J. MATER. CHEM., 1994,4(5), 675-678 Electroluminescence of Organic Thin Films based on Blends of Polystyrene and Fluorescent Dyes Peter Frederiksen, Thomas Bjarnholm,* Hans Georg Madsen and Klaus Bechgaardt Centre for Interdisciplinary Studies of Molecular Interactions, Department of Chemistry-Symbion, University of Copenhagen, Fruebjergvej 3, DK-2 700 Copenhagen 0,Denmark Blue and red light-emitting diodes were constructed as devices in which a polystyrene/dye blend was sandwiched between a calcium electrode and a transparent indium-tin oxide electrode. As dye molecules, four alkoxy-substituted derivatives of 2,5-dialkoxy-l,4-bis(2-phenylvinyl)benzene (inemp-phenylene vinylene oligomers), a derivative of bisanthra- cene (2,2’,3,3’-tetrachloro-9,9’-dimethoxy-l0,1O’-bianthracene) and the laser dye 4-dicyanomethylene-2-methyl-6-(p-dimethylaminophenylvinyl)-4H-pyran (DCM) were used.The critical field for light emission in devices using 20 wt.% PPV oligomers in polystyrene was found to be of the order of 2 x 1O8 V m-’. The dependence of the device conductivity on the concentration of dye molecules in the polystyrene film is described by simple percolation theory. The synthesis of 2,2’,3,3’-tetrachloro-9,9’-dimethoxy-l0,1O’-bianthracene is reported. In recent years two new types of light-emitting diode (LED) have been reported. One type uses thin molecular films of simple fluorescent molecules as the light-emitting layer,’-5 the other type uses conjugated polymers, e.g. poly(p-phenylene vinylene) (PPV) as the active polymer in thin-film based devices.&14 In both cases, electrons injected at the negative electrode and holes injected at the positive electrode recombine in the organic thin film, forming excited states which decay radiatively.The present paper is concerned with an intermediate situ- ation in which a blend containing a fluorescent PPV oligomer, or other fluorescent molecules in a neutral host polymer, e.g. polystyrene, forms the electroactive layer. Such films have recently been demonstrated to be electroluminescent.’5~16We present here investigations describing the dependence of the electroluminescence on the type of fluorescent molecule, the loading of the fluorescent molecules in the host polymer film and the nature of the polymer host.The results show that well dispersed blends of the fluorescent guest molecules in polystyrene form efficient light-emitting diodes compared to diodes using saturated polymer hosts like poly(methy1 meth- acrylate). The colour of the diode is readily changed by changing the guest molecules. Experimental The PPV oligomer dyes (Fig. 1) were synthesized by a Horner-Emmons variation of the Wittig reaction, between a suitable p-xylylene diphosphonate and a suitable benzal- dehyde, in a similar manner, as described in ref. 17 and 18. Mixtures of cis and trans isomers were analysed by HPLC, and the fraction of the all-trans isomer was found to be 50-100% (Table 1). of CH2C12, dried over MgSO, and filtered through ;ishort column of silica gel.The resulting solution was evaporated to dryness to yield 10.0 g (95%) of BA as a glass. To obtain crystals, the product was dissolved in CH,Cl, and precipitated by adding petroleum ether. Mp: 304-305°C.1 H NMR (60 MHz, CDCl,, standard TMS) 6: 4.3 (s, 6H); 6.8-7.7 (m, 8H); 8.3-8.7 (m, 4H). Calcd. for C3,H18C1,02: C. 65.24; H, 3.29; Cl, 25.68%. Found: C, 65.00; H, 3.37; C1, 26.00Y0. The laser dye, DCM, and the various polymers were commercially available. The molecular weight distribution of the polymers did not have a pronounced effect on the proper- ties of the diodes. LED devices were constructed by using an indium-tin oxide (ITO) coated glass substrate, where the IT0 glass serves as a transparent hole-injecting electrode.The glass/ITO substrates were cleaned by washing them with water and propan-2-01, and finally vapour, degreasing them in chlorobenzene. On this substrate a layer of the polystyrene/dye blend was formed by spin-coating a solution in chlorobenzene of the desired mixture of guest and host. On this film, a layer of calcium was vacuum deposited ( Torr) through a mask, to obtain round spots with areas of ca. 12mm2. Finally, a layer of aluminium was vacuum deposited on the calcium electrode to protect it from oxidation and corrosion. A typical film containing 20 wt.