Pre-annihilation electrochemiluminescence

EIectrochitnica Acta. 1968. Vol. 13. pp. 1209 to 1220. Paattton

PRE-ANNIHILATION D.

Pms.

Pkned

in Nonhcm

Inland

ELECTROCHEMILUMINESCENCE*

L. MARICLE, A. ZWEIG, A. H. MAURER and J. S. BRINEN American Cyanamid Company, Stamford, COM., U.S.A.

Abstract-Electrochemihuninescence generated by alternating electrolysis of solutions of fluorescent aromatic hydrocarbons is attributed primarily to radical-anion-cation annihilation. However, light can also be detected while oxidizing the anion or reducing the cation of certain molecules at potentials insufficient to generate the oppositely charged radical. The mechanism of the pm-annihilation ccl has been studied for two aromatic hydrocarbons. The light pulse produced while oxidizing the rubrene anion in a double-potential-step experiment was found to be singlet emission (spectrum identical to fluorescence) even though the overnotentials applied were considerably less th&r the 2.3 eV singlet energy of this molec;le. Possible energy doubling mechanisms are discussed. Most of the nre-annihilation light eenerated from the rubrene anion icshown to result from the oxidation of &r anion decomp&iti& product, followed by a homogeneous chemiluminescent reaction. Phenanthrene, unlike most molecules, exhibits preamihilation luminescence which is not the same colour as its fluorescence (violet). Oxidation of the anion at overvoltages of about 2.6 V produces a green emission with a maximum between 4900 and 5000 A. This corresponds closely to the published room temperature solution phosphorescence. The emission is quenched by appropriate triplet quenchers. This is believed to be the first positive evidence for electrochemically generated triplets. R&urn&-L’Clectrochimiluminescence produite par l’electrolyse alternative des solutions d’hydrocarbones aromatiques fluorescents est attribuCe essentiellement B l’annihilation mutuelle d’anions et cations radicalaires. Cependant, la lumiere peut aussi etre d&e&e pendant l’oxydation d’anions ou la reduction de cations de certaines molecules a des potentiels insuffisants pour produire des radicaux de charges oppos&es. Le. rn6canisme de cette pr&annihilation ccl a CtC ttudie pour deux hydrocarbones aromatiques. L’impulsion de lumiere produite en oxydant l’anion rub&e dans une experience de double saut de potentiel a Cte reconnue dtre l’emission dun singlet (spectre identique a celui de fluorescence), bien que les surtensions appliquees aient considerablement btb moindres que celle correspondant a l’energie de singlet, soit 2,3 eV de cette molecule. Discussion des possibilitib d’energies de double mCcanisme. 11est montre que la plus grande partie de la lumiere de prtannihilation engendr6e par l’anion rubrene resulte de l’oxydation d’un produit de d6composition d’un anion, suivie d’une reaction de chimiluminescence homogene. Le phenanthrene, contrairement a la plupart des mol&ules, manifeste une luminescence de pre-annihilation qui n’est pas de mt?me couleur que celle de sa fluorescence (violette). L’oxydation de l’anion a des surtensions exc6dant 2,6 V produit une emission verte, avec un maximum entre 4 900 et 5 000 A. Ceci correspond exactement aux don&s bibliographiques sur la phosphorescence de la solution a la temp6rature du laboratoire. L’emission est eteinte par des extincteurs de triplets appropribs. Ceci est consid&& comme la premiere preuve effective de triplets engendres &ctrochimiquement. Zusannnenfassung-Die bei der wchselweisen Elektrolyse der Losungen fluoreszierender aromatischer Kohlenwasserstoffe auftretende Elektroluminiszenz bird vor allem einer Annihilierung zwischen Radikal-Anionen und -Kationen zueeschrieben. Lichtemission kann iedoch such bei der Oxvdation des Anions bzw. der Reduktion de: Kations von gewissen Molekiiien bei Potentialen beobachtet werden, welche fur die Bildung des entgegengestzt geladenen Partners ungentigend sind. Man untersuchte den Mechanismus der Vorannihilierungs-ECL an zwei aromatischen Kohlenwasserstoffen. Es konnte festgestellt werden, dass der bei der Oxydation des Rubren-Anions in einem DoppelPotentialstufenexperiment entstehende Lichtimpuls durch Singlett-Emission bedingt ist (Spektrum identisch mit dem Fluoreszenzspektrum), obschon die verwendeten Uberspannungen bedeutend kleiner sind als die 2,3 eV der Singlett-Energie dieses Molekiils. Man diskutiert mogliche Mechanismen einer Energieverdoppelung. Der Hauptteil der Vorannihilierungs-Lichtemission des Rubrenanions wird als von einer Oxydation eines Zerfallsproduktes des Anions herrtihrend ermittelt, gefolgt von einer homogenen Chemiluminiszenzreaktion. Phenanthren zeigt im Gegensatz zu den meisten * Presented at the 18th meeting of CITCE, Elmau, April 1967; manuscript 1967. 1209

