Green Synthesis of Lactone-Based Conjugated Polymers for n-Type Organic Electrochemical Transistors

As new and better materials are implemented for organic electrochemical transistors (OECTs), it becomes increasingly important to adopt more economic and environmentally friendly synthesis pathways with respect to conventional tran-sition-metal-catalyzed polymerizations. Herein, a series of novel n-type donor– acceptor-conjugated polymers based on glycolated lactone and bis-isatin units are reported. All the polymers are synthesized via green and metal-free aldol polymerization. The strong electron-deficient lactone-building blocks provide low-lying lowest unoccupied molecular orbital (LUMO) and the rigid backbone needed for efficient electron mobility up to 0.07 cm 2 V − 1 s − 1 . Instead, polar atoms in the backbone and ethylene glycol side chains contribute to the ionic conductivity. The resulting OECTs exhibit a normalized maximum transconductance g m,norm of 0.8 S cm − 1 and a μ C* of 6.7 F cm − 1 V − 1 s − 1 . Data on the microstructure show that such device performance originates from a unique porous morphology together with a highly disordered amorphous microstructure, leading to efficient ion-to-electron coupling. Overall, the design strategy provides an inexpensive and metal-free polymerization route for high-performing n-type OECTs.


Introduction
Organic electrochemical transistors (OECTs) are electronic devices that take advantage of the mixed ionic and electronic transconductance g m , and by the product of electronic charge carrier mobility and volumetric capacitance μC*; parameters that are linked by Equation (1) (1) where W is the channel width, d is the channel depth, L is the channel length, V TH is the threshold voltage, and V G is the gate voltage. The maximum transconductance normalized by channel geometry g m,norm provides information on the amplification capabilities of the device. Moreover, the product μC* captures the steady-state ionic/electronic transport of the materials independently on device geometry and can be further decoupled to estimate the individual contribution of electronic mobility μ and volumetric capacitance C* to OECT performance. Most OECTs developed so far are built using p-type conjugated polymers (hole conductors) as core components. [9][10][11][12][13][14][15][16][17] In contrast to n-type materials (electron conductors), p-type conjugated polymers offer to date better device performance, with μC* over 300 F cm −1 V −1 s −1 for, as an example, all p(g2T2-g4T2)-based polymers, [18] poly(2-(4,4′-bis(2-methoxyethoxy)-5′-methyl-[2,2′bithiophen]-5-yl)-5-methylthieno [3,2-b]thiophene) p(gBTTT), [19] and p(DPP-T2)-based donor-acceptor (D-A) copolymers. [20] The superior performance is ascribed to their better stability in contact with air or aqueous media, higher charge carrier mobility, and/or electrochemical stability with respect to n-type polymers. However, both n-and p-type materials are needed to increase the level of sophistication of OECT-based technologies and to develop complementary OECT amplifiers, [21] whose operation is currently limited by the comparatively low performance of n-type polymers.
The jump in μC* from 0.06 F cm −1 V −1 s −1 for NDI-based polymers to 25.9 F cm −1 V −1 s −1 for BBL152 was enabled by strong π-π interaction and high crystallinity by improved molecular weight. [31] Overall, electronic transport is facilitated mostly by building blocks that provide good solubility, electronwithdrawing capabilities, and planar backbone. Instead, ionic transport is enabled by low crystallinity and polar atoms both in the backbone and side chains, aiding the transport of hydrated electrolyte ions. In addition to ionic and electronic charge transport, materials for n-type OECTs require low-lying lowest unoccupied molecular orbital (LUMO) level, providing an ionization potential that is low enough for conjugated polymer reduction in contact with aqueous electrolytes.
