Universality in Solvent-Dependent Conductivity of Conjugated Thiophene-Based Copolymers with Expanded Core Monomer Designs

Introduction

Conjugated polymers (CPs) are promising materials for thermoelectric devices, sensors, and printed electronics due to their electrical and mechanical properties. [1][2][3][4] However, CPs are typically semiconducting or insulating in nature and must be doped to attain useful electrical conductivities.

Therefore, molecular dopants are used to either oxidize (p-doping) or reduce (n-doping) CPs to generate charged carriers, resulting in greater conductivity. [4][5][6] P-type doping is more prevalent in organic semiconductors due to the greater stability of p-type dopants in ambient conditions and electron-donating π-systems of common CPs.

One of the most extensively studied p-type systems is that of poly(3-hexylthiophene) (P3HT) and 2,3,5,6-tetrafluoro-7,7,8,8-tetracyanoquinodimethane (F4TCNQ). F4TCNQ is particularly well suited for p-doping due to its deep LUMO level (-5.24 eV), which aligns well with the HOMO level of P3HT (-5.2 eV). 6,7 This energy level alignment allows for facile electron transfer from the polymer to the dopant, resulting in the formation of a charged complex. However, the regioregularity of P3HT significantly impacts the doping process and resultant conductivity. Hynynen et al. showed that P3HT with a regioregularity of 28% had a very low degree of doping (as evidenced by the lack of anionic F4TCNQ formation) while P3HT samples with regioregularity above 84% could be readily doped with F4TCNQ. 8 Guchait et al. found that charge conductivity in regioregular P3HT was roughly 50 times higher than that of regiorandom P3HT due to the increased doping level associated with regioregular P3HT. 9 The change in electrical performance with higher degrees of regioregularity is not only due to the increased doping levels, but also to enhanced microstructural organization. For instance, Noriega et al. studied doping of blends of regiorandom P3HT with regioregular P3HT nanofibrils and determined that within the heterogenous microstructure adopted by the blends, charge carriers were confined to the ordered nanofiber regions due to energetic offsets at order/disorder interfaces. 10 Additionally, Pingel et al. found that with P3HT, the macroscopic charge mobility was two orders-of-magnitude lower than the local mobility due to heterogeneity in the microscale organization. 11 The macroscopic mobility was reduced (compared to the local mobility) due to the disordered regions in the system. 11 These studies highlight that regioregular P3HT is superior compared to regiorandom P3HT in that higher doping levels, greater microstructural organization, and higher electrical performance can be achieved.

To improve their solubility, conjugated polymers are typically decorated with aliphatic side chains; however, these side chains may inhibit polymer-dopant interactions due to increased steric demand. Therefore, it has become conceptually attractive to alter polymer design to increase the "spacing length" between alkyl chains in the backbone of CPs to allow for increased polymerdopant interactions. For example, Li et al. studied the incorporation of a thienothiophene spacer unit in a quaterthiophene-based polymer and found that separating side chains along the backbone improved dopant integration. 12 This in turn improved ordering and π-stacking while also increasing doping efficiency and conductivity. 12 Similarly, Nam et al. showed that introducing thiophene spacing units in diketopyrrolopyrrole (DPP) copolymers enhanced intercalation of the dopant into the lamellae, thereby enhancing doping efficiency. 13 These works highlight how polymer design can be strategically tuned to improve doping efficiency and enhance conductivity, though it should be appreciated that decreasing the number density of side chains alters solubility.

In addition to altering polymer design, the method of dopant integration is known to have a significant impact on doping efficiency and, therefore, conductivity. Specifically, solution and sequential doping have been extensively studied, especially with the polymer-dopant system of P3HT and F4TCNQ. In the case of solution doping, polymer and dopant solutions are directly mixed and then cast to form doped polymer thin films. 6,7 This approach typically results in a significant number of aggregates due to the formation of charged complexes that are too polar to be dissolved in the same solvent that is used to dissolve the neutral polymer. 7,14,15 These ill-defined aggregate structures result in heterogenous microstructures that restrict charge mobility and decrease film conductivity. For example, Jacobs et al. showed that once solutions containing aggregated assemblies generated via solution doping were cast into films, there were highly conductive crystalline domains; however, these domains were not connected, thus the conductivity of the film was diminished. 7 With sequential doping, the polymer film is created first by casting, and then the dopant is added by exposing the film to a dopant solution that uses a marginal solvent for the polymer that swells, but does not dissolve the polymer. 7,14 This method avoids the formation of aggregates and minimizes disruption of the microscale organization of the pristine polymer film.

