704
ROBERT LIVINGSTON AND V. SUBBARAO
Vol. 63
A FURTHER STUDY OF THE PHOTOCHEMICAL AUTOOXIDATION OF ANTHRACENES BY ROBERT LIVINGSTON AND V . SUBBARAO Contribution from the Diwision of Physical Chemistry, Institute of Technology, University of Minnesota, Minneapolis, Minn. Received November 10, 1068
The quantum yields of the photoautooxidation of anthracene and diphenylanthracene were measured in several solvents. The results of these measurements demonstrate that the fluorescent state as well as the triplet state of the hydrocarbon can 2 to form an intermediate, leading to the formation of a peroxide. In strongly fluorescent solutions (e.g., direact with 0 phenylanthracene in benzene), the interaction of the fluorescent state with 02 is the principal step. For weakly fluorescent solutions, the triplet state plays the dominant role. Diphenylanthracene sensitizes the oxidation of anthracene. This process was studied by illuminating a solution, containing both substances with light which was absorbed by diphen 1anthracene only. An analysis of these measurements indicates that the dacltsky mechanism, which involves molecurar oxygen in a singlet state as a reaction intermediate, is not compatible with our results.
The photochemical autooxidation of anthracene test of the reliability of the proposed mechanism of and its derivatives has been studied quantitatively these reactions. and extensively by Bowen and his c o - w o r k e r ~ . ~ ~ ~ Experimental They determined the quantum yields of fluoresMaterials .-Benzene and carbon disulfide were highcence, dimerization and autooxidation of several anthracenes in a variety of solvents and over a grade commercial samples and were used without further Bromobenzene was twice distilled under niwide range of hydrocarbon concentrations. How- purification. trogen. Quinone was sublimed shortly before it was used. ever, their measurements were limited to oxygen The final purification of anthracene and diphenylanthracene concentrations corresponding to solutions saturated was accomplished chromatographically on activated aluwith pure 02,.air or pure N2, each a t 1 atm. The mina. Flash-photolytic Measurements.-A few preliminary present experiments include the results of measure- measurements of the half-life of the triplet state of diphenylments of the oxidation of diphenylanthracene in anthracene and of the quenching of the triplet of anthracene benzene at oxygen concentrations ranging from 1.00 by quinone were made, using the apparatus and technique by Livingston and Tanner .6 X lou2 t o 2.5 >( M and of anthracene in described Measurements of Photochemical Auto6xidation.-A bromobenzene a t concentrations from 8 X lo+ modification of the method of Bowen and Tanner2 was used to 8 X M . In addition, the effect of the addi- to measure the rates of photochemical autooxidation. tion of small amounts of quinone upon the rate of Two 3 -00-ml. samples, containing identical concentrations autooxidation of anthracene was studied. The of anthracene or diphenylanthracene, were exposed simulto the complete radiation (filtered only by a 5photochemical autooxidation of anthracene sensi- taneously cm. layer of a 2% CuSO, solution) from a GE AH5 mercury tized by diphenylanthracene also was investi- arc. One sample, which was free from quenchers and was in equilibrium with air a t 1 atm., was used as the reference gated. The ratio of the change in anthracene concentraE. J. B o w e ~ Phas demonstrated that, under solution. tion in this solution to that in the other solution, which certain conditions, the oxidation of anthracene is contained a quencher or was in equilibrium with a different, the consequence of a reaction between an oxygen known concentration of oxygen, was a measure of the effect molecule and a molecule of the hydrocarbon in its of the variable upon the rate ( L e . , the quantum yield) of lowest triplet state; he has suggested that the photooxidation. reaction cells were Pyrex cylinders, about 4.0 cm. photochemical autooxidations of all anthracenes in The diameter and 10.0 cm. long. At each end, the cylinders proceed, always, by way of this step. Our results were fitted with projecting axially-centered tubes, 0.6 cm. in show that the interaction of an oxygen molecule diameter and 5 cm. long. One such tube was sealed: the with one of the hydrocarbon in either its triplet or other ended, for the reference cell, in a male T joint, .which be closed by the external member of the joint without fluorescent state can induce the formation of the could danger of contaminating the contents of the cylinder with peroxide. I n strongly fluorescent solutions (e.g., (silicone) lubricating grease. Cells which were used with diphenylanthracene in benzene) the important air or oxygen at pressures other than 1 atm. had, in place intermediate is the fluorescent state, while in of the seal, a constriction followed by a T joint, by which could be connected to an ordinary vacuum line. T o practically non-fluorescent solutions of anthracene they fill the cylinder with gas at a known pressure, it was clamped in bromobenzene the dominant step is the inter- in a vertical position and connected to the vacuum line. action of an oxygen molecule with one of anthracene The 3.00-ml. sample was pi etted into the small, closed-end tube. The solution was txen frozen with liquid nitrogen in its triplet state. and tho cylinder evacuated. The solution was degassed by The publication of the results of measurements of successive freezing, pumping and thawing; after which, the half-life of the triplet state of anthracene and of with the solution frozen and the cylinder at room temperathe rate constant for its bimolecular reaction with ture, gas was admitted t o the desired pressure. The cylinwas then sealed a t the constriction and separated from oxygen4vs offers the opportunity for an additional der the vacuum line. Since the volume of the solution was (1) This work was in part supported by grants from the National Scienoe Foundation (NSF-G 1449) and from the Graduate School of the University of Minnesota, for which the authors are grateful. (2) E. J. Bowen and D. W. Tanner, Trans. Faraday SOC.,61, 475
(19.55). (3) (a) E. J. Bowen, Disc. Faraday SOC.,14, 143 (1953); 6)E. J. Bowen, TTans. Faraday Soc., 60, 07 (1954). (4) G. Porter and M. W. Windsor, Disc. Faraday Soc., 17, 178 /1964);Pro& Boy. 800. (London), 24158, 238 (19581.