In a parallel trend, the need for greener and bio-friendly materials is pushing researchers to think beyond standard synthesis methods and integrate polymerization strategies free from highly toxic reagents, such as the hazardous monomers and transition-metal catalysts used in Stille and Suzuki couplings. [32,33] Alternative synthesis pathways would enable both sustainable products manufacturing and meet the strict regulations concerning the use of chemicals in biomedical devices. [33,34] To date, most semiconducting polymers for OECTs are synthesized via transition-metal-mediated cross-coupling polymerizations, producing stoichiometric quantities of hazardous byproducts containing heavy metals, such as organostannane. [29,35] To address these critical issues, future materials design should integrate monomers enabling both high performance and synthesis via environmentally friendly and economic polymerization routes (e.g., aldol polycondensation). [33] Conjugated polymers comprising bisistain and lactone building blocks were introduced for organic field-effect transistors (OFET) and thermoelectrics. Their rigid backbone and low-lying LUMO level led to materials with high electronic mobility (up to 7.4 cm 2 V −1 s −1 ). [36][37][38][39] Our work is inspired by these developments to address the need for greener and better performing materials for n-type OECTs. We postulated that the rigid structure and low-lying LUMO level of lactone monomers, combined with ethylene glycol (EG) side chains, will provide the balance of electronic and ionic mobility needed for highperforming n-type OECTs. Thiophene donor derivatives have been introduced into the backbone to further regulate intermolecular interactions and to tune the ionic and electronic transports of the polymers.
The lactone building blocks enabled synthesis via metal-free and bio-friendly aldol polymerization (shown in Scheme 1). Moreover, they provided the electron-withdrawing capability, planarity, and effective electron transport properties, leading to electronic mobility μ e as high as 0.07 cm 2 V −1 s −1 . The short packing distance, combined with porous microstructure, led to a path for ionic conductivity and a volumetric capacitance C* up to around 100 F cm −3 . Overall, OECTs exhibited a normalized transconductance g m,norm of 0.8 S cm −1 and a μC* of 6.7 F cm −1 V −1 s −1 for p(C-T), showing that these are excellent units to push toward the next-generation OECTs combining high performance and benign polymerization strategies.

Results and Discussion
Glycolated 6-bromo-isatin and 6-bromo-azaisatin were afforded from 6-bromo-isatin and 6-bromo-azaisatin by nucleophilic substitution of the desired tosylated-EG side chain. The length of the polar EG side chains (n = 6 and n′ = 2) was chosen to ensure enough polymers solubility in organic solvents for processing, film stability in contact with aqueous solvents, and the ionic conductivity needed for effective doping and de-doping. The chemical structures of the monomers and polymers are shown in Scheme 1. Dilactone was synthesized by a dehydration reaction using commercially available 2,2-(2,5-dihydroxy-1,4-phenylene)diacetic acid and acetic anhydride. [40] The other key bis-isatin monomers (M1, M2, M3, and M4) were synthesized by Stille coupling between the glycolated bromo-isatin or bromo-azaisatin and corresponding difunctionalized tin reagents. Although Stille coupling was used for bis-isatin monomers' preparation, it is anticipated that more atom-economic CH functionalization is possible. [32] The polymers p(C-T), p(N-T), p(C-2T), and p(C-g2T) were ultimately synthesized by aldol polycondensation between the enolizable carbonyl unit of the dilactone monomer, and the electrophilic carbonyl unit of the corresponding bis-isatin comonomers catalyzed by p-toluenesulfonic acid monohydrate (PTSA). The detailed procedures are described in the Supporting Information. Compared to conventional expensive transition-metal-catalyst polymerization for constructing D-A polymers, this route offers the advantages of using benign conditions and avoiding the production of stoichiometric quantities of highly toxic organostannane byproducts. All polymers exhibit good solubility in warm chloroform and chloroform:o-dichlorobenzene, allowing for solution processing. The molecular weight of the polymers was estimated via matrix-assisted laser desorption ionization time of flight (MALDI-TOF). We observed peaks around 20 