IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 4, NO. 1, JANUARY/FEBRUARY

Transcription

1 IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 4, NO. 1, JANUARY/FEBRUARY Electrical Characterization of Polymer Light-Emitting Diodes Paul W. M. Blom and Marc J. M. de Jong Abstract This paper presents a device model for the current and light generation of polymer light-emitting diodes (PLED s). The model is based on experiments carried out on poly(dialkoxyp-phenylene vinylene) (PPV) devices. It is demonstrated that PLED s are fundamentally different as compared to conventional inorganic LED s. The hole conduction in PPV is space-charge limited with a low-field mobility of only m 2 /Vs, which originates from the localized nature of the charge carriers. Furthermore, the hole mobility is highly dependent on the electric field and the temperature. The electron conduction in PPV is strongly reduced by the presence of traps. Combining the results of the electron- and hole transport a device model for PLED s is proposed in which the light generation is due to bimolecular recombination between the injected electrons and holes. It is calculated that the unbalanced electron and hole transport gives rise to a bias dependent efficiency. By comparison with experiment it is found that the bimolecular recombination process is of the Langevin-type, in which the rate-limiting step is the diffusion of electrons and holes toward each other. This is in contrast to conventional semiconductors, in which the bimolecular recombination is governed by the joint densityof-states of electrons and holes and is not limited by charge transport. The occurrence of Langevin recombination explains why the conversion efficiency of current into light of a PLED is temperature independent. The understanding of the device operation of PLED s indicates directions for further improvement of the performance. Index Terms Charge transport, device physics, polymer lightemitting diodes. I. INTRODUCTION SINCE their discovery in 1990 polymer light-emitting diodes (PLED s) have been considered promising candidates for large-area applications as a result of both easy processing and mechanical flexibility [1], [2]. A typical PLED consists of a thin layer of undoped conjugated polymer sandwiched between two electrodes on top of a glass substrate. Experimentally, attention has especially been focused on PLED s that contain the conjugated polymer poly(phenylene vinylene) (PPV) or its derivatives which may have an external conversion efficiency larger than 1% photons/charge carrier. The PPV is spin-coated on top of a patterned indium tin oxide (ITO) bottom electrode, which forms the anode. The cathode on top of the polymer consists of an evaporated metal layer Manuscript received September 22, 1997; revised February 14, This work was supported by the European Commission (BRITE-EURAM) under Contract POLYLED BRE2-CT and by ESPRIT Contract LEDFOS The authors are with the Philips Research Laboratories, 5656 AA Eindhoven, The Netherlands. Publisher Item Identifier S X(98) Fig. 1. Schematic band diagram of a poly(dialkoxy p-phenylene vinylene)-based PLED under forward bias using ITO as a hole- and Ca as an electron injector. for which Ca is used. Under forward bias electrons and holes are injected from the cathode and the anode, respectively, into the polymer, as schematically indicated in Fig. 1. Driven by the applied electric field, the charge carriers move through the polymer over a certain distance until recombination takes place. The device operation of a PLED is thus determined by three processes: charge injection, charge transport, and recombination [3]. Studies of the current density versus voltage characteristics of PPV-based LED s have yielded conflicting information about which process dominates the operation of a PLED. From the absence of a thermally activated behavior as well as a highly nonlinear characteristic at high-electric field, it has been suggested [4], [5] that the dependence of on resembles that of Fowler Nordheim tunneling through a barrier. This suggests that the operation of a PLED is limited by the tunneling of electrons and holes through contact barriers arising from the band offset between the polymer and the electrodes. However, quantitatively the currents predicted by the Fowler Nordheim theory exceed the experimentally observed currents by several orders of magnitude. Furthermore, at low electric fields the tunneling model was found not to be applicable to the experimental characteristics, which has been attributed to space-charge effects [4], the contribution of thermionic emission to the current [5], band-bending effects at the interface [6], and backflow of injected carriers into the injecting contacts as a result of the low mobility of the organic materials [7]. Following the tunneling model the device performance of PLED s can be improved by balancing and reducing the contact barriers of both electrons and holes X/98$ IEEE