less than 2.5% of that of the gas in the reaction vessel, the oxygen was seriously depleted only in those few experiments where its initial partial pressure was less than 1 mm. (see Tahle I). The two cylinders were supported side by side on a set of three horizontal, rubber-covered rollers and were rotated
(5) R . Livingston and D. W. Tanner, Trans. Faraday Soc., 6 4 , 765 (1958).
c
June, 1959
PHOTOCHEMICAL AUTOXIDATION OF ANTHRACENES
about their axes at about 150 r.p.m. The rotation of the cylinders spread the solutions on their inner walls as thin, continuously renewed films. Bowen and Tanner demonstrated,* and it was confirmed under our experimental conditions, that the oxygen in the solution was maintained in equilibrium with the gas phase by this procedure. A series of calibrating experiments showed that the rate of oxidation was sensibly independent of what cylinder was used and of which position on the rollers it occupied. The solutions were analyzed spectrophotometricaIly, before and after irradiation. Since the solutions were much too concentrated to be analyzed spectrophotometrically in a cell of normal thickness, samples were diluted with the aid of a calibrated micro-dilution pipet. The resulting solution was further diluted to 10 ml. and its optical density determined with a Cary spectrophotometer using matched 2.00-cm. silica cells. For the solutions used in these exmeasurement periments, the complete analysis (dilubion of optical density) was reproducible t o better than 0.5% I n a few cases, the decrease in anthracene concentration was checked using Bowen and Tanner’s analysis for anthracene peroxide; however, in our hands this method proved to be less precise than the spectrophotometric measurements.
+
Experimental Results The Photoiixidation of Anthracene in Bromobenzene.-The results of measurements of the rate of disappearance of anthracene, in bromobenzene containing various concentrations of oxygen, are summarized in Table I. Measurements were made of the change of concentration of anthracene in pairs of reaction vessels, as is described in the preceding section of this paper. I n each pair of vessels, the concentration of anthracene was initially the same and the vessels were exposed simultaneously to light of the same intensity for the same length of time. The solubility of oxygen in bromobenzene is not known. The tabulated concentrations of oxygen were calculated upon the arbitrary assumption that bromobenzene saturated with air a t 1 atm. contains 2.0 X M oxygen, which is approximately the solubility of oxygen in benzene. In contrast to the report of Bowen and Tanner12 we observed an appreciable photodimerization of anthracene in bromobenzene. The rate of disappearance of anthracene from an illuminated, oxygen-free solution was about one-fourth that observed in the reference solution, but the standard analytical method2 fails to indicate the presence of any peroxide in such an anaerobic solution. Direct evidence for dimerization was obtained. A 3 X loF2 M deoxygenated solution was illuminated for an hour. The resulting crystalline precipitate was filtered, washed and dissolved in benzene. This solution was transparent for light of wave lengths greater than 3000 A., but showed an absorption maximum a t 2830 A. Its absorption spectrum was similar to that reported for the anthracene dimer by Noland.6 It is apparent from the data listed in the first five columns of Table I that the rate of the photochemical reaction of anthracene in bromobenzene is practically independent of the oxygen concentration, in the range from 8.2 X to 8.3 X M . At lower concentrations, the rate of peroxide formation decreases rapidly. The values in column six have been (approximately) corrected for the effect upon the rate of the differently changing concentrations of anthracene. The rates of both dimerization and peroxide (6) W. E. Noland, J . Am. Chsm. Soc., 18, 188 (1956).