kDa for all polymers (see the Supporting Information), indicating that the polymer series have moderate molecular weights. We could not gather realistic results from gel permeation chromatography (GPC), likely due to strong aggregation caused by the high polarity and rigid backbone of these polymers. [23] UV-vis-NIR absorption spectra were performed on the spincoated polymer films on glass (Figure 1a; Table S1, Supporting Information). Similar to their alkylated counterparts, [41,42] these polymers exhibit the two typical bands that characterize D-A copolymers; the high-energy band between 400 and 600 nm is attributed to the π-π* transition, while the high-energy band between 600 and 1400 nm is ascribed to intramolecular charge transfer (ICT). p(C-T), p(C-2T), and p(N-T) display similar absorption maximum wavelength at around 771 nm and optical bandgaps (E g,opt , derived from the onset of the absorption spectra of the polymer films) around 1.28 eV. In contrast to p(C-T) and p(C-2T), p(N-T) exhibits a blueshifted π-π* transition band (around 440 nm) and a shoulder peak at 834/774 nm, attributed to a stronger electron-deficient ability of azabisisatin. In comparison with p(C-T), the maximum absorption wavelength of p(C-2T) shows a non-negligible blueshift, leading to a slightly smaller bandgap of 1.24 eV with respect to p(C-T) (1.29 eV). The addition of glycol side chains to the bithiophene units of p(C-T) afforded p(C-g2T). p(C-g2T) exhibits the most redshifted absorption maximum at 922 nm and the narrowest optical bandgap of 0.92 eV within the polymer series. These features are ascribed to the presence of electron-donating methoxy substituents on the bithiophene unit in p(C-g2T). [20] To estimate the electrochemical properties of the polymers, cyclic voltammetry (CV) of the films spin-coated on indium-tin oxide (ITO)-coated glass was conducted using either an organic or an aqueous electrolyte (Figures S1 and S2, Supporting Information). The onset of the first reduction potential in 0.1 m tetrabutylammonium phosphate in acetonitrile (E red,org ) and in 0.1 m aqueous NaCl (E red, aq ), as well as LUMO and highest occupied molecular orbital (HOMO) energy levels, are Notably, all polymers have low-lying LUMO, below −4.02 eV, a criterion needed to avoid electron trapping by water and oxygen. [22,29,43] We then estimated the HOMO level by subtracting the optical bandgaps (E g,opt ) from the LUMO levels, leading to −5.53 eV for p(C-T), −5.76 eV for p(N-T), −5.43 eV for p(C-2T), and −4.98 eV for p(C-g2T), respectively.
Overall, the results show that the highly electron-deficient nature of the bis-azaisastin with respect to bis-isastin leads to the lowest-lying LUMO and HOMO levels of p(N-T) among these polymers. In contrast, p(C-g2T) shows the highest LUMO and HOMO levels due to the strong electron-donating ability of methoxy bithiophene. [20] The LUMO level value of p(C-T) is slightly lower with respect to p(C-2T), due to the relatively weaker electron-donating ability of thiophene with respect to bithiophene. The observed trend in the LUMO level results is consistent with the values previously observed for their alkyl chain counterparts. [41,42] Switching from organic media to aqueous media led to the same trend in reduction onset ( Figure S1b, Supporting Information). The difference between the onset in organic E red,org and aqueous media E red,aq (ΔE re ) can be used to estimate favorable energetics of ion bulk doping into the semiconducting polymers. The ΔE re values of p(C-T), p(N-T), p(C-2T), and p(C-g2T) are 0.02, 0.04, −0.03, and 0.06 V, respectively. The relatively small differences suggest that cations can penetrate easily into the semiconducting polymers from both organic and aqueous media. All the polymers exhibit a positive voltage shift ΔE re apart from p(C-2T). This difference might be attributed to the lower density of glycol side chains, leading to a less favorable Na + penetration into the polymer film with respect to the other three polymers. Five charging and discharging cycles are presented in Figure S2 (Supporting Information). p(C-T), p(N-T), and p(C-2T) exhibit highly reversibility in aqueous media, while p(C-g2T) shows lower stability, which might negatively impact its usability as the active material for OECTs.
UV-vis-NIR spectro-electrochemistry was performed to monitor the reduction behavior of polymer films upon application of electrochemical potential in an aqueous electrolyte.