2 106 IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 4, NO. 1, JANUARY/FEBRUARY 1998 Fig. 2. Experimental and calculated (solid lines) J V characteristics of ITO/PPV/Au hole-only devices with thicknesses L = 130 nm (squares), 300 nm (triangles), and 700 nm (dots). The device structure is indicated in the inset. The hole transport for all samples is described by SCLC [see (1)] with a hole mobility p = cm 2 /V1s and a dielectric constant r = 3. An unbalanced injection results in an excess of one carrier type and, thus, in a current flow, which partially does not contribute to light emission. On the other hand, it has been found that the transport properties of electrons and holes in PPV are highly asymmetric. Time-of-flight measurements by Antoniades et al. [8] have shown that electrons, in contrast to holes, are severely trapped in PPV. Clearly, the resulting unbalanced electron and hole transport must affect the characteristics and efficiency of PLED s. As a result of the reduced electron transport the electroluminescence is confined to a region close to the cathode. Since metallic electrodes are very efficient quenching centres for the generated electroluminescence (EL) this confinement is expected to strongly decrease the device performance. It is important to have a model that accurately describes the device operation of PLED s. A thorough understanding of the device characteristics of PLED s will give insight in how to improve the performance of present and future devices. Recently, we have performed elaborate measurements and model calculations on PPV based devices. The combination of the experimental results and the calculations have revealed that the device properties of our PLED s are determined by the bulk-conduction properties of PPV and not by charge injection. This paper provides an overview of our results obtained so far. II. SINGLE CARRIER DEVICES A. Experimental Setup The devices that we have studied consist of a single polymer layer sandwiched between two electrodes on top of a glass substrate. The polymer is soluble poly( dialkoxy-p-phenylene vinylene) [9], shown in the inset of Fig. 1 with R CH and R C H, which is spin-coated on a patterned bottom electrode. In order to study the transport properties of holes an ITO bottom contact and an evaporated Au top contact are used. In these hole-only devices [5], the work functions of both electrodes are close to the valence band of PPV preventing electron injection from the negatively biased electrode, as indicated in the inset of Fig. 2. The transport properties Fig. 3. Experimental J V characteristics of a ITO/PPV/Au hole-only device with thickness L = 125 nm for various temperatures. In the low field part the predicted J V characteristics (dotted lines) according to the conventional SCLC model [(1)] are shown. Also shown are the calculated J V characteristics as predicted by a SCLC model using the field dependent mobility defined by (2) and (3) using 0 = m 2 /V1s 1=0.48 ev, G = ev(v/m) 0(1=2) and T 0 = 600 K. of electrons have been investigated using an electron-only device consisting of a PPV layer sandwiched between two Ca electrodes. Ca has a work-function close to the conduction band of PPV, as shown in the inset of Fig. 5. Finally, for the double-carrier PLED s ITO is used as a hole injector and Ca as an electron injector, as indicated in Fig. 1. The measurements have been performed in a nitrogen atmosphere in a temperature range of 200 K 300 K using a HP 4145A semiconductor parameter analyzer. The detection limit of our setup is about a few pa, which corresponds to 10 A/m for the electrode area of 10 m. B. Hole-Only Devices In Fig. 2 the characteristics of several ITO/PPV/Au hole-only devices are presented [10]. From the slope of the versus plot we observe that the current density depends quadratically on the voltage. This behavior is characteristic for space-charge limited current (SCLC) in which case [11] where is the permittivity of the polymer, the hole mobility, and the thickness of the device. Assuming 3, we find that the characteristics of our devices, with 130, 300, and 700 nm, respectively, are described well by (1) using 5 10 m /V s. The importance of the observation of space-charge limited current in our hole-only devices is that it clearly demonstrates that the hole current is bulk-limited and not injection-limited, as proposed before [4], [5]. The discrepancy with these earlier results might result from differences in the processing conditions or the materials used. At high electric fields ( 3 10 V/cm) we observe a gradual deviation from (1) as demonstrated in Fig. 3 where the experimental characteristics of a hole-only device with 125 nm are plotted for six temperatures between 209 K 296 K. The behavior according to (1) is also plotted at low fields as dashed lines. At higher voltages, it is (1)

