Effect of Solvents on the Quantum Efficiency of Rubrene Electrogenerated Chemiluminescence

Transcription

1 Effect of Solvents on the Quantum Efficiency of Rubrene Electrogenerated Chemiluminescence A. PIGHIN Bell-Northern Research, Otta~va, Canada KI Y 4H7 Received May 3, 1973 A simple and precise method for measuring the efficiency of electrogenerated chemiluminescence (e.g.c.1.) in terms of photons/(faradaic electron)(@,,.,) is described and applied to rubrene e.g.c.1. A symmetrical square wave potential having a frequency of Hz was employed to produce the e.g.c.1. The electrogenerated ion radicals were shown to be completely stable at these frequencies; consequently, was determined in the absence of side reactions. Change of solvent from N,Ndimethylformamide (DMF) to equivolume benzene-dmf resulted in an increase in from 0.67 to 1.21% at 20 decreased with increasing frequency. Varying the electrolyte (tetra-n-butylammonium perchlorate, TBAP) concentration did not Une methode simple et precise pour mesurer I'efficacitk de la chimiluminescence electroengendree (c.1.e.e.) en termes de photons par electron faradaique a Cte decrite et appliquee au rubrene c.1.e.e. Un potentiel d'onde a symetrie carree ayant une frequence de Hz a ett utilisee pour produire la c.1.e.e. On a montre que les ions radicaux electroengendres etaient completement stables a ces frequences, par consequent, a,,., a kt6 determine en I'absence de reactions parasites. Le passage de la N,N-dimkthylformamide (DMF) au melange equivolume benzene-dmf a pour resultat d'accroitre a,,.l de 0.67 a 1.21% a 20 Hz. a,,,,, decroit lorsque la frequence croit. La variation dans la concentration en electrolyte (perchlorate de tetra-n-butylammonium, PTBA) n'affecte pas [Traduit par le journal] Can. J. Chern., 51, 3467 (1973) Introduction The efficiency of electrogenerated chemiluminescence (e.g.c.l.),' in this article expressed in photons/(faradaic electron) and represented by Q,,,,, can be a useful parameter in elucidating the e.g.c.1. mechanism. Bard et al. (1) have recently reviewed the reported e.g.c.1. efficiencies and discussed the difficulties in these measurements. Rubrene is a popular electrochemiluminescent compound and its reported e.g.c.1. efficiencies range from a low of 0.1% (1) which has been corrected for lost electrogenerated species to a high of 8.7x (2) without any corrections. Bezman and Faulkner (3) obtained intermediate values and recorded greater efficiencies in benzonitrile than in N,N-dimethylformamide (DMF). Other e.g.c.1. systems have shown either a solvent composition or electrolyte concentration dependence of their efficiencies (1). In this article, we present evidence of a large solvent effect, a frequency dependence, and no electrolyte concentration effect on the quantum efficiency of stable rubrene e.g.c.1. '~lectrogenerated chemiluminescence is abbreviated as ECL in the existing literature. To avoid confusion with Emitter-Coupled Logic we employ e.g.c.1. Experimental The rubrene was purchased from the Aldrich Chemical Co. and recrystallized repeatedly from xylene-methanol and then dried by vacuum heating. Tetra-n-butylammonium perchlorate (TBAP), polarographic grade from MC&B, was recrystallized repeatedly from ethanol and dried by vacuum heating. The two solvents, Fisher D-133 DMF and Fisher B-414 benzene. were dried with molecular sieves and outgassed. The solutions were prepared and used inside a nitrogen-filled dry box. The apparatus used to measure is illustrated in Fig. 1. The electrolysis cell was formed from a TS Pyrex ground glass joint while the electrodes were two parallel in. diameter platinum wires rigidly fixed about 1 mm apart by beads of soft glass. A symmetrical square wave potential was employed to electrolyze the solution in the cell. Two identical pulses of light per cycle from each electrode were observed and the overall intensity remained constant for at least several hours or continuous electrolysis. The driving potential was restricted to those values which produced stable e.g.c.1. The bottom of the electrolysis cell rested inside an open box whose walls and floor were made of five 1 cm2 solar cells connected in parallel. The level of the solution inside the electrolysis cell was kept below the top of the solar cell box. The solar cells were operated in the current mode and their output measured with a picoammeter. The solar cell box was calibrated using a red light-emitting diode of known efficiency (traceable to the National Research Council of Canada). The correction factor was 1.7, i.e. (measured current x 1.7) = corrected current. The electrical current through the electrolysis cell was converted