The films were spin-coated on ITO-coated glasses, immersed in 0.1 m aq. NaCl, and electrochemically switched from fully neutral to n-doped forms using Ag/AgCl as a reference electrode. The biggest spectral changes were found in the region above 600 nm. As the potential magnitude increases, the absorption arising from the ICT peaks between 600 and 1000 nm begins to noticeably decrease in intensity, while new absorption features appear above 1100 nm, ascribed to the formation of polaronic/bipolaronic species. p(N-T) exhibits three isosbestic points at around 890, 600, and 450 nm upon doping. In contrast, p(C-T) and p(C-2T) show a clear isosbestic point at around 800 nm and smaller changes at around 600 and 450 nm. We ascribed this difference to the presence of electronegative nitrogen atoms in p(N-T) backbone, which are absent in the structure of the other two polymers. Spectro-electrochemical data for p(C-g2T) ( Figure S3, Supporting Information) shows the least changes, ascribed to its highest LUMO level in which electrons can be easily trapped by water and oxygen. The overall relative changes of the ICT and polaron peaks' absorption spectra of each polymer film upon electrochemical doping at various voltages are shown in Figure 2d. p(C-T), p(N-T), and p(C-2T) are completely reduced potentials above −0.4 V. p(N-T) exhibits the biggest changes in absorbance at the lowest electrochemical potential, in agreement with its low E red,org , E red,aq , and LUMO level, which is attributable to the presence of electron-deficient pyridine units in the polymer backbone. In contrast, p(C-g2T) exhibits negligible changes in absorbance, which can be related to its high LUMO level (the highest within the polymer series). We thereby decided to exclude this polymer from further characterization. Overall, the results indicate that the energetics associated with doping from hydrated Na + ions become increasingly less favorable upon increasing of LUMO level.
We then tested the polymers as active materials for OECT having 0.1 m aq. NaCl as the supporting electrolyte and a silver pellet as the gate electrode. The devices were fabricated using previously reported methods (details provided in the "Experimental Section") [11] and had a channel width of 500 μm (W), length of 20 μm (L), and a comparable thickness of around 70 nm. The output and transfer characteristics are shown in Figure 3, while the steady-state OECT performance is summarized in Table S2 (Supporting Information). As shown by spectroelectrochemical data, p(C-g2T) exhibits low doping levels and severe instability, preventing the acquisition of reliable OECT data. Figure 3 shows that all other polymers led to devices working in accumulation mode upon application of the positive drain voltages, consistent with their n-type semiconducting character. p(N-T) reached a maximum drain current of 22.5 μA (V G = 0.45 V), which is approximately two times higher than   Table 1).
The threshold voltage (V TH ) is also an important parameter for OECTs, as a low V TH translates into a lower onset for chargecarrier generation, leading to lower power consumption and noise. Interestingly, the averaged transfer curves show negative OFF currents for the investigated polymers, which is probably due to the trapping of minority carriers. [44] We used the OFF current rather than the zero current to estimate the threshold voltage V TH (as shown in Figure 3d Table S2, Supporting Information). The observed trend for V TH values correlates well with the onset potential in UV-vis-NIR spectro-electrochemistry (Figure 2d), and the reduction onset observed in CV curves in both organic and aqueous media, confirming that the low-lying LUMO levels decrease the energetic barrier of ion injection into conjugated polymers.
We then calculated the μC* by Equation (1). Data are summarized in Figure 4 and Table 1. Polymers exhibit a high μC* value of 6.4 F cm −1 V −1 s −1 for p(C-T), 4.7 F cm −1 V −1 s −1 for p(C-N), and 1.2 F cm −1 V −1 s −1 for p(C-2T), respectively, consistent with normalized transconductance data. When compared with the current figure of art n-type OECT polymers, around 4 and 3.5 times higher normalized transconductance for p(C-T) and p(N-T) than that of PgNaN. The μC* for p(C-T) is 1 magnitude higher than that of PgNaN and BBL, and are outcompeted Extracted from the slope of OECT transfer curves as a function of VG; b) Maximum transconductance normalized by channel thickness and aspect ratio; c) Estimated using Equation (1), at maximum g m ; d) Volumetric capacitance measured with electrochemical impedance spectroscopy; e) OECT electron mobility of the films calculated using Equation (1) by using the C* values.  Table S2 in the Supporting Information).