795
TABLE I EFFECT OF OXYGENCONCENTRATION UPONT H RATE ~ OF FORMATION OF ANTHRACENE PEROXIDE IN BROMOBENZENE
2;:
AlAl 108
[Oib,X
x
(air sat.)
-Ratio
Obsd.
of rateCor. Calcd.
1.00 625.5 82.0 1.40 0.93 0.93 1.03 1.00 97.6 12.8 2.64 1.03 1.03 0.99 1.00 6.0 .79 2.09 0.97 0.97 .98 1.00 2.4 .31 1.96 .01 .91 .96 0.36 1.7 .22 1.58 1.00 1.00 .96 0.36 0.6 .083 0.93 0.91 0.91 .96 .69= 1.05 0.26 ,0260 4.51 .33 .27 1.03 00 0 4.56 .26 -22 .22 1.03 Ob 0 4.47 .24 .20 .22 Ob 0 4.60 .26 .22 .22 1.03 0 Oxygen removed by successive freezing, um ing and melting. Vessel finally filled with purified I?z BOxygen . removed as stated above. Vessel sealed off while evacuated. C T h e total number of moles oxygen present is several-fold less than the number of moles of anthracene consumed. An approximate calculation indicates that the average concentration of oxygen was 1.6 X 10-6 M. In calculatin a value for the ratio of the rates, a constant value of [&I = 1.5 X 10-8 M was used.
formation increase with increasing concentration of the hydrocarbon. For the peroxide formation in bromobenzene, the rate is, within the limits of precision, directly proportional to the anthracene concentration, a t least to 0.02 M . I n the last four experiments listed in Table I, the disappearance of anthracene was much faster in the reference vessel than in the other vessel. To correct approximately for this difference, the observed ratio of the rates has been multiplied by the ratio of the average concentrations in the reference and reaction vessels, respectively. The Photooxidation of Diphenylanthracene in Benzene.-The photochemical properties of 9,lOdiphenylanthracene in benzene differ markedly from those of anthracene in bromobenzene. Diphenylanthracene is not detectably photodimerized.2 Its fluorescence has a high yield (about 0.8 in benzene) and is not measurably self-quenched. In view of these differences, it is not surprising that, unlike anthracene in bromobenzene, its rate of photooxidation is strongly dependent upon the concentration of oxygen. Evidence for this dependence is summarized in Table 11. Since for these experiments the decrease in the concentration of hydrocarbon was quite different in the reference and reaction vessels, the observed ratio of the rates were corrected, approximately, by multiplying the observed values by the ratio of the average concentrations of the hydrocarbon in the reference and reaction vessels. The corrected values are given in column six. Carbon disulfide efficiently quenches the fluorescence of the anthracenesa but has little effect upon the half-life of the triplet state of anthracenen6 I n terms of a possible mechanism for these reactions (which is discussed in a later section of this paper) i t is to be expected that the photooxidation of diphenylanthracene should be practically independent of oxygen concentration (except a t very low concentrations) in solutions containing carbon
ROBERTLIVINGSTON AND V. SUBBA RAO
796
Vol. 63
TABLE 11
renders any quantitative analysis of their results
BENZENE
anthracene with Iight from a mercury arc after it had passed through 2 cm. of a concentrated solution of anthracene in toluene. Since the anthracene in the filter was rapidly dimerized and oxidized, it was renewed before each measurement. Under these conditions, pure anthracene in benzene did not react detectably. In mixtures, both anthracene and diphenylanthracene were consumed. The reaction mixtures were analyzed spectrophotometrically. The concentration of diphenylanthracene was calculated from the optical density measured at 3930 8.,at which wave length the absorption of anthracene is negligible. Optical densities were also measured at shorter wave lengths, corresponding to three maxima and two minima of the anthracene absorption spectrum. Knowing the diphenylanthracene concentration and the extinction coefficients of each hydrocarbon, a value for the anthracene concentration can be calculated from the measurements at each of the five wave lengths. The concentration was taken as the average of five such determinations. We found this to be a more reliable and reproducible way of measuring the anthracene concentration in the mixture than the use of the iodimetric method2 of analysis for anthracene peroxide. In each experiment two tubes were exposed side by side. They both contained diphenylanthracene a t the same concentration but one had, in addition, a known concentration of anthracene. All solutions were saturated with air. The results of such measurements are presented in Table V. The values of A [A]/A [A&.] which are listed in column three are the ratios of the changes in the concentrations of anthracene and diphenylanthracene, respectively, based upon the analysis of the solution contained in the reaction cell. The ratios A[A+2]o/A[A$,], listed in column five, were obtained by dividing the decrease in the concentration of diphenyIanthracene in the reference cell by the corresponding decrease which occurred in the reaction (ie., mixture-containing) vessel. It is noteworthy that the values of A[A42]o/ A[A@2]are all close to unity; in fact, their average is 0.98 f 0.06. For the range of concentrations investigated, these data demonstrate that the addition of anthracene to a diphenylanthracene solution has little, if any, effect on the rate of disappearance of diphenylanthracene, although the anthracene itself is photooxidized.