received 23 August

D. L. MARICLE, A. ZWEIG,

1210

MAURER and J. S. BRINEN

A. H.

Verbindungen eine Vorannihilierungs-Luminiszenz von anderer Wellenhinge als seine Fluoresxenz (violett). Die Oxydation des Anions bei ca. 2,6 V ist von einer Emission im griinen Spektralbereich begleitet, deren. Maximum zwischen 4900 und 5000 A hegt. Dies entspricht weit gehend der publixierten Phosphoreszenx der Liisung bei Raumtemperatur. Die Emission lasst sich durch entsprechende Triplet&Quencher unterdrticken. Diese Tatsache wird als erster positiver Beweis fiir die elektrochemische Triplett-Erxeugung gewertet. INTRODUCTION

of luminescence resulting from electrolysis of solutions of fluorescent aromatic molecules have recently been published.1-5 In most cases, the luminescence has been attributed to a radical-anion-radical-cation annihilation reaction which leaves one of the resulting neutral molecules in an excited state, Swmu

reports

R++R’-+R*+R, R* + R + hv.

(1) (2)

Alternative explanations for light-emitting reactions seen under some conditions have involved luminescent oxidation of the anion by either cation decomposition product$ or solvent oxidation products. 4 However, light can also be generated under conditions such that none of the above explanations is pertinent. Specifically, with certain aromatic hydrocarbons light can be detected while the anion is oxidized or the cation reduced at potentials insufficiently positive or negative, respectively, to generate the oppositely charged radical-ion and at potentials where there is no appreciable background electrolysis.‘js The behaviour of rubrene,6 a highly electrochemiluminescent molecule, typifies this kind of pre-annihilation luminescence in that only singlet emission is observed. That is, the electrochemiluminescent (ccl) emission spectrum is identical to the fluorescence emission spectrum. Recently, however, we reported in a preliminary communication’ that pre-annihilation emission from phenanthrene occurs at distinctly longer wavelengths than the normal singlet emission, and we attributed this to emission from electrogenerated phenanthrene triplet molecules. This paper presents the details of the observations made on the pre-annihilation ccl of both of these molecules and considers the problem of energy doubling in ccl processes. EXPERIMENTAL

TECHNIQUE

Reagents Reagent grade DMF was heated over NaOH at 90°C for several hours, passed through a column of 4A molecular sieves and then fractionated under vacuum. The resulting material contained ca 40 ppm H,O. The acetonitrile was stored over Na&O,, decanted, distilled from KMNO,. acidified with H,SO, and distilled again, and distilled a final time from P,Or,. This solvent contained 10-20 ppm H,O. Polarographic grade tetrabutylammonium perchlorate was obtained from Southwestern Analytical Chemicals and dried with our P,O,. The phenanthrene was purified by the method of Kooyman and Farenhorst. s The rubrene was recrystallized from degassed heptane in subdued light. 1,3,5-trans-hexatriene and 2,3-dimethylbutadiene were purified by distillation. The biphenyl was recrystallized from ethanol.

Apparatus

and procedures

An operational-amplifier-based potentiostat with a Krohn-Hite DCA-10 wide band ac-dc amplifier employed as a booster was used for the cyclic voltammetry and

Preannihilation

electrochemiluminescnce

1211

double-potential-step experiments. The light pulse resulting from the double-potentialstep experiments was recorded with an Aminco photomultiplier microphotometer equipped with a lP21 photomultiplier, and a Tectronix 502A dual beam oscilloscope with a Polaroid camera. The oscilloscope was connected directly to the output of the photomultiplier, thus bypassing the relatively slow Aminco-amplifier. The voltage wave-form appearing between the reference and working electrode was recorded simultaneously with the light pulse to make sure no ringing or overshoot occurred. The phenanthrene pre-annihilation emission spectrum was photographed by focusing a 1: 1 image of the electrode on the slit of a fast grating Raman spectrographlo using Eastman Kodak 103 af spectrographic plates. The plates were microphotometered on a Jarrell Ash Model 23-100 recording microphotometer and were corrected for the spectral sensitivity of the emulsion. The potential of the platinum working electrode was cycled continuously between -2.6 and $-l-O V for several minutes while the spectrum was recorded. The square-wave voltage input to the potentiostat for this and the double-potential-step experiment was derived from a Hewlett-Packard 202A signal generator. In carrying out the triplet quenching experiments a cyclicvoltammogramwas runon the same electrode and in the same solution as was used for the light-intensity measurement, both before and after the quencher was added. In this way it was possible to ascertain readily whether the triplet quencher had any effect on the electrochemical process or the stability of the radical-ions. The cell employed is shown in Fig. 1. It is designed to contain 50 ml of test solution with the liquid level well below the O-ring seal. Figure 1 is generally self-explanatory. The final non-aqueous salt-bridge compartment terminates with an ultrafine glass frit Nitrogen purging blanketing inlet

and

n Working

,Nitrogen

outlet non-aqueous bridge comportment

electrode

L__--_-I FIG. 1. Electrochemical cell.