To better compare device performance, we further decoupled μC* into the charge mobility μ and volumetric capacitance C*. We first conducted electrochemical impedance spectroscopy (EIS) to obtain the volumetric capacitance C* and estimate ion uptake. The data were acquired at −0.4 V and then fitted in two steps to calculate the volumetric capacitance: i) fit every impedance data with a model circuit R s (R p ||C) to find the capacitance; ii) fit the capacitance data with a linear fit to extract the volumetric capacitance. Experimental Bode plots (log|Z| vs frequency, and phase vs frequency) agree well with theoretical Bode plots, as shown in Figure S5 (Supporting Information). The final fit shows that the volumetric capacitance decreases from 97 F cm −3 for p(C-T), 73 F cm −3 for p(N-T), and 53 F cm −3 for p(C-2T) (Figure 4c and Table 1). The lowest C* for p(C-2T) among the three polymers is attributed to its chemical design having the lowest ratio of EG in the conjugated polymer backbone and a lower degree of ion hydration and uptake. Charge mobility μ was extracted by μ = [μC*]/C*, leading to 0.069 cm 2 V −1 s −1 for p(C-T), 0.059 cm 2 V −1 s −1 for p(N-T), and 0.019 cm 2 V −1 s −1 for p(C-2T), respectively.
Device stability was established by continuous ON-OFF cycles via repeated gate pulsing, as shown in Figure S6 (Supporting Information). OECTs retained around 12.5%, 63.0%, and 72% for p(C-T), p(N-T), and p(C-2T), respectively, after 30 min cycling (1000 cycles for p(C-T) and p(C-2T), and 4200 cycles for p (N-T)). These differences could be related to the LUMO energy level of the polymers and/or to irreversible changes in the morphology of the films upon electrochemical doping. We found that the trend observed in device stability (p(C-T) ≪ p(N-T) < p(C-2T)) correlates well with the volumetric capacitance C* of the polymer films, which follows the inverse trend p(C-T) > p(N-T) > p(C-2T). Indeed, we expect that the polymers showing the highest values of C* can accommodate the largest number of hydrated electrolyte ions. However, this could lead to changes in the microstructure of the film and to the lower stability observed for p(C-T).
In contrast, p(C-2T), with a lower density of EG side chains, exhibits the smallest volumetric capacitance (and penetration of hydrated electrolyte ion) among the three polymers. Such results indicate that irreversible changes in the packing of the polymer films during doping might be the leading cause of instability.
Previous research also showed that OECT characteristics, such as response time and transconductance, are linked to channel geometry. [45] Depositing the polymers on interdigitated electrodes having a channel length of L = 20 μm and a width of W = 39 000 μm (39 parallel channels having W = 1000 μm) led to better device stability, retaining 15.9% for p(C-T), 71.3% for p (N-T), and 67.5% p(C-2T) of maximum current under square wave gate voltage for 1 h (2000 cycles)-around double with respect to the single channel ( Figure S7, Supporting Information). This difference indicates that interdigitated channels, while suitable for certain applications, might not be the best choice to evaluate materials' parameters independently of channel geometry, such as stability.
The interfacial area between the electrolyte and the active layer is a critical OECT parameter since it directly affects the exchange of ions between the two phases. Increasing the electrolyte-semiconductor interfacial area is expected to improve ion exchange similar to what is observed for catalytic and gas sensor systems. [46,47] Ex situ atomic force microscopy (AFM) was performed to gain an insight into the surface morphology of the polymer films both in dry and hydrated forms, at different doping levels. Figure 5a and Figure S8 for p(N-T). In contrast, p(C-2T) and p(C-g2T) appear dense with an even distribution across the surface with much lower RMS values of 7.35 and 1.93 nm (Figure 5a; Figure S8, Supporting Information). The data correlate well with volumetric capacitance (C*) of the polymers, which decrease in the order p(C-T) > p(N-T) > p(C-2T), likely due to a decrease in interfacial area and ion transport capacity.
