ON LUMINESCENCE DECAY TIMES AND THEIR RELATION TO MECHANISMS OF ENERGY TRANSFER IN RADIATION CHEMISTRY OF LIQUIDS* BY MILTON BURTON AND HERBERT DREESKAMP Dept. of Chemistry, University of Notre Dame, Notre Dame, Indiana Received 2nd April, 1959 Luminescence against time curves, obtained by a newly developed technique, are given. Decay times of luminescence processes in mixed cyclohexane + benzene scintil- lator systems (with air present) lie in the range around 2 x 10-9 sec. Addition of benzene to a cyclohexane solution of p-terphenyl causes a sharp rise in the decay time to a maximum at low benzene concentrations. The decay time is attributed tentatively to the energy transfer process from the lowest excited state. The estimated fraction of such low excited states transferring excitation to the scintillator is only 0-04 to 0.08.On the other hand, the scintillator appears also to quench the excited solvent. With the preliminary data at hand, the luminescence decay times are used as a tool for interpretation of the mechanism of protection in radiolysis of mixed liquid systems. 1. INTRODUCTION 1.1. ON PROTECTION Numerous experiments on radiolysis of liquid mixtures containing cyclo- hexane and benzenel-4 indicate that the 100eV yield of hydrogen from cyclo- hexane may be decreased by two effects attributable to the benzene. One may be true free-atom scavenging. The other appears to be a true protection, in the sense that energy primarily absorbed in a cyclohexane molecule is transferred to a benzene molecule, where it is there dissipated. One of the first suggestions of the nature of the energy-transfer process in this case was that both ionization- transfer and excitation-transfer may be involved, because both the ionization potential and the lowest (singlet) excitation potential of cyclohexane, determined for the gas, lie higher than the respective quantities for benzene.1 There are numerous difficulties in establishment that a protective effect in a liquid mixture is properly attributable to ionization transfer.Not the least is the difficulty in following experimentally the behaviour of an ion in such cases. Evidence does exist of ionization transfer in gases. However, in view of the various possible complexities of ion-molecule behaviour in liquids as seen, in part, from behaviour of gases, and, in part, from pure speculation, extrapolation of such evidence to liquid cases can be a quite unjustifiable procedure.5 A further difficulty in the assumption of an ion-transfer mechanism seems t o be wholly theoretical ; for pure hydrocarbon systems like cyclohexane + benzene the ion may have too short an independent life to participate in an ionization- transfer mechanism representable by such a simple reaction as Cf + B -+ C + B+, (1) if we think of the two ions as having separate existence.On the other hand, there is a possibility that the combination of initial positive-hole and electron, * Contribution from the Radiation Project operated by the University of Notre Dame and supported in part under Atomic Energy Commission contract AT(11-1)-38.64M. BURTON AND H. DREESKAMP 65 produced in a spur, may not be localized but may itself move very rapidly so that independently of what is initially ionized the neutralization process does involve a benzene molecule. In such a sense an ionization transfer process does occur but the mechanism has connotations somewhat different from what the chemist ordinarily means by reaction (1). According to an alternative view, the neutralization process does involve a truly primarily-ionized molecule. For a cyclohexane + benzene mixture rich in cyclohexane, this means that the predominant neutralization process would be C+ + e -+ C+ where C+ represents a highly excited state of cyclohexane, either singlet or triplet.5 In any event a significant number of excited cyclohexane molecules are produced primarily in a variety of excited states (2) c -c*.(3) According to Kasha,6 in most liquid cases the more highly excited states initially produced (i.e. either by reactions (2) or (3)) internally convert quite rapidly (i.e. in about 10-13 sec) to the lowest excited state permitted by the selection rules or by considerations of stability. Thus, if the initial highly excited state is singlet, the final excited state is singlet. On the other hand, there is a possibility (not discussed in the paper by Kasha), for triplet states particularly, that the course of internal conversion may lead into a repulsive state and directly into decomposition, thus interrupting the excitation degradation at an intermediate point. Such a mechanism may be the path of radiolysis