D. L. MARICLE, A. ZWEIG, A. H. MAURERand J. S. BRINEN

1212

and is filled with the test solution. An intermediate non-aqueous salt bridge filled with DMF containing 1.0 M TBAP connects the aqueous see to the final salt-bridge compartment. The intermediate bridge uses a thin film of polyvinylalcohol as an ion-permeable membrane at the non-aqueous end. This film has been found to provide an inert, moderate resistance, separator useful in a wide variety of non-aqueous solvents.ll For experiments which resulted in appreciable electrolysis, eg the purging experiments, the cell bottom shown in Fig. 3 was replaced with one of the usual Hcell design employing double glass-frit separators. The working electrodes were a O-025-in dia. platinum wire 1 cm in length used for the cyclic voltammetry and the double-potential-step experiments and a 3 x IO-mm platinum foil electrode used to obtain the phenanthrene emission spectrum. This latter electrode was coated on one side with RTV-60 (General Electric) so that only the side facing the spectrograph was exposed to the solution. The purified nitrogen was passed through a hot copper scrubber to remove traces of OS and only glass tubing was used between the scrubber and the cell (except for short polyethylene connectors). All solutions were prepared and transferred to the cell in a dry nitrogen atmosphere. RESULTS

AND

DISCUSSION

1. Rubrene The potential/light-intensity relationships observed for this molecule are illustrated in Fig. 2. The lower curve is a cyclic voltammogram of 1 x lo-3 M rubrene in dimethylformamide (DMF) O-1 M in tetrabutylammonium perchlorate (TBAP)

FIG. 2. Cyclic voltammogram and potential w light-output data for rubrene in DMF.

on a platinum electrode at a scan rate of 5 V/mm. This curve is shown in the same figure as the light output curve in order to illustrate the electrochemical behaviour of rubrene in this system. A moderately stable anion is formed at E1,2 = -1.37 V (see) (measured polarographically) and a considerably less stable cation (ca 5 to 10 s lifetime) is produced at E 1l2 = +0*95 V (see) (estimated from the cyclic voltammogram). The upper curve is the light output detected by a photomultiplier 0.2-s after initiation of the second step of a double-potential-step experiment. The anodic (left hand) branch of the light intensity curve was obtained by holding the potential of the electrode at -1.6 V (generating anion) for 5-s and then switching to various positive potentials as indicted on the potential axis, (generating cation and/or oxidizing anion) for the second 5-s step. The cathodic branch of the light intensity curve was recorded

Pre-annihilation

ekctrochemiluminescence

1213

recorded similarly by holding the electrode at + 1-OV and then switching to various cathodic potentials. It is readily apparent that light can be seen while oxidizing the anion from about -0-2 V and reducing the cation from about -0.95 V. In both cases, these threshold potentials are sufficiently far removed from the potentials at which the oppositely charged ion is produced to rule out a simple annihilation reaction as the source of light. It is also apparent that light is produced while oxidizing the anion and reducing the cation at overvoltages well below the energy of the rubrene O-O singlet transition (2.3 ev). This requires a mechanism more involved than direct excitation to the singlet. One possibility is that this pre-annihilation light-emitting process initially generates triplets which then undergo a T-T annihilation reaction, producing singlets. Of the possible triplet producing reactions occurring under pre-annihilation conditions, the simplest is generation of 3R by direct heterogeneous electron transfer at the electrode surface, RL-t3R+e-, (3) or R+ + e--t3R (4) followed by 3R+3R+1R+R, lR + R + hv.

(5) (6)

This has been held improbable on theoretical grounds12 but it is difficult to rule out low level luminescence on this basis alone. A second explanation involves oxidation or reduction of impurities in the system, followed by homogeneous chemiluminescent electron-transfer reactions. This simple impurity explanation is not reasonable in view of the different threshold voltages observed for pre-annihilation luminescence for various compounds.* However, it is possible that some common impurity (eg H,O) reacts with R’ and Ri to form new species (eg RH-) which oxidize or reduce at potentials which do vary for different fluorescers but are generally less positive or less negative respectively, than the potentials required to oxidize and reduce the original fluorescer. The apparent luminescent oxidation of the anion at potentials below that required for cation formation could then be attributed to oxidation of an anion decomposition product to an oxidized species (eg RH+) which in turn undergoes a luminescent electron-transfer reaction with unreacted anion. An analogous reaction can be visualized for the cation. This type of impurity mechanism was shown to be responsible for a major fraction of the pre-annihilation luminescence of the rubrene anion illustrated in Fig. 2, by the following experiments. In a l-7 x lop3 M solution of rubrene in DMF, 0.1 M in TBAP, the R’ species was generated at constant potential until a slight excess of RL was present in the bulk of the solution (as indicated by a persistent green colour and by the voltammetry of the solution). Thus, impurities which react with R’ were essentially purged by titration with R’. Prior to the end point of this titration, the R’ reacted very rapidly (lifetime go.1 s). after the end point the lifetime was of the order of many min. The re-oxidation of R’ shown in the cyclic voltammogram