2 CAN. I. CHEM. VOL. 51, 1973 TABLE 1. Physical properties at 25 "C Diffusion Viscosity coefficient of Dielectric (cp) rubrene (crnz/s) constant P M TBAP-DMF x M TBAP - equivolume x benzene-dmf Benzene 2.274* DMF 'Reference 4.?Reference 5. FUNCTION GENERATOR. WAVETEK MODEL 130 k@-m AMPLIFIER. HP HARRISON 6823A VOLTMETER.HP34WA TRUE RMS P U PICOAMMETER, KEITHLY MODEL 410A CURRENT TO VOLTAGE CONVERTERIAMPLIFIER LINEAR RECTIFIER INTEGRATING DIGITAL VOLTMETER. DATA TECHNOLOGY MODEL 3W FIG. 1. Experimental apparatus for measuring cd,. and amplified by a factor of 1000 to a voltage, rectified with a fast linear rectifier, and then measured with an integrating digital voltmeter. This signal processing of the electrical current yields the rectified average value of the current. Specifically, 1 V d.c. was equivalent to 1 ma average rectified a.c. The cd, experiments were performed at 23? 1 "C. Electrochemical measurements were performed with a Princeton Applied Research Model 170 Electrochemistry System and a three electrode cell. A silver wire immersed in the test solution was used as the reference electrode; it was contained inside a compartment whose Luggin tip extended to within 2 mm of the platinum bead working electrode. The counter electrode consisted of a large platinum wire helix around the working electrode and Luggin tip. ir compensation was employed in the recordings. Results and Discussion The two solvents employed in this investigation were DMF and equivolume benzene-dmf and were selected for their physical properties (Table 1). The solution viscosities are only slightly different but the dielectric constant of DMF is much greater than that of benzene. It was therefore expected that the bulk dielectric constant of the mixed solvent solution would be significantly less than that of the DMFsolution. The viscosities were measured with an Ostwald viscometer that was calibrated with water and the diffusion coefficients of the 1 mm rubrene in M TBAP solutions were determined from chronoamperograms of the reductions at a shielded platinum disk electrode of known area. Note that the products of (viscosity) (diffusion coefficient) areessentially the same in obeyance of the Stokes-Einstein equation (6). Table 2 contains the potential differences between the half-wave oxidation and reduction potentials of rubrene in both solutions at selected temperatures. These data were derived from cyclic voltammograms of the oxidation and reduction at a scan rate of 100 mv/s with ir compensation. The peak potentials were measured from the cyclic voltammogram, 1.109RTlnF (7) V were subtracted from their absolute values to obtain the half-wave potentials, and then the TABLE 2. Variation in the potential difference between the half-wave oxidation and reduction potentials of rubrene with temperature MTBAP- Temperature equivolume ("C) M TBAP-DMF benzene-dmf

3 PIGHIN: RUBRENE ELECTROGENERATED CHEMILUMINESCENCE TABLE 3. Thermodynamic and spectroscopic data of rubrene Relative A 9 -AHzs oc Els* fluorescence (cal/mol "K) (kcal/mol) (kcal/mol) efficiency M TBAP-DMF O M TBAP - equivolume O benzene-dmf half-wave reduction potential was subtracted from the half-wave oxidation potential.nemec's (8) equation for calculating ir losses for spherical microelectrodes was used to estimate our ir losses in the absence of ir compensation. The peak currents were about 20 pa in both solutions at 25 "C, the radius of our electrode was about 0.5 mm, and the Luggin tip was positioned about 2 mm from the electrode's surface. The specific conductances of the M TBAP solutions were measured and were ohm-' cm-' for DMF and ohm-' cm-' for equivolume benzene-dmf at 25 "C. When all these values are inserted into Nemec's eq. 2, and multiplied by 2 since the ir losses occur during both the oxidation and reduction, and are therefore additive, the ir losses are calculated to be about 12 mv for the DMF solution and 20 mv for the mixed solvent solution. Assuming that the ir compensation is 80% effective, we see that the residual, uncompensated ir is only a few mv in both cases. The temperature coefficient of the conductance is negative and greater in magnitude than the positive temperature coefficient of the peak current. Consequently, the amount of residual, uncompensated ir tends to decrease with increasing temperature. From the temperature coefficient of the potential difference between the half-wave oxidation and reduction potentials (Table 2), the standard entropy of the homogeneous electron transfer reaction in each solution was calculated using the following equation in which the symbols have their usual significance: The temperature coefficient was assumed to be constant and the slope was calculated using linear regression analysis. The results of these calculations are listed in Table 3 along with other relevant thermodynamic and spectroscopic data of rubrene in both solvents. It should be noted that because the amount of residual, uncompensated ir decreases with increasing temperature then the above entropies are slightly smaller than the true values. The change in enthalpies in Table 3 are for the homogeneous electron transfer reaction between the monoanion and monocation radicals of rubrene and were calculated from the equation The energy of the lowest excited singlet (Els*) was determined from the crossover point between the absorption and fluorescence spectra of rubrene in each conducting solution. The relative fluorescence efficiencies were obtained by comparing the intensities of the fluorescences of mm rubrene in each conducting solvent. The fluorescences were excited with 4500 A radiation. Neither solvent absorbs at this wavelength so that the fluorescence intensities did not have to be corrected for differences in absorption but only for the ratio of the squares of the refraction indices of their solutions (9). An important conclusion can be made from the data in Table 3. The change in enthalpy of the homogeneous electron transfer reaction in both solutions is clearly less than the energy of the emitting species, El,*. Residual, uncompensated ir in the electrochemical measurements used to calculate the -AHYs have made the calculated values of -AH slightly larger (more negative) than the true values. Rubrene e.g.c.1. is therefore "energy deficient" and must proceed in both solutions via the "triplet route" (10). This mechanism is summarized in Table 4. In this scheme, each homogeneous electron transfer reaction produces an average of 314 triplets. This fraction is given by spin statistics assuming that the solvent does not preferentially enhance either the formation of triplets or singlets (11). This seems probable considering rubrene is a simple aromatic hydrocarbon (12). Triplet-triplet annihilation can then produce either triplets, singlets, or

