Synthesis of T2Cl2 (1)

Under argon, a mixture of 3,3′,5,5′-tetrachloro-2,2′-bithiophene (3.04 g, 10.0 mmol), zinc powder (1.95 g, 30.0 mmol), isopropylalcohol (50 mL), and acetic acid (15 mL) was refluxed for 24 h. After thin-layer chromatography test indicated the incompletion of reaction, more zinc powder (1.44 g, 22.2 mmol) and acetic acid (12 mL) was added and the reaction mixture was stirred under nitrogen for an additional 38 h. Upon cooling to room temperature, the reaction mixture was precipitated in methanol, and the crude product was subject to column chromatography on silica gel to afford a white solid as the product (0.27 g, 11.5%). ¹H NMR (CD₂Cl₂, 400 MHz): δ 7.43–7.44 (d, J = 5.2 Hz, 2H), 7.04–7.05 (d, J = 5.2 Hz, 2H).

Synthesis of 5,5′-bis(trimethylstannyl)-T2Cl2 (2)

Under argon, a solution of T2Cl2 (0.26 g, 1.1 mmol) in anhydrous tetrahydrofuran (THF) (15 mL) was cooled to −78 °C, and a solution of n-BuLi in hexane (0.90 mL, 2.24 mmol) was then added dropwise. The resulting solution was stirred at this temperature for about 30 min and at room temperature for an additional 1 h. The reaction mixture was then cooled to −78 °C again, and a solution of trimethyltin chloride (4.5 mL, 1 m in hexane, 4.5 mmol) was added slowly. This mixture was allowed to warm to room temperature and stirred at room temperature overnight. After the reaction was diluted with hexane (~100 mL), it was washed with water (3 × 30 mL), dried over anhydrous sodium sulfate, and concentrated in vacuo. The residue was recrystallized from isopropyl alcohol, leading to a pale solid as the product (0.39 g, 63.2%). ¹H NMR (CDCl3, 400 MHz): δ 7.03 (s, 2H), 0.40 (m, 18H).

Synthesis of P(NDI2HD–T2Cl2)

Under argon, a mixture of NDI2HD-Br₂ (669.1 mg, 0.77 mmol), 5,5′-bis(trimethylstannyl)-T2Cl2 (2) (430.5 mg, 0.77 mmol), Pd₂dba₃ (14.1 mg, 0.015 mmol), and P(o-tol)₃ (37.5 mg, 0.12 mmol) in anhydrous chlorobenzene (60 mL) was stirred at 90 °C for 18 h. Bromobenzene (0.5 mL) was then added, and the resulting mixture was stirred for an additional 6 h. After cooling to room temperature, the reaction mixture was poured into methanol (200 mL). The resulting suspension was stirred for 20 min, before the precipitate was collected by filtration and washed with methanol. This crude product was then subjected to Soxhlet extraction under nitrogen with methanol, acetone, and hexane. Finally, it was extracted with chlorobenzene (about 200 mL). Upon cooling to room temperature, the chlorobenzene extract was precipitated into methanol (400 mL). The precipitate was collected by filtration, washed with methanol, and dried under vacuum, leading to a dark solid (675.0 mg, 93.1%). ¹H NMR (CDCl₂CDCl₂, 500 MHz, 120 °C): δ 8.91 (s, 2H), 7.40 (s, 2H) , 4.22 (s, br, 4H), 2.12 (s, br, 2H), 1.25–1.56 (m, br, 48H), 0.89–0.97 (m, br, 12H).GPC: M _n = 20.4 kDa, PDI = 2.2. Elemental Analysis (calc. C, 68.55; H, 7.46; N, 2.96): found C, 68.44; H, 7.41; N, 3.01.