of benzene which decomposes with reasonable yield 1 (e.g.G(H2) = 0.037) in spite of the fact that the photochemical quantum yield of decomposition, in contrast, is negligible in the absorption region near 2500 A and detectable only on irradiation 7 9 8 near 1800 A. The mechanism of internal conversion suggested by Kasha has not yet been elucidated nor has it been established whether a presumed protective process like C* + B -+ C + B* can occur during the time of the excitation degradation or is necessarily delayed until the degradation to the lowest excited state of cyclohexane is complete. Furthermore, in a certain sense, mechanisms have been suggested for the simul- taneous resistance of benzene to radiolysis and its failure to show any large fluorescence.They involve the notion of an extremely rapid internal conversion from the lowest excited state to the ground state. However, such an interpretation appears contrary to the known great effectiveness of benzene as a scintillator solvent 9 and to the data of the present paper. A major difficulty in an attempt to solve the problems involved in mechanisms of protection results from the various possible effects of trace quantities of im- purity.5 They may act as scavengers, as negative-ion formers, as protective agents and as inducers of internal conversion. When an effort is made to establish the behaviour of a pure substance as a reference material, the question of the possible determinative effect of an unknown impurity enters. While this method of approach to the solution of the problems of energy transfer cannot be ignored and must continue to be pressed, the most recent endeavour in our laboratories has been to avoid such complicating features wherever we can.(4) 1.2. O N LUMINESCENCE OF MIXTURES An obvious approach to the problems of the behaviour of excited states in radiolysis is through the study of luminescence as a function of the composition of multicomponent mixtures. Some of our results have already been reported 10-13 and the evidence can be interpreted to indicate (as might have been expected) C66 LUMINESCENCE DECAY TIMES that in certain mixtures, like cyclohexane + benzene + 0.6 mole % p-terphenyl, the path of excitation transfer may be represented by C* + B 3 C + B*, B* + T --f B + T*, (4) (5) T * + T + h , (6) followed by light emission from the scintillator and accompanied by a number of other energy degradation and quenching processes.One rather significant result obtained in such cyclohexane + benzene + p-terphenyl mixtures containing oxygen, phenyl bromide or methyl bromide as quencher is the existence of a luminescence minimum at benzene concentration near 1 volume percent. Existence of such a minimum is not a general property of quaternary scintillator solutions and is even affected by the nature of the quencher. We have made repeated efforts at understanding of these results by both chemical and physical experiments. I0 r I 0. I 0 10 20 t, units of 10-9 sec FIG. 1.-Luminescence intensity as a function of time for an aerated solution containing 5 g p-terphenyl per litre of benzene.I is in arbitrary units. The present paper represents a brief report of current work on a number of physical features of luminescence studies and their possible significance for radi- ation chemistry. Only the general features of the preliminary results are related. Details will be published in forthcoming papers. 2. TIME STUDIES OF LUMINESCENCE A technique developed in our laboratory 14 makes possible the study of decay processes which go to completion in only a few millimicroseconds. Fig. 1 shows the relative luminescence yield from a liquid solution containing benzene +M. BURTON AND H. DREESKAMP 67 terphenyl + air, as a function of time after switching on the beam from a 30 kV X-ray tube and switching it off about 1 1 millimicroseconds later.The rise-time of the X-ray beam and the fall-off time are each estimated to be about 10-9 sec. Attention is called to the fact that, after the beam is turned off, the decrease of luminescence fails to follow an exponential law for ca. 1 x 10-9 sec ; thereafter the fall-off is exponential over about two decades of intensity and the decay time, T [defined by I = I0 exp (- t / ~ ) ] , in this case appears to be 2.2 x 10-9 sec. A more careful calibration of the time scale may change the latter value by f 5 %. I I t, units of 10-9 sec FIG. 2.