Effective Solution‐ and Vacuum‐Processed n‐Doping by Dimers of Benzimidazoline Radicals

Abstract

Benzimidazoline radical dimers that can be handled in air but that function as powerful reductants are reported and evaluated as n-dopants by solution- and vacuum processing. In several host materials, one of these dimers is found to have a more-consistent doping effect than a hydride-donor dopant analog. Notably, a record high room-temperature conductivity of 12.0 S cm−1 is obtained for doped C60. The p- and n-doping of organic semiconductors by chemical agents to modulate the carrier density is important due to the use of doped layers and interfaces in devices such as organic solar cells, transistors, light-emitting diodes (LEDs), Zener diodes, and thermoelectrics.1-4 Early work on the n-doping of organic semiconductors mainly employed the alkali metals, but their highly reactive nature, as well as the high diffusivity of the corresponding ions, has limited their implementation.5-8 More recently a diverse array of molecular n-dopants have been reported, but there are few examples of air-stable n-dopants that may be both solution- or vacuum-processed and that are capable of doping host materials with low electron affinities (EAs).3 Molecular n-dopants achieve doping either by reduction of the host material by direct electron transfer from the dopant or by decomposition of a stable dopant precursor to an intermediate that is capable of reducing the host.9, 10 Air-stable dopant precursors based upon cationic and hydride-reduced π-conjugated small molecules were first studied as dopants by Leo.11, 12 Two examples of these types of dopants that have been studied recently are the cationic benzimidazolium iodide salts (DMBI-I, Scheme 1), which release unidentified reactive intermediates during a high-temperature thermal deposition process, and the 2,3-dihydro-1H-benzimidazoles (DMBI-H, Scheme 1) which react with PC61BM by an initial hydride-transfer step.13, 14 In either case, the benzimidazolium cation/host radical anion pair, DMBI+/A•−, is believed to be the doping product that is responsible for the additional free carriers (Scheme 1). However, for DMBI-I based salt dopants, strong n-doping has only been observed for vacuum deposited films, while for DMBI-H dopants, doping is necessarily accompanied by hydride or hydrogen atom transfer. Stable dopant precursors have also been obtained by isolating dimers of highly reducing neutral radicals; this approach has been applied to the dimers of various 19-electron sandwich compounds, which react in either solution or by vacuum deposition with minimal side products formed,15, 16 while dopants based on dimers of organic radicals, including those obtained by reduction of imidazolium and pyridinium species, have been described in the patent literature.17, 18 Here we report the neutral benzimidazoline-radical dimers (2-Cyc-DMBI)2, (2-Rc-DMBI)2 and (2-Fc-DMBI)2 (Scheme 1) and their use to form the doped state DMBI+/A•− in a similar way. The distinct properties of the (DMBI)2 dopants relative to the DMBI-H dopants is evident from the solution doping rates, doped thin-film polaron band intensities, Fermi level shifts, and conductivities achieved using (2-Cyc-DMBI)2 and its DMBI-H analog, 2-Cyc-DMBI-H, as dopants (Scheme 1). Greater variability of the doping effect in the different hosts was found for the DMBI-H compound than for the (DMBI)2 compound, which is attributed to the distinct reactions by which the two classes of dopants function. Notably, a very high room-temperature conductivity of 12.0 S cm−1 was obtained for C60 doped with (2-Cyc-DMBI)2. DMBI dimers with 2-alkyl and 2-metallocenyl substituents were synthesized as shown in Scheme 1 (see the Supporting Information for details and characterization data), but several 2-aryl analogues were less stable and could not be isolated. The three (DMBI)­2 compounds are reasonably stable in air as solids relative to other highly reducing compounds such as decamethylcobaltocene or W2(hpp)4. For instance, elemental analysis of (2-Cyc-DMBI)2 after one week of storage in air provided no evidence for