Synthesis of P(NDI2OD–T2Cl2)

Under argon, a mixture of NDI2OD-Br₂ (655.3 mg, 0.67 mmol), 5,5′-bis(trimethylstannyl)-T2Cl2 (2) (373.0 mg, 0.67 mmol), Pd₂dba₃ (12.2 mg, 0.013 mmol), and P(o-tol)₃ (32.4 mg, 0.11 mmol) in anhydrous chlorobenzene (60 mL) was stirred at 90 °C for 20 h. Bromobenzene (0.5 mL) was then added, and the resulting mixture was stirred for an additional 5 h. After cooling to room temperature, the reaction mixture was poured into methanol (150 mL). The resulting suspension was stirred for 20 min, before the precipitate was collected by filtration and washed with methanol. This crude product was then subject to Soxhlet extraction under nitrogen with methanol, acetone, and hexane. Finally, it was extracted with chlorobenzene (about 350 mL). Upon cooling to room temperature, the chlorobenzene extract was precipitated into methanol (500 mL). The precipitate was collected by filtration, washed with methanol, and dried under vacuum, leading to a dark solid (687.2 mg, 96.9%).¹H NMR (CDCl₂CDCl₂, 500 MHz, 130 °C): δ 8.92 (s, 2H), 7.41 (s, 2H) , 4.23 (s, br, 4H), 2.13 (s, br, 2H), 1.26–1.58 (m, br, 64H), 0.88–0.99 (m, br, 12H). GPC: M _n = 17.0 kDa, PDI = 1.8. Elemental Analysis (calc. C, 70.36; H, 8.19; N, 2.65): found C, 70.59; H, 8.28; N, 2.81.

TFT device fabrication and measurements

OTFTs with TG/BC architecture were fabricated on Corning Eagle 2000 glass substrates. The source–drain (S/D) electrodes were thermally evaporated Au contacts (40 nm thick) with L = 50 µm and W = 1 mm (for the devices of Table II) or photolithography-defined Ni (3 nm)/Au (13 nm) with L = 2–60 µm and W = 1 mm (for contact resistance measurements and devices presented in Table III). The defined substrates were cleaned in an ultrasonic bath with acetone, isopropanol, and de-ionized water, respectively (10 min each), and then post-baked at 110 °C for 30 min. N-channel polymer semiconductor formulations of P(NDI2HD–T2Cl2) and P(NDI2OD–T2Cl2) were prepared by dissolving the corresponding polymers in anhydrous dichlorobenzene at 1, 5, and 10 mg/mL concentrations. The non-chlorinated parent polymers, P(NDI2HD–T2) and P(NDI2OD–T2) (Polyera ActivInk^™ N2200) were dissolved in p-xylene to obtain 10 mg/mL reference semiconductor solutions. The semiconductor solutions were filtered through 0.2 µm polytetrafluoroethylene syringe filter, and spin-coated at 2000 rpm for 60 s in an N₂-filled glove box. The spin-coated P(NDI2HD–T2Cl2) and P(NDI2OD–T2Cl2) films were thermally annealed at 150 °C for 30 min and 110 °C for 20 min, respectively, in a glove box with a N₂ atmosphere to remove residual solvents. Poly(methyl methacrylate) (PMMA) (Aldrich, M _W = 120 kDa) or a CYTOP solution (Asahi) ware used as the dielectric material. PMMA was dissolved in n-butylacetate to obtain 80 mg/mL dielectric solution, and it was spin-coated at 2000–3000 rpm for 60 s on the semiconducting polymer film. Afterwards, the devices were baked at 80 °C for 2 h to remove residual solvent under N₂. The OTFTs were completed by depositing Al TG electrodes via thermal evaporation with the aid of metal shadow mask. The fundamental OTFT electrical characteristics were measured using a Keithley 4200 semiconductor characterization system in an N₂-filled glove box. The conventional saturation mobility (μ _sat) and corresponding V _Th were extracted at the saturation region (at a drain voltage V _DS of +80 V) using gradual channel approximation equation for a standard metal oxide semiconductor TFT (MOSFET): μ _sat = (2I _DS L)/[WC_i(V _G−V _Th)²], where I _DS is the S/D saturation current, C _i is the gate dielectric capacitance (per area), and V _G is the gate voltage.