-Luminescence intensity as a function of time in various crystals and plastics. I is in arbitrary units. The samples are: (1) anthracene, (2) stilbene, (3) Sintilon, (4) an unidentified plastic.A variety of plastic scintillators, some of unknown composition, has also been tested by Dreeskamp, together with Ghosh and Yguerabide, in our laboratory and it has been found that no such scintillators show a simple first-order decay law. Fig. 2 displays the nature of some of the results. To be sure, all the plastic scintillators were characterized by the fact that the bulk of the energy absorption must have been in the major, essentially non-luminescent, component and that energy must have been transferred at some time to the luminescent component (perhaps, in some cases, components). However, fig. 2 also indicates that the processes measured by luminescence decay in this way, were not first-order decay processes of a single component. Fig. 3 shows the type of luminescence curve obtained for a quaternary solution of cyclohexane + benzene + p-terphenyl + oxygen.13 Luminescence rise and decay curves, such as shown in fig.1, have been obtained for the complete range of cyclohexane/benzene ratios shown in fig. 3. In their general features they rescmble fig. 1 but the decay times are different. The results are summarized in68 LUMINESCENCE DECAY TIMES fig. 4. Several facts are noteworthy. The decay times are functionally related to the relative concentrations of the two solvents but remain always first-order. 5 4 3 I 2 I I I I I 1 I 1 I I I 0 20 40 60 00 100 vol. % B FIG. 3.-Co6o-gamma-induced luminescence in air-saturated cyclohexane + benzene + p-terphenyl according to Nosworthy.13 I is luminescence intensity in arbitrary units.21 1 1 I I I I I I I I 0 50 100 vol. % benzene FIG. 4.-Luminescence decay time in aerated solution of cyclohexane + benzene + p-terphenyl. The decay time rises sharply in the region where the luminescence goes through a minimum and attains a maximum at low benzene concentration. The decay time in such solutions is about twice that in cyclohexane solutions.M. BURTON AND H. DREESKAMP 69 3. POSSIBLE INTERPRETATION OF LUMINESCENCE AGAINST TIME CURVES As already stated, this paper is concerned only with preliminary interpreta- tions. The region A-B in fig. 1 seems characteristic of all the luminescence- time curves and may be representative exclusively of the cut-off time of the X-ray tube. If it is, two alternative interpretations of the result lead to the opposed conclusions that the measured T is representative (a) of the decay time of the scintillator or (b) of the time involved in energy transfer.The first interpretation is that the time required for the processes C* + T -+ C + T* (7) i n cyclohexane solution and in benzene solution, as well as the possible succession of processes, B* -t T + B + T* C* + B -+ C + B* B* + T - t B + T* in cyclohexane + benzene solutions does not exceed 10-9 sec.* A change of T of the scintillator itself (as might be measured by u.-v. excitation with light which is absorbed by the scintillator alone) may be the result of one of two effects. (i) A change in T in such case could be the result of different solvent surroundings and would be revealed by a shift in the emission spectrum; un- published results of S.F. Pensabene indicate that small concentrations of benzene such as cause a sharp rise in T (cf. fig. 4) are without significant effect on the fluorescence spectrum. (ii) On the other hand, change in T of the scintillator might be the result of quenching of the scintillator. However, it has been shown 10 that, in the range of ca. 10-3 M quencher, only the solvent is effectively quenched. The result is confirmed by the data of Knau 15 for the system benzene + anthracene +nitrobenzene as quencher, irradiated with 3600A u.-v. or with less than 50 keV electrons, which indicate an effect on the solvent (or the energy transfer system) appearing at < 10-3 M quencher. An effect on the solute itself begins at 6 x 10-3 M quencher concentration.In contrast, benzene (present as second solvent) does not act as a quencher at all, as shown for low concentrations in fig. 4 by the sharp rise in T and for high concentration in fig. 3 by the increased luminescence. We are, therefore, impelled to an interpretation leading to the conclusion (b), much more interesting from the viewpoint of radiation chemistry. It is that the decay times shown by curves such as represented by fig. 1 (and given in fig. 4) are characteristic of the excitation transfer processes shown by reactions (7) or (9, or the combination (4), (3, and not of the light-emission process, Of course, this interpretation states nothing about the time required for the initially ionized and excited molecules to degrade to the low-excited states, C* or B*, represented in these equations; if the views already summarized (e.g.those of Kasha) are applicable, the latter time is ca. 10-13 sec. A more convincing demonstration that a quencher (to be distinguished from a second solvent like benzene) acts to decrease the life of excited solvent molecules is given by some unpublished results of S. F. Pensabene and A. Ghosh. The former has preliminary evidence that azulene cannot quench u.