% of the dye and having a thickness of 150nm was obtained by spin-coating using a solution of 10mg of the dye, and 40mg of polystyrene in 1 ml of chlorobenzene. The thickness of the films was calculated using the molar absorption coefficient for dilute solutions and the measured absorbance at the absorption peak of the dye.$ The resulting thicknesses agreed with profile measurements and thicknesses obtained from analysis of the fringes resulting from optical measurements of the films.Synthesis of 2,2’,3,3’-tetrachloro-9,9’-dimethoxy-10,10’-bianthracene (BA): A slurry containing 2,2’,3,3’-tetra-chloro-10,lO’-bianthronyl(10.0 g, 19.9 mmol) (19), 1 ml of absolute ethanol and 200ml of dry DMF was prepared. Sodium hydride (1.26 g, 2.20 equiv., 80% in mineral oil) was added under nitrogen. A red homogeneous solution was formed as the bianthronyl dissolved. When no more gas evolved, dimethyl sulfate (4.0 ml, 2.2 equiv.) was added and the solution was stirred for 24 h.The yellow opaque solution containing the product was poured into 500 ml of water. The precipitate was filtered, washed with water, dissolved in 150 ml t Present address: Department of Physics, RIS0 National Laboratory, DK-4000 Roskilde, Denmark. The current-voltage characteristics were measured using a standard power supply and an X/Y recorder or with a standard digital voltmeter and amperemeter (Fig. 2). Light emission was analysed using a Perkin-Elmer LS 5 spectrometer. Results Electroluminescence was observed from devices based on polystyrene blends of the fluorescent guest molecules shown $d=103(AM,)/psw, where d is the film thickness in pnm, A is the absorbance of the film, M, is the molecular weight of the guest molecule, w is the weight percent of the guest in the host, E is the absorption coefficient measured in dilute solution (in dm3 mol-I cm-l) and p is the density of guest and host (in g cmP3).J. MATER. CHEM., 1994, VOL. 4 I I ,o P 0‘O’ I I PPVO1 PPVO2 I 01 I/. -\ \/ \ \I d \ \/\\Ilo--0 0 -0 -0 I /-O PPVO3 PPVO4 0’ 1 CI clyJ-pCI yo BA DCM polymer host molecules: CI polystyrene poly(rnethyi methacrylate) poly(viny1 chloride) Fig. 1 Guest and host molecules used in the investigation. Abbreviations as follows: PPVO 1, 2,5-dimethoxy- 1,4-bis( 2-phenylviny1)benzene; PPVO2, 1,Cbis[244-methoxyphenyl)vinyl]-2,5-dimethoxybenzene;PPVO3, 1,Chis[24 2,4-dimethoxyphenyl) vinyl]-2,5-dimethoxybenzene; PPVO4, 1,4-bis[2-(2,5-dimethoxyphenyl)vinyl]-2,5-diethoxybenzene; BA, 2,’3,3’-tetrachloro-9,9-dimethoxy-10,lO-bianthracene; DCM, 4-dicyanomethylene-2-methyl-6-( p-dimethylaminophenylvinyl)-4H-pyran Table 1 Electroluminescence from devices based on polystyrene blends of fluorescent guest molecules of Fig.1 Amax (electroluminescence)/ €1 guest nm lo4dm3 mol-’ cm-’ PPVOl 484 388 2.7 PPVO2 482 394 4.6 PPVO3 494 398 4.4 PPVO4 461 402 BA“ 458 27 1 7.6 DCM 620 472 4.2 “The transition with lowest energy appears at 11=415 nm (E= 1.8 x lo4dm3 mol-’ cm-I). Ratio of cisltrans isomers (based on HPLC measurements): PPVOl and PPVO2 100% trans-trans; PPVO3: trans-trans 72%; cis-trans 28%. PPVO4: trans-trans 48%, cis-trans 33YO; cis-cis 19%.in Fig. 1 (Table 1). Different batches of polystyrene were used All electroluminescent devices exhibit an electroluminesc- with no pronounced effect on the quality of the devices. Under ence spectrum which is very similar to the photoluminescence the same experimental conditions high-quality films of the spectrum of the same films (see Fig. 3 as a representa-dyes could be produced using PMMA and PVC as the host tive example). Since the excitation spectrum of the films polymer, but electroluminescence could not be observed from additionally resembles the absorption spectrum of the guest devices based on these films. molecules (Fig. 3), it is evident that the electroluminescence J. MATER. CHEM., 1994, VOL.4 I calcium II Fig. 2 Schematic illustration of the device configuration I --1 350 400 450 500 550 600 wavelengthhm Fig.3 Optical properties of a film of PPVO3 in polystyrene. (-) Absorption spectrum, (---) excitation spectrum, electroluminescence spectrum (EL) and photoluminescence spectrum (PL) of the film. The dashed thin line shows the absorption spectrum of a dilute solution of PPVO3 in CH,Cl,. The properties are representative of the films investigated. occurs by radiative decay of an electronically excited guest molecule. The absorption spectrum of the guest molecules densely packed in the polymer films (20wt.