Computational modeling

Chlorine incorporation into the parent polymers P(NDI2HD–T2) and P(NDI2OD–T2) was motivated by the aforementioned electronic and solubility advantages. However, the regiochemistry of the substitution must take into account the synthetic strategy and the far larger Van der Waals radius of Cl (1.75 Å) versus H (1.20 Å) and F (1.47 Å) to avoid sterically congested π-backbone distortions between NDI acceptor and bithiophene donor units. Thus, geometry optimizations and molecular orbital (MO) computations on several polymer segments (model structures I–V) incorporating chlorine atoms at different locations were performed at the DFT level to guide the polymer design. Figure 3 and S1 collect the computed structures, MO topologies and energies, and relevant torsional angles (θ) for the model units I–V. The computed MO energies (E _H = 5.43 eV and E _L = −3.31) and θ values (θ _NT = 46° and θ _TT = 19°) for the unsubstituted system (structure I) are in agreement with previous computations on similar backbones. Note, it was found experimentally that NDI-based polymers such as P(NDI2OD–T2) exhibit a torsional angle of ~35° between the NDI and the T2 units, which does not prevent efficient electron transport.^[ Reference Zhan, Facchetti, Barlow, Marks, Ratner, Wasielewski and Marder ^{68 ,} Reference Jones, Ahrens, Yoon, Facchetti, Marks and Wasielewski ^{69 ]} Our geometry optimizations indicate that chlorine functionalization at both NDI 3,7- (structure II) and thiophene 4-positions (structure III) considerably increases the NDI-T interring torsional angles (θ _NT) to 68° and 64°, respectively. In all cases, the interring torsional angles between the two adjacent thiophene units (θ _TT) are similar (18–23°). However, when the chlorine atoms are introduced at the 3-positions of the thiophenes on both sides of NDI (structure IV) or at the 3,3′-positions of the bithiophene moiety (structure V), θ _NT values are identical to those of the unsubstituted systems (45–46°), which is critical to maintain a good donor–acceptor (D–A) electronic communication in the polymer backbone. The thiophene–thiophene torsional angle, θ _TT, is unaffected for the structure IV having one chlorine at either side (22°), whereas it increases to ~63° in the case of the head-to-head 3,3′-dichloro substituted structure V. It is noteworthy that the addition of chlorine substituents into the polymer π-backbone may induce intrachain (thienyl)Cl···S(thienyl) and interchain (thienyl)Cl···O(imide), non-covalent interactions,^[ Reference Ji, Xia, Ren, Wang, Wang and Diao ^{70 ,} Reference Adhikari and Scheiner ^{71 ]} which could promote rigidification and solid state stacking. Chlorine incorporation also affects the frontier MO energies of the structures I–V. Considering the MO topologies (Fig. S1) with the LUMO localized on the NDI acceptor and the HOMO more delocalized along the thiophene units, Cl insertion on the NDI unit generally more effectively reduces the LUMO energy (−3.53 eV) versus the parent structure I (−3.31 eV), whereas this effect is weaker for structures III and V. On the other hand, the HOMO level is found to be effectively stabilized (Δ = 0.23 eV) in structure V since chlorine substitution is on the donor bithiophene moiety. Considering the synthetic accessibility, acceptable NDI-Thiophene interring torsional angle, and regioregularity of the monomers, structure V was selected for the synthesis of the new chlorinated polymers.

Figure 3.

Computed molecular structures of polymer model structures I–V based on NDI-thiophene.

