4
E. J. BOWEN AND J. SAHU
peak a t 40" that precedes the main light emission. The equations of Randall and Wilkins3 with the value of K as 2.5 X lo9 set.-' give the activation energy for this peak as 0.69 e.v. Unless the sample is cooled this peak is largely annealed out during the illumination. During the course of these experiments a number of qualitative observations were made that bear on the mechanism whereby dried chloroplasts store energy. They are summarized in the following statements. (1) The composition of the atmosphere surrounding the sample during the heating seems to have no effect on the glow curve. (2) Samples illuminated in air or 0 2 become charged, that is, store energy. (3) Samples illuminated in N2 or SO2 do not become charged. (4) Neither water vapor or COz seem to have any effect on the charging. ( 5 ) Air and 0 2 illuminated with short wave length ultraviolet light can charge the dried chloroplasts by flowing over the surface in the dark, ( 6 ) Nitrogen illuminated by ultraviolet does not charge the samples. (7) Air illuminated by ultraviolet and passed through a strong electric field to remove ions will still charge the samples. Discussion The two most reasonable explanations of the storage of energy in dried chloroplasts are the trap-
Vol. G3
ping of electrons or the formation of some high energy chemical compound. The trapping of electrons is suggested by the following: (1) at least five different activation energies are involved in the glow curve. A multiplicity of electron trap levels is common in the storage of energy in crystals. (2) The spikes in electrical conduction shown in Fig. 4 and 5 of the previous paper2 would seem to imply the freeing of electrons. (3) The value of the frequency factor K of 2.5 >( lo9set.-' is near to the l O S - l O g set.-' suggested by Randall and WiJkins3 as appropriate for the freeing of electrons. On the other hand the chemical storage is suggested by: (1) oxygen is needed in order for the energy to be stored. If all that oxygen did was to trap an electron then SO2 would be expected to serve as well, and it does not. (2) The fact that air irradiated with ultraviolet light can be used to charge a chloroplast sample would indicate a chemical reaction. Finally it might be argued that the storage represents a mixture of electron trapping and chemical compound formation. The occurrence of the spike in electrical conduction2 at a temperature considerably below the temperature of maximum light emission may indicate that these two effects are different.
THE EFFECT OF TEMPERATURE O N FLUORESCENCE OF SOLUTIONS BY E. J. BOWEN AND J. SAHU Physical Chemistry Laboratory, Oxford University,England Received M a y 10,1068
Measurements have been made of the effect of temperature on the fluorescence yields of substituted anthracenes dissolved in several solvents. 9-Substituted anthracenes show high yields and steep temperature dependencies, while side-substituted derivatives have low yields and small temperature variations. The results are interpreted in terms of the following concepts. There appear t o be two processes of energy degradation of the excited molecules, a substantially temperatureindependent one which probably is associated with the singlet-triplet conversion, and a temperature-dependent one having a heat of activation of degradat'ion and a dependence on solvent viscosity which may be associated with a direct transition from excited to the ground state. Values also are given of fluorescence quenching constants of dissolved oxygen, bromobenzene and carbon tetrachloride for these anthracene derivatives.
Measurement of true temperature effects of the fluorescence of solutions is rendered difficult by the smallness of the changes and of the complications inherent in the estimation of the total light emitted. Previous attempts to obtain reliable results have been subject to uncertainties of magnitudes approaching those of the quantities measured. Apart from experimental errors due to inconstancies of the apparatus, changes in the spatial distribution of the fluorescence caused by refractive index and polarization effects can occasion misleading interpretations of the results. 1.2 The correction to be applied for the refractive index effect can be assessed only roughly, as it depends on the detailed geometry of the apparatus, and it is sufficiently large in some instances to reverse the sign of the change of fluorescence with temperature. Measurements are here given which were made with a new form of apparatus designed par( I ) , E. J. Bowen and K. West, J . Chem. SOC.,4394 (1955). (2) E. J. Bowen and D. M. Stebbens, ibid., 360 (1957).
ticularly to minimize refractive index and polarization errors.