%) resembles the spec- trum of dilute solutions of the guest molecules (Fig. 3). Measurements of the dichroic ratio at normal incidence and skew angles of incidence2' of some of the samples revealed a highly isotropic distribution of the guest molecules in the polymer films.Comparison of the absorption at Laxof dilute solutions in 1 cm cuvettes and films of known thicknesses revealed that within a 10% margin, the molar absorption coefficient of the guests in the films was identical to the absorption coefficient in dilute solution (Table 1). These results are all indicative of a molecularly dispersed mixture of guest and host. In the following we therefore assume that no strongly bound aggregates of the molecules are formed. The current-voltage characteristics of a representative device show diode behaviour, allowing a current to run through the device when the Ca electrode is negatively biased above a certain critical value, but not vice versa.The onset of light emission occurs at slightly higher bias voltages, as shown by the inset in Fig.4.With a 50 mA current running through the device at ambient conditions, the half-life was 90 min. The LEDs were stable for weeks when they were kept under dry argon and zero current. The p-phenylene vinylene oligomer (PPVO) based diodes were the most stable of the devices studied here. Investigations of samples of different thicknesses of PPVO3 2.0 1.5 2 1.0 2 0.5 bias voltageN I0 0.0 -0.5 Fig. 4 Current-voltage characteristics of a device using 20 wt.% PPVO3 in polystyrene. Inset: (-), current; (---), luminescence reveal a linear dependence of the driving voltage on the film thickness (Fig.5). From the slope of the plot in Fig. 5 the critical field for light emission is estimated to be 1.7 x lo8V m-'. The dependence of the current-voltage characteristics on the loading of guest in the host polymer is shown in Fig. 6. At zero loading the diode characteristics are absent and it is not possible to drive a current through the device using voltages below the critical voltage for dielectric breakdown. The clear signature of the diode characteristics occurs at loadings higher than ca.10wt.%. The same data are replotted in Fig. 7 to show the dependence of the current on the loading at constant bias voltage. A steep rise in the current appeared at ca. 15 wt.% of guest molecules.This behavior is typical of percolation behaviour as discussed further in ref. 21. I 1 0 100 200 300 400 500 600 film thicknesshm Fig.5 Drive voltage us. film thickness of devices using 20wt.% PPVO4 in polystyrene. Slope = 1.72 x lo8V cm-l 1200 20%:Innrr I/15% 2001 0-0% I I 1 I I J. MATER. CHEM., 1994, VOL. 4 PVC since the latter does not contain an appreciable 7c-electron density. The mean molecul$r separation at the onset of the rise in Y 2 150 I? 4001 I I0 4 8 12 16 20 IE 300-d/B, I I 30 2001 I I I I I I 100-------*-- - - -4 I L weight (%) Fig. 7 Current us. wt.% at constant bias voltage (32 V) for PPVO3 in polystyrene (the inset shows the dependence of the current on the mean molecular separation of the dye molecules in the polymer) Discussion LEDs of a number of blends of fluorescent molecules and polystyrene have been constructed and investigated.The electronic properties of the fluorescent molecule govern the colour of the device (Fig. 3) and the miscibility of the guest molecules in polystyrene governs the quality and efficiency (Fig. 7). Our results indicate that it is possible to create a large variety of colours by using these types of device. The luminescence of the device resembles the fluorescence of the dye (Fig. 3). The clear evidence of a critical field for light emission (Fig. 5) indicates that the diodes are tunnel diodes” and not Schottky diodes. The quantum efficiency of the devices has not been measured, but the dependence of the current (and luminescence) on the amount of guest in the host polymer (Fig.6) shows that optimization of the loading can improve the efficiency of the device considerably. Reasonable efficiencies have been reported for similar system^.'^ At constant electric fields the conductivity behaviour (i.e. current-voltage) shown in Fig. 7 clearly resembles the behav- iour of systems of metal-insulator blends described by per- colation theory.21 In three dimensions the critical volume fraction for conductivity in such systems is 16%21 and, assuming that the densities of guest and host in our systems are similar; this corresponds to a critical amount of 16 wt.