For instance, Scholes et al. showed that with sequential doping of P3HT with F4TCNQ, the size and orientation of the crystallites were unchanged in comparison to the as-cast P3HT films. 15 Although the sequential doping method prevents aggregation of the copolymer, it relies heavily on finding an appropriate solvent. 16,17 Yoon et al. showed that the degree of solubility of the CP with respect to the dopant solvent impacts resultant electrical conductivity in sequentially doped films. 16 Specifically, greater charge carrier generation was seen with dopant solvents that offer moderate solubility for the polymer and good wettability, which enhanced dopant diffusion into the film. 16 Overall, most studies that focus on P3HT and F4TCNQ have defined a heuristic that solution doping is absolutely inferior to sequential doping, despite the convenience afforded by this method. 7,14,15 Herein, we report the synthesis and characterization of thiophene-based copolymers in which the π-conjugated repeat unit is systematically altered to increase the spacing length between alkyl side chains. We examine the impact that copolymer design has on optoelectronic properties, solubility, and electrical performance (i.e., conductivity) as functions of doping level and method.

We find that the heuristic defined for the benchmark P3HT/F4TCNQ system is not universal; however, the dependence on solvent quality with sequential doping appears to follow a consistent pattern of behavior across the spectrum of thiophene-based copolymers studied.

Experimental Section

Materials. 2,5-bis(trimethylstannyl)thiophene (97%, Sigma), 2,5-bis(trimethylstannyl)thieno [3,2b]thiophene (97%, Sigma), 5,5'-dibromo-2,2'-bithiophene (98%, TCI), 5,5''-dibromo-2,2':5',2''terthiophene (97%, Sigma), 2-bromo-3-hexylthiophene (97%, Sigma), 3-hexylthiophene-2boronic acid pinacol ester (95%, Sigma), 3-hexyl thiophene (99%, Sigma), Pd(PPh3)4 (99%, Sigma), Pd(OAc)2 (98%, Sigma), potassium carbonate (99.7%, Fisher), neodecanoic acid (NDA, Sigma), magnesium sulfate (MgSO4, 99.5%, Alfa Aesar), 2,3,5,6-Tetrafluoro-7,7,8,8tetracyanoquinodimethane (F4TCNQ, 99%, Ossila), lithium iodide (LiI, 99.999%, ultra dry, Alfa Aesar), toluene (99.85%, extra dry, Thermo Scientific), diethyl ether (anhydrous, Fisher), ethyl acetate (99%, Fisher), hexanes (98.5%, Fisher), dimethyl formamide (DMF, 99.8%, Thermo Scientific), N,N-dimethylacetamide (DMAc, 99.5%, extra dry, Thermo Scientific), methanol (99%, Fisher), acetone (99%, Fisher), chloroform (98%, Fisher), chlorobenzene (99.8%, anhydrous, Sigma), acetonitrile (99%, Fisher) and graphite conductive adhesive (aqueous based, Alfa Aesar) were used as received. N-bromosuccinimide (NBS, 99%, Sigma) was recrystallized in water and vacuum dried prior to use. Synthetic Procedures. 2,5-dibromo-3-hexyl thiophene was synthesized based on reported procedures, 18,19 as summarized in the Supporting Information. Comonomers featuring different spacing units (thiophene, 2,2'-bithiophene, 2,2':5,2"-terthiophene, and thieno [3,2-b]thiophene) between alkyl side chains were synthesized via Suzuki-Miyaura and Stille cross-couplings, and copolymers were synthesized via direct arylation polymerization (DArP). These procedures are discussed in detail in the Supporting Information. Macromolecular Characterization. 1 H and 13 C NMR spectra of each monomer and the 1 H NMR spectrum of each copolymer were acquired using a Varian VNMRS 500 MHz NMR. Gel permeation chromatography (GPC) was used to determine the number-average molecular weight (Mn) and dispersity (Ð) of each copolymer based on universal analysis. An Agilent 1260 Infinity II system was used for these measurements with a mobile phase of THF at 25 °C and a flow rate of 1 mL/min. Solutions for GPC measurements were made at a concentration of 3-5 mg/mL in THF and filtered through a 0.2 μm PTFE filter prior to injection. Thin Film Preparations. Solution Doping. Solutions of 0, 6, 10, 17, and 23 mol% F4TCNQ (relative to the total number of moles of thiophene repeat units) were made by mixing premade solutions of the copolymer and F4TCNQ in chlorobenzene. These solutions were prepared so that after mixing the final copolymer concentration was 2.0 mg/mL. The resultant polymer/dopant solution was heated to 70 °C prior to film deposition. Glass slides (1 in × 1 in) were cleaned with KimWipes TM , then washed with acetone and methanol, and dried with a stream of dry N2 gas. Two layers of polymer/dopant were spin-cast onto the cleaned glass substrate. First, 200 μL of the polymer/dopant solution were deposited statically and then spun at 1000 rpm for 40 s. Immediately after, 200 μL of the same polymer/dopant solution was added dynamically under the same conditions. The resultant polymer film was allowed to dry for at least 10 minutes. Sequential Doping. For sequential doping, the copolymer was dissolved in chlorobenzene (2.0 mg/mL) at 70 °C while F4TCNQ was dissolved in the indicated dopant solvent (1.0 mg/mL) at room temperature. Clean glass slides were prepared as described above. A copolymer thin film was created on the glass substrate by spin casting. 200 μL of the polymer solution were deposited statically and then spun at 1000 rpm for 40s. Then, another 200 μL were deposited dynamically at the same conditions (1000 rpm for 40 s). Immediately after film deposition, the dopant was spin cast onto the polymer thin film following the same two-step procedure. The resulting doped copolymer films were allowed to dry for at least 10 minutes.