-v.-induced fluor- escence of p-terphenyl in cyclohexane solvent. The latter finds, also in preliminary experiments, that 1 /T varies approximately linearly (see fig. 5 ) with azulene * It should be noted that the possibility of an ionization-transfer mechanism being involved in the luminescence is specifically omitted on the basis of results reported in a forthcoming paper by S.Lipsky and M. Burton. T*-+T+hv. (6)70 LUMINESCENCE DECAY TIMES (quencher) concentration in such a solution, irradiated with Co6O-y, just as is required for normal quenching action. The reasonably admissible conclusion is that quenching itself involves the excited solvent directly. The results of fig. 5 show that the decay time of the excited state of p-terphenyl in this work is definitely less than 2 x 10-9 sec. It is a reasonable inference that quencher fails to act directly on the scintillator, in this case at least, because of its very short life. If such very short life is a general property of scintillators excited indirectly by high-energy radiation, a legitimate next question concerns the mechanism of two different effects: (a) the difference in luminescence efficiencies of various scintillators ; (6) the effect of concentration of scintillator on total luminescence.0.50 4 I 8 OI 4 ru 0 0.46 0.42 0 2.5 5.0 azulene, units of 10-4 M FIG. 5.-Reciprocal decay time as a function of azulene (quencher) concentration in a solution of 1.2g p-terphenyl per litre of cyclohexane, according to A. Ghosh. The quenching constant y appears to have a value in the range 320 to 440 mole-1. 3.1. EFFECT OF CHANGE OF SCINTILLATOR Preliminary data obtained by A. Ghosh for a solution of diphenylhexatriene in cyclohexane irradiated with 30 kV X-rays are similar to those shown in fig. 1 . However, the decay time, T, in this case is ca. 8 x lo-gsec, in contrast with T N 2.2 x 10-9 sec shown for p-terphenyl scintillator at the r ime concentration.It is known from studies of Co60-y induced luminescence that p-terphenyl is a more efficient scintillator than is diphenylhexatriene.16 The fact that the less efficient luminescence emission is associated with the longer decay time indicates that T represents the rate controlling (i.e. slowest step) in transfer of energy from excited state of solvent to scintillator. Although r might, on a naive basis of arithmetic alone, be representative of a longer life of excited cyclohexane, there is no apparent process by which the intrusion of a second component can prolong the life of a state. A simpler interpretation is that in the presence of a less effective energy acceptor, the energy transfer process itself is prolonged.The longer r is in such cases, the greater is the probability of interception of the excited state of the solvent by some type of energy degradation process. 3.2. EFFECT OF SCINTILLATOR CONCENTRATION According to Swank, Phillips, Buck and Basiie,l7 increase of p-terphenyl concentration in toluene solvent results in decrease of r to a limit determined byM. BURTON AND H. DREESKAMP 71 the decay time of the scintillator itself ; this figure as given by them is about 2.2 x lO-9sec. Fig. 1 and 4 of this paper taken together show likewise that T decreases with increasing scintillator concentration in benzene solution. The lower value shown is also 2.2 x 10-9 sec. However, the experiments with quencher present (cf. fig. 5 ) give T as low as 1.8 x 10-9 sec, which value must be 2 T for excited p-terphenyl. Thus, the effect of p-terphenyl is two-fold : to decrease T of the excitation transfer process and simultaneously to increase luminescence output," just as found by Swank et aZ.17 If the fraction of excited states of the solvent which terminate in luminescence of the scintillator is large, this effect is readily understood because decay time and luminescence yield would appear just on the basis of very simple kinetics to be reciprocally related.However, if that fraction is small, the effect of scintillator concentration or type on decay time must be explained for the most part as quenching of the excited solvent by the scintillator, as described by Birks and Cameron.18 In $4, a rough calculation indicates that but a relatively small fraction of excited molecules transfer their energy to the scintillator.Tests of the effect of