4 CAN. J. CHEM. VOL. 51, 1973 TABLE 4. Rubrene e.g.c.1. mechanism -- - R-,R++e R+e-,R- Oxidation of rubrene at electrode Reduction of rubrene at electrode R+ + R- -, 314 3R* R Homogeneous electron transfer reaction/ triplet formation 3R* + 3R* -, 114 'R* R* + R Triplet-triplet annihilation/singlet formation 'R* -, R + Iiv quintets. If we make the usual assumption that the formation of quintets is energy-forbidden, then spin statistics predict an average yield of 114 singlets per triplet-triplet annihilation. The singlets immediately radiate to their ground state with the emission of light. The excited triplets produced by the triplet-triplet annihilation reaction engage in further triplettriplet encounters, thereby increasing the singlet yield per triplet-triplet annihilation from 114 to 215. Thus, two electrons (one homogeneous electron transfer reaction) ultimately results in the emission of an average of 3/20 photons. However, about 8 (3) to 17% ((2, say 12.5%, of the ion radicals are neutralized by the reversal of the polarity of the electrode potential. The maximum theoretical efficiency of an energy deficient e.g.c.1. system, such as rubrene e.g.c.l., using square wave stimulation is 6.6% photons/electron. Figure 2 illustrates a typical current-voltage curve of the electrolysis cell. Prior to the appearance of e.g.c.l., the current increases linearly with applied voltage and at constant voltage this Radiative decay of excited singlet current was nearly directly proportional to the frequency. Since these characteristics are peculiar to capacitive currents, then this current must be the current that is needed to polarize the electrode-solution interphase. The abrupt rise in current at about 2.3 V (peak) coincides with the detection of e.g.c.1. Assuming that the Faradaic processes do not alter the double layer capacity and that the double layer capacity remains constant in the potential region for e.g.c.l., then by extrapolating the capacitive current into the region for e.g.c.1. and subtracting it from the total at that potential, the Faradaic current at that voltage is obtained. A plot of the corrected photon current us. the Faradaic current is then made and O,,,, calculated from the slope of this plot. Linear regression analysis was used to obtain the slopes. Figure 3 illustrates one of these plots. The photon current measured with the picoammeter is twice that emitted at one electrode. 0 I I I I average Faradsic currant, ma FIG. 2. Typical electrical characteristics of theelectroly- FIG. 3. Typical variation in photon current with sis cell. The 1.01 mm rubrene in M TBAP - Faradaic current. The 1.01 mm rubrene in M equivolume benzene-dmf was electrolyzed with 100 Hz TBAP - equivolume benzene-dmf was electrolyzed square wave potential. with 100 Hz square wave potential.

