Polymer synthesis and characterization. Commercially available TDPP (1) was used as the acceptor moiety. Compared with other n-type materials with long synthetic steps, the bromo-substituted TDPP (3) monomer used for polymerization only needs two steps to be obtained11,15,25,36,37. To compare the different electronic structures of the doped states, 2,5-bis(trimethylstannyl)thiophene (T) and 2,5-bis(trimethylstannyl)-3,4-difluorothiophene (2FT) were chosen to construct two similar polymers, namely P(gTDPPT) and P(gTDPP2FT) (Fig. 1). Ethylene glycol side chains (R) with farther branched positions were chosen for a closer π−π stacking distance and potentially enhanced charge carrier mobility, as we reported before38,39. Both polymers were obtained via Pd-catalyzed Stille coupling reactions in the presence of CuI as the co-catalyst40. Both polymers were purified by Soxhlet extraction and finally collected by chloroform. The molecular weights of the polymers were evaluated by gel permeation chromatography (GPC) using hexafluoroisopropanol (HFIP) as the eluent4. P(gTDPP2FT) shows a Mw/Mn = 65.04/30.70 kDa, comparable to other p- or n-type polymers4. Both polymers exhibit good thermal stability with high decomposition temperatures (Fig. S1, S2).
The optoelectronic properties of both polymers were evaluated using UV-Vis-NIR absorption spectra, cyclic voltammetry (CV), and spectroelectrochemistry (Fig. 2). The spectrum of P(gTDPP2FT) shows very similar maximum absorption peak (832 nm) and bandgap (1.34 eV) to that of P(gTDPPT) (836 nm, 1.36 eV) (Fig. 2a & b). Both spectra of P(gTDPPT) and P(gTDPP2FT) show a redshift in film and annealed film compared with the solution state, largely due to the further aggregation of the polymers. The 0-0/0-1 vibrational absorption peak ratio of P(gTDPP2FT) is larger than that of P(gTDPPT), suggesting a more planar backbone structure. The spectra results are consistent with the relaxed potential energy surface (PES) scan calculations. The PES scans at the dihedral angles of the TDPP-T/2FT show that both polymers have similar torsional barriers. P(gTDPP2FT) exhibits a dominant conformation at 0o at the TDPP-2FT dihedral angle, while (P(gTDPPT) exhibits a 30o dihedral angle (Fig. 2c).The measured ionic potentials (IP) and electron affinities (EA) of P(gTDPP2FT) are estimated to be 5.20 eV and 3.86 eV, higher than that of P(gTDPPT) (4.86 eV, 3.69 eV) (Fig. S3a & b), which is consistent with the DFT calculation results (Table S1, Fig. S4). Continuous CV sweep measurements of two polymers were explored in 0.1 M NaCl aqueous solution as the electrolyte, and both show good electrochemical stability (Fig. S3c & d). Interestingly, the DFT calculated torsion barriers of both polymers increase further after being n-doped. Besides, the bond length of TDPP-2FT (1.449 Å) is shorter than TDPP-T (1.454 Å), suggesting the enhanced conjugation of P(gTDPP2FT) (Fig. 2d).
Spectroelectrochemistry was performed to investigate the electrochemical characteristics of both polymers (Fig. 2e & f). Since P(gTDPPT) is a p-type OECT material, it was charged by oxidization, while P(gTDPP2FT) was charged by reduction. Driven by the positive voltage, chloride ions in the electrolyte penetrate into the P(gTDPPT) film to keep the charge neutrality of the polymer film. On the contrary, P(gTDPP2FT) was reduced by the negative voltage, and sodium ions are the counter ions. The electrochemical doping process generated polarons/bipolarons in both polymers films, showing new absorption bands in the long-wavelength region. The neutral polymers’ absorption bands (700-900 nm) decrease, and the polaron/bipolaron absorption bands (900-1200 nm) rise.