Thin Film Characterization. AFM Measurements. An Asylum Atomic Force Microscope was used in tapping mode to probe the topography of thin films. AC200TS tips with a spring constant of 9 N/m and resonance frequency of 150 kHz were used to acquire images of 20 μm × 20 μm. To measure the thickness of each film, a scratch test was performed. A razor was used to scratch the film, thereby displacing the polymer film and exposing the glass substrate. Images were then acquired at the scratch interface, and film thickness was taken as the height difference between the polymer film and glass substrate.

Conductivity Measurements. Conductive graphite adhesive was used to add contact points at each corner of the solution doped and sequentially doped thin films. A Keithley 2450 Sourcemeter was used to perform current-voltage (I-V) sweeps from 0 to 10 V. Voltage was applied at the contact points across one edge and the current was measured along the contact points on the opposite edge. This was repeated along each of the four edges of the film in accordance with the van der Pauw method [20][21][22] and a total of 3 replicate samples were measured for each reported conductivity value.

The sheet resistance, Rs, was calculated using the slope of the I-V curve in the initial linear region (0-2 V). Rs, along with the film thicknesses, t, obtained from AFM measurements, were used to calculate the film resistivity, ρ. Subsequently, conductivity, σ, was calculated using: σ = ρ -1 .

Absorbance Measurements. Thin Film Measurements. UV-Vis-NIR spectra were acquired for as cast and doped thin films using a Cary 5000 UV-Vis-NIR Spectrophotometer in the wavelength range of 350 to 2000 nm. A clean glass slide was used as the background spectrum to baseline correct the spectra acquired for thin films. Each spectrum was then normalized by film thickness, as practiced by Li et al. 12 Solution Measurements. To determine the effective conjugation length (ECL) of each copolymer, samples of different molecular weight were collected. This was done by recovering the copolymers resulting from Soxhlet extractions that used acetone and hexanes in the overall process of purifying each copolymer. The solids obtained from these fractions were dried and the degree of polymerization (Xn) and number of thiophenes per chain was calculated from the Mn determined via GPC using universal calibration analysis. The UV-Vis spectra were acquired for these samples in chloroform across a wavelength range of 300-600 nm using a Thermo Scientific Evolution 300 spectrophotometer. The optical band gap was calculated using Tauc plot analysis. 23 The optical band gap of the monomers and purified copolymers were measured in the same manner.

Hansen Solubility Parameters. The Hansen solubility parameters (HSPs) of each polymer was determined as described by Díaz de los Ríos and Ramos. 24 In brief, 1 mg of the polymer was added to 1 mL of a particular solvent (or solvent blend). Twenty-three (23) different solvents were used, and the homogeneity of the solution was evaluated based on visual observations. Each solution was given a score of 1 (soluble) or 0 (not soluble) and entered in the Excel spreadsheet accessible from Díaz del los Ríos and Ramos' work. 24 For those solutions deemed partially soluble at room temperature, the sample was heated gently. If the copolymer subsequently dissolved and the solution remained homogenous for more than 1 h, it was considered to be soluble and given a score of 1. Using the embedded formulae and data solver in Excel, the HSPs of each copolymer were calculated based on these observations.