1214

D. L. MARICLE,A. ZWEIG, A. H. MAURER and J. S. BRINEN

(recorded in an unpurged solution) presumably reflects the excess R” over the impurity, produced in the diffusion layer surrounding the electrode. The charge passed to reach the end-point represent an impurity level of lo4 to lo4 equiv/l. When the voltammetry of a freshly titrated solution was examined, an oxidation process commencing at -0.2 V and extending to the cation-formation region was clearly evident. Furthermore, when the potential of the electrode was maintained in this region, light could be detected continuously (solution stirred) with no switching of the voltage required. No such oxidation process or light emission could be detected prior to the purging titration. After the solutions aged (20 min) both the oxidation process and the light emission disappeared. Further generation of anion failed to revive either. Clearly the rubrene radical-anion reacts with some impurity (perhaps H,O) to produce a moderately unstable decomposition product, which when oxidized produces a chemiluminescent reaction. Whether this chemiluminescent reaction necessarily involves traces of R’ is difficult to determine, but it is known that the decomposition of the rubrene cation yields low levels of light in the absence of R’ .13 A similar purging experiment with the cation was impossible because of the instability of this species in the solvent employed. The purged and aged solutions exhibiting anion lifetimes of many min (at least 20) and exhibiting no continuous luminescence in the -0.2 to f0.9 V region were examined by a double-potential-step experiment identical to the one used to study the unpurged solution (Fig. 1) except that l-s pulses were used. Apparent luminescent oxidation of the anion was still detected. However, the threshold voltage was shifted from -0.2 to +0*5 V (see). The intensity of the luminescence was about two orders of magnitude below the intensity produced by the annihilation reaction. It resembles the pre-annihilation light observed for cation reduction both in intensity and in overvoltage required, and is only observed in a double-pulse or ac experiment while substantial anion oxidation current is passing. Low level pre-annihilation emission has also been reported for this molecule in benzonitrile,14 a solvent in which both the anion and the cation are stable. Although the mechanism of this residual luminescent process remains unproven, it may be due to direct generation of an excited-state species in a heterogeneous electron-transfer step. Even though a primary pre-annihilation light-producing process observed for rubrene is a following homogeneous chemiluminescent reaction, rather than direct excitation during a heterogeneous electron-transfer step, an energy-doubling mechanism is still required in order to reach the 2.3 eV singlet state. Triplet-triplet annihilation is again a possible process but no direct evidence of electrochemical excitation to a triplet state by either a heterogeneous electron-transfer step or a following chemical reaction has been available. In fact, evidence that triplets are not involved in the pre-annihilation ccl emission from isobenzofurans and tetracene (also singlet emission) has been established.8 2. Phenanthrene Examination of the pre-annihilation emission obtained during the oxidation of the phenanthrene radical-anion in DMF and acetonitrile provided the first positive identification of a triplet molecule generated in an electrochemiluminescent system. Figure 3 shows the cyclic voltammetry and the light-output/potential data obtained in

Pre-annihilation

electrochemiluminescence

1215

a double-potential-step experiment (l-s pulses) for purified 4 x 10W M phenanthrene in acetonitrile containing O-1 M (Bu),NClO, on a platinum electrode. Light can be detected during the oxidation of the stable anion (E 1,2 = -2.47 V(sce)) at voltages positive to + 0.15 V up to ca +1*8 V, where cation formation commences. In contrast to rubrene, the extremely unstable cation undergoes competing decomposition reactions which apparently eliminate the chemiluminescent annihilation reaction normally observed for fluorescent aromatic hydrocarbons. No light was detected in double-potential-step experiments when the cation was generated first, presumably for _ 6”r

FIG. 3. Cyclic voltammogram

and potential vs light-output acetonitrile.

data for phenanthrene

in

the same reason. The ca 2.6 V overvoltage required to obtain light during anion oxidation corresponds to ca 60 Kcal/mol, which (see later, Table 2) approximates the triplet energy of phenanthrene. The green colour of the ccl emission from this molecule is readily distinguished from the normal violet fluorescence emission. However, eximer emission is also known for this molecule15 and was claimed by Chandross et all6 to be the emission produced by the annihilation reaction. Table 1 shows the published spectral emission TABLE

1.

PHENANTHRENE

EMISSION SPECIRA

Intensity maximum KK Eximer emission*s Triplet emissionl’ Preannihilation ccl emission

23.3 ca 19.2 20.0

*Band width KK 5.4 ca 3.8

3.3

* Band width at half intensity. for the eximer and the triplet, and our data for the pre-annihilation ccl emission obtained from a 4 mM solution of phenanthrene in DMF. The ccl emission is a broad structureless band that resembles much more closely the phosphorescence spectrum also measured in room temperature fluid solutions by Parker and Hatchardl’ then it does the eximer emission. Both the position of the band maximum, and particularly the band width, support this conclusion. Further indication that triplet emission is being seen was obtained by the use of triplet quenchers. As shown in Table 2, addition of various substances which are electro-inactive between -2.5 and + 1.5 V(sce) in DMF to the green-emitting phenanthrene system affects the emission in a manner which can be related to the triplet energies of the substances. The presence of 1,3 ,Shexatriene does decrease the stability of the phenanthrene anion but not sufficiently to account for the total loss of emission. data