Synthesis of monomers and polymers

Scheme 1 shows the synthesis of the chlorinated bithiophene monomer 2, and the corresponding copolymers P(NDI2HD–T2Cl2) and P(NDI2OD–T2Cl2). The building block 2 was prepared from 3,3′,5,5′-tetrachloro-2,2′-bithiophene via a sequence of reactions involving selective dehalogenation and lithiation/stannylation. Although enough amount of 2 was obtained in high purity, it is noteworthy that the yield over two steps was found to be <10%, which may be attributed to poor selectivity of dechlorination and the formation of decomposition products.^[ Reference Chan, Ng, Seowa and Moderscheimb ^{72 ]} The syntheses of N,N-dialkylated 2,6-dibromonaphthalene diimides, NDI2HD-Br₂ and NDI2OD-Br₂,were performed according to literature procedures (see Supporting Information), and they were purified by column chromatography. Two different alkyl substituents, 2HD and 2OD, were used to record the effects of alkyl substituents on polymer processibility and optoelectronic properties. The chlorinated polymers, P(NDI2HD–T2Cl2) and P(NDI2OD–T2Cl2) were synthesized via conventional Stille polycondensation protocols by reacting the chlorinated bithiophene monomer 2 with NDI2HD-Br₂ and NDI2OD-Br₂, respectively, in chlorobenzene using Pd₂(dba)₃/P(o-tol)₃ as the catalyst/ligand system. The new polymers were found to be soluble in common non-protic organic solvents including THF, toluene, xylenes, chloroform, and chlorobenzene, and they were purified by sequential soxhlet extractions and multiple dissolution–precipitation processes (93–97% yields). The observed high yields for the polymerization reactions are consistent with the previous reports on different NDI polymerizations, and it indicates the favorable structural/electronic properties of the present monomers for effective Pd-mediated coupling reactions.^[ Reference Guo, Kim, Seger, Jenekhe and Watson ^{73 ]} The final polymer structures and purities were characterized by ¹H NMR spectroscopy, elemental analysis, and gel permeation chromatography. The number average molecular weights (Mn's) of the new chlorinated polymers were determined to be 20.4 kDa (PDI = 2.2) for P(NDI2HD–T2Cl2) and 17.0 kDa (PDI = 1.8) for P(NDI2OD–T2Cl2). To better understand the effects of chlorination on NDI-based polymers without any contribution from molecular weight differences, the non-chlorinated polymers, P(NDI2HD–T2) and P(NDI2OD–T2), were also synthesized with comparable number average molecular weights of 18.5 kDa (PDI 1.9) and 28.6 kDa (PDI = 2.2), respectively.

Scheme 1.

Synthesis of chlorinated bithiophene monomer 2, and the corresponding chlorinated polymers, P(NDI2HD–T2Cl2) and P(NDI2OD–T2Cl2).

Optical and electrochemical properties of polymers

The optical absorption and electrochemical redox properties of the new chlorinated polymers along with the non-chlorinated parent polymers were investigated by thin-film UV–vis absorption and cyclic voltammetry measurements. The optical spectra and voltammograms are shown in Fig. 4, and data are collected in Table I. P(NDI2HD-T2Cl2) and P(NDI2OD–T2Cl2) exhibit two main absorptions corresponding to the π − π* and D–A intramolecular charge transfer (ICT) transitions of the naphthalene diimide–dichlorobithiophene backbone, which are located at λ _max = 376/613 nm and λ _max = 371/607 nm, respectively (Fig. 4(a)).

Figure 4.

(a) Optical absorption spectra of spin-coated P(NDI2HD–T2Cl2) (magenta), P(NDI2OD–T2Cl2) (red), P(NDI2HD–T2) (blue), and P(NDI2OD–T2) (black) films (~30 nm thick) on glass. (b) Thin-film cyclic voltammetries of (NDI2HD–T2Cl2) (magenta), P(NDI2OD–T2Cl2) (red), P(NDI2HD–T2) (blue) and P(NDI2OD–T2) (black) thin films on a Pt working electrode [Fc/Fc⁺ (0.54 V versus SCE) as an internal standard]. (c) HOMO/LUMO energy levels for the present chlorinated and non-chlorinated NDI-based copolymers.

Table I.

Physicochemical properties [molecular weight (Mn, kD), polydispersity (PDI), film optical absorption maxima (λ _max ^film, nm), energy gaps (Eg^opt, eV), electrochemical reduction potentials (E _1/2 ^red,V), and HOMO/LUMO energy levels (eV)] of polymers P(NDI2HD–T2Cl2), P(NDI2OD–T2Cl2), P(NDI2HD–T2), and P(NDI2OD–T2).