Experimental Figure 1 shows the arrangement in section and in plan. A 5-liter glass flask was fitted with a tube A for evacuation, and with a tube B, reaching to the center, to contain the solution. The flask was lightly smoked internally with esium oxide and painted over the external surface %,f?barium sulphate-Cellofas B mixture, except for holes to admit the exciting li h t and to observe the fluorescence. After evacuation the h s k was thus both a Dewar vessel and a0n integrating sphere. A beam of filtered mercury 3660 A. light was focussed by lens C on to the tube B, and the fluorescence collected from the flask wall a t E allowed to fall on a spectrometer slit a t D. The concentrations of the solutions were such as to give practically total light absorption in tube B, and the liquid level was above the top of the light beam to eliminate any effect from thermal expansion. This arrangement avoids errors due to changes in the spatial distribution of the fluorescence emitted from the solution, whether due to refractive index or to polarization effects. The collected fluorescence light waa dispersed by a Hilger D 246 spectrometer, with glass prism, to .the exit slit of which was attached a 13 stage photo-multiplier connected
h
EFFECT OF TEMPERATURE ON FLUORESCEWE OF SOLUTIONS
.J:m., 1059
A
to a microammeter. The spectrometer-multiplier combination was calibrated by reference to a tungsten lamp of known color temperature so that fluorescence spectra could be plotted as relative quanta emitted per equal interval of wave number against the wave number. The areas of such curves are proportional to quantum yields of fluorescence. For the substances used in this work changes of band shape with temperature or solvent were slight, and it was usually sufficient to com are fluorescence emissions by measurement of the heigft of a band maximum. As a reference standard the quantum yield of a dilute, de-aerated benzene solution of anthracene was taken as 0.29.3 The solutions contained in tube B, from which dissolved oxygen was removed by a fine stream of pure nitrogen, were heated or cooled by the temporary insertion of an inner glass tube, terminated by a closed copper-glass seal, into which hot water or liquid oxygen could be introduced. Temperatures were determined by a dipping thermo-junction. The fluorescent solutes investigated were anthracene derivatives dissolved in several solvents. The concentrations necessary to give practically total light absorption were fortunately just low enough to make concentration quenching changes unimportant. Temperatures ranged downwards to -70’ for non-freezing solvents and upwards to f80” for viscous paraffin.
Results Figure 2 shows the curves obtained for two methyl anthracenes, and Fig. 3 those for two dichloroanthracenes. It was found that all the 9substituted anthracenes studied gave yields which tended close to unity a t low temperatures and which fell off steeply a t higher temperatures, except for the 9-cyano- and 9,IO-diphenyl derivatives, which gave unit yields in all solvents over the whole temperature range. In contrast to this the sidesubstituted derivatives showed flattish yieldtemperature curves whose zero extrapolations could not be determined, but which probably do not reach unity. We may write for the 9-substituted derivatives the simple scheme A + ku+ A” A’ +A h.v’ Ax+ A
+
(1) (2) (3)
From this it follows that (1/F - 1) = the ratio of rates of procemes 3 and 2 , where F is the fluorescence yield. This ratio may be set equal to k exp(- E I R T ) , where E is the activation energy of process 3 and k is its pre-exponential constant multiplied by the mean life of process 2. Values of k and E (cal./mole) so determined are given in Table I. For the derivatives where the curves are flattish and the course a t lower temperatures uncertain, part of process 3 may be temperature independent and the equation may need to be of the form 1/F
-K
= k
exp(-EjRT)
As the experimental data are not accurate enough to determine K in this way the values of IC and E given for them in Table I are evaluated on the simpler formula to serve merely as an empirical representation of the measurements. The E values for the 9-substituted derivatives depend upon the nature of the anthracene molecule, and are greater for those of higher degree of electron delocalization. The variation with solvent is comparatively small, but when large changes are (3) F. Weber and F. W. J. Teale, Trans. Faradaz, Soc., 63, 646 (1957).