D. L. MARICLE, A. ZWEIG, A. H. MAURER and J. S. BRINEN

1216

The other added substances have no detectable effect on the phenanthrene radical stability.

anion

TABLE 2. EFFECT OF ELECTRO-INACXIVE TRIPLET QUENCHERSON PHENANTHRENE ELECTROCHEMILUMINESCENCE El Kcal/mol 1 ,3,5-trans-hexatriene’B 2,3-dimethylbudtadieneln Phenanthrene16 Biphenylso

47 59 62 65

% Quenched 100 87 0

These results together with the observation that triplet quenchers had no effect on thejuorescent electrochemiluminescence of 1,3 ,4,7-tetraphenylisobenzofuran8 indicate that in the acceptor-phenanthrene systems, irreversible triplet energy transfer is responsible for the light diminution and that phosphorescence is being seen. Purging experiments similar to those conducted with rubrene failed to prove the involvement of any anion decomposition products in the light-emitting reaction. The pre-annihilation emission was similar in both purged and unpurged solutions. Prolonged anion generation did eventually build up residual oxidation currents, but no light could be detected during oxidation of these species. Therefore, the available experimental evidence neither supports nor contradicts the hypothesis that the triplet species is produced as a result of a heterogeneous electron-transfer process. However, in view of the improbable nature of such a process, we think it more likely that a rapid following homogeneous electron-transfer process, probably involving traces of a relatively unstable anion decomposition product, produces the excited species. Considerable evidence for generation of triplet species by chemical oxidation of radical-anions has recently been obtained by Weller.21 These reactions are analogous to pre-annihilation electrochemiluminescence in that the oxidant employed (Wiirster’s blue perchlorate) is not sufficiently oxidizing to produce the radical-cation of the aromatic hydrocarbon involved, and yet singlet emission is usually detected. CONCLUSIONS

clear that triplet species can be produced by electrochemical excitation at or very near electrode surfaces. In the case of phenanthrene the triplet lifetime is sufficient to allow substantial radiative decay and therefore detection of the species by means of its characteristic phosphorescent emission. It seems reasonable to assume, therefore, that triplet species produced under pre-annihilation ccl conditions would have sufficient life time to undergo a bi-molecular T-T annihilation reaction and be responsible for singlet emission. This does not mean that T-T annihilation is necessarily responsible for pre-annihilation emission from all systems, but it is the most likely explanation unless specific information to the contrary is available.8 It is

REFERENCES 1. D. M. HERCULES,Science 145, 808 (1964). 2. M. M. RAUHUT et al, Chemiluminescent Materials. American Cyanamid Co. Technical Report No. 5 to the Office of Naval Research and the Advanced Research Projects Agency, Clearinghouse AD 606989 (1964). 3. R. E. VISCOand E. A. CHANDROSS,J. Am. them. Sot. 86, 5350 (1964). and A. J. BARD, J. Am. them. Sot. 87, 139 (1965). 4. K. S. V. SANTHANAM

Pm-annihilation

electrochemiluminesnce

1217

5. A. ZWEIG, G. METZLER,A. MAURER and B. G. ROBERTS,J. Am. them. Sot. 88,2864

6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21.

(1966). D. L. MARICLEand A. H. MAIMER, J. Am. them. Sot. 89, 188 (1967). A. ZWEIG, D. L. MARICLE, J. S. BRINEN and A. H. MAUIU~R,J. Am. them. Sot. 89,473 (1967). A. Z\NEIG. A. K. HOFFMANN,D. L. MARICLEand A. MAIJFUR,Chem. Commun. 106 (1967). E. C. K&VMAN and E. FARENHORST,Trans. Faraday Sot. 49, 58 (1943). R. F. STAMMZnd. ew. Chem.. Anal. Ed. 17,318 (1945). D. L. MARICLEand “J. P. MORNS, unpublished work. R. A. MUCUS, J. them. Phys. 43,2654 (1965). D. M. HERCULES,R. C. LANSBURYand D. K. ROE, J. Am. them. Sot. 88,457s (1966). J. CHANG, D. M. HERCULESand D. K. ROE, Electrochim. Acta 13,1197 (1968). T. AZUMI and S. P. MCGLYNN, J. them. Phys. 41,313l (1964). E. A. CILUDROSS, J. W. LONGWORTHand R. E. Vrsco, J. Am. them. Sot. 87,3259 (1965). C. A. PARKERand C. G. HATCHARD,J. phys. Chem. 66,2506 (1962). D. F. EVANS, J. them. Sot. 1735 (1960). R. E. KELUX~Gand W. T. SIMPSON,J. Am. them. Sot. 87,423O (1965). V. L. ERMOLAW, Usp. Fiz. Nauk. 80, 3 (1963). A. WELLER, K. Z. ACHAREASSE,J. Chem Phys, 46,4984 (1967).