^a From optical absorption as spin-coated thin film on glass, optical band gap is estimated from the low-energy band edge of the UV–vis spectrum.

^b As thin film with 0.1 m Bu₄N⁺PF₆ ⁻ in acetonitrile (versus SCE).

^c E _HOMO is calculated from E _g ^opt = E _LUMO − E _HOMO.

^d Estimated from the equation E _LUMO = −4.44 eV − E _1/2 ^red−1.

Both the low-energy and high-energy maxima are shifted to shorter wavelengths by ~13–92 nm versus those of their parent non-chlorinated polymers P(NDI2HD–T2) (λ _max = 389/700 nm) and P(NDI2OD–T2) (λ _max = 392/699 nm). The corresponding optical band gaps (E _g's) are estimated from the low-energy band edges as 1.69 eV for P(NDI2HD–T2Cl2), 1.69 eV for P(NDI2OD–T2Cl2), 1.45 eV for P(NDI2HD–T2) and 1.47 eV for P(NDI2OD–T2). The observed band gap increases for the present chlorinated NDI polymers is consistent with the hypochromic shifts in the absorption maxima. However, the absorption shifts upon chlorination on the lower energy maxima are more profound (∆λ _max = 87–92 nm) compared with those of high energy maxima (∆λ _max = 13–21 nm). This result may be attributed to the different origin of these transitions, and how they are affected by the chlorine substitutions. Since ICT band is affected by the strength of D–A coupling in the polymer backbone, the weaker donor character of the T2Cl2 unit versus T2 reduces D–A  interactions.^[ Reference Anthopoulos, Anyfantis, Papavassiliou and de Leeuw ^{52 ]} On the other hand, the π − π* transition is mostly governed by the extent of π-conjugation. Since the polymer backbone in NDI-bithiophene polymers is already twisted, the larger steric demand of Cl versus H may slightly reduce the backbone planarity on the bithiophene moiety, as supported by the MO computations. This reduces the π-conjugation length and increases the π − π* transition energies.^[ Reference Li, Vamvounis and Holdcroft ^{74 –} Reference de Oliveira, Camilo, da Silva-Filho and Lavarda ^{76 ]} Although the nature of the N-alkyl substituent (2HD versus 2OD) is found to have negligible effects on the polymer absorption maxima and band gap, slight differences were observed in peak vibronic features, which are possibly due to different polymer chain packings in the solid-state.

Cyclic voltammetry measurements provide further insights into the redox properties and energy level changes of the new chlorinated polymers. Two reversible reductions [Fig. 4(b)] are observed for both P(NDI2HD–T2Cl2) and P(NDI2OD–T2Cl2) with the first/second reduction potentials located at −0.40/−0.73 V and at −0.41/−0.75 V, respectively. The reversibility of both reductions demonstrates the redox stability of the new chlorinated polymer backbone. The reduction potentials exhibit anodic shifts compared with those of non-chlorinated polymers [P(NDI2HD–T2) = −0.55/−1.04 V and P(NDI2OD–T2] = −0.49/−0.96 V) when measured in the same electrochemical set-up. This is consistent with the stronger electron-withdrawing characteristics of –Cl compared with –H (vide supra), which facilitates the reduction processes. In addition, chlorine functionalization at NDI-T2 backbone decreases the voltage gap between the first two reductions (E _1/2 ^red−1 − E _1/2 ^red−2) by 0.2 V, further supporting the more electron-accepting nature of these derivatives, which renders the second reduction more facile.