decomposition, but 1H-NMR after 3 months in ambient indicated ∼5 mol% conversion to a decomposition product. Notably, deoxygenated solutions of (2-Cyc-DMBI)2 in heptane show no decomposition, but all of the dimers decompose in non-deoxygenated solvents (Figure S5). The moderate air-stability of the (DMBI)2 solids can be beneficial for weighing and handling compounds in air; however, it is important to recognize that the air sensitivity of an n-doped thin film is more dependent on the host material EA than the dopant properties. The dimers were first evaluated in solution-phase doping reactions with PC61BM (Figure 1). The growth of the PC61BM radical-anion absorption (1030 nm) was monitored by UV-Vis-NIR spectroscopy (Figure S6).16 All three dimer dopants, as well as 2-Cyc-DMBI-H, were effective reducing agents for PC61BM, and quantitative conversion to DMBI+/PCBM•− was observed in minutes for the dimers. The rate of formation of PC61BM•− is much more rapid for (2-Cyc-DMBI)2 than for the hydride analogue, 2-Cyc-DMBI-H (Figure 1). Monitoring the change of the 1030 nm absorbance over time for conditions with (2-Cyc-DMBI)2 in excess and with PC61BM in excess suggests a reaction that is first order in both fullerene and dimer, consistent with the rate-determining step being an electron transfer from (2-Cyc-DMBI)2 to PC61BM rather than dimer C-C homolysis (Figure S7).16 Additionally, all of the dimers were capable of reducing 6,13-bis(triisopropylsilylethynyl)-pentacene (TIPS-pentacene, EA ∼3 eV) to its radical anion in solution, whereas no DMBI-H compounds were capable of forming a detectable concentration of the radical anion of this comparatively weak electron acceptor.19 Photothermal deflection spectroscopy (PDS) was used to compare the intensities of the broad polaron absorptions, which are proportional to carrier densities in doped organic semiconductors, of doped P(NDI2OD-T2) and PC61BM thin films formed by drop-casting (Figure 1).20 Compared to traditional UV-VIS spectroscopy, PDS has a superior sensitivity, and is well suited for the detection of very weak sub-gap absorption features in thin films.21 All combinations of dopants and hosts studied produced sub-gap features consistent with the polaron absorptions of a doped state. In P(NDI2OD-T2), at a fixed doping concentration of 9.8 mol%, the absorption intensity at 1 eV was comparable for all three dimer dopants suggesting similar extents of electron transfer to the polymer, while the absorption intensity imparted by the hydride dopant, 2-Cyc-DMBI-H, was over an order of magnitude lower. In contrast, the PDS spectra of 9.8 mol% doped PC61BM revealed similar polaron absorption intensities across the spectrum when either 2-Cyc-DMBI-H or (2-Cyc-DMBI)2 was the dopant, and in this case 2-Cyc-DMBI-H actually exhibited a slightly greater polaron absorption intensity than (2-Cyc-DMBI)2. These results demonstrate that the hydride dopant 2-Cyc-DMBI-H is less efficient than (2-Cyc-DMBI)2 at generating absorbing polarons in the conjugated polymer P(NDI2OD-T2), but that the two dopants generate absorbing polarons in PC61BM with similar efficiencies, despite the comparable EAs of these two hosts.16, 22 The doped P(NDI2OD-T2) films were studied by ultraviolet photoelectron spectroscopy (UPS) (Figure 1). All three dimer dopants led to large shifts of the Fermi level EF to ca. 1.7 eV away from the onset of ionization from filled states (EHOMO) at a doping ratio of 2.4 mol%, whereas, at all molar doping ratios studied, 2-Cyc-DMBI-H induced a significantly smaller shift of EF than (2-Cyc-DMBI)2. The saturation of the shift (pinning) of EF with increasing doping levels has been observed previously for both p- and n-doped organic semiconductors, and it is thought to reflect the approach of EF to the tail of the Gaussian density of states near the band edges.23 Indeed, the transport gap for this polymer determined by UPS and inverse photoelectron spectroscopy is ca. 1.6 eV, indicating that the pinning in the present case occurs at or very near to the onset of empty LUMO-derived states.15 The UPS and PDS data suggest that the dimers efficiently dope both PC61BM