u u1 W
0.3
g 0.2
-
PETROLEUM
VISCOUS
PARAFFIN
J%g& ETHYL
-ETHYL KETONE
ALCOHOL
ACETATE
2-METHYL ANTHRACENE
3
0.1 -
0 -
I
I
I
I
I
E. J. BOWEN AND ,J. SAHU
G
Vol. 63
TABLE I Petroleum ether
Anthracene
Unsubstituted
k
13 1100 2.24 Small 11 260 24.5 480 253 2840 275.5 2920 14000 5310 40.5 2220 1980 3370 842 4100
E k E k E
2-Methyl 1-Chloro 1,5-Dichloro
k
E 9-Methyl
k
9-Ethyl
E k E
9-Methoxy
k
E 9-Phenyl
k
E 9-Chloro
k:
E 9,lO-Dichloro
Ethyl alcohol
k:
E
Ethyl acetate
Solvent Acetone
Chloroform
Toluene
7 10.5 7 610 860 570 3.54 4.7 10.7 Small Small 570 9 10.5 16.5 Small Small 260 21 22.5 24 310 310 360 100 90 23 2580 2550 2120 150 75 2830 2810 5400 4780 3700 4500 4290 4850 34.5 23.5 29 2300 2000 2320 1170 725 550 3450 3250 3500 842 404 104 96 4100 3730 2950 3300 Have a yield of unity in all the solvents in the temperature range of -70 20" 11 890 3.24 Small 13 260 26.5 480 91 2380 89 2400 10200 4900 34.5 2140
9-Cyano 9,lO-Diphenyl
9.5 230 12 Small 5 310 31 2120
3260 4500 43 2480 134 3500
+
TABLE I1 Anthracene
9-Methyl
k
E
IC
9-E t,hgl
E k E
9-Methoxy
IC
9-Phenyl
E 9-Chloro
k
E 9,lO-Dichloro
k E
A
B
430 3900 645 4175 47000 6450 110 3300 2000 3900 1234 4800
420 3700 630 3950 45000 6350 104 3140 2000 3700 928 4500
F
342 3380 565 3730 42500 6200 92 2910 2000 3650 904 4400
TABLE I11 A
B
E
F
mate agrcement with the measured fluorescence yields is given by the formula
- 1 = k exp(-R/RT)
X T/v/(C
+ T/v)
where C is a constant almost independent of the solvent. Table I V gives numerical values of the constants. This equation can be derived from the more elaborate scheme (a) (b) (c) (d) (e)
+
A hv +A" A" +A hr' A" -+ A1 A1 +Ax A1 Solv. -+ A
+
+
335 3160 392 3320 37200 5950 62 2530 2110 3600 850 4230
253 2840 276 2920 14000 5310 41 2220 1980 3370 842 4100
TABLE IV
Solvent C D
Viscosity constants a X 10-6 57500 43.8 4.61 3.26 5.66 5 . 6 b, cal./mole 9550 5360 3600 2780 3200 2560
1/F
340 3200 553 3580 39200 6050 86 2810 2110 3600 862 4250
Rate constant 1
kr kl exp(--E/fiT) kz kr(T/v)
which gives IC = kl/kf and C = kz/k3. Process (c) represents an excited molecule A", stabilized by the solvent on its zero vibrational level, receiving thermal energy and reaching a "crossing
Anthracene derivative
9-Methyl 9-Ethyl 9-Methoxy 9-Phenyl 9-Chloro 9,lO-Dichloro
k
65 50 10,000 2.4 600 270
E (cal./ mole)
2000 1900 5000 1900 2600 3400
C
4
x
7
x
104 104 104 103 104 102
point" to a lower level on the potential energy surface; in process (d) it simply drops back to its zero vibrational level. In process (e) we imagine the molecule A', while executing vibrational movements at the crossing point, being trapped by inward diffusion of solvent molecules during one of its compression phases, so that it loses vibrational energy to the solvent and reverts electronically to the lower level. The vapors of anthracene and of 9,lO-diphenylanthracene have fluorescence yields of nearly unity, although the viscosity is exceedingly small.4 This may be ascribed to a much smaller value of the constant k3 (which depends on collisional rate) for gases than for liquids. (4) B. Stevens, i b i d . , 51, G10 (1955).
. ,
EFFECT OF TEMPERATURE ON FLUORESCENCE OF SOLUT~ONS
Jan., 1959
The expression given above covers relationships found by G. Oster and Y. Niehijima5 for the fluorescence of diphenylmethane derivatives and by Porter and Windsor for the rate-constants of decay of the triplet level of anthracene in solution.6 Excited molecules may revert to a triplet level or directly to the ground level. It is likely that temperature-independent changes which appear to play a large part for side substituted anthracenes characterize the first of these, and that temperature-dependent ones of the kind discussed above relate to the second mode of deactivation. Measurements also have been made of the quenching of the fluorescence of anthracenes by dissolved oxygen, bromobenzene and carbon tetrachloride. The Stern-Volmer constants were determined and the rate constants for the quenching process were evaluated in 1. mole-' set.-' units by dividing by the fluorescence yield and by the mean life of radiation, the latter value being taken as sec. in the absence of more precise data.' The constants for oxygen are given in Table V. For solvents of viscosities of those of Table I1 the La Mer theorys for the diffusion controlled bimolecular reaction predicts absolute rates of about 100 X los; the values found are much larger, and the discrepancy cannot be accounted for by lack of precision in the value assumed for the mean radiational life. It is probable that the excited oxygen molecule has a large range of effective action. TABLE V
O2QUENCHING CONSTANTS, L. MOLE-1 Anthracene
Unsubst. 2-Methyl 9-Methyl 9-Ethyl 9-Methoxy %Cyano 9-Phenyl 9,lO-Diphenyl 1-Chloro 1,5-Dichloro 9-Chloro 9,lO-Dichloro
SEC.-~
X 10-8
-Solvent
r
Petro- Ethyl leum alco- Ethyl Ace80-10do hol acetate tone
760 585 460 658 680 173 510 350 435 320 480 271
635 454 545 510 515 223 354 320 326 336 345
767 670 573 660 655 243 440 355 382 404 525 415
748 715 785 620 765 460 380 463 447 555 477
Tolu- Chloroene form
515 640 607 550 515 210 352 271 440 380 400 390
350 485 393 440 130 315 226 330 212 286 268
Tables VI and VI1 give the constants for bromobenzene and carbon tetrachloride as quenching agents. The values are of comparable magnitude and about a hundred-fold lower than those for oxygen. Table VI11 shows the effect of temperature. These bimolecular reactions lie just on the (5) G . Oster and Y.Nishijima, J. Am. Chem. Soc., 78, 1581 (1956). (6) G. Porter and M. W. Windsor, Disc. Faraday Soc., IT, 178 (1954). (7) E. J. Bowen, Trans. ParadaV SOL,SO, 97 (1954). ( 8 ) J. Q. Umberger and V. K. La Mer, J. A m . Chem. Soc., 67, 1099 (1954).