%. As seen from Fig. 7, the agreement with this value is remark- able, indicating that percolation phenomena govern the con- ductivity of the devices. The resemblance between the photoluminescence and the electroluminescence in the films shows that localized exci- tations of dye molecules are important.The dependence of the device conductivity on loading shows that conduction is mediated by the guest molecules, and that the intrinsic conduc- tivity of the polymer host is very low (Fig. 6). It is therefore plausible to assume that the mechanism for conductivity occurs by electron or hole hopping from dye molecule to dye molecule. In this context the polymer host plays the role of a tunnelling barrier. Since efficient devices could only be made using the polymer with a relatively high n-electron density it is plausible that tunnelling from guest to guest is the mechan- ism of conduction because n-electrons in the polymer barrier between guests (as in polystyrene) offers lower-lying states than in the corresponding saturated polymers. Polystyrene will hence act as a lower energy barrier than PMMA and conductivity is ca.8 A as seen from the inset in Fig. 7. Since this separation represents an upper limit for the separations found in the possible percolation cluster responsible for con- ductivity, mean molecular separations are indeed small enough to allow significant tunnelling probabilities for the hopping process from guest to guest. We wish to thank K. Schaumburg Ib Johannsen and Ole Kramer for useful discussions and the Danish Materials Research Program for funding.References 1 C. W. Tang and S. A. VanSlyke, Appl. Phys. Lett., 1987,5,913. 2 C. W. Tang, S. A. VanSlyke and C. H. Chen, J. Appl. Phys., 1989, 65, 3610. 3 C. Adachi, T. Tsutsui and S. Saito, Appl. Phys. Lett., 1989, 55, 1489. 4 C. Adachi, T. Tsutsui and S. Saito, Appl. Phys. Lett., 1990,56,799. 5 C. Adachi, T. Tsutsui and S. Saito. Appl. Phys. Lett., 1990,57,531. 6 J. H. Burroughs, D. D. C. Bradley, A. R. Brown, R. N. Marks, K. Mackay, R. H. Friend, P. L. Burn and A. B. Holmes, Nature (London), 1990,347,539. 7 D. Braun and A. J. Heeger, Appl. Phys. Lett., 1991,58,1982. 8 D. D. C. Bradley, A. R. Brown, P. L. Burn, J. H. Burroughes, R. H. Friend, A. B. Holmes, K. D.Mackay and R. N. Marks, Synth. Met., 1991,41-43,3135. 9 A. R. Brown, N. C. Greenham, J. H. Burroughes, D. D. C. Bradley, R. H. Friend, P. L. Burn, A. Kraft and A. B. Holmes, Chem. Phys. Lett., 1992,46,200. 10 P. L. Burn, A. B. Holmes, A. Kraft, D. D. C. Bradley, A. R. Brown, R. H. Friend and R. W. Gymer, Nature (London), 1992,47,356. 11 G. Gustafsson, Y. Cao, G. M. Treacy, F. Klavetter, N. Colaneri and A. J. Heegei, Nature (London), 1992,357,477. 12 R. H. Friend, D. D. C. Bradley and A. B. Holmes, Phys. World, Nov. 1992, 42. 13 D. D. C. Bradley, Adu. Muter., 1992,4,756. 14 D. A. Halliday, P. L. Burn, D. D. C. Bradley, R. H. Friend, 0. M. Gelsen, A. B. Holmes, A. Kraft, J. H. F. Martens and K. Pichler, Adu. Muter., 1993,5,40. 15 See W.Tachelet and H. J. Geise, Abstract book, ICSM’92, Gothenburg, Sweden, 1992; W. Tachelet, H. J. Geise and J. Gruner, to be published. 16 H. Vestweber, J. Oberski, A. Greiner, W. Heitz, R. F. Mahrt, H. Bassler, Adu. Muter. Opt. Electron 1993, 2, 197; H. Vestweber, A. Greiner, U. Lemmer, R. F. Mahrt, R. Richert, W. Heitz and H. Bassler, Adu. Muter., 1992,4,661. 17 Z. Yang, H. J. Geise, M. Mehbod, G. Debrue, J. W. Visser, E. J. Sonneveld, L. Van’t dack and R. Gijbels, Synth. Met., 1990, 39, 137. J. Nouwen, D. Vanderzande, H. Martens, J. Gelan, Z. Yang and H. J. Geise, Synth. Met., 1992,46,23. 18 S. Jacobs, W. Eevers, G. Verreyt, H. J. Greise, A. De Groot and R. Domisse, Synth. Met., in the press. 19 E. Barnett, Berichte 1932,65B, 1563. 20 J. Michl, E. W. Thulstrup, Spectroscopy with Polarised Light, VCH, Weinheim, 1986; T. Bjerrnholm, N. B. Larsen, F. E. Christensen, P. Sommer-Larsen, T. Skettrup and M. Jnrrgensen, Synth. Met., 1993,57, 3813. 21 D. J. Phelps and C. P. Flynn, Phys. Rev. B, 1976, 14, 5279; R. Zallen, The Physics of Amorphous Solids, John Wiley, Chichester, 1983,4, p. 187. 22 I. D. Parker, J. Appl. Phys., submitted. 23 W. Tachelet, H. J. Geise and J. Gruner, personal communication. 24 See eg., Metal Ions in Biological Systems, ed. H. Sigel and A. Sigel Marcel Dekker, New York, 1991, vol.27; R. R. Dogonadze, Reactions of Molecules at Electrodes, ed. N. S. Hush, John Wiley, Chichester, 1971,p. 135. Paper 3/07313K; Received 10th December, 1993