Once immersed into 0.1 m NaCl (hydrated state), the polymers exhibit different degrees of swelling ability. p(C-T), p(N-T), and p(C-g2T) feature rougher surface morphology in their hydrated state with respect to their dry state, with increased RMS. Conversely, p(C-2T) exhibits a smoother morphology with decreased RMS. When subjected to an applied voltage of −0.4 V, p(C-T) and p(C-g2T) show the greatest changes in film morphology and a lower RMS with respect to their hydrated phase (Figure 5a; Figure S8, Supporting Information). These data suggest that ion and water uptake during electrochemical cycling leads to irreversible changes in the morphology of p(C-T) and p(C-g2T) films, resulting in the observed poor operational stability. On the contrary, p(N-T) and p(C-2T) show relatively small morphological changes between hydrated and doped films, correlating well with their relatively higher operational stability. Overall, AFM data for the doped films corroborate the correlation between device stability and volumetric capacitance, and confirm that irreversible changes in film morphology upon doping are the major cause of instability for these materials.
Grazing incidence wide angle X-ray scattering (GIWAXS) was then used to evaluate the molecular packing of the polymers. As shown in Figure 5b-d and Figure S9 (Supporting Information), the line cuts of p(C-T) and p(N-T) display similar π-stacking scattering intensity both in plane and out of plane ( Figure S10a,b, Supporting Information), confirming the presence of mixed edge-on and face-on crystallite orientations. Conversely, p(C-2T) and p(C-g2T) displayed a strong in-plane (010) peak from π-π scattering, two out-of-plane lamellar scattering peaks (100) and (200), and a still identifiable third-order lamellar peak (300), revealing edge-on packing texture ( Figure S10c,d, Supporting Information). Quantitative fitting of persisting peaks yielded to peak positions and calculated d-spacing. The coherence length (L c ) was determined by the half peak breadths; these data are summarized in Table S3 (Supporting Information). p(C-T) and p(N-T) displayed a shorter π-stacking spacing of around 3.45± 0.02 Å with respect to 3.50 Å for p(C-2T) and p(C-g2T), which could contribute toward the higher electron transport ability observed for p(C-T) and p(N-T). and p(C-g2T), respectively, indicating that EG chains in the bithiophene donor unit disrupt the π-stacking to some degree. In contrast with p(C-T), p(N-T) shows larger L C and slightly shorter π-π stacking distance both in plane and out of plane (010), indicating that the more planar backbone of p(N-T) can enhance its intramolecular and intermolecular interactions. Overall, the amorphous and porous microstructures for p(C-T) and p(N-T) observed by AFM and GIWAX are expected to be the main driving force promoting 3D bulk doping needed for applications in OECT and thermoelectric devices. [29,48,49] On one hand, we postulate that electron charge transfer occurs in an interconnecting network of ordered islands surrounded by porous structure, yielding to the observed high mobility μ. On the other hand, the amorphous and disordered sites provide more free volume, accounting for the high ion transport for p(C-T) and p(N-T). Further investigations will be focused on elucidating the role of the unique porous microstructure and amorphous features to boost further the performance of n-type OECT.