Based on solid-state optical and electrochemical data, the HOMO/LUMO energies are estimated as −5.73/−4.04 eV for P(NDI2HD–T2Cl2) and −5.72/−4.03 eV for P(NDI2OD–T2Cl2). The HOMO and LUMO energy levels of the new chlorinated polymers are ~0.1–0.4 eV lower than those of parent non-chlorinated polymers, P(NDI2HD–T2) (−5.34/−3.89 eV) and P(NDI2OD–T2) (−5.42/−3.95 eV).^[ Reference Lei, Dou and Pei ^{36 ]} The reduction of the LUMO energies (ΔE _LUMO = 0.1–0.15 eV) is less pronounced than those of the HOMO energies (ΔE _HOMO = 0.3–0.4 eV), in agreement with the electronic structure of these polymers. This result is mainly because the LUMO is primarily localized on the NDI acceptor unit whereas the HOMO is more delocalized toward the bithiophene units where the chlorine atoms are located (Fig. S1). The lower LUMO energy level of the present polymers is expected to stabilize the field effect-induced electrons, while the far reduced HOMO energy should reduce the ambipolar characteristics when using certain dielectrics.^[ Reference Kim, Hong, Oh and Yang ^{77 ]} The interesting question is how the considerably decreased NDI–T2 intramolecular electronic coupling and slightly reduced π-conjugation will affect charge transport in OTFTs.

Charge transport measurements

The comparative charge transport properties of the new NDI-based copolymers were investigated in TG/BC transistor geometries. The experimental details are described in the Supporting Information. Briefly, thermally evaporated Au contacts (40 nm) or photolithography-defined Ni (3 nm)/Au (13 nm) were used as the S/D electrodes (L = 2–60 µm; W = 1 mm). The polymer semiconducting layers were deposited by spin-coating 1–10 mg/mL formulations in 1,2-dichlorobenzene or xylene for 1 min, and the films were annealed at 100–300 °C for 30 min. Next, the polymeric dielectric layer (PMMA or CYTOP) was spin-coated, and the devices were annealed to remove residual solvents and induce thin-film ordering. The OTFT devices were completed by depositing 50 nm thick Au or Al TG electrodes via thermal evaporation using a shadow mask. Device characteristics were evaluated under positive or negative gate bias in ambient conditions to explore the majority charge carrier type and device performance. The device carrier mobilities (μ) were calculated both in the linear and saturation regimes using conventional MOSFET equations.

In preliminary experiments, the OTFT characteristics of the chlorinated polymers were investigated with respect to the semiconductor solution concentration and semiconductor film post-deposition annealing temperature (Table S1, Supporting Information). Three different semiconductor solution concentrations (1, 5, and 10 mg/mL) were used resulting in three different active layer thicknesses of ~12, 17, and 23 nm, respectively. In addition, thermal annealing at four different temperatures (100, 150, 200, and 300 °C) were also explored. The output curves reveal the existence of some degree of contact resistance, which was measured for P(NDI2OD–T2Cl2) using the Y-function method.^[ Reference Xu, Minari, Tsukagoshi, Chroboczek and Ghibaudo ^{78 ]} The analysis results (Fig. S3) indicate that the contact resistance of the OTFT devices increases with increasing active layer thickness from 12 nm (~0.5–1 MΩ) to 17 nm (~1 MΩ) and 23 nm (~2.5 MΩ). These values are in the same order of magnitude of that measured for P(NDI2OD–T2) devices (~1 MΩ),^[ Reference Lee, Lee, Shin, Cho and Choi ^{34 ]} and the lowest OTFT resistance (0.5–1 MΩ) was obtained with 1–5 mg/mL spin-coated solutions, which correlates well with the observed higher mobilities for these samples. From this study, it was identified that a concentration of 5 mg/mL affords the best performance whereas semiconductor film thermal annealing at temperatures >150 °C does not significantly improve the device performance, in agreement with negligible variations in film microstructures as accessed by x-ray diffraction experiments (Fig. S2). Thus, these conditions were used to study vis-à-vis the transport properties of the four polymers.