TABLEVI CoH5Br QUENCHING CONSTANTS, L. MOLE-'
7 sEC.-' X
-----
Solvent-Petro- Ethyl Ethyl leum, alcoaceAce80-100" hol tate tone
Anthracene
Unsubst. 2-Methyl 9-Methyl 9-E t h y 1 9-Methoxy 9-Cyano 9-Phenyl 9, IO-Diphenyl I-Chloro 1,5-Dichloro 9,lO-Dichloro
5.8 6.4 2.2 1.5 0 0 1.02 0 4.5 . 5.7
7.5 7.5 2.48 1.74 0 0 2.64 0 6.0 6.25
0
0
Tolu- Chloro-
4.1 5.7 8 . 3 1.45 1.25 0.76 0.50 0 0 0 0 0.82 0.80 0 0 4.2 5 . 8 4.1 4.2 0 0 4.4
TABLE VI1 CC14QUENCHING CONSTANTS, L. MOLE-'
ene
form
4.53 7.3 0.4 0.38 0 0 0.33 0 4.2. 2.9 0
4.5
SEC.-'
X
0.8 0.29 0 0 0.24 0 4.3 3 1 0
lo-*
Ethyl alcohol
Ethyl acetate
Acetone
Tolu-
Chloroform
Unsubst. 3 .4 66.0 4 . 5 36.6 2-Methyl 3 .95 47.0 9-Methyl 2 . 9 37.6 9-Ethyl 9-Methoxy 15. 6 05.5 9-Cyano 0 0 9-Phenyl 1.68 20.2 9, IO-Diphenyl 2 . 5 12.5 I-Chloro 0 2.5 1,5-Dichloro 0 0 9,lO-Dichloro 0 0
45.4 41.0 49.5 40.0 61.0 0 18.0 10.9 6.5 0 0
68.3 69.0 90.0 66.5 97.0 0 31.0 19.7 13.8 1.8 0
18.9 35.0 30.6 23.0 46.4 0 8.0 4.85 2.2 0 0
12.8
Anthracene
Petroleum,
--?
80-100°
ene
24.0 15.8 53.7 0 8.8 4.4 1.65 0 0
TABLE VI11 EFFECTOF TEMPERATURE O N QUENCHING CONSTANTSQ Solvent
-50
Petroleum, b.p.80-
B C
Kerosene
B C B C B C
looo
Viscous paraffin Ethyl alcohol
5.8 1.4
-30
5.9 1.9
-10
+20
5.8 2.6
5.9 3.4
+40
6.2 6 . 2 2.3 3.4 1.9 2.8 3.4 3.9 6,4 6.8 7.0 7.5 19 29 41 66
4-60'
6.3 5.1 3.2 4.8
B = bromobenmene and C =carbon tetrachloride,
point where the controlling influence ceases to be diffusion. The bromobenzene constants are not very dependent on solvent or on temperature, in contrast with those for carbon tetrachloride. This would indicate that polarizability and the need for appreciable activation energy plays a larger part with the latter substance, and this 'may be correlated with the fact that bromobenzene quenching is a spin-reversal effect due to the heavy bromine atom while carbon tetrachloride quenching involves actual chemical reaction. (9) E. J. Bowen and I