Dynamic Behaviors of Exciplex States in Rubrene/${\mathrm{C}}_{60}$-Based OLEDs with Sub-Band-Gap Turn-On Electroluminescence

Although the origin of sub-band-gap electroluminescence has been extensively studied for organic light-emitting diodes (OLEDs) with a planar heterojunction of rubrene-fullerene (${\mathrm{C}}_{60}$), the dynamic behaviors of exciplexes formed at the $\mathrm{rubrene}/{\mathrm{C}}_{60}$ interface are still vague. Herein, employing magnetoconductance (MC) and magnetoelectroluminescence (MEL) detection techniques, we find an unreported reverse intersystem crossing (RISC, ${\mathrm{EX}}^3$ → ${\mathrm{EX}}^1$) from triplet to singlet exciplexes dominated at the $\mathrm{rubrene}/{\mathrm{C}}_{60}$ interface. Besides this RISC evolution channel, the triplet exciplexes also participate in another two processes: one is the scattering channel of triplet-charge annihilation (TCASC, ${\mathrm{EX}}^3$ + $\overleftarrow{e}$ → $S_0$+ ↓e) with excessive charge carriers, which is disadvantageous for RISC to occur, and the other is Dexter-energy transfer to produce a triplet exciton (${\mathrm{EX}}^3$ → $T_1$) in rubrene followed by triplet-triplet annihilation (TTA, $T_1$+ $T_1$ → $S_1$+ $S_0$ → hν↑ + 2$S_0$). This TTA is the physical origin of sub-band-gap EL because TTA needs no respective injection of electrons and holes into the LUMO and HOMO energy levels and requires no direct recombination of LUMO electrons with HOMO holes in rubrene. Moreover, both the bias-current dependences of MC and MEL traces reveal that the RISC process is more obvious in relatively balanced devices due to weak TCASC processes, and they show a monotonic decrease and nonmonotonic variation in unbalanced and relatively balanced $\mathrm{rubrene}/{\mathrm{C}}_{60}$-based OLEDs, respectively, because of the combined effects of the applied electrical field and the TCASC process. Theoretically, a stronger RISC process should be observed in devices by inserting a thin bathocuproine (BCP) interlayer to modify the $\mathrm{rubrene}/{\mathrm{C}}_{60}$ interface due to the increased separation distance between electron and hole in exciplex states, but the opposite experimental results are obtained because of strong TCASC channels simultaneously occurred with the insertion of a BCP interlayer. The RISC process decreases monotonically upon lowering the operational temperature due to its endothermic property. This work deepens our comprehensive understanding of the dynamic behaviors of exciplexes and the physical origin of sub-gap EL features in $\mathrm{rubrene}/{\mathrm{C}}_{60}$-based OLEDs.

Figure 1

(a),(b) Schematic diagrams of energy-level alignments and carrier-injection processes for Dev. 1 and Dev. 2, respectively. (c) Normalized EL spectra of Dev. 1, Dev. 2, and a control device. Inset shows the molecular structure of rubrene; molecular structures of other functional materials are displayed in Appendix pp3. (d) Luminance-voltage characteristics of Dev. 1 and Dev. 2 operating at room temperature. Inset shows luminance as a function of current density in these two devices.

Figure 2

(a),(b) Current dependences of MC curves for Dev. 1 and Dev. 2, respectively, within magnetic field ranges of |B| < 150 mT. These MC curves are offset for clear observation of hyperfine structures within the field range of |B| < 10 mT. White and black solid lines are fitted curves. Corresponding raw data without shifts are shown in Appendix pp5. (c),(d) Current dependences of intensity factors (C₁, C₂, and the sum of C₁+ C₂) in Dev. 1 and Dev. 2, respectively. (e),(f) Current dependences of intensity factor C₃ in Dev. 1 and Dev. 2.

Figure 3

(a),(b) Current dependences of MEL curves for Dev. 1 and Dev. 2 within magnetic field ranges of |B| < 150 mT, respectively. White and black solid lines are fitted curves and MEL curves, respectively, of Dev. 1 and Dev. 2, which are offset for clear observation. (c),(d) Current dependences of the intensity factors (C₄, C₅, and the sum of C₄+ C₅) in Dev. 1 and Dev. 2, respectively. (e),(f) Current dependences of intensity factor C₆ in Dev. 1 and Dev. 2, respectively.

