Observation of Electroluminescence and Photovoltaic Response in Ionic Junctions

See allHide authors and affiliations

Science  08 Sep 2006:
Vol. 313, Issue 5792, pp. 1416-1419
DOI: 10.1126/science.1128145


Electronic devices primarily use electronic rather than ionic charge carriers. Using soft-contact lamination, we fabricated ionic junctions between two organic semiconductors with mobile anions and cations, respectively. Mobile ionic charge was successfully deployed to control the direction of electronic current flow in semiconductor devices. As a result, these devices showed electroluminescence under forward bias and a photovoltage upon illumination with visible light. Thus, ionic charge carriers can enhance the performance of existing electronic devices, as well as enable new functionalities.

Junctions between n-type and p-type semiconductors are the cornerstone of modern electronic materials technology because they enable a variety of solid-state devices such as transistors, light-emitting diodes, and photovoltaic cells (1). Diffusion of electronic charge across a pn junction sets up a built-in potential, which allows current to flow preferentially in one direction (rectification). Rectification is also observed in junctions of ionic conductors, where a built-in potential arises from diffusion of anions (N) and cations (P) across a membrane. Such ionic junctions play an important role in biology. In bilayer membranes, for example, voltage-gated ion channels control the direction of ion transport into and out of cells (2). Although such systems provide an inspiration for rectifying devices, their inherent complexity makes it difficult to study systematically. Electrolytic junctions formed by using positively and negatively charged polymer solutions offer a synthetic alternative. For example, by defining the interface with a permeable membrane, a junction was formed between a solution of a polymeric acid and a solution of a polymeric base, with mobile protons (H+) and hydroxide ions (OH), respectively. Under an applied alternating current, this type of junction showed considerable rectification of ionic current (3).

The analogy between electronic (pn) and ionic (PN) junctions has not been exploited in solid-state devices. In addition to the fundamental merit of exploring the coupling between ionic and electronic carriers, such junctions may result in devices with improved performance and new functionalities. Progress toward this goal has been hampered by a lack of materials that combine semiconducting behavior with appreciable ionic conductivity. The addition of salts in films of conjugated polymers led to devices in which ion concentration gradients enhance the injection of electronic charge and lead to efficient electroluminescence (4). These gradients in ion concentration are established by an applied bias. As a result, these devices do not show inherent rectification, and the ion gradients tend to disappear when the bias is removed (4). The electrochemical deposition of junctions of polyacetylene films with oppositely charged ionomers has led to rectification (5). However, the rectification was attributed to an asymmetry in the ion polarization processes near the metal electrodes and the inability of one of the layers to transport electrons. Removal of the mobile ions led to the formation of purely electronic pn junctions (6). Electroluminescence or photovoltaic response was not reported in these devices. Electrochemical deposition also does not produce abrupt junctions and is of limited applicability to a subset of organic materials.

Over the past few years, several organic semiconductors with intrinsic ionic conductivity have emerged. A particular example is ionic transition metal complexes, such as [Ru(bpy)3]2+(PF 6)2, where bpy is 2,2′-bipyridine (Fig. 1). These materials, studied extensively in electrochemistry and spectroscopy (7), have recently attracted renewed interest in solid-state electroluminescent devices (810). The [Ru(bpy)3]2+ ion is an intrinsic semiconductor with a negligible concentration of mobile electrons and holes at room temperature due to a large HOMO-LUMO (highest occupied molecular orbital–lowest unoccupied molecular orbital) gap (7). However, substantial concentrations of electrons and holes can be injected into a film from metal electrodes where they migrate by hopping and recombine to give rise to light emission (11). An identifying feature of these materials is the mobile counter ions (PF 6), which redistribute under the application of an applied bias and assist the injection of electronic charge (11). The redistribution of counter ions is usually slow, giving rise to a characteristic delay between the application of a bias and the emission of light (12).

Fig. 1.

Structure and energetics of the PN junction. (A) Structure of the PN junction. (B) Detail of the junction area. (C) Relevant energy levels at the junction before (dotted lines) and after (solid lines) ion diffusion. The arrows in (B) indicate the direction in which the counter ions diffuse to form the junction when the two layers are brought in contact.

In addition to ionic transition metal complexes, any organic semiconductor can, in principle, be modified with the addition of appropriate groups that endow ionic conductivity. Considerable interest has focused on conjugated polymers with ionomers attached on side chains (13). Small molecules can be similarly modified. One example is the molecule 9,10-diphenylanthracene-2-sulfonate with sodium counter ions (DPASNa+) (Fig. 1), which has been used as a water-soluble fluorescence probe and for the generation of electrogenerated chemiluminescence in solution (14, 15). The ability to introduce ionic conductivity to practically any organic semiconductor enables the synthesis of materials with a broad range of electronic and ionic transport characteristics.