DISCUSSION R. A. Marcus-A detailed theoretical discussion of the relative rates of formation of an excited molecule (P*) on one hand, and an unexcited molecule (P) and an excited electrode on the other, involves (1) a calculation of intrinsic rates of formation of P* and P, ignoring a preclusion question, (2) a study of the competition and the question of whether the occurrence of one precludes that of the other, and (3) an investigation of the subsequent kinetic fate of the products. Consideration of (1) for a very exothermic electrode reaction led to a result* given by KP exp (-&/4RT) (at zero overpotential for P*) and by KP for the probabilities of formation of P* and of P, respectively, per collision with the electrode. For aromatics, this ratio might be about lo-’ to 0.03, depending on the estimate* of &r/4. A favorable overpotential would increase it. The product surface leading to P* is seen to occur at higher energies than the leading one to P. If they have the same reaction co-ordinate (ie if they are reached by the same type of fluctuation of co-ordinates), as one would expect, the formation of P would tend to prevent that of P*, making its probability of formation even less than that calculated above. This preclusion would be complete when the transition probability K w 1 and would be small when K w 0. * Eg, if &i/4 has one half the value for homogeneous exchanges, and if KP w 1 for the latter, then* exn (-&/4RT)w 2/1O-8. If, instead, the rate constant is about 1 cm/s (compare M. E. PEOVER&d B.~S. W&E, j. electro&al. Chem. 13, 93 (1967) and P. MALACHESKY,T. -MILLER,T. LAYHOEPand R. ADAMS, Symposium on Exchange Reactions. p. 157, I.A.E.A., BrookhavenNational Laboratory, June 1965), then KP exp (-&1/4RT) is about lo-‘. M. E. Peover-It is unlikely that the pm-annihilation reaction can be due to a mechanism involving the species RH- produced as a result of proton transfer to the radical-anions. From a study of the spontaneous monoprotonation of dinegative ions of polycyclic aromatic hydrocarbons by cyclic voltammetry, we have been able to measure the oxidation potentials of the species RH- in the sequence of reactions R + 2e- -+ RB-, R*- + solvent -+ RH-, RI-

+ RH’ + e-.

The carbonion RH- contains the charge in a non-bonding molecular orbital for altemant-hydrocarbons, of approximately constant energy. The oxidation potentials of RH- are therefore very similar to different R, and lie close to -1-O V(sce). This potential is thus too negative to account for the observed pm-annihilation chemilumininescence, which occurs in the region 0 to +0*5 V. A similar experimental result is found for 9,10-diphenylanthracene and is to be expected for rubrene also. D. L. Maricle-We have no proof that water is the impurity being titrated, and your conclusion is no doubt correct. However, it is still an anion reaction product of some kind which is responsible for pre-annihilation ccl for rubrene in this solvent system. R. A. Marcus-It is sometimes assumed in the literature that when the standard free energy (AF”) or energy (Au”) of a reaction is positive the reaction cannot go. This assumption in inaccurate, because the reactants can acquire sufficient thermal energy from the environment to overcome any positive

1218

D. L. MARICLE, A. ZWEIG, A. H. MAURER and J. S. BR~NEN

AF” or AU”. The chance of their doing so during their lifetime depends on the free energy deficit to be overcome, and on the rates of competing reactions. The equations in my paper B illustrate this point :

k = ZKp e -AF*IRT

(4

AS AF* = w’ + ; 1+, , Cs) ( ) whereA=m’+wP-wr. The common assumption is that k = 0 when AF” > 0. Instead, these equations illustrate, k, is non-zero for any p. The AU” criterion is at least equally deficient. As a spechic example, the formation of an excited state of tetracene from a radical-cationradical-anion annihilation will be considered. The free energy deficit, m , was estimateda to be about +O.l eV 2.3 kcal/mol. For these reactions A/4 is estimated* to be about 4 kcal/mole*. When, for appreciable occurrence of the reaction in the presence of competing ones, one needs k > 1V l/mol/s, then one requires from (1) that AF* < RTln 101l/lo(, ie 7 kcal/mol. If k is to exceed lo8 l/mol/s then AF* should be less than 4 kcal/mol, and so on. If for the moment, wp and w’are ignored, (B) shows that AF* = 4(1 -t AF”/16)‘, and that AF* < 7 when AF” Q +6Kcal/mol, while AF* < 4 when AF” < 0. In the former case it is clear that quite a large standard free energy deficit can be tolerated and reaction still occur. Presumably wb is small, but wr is moderately negative. When these values are included the numerical values above are altered somewhat. * This A/4 followed from (A) and (B) for a typical electron exchange between aromatics AF” = 0, where k m lo8 l/mol/s; Z M 10” l/mol/s; KP M 1 (assumed).