3 BLOM AND DE JONG: ELECTRICAL CHARACTERIZATION OF POLYMER LIGHT-EMITTING DIODES 107 Fig. 4. Temperature dependence of the Poole Frenkel factor as obtained from the high-field J V characteristics of Fig. 3. Also shown is the empirical dependence (solid line) [(3)] as proposed by Gill [16]. Using this empirical description we obtain for our PPV a T0 of 600 K and a constant G of ev(v/m) 0(1=2), which are both in close agreement with the values reported by Gill [16] for PVK. observed that the current density is larger than as is expected from (1). This suggests that the carrier mobility increases with electric field [12]. In 1970, Pai [13] demonstrated by time-of-flight (TOF) measurements that at high-electric field the mobility of photo-injected holes in poly -vinylcarbazole) (PVK) can be described by with a temperature independent prefactor, activation energy, Boltzmann s constant, and temperature. The field dependence (2) through the coefficient is comparable to the Poole Frenkel effect [14]. However, it was soon recognized that the physical origin of (2), which appears to be applicable to a large variety of molecular doped polymers as well as molecular glasses, is not the presence of traps, but it is related to the intrinsic charge transport in disordered materials [15]. In order to describe the hole conduction in PPV at both low and high fields, we combine the SCLC, (1), with the field-dependent mobility of (2). We observe that the hole mobility is thermally activated and that it decreases over more than three orders of magnitude while going from 296 K 209 K. The low-field mobility can be described by (2), using 0.48 ev and m /Vs. At higher voltages, we extract for our PPV at each temperature. Fig. 4 suggests that the observed temperature dependence of can accurately be described by an empirical expression: This result has also been found in PVK by Gill [16], with ev(v/m) and 520 K 660 K, depending on the molecular dopant density. For our PPV we obtain ev(v/m) and 600 K. In Fig. 3, it is demonstrated that the combination of SCLC and a field-dependent mobility provides an excellent agreement with the experimental curves in both the low- and high-voltage regime [12]. (2) (3) Fig. 5. Thickness dependence of the electron only J V characteristics at L = 220, 310, and 370 nm. The solid lines represent the predicted J V characteristics according to (5). For this set of samples the trap density N t = cm 03. For comparison also the J V characteristics of a hole-only device with thickness L = 0.3 m is shown. A microscopic theory for the observed mobility according to (2) and (3) is still lacking. Monte Carlo simulations by Bässler [15] of hopping between sites that are subject to both positional and energetic disorder agree with (2) and (3) in a limited-field regime. Taking into account spatial correlations in the energetic disorder improves agreement with experiment [17], [18]. A theoretical explanation for the mobility, which seems to be applicable to a large class of disordered materials is highly wanted, since it would clarify the physical meaning of the invoked parameters, determined empirically here. C. Electron-Only Devices The transport properties of electrons are investigated using an electron-only device consisting of a PPV layer sandwiched between two Ca electrodes, which have a work function close to the conduction band of PPV (see inset of Fig. 5). In Fig. 5, the characteristics of electron-only devices with thickness 220, 310, and 370 nm are shown together with the characteristic of a hole-only device with 300 nm. The electron current is smaller than the hole current, especially at low bias where the difference amounts to three orders of magnitude. Furthermore, the electron current shows a very strong field-dependence which is characteristic for the trapfilled limit (TFL) in an insulator with traps [11]. For trap levels located at a single energy there is an extremely sharp transition, at which the current directly switches to the SCLC. The more gradual increase, as observed in Fig. 5, points to a distribution of trap-level energies. Actually, this is what one would expect for a disordered system, such as a spin-coated conjugated polymer. The characteristics in Fig. 5 are well described using an exponential distribution of traps with the trap density of states at energy the energy of the conduction band, the total density of traps, and an energy characterizing the trap distribution. The trap distribution (4) implies for the characteristic in the TFL (4)