Results and Discussion

Comonomer Synthesis and Characterization. A series of thiophene-based copolymers were used to investigate how repeat unit design alters key properties, such as solubility and conductivity of chemically-doped thin films. Thiophene comonomers were synthesized via Suzuki-Miyaura (M3, M4) and Stille (M5, M6) cross-couplings, as depicted in Scheme 1. The structure of each comonomer was verified via 1 H NMR and 13 C NMR spectroscopies. (See Figures S1 -S8.) It should be noted that each comonomer is flanked with hexyl thiophenes to improve solubility of the monomer and the resultant copolymer. 25 Thiophene (M3), 2,2'-bithiophene (M5), and 2,2':5':2"-terthiophene (M6) were chosen as "cores" to compare how decreasing the steric demand along the backbone by increasing the spacing between alkyl side chains impacts copolymer solubility, dopant integration, and film conductivity. The fused ring core of thieno[3:2,b]thiophene (M4) was selected to provide insight into how increasing monomer rigidity and planarity affects solubility and conductivity. To first evaluate how comonomer design impacts optoelectronic properties, the absorbance spectrum of each comonomer was measured, as shown in Figure 1, and the corresponding optical band gap was calculated via Tauc plot analysis. 23 Table 1 summarizes the absorbance maximum, λmax, and optical band gap, E gap opt , of each of the thiophene-based comonomers in solution. The λmax values, which correspond to the π-π* transition, range from 339-405 nm and show a red-shift as the core size of the comonomer increases. This red-shift is indicative of increased electron-richness in the comonomer with the addition of more thiophene units. Additionally, the band gap decreases as the core size increases due to increased conjugation.

These two trends are seen most clearly across the series of M3 to M5 to M6. On the other hand, when the core is changed from thiophene (M3) to thieno[3:2,b]thiophene (M4), the changes are less pronounced, as the red-shift is only 9 nm and E gap opt decreases by only 0.07 eV. These comparisons suggest that adding a few thiophene units produces a greater impact on optoelectronic properties than increasing the planarity/rigidity through the introduction of a fused-ring structure in the backbone. However, as described later, the addition of thiophene rings to the core decreases the solubility of the resulting copolymer. Copolymer Synthesis and Characterization. As shown in Scheme 2, each of the comonomers M3-M6 were copolymerized with 2,5-dibromo-3-hexylthiophene via direct arylation polymerization (DArP) using conditions typically reported [26][27][28][29] for thiophene-based conjugated polymers. The AA+BB nature of this step-growth polymerization along with the lack of symmetry in 2,5-dibromo-3-hexylthiophene results in regio-irregular copolymers; however, due to the symmetry of comonomers M3-M6, the regioregularity of each copolymer (P3-P6) can be assumed to be 50%. The structures of the thiophene-based copolymers were confirmed via 1 H NMR, as shown in Figures S11-S14. For comparisons to the thiophene-based copolymers P3-P6, DArP was used to synthesize P3HT via AB and AA+BB type step-growth polymerizations. P3HT (AB) was chosen because it is a benchmark for thiophene-based polymers and P3HT (AA+BB) was used for comparison due to the regio-irregularity of the core-expanded copolymers. From 1 H NMR it was determined that the P3HT synthesized by an AB polymerization was 93% regioregular while the P3HT synthesized via the AA+BB copolymerization was 57% regioregular. The molecular weight of each polymer was determined by GPC and the traces are shown in Figure S17. As shown in Table 2, the number-average molecular weight (Mn) of the copolymers, as determined via universal calibration analysis, range from 5.0-11.2 kg/mol. Therefore, to enable comparisons between copolymers of different designs, the effective conjugation length (ECL) of each of the thiophene-based copolymers was assessed. The ECL, which is identified as the chain size at which the addition of repeat units no longer alters optoelectronic properties, 30 was determined based on the number of thiophene units per chain, rather than the degree of polymerization (Xn) because the copolymers studied here have different repeat unit structures. For example, while the Mn of P3 (11.2 kg/mol) equates to a Xn of 19 based on repeat unit structure, this copolymer has 76 thienyl rings per chain on average. Also, for this analysis, the thieno[3:2,b]thiophene subunit in P4 is counted as one thiophene unit, as it is one conjugated unit.