Conclusions
In this study, we designed four novel types of n-type lactonebased glycolated D-A-conjugated polymers. Our design strategy offers the advantage of synthesis via aldol polycondensation, thereby avoiding the use of toxic and environmentally harmful compounds during the polymerization step. The strong electron-withdrawing lactone-building blocks facilitate electron injection and transport, leading to deep-lying LUMOs below −4.0 eV and enabling fabrication of OECTs exhibiting a μC* of 6.7 F cm −1 V −1 s −1 . We used AFM and GIWAXS to have information on the microstructure and molecular packing of the polymer films. p(C-T) and p(N-T) showed porous features with interconnected ordered islands, a morphology that is expected to maximize the electrolyte/polymer interface and promote ion transport. The less-ordered semiconducting polymers exhibited the highest electron mobility, a behavior tentatively ascribed to enhanced rigid backbone-to-backbone contacts observed by GIWAXS. GIWAX data showed that, in addition to a porous morphology, p(C-T) and p(N-T) possess bulk molecular packing (mixed edge-on/face-on texture) and denser π-π contact, resulting in high charge-carrier mobility and further yielding to overall superior OECT performance with respect to the other polymers. Overall, p(C-T) exhibited an electron mobility of 0.069 cm 2 V −1 s −1 and a volumetric capacitance of 97 F cm −3 , resulting in the highest μC* of 6.7 F cm −1 V −1 s −1 and the normalized transconductance g m,nom of 0.80 F cm −3 within the polymer series. p(N-T) showed similar OECT performance with respect to p(C-T), with a comparable μ of 0.059 cm 2 V −1 s −1 , a C* of 73 F cm −3 , a μC* of 6.7 F cm −1 V −1 s −1 , and a g m,norm of 0.72 S cm −1 . As such, we demonstrated that electron-deficient lactonebased polymers are an excellent choice for future n-type OECTs, combining the mixed electronic and ionic conductivities needed for high performance with an environmentally benign polymerization strategy. This is relevant both for sustainable materials' design, but also for envisaged applications in bioelectronics, where limiting the use of toxic metals during synthesis would prevent lengthy purification steps and the risks of dangerous residuals leaching during device operation. [50] We expect that our design route will provide a positive contribution to the field, pushing toward design strategies that are not only high performing but also more sustainable.

Experimental Section
Materials: Reagents for synthesis were purchased from commercial suppliers unless otherwise indicated. Solvents for spectroscopic studies were of spectroscopic grade, purchased from Sigma-Aldrich, and used as received. Synthetic intermediates and final monomers were purified by column chromatography with silica gel (General-Reagent, 200-300 mesh) using dichloromethane, ethyl acetate, and methanol, or their mixture as described in detail in Section S9 (Supporting Information). Polymers were purified by sequential Soxhlet extractions in various solvents; the details are shown in Section 9 (Supporting Information). 1 H and 13 C NMR spectra were recorded in CDCl 3 or 1,1,2,2-tetrachloroethane-d 2 (TCE-d 2 ) with a 400 or 500 MHz Bruker Avance III spectrometer.
UV-Vis-NIR Absorption Spectroscopy: UV-vis-NIR absorption spectra were measured with an Agilent Technologies (Cary 5000/6000i) Cary Win UV-vis-NIR spectrophotometer. Solution UV-vis-NIR spectra were recorded in CHCl 3 at a concentration of 1.0 × 10 −5 m in a quartz cuvette with a 1 cm path length. Polymer samples in film form were prepared by spin-coating 6 mg mL −1 in chloroform:o-dichlorobenzene (10:1) onto ITO-coated glass substrates. Before the spin coating, the ITO slides were cleaned by subsequent sonication steps in soapy water, ethanol, and isopropyl alcohol for 15 min each, followed by a drying step using a nitrogen gun.
Cyclic Voltammetry: CV measurements were carried out using a Shanghai Chenhua Instrument CHI760E potentiostat with a threeelectrode setup. An Ag/AgCl (saturated KCl aqueous solution) was used as the reference electrode, a Pt wire as the counter electrode, and a polymer film spin-coated on ITO-coated glass as the working electrode. The polymer films were prepared analogously to those used for UV-vis-NIR absorption spectroscopy. Measurements were carried out either in an organic-or aqueous-based electrolyte: the organic electrolyte consisted of a 0.1 m tetrabutylammonium hexafluorophosphate (TBAPF 6 ) solution in anhydrous acetonitrile. For CV measurements in the organic electrolyte, an Ag/AgCl (saturated KCl aqueous solution) was used as the reference electrode, a Pt wire as the counter electrode, and the ferrocene/ferrocenium couple (Fc/Fc + ) was used as an external standard to calculate the E LUMO . Aqueous CV measurements were instead conducted in a 0.1 m aqueous sodium chloride solution.