Representative n-channel operation transfer and output plots are shown in Fig. 5 and S4, whereas the OTFT device characteristics are summarized in Table II. The electron mobilities both in the linear and saturation regime of the parent non-chlorinated polymers (0.14–0.63 cm²V⁻¹s⁻¹) are found to be 5–10× larger than those of the corresponding chlorinated derivatives (0.02–0.05 cm²V⁻¹s⁻¹). The reduced electron mobilities most probably reflect backbone distorsions and the far greater band gap of the chlorinated semiconductors. V _on (~0 V) and V _th (~10–20 V) were found to be similar for all polymers. Significantly, the transfer characteristics of the representative optimized devices for all new semiconductors in the linear regime (V _DS = 10 V) exhibit seven to eight order of magnitude current modulations, typical of high-performing organic polymer semiconductors.

Figure 5.

Representative n-channel transfer curves for TG/BC OTFT devices of the indicated semiconductors.

Table II.

Electrical parameters measured in ambient (RH ~ 40%) for TG/BC OTFTs (glass/thermally evaporated Cr–Au/semiconductor/PMMA/Au, L = 50 µm, W = 1000 µm) based on the indicated polymers. The results are given for the average of at least five devices.

Interestingly, the ambipolar characteristics of the chlorinated polymers are greatly suppressed (Fig. S5), resulting in hole mobilities of ~10^–5 cm²V⁻¹cm⁻¹ in the saturation regime, which are about two orders of magnitude lower than those of the parent non-chlorinated systems (~10⁻³ cm²V⁻¹cm⁻¹). These results clearly reflect the stabilization of HOMO energies in the new polymers due to chlorination.

Device stability and OTFT operation in water

The present chlorinated NDI-based copolymers comprise a unique family of n-channel semiconductors exhibiting lowered LUMO energy levels (∆E _LUMO = 0.1–0.2 eV) compared with the parent non-chlorinated systems and typical NDI-based copolymers. In general, the relative positions of the LUMO energies with respect to H₂O- and O₂-redox reactions determines the ambient stabilities of the electrons during the charge transport process in polymer thin films, and a LUMO energy level of <−4.0 eV combined with an hydrophobic gate insulator, is typically required for stable n-channel OTFT device operation.^[ Reference Usta, Risko, Wang, Huang, Deliomeroglu, Zhukhovitskiy, Facchetti and Marks ^{79 ,} Reference Park, Lee, Lee, Lee, Lee, Yoon, Sung and Im ^{80 ]} To this end, numerous small molecular n-channel semiconductors have been developed and exhibit good stabilities.^[ Reference Usta, Kim, Wang, Lu, Huang, Facchetti and Marks ^{50 ]} However, the number of n-channel polymers operating in ambient is still very limited and their operational stabilities are insufficient for most applications. More surprisingly, TFT operation under high-humidity and aqueous solutions has only been tested for a handful of polymers.^[ Reference Hwang, Fuentes-Hernandez, Fenoll, Yun, Park, Shim, Knauer, Dindar, Kim, Kim, Kim, Cheun, Payne, Graham, lm, Anthony and Kippelen ^{81 ]} Therefore, it remains very important to develop and understand the performance of new semiconductor materials with superior stabilities, including under the aforementioned extreme conditions, which will extend the scope of n-channel OTFTs in chemical and biological sensor applications in aqueous media.^[ Reference Yun, Sharma, Fuentes-Hernandez, Hwang, Dindar, Singh, Choi and Kippelen ^{82 ]} When compared with non-chlorinated NDI-based copolymers, the present copolymer semiconductors should exhibit better ambient stabilities. Thus, we investigated TG/BC transistor performance of the devices based on non-chlorinated P(NDI2OD–T2) versus those of the chlorinated P(NDI2HD–T2Cl2). The TFT structure for this study is glass/photolithography-Ni–Au/semiconductor/dielectric/Al, the gate dielectric is either PMMA (thickness ~500 nm, C _i  ~ 6.19 nF/cm²) or CYTOP (thickness ~ 380 nm, C _i  ~ 4.89 nF/cm²), and L = 20 µm/W = 1000 µm. The devices were first tested in a nitrogen filled glove box, and then stored under an average relative humidity of 60.5 ± 2.8% at 24.6 ± 1 °C for 1–3 weeks, during which time the devices were periodically measured.