4 108 IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 4, NO. 1, JANUARY/FEBRUARY 1998 regime [11] (5) with, the effective density of states in the conduction band estimated as cm and. The characteristics in Fig. 5 are then determined by three parameters: the carrier mobility, the trap density, and the trap distribution parameter. The latter follows directly from the slope of the versus characteristic in the TFL regime, according to (5), which yields 1500 K. Assuming [19], we find that 1 10 cm. Using these parameters we find good agreement between our experimental and theoretical results, as shown in Fig. 5. The observed thickness dependence, which is characteristic for an exponential trap distribution, clearly proves that the electron current is determined by the bulk-transport properties of the PPV. In summary: our results confirm the severe trapping of electrons as observed in time-of-flight measurements by Antoniades et al. [8]. III. DOUBLE-CARRIER DEVICES A. Device Model The results obtained in this study so far enable us to further investigate the relevance of the unbalanced electron and hole transport to the device efficiency of a polymer LED. Now we propose a model for the and efficiency versus behavior of a PPV-based PLED taking into account the effects of both space-charge (holes) and trapping (electrons). For a double-carrier device two additional phenomena become of importance, namely recombination and charge neutralization. We will assume that the recombination is bimolecular, i.e., that its rate is proportional to the product of the electron and hole concentrations. Due to charge neutralization the total charge may far exceed the net charge. As a result, the current density in a double-carrier device can be considerably larger than in a single-carrier device. In the simple case without traps and a field-independent mobility the double carrier current (plasma limit) is given by [11] with the bimolecular recombination constant. Upon increasing the amount of neutralization decreases, so that becomes smaller. The difference in current between a singleand a double-carrier device [see (1) and (6)] provides direct information about the strength of the recombination process. We compare the characteristics of an ITO/PPV/Au hole-only device with an ITO/PPV/Ca double injection device in Fig. 6 at room temperature. The device thickness 130 nm. Unfortunately, (6) is not directly applicable to the experimental results of Fig. 6, since electron traps and a fielddependent mobility are not included. Therefore, we present a device model which is characterized by the current-flow (6) Fig. 6. Experimental and calculated J-V characteristics for a single- (ITO/PPV/Au) and double- (ITO/PPV/Ca) injection device with a thickness of L = 130 nm. From the difference betwee the single- and double carrier current a recombination constant B = cm 3 /s has been obtained. equation: the Poisson equation: and the particle-conservation equations: In the above equations, and represent the density of mobile holes and electrons, the density of trapped electrons, and the electric field as a function of position. For the boundary conditions we assume that the injection is diffusion limited [10]. Furthermore, the electron trap distribution as a function of energy is assumed to be exponential with a characteristic energy, according to (4). To determine we assume quasi-equilibrium between the trapped and free electrons [11]. The set of (7) (9) can be solved numerically. As stated above, the double-injection current now only depends on the bimolecular-recombination strength. The experimental results in Fig. 6 for the double-carrier device can be modeled from (7) to (9), with B as the only unknown parameter. Agreement with experiment is obtained for a recombination constant of 2 10 cm /s. The total number of recombination inside the PLED as determined from the characteristics is now compared with the actual light output. We obtain that only 5% of the total recombination contributes to the light output of the device. This is in agreement with the estimate that due to spin statistics the maximum EL yield will be no more than 1/4 of the photoluminescence (PL) yield, which amounts to 15% for our PPV. The recombination processes in a PLED are thus mainly nonradiative. B. Conversion Efficiency An important device parameter of a PLED is the conversion efficiency (CE), defined as photon/charge carrier. In Fig. 7, (7) (8) (9)