UV-Vis measurements presented in Figure S18 show that the optical band gap of each polymer becomes independent of chain size above ~20 thiophene units per chain, which sets the ECL at 20 thiophene units. This result is in agreement with characterization of the ECL of P3HT and other polythiophene derivatives by de Oliveira et al. 31 and Nakanishi et al. 32 Therefore, we posit that any differences in optoelectronic and/or conductive properties observed in the subsequent studies are attributed to structural design, rather than a function of molecular weight differences, as each polymer contains more than 20 thiophene units per chain. Mn (kg/mol) a Ð Xn Thiophene units b Solution λmax (nm) c Solution 𝐄 𝐠𝐚𝐩 𝐨𝐩𝐭 (eV) d Thin Film λmax (nm) c Thin Film 𝐄 𝐠𝐚𝐩 𝐨𝐩𝐭 (eV) d P3 11.2 2.2 19 76 459 2.38 542 1.99 P4 8.4 2.4 13 52 453 2.40 516 2.03 P5 5.0 1.8 8 40 465 2.36 545 1.94 P6 7.5 1.9 10 60 471 2.33 533 2.02 P3HT (AB) 9.0 2.2 54 54 435 2.45 506 2.07 P3HT (AA+BB) 9.7 2.3 58 58 425 2.54 436 2.45 a M n based universal calibration analysis. b Number of thiophene units per chain calculated from X n and the number of conjugated thiophene units in the repeat unit. c λ max values determined from UV-Vis measurements using CHCl 3 solutions (< 1 mg/mL) and thin films cast from chlorobenzene (at 2 mg/mL). d 𝐸 𝑔𝑎𝑝 𝑜𝑝𝑡 values were determined from Tauc plot analyses of UV-Vis spectra measured from solutions and thin films.

The absorbance maxima and optical band gaps of the copolymers in solution (CHCl3) and thin film form were determined from UV-Vis measurements, and the results are summarized in

Table 2. The solution and thin film spectra are shown in Figures S19-S20, respectively. Results are consistent with well-known behaviors in that a red-shift in λmax and a corresponding decrease in E gap opt is observed from solution to solid-state due to the organization in the solid-state brought on by π-stacking interactions. More specific behaviors can be attributed to chain microstructure and monomer design. For example, when comparing photophysical characteristics of P3HT (AA+BB) to P3HT (AB) measured in solution, there is a significant increase in the E gap opt (0.11 eV)

and a blue-shift in the λmax of 10 nm. These differences in optoelectronic properties are related to the degree of regioregularity of the two P3HTs. The lower degree of regioregularity of P3HT (AA+BB) leads to greater rotational freedom along the backbone, thereby decreasing the persistence length and increasing the resultant E gap opt . 33 In addition, the decrease in E gap opt for the thin film compared to the solution of P3HT (AA+BB) is not as significant as that seen with P3HT (AB). This smaller change in E gap opt is attributed to poor ordering in the film due to more rotational degrees of freedom caused by the regio-irregularity of P3HT (AA+BB).

Despite the regio-irregularity inherent in the series of "core expanded" copolymers that arises from the synthetic design (symmetrical + asymmetrical monomer), there is no obvious detrimental impact seen in the solution or thin film measurements like those observed for P3HT (AA+BB). Therefore, it is believed that the spacing units minimize the impact regio-irregularity has on ordering and π-stacking interactions. Furthermore, within the series of copolymers, a slight red-shift and decrease in E gap opt is observed in solution as thiophenyl moieties are added to the repeat unit design (P3 to P5 to P6) and side chain densities are reduced. These changes are attributed to decreasing the steric demand along the backbone with the incorporation of the spacing units, thereby allowing for more π- interactions. However, measurements of P4 in solution reveal a small blue-shift and an increase in E gap opt compared to P3, P5, and P6. According to Danielsen et al., monomers that impart a non-zero deflection angle along the backbone decreases the persistence length of conjugated polymers. 34 Thus, despite the planarity of the fused ring structure, it is believed that that the persistence length of P4 is reduced compared to the other copolymers due to the introduction of the thieno[3:2,b]thiophene core, which alters the deflection angle compared to P3HTs and the other copolymers. This decrease in persistence length would cause a subsequent increase in the E gap opt of P4 compared to the other copolymers. It is also noted that for P6, the decrease in E gap opt from solution to solid-state is not as pronounced as it is with the other copolymers (P3-P5) and P3HT (AB). The smaller change in E gap opt is attributed to poor ordering in the film, which may be due to aggregation or increased rotational freedom resulting from the increased spacing length between alkyl side chains (by using terthiophene as the central core). Furthermore, it is observed that in the solid state, E gap opt is higher and λmax is lower for P6 in comparison to P5.