Figure 4

Schematic diagrams of two different operational mechanisms used for Dev. 1: (a) TTA mechanism; (b) Auger-electron-assisted up-conversion mechanism.

Figure 5

(a) Schematic diagram of formation processes and evolution pathways of excited states existing at the rubrene/${\mathrm{C}}_{60}$ interface and in the rubrene bulk layer. (b) Schematic diagram of the mutual transformation and energy-transfer process between excited states, the dissociation and scattering processes of exciplex states, and the fluorescent emission process.

Figure 6

(a),(b) Temperature-dependent MC curves of Dev. 1 and Dev. 2 at a bias current of 500 µA, respectively. All MC curves are offset for clear observation; white solid lines are fitted curves. (c),(d) Temperature dependences of intensity factor C₃ for Dev. 1 and Dev. 2, respectively. (e),(f) Temperature dependences of intensity factors (C₁, C₂, and the sum of C₁+ C₂) for Dev. 1 and Dev. 2, respectively.

Figure 7

(a),(b) Current-dependent MEL curves of Dev. 3 and Dev. 4 measured at room temperature within the field ranges of |B| < 150 mT, respectively. (c),(d) Current-dependent MC curves of Dev. 3 and Dev. 4 acquired at room temperature within the field ranges of |B| < 150 mT, respectively. MEL and MC curves of Dev. 3 and Dev. 4 are offset to be clearly observed, and their corresponding raw data without shift are shown in Appendix pp12. (e),(f) Schematic diagrams of rubrene donor (yellow hexagons) and ${\mathrm{C}}_{60}$ acceptor (gray circles) molecules at the rubrene/${\mathrm{C}}_{60}$ interface without and with a BCP (blue trapezoid) interlayer, and the relationship between the relative energy-level difference of EX states (EX¹ and EX³) and the intensity of the RISC process.

Figure 8

Fingerprint MEL or MC curves of B-mediated ISC, RISC, TTA, SF, and TCASC processes. For B-mediated ISC and RISC processes, as shown by black and red lines, respectively, their respective MEL and MC traces have inverted and upright Lorentzian line shapes with featured fields of several mT, which is in the order of the hyperfine field. For the MC fingerprint curve determined by the TCASC process, as demonstrated by the olive line, its line shape is non-Lorentzian and has a characteristic field of about 125 mT. For TTA- or SF-induced MEL fingerprint curves, they have the same line shapes but with opposite signs and their low- and high-field characteristic fields are found to be 20 and 105 mT (or 25 and 110 mT), respectively.

Figure 9

Molecular structures of PEDOT:PSS, HAT-CN, ${\mathrm{C}}_{60}$, and BCP.

Figure 10

(a),(b) Current-dependent MC curves of Dev. 1 and Dev. 2 operated at room temperature within the field ranges of |B| < 150 mT, respectively.

Figure 11

(a),(b) Current-dependent MC and MEL curves, respectively, of the reference device at room temperature within the field ranges of |B| < 150 mT. (c) Schematic diagram of microscopic mechanisms of the reference device.

Figure 12

(a) Temperature-dependent MC curves of Dev.1 at a bias current of 50 µA. These MC curves are offset for clear observation. (b) Temperature dependences of the intensity factors (C₁, C₂, and the sum of C₁+ C₂) for Dev. 1. (c) Temperature dependences of the intensity factor (C₃) for Dev. 1.

Figure 13

(a),(b) Luminance-voltage characteristic curves of Dev. 3 and Dev. 4, respectively.

Figure 14

(a),(b) Schematic diagrams of energy-level alignments and the formation of exciplex states in Dev. 3 and Dev. 4, respectively.

Figure 15

Current-dependent MEL curves of Dev. 3 and Dev. 4 at room temperature within the field ranges of |B| < 150 mT, respectively. (c),(d) Current-dependent MC curves of Dev. 3 and Dev. 4 at room temperature within the field ranges of |B| < 150 mT, respectively.