Figure S6 shows representative transfer plots whereas Figs. 6(a) and 6(b) report the mobility variations of these devices over time. Table III summarizes major TFT parameters for these devices. Although both P(NDI2HD–T2Cl2) and P(NDI2OD–T2) devices with PMMA as gate dielectric exhibit some degradation over time, overall P(NDI2HD–T2Cl2)-based devices showed better stabilities than P(NDI2OD–T2). Regardless of the lower electron mobilities of P(NDI2HD–T2Cl2) versus the well-studied P(NDI2OD–T2), P(NDI2HD–T2Cl2) devices exhibit a ~20% electron mobility degradation after ~1 week of storage, whereas higher degradation (35%) is recorded for P(NDI2OD–T2) (Fig. 6). Both devices exhibits a considerable V _th shift from ~+(10–20) V to ~+(40–45) V after ~1 week of storage indicative of formation of deep traps. However, after these performance variations, the I–V characteristics stabilize and remain constant for the subsequent ~2 weeks. The TFT operation and improved stability of P(NDI2HD–T2Cl2)-based devices was further confirmed by submerging tests in water. As shown in Fig. S7, Fig. 6(c) and 6(d), the performance of P(NDI2HD–T2Cl2)-based OTFT devices with PMMA as the gate dielectric exhibit far lower degradation after 3 weeks submergence in water, whereas those based on P(NDI2OD–T2) exhibit larger degradation.

Figure 6.

Device stabilities of P(NDI2HD–T2Cl2)-and P(NDI2OD–T2)-based OTFTs with PMMA in various conditions: ambient storage (RH: 60.5 ± 2.8 %,Temp.: 24.6 ± 1 °C) (a, b), H₂O immersion (DI water) (c, d). (e) Optical image of a submerged device array. (f) Operational stability under supplied bias (V _GS = 60 V, V _DS = 20 V).

Table III.

Electrical parameters of TG/BC OTFTs fabricates with P(NDI2OD–T2) and P(NDI2HD–T2Cl2) during stability measurements.

It is important to note that the devices fabricated with CYTOP dielectric are found to be highly unstable and quickly become unfunctional upon immersion in water. At first, this result appears to be quite surprising, because several groups reported excellent device stabilities for CYTOP devices in highly moisturized conditions as a result of the dielectric's high hydrophobic nature.^[ Reference Yun, Sharma, Fuentes-Hernandez, Hwang, Dindar, Singh, Choi and Kippelen ^{82 –} Reference Cheng, Caironi, Noh, Wang, Newman, Yan, Facchetti and Sirringhaus ^{84 ]} It is likely that this behavior originates from the poor adhesion of CYTOP to both conjugated polymer films. In fact, the CYTOP/gate layer of the OTFTs immersed in water for one week was found to pill-off when the devices were blown with N₂ gas to dry them before the electrical tests (Fig. S8). This result suggests that there could be penetration of ambient moisture into the solid films, which might cause delamination of the dielectric from the channel and result in the observed significant degradation.

Finally, bias-stress stability tests for the P(NDI2OD–T2)/PMMA and P(NDI2HD–T2Cl2)/PMMA devices were carried out in an N₂-filled glove box to investigate the intrinsic stability of these devices when water/air is excluded. For this study, the TFT structure is glass/Au/semiconductor/dielectric/Al (W/L = 1000/20 µm). Under these conditions, both polymers exhibit excellent operational stabilities with <3% I _on current variation [Fig. S9 and Fig. 6(f)].^[ Reference Khim, Baeg, Kim, Kang, Lee, Chen, Facchetti, Kim and Noh ^{85 ]} Overall, the improved stability in high-humidity/aqueous conditions for the devices based on P(NDI2HD–T2Cl2) indicate that chlorination is a facile modification strategy to broaden the application of NDI-based polymer semiconductors.