at

W. Me/&-The experiments of Weller and Zachariasse (presented at the 66th meeting of the Bunsengesellschaft , Cologne, May 1967) seem to indicate that the formation of excited tetracene singlet states is possible by triplet-triplet recombination. In a simple flow apparatus a solutionoftetraceneanionsin dimethoxymethane was mixed with a solution of Wtirster’s blue cations, resulting in emission of the fluorescence spectrum of tetracene. Energetic considerations show that the energy liberated by the electron-transfer process is sufficient to populate the triplet levels but not the first excited singlet. The interesting point about this observation is that the energy liberated by t-t recombination is, for tetracene, slightly less than the energy of the first excited singlet, so that its formation must be an activated process. According to Marcus (personal communication) the free energy of activation for this process is given by AF*=i(l+$* where A is the difference between the level of the first excited singlet state and the energy liberated by the t-t recombination process (for tetracene A R+ 0.06 ev). Grubowski--Prof. MarcusB discusses the influence on the rate constant of an electron transfer of increasing difference of AF” (either in electrode or homogeneous reaction), in a way which is illustrated in Fig. A. Curve 1 represents the substrates A- + B, curve 2 the products A + B-. Increasing AF” (eg, by changing the electrode potential or the redox potential of the system) we obtain a lowering of AF* down to zero (curve 3), and then a new increase of AF* (curve 4). This

1

A-+

0 A+B-

FIG. A.

Reaction

parameter

Pre-annihilation

electrochemilumines

1219

physical picture, marginally, seems to contradict the view and evidence. presented by KrishtalikD. If it is true, however, then we can expect to obtain an emission under conditions represented by curves 1 and 4. The initial state, A- + B, is already an electronic excited state-an excited charge transfer state-of the tinal ground state, A + B-, which can be achieved either through thermal activation, or by emission of a luminescence photon, as indicated by the arrow. This will be, however, not an emission characteristic for a luminescence of any of the reactants, as their individual excited states need not to be involved, but a chargetransfer luminescence, broad and structureless, of wavelength variable with redox potential or electrode potential. The papers concerning chemihuninescence suggest strongly, that if an electron transfer is thermodynamically allowed in two ways, namely (a) to products in their ground states, with a large energy gain, or (b) to an electronically excited product, with energy change nil or nearly so, then the iso-energetic way to an excited state is favoured. This may perhaps explain some anodic oxidations of radical-anions, eg the pm-annihilation ccl discussed by Markleo. An anion, oxidized to the ground state hydrocarbon in a diffusion-controlled reaction, at more positive potentials may be oxidized preferentially to an excited species, usually a triplet, with no observable electrochemical effect. * Organic molecules containing large conjugated n-electronic systems, undergo relatively little structural chances bv eain of loss of an electron. Solvation of the corresnondine ions is also usuallv, . less than that aufthe’w’ell studied inorganic complexes. It seems to me that the preferential reaction paths to the excited product in the condition of energetic near-resonance with the substrate, might be. easilv described for these systems bv one of qua&m-mechanical approximations, eg by &reeling 0; by perturbation treatment of nearly degenerate initial and finite states, taking into account the vibrational (Franck-Condon) overlap integrals, as in the theory of radiationless transitions of Robinson and Frosch. R. E. Visco and E. A. Chandross (communicated)--The scheme proposed by Maricle et ale as a possible explanation for pre-annihilation ccl of rubrene on the basis of the presence of water in the solvent involves Ar- + Ha0 --f ArH’ + OH-, ArH’ -+ ArH+ + e- (electro-oxidation), Ar- + ArH+ -f Ar* + ArH’. They suggest that the ArH+/ArH’ couple lies below the Ar+/Ar couple, as it must to be electro-active in this potential range. They apparently have not considered further the energy provided by the third reaction. In rubrene, the reaction of Arf with Ar- provides just enough energy to allow the formation of lAr*. It is fortunate that TAS % 0 as discussed in our paper. If the ArH+/ArH couple lies appreciably below the Ar+/Ar couple, the third reaction cannot produce rAr* directly. The potentials for the various possible ArI-I radicals are unknown but the oxidation of a related species, the triphenyhnethyl radical occurs at about 0.4 V below that of tetracenex. It is likely that the corresponding ArH’ derived from rubrene would be oxidized even more easily with respect to rubrene. Thus the formation of rAr* via such a process in unlikely; one would then have to invoke the intermediacy of aAr* and triplet-triplet annihilation. Another point to be considered is the energy of the lowest excited state of the other product, ArI-I, of the reaction in question. This cannot be predicted with certainty, but two related cases should be considered. The radical derived from rubrene by the formal addition of a hydrogen atom is likely to be related to the triphenyhnethyl radical. The lowest excited state of (GHJ,C’ lies at 2.4 eV.B The diphenyl-a-naphthylmethyl radical would probably be a better model for the rubrene H’ radical. It has an appreciably lower lying excited state (l-9 eV)x than rubrene (2.3 ev), so that singlet excited rubrene could not be produced by the reaction suggested. Further, the reaction of Ar- with ArH+ could not produce sAr* because it requires that a triplet excited hydrocarbon, *Ar*, and other paramagnetic species, ArH’, be formed in the same solvent cage upon collision of the precursors At- and ArH +. This should lead to efficient quenching of the triplet. We feel that other reactions are probably involved in pm-annihilation ccl but we must admit that we have no concrete suggestions to offer. It appears that simple electron-transfer reactions alone cannot provide sufficient energy to explain preannihi1ation ccl. The energy deficiency could be satisfied by coupling other exothermic reactions which must involve bond formation and/or cleavage. We note that the impurity level in these solutions as determined by the titration technique (lo-’ to lo-” equiv/l) is not a negligible fraction of the initial rubrene concentration (1.7 x lo-* M). Also, as discussed in our paper, the triplet level of rubrene is unknown and it may be too low to permit triplet-triplet annihilation to yield an excited singlet. We have showna by simple thermodynamic arguments involving only one kinetic assumption (that electron transfer to and from “Ar* is as fast as that to Ar) that it is impossible to generate