OECT device fabrication and characterization. The OECT devices were fabricated using a photolithography and parylene patterning method35. The polymers were deposited using their chlorobenzene solution by spin-coating (see SI for more details). To evaluate the performance of an OECT material, the following equation based on the Bernards’ model is often used (Eq. 1):41
gm = (W/L) ∙ d ∙ μ ∙ C* ∙ |(Vth - VGS)| (Eq. 1)
where W, L, and d are the channel width, length, and film thickness, respectively, µ denotes the charge carrier mobility, C* denotes the capacitance of the channel per unit volume, and Vth is the threshold voltage.
We applied both positive and negative gate voltages for both polymer devices. P(gTDPPT) shows typical p-type OECT behaviors, with a μC* of up to 65.1 F cm−1 V−1 s−1, while P(gTDPP2FT) shows an outstanding pure n-type OECT behaviors, with a high μC* of up to 54.8 F cm−1 V−1 s−1 (Fig. 3a-d, Table 1), which is a record value in the literature reported to date. Both polymers show a similar threshold voltage with an absolute value of around 0.6 V. To exclude the potential side-chain effects, we also synthesized P(lgTDPP2FT), with the same backbone as P(gTDPP2FT) but a linear side chain (Fig. S5). P(lgTDPP2FT) also shows n-type OECT behaviors with a high μC* of 20.4±1.0 F cm−1 V−1 s−1, which demonstrates the high n-type OECT performance of P(gTDPP2FT) come from the introduction of F atoms, not the side chains. The volumetric capacitance (C*) was measured by electrochemical impedance spectrum (EIS) (Fig. S6). The maximal C* was extracted with an average value of 161 F cm−3 for P(gTDPPT), and 156 F cm−3 for P(gTDPP2FT). Based on the μC* and C* values, the hole/electron mobility (μ) was calculated to be 0.40 cm2 V−1 s−1 for P(gTDPPT) and 0.35 cm2 V−1 s−1 for P(gTDPP2FT). Furthermore, we tested the transient characteristics of their OECT devices to evaluate the response speed of both polymers (Fig. 3e & f). A pulse was applied to the gate electrode, and a DC voltage with an absolute value of 0.6 V was applied to the drain electrode. The response time was estimated by an exponential fitting of the IDS. P(gTDPPT) and P(gDPP2FT) both exhibit short response times, with τon/τoff of 0.46/0.08 ms and 1.75/0.15 ms, respectively. The high μ and fast response characteristics make P(gTDPP2FT) a promising material for real-time high-speed sensing applications. The two polymers were used to fabricate complementary inverters because of their matched operating voltage and device performance. When the supply voltage (VDD) is set to 0.8 V and the input voltage (VIn) is swept from 0 to 0.8 V, a relatively high gain value (∂VOut/∂VIn) of 26.8 was obtained (Fig. 3g). The μC* and μ of P(gTDPP2FT) are both record values, and the response times are among the shortest in n-type OECT materials (Fig. 3h & i)25,27-30,38,42.
Table 1. Summary of the OECTs Performance and Molecular Packing Parameters of the Polymers.
|
Type
|
gma
(mS)
|
d
(nm)
|
Vthb
(V)
|
Ion/Ioff
|
μC* c
(F/cm V s)
|
μd
(cm2/V s)
|
C*
(F cm−3)
|
τon
(ms)
|
τoff
(ms)
|
dlamellar
(Å)
|
dπ-π
(Å)
|
P(gTDPPT)
|
p
|
1.18
|
60.5
|
-0.60
|
5×106
|
65.1
(45.9±13.7)
|
0.40
|
161
|
0.46
|
0.08
|
26.18
|
3.68
|
P(gTDPP2FT)
|
n
|
0.93
|
60.6
|
0.62
|
5×106
|
54.8
(42.2±6.5)
|
0.35
|
156
|
1.75
|
0.15
|
27.32
|
3.68
|
aThe W/L of all the devices is 100/10 μm. All the OECT devices were operated in a 0.1 M NaCl aqueous solution. bVth was determined by extrapolating the corresponding IDS1/2 vs. VGS plots. cFour devices were tested and computed for each polymer. μC* was calculated according to Eq. 1. dμ was calculated from the μC* and the measured volumetric capacitance C*.