p-terphenyl concentration on T in cyclohexane solution will assist in elucidation of the mechanism and in a test of Birks and Cameron's suggested quenching mechanism; in this case, inter- pretation of the results will be simplified because p-terphenyl is known not to protect cyclohexane from radiolysis.19 4. SIGNIFICANCE FOR RADIATION CHEMISTRY For multicomponent systems of the type here discussed, a variety of factors determines the probability that an initial excitation or ionization will result in an eventual emission of a photon as luminescence. For scintillator concentrations such as employed in this work we can assume all the initial energy absorption to be in solvent molecules; we can neglect also the trivial contribution to lumin- escence made by such molecules.ne = the number of scintillator molecules which emit luminescence, per unit of energy primarily absorbed in the solvent ; no = the number of solvent molecules primarily ionized or excited, per unit of energy primarily absorbed in the solvent ; #d = fraction of primarily ionized or excited molecules which do not complete the course of internal conversion to the lowest excited (singlet or triplet) state. The intercepting processes can include both chemical reaction and, otherwise undefined, physical quenching. The fraction which reaches the lowest excited state is then = 1 - $ d ; #t = fraction of molecules, in lowest excited state of solvent, which transfer energy to scintillator molecules.The fraction 1 - $t, which do not transfer energy in this way, may lose energy both in chemical reaction and in physical quenching. Thus, the nature and the quantity of impurities present affect the value of #t ; #e = probability that an excited scintillator molecule emits its energy as luminescence. The probability that an initial excitation or ionization results in a photon Let emission is, on this very simple model, given by An equation as simple as the above is restricted for useful application to one- solvent systems. In $ 1.1, we have seen that that portion of the Kasha excitation * The effect on luminescence output is, of course, an old and well-known effect in these t Note that ne/no 2: 2 x the fraction of absorbed energy re-emitted as luminescence.cases.72 LUMINESCENCE DECAY TIMES degradation 6 which can involve transition to repulsive states may terminate in decomposition before the lowest excited state is attained. Thus, $d can be a rather large quantity. In the specific case of pure cyclohexane, we can roughly estimate the lower limit of the yield of molecules initially ionized or excited from the relationship 20 50 50 Ic Ec’ G * & - - + - where Ic N 11 eV,21 the ionization potential, and Ec = 7.1 eV,22 the excitation potential-both in the gaseous state-are used in lieu of information regarding the liquid. Thus G* 2 11.5. Using the 100 eV yield of total gas,l G(gas) N 6, as representative of the decomposition of cyclohexane, it follows that $d can be as large as 0.5 if all the decomposition processes occur before the lowest excited states are reached.On the other hand, it is interesting to note that the value of gc(H2), the yield of H2 per 100 eV of energy actually absorbed in the cyclohexane, goes to a limiting value of 0.5 in mixtures containing increasing concentration of benzene.3 The yield of cyclohexane molecules decomposed per 100 eV of energy actually absorbed by the cyclohexane is then approximately given by where the last term, the yield of H atoms scavenged (in the same units) is estimated on the assumption 3*4 gc(H) - G(H) = 2.2, (4) and the value 2-2 is estimated from results with iodine or other scavenger.39 49 23 Thus, gc(-C6H12) N 2.7 and $d - 2.7/115 = 0.23 on the unlikely assumption that decomposition is complete before the lowest excited state is reached.On the other hand, $d - 0.5/11.5 = 0.04 if rupture into H atoms does not contribute at low cyclohexane concentration. Under Co60-y irradiation conditions ne/no I 0.02 for the systems employed in this work.* Also, evidence from u.