5 BLOM AND DE JONG: ELECTRICAL CHARACTERIZATION OF POLYMER LIGHT-EMITTING DIODES 109 Fig. 7. Efficiency (photons/carrier) versus voltage V for a ITO/PPV/Ca double injection device from experiment (dots) and model (lines). Agreement with experiment is obtained by taking into account quenching of the excitons at the cathode in a region with a length Lq = 10 nm. the normalized efficiency CE/CE is shown as a function of applied voltage. By taking into account out coupling losses the internal efficiency of 5% is further reduced to an external CE of 2%. The applied voltage is corrected for the builtin voltage ( 1.5 ev), which originates from the work function difference between ITO and Ca. It is observed that the CE is dependent on the applied voltage, which is a fundamental difference between a PLED and a conventional semiconductor LED. The CE gradually increases from zero at the turn-on voltage to its maximum value at 5 V. In order to understand this behavior, we have compared the calculated charge carrier distributions at both low- V) and high voltage (5 V), as shown in Fig. 8. At 1V it is observed that most of the electrons injected at are trapped and the free electron density, which is responsible for the light output, decreases strongly with increasing distance from the cathode. At 5 V, however, nearly equals and the positional variation of is now considerably reduced. As a result, as demonstrated in Fig. 9, at 1 V the corresponding light-output is mainly generated in a region close to the cathode, whereas at 5 V the light-output is generated more homogeneously in the PLED. We suggest that the reduced efficiency at low voltage is a result of nonradiative recombination losses at the interface, where a large number of exciton quenching centres are present. In our device model this quenching behavior can be approximated by neglecting the light-output in a region between and, with representing the width of the quenching region. In Fig. 7 the calculated CE as a function of voltage is shown for several values of. Agreement with experiment is obtained for 10 nm, which is in accordance with the exciton diffusion length of 9 1nm as obtained in PPV-based photovoltaic devices [20]. C. Recombination Process in a PLED Our studies on single carrier hole-only devices (Fig. 3) have revealed that, in contrast to conventional semiconductor LED s, the charge transport in a PLED is strongly dependent on temperature. For applications it is important to know and understand how this thermally activated transport affects Fig. 8. Calculated charge distributions as a function of distance at low (1 V) and high (5 V) bias voltage. At low voltage, most of the electrons are trapped and the free electrons are localized in a region close to the cathode (x = L). Fig. 9. Calculated distribution of the light generation in a PLED. At low voltage most of the light output is generated in a region close to the cathode, at higher voltage the light output is generated more homogeneously in the device. the device performance of a PLED. Let us assume that the mobility is thermally activated and that. In the most simple case without traps and a field-independent mobility the temperature dependence of the hole-only current [see (1)] is directly governed by whereas the LED current [see (6)] is determined by. Thus for a constant the relative difference between a single- and a double-carrier device is expected to decrease with decreasing temperature. The temperature dependence of the charge transport in polymer devices is investigated by performing measurements on both hole-only and double-carrier devices [21].

6 110 IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 4, NO. 1, JANUARY/FEBRUARY 1998 In Fig. 10, the characteristics of both an ITO/PPV/Au device and an ITO/PPV/Ca device are shown for various temperatures. The thickness of the polymer layer amounts to 280 nm. It is obtained that the relative difference between the single- and double-carrier current is rather independent of the temperature, even though according to (1) and (6) it is expected to decrease with decreasing temperature. Theoretical results are plotted as solid lines in Fig. 10. Clearly, our theory is in good agreement with experiment over a large voltage and temperature range. The obtained values for the bimolecular recombination constant are shown in an Arrhenius plot in Fig. 11. For our present PPV device ( 280 nm) again a 2 10 cm /s has been obtained at room temperature. We observe that is thermally activated with an activation energy equal to that of the charge carrier mobility [21]. This result strongly indicates that the recombination strength in a PLED is of the Langevin type [22], i.e., diffusion controlled. Such a behavior is characteristic for materials in which the mean free path of the charge carriers is smaller than a critical distance at which the Coulomb binding energy between an electron and hole equals. For an organic hopping system with a low-carrier mobility, the bimolecular recombination mechanism is expected [23] to be of the Langevin type and has recently been included in an analytical model for PLED s [24]. In our PPV, the mean free path is approximately equal to the distance between the conjugated sites which is of the order of 10 Å, whereas 185 Åat 300 K ( 3). As a result the rate-limiting step in the bimolecular recombination process is the diffusion of electrons and holes toward each other in their mutual coulomb field. This implies for the recombination [22] Fig. 10. Experimental and calculated (solid lines) J V characteristics for an ITO/PPV/Au hole-only device (squares) and an ITO/PPV/Ca double-carrier device (circles) with a thickness of L = 280 nm at T = 303 K, T = 252 K and T = 212 K. The double-carrier J V characteristics are corrected for a built-in voltage V bi of 1.5 ev which arises from the work function difference between the ITO and the Ca contact. The hole-only device is modeled using SCLC in combination with a field dependent mobility according to (2) and (3), the double injection current is numerically solved from (7) to (9). (10) which is also plotted in Fig. 11. The as determined from our characteristics are slightly (factor 3 4) larger than the values predicted by (10). A possible explanation for this small discrepancy might be the fact that recombination from trapped electrons may play a role as well [25]. It should be noted that by including Langevin recombination [(10)] into (6), the double-carrier current J becomes proportional to, which for implies that is proportional to instead of. An equal temperature dependence for the hole-only and the double-carrier device is the result, in agreement with the experimental results shown in Fig. 10. In fact, the temperature independence of the difference between the single- and double-carrier current is a direct demonstration that the recombination process is proportional to the charge-carrier mobility. Now, we discuss the consequences of Langevin recombination for the device performance of PLED s. In Fig. 12, the measured external conversion efficiency CE, defined as photon/charge carrier, is shown for a ITO/PPV/Ca PLED with 110 nm at various temperatures. It appears that the CE of a PLED is independent of the temperature, in contrast to conventional semiconductor LED s. This typical behavior of a PLED is a direct consequence of the occurrence of Langevin recombination. In a space-charge limited device the number Fig. 11. Temperature dependence of the bimolecular recombination constant (circles) as directly obtained from the J-V characteristics as shown in Fig. 10 using the device model defined by (7) (9). Also shown is the temperature dependence of the Langevin bimolecular-recombination constant (solid line) as predicted by (10) with n = p. of charge carriers is mainly determined by the applied voltage and not by the temperature. As a result, the device current is only dependent on temperature by means of the chargecarrier mobility. Since the recombination mechanism is of the Langevin type, both the recombination strength and the device current are thermally activated by means of the charge carrier mobility. Since in PPV the photoluminescence (PL) efficiency is relatively weakly dependent [26], [27] on temperature in the range of 200 K 300 K, it is clear that the resulting CE of a PLED must also exhibit only a weak temperature dependence. It should be noted, however, that in other studies a stronger variation of the PL efficiency of PPV-related materials with temperature has been observed [1], [28], which might be related to the microscopic properties of the materials [28]. D. Enhancement of the Device Performance Our study has revealed several ways to improve the device performance of a PLED. Firstly, by decreasing the large