THEEFFECT O F OXYGEN CONCENTRATION UPON THE RATE difficult. To avoid this difficulty, we illuminated solutions (in benzene) of anthracene and diphenylOF FORMATION OF DIPHENYLANTHRACENE PEROXIDE IN A[AQel [A+zI
X 102
Pol, mm.
[O;d3X
x
10s
(in air)
-Ratio Obsd.
of the ratesCor. Calcd.
1 . 0 3 1049.0 10.05 1.59 2.27 2.51 2.38 1.02 652.1 6 24 1.64 2.11 2.28 2 . 0 1 1 . 0 4 619.3 5 . 9 3 1.84 2.02 2.24 1.96 1.05 314.8 3.01 2.32 1.28 1.36 1.46 1 . 0 5 205.7 1.97 1 . 9 5 1.32 1.36 1.17 1.07 140.0 1.34 2.39 1 . 0 3 1.06 0 . 9 6 0.98" 58.6 0 . 5 6 1.35" 0 . 5 2 0.50 .63 1.03 58.5 .56 1.69 .53 .49 .63 1.04 26.1 .25 1 . 8 6 .33 .32 .47 a The intensity of the incident light was reduced about 2.5-fold by the use of a neutral fiIter (screen) and the time of illumination was increased to compensate partially for this change.
disulfide. The data of Table 111 confirm this prediction. TABLE I11 THEEFFECT OF OXYGENCONCENTRATION UPONTHE RATE OF FORMATION OF DIPHENYLANTHRACENE PEROXIDE IN BENZENE CONTAINING CARBONDISULFIDE Vol. % of CSe
[Arne I x 101
Poe, mm.
A[ACl x 10s (in air)
Ratio of the rates (obsd.)
1
1.02 1.02 1.02 1.06 1.02 1.02
602.0 10.7 4.9 642.0 9.9 4.2
3.20 3.15 3.02 3.65 3.14 3.38
1.14 0.98 1.00 1.01 1.00 0.88
1 1 9 9 9
Effect of Quinone upon the Photolixidation of Anthracene.-Using the apparatus and technique of Livingston and Tanner6 we have shown that p benzoquinone quenches the triplet state of anthracene about as efficiently as oxygen. In bromobenzene, the bimolecular quenching constant is approximately 1.6 X lo9 set.-'. A similar value has been reported' for the quenching of the triplet state of chlorophyll a by quinone. Schenck observeds that quinone inhibits photoautooxidations sensitized by a number of dyes and pigments and interprets this effect as the consequence of the quenching of the triplet state by quinone. As is illustrated by the data of Table IV, quinone has a similar retarding effect upon the photochemical autooxidation of anthracene in bromobenzene. However, it should be remembered that the observed rate of disappearance of anthracene is the sum of the rates of photooxidation and dimerization. The PhotoSxidation of Anthracene Sensitized by Diphenylanthracene.-It has been demonstrated by Bowen and Tanner2 that in solutions containing both anthracene and diphenylanthracene either substance can sensitize the photooxidation of the other. Most of their measurements were made with systems in which each substance absorbed an appreciable fraction of the incident light. This fact (7) E. Fujimori and R . Livingston, Nature, 180, 1036 (1957). (8) G. Sohenok and K . Kinkel, Nalurmissen., 38, 355 (1951); G. Sohenok and K. H. Ritter, ibid., 41, 334 (1954).