Film microstructure characterization. Grazing incidence wide-angle X-ray scattering (GIWAXS) and atomic force microscope (AFM) were employed to explore the molecular packing and morphology. Both P(gTDPPT) and P(gTDPP2FT) show typical face-on molecular packings (Fig. 4a & b). The two polymers have the same π-π stacking distance of 3.68 Å, and P(gTDPP2FT) shows a slightly larger lamellar distance (Table 1 and Fig. S7). P(gTDPP2FT) exhibits three in-plane lamellar scattering peaks with narrower half peak width, (100), (200), and (300), suggesting its more ordered molecular packing and higher crystallinity. Both polymer films are smooth with small root-mean-square (RMS) roughness in atomic force microscope (AFM) height images (Fig. 4c & d). P(gTDPP2FT) film shows fiber-like textures, while P(gTDPPT) film is more amorphous. The AFM results are consistent with GIWAXS results.
Understanding of the “doped state engineering” strategy. The outstanding OECTs performances of P(gTDPP2FT) are out of our expectations because its LUMO energy level is rather high compared with several typical n-type OECT materials (Fig. 5a). P(gPyDPPT) was also synthesized for comparison (Fig. 5b). P(gPyDPP-T2) with bithiophene as the donor moiety was reported by Giovannitti et al. It showed very poor p-type OECT performance34. We synthesized P(gPyDPPT) here, and it shows poor n-type OECT performance with μC* of 0.07 F cm−1 V−1 s−1 (Fig. S8). The introduction of pyridine and F atoms both reduce the LUMO energy level of P(gTDPPT) and the difference value between the LUMO energy level of P(gPyDPPT) and P(gPyDPP2FT) is less than 0.1 eV (Fig. 5c). We summarized the relationship between LUMO energy level and device performance of several n-type OECT polymers (Fig. 5a). The μC* value is not correlated well with LUMO energy levels. These results indicate that high-performance n-type OECT materials cannot be simply obtained by lowering the LUMO energy levels.
OECT materials usually work under highly doped states. We propose that the molecular properties under highly doped states might significantly affect the charge transport properties. Therefore, we calculated the properties of the three polymers’ doped states. The energy difference between the neutral and negatively charged state (ΔE = Enegative − Eneutral) of P(gPyDPPT) is −48.93 kcal/mol, which is much smaller than that of P(gTDPP2FT) (−52.99 kcal/mol), and even smaller than that of P(gTDPPT) (−49.01 kcal/mol). These results suggest that the negative polarons on P(gTDPP2FT) backbone are more stable, and the stability is not related to the polymer LUMO energy levels (Fig. 5a, Fig. S9). We also calculated the charge distribution of the three polymers relative to their neutral state (Fig. 5d-f, Fig. S10). The negative charges of n-doped [P(gTDPP2FT)]1− distribute on the whole polymer chain, whereas the negative charges of [P(gPyDPPT)]1− are mostly localized in the center of the chain. Compared to [P(gTDPPT)]1−, the negative charges of [P(gTDPP2FT)]1− on the T moieties next to DPP are shared by the donor moiety 2FT (black and red arrows in Fig. 5d & f), which makes the charge distribution more balanced. On the contrary, the positive charges of P(gTDPPT) are the most delocalized, while the positive charges of the other two polymers are located on one end of the chain. Besides, the distribution of dihedral angles between fragments is quite different for the three polymers (Fig. 5g-i). P(gTDPP2FT) shows the smallest dihedral angles along the polymer backbone at the neutral state, which decrease further after being negatively charged. Relatively large dihedral angles of P(gTDPPT) decrease a little after being negatively charged. Conversely, P(gPyDPPT) exhibits the largest dihedral angles, which do not change much after being both positively and negatively charged. All these charge and dihedral angle distributions prove that the introduction of fluorine atoms in P(gTDPP2FT) not only lowers the LUMO energy level but also enhances the polymer backbone planarity, delocalizes, and stabilizes the negative polaron. These favorable factors might explain the n-type charge transport behavior and high electron mobility of P(gTDPP2FT) under strong electrochemical n-doping.