7 BLOM AND DE JONG: ELECTRICAL CHARACTERIZATION OF POLYMER LIGHT-EMITTING DIODES 111 Fig. 12. External quantum efficiency (photons/carrier) versus voltage measured from an ITO/PPV/Ca double-carrier device with a thickness of L = 110 nm at 300, 261, and 242 K. The maximum quantum efficiency amounts to 2% and is independent of temperature. the bulk conduction properties of the polymer, and not by the injection properties of the contacts. The conduction of holes in a film of the conjugated polymer PPV is governed by a combination of a field-dependent mobility and space charge effects. The electron transport is limited by traps which are exponentially distributed in energy. A device model for PLED s is proposed which demonstrates that the recombination in a PLED is mainly nonradiative, so that only 5% of the total number of recombinations contributes to the light output of the device. From the temperature dependence of the characteristics we have demonstrated that the recombination mechanism in a PLED is of the Langevin type. Due to this diffusion- controlled recombination PLED s exhibit, in contrast to conventional inorganic LED s, a temperature independent recombination efficiency. ACKNOWLEDGMENT The authors are grateful to S. Breedijk, C. T. H. F. Liedenbaum, M. G. van Munster, and J. J. M. Vleggaar for their contributions to the work in this paper. REFERENCES Fig. 13. Applied voltage at which the light output amounts to 100 cd/m 2 as a function of temperature. The observed voltage increase with decreasing temperature originates from the decreasing carrier mobility. amount of nonradiative recombinations the maximum conversion efficiency can be enhanced. Secondly, the use of an electron transport layer will shift the recombination zone away from the metallic cathode, resulting in an efficiency increase at low voltages. Due to the occurrence of Langevin recombination the device efficiency (light output/current) of a PLED is not dependent on the temperature. However, with decreasing temperature the voltage required to maintain a certain amount of light-output (or corresponding current) increases due to the decreasing charge carrier mobility. In Fig. 13, it is demonstrated that the voltage required to maintain a light output of 100 cd/m, which is typical for applications, increases from 3.5 V at room temperature to over 8 V at 180 K. As a result, a PLED fabricated from a polymer with a low carrier mobility will only provide sufficient light output at high voltages, giving rise to a poor power efficiency. Thus for an optimum device performance both the PL efficiency and the carrier mobility of the polymer should be increased. IV. CONCLUSION We have demonstrated that the electron and hole currents in PPV devices with low contact barriers are determined by [1] J. H. Burroughes, D. D. C. Bradley, A. R. Brown, R. N. Marks, K. Mackey, R. H. Friend, P. L. Burn, and A. B. Holmes, Light-emitting diodes based on conjugated polymers, Nature, vol. 347, pp , [2] D. Braun, and A. J. Heeger, Visible light emission from semiconducting polymer diodes, Appl. Phys. Lett., vol. 58, pp , [3] D. D. C. Bradley, A. R. Brown, P. L. Burn, R. H. Friend, A. B. Holmes, and A. Kraft, Electro-optic properties of precursor route poly(arylene vinylene)polymers, Electronic Properties of Polymers, Solid State Sciences. Heidelberg, Germany: Springer, 1992, vol. 107, pp [4] R. N. Marks and D. D. C. Bradley, Charge injection and transport in poly(p-phenylene vinylene) light-emitting diodes, Synth. Metals, vol. 57, pp , [5] I. D. Parker, Carrier tunnelling and device characteristics in polymer light-emitting diodes, J. Appl. Phys., vol. 75, pp , [6] E. Ettedgui, H. Razafitrimo, Y. Gao, and B. R. Hsieh, Band bending modified tunneling at metal/conjugated polymer interfaces, Appl. Phys. Lett., vol. 67, pp , [7] P. S. Davids, Sh. M. Kogan, I. D. Parker, and D. L. Smith, Charge injection in organic light-emitting diodes: tunneling into low mobility materials, Appl. Phys. Lett., vol. 69, pp , [8] H. Antoniades, M. A. Abkowitz, and B. R. Hsieh, Carrier deep-trapping mobility-lifetime products in poly(p-phenylene vinylene), Appl. Phys. Lett., vol. 65, pp , [9] D. Braun, E. G. J. Staring, R. C. J. E. Demandt, G. J. L. Rikken, Y. A. R. R. Kessener, and A. H. J. Venhuizen, Photo- and electroluminescence efficiency in poly(dialkoxy-p-phenylene vinylene), Synth. Metals, vol. 66, pp , [10] P. W. M. Blom, M. J. M. de Jong, and J. J. M. Vleggaar, Electron and hole transport in poly(p-phenylene vinylene) devices, Appl. Phys. Lett., vol. 68, pp , [11] M. A. Lampert and P. Mark, Current Injection in Solids. New York: Academic, [12] P. W. M. Blom, M. J. M. de Jong, and M. G. Van Munster, Electric-field and temperature dependence of the hole mobility in poly(p-phenylene vinylene), Phys. Rev. B, vol. 55, pp. R656 R659, [13] D. M. Pai, Transient photoconductivity in poly(n-vinylcarbazole), J. Chem. Phys., vol. 52, pp , [14] J. Frenkel, On pre-breakdown phenomena in insulators and electronic semiconductors, Phys. Rev., vol. 54, pp , [15] H. Bässler, Charge transport in disordered organic photoconductors, Phys. Stat. Sol. B, vol. 175, pp , [16] W. D. Gill, Drift mobilities in amorphous charge-transfer complexes of trinitrofluorenone and poly-n-vinylcarbazole, J. Appl. Phys., vol. 43, pp , 1972.

