Viologens are organic bipyridinium compounds that are widely studied and utilized for their electron-donating,24–26 electrochromic,27 and catalytic,28 properties; in addition, their high solubility in common solvents render them compatible with cost-efficient printing and coating fabrication.24, 29 One member of the bipyridinium class is the neutral reduced benzyl viologen (r-BV0) molecule, which comprises a central 4,4′−bipyridine core surrounded by two benzyl groups (see Figs. 1a,b). r-BV0 is a notably strong electron donor, i.e. chemical reductant,26 which has been employed for the n-type doping of a wide variety of semiconductors, including conjugated polymers, carbon nanostructures, graphene, MoS2 and PbS.25, 26, 30–35
We herein synthesized r-BV0 by reducing a benzyl viologen dichloride salt (BV2+Cl-2) with the aid of a large surplus of a NaBH4 salt as the reducing agent31–35 in a two-component water:toluene solution, as schematically depicted in Fig. 1(a). The exothermic reaction proceeds in two reduction steps, with the initial step featuring an electron transfer from NaBH4 to the nitrogen on one of the two pyridine rings of the BV2+ di-cation for the formation of a r-BV+ mono-cation; while the final step constitutes an electron transfer from NaBH4 to the nitrogen on the second of the pyridine rings for the formation of the desired neutral r-BV0 end product. The chemical structures of these three different oxidation states of the benzyl viologen (BV) molecule, and their associated colors, are displayed in Fig. 1(b).
The progression of this chemical reaction can be observed visually, since the two reactants (BV2+Cl− 2, and NaBH4) are colorless, while the r-BV+ intermediate is colored violet, and the final r-BV0 product is pale yellow.25, 31, 34–41 The conclusion of the reaction can accordingly be determined by the reaction solution shifting color from violet to yellow, in combination with the disappearance of the H2 gas bubbles from the reaction solution (see Fig. 1a). The completion of this double-reduction reduction was further confirmed by the recording of a silent EPR spectrum (data not shown). The single-reduction product r-BV+ would in contrast have exhibited a characteristic EPR spectrum.42 The employment of the poorly miscible two-component water:toluene reaction solution enabled for the practical collection of the neutral r-BV0 product from the upper (and lighter) toluene phase, since both reactants and all other non-gaseous products are solely soluble in the lower water phase. More details on the synthesis are available in the Methods section.
The extracted “doping solution” (i.e., r-BV0 dissolved in toluene) exhibited a r-BV0 solute concentration of ~ 4 g·l− 1. This solution was used as the chemical reductant in the formulation of the ”active-material inks”, which also comprised the electroluminescent and polymeric organic semiconductor termed Super Yellow, the ion transporter hydroxyl-capped trimethylolpropane ethoxylate (TMPE-OH), and the salt KCF3SO3 in a constant mass ratio of 1:0.1:0.03. The r-BV0 concentration was varied in the different inks, as quantified by the mass fraction of r-BV0 with respect to Super Yellow.
Figure 1(c) is an electron-energy diagram, which presents values for the HOMO of r-BV0 to the left and the LUMO, HOMO and an electron-trap level of Super Yellow43–45 to the right. The HOMO value for r-BV0 was derived from a cyclic voltammetry (CV) measurement vs. the standard hydrogen electrode (SHE), and by using the convention that the SHE electrode is offset by 4.44 eV from the vacuum level.26, 46, 47 The HOMO and LUMO of Super Yellow were similarly determined from CV measurements,48 while a major and generic electron trap level in organic semiconductors was identified and rationalized by Blom et al.43, 45 A recent LEC study further suggest that the inclusion of the TMPE-OH ion transporter into the active material of our LEC devices can result in the formation of a significant concentration of additional electron traps,21 but their exact energy level is currently unknown. Nevertheless, the presented electron-energy diagram implies that an electron transfer from the HOMO of r-BV0 to the established electron-trap level of Super Yellow should be possible, as indicated by the arrow in Fig. 1(c).
The photoluminescence quantum yield (PLQY) is a sensitive indicator of the existence of “trap impurities” in organic semiconductors, since the singlet exciton typically diffuses a significant distance of ~ 10 nm during its lifetime in neat organic semiconductors,49 and since the exciton commonly is efficiently quenched by such trap impurities.50 51 Fig. 2(a) presents the evolution of the PLQY of the active-material film as a function of the added r-BV0 concentration. Interestingly, we find that the PLQY first increases up to ~ 2 mass% of r-BV0, and thereafter decreases monotonously with further increasing r-BV0 concentration. The initial increase of the PLQY implies that the first added r-BV0 molecules effectively eliminates “dark” electron traps, presumably by donating electrons and thereby filling of electron trap levels, as schematically depicted in Fig. 1(c).
The influence on the electronic conductivity of Super Yellow by the addition of r-BV0 was investigated by the fabrication and characterization of glass/ITO/Al/(Super Yellow + r-BV0)/Ca/Al electron-only devices. Figure 2(b) shows that the current density increases by more than one order of magnitude over the entire probed voltage interval when the r-BV0 concentration is increased from 0 to 3 mass%; but that trend is reversed at higher r-BV0 concentrations exceeding 3 mass%, for which the electronic conductivity is reverting back to the lower level of intrinsic Super Yellow.