5 PIGHIN: RUBRENE ELECTROGEI VERATED CHEMILUMINESCENCE 3471 Therefore, calculated from this plot must be divided by 2 to obtain the value at each electrode. The results of measurements are given in Fig. 4. These results, when corrected for the fraction of ion radicals lost by the reversal of the electrode potential (8-17%), are at most 20% of the maximum theoretical value previously calculated. A frequency dependence is seen in all cases, in contradiction to theory (2). If a quencher were present in the bulk of the solution, the fractional probability of quenching will decrease with increasing frequency because the triplet concentration would increase within a steadily more compact reaction zone. It has been suggested (1) that the triplet state may be quenched by the parent ion radicals and by the electrode. The effectiveness of the former mode is frequency independent because the ratio ion radicals: triplets remains constant with changing frequency whereas the effectiveness of the latter mode increases with increasing frequency because the reaction layer becomes more compact at the electrode. Our frequency dependence of I,,,, is consistent with quenching of the triplets by the electrode. Reducing the electrolyte concentration from 0 1.WmM RUBRENE IN0.100M TBAP-DMF 1.W mm RUBRENE IN 0.020M TBAP - DMF 1.01 mm RUBRENE IN O.1WM TBAP - EOUIVOLUME BENZENE-DMF J to M TBAP in DMF did not affect the efficiency of the rubrene e.g.c.1. which is in contrast to the increase reported by Bard et al. (1) in the e.g.c.1. efficiency of 9,lO-dimethylanthracene-tri-p-tolylammonium perchlorate with decreasing electrolyte concentration in TBAPtetrahydrofuran. In this latter case, ion pairing was suggested as a cause. In our case, TBAP is less dissociated in equivolume benzene-dmf than in DMF (based on conductivity measurements) but the absence of an electrolyte effect on of rubrene in TBAP-DMF renders the degree of electrolyte dissociation unimportant. Addition of one part of benzene to DMF increased of rubrene from 0.67 to 1.21% at 20 Hz for the same concentration of fluorescor (1 mm) and electrolyte (0.100 M). The luminescences obtained were spectrally identical and the fluorescence efficiencies are the same in both solutions (Table 3). We established earlier that the e.g.c.1. proceeds in both solutions via the "triplet route". The greater potential differences between the oxidation and reduction potentials of rubrene in the mixed solvent solution (Table 2) indicates that the rubrene is more strongly solvated in the mixed solvent solution than in the DMF solution. We obtained e.s.r. spectra of the electrogenerated monoanion radicals in both solutions and the spectra were identical : the difference in solvation of the monoanions was either apparently not sufficient to cause a shift in the e.s.r. spectrum or the monoanions were similarly solvated (by DMF) in both solutions. Because of the heteropolar nature of the mixed solvent, i.e. one aromatic and one polar component, it is likely that the benzene fraction will preferentially solvate the neutral rubrene, including excited forms, while the DMF fraction will preferentially solvate the ion radicals. If the triplets were quenched by the parent ion radicals and/or electrode, their greater solvation in the mixed solvent system could shield them from the quenching agents thereby accounting for the in the mixed solvent solution. The common element between our solvent effect on Q,,,, and those reported by Bezman and Faulkner (3) and Bard et al. (1) is that we all obtained higher efficiencies in the less polar, or more aromatic, solvent. FIG. 4. Variation of rubrene in selected solutions with frequency of driving potential. This work was supported in part by the National Research Council of Canada through its Industrial

6 3472 CAN. J. CHEM. VOL. 51, 1973 Research Assistance Program. The assistance of R. M. van Dyk in the design of the electronic circuitry and of D. P. Malanka in performing the experiments, and the helpful criticism of Dr. M. A. Kabayama are gratefully acknowledged. 1. A. J. BARD, C. P. KESZTHELYI, H. TACHIKAWA, and N. E. TOKEL. 112 Chemiluminescence and bioluminescence. Edited by M. J. Cormier, D. M. Hercules, and J. Lee. Plenum Press, New York, N.Y p P. M. SCHWARTZ, R. A. BLEKELEY, and B. B. ROBINSON. J. Phys. Chem. 76, 1868 (1972). 3. R. BEZMAN and L. R. FAULKNER. J. Am. Chem. SOC. 94, 6324 (1972); 95, 3083 (1973). 4. R. C. WEAST. (Editor). Handbook of chemistry and physics, 49th ed. The Chemical Rubber Co., Cleveland, Ohio, p. E58. Du Pont DMF Product Information Bulletin. E. I. Du Pont de Nemours and Co., Wilmington, Delaware. p. 3. C. A. PARKER. Photoluminescence of solutions. Else,vier, London p. 76. R. S. Nlc~oLsoN and I. SHAIN. Anal. Chem. 36, 706 (1964). L. NEMEC. J. Electroanal. Chem. 8, 166 (1964). J. N. DEMAS and G. A. CROSBY. J. Phys. Chem. 75, 991 (1971). L. R. FAULKNER, H. TACHIKAWA, and A. J. BARD. J. Am. Chem. Soc. 94, 691 (1971). G. J. HOYTINK, Discus. Faraday Soc. 45, 14 (1968). G. J. HOYTINK, III Chemiluminescence and bioluminescence. Edited by M. J. Cormier, D. M. Hercules, and J. Lee. Plenum Press, New York, N.Y p. 147.