8 112 IEEE JOURNAL OF SELECTED TOPICS IN QUANTUM ELECTRONICS, VOL. 4, NO. 1, JANUARY/FEBRUARY 1998 [17] Yu. N. Gartstein and E. M. Conwell, High-field hopping mobility in molecular systems with spatially correlated energetic disorder, Chem. Phys. Lett., vol. 245, pp , [18] D. H. Dunlap, P. E. Parris, and V. M. Kenkre, Charge-dipole model for the universal field dependence of mobilities in molecularly doped polymers, Phys. Rev. Lett., vol. 77, pp , [19] P. Gomes da Costa and E. M. Conwell, Excitons and the band gap in poly(phenylene vinylene), Phys. Rev. B, vol. 48, pp , [20] N. C. Greenham, Charge separation, recombination and exciton decay in conjugated polymer photodiodes and LED s, presented at the MRS Spring Meet., San Francisco, CA, [21] P. W. M. Blom, M. J. M. De Jong, and S. Breedijk, Temperature dependent electron-hole recombination in polymer light-emitting diodes, Appl. Phys. Lett., vol. 71, pp , [22] P. Langevin, Sur la loi de recombination des ions, Ann. Chim. Phys., vol. 28, pp , [23] U. Albrecht and H. Bässler, Langevin-type charge carrier recombination in a disordered hopping system, Phys. Statist. Sol. (B), vol. 191, pp , [24] J. C. Scott, S. Karg, and S. A. Carter, Bipolar charge and current distributions in organic light-emitting diodes, J. Appl. Phys., vol. 28, pp , [25] P. E. Burrows, Z. Shen, V. Bulovic, D. M. McCarty, S. R. Forrest, J. A. Cronin, and M. E. Thompson, Relationship between electroluminescence and current transport in organic heterojunction light-emitting diodes, J. Appl. Phys., vol. 79, pp , [26] M. Furukawa, K. Mizuno, A. Matsui, S. D. D. V. Rughooputh, and W. C. Walker, Time-resolved excitonic luminescence processes in poly(phenylenevinylene), J. Phys. Soc. Jpn., vol. 58, pp , [27] K. Pichler, D. A. Halliday, D. D. C. Bradley, P. L. Burn, R. H. Friend, and A. B. Holmes, Optical spectroscopy of highly ordered poly(pphenylene vinylene), J. Phys.: Condens. Matter, vol. 5, pp , [28] C. M. Heller, I. H. Campbell, B. K. Laurich, D. L. Smith, D. D. C. Bradley, P. L. Burn, J. P. Ferraris, and K. Müllen, Solidstate-concentration effects on the optical absorption and emission of poly(p-phenylene vinylene)-related materials, Phys. Rev. B, vol. 54, pp , Paul W. M. Blom was born in Maastricht, The Netherlands, in He received the Ir. degree in physics in 1988 and the Ph.D. degree in 1992, respectively, from the Technical University Eindhoven, Eindhoven, The Netherlands. His thesis work was on picosecond charge carrier dynamics in GaAs quantum wells. At Philips Research Laboratories, he was engaged in the electrical characterization of various oxidic thin-film devices. His current research interests are the electrooptical properties of polymer LED s. Marc J. M. de Jong was born in Herten, The Netherlands, in He received the Ir. degree in physics in 1992 from the Delft University of Technology, Delft, The Netherlands, and the Ph.D. degree in 1995 at the Leiden University, Leiden, The Netherlands. He has worked on shot noise and electrical conduction in mesoscopic systems, on optoelectronic properties of polymer LED s, and on sensors for automotive applications. His current activities are in the field of broad-band communication and access networks.