-v. studies indicate that $e can equal ca. 0.5. for solutions of p-terphenyl in cyclohexane.24 Assuming that this $e value should be used and using $d = 0.04 to 0-5 and $e - 0.4 to 0.8 for Co60-y irradiation conditions, must lie in the range 0.04 to 0.08. Obviously, some future efforts in the study of these systems can well be addressed to better establishment of the value of $d but it is nevertheless apparent that $t is a small quantity. The presently most reasonable explanation of the results of fig.4 appears to be that it is the rate of the excitation-transmission process as reflected by t,ht which is affected by change in the solvent. The assumption in the use of eqn. (1) for estimation of $l is that the transmission process involves exclusively lowest excited states of the solvent. It is a reasonable conclusion from the Kasha effect and the considerations of the previous paragraphs that the same lowest excited states make the principal contribution to the chemical processes; i.e. to the actual radiolysis. It would simplify interpretation for radiation chemistry if one might didactically state that change in half-life of luminescence decay is the same thing as change in half-life of the excited state involved.Actually, the effects shown in fig. 4 probably represent a more complicated situation. Addition of a very small amount of benzene to a cyclohexane + p-terphenyl mixture vauses a sharp increase in T. The inference for radiation chemistry is that the observed effect is the result of transfer of excitation represented by reaction (4). The state produced has T greater than that in benzene + p-terphenyl. The implication is that isolated excited benzene molecules have larger half-lives than those in the immediate *This figure is based on an estimate 10 for the system benzene + terphenyl. It is definitely on the high side.M. BURTON AND H. DREESKAMP 73 environment of other benzene molecules. The mechanism and consequences of a presumed energy transfer, involved in such interaction, for radiation chemistry remain to be established.4.1. THE SCINTILLATOR AS AN INDICATOR It has been shown 19 that the presence of p-terphenyl in benzene has no de- tectable effect on the 100 eV yield of H2 when the liquid is irradiated with Co60-y rays. According to the suggested interpretation of the results in the present paper, the explanation is that only a small fraction of the excited states (repre- sented by i,!~t = 0.04 to 0.08) transfers energy to the p-terphenyl. Thus the latter is without effect on the major course of events (non-radiative deactivation, decom- position, etc.) in which the lowest excited states of the solvent can be involved. In other words, the scintillator, in luminescent solutions of the type here discussed, may be used as an indicator by which one may examine the life of such states.4.2. CHEMICAL EFFECTS OF THE QUENCHER A related conclusion refers to the effect of quencher on the decay times, a factor which we have not yet been able to study in detail. It may be anticipated that a quencher will affect T and, in mixed solvents, will affect 7 of each solvent in- dependently. While we are not prepared to say what the effect will be on the radio- lysis process, it appears reasonable to expect that such an effect (i.e. as a quencher, not merely as a chemically reactive agent) will ultimately be found-if not on the overall results, certainly on the detailed elementary processes. It is interesting in this connection that Bach and Sorokin 25 report that oxygen, a typical quencher, increases yield of H2 in radiolysis of ethanol. 1 Manion and Burton, J, Physic. Chern., 1952, 56, 560. 2 Burton and Patrick, J . Physic. Chem., 1954, 58, 421. 3 Burton, Chang, Lipsky and Reddy, Radiation Research, 1958, 9, 203. 4 Meshitsuka and Burton, Radiation Research, 1959, 10,499. 5 Burton, Hamill and Magee, Proc. Second Int. Congress Peacefitl Uses of Atomic 6 Kasha, Faraday SOC. Discussions, 1950, 9, 14. 7 Wilson and Noyes, Jr., J. Amer. Chem. SOC., 1941, 63, 3025. 8 Krassina, Acta physiochim., 1939, 10, 189. 9 Kallmann and Furst, Physic. Rev., 1952, 85, 816. 10 Burton, Berry and Lipsky, J. Chim. Physique, 1955, 52, 657. 11 Berry and Burton, J. Chem. Physics, 1955, 23, 1969. 12Berry, Lipsky and Burton, Trans. Faraday SOC., 1956, 52, 311. 13 Nosworthy, Magee and Burton, Radiation Research, 1958, 9, 160; also, forth- 14 Dreeskamp and Burton, Physic. Rev. Letters, 1959, 2, 45. 15 Knau, Z. Naturforschung, 1957, 12a, 881. 16 P. J. Berry, Thesis (University of Notre Dame, 1955). J. L. Kropp, unpublished 17 Swank, Phillips, Buck and Basile, IRE Transactions Nuclear Science, 1958, NS-5, 18 Birks and Cameron, Proc. Physic. SOC., B, 1958, 72, 53. 19 Burton and Patrick, J. Chem Physics, 1954, 22, 1150. 20 cf. Burton and Kurien, J. Physic. Chem., 1959, 63, 899. 21 Hustrulid, Kusch and Tate, Physic. Rev., 1938, 54, 1037, 22 Pickett, Muntz and McPherson, J. Amer. Chem. Soc., 1951, 73, 4862. 23 Dewhurst, J. Physic. Chem., 1959, 63, 813. 24 Bowen and Williams, Trans. Faraday SOC., 1939, 35, 765. 25 Bach and Sorokin, Sbornik Rabot Radiatsonnoi Khimi, Akad. Nauk., U.S.S.R., 1955 1, 163 ; English translation : Symposium on Radiation Cheniistry, Acad. Sci. U.S.S. R., 1955, 1, 135. Energy, 1958. coming publication. work. 183.