Spectro-Electrochemistry: Spectro-electrochemical measurements were performed by fitting the CV setup inside a quartz cuvette in the UV-vis-NIR absorption spectrophotometer. Polymer film spin-coated on ITO-coated glass were used for the measurement. The polymers were spin-casted onto the substrates from polymer solutions in chloroform:odichlorobenzene (10:1) with a concentration of 6 mg mL −1 . Potentials were applied for 15 s prior to recording the UV-vis-NIR absorption spectrum at each voltage to ensure stabilization of the optical trace. Potential steps of 0.1 V were then used to evaluate the electrochromic behavior of the polymers. All spectro-electrochemical measurements were conducted using a 0.1 m aqueous sodium chloride solution as the supporting electrolyte.
Atomic Force Microscopy: AFM measurements were performed in tapping mode, using a Bruker Dimension Icon atomic force microscope with a silicon tip on nitride lever (cantilever: T = 650 nm, L = 115 μm, W = 25 μm, f 0 = 70 kHz, k = 0.4 N m −1 ). Pristine polymer thin film samples were prepared analogously to those used for UV-vis-NIR absorption spectroscopy. The polymers were spin-casted onto the substrates from polymer solutions in chloroform:o-dichlorobenzene (10:1) with a concentration of 6 mg mL −1 . Hydrated polymer samples were prepared by immersing the pristine polymer films into a 0.1 m aqueous sodium chloride solution for 30 min followed by rinsing with deionized water and then drying under a stream of nitrogen prior to recording the AFM images. AFM images of electrochemically doped polymer films were subsequently obtained by the electrochemical workstation applied at a constant voltage of −0.4 V for 15 min for the pristine polymer. Films were then washed with deionized water and dried with nitrogen before recording their images.
GIWAXS: Samples for X-ray scattering were prepared analogously to those used for UV-vis-NIR absorption spectroscopy albeit using p-doped silicon wafers as the substrates. An incidence angle of 0.18 ○ and a photon energy of 8 keV were used to record the scattering patterns. The 2D GIWAXS patterns were collected from films that were spincoated as previously described for absorption spectroscopy on p-doped Si wafer substrates with a resistivity of 0.001-30 Ω cm −1 .
Device Fabrication: Electrode contacts were prepared by subsequently evaporating titanium (as an adhesion layer) and gold on glass using a custom-made mask. The channel was L = 20 μm and width W = 500 μm or (only for device stability shown in Figure S7 in the Supporting Information) interdigitated electrodes having a channel length of L = 20 μm and a width of W = 39 000 μm (39 parallel channels having W = 1000 μm). Solutions of the conjugated polymers (concentration = 6 mg mL −1 ) in chloroform:o-dichlorobenzene (10:1) were spin-coated (2500 rpm for 60 s) on the electrodes and cured at 120 °C for 10 min. Film thicknesses were measured with a Tencor-P15 stylus profilometer. A solution of 0.1 m NaCl in water was used as the electrolyte and a silver pellet as the gate. Device characterization was performed in air at room temperature using a Keithley 4200A-SCS parameter analyzer. Data analysis was performed using OriginPro 2021, and transconductance lines were smoothed by adjacent averaging to attenuate instrument and environmental noise. Impedance Spectroscopy: Electrodes of different sizes were built by evaporating titanium (as an adhesion layer) and gold on glass using a custom shadow mask. The mask was made by patterning cellulose filter paper Grade 50 from Whatman (GE Life Sciences; now Danaher Corp., Washington, DC, USA) using a VLS 2.30 CO 2 laser cutter/ engraver (10.6 μm wavelength; Universal Laser Systems, Scottsdale, AZ, USA). Polymers were spin-coated on the electrodes as described for device fabrication. Impedance data were acquired using a BioLogic multichannel potentiostat. The polymers on gold were used as working electrode, Ag/AgCl 3 m KCl as reference electrode, and a platinum plate as a counter electrode. A BioLogic EC-Lab Software was used to perform the fitting of impedance data.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.