1220

D. L. MARICLE, A. ZWEIG,A. H. MAURERand J. S. BRINEN

SAr* by charge transfer to or from an electrode. Since the only process that could have occurred at the electrode in the phenanthrene experiment in the absence of impurities or decomposition is oxidation of Ar- to Ar, the conclusions reached are questionable. The experiments with triplet quenchers are not conclusive. The two substances that reduced the luminescence intensity are chemically reactive species, which might be effective because they react with the impurities or decomposition products responsible for the chemiluminescence. It is unlikely that it will be possible to find a triplet quencher that can satisfy the requirements of electrochemical inactivity and chemical inertness. D. L. Mark&-The discrepancy between the energy provided by a singlet electron transfer step under pre-annihilation conditions (e.g. equation 3, your remarks) and that required to reach the 2.3 eV rubrene singlet energy level has, of course, been recognized and discussed at this meeting and in all our publications dealing with this subject (see e.g. this paper, Results and Discussion section, paragraph 10. In fact, curiosity about what energy doubling processes are occurring, was one of the major stimulants to pursue this work. However, when it was found that the major preannihilation process for rubrene resulted from an anion decomposition reaction, this was not considered of primary importance and our efforts were turned to investigating the phenanthrene system. Therefore, as indicated in resnonse to Dr. Peovers’ comments. we did not identify nositivelv the imnuritv reacting with the rtibrene anion, and in view of his measurements, now r;grke that i’t may will not have been water and that RH+ may not have been the electrogenerated oxidant. We agree that other reactions may be involved in proannihilation ECL processes, and have suggested ion aggregate redox processes as an alternative to T-T annihilation to account for the energy discrepancy (reference 8 this paper). Reactions involving bond formation and/or cleavage may also be involved, but experimental evidence or even the suggestion of a specific reaction would be most welcome. In regard to the question of the assignment of phenanthrene preannihilation emission to phosphorescence, the following points need to be reemphasized. A variety of impurity type processes may be occurring simultaneously with the heterogeneous oxidation of R- to R (eg. this paper, Results and Discussion, paragraph 15). The purging experiment failed to produce positive evidence for the type of impurity mechanism found for rubrene, but this by no means eliminates all possible following homogeneous electron transfer reactions. Our conclusion that such a reaction is responsible for excitation to the triplet state has received considerable support as a result of the recent observation by Professor Weller of triplet emission from Chrysene following chemical oxidation of the anion (see A. WELLERand K. ZACIURLUSE,J. them. Phys. 46, 4984 1967). Finally, in order to adopt the interpretation that diminution of the phenanthrene pre-annihilation ECL by triplet quenchers result from reaction with species involved in the following homogeneous electron transfer process, it is necessary to assume that the efficiency of such a chemical reaction is a function of the triplet energy of the quenchers, and that the quenchers react in this way with the intermediates involved in phenanthrene pre-annihilation ECL, but not at all with species responsible for pm-annihilation ECL of 1,3,4,7-tetraphenylisobenzofuran. In view of the spectral evidence supporting triplet emission, and the similarity between Professor Wellers’ observations and ours, we hold that triplet energy transfer is a much more reasonable explanation of the light diminution and that the phenanthrene ECL emission is phosphorescence. REFERENCES A. R. A. MARCUS, J. &em. Phys. 43, 2654 (1965). (Note: in Fig. lc the upper R and lower P branches should be joined, as should the upper P and lower R branches.) B. R. A. MARCUS. EIectrochim. Acta, 13,995 (1968). C. D. L. MAR&, A. ZW~G, A. H. MAURER and J. S. BRINEN,Electrochim. Acta, 13,1209 (1968). D. L. I. KRISCHTAL~K,Electrochim. Acta 13,1045 (1968). E. G. J. Horrrrmr, Iti. Chim. Be&. 1371 (1963). F. G. N. LEWIS, D. Lr~ru~ and T. T. MAGEL,J. Am. them. Sot. 66,1579 (1944). G. E. A. CHANDROSSand R. E. Vrsco, submitted to J. phys. Chem.