2756. (27) Fbrster, Th. Ann. Phys. 1948, 2, 55. (28) To com Ute the transfer probability, we have utilized the F h t e r rcaultm (pr = r&j& erfc r), where r) = (22/2/3)cR30. (29) Fbrster, Th. Z . Naturforsch. 1949, l a , 321. (30) Unfortunately, the spectral distribution of our lamp (see F W e 1) precluded our making extensive measurements at wavelengths other than 161
-
-
+
nm. (31) Ostafin, A. E.;Lipsky, S . 1.Phys. Chem.,submitted for publication. (32) At longer wavelengths, closer to the absorption thrslhold, small impurity absorptions and geometrical effects on the collection SlIiOieacy of the emitted light preclude accurate measurement of the fluotsscencequantum yield. (33) Robin, M. B. Higher Excited States of Polyatomic Molecules. Academic Press: New York, 1924; Vol. I, Chapter 3. (34) Johnson, K.E.;Kim, K.;Johnston, D. B.; Lipsky, S.J. Chem. Phys. 1979, 70, 2189.