Discussion The following generalized mechanism is consistent with the measurements of the maximum yield of fluorescence, the quenching of fluorescence, the natural life and quenching of the triplet ~ t a t e , ~ , ~ the photodimeri~ation,~~~ the photoaut~oxidation~~~ and the photosensitized oxidation of the anthracenes. The roles played by the fluorescent and lowest triplet states (H* and H',respectively) are well established and are represented by the first seven steps of the mechanism. The third intermediate, X, has not been detected directly and its
r
J
PHOTOCHEMICAL AUTOXIDATION OF ANTHRACENES
June, 1959
797
TABLE IV
EFFECT OF QUINONE UPON IQ1/ P a l 1.0 2.1 2.1 3.1 5.2 5.2 6.9 9.5 10.4 10.4
[021 A 106
[.4] X 102
9.60 9.60 9.60 9.60 9.60 9.60 2.88 2.11 1.92 1.92
3.20 1.63 3.18 3.22 3.15 3.25 3.27 3.20 2.84 3.14
THE PHOTOOXIDATION OF
-RateRef.
1.14 0.67 1.10 1.20 1.19 1.21 0.78 .68 .44 .50
ANTHRACENE IN BROMOBENZENE
With Q
-Ratio Obsd.
= RRef/RQ-
Cor.
Calcd.
0.72 .32 .56 .57 .41 .48 .36 .22 .I9 .20
1.58 1.97 1.96 2.14 2.90 2.52 2.16 3.09 2.32 2.50
1.61 2.05 2.05 2.21 3.03 2.61 2.16 3.09 2.33 2.56
1.51 1.90 1.87 2.15 2.52 2.52 2.72 2.95 3.05 3.05
hv+H+H* H* +H H* -+ H' H*+H H H* +2H H H* +Hz H' +H
TABLE V THE PHOTOOXIDATION OF ANTHRACENE SENSITIZEDBY DIPHENYLANTHRACENE IN BENZENE ld2.41
X 102
[AI/[Adel
1.05
1.42 1.44 2.44 3.02 3.36 3.75 4.03 4.96 5.36 5.78 5.82 6.33 7.52 7.78 8.15 15.08 15.30
1.11
0.34 1.06 1.09 1.12 0.66 1.22 1.08 1.09 0.87 .I4 .27 .68 .65 .34 .34
AIA1/A[%!!d, Obsd.
0.17 .36 .46 .42 .54 .86 .75 1.31 1.53 1.44 1.05 2.15 1.51 1.72 1.62 2.13 2.51
0.30 .32 .51 .62 .70 .76 .80 1.00 1.08 1.16 1.16 1.25 1.47 1.55 1.58 2.64 2.68
+ hu~
AfAd21°/AIA 21 Obsd. Catd.
0.94
...
...
1.03 0.90 .91 1.02
1.00
... *..
...
1.11
1.02 1.00 0.92
1.02 0.95
--j. X OzfH'+X X+02+H H X +H KO2 R X +H ROz
0 2
1.01 1.01 1.01 0.98
...
...
...
...
...
1.02
0.98
...
...
... ...
0.91
.88
+ + + H*+ +
+
+
+
(1) (2) (3) (4) (5) (6) (7) (8) (9) ( 10) (1W (11r)
Values of k2 were calculated by B ~ w e from n ~ ~the integrated extinction coefficients. Values of ka k4 were derived by combining k2 with the maximum3b quantum yields of fluorescence. The ratio, k3/(k3 k4), was estimated from the limiting, maximum yields of photooxidation.2 The sum, lcg k6, was obtained from the self-quenching constant for f l ~ o r e s c e n c eand ~ ~ ~ICz.~ The maximum yield of dimerization2 was used to calculate k6/(k5 k6). The value of k7 was obtained directly by flash-photolytic measurement^.^^^ k8 was derived from the Stern-Volmer quenching constant for oxygen3b; lcg, from flash photolytic ~ t u d i e s . ~ ' ~ The ratio, klo/kll, was obtained from the variation of the yield of photooxidation with the concentration of the hydrocarbon.'
+
+
+
+
nature is in dispute. It has been postulated that it is either a reactive, labile moleoxide, 02H',2s9or else an oxygen molecule in a metastable singlet state.2v10 An alternative mechanism, involving a labile reactive dimer of the hydrocarbon," can be rejected in terms of kinetic evidence2 and of the measured properties4v5of the triplet state. TABLE VI The kinetics of the autooxidations are consistent SUMMARY OF THE VALUESOF THE RATECONSTANTS with either of the postulated descriptions of the HydroAnthracene Diphenylcarbon Anthracene Bromoanthracene intermediate X. However, u e believe that the solvent Benzene benzene Benzene measurements of the photochemical oxidation of 7.4 x 107 7 . 4 x 10' 1 . 3 X 108 anthracene sensitized by diphenylanthracene are 2 . 4 X 108 4 . 0 X IOg 3.2 x 107 incompatible with the view that X is a metastable, 0.67 0.90 0.5 singlet state of 02. 5 x 109