In analogy to semiconductors, ionic conductivity can be classified as N-type in materials such as [Ru(bpy)3]2+(PF 6)2, where the anions are the predominant carriers of ionic current, and P-type in materials such as DPASNa+, where the opposite is true. Because both N-type and P-type ionic conductors are available, ionic (PN) junctions can be fabricated in direct analogy to conventional semiconductor electronic (pn) junctions (1). Mechanistically, such a PN junction is similar to a pn junction: Gradients in the concentration of mobile ions at the junction result in diffusion of cations into the N-type material and anions into the P-type. As ions diffuse, a built-in potential that opposes diffusion is established. Equilibrium is reached when the flux due to ion diffusion is balanced by the flux due to ion drift caused by the built-in potential. This built-in potential modifies the energy levels for electrons and holes at the junction, in a manner analogous to band bending in crystalline semiconductors. Similar to a pn junction, the built-in potential directs photogenerated electronic carriers toward opposite electrodes. The interplay between junction energetics (determined by the semiconductor molecules) and built-in potential (induced by the redistribution of ions) provides a powerful means of controlling device performance.

The fabrication of such junctions is challenging. Vapor deposition techniques, which are usually used to fabricate high-quality heterojunctions of organic semiconductors, are of limited use due to the low vapor pressure of ionic materials. Successive deposition from solution is also challenging because dissolution of the bottom layer leads to intermixing of the two materials, generally precluding the formation of an abrupt junction. These problems were circumvented by using the technique of soft-contact lamination (16, 17). A DPASNa+ film was deposited on a glass substrate with patterned indium tin oxide (ITO) electrodes. A [Ru(bpy)3]2+(PF 6)2 film was deposited on a polydimethylsiloxane (PDMS) substrate with patterned Au electrodes. The two substrates were brought together, laminating the two organic layers to yield the structure ITO/DPASNa+//[Ru(bpy)3]2+(PF 6)2/Au (Fig. 1), where // indicates the location of the laminated interface. The elastomeric PDMS substrate enabled the lamination by conforming to the contours of the bottom half of the device, and the transparent ITO electrode allowed optical access to the junction (18).

To better understand the operation of the PN junction, it is important to consider the behavior of its components. The temporal evolution of the current and radiant flux of ITO/[Ru(bpy)3]2+(PF 6)2/Au devices at ±3V are shown in Fig. 2A. The curves show the usual response for organic semiconductor devices with mobile ions, with a slow turn-on time consistent with redistribution of the PF 6 counter ions (11). At steady state, the device does not show rectification, consistent with previous reports (11). The lack of rectification is a result of the ionic nature of this material. When ITO is biased positive with respect to Au, the PF 6 counter ions accumulate near the ITO electrode, leaving uncompensated [Ru(bpy)3]2+ near the Au electrode. The ionic charge leads to high electric fields near the electrodes, which help inject holes from ITO and electrons from Au into the [Ru(bpy)3]2+ molecules. Reversing the bias leads to accumulation of the PF 6 counter ions near the Au electrode, and the electric fields at the interface assist the injection of electrons from ITO and holes from Au (11). Similar characteristics were obtained for the P-type layer (Fig. 2B). The emission from the ITO/DPASNa+/Au device was below the noise threshold of the integrating sphere/photodetector assembly. It was estimated in a separate experiment to be on the order of 10–11 and 10–10 W for reverse and forward bias, respectively. These results show that both electrons and holes are injected and are mobile in [Ru(bpy)3]2+(PF 6)2 and DPASNa+ films.

Fig. 2.

Temporal response of the current and radiance of the various devices at forward and reverse bias. (A) ITO/[Ru(bpy)3]2+(PF6)2/Au, (B) ITO/DPASNa+/Au, and (C) ITO/DPASNa+//[Ru(bpy)3]2+(PF6)2/Au.