Accordingly, the summary conclusion from the data presented in Fig. 1(c) and Fig. 2 is that the first addition of r-BV0 molecules (up to 2–3 mass%) results in a filling of electron traps of Super Yellow, which is manifested in an increased electron mobility (and thereby an improved electronic conductivity)52 and an improved PLQY. A higher concentration of added r-BV0 molecules results in transport- and emission-damaging effects that obviously should be avoided in devices. We now turn our attention to the investigation of the effects of this dopant-enabled electron-trap filling approach on the performance of LEC devices.
Figure 3(a) presents the simulated steady-state exciton profile in the interelectrode gap for ITO/active-material/Al LECs, which are solely distinguished by the non-filled electron-trap concentration on the organic semiconductor in the active material, as identified in the inset. The trap-free electron/hole mobility ratio was set to 3 in the simulation, and the anodic (cathodic) interface is positioned at a normalized interelectrode position of 0 (1). The peak exciton concentration can be considered to represent the position of the emissive p-n junction.
The simulation shows that the position of the p-n junction shifts from the center of the active material towards the positive anode with decreasing concentration of electron traps, i.e., with increasing concentration of chemical-reductant molecules, as highlighted by the arrow. This behavior is explained by that the steady-state position of the p-n junction in LEC devices is determined by the effective electron/hole mobility ratio, with a higher electron (hole) mobility resulting in a p-n junction positioned closer to the positive anode (negative cathode). The filling of electron traps will increase the effective electron mobility, and it is accordingly expected to result in the observed shift of the p-n junction towards the positive anode.
We have experimentally investigated the effects of the addition of the r-BV0 chemical reductant on the p-n junction position by measuring the angle-dependent EL spectrum and intensity for ITO/(Super Yellow + TMPE-OH + KCF3SO3 + r-BV0)/Al LECs and thereafter simulating the same data. The sole free parameter in the simulation was the position of the emissive p-n junction, and by systematically determining the value for this parameter that produced the best agreement between the measured and simulated data, we could establish the position of emissive p-n junction with high accuracy;21, 53, 54 see Methods section for more details.
Figure 3(b) presents the experimentally determined steady-state position of the p-n junction in the interelectrode gap as a function of the r-BV0 concentration. The LEC device void of r-BV0 exhibits a steady-state p-n junction in the exact middle of the active material at 0.50, while the addition of the r-BV0 chemical reductant to the active material results in a gradual and significant anodic shift of the p-n junction position to 0.43 at [r-BV0] = 1 mass% and 0.39 at [r-BV0] = 3 mass%. This experimental finding is thus in excellent agreement with the simulation results displayed in Fig. 3(a). At a higher r-BV0 concentration ([r-BV0] ≥ 5 mass%) the anodic shift is halted, which is in agreement with our earlier finding that these higher r-BV0 concentrations cause transport (and emission) limiting effects.
We have finally investigated the effects of this chemical pre-doping approach, and its induced p-n junction shift, on the performance of LEC devices. Figures 3(c) and 3(d) display the measured voltage and luminance transients, respectively, of champion ITO/(Super Yellow + TMPE-OH + KCF3SO3 + r-BV0)/Al LEC devices during electrical driving by a constant current density of 7.75 mA·cm− 2. The r-BV0 concentration is identified in the inset of Fig. 3(c). We have measured eight independent devices for each r-BV0 concentration, and Figure S1 and Table S1 present the average and standard deviation for key device metrics.
All investigated LEC devices feature a decreasing voltage and an increasing luminance during the initial operation, and a fast turn of < 2 s to a luminance exceeding 100 cd·m− 2. This implies that the LEC-characteristic in-situ electrochemical doping capacity of the organic semiconductor Super Yellow is not significantly damaged or affected by the addition of rBV0.55 We find that the LEC devices with a r-BV0 concentration of ≤ 3 mass% exhibit a markedly better performance than the LECs with a higher r-BV0 concentration (see also Figure S1 and Table S1). This supports our earlier finding that a too high r-BV0 concentration causes transport and emission limiting effects within the active material.
Importantly, we find that the LEC with 1 mass% r-BV0 exhibits ~ 10% higher peak luminance and ~ 70% longer operational lifetime than the reference LEC, which is void of r-BV0. (The operational lifetime is here defined as the total time that the device emits with a luminance exceeding 100 cd·m− 2.) We assign these improvements to the combined effects of a slightly increased PLQY (Fig. 2a), and the associated suppression of weakly emissive trap-assisted electron and hole recombination,45 in combination with the anodic shift of the emissive p-n junction following the addition of 1 mass% r-BV0 to the active material (Fig. 3b). An anodic shift of the exciton population can for this particular device be attractive from an emission-efficiency viewpoint since the ITO anode is a less potent exciton quencher than the Al cathode.56, 57 Accordingly, a shift of the emissive p-n junction from the center of the active material towards the positive ITO anode can be expected to result in lowered losses due to exciton-electrode quenching. We finally note that exciton quenching can result in severe self heating58, 59 and/or the formation of highly localized high-energy species on the organic semiconductor, which in turn can cause material and device degradation. Thus, the suppression of exciton quenching reactions by the electron trap filling of the organic semiconductor by the chemical reductant can also rationalize the observed prolongation of the device lifetime.