and P(NDI2OD-T2), whereas 2-Cyc-DMBI-H is a much more efficient dopant for the fullerene than the polymer, despite similarity of the EAs and reduction potentials of PC61BM and P(NDI2OD-T2) (ca. 3.9 ± 0.1 eV and ca. −1.0 V vs. FeCp2+/0).10, 15, 24 These results are consistent with the different chemistry associated with doping by DMBI dimers and DMBI-H derivatives. For the dimer dopants the thermodynamic feasibility of doping (to give two moles of A•−) is given by ΔGdoping = ΔGC-C + 2F(E0/–A – E+/0D)16 where ΔGCC is the free energy for homolysis of the dimer central C-C bond and E0/−A and E+/0D are the reduction potentials of the acceptor and dopant cation, respectively. The driving force for electron transfer, E0/–A – E+/0D, from a 2-aryl/alkyl-benzimidazolium radical to any arbitrary acceptor with an EA of −3.7 eV can be estimated at > 1 eV which suggests the doping reaction is highly exergonic for any reasonable estimate of ΔGC-C. However, for DMBI-H the free energy for doping (to give one mole of A•−) is given by ΔGdoping = ΔΔGC-H + F(E0/–A – E+/0D) where ΔΔGCH is the difference in the free energies of cleavage of the donor-H bond and the acceptor–H bond in the H-reduced acceptor; thus, the ability to dope an acceptor depends on both the EA and the hydrogenation thermodynamics. Evidently, the hydrogen-accepting properties of P(NDI2OD-T2) are poorer than those of PC61BM. The UV-Vis-NIR, PDS, and UPS data suggest that all three (DMBI)2 dimers give comparable yields of acceptor radical anions in both P(NDI2OD-T2) and PC61BM. However, for many practical applications, it is the enhancement of conductivity that is of relevance, which depends on both the number of carriers introduced in addition to the impact of the dopants on the effective carrier mobility. Accordingly, conductivity changes induced by the dimers and by 2-Cyc-DMBI-H when doped into spin-coated P(NDI2OD-T2) and PC61BM, and of the dimers for vacuum-deposited C60, were measured using a four-point probe technique. For all material combinations, the doped films exhibited greater conductivity than the intrinsic host. Similar maximum conductivities on the order of 10−2 S cm−1 were observed for both (2-Cyc-DMBI)2 and 2-Cyc-DMBI-H in PC61 BM,25 whereas in P(NDI2OD-T2) over an order of magnitude higher conductivity was obtained for the dimer dopant (2.8 × 10−3 S cm−1 at 5.0 wt%/11 mol% vs. 3.4 × 10−4 S cm−1 at 20 wt%/53 mol%), and the maximum conductivity was observed at a much lower doping concentration for the dimer (Table 1), consistent with the PDS and UPS data.26-29 The DMBI dimers with organometallic substituents were also evaluated as dopants for both P(NDI2OD-T2) and PC61BM. The Rc-substituted dimer behaved in a similar manner to its Cyc analogue, giving a maximum conductivity of 3.0 × 10−3 S cm−1 at 10 wt% (12 mol%) in P(NDI2OD-T2) and 1.6 × 10−2 S cm−1 at 10 wt% (11 mol%) in PCBM (Table 1). However, (2-Fc-DMBI)2 gave lower maximum conductivities than either of the other two dimers or the hydride dopant, reaching maxima of only 7.6 × 10−5 S cm−1 at 20 wt% (26 mol%) and 1.9 × 10−3 S cm−1 at 5 wt% (6 mol%) in P(NDI2OD-T2) and PC61BM respectively. Given the similar levels of charge-carrier generation for the different dimers suggested by both the PDS and UPS measurements, we examined film morphology as a possible explanation for the variation in the conductivities of the dimer doped films. Optical microscopy images of the PC61BM doped films reveal a clear trend in morphology. Particle formation on the smooth PC61BM surface is observed upon addition of any of the dopants studied. The particles range in size from only a few nanometers in diameter for 2-Cyc-DMBI-H doping to micron-sized particles present on all of the dimer-doped films at concentrations above 3 wt% (Figure S10). The aggregation is most extensive for the Fc- and Rc- substituted dopants at all concentrations (Figure S10). Previously, both DMBI+ and PC61BM ions have been identified in the precipitate isolated from DMBI-H-doped PC61BM solutions, and the precipitate has been identified as a DMBI+/A•− salt that is sparingly soluble in organic solvents.10 In the present