In contrast to its constituent layers, the PN junction shows pronounced rectification. Figure 2C shows the current in ITO/DPASNa+//[Ru(bpy)3]2+(PF6)2/Au to be more than 104 times higher under forward bias (ITO positive) than under reverse bias (±5 V, 20 min after application of a bias). At the junction, counter ion diffusion is expected to lead to excess negative ionic charge in the P-side of the junction and excess positive ionic charge in the N-side of the junction. The resulting built-in potential raises the HOMO and LUMO levels of DPAS relative to [Ru(bpy)3]2+ (Fig. 1C), which suppresses the transport of holes from DPAS to [Ru(bpy)3]2+ (and the transport of electrons in the opposite direction). However, at short-circuit (zero applied bias), the Fermi levels of the two metals will align. This will be accommodated by additional potential drops near the metal contacts, caused by ion redistribution (18). These potential drops assist the injection of holes from ITO in DPAS and electrons from Au in [Ru(bpy)3]2+. Under an applied bias, the steady-state current is determined by the complex interplay between ion-induced potentials, the potentials associated with injected electronic carriers, and the energy levels of the constituent materials. Device simulations (fig. S2) show that the magnitude and the spatial extent of the built-in potential vary with bias as in a pn junction. To the first order, the application of forward bias suppresses the built-in potential and allows holes from DPAS to be injected in [Ru(bpy)3]2+ (and vice versa for electrons). At the same time, enhanced charge injection from the metal contacts takes place. Reverse bias increases the magnitude of the built-in potential at the junction. As a result, only a small fraction of the applied bias drops near the contacts to help inject electronic carriers, leading to poorer injection and an overall lower current than under forward bias. As expected, the direction of current flow was reversed in control devices in which the order of the two layers was reversed (ITO/[Ru(bpy)3]2+(PF6)2//DPASNa+/Au).

The PN junctions show intense light emission under forward bias, with a luminance of 500 cd/m2 at 5 V. Their spectral response (Fig. 3) reveals that emission arises from the [Ru(bpy)3]2+(PF 6)2 side of the junction. This is consistent with the energy-level offsets at the heterojunction (Fig. 1C). The barrier for hole injection from DPASNa+ into [Ru(bpy)3]2+(PF 6)2 is 0.7 eV smaller than that for electron injection from [Ru(bpy)3]2+(PF 6)2 into DPASNa+. No light emission was detected at reverse bias (the radiant flux was lower than 10–12 W). In the absence of mobile ions, a device with energy levels as in Fig. 1C (dotted lines) and with ITO and Au electrodes would show hole-only current without substantial rectification or light emission. Rectification and light emission in traditional organic light-emitting diodes are associated with the use of an anode and a cathode with a high and a low work function, respectively.

Fig. 3.

Electroluminescence spectra of the various devices.

Illumination of the junctions resulted in a photovoltaic response. Under 100 mW/cm2 excitation from a halogen lamp, an open circuit voltage of 0.45 V and a short-circuit current density of 0.15 μA/cm2 were measured (Fig. 4). Upon sudden illumination the photocurrent took a few seconds to reach steady state, presumably due to redistribution of the ionic carriers under the influence of the photogenerated electrons and holes. The data of Fig. 4 were acquired at steady state. The intensity dependence of the photocurrent was measured up to 500 mW/cm2 and was found to be sublinear, with an exponent of 0.85. The internal quantum efficiency (electrons collected per photon absorbed) was estimated to be on the order of 5 × 10–5. This is low compared to state-of-the-art organic photovoltaics but expected given the low exciton dissociation yield in these materials. Indeed, devices based on the constituent layers did not show any photovoltaic response, confirming negligible exciton dissociation in the bulk and at the electrode interfaces.

Fig. 4.

Photovoltaic response of the PN junction.

The photovoltaic response of the junctions is consistent with the built-in potential assisting the dissociation of excitons and the separation of electrons and holes in the [Ru(bpy)3]2+(PF6)2 and the DPASNa+ layers, respectively (Fig. 1C). Estimating the magnitude of the built-in potential is difficult, because the ion density near the junction interface is hard to predict due to steric effects associated with ion packing. The value of the open circuit voltage represents a lower limit for the built-in potential, considering that there is some resistive loss across the two layers and at the contacts. It should be mentioned that the use of mobile ions in photovoltaics is well established in Grätzel cells (19). In these devices, ions in an electrolyte extract and transport holes from charge-transfer dyes attached to titanium dioxide. In the ionic junctions reported here, the ions establish a built-in potential across a heterojunction.

The observations of rectification in current and light emission, and of a photovoltaic response in the PN junction and not in the constituent layers, demonstrate that mobile ionic charge can be used to control the flow of electronic current in solid-state devices. In principle, any organic semiconductor can be modified with ionomers to endow unipolar ionic conductivity, and the technique of lamination described here can be easily applied to fabricate PN junctions from these materials. Such junctions might help decrease recombination and increase the efficiency of organic heterojunction solar cells (20). Moreover, given that ionic and electronic mobilities often differ by several orders of magnitude, the possibility of realizing ionic junctions that can be reconfigured by the prolonged application of a bias and be used at faster time scales to rectify electronic current is exciting.

Supporting Online Material

Materials and Methods

Figs. S1 and S2

References and Notes

Stay Connected to Science

Navigate This Article