study, a comparison of the 2-Cyc-DMBI-H- and (2-Cyc-DMBI)2-doped PC61BM film morphology reveals much larger particles for the dimer-doped films suggesting a greater extent of the doping reaction during film formation which is consistent with the faster doping rate observed for the dimer relative to the hydride dopant. Particles are also observed by AFM on the polymer surface of P(NDI2OD-T2) when doped, and again (2-Fc-DMBI)2 and (2-Rc-DMBI)2 generate the largest aggregates. The low conductivities obtained for (2-Fc-DMBI)2 doped PC61BM and P(NDI2OD-T2) relative to the (2-Cyc-DMBI)2 doped films, despite the comparable doping efficiency that might be anticipated from the UPS and PDS data, is at least partly explained by the rough and heavily aggregated morphology that is observed when using the (2-Fc-DMBI)2 dopant, which may be influenced by several factors including: the fast reaction rate for the dimers, a low solubility for the product DMBI+/A•− salt, and a low miscibility of the product DMBI+/A•− salt in the host material. A thorough mechanistic study is underway to further explore the possible mechanisms and reaction rates of the dimers. The dimer dopants can also be readily evaporated under high vacuum. To evaluate the dimers as vacuum processed n-dopants, doped C60 films were deposited by co-deposition under high vacuum (10−6 to 10−7 torr). The thicknesses and the doping concentrations of the films were confirmed by AFM and XPS respectively. All three dimer dopants were competent dopants for C60 (Figure S9). The maximum average conductivity of 12 S cm−1 was obtained at 18 wt% (26 mol%) of (2-Cyc-DMBI)2 (Table 1); to the best of our knowledge, this is the highest room-temperature conductivity that has been reported for a molecularly n-doped C60 film. The maximum conductivity was several orders of magnitude lower for the other two dimers (Table 1). AFM height profiles of the 20 wt% doped C60 films are shown in Figure S11; the root-mean-square roughnesses, Rrms, of the (2-Cyc-DMBI)2-, (2-Rc-DMBI)2-, and (2-Fc-DMBI)2-doped films are 0.21, 0.87, and 1.09 nm respectively indicating a similar morphological trend in the dimer-doped C60 as seen for the solution-processed films. In summary, neutral (DMBI)2 dimers are effective solution- and vacuum-processable molecular n-dopants that are ambient stable as solids. Compared with previously reported DMBI-H and DMBI-I dopants, these dimers exhibited a stronger doping effect in a more diverse array of materials. The variation in the film conductivities, especially for the case of PC61BM and C60 doping, is attributed to the tendency of the Fc- and Rc- substituted dimers to form rougher films, presumably with more tortuous percolation pathways. A comparison of the 2-cyclohexyl substituted dimer and hydride dopants found comparable film conductivities when doped in the C-H acceptor host, PC61BM, but much greater conductivities for the dimer when doped into a weaker C-H bond acceptor such as P(NDI2OD-T2). This result is consistent with the thermodynamics for doping by DMBI-H and (DMBI)2 compounds. For C60 doping, (2-Cyc-DMBI)2, yielded the most conductive molecularly n-doped C60 film reported to date with a maximum average room-temperature conductivity of 12 S cm−1. Notably, n-doped C60 layers have been the topic of several studies involving both thermoelectrics and p-i-n organic solar cells.30, 31 Future work will investigate the detailed mechanism of the n-doping process for the (DMBI)2 dimers with different acceptors. B. D. N. gratefully acknowledges the National Defense Science and Engineering Graduate Research Fellowship, and the National Science Foundation for a Graduate Research Fellowship. This research was also funded by the National Science Foundation (Materials Network Program, DMR-1209468; MRSEC Program, DMR-0820382; and DMR-1305247) and Air Force Office of Scientific Research (FA9550–12–1–01906). We also thank Tissa Sajoto for synthesis of P(NDI2OD-T2). Note: Minor corrections to Scheme 1, Figure 1, and the Supporting Information were implemented after initial online publication. As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.