These findings suggest that there may be a limit on the size of the spacing unit before it is no longer advantageous for use in thin films and indicate that there are complex trade-offs involving spacing length, conformational flexibility, and solubility that impact properties and performance of the copolymers.

To gain insight into how spacing length between side chains and planarity impacts polymer solubility, the Hansen solubility parameters (HSPs) of each copolymer were evaluated following the method described by Díaz de los Ríos and Ramos. 24 This method relies on assessing the solubility of the polymer in various solvents for which values of the dispersive, polar, and hydrogen-bonding contributions, δD, δP, δH, respectively, are known. The HSPs determined for the thiophene-based copolymers are summarized in Table 3, and the underlying assessments of their solubility in 23 different solvents/solvent blends are listed in Table S1. The tabulated results show that for the two P3HTs and all of the copolymers the main contributor to the solubility is dispersive interactions (δD), since the polymers are non-polar. We also note that the HSPs defined for P3HT

(AB) are in agreement with those reported by Yoon et al. 16 and Machui et al. 35 There are clear differences in the HSPs between P3HT (AB) and P3HT (AA+BB), reflecting the fact that regioregularity impacts solubilizing behavior. Specifically, P3HT (AA+BB), which is 57% regioregular, has lower HSP energies, and the values are consistent with those reported by King et al. 36 This difference can be attributed to P3HT (AB) having stronger interchain π-stacking interactions due to its higher regioregularity, which increases the cohesive energy density of the polymer. Table 3 also shows that the values of δD, δP, and δH increase with an increase in the spacing length from one (P3) to two (P5) to three (P6) thiophene units, although the changes are very small from P3 to P5. We posit that the increased spacing length that results from expanding the comonomer core facilitates interchain π-stacking interactions, thus increasing the dispersive energy, δD. The small increases in δP and δH, which reflect an increase in polarity and hydrogen bonding capacity, are most likely the result of the decreased density of alkyl side chains along the backbone. Finally, we note that the increase in HSPs of the copolymers compared to P3HT (AB) is more drastic with the addition of thieno[3:2,b]thiophene (P4) than with thiophene (P3) or bithiophene (P5). This pattern of behavior indicates that the fused ring core alters the solubility more significantly than its non-fused ring counterparts. Although the HSPs of each copolymer offer significant information on polymer-solvent interaction energies, it does not directly indicate which polymer is considered more soluble than others. However, it can be inferred that the solubilities of P3, P4, P5, and P6 are decreased compared to P3HT (AB), as the solubility scores listed in Table S1 show that these core expanded copolymers dissolve in fewer solvents (11, 9, 10, and 7 solvents, respectively) compared to P3HT (AB) (which dissolves in 13 solvents). This decrease in solubility is consistent with expectations, as increasing the number of thiophene units in the comonomer core reduces the density of alkyl side chain in the repeat unit, thereby reducing their effectiveness in promoting solubility of the conjugated polymer. Solution Doping of Thiophene-based Copolymers. F4TCNQ was used as the dopant due to its compatibility with thiophene-based polymers, especially P3HT. 15,37,38 In these solution doping studies, each of the copolymers was doped at levels ranging from 0 to 23 mol% F4TCNQ, which is based on the moles of thiophene in the polymer solution. Adopting this basis provides consistency across all copolymers despite differences in their repeat unit design. One drawback frequently noted with solution doping is the appearance of aggregates that form upon mixing due to the creation of charged complexes. 39 Therefore, because polymer thin films deposited from solution inherit structural characteristics from their parent solution, to assess the presence of aggregation in the solid-state, the solution doped thin films were imaged by atomic force microscopy (AFM). S2. There are several salient points to highlight: First, there are significant differences in the film topography of P3HT (AB) and P3HT (AA+BB) (Figures S25 and S26, respectively). As documented in Table S2, P3HT (AA+BB) has a higher surface roughness compared to its regioregular counterpart, P3HT (AB), at all doping levels. In addition, the large roughness measured for P3HT (AA+BB) at 0 mol% doping indicates that the regio-irregular nature of the chain microstructure disrupts the organization of the polymer film, thus decreasing the effectiveness of π-stacking interactions that drive interchain organization. 40,41 The P3HT (AA+BB) film cast from a solution of 6 mol% F4TCNQ also displays a high surface roughness, indicating the incorporation of F4TCNQ causes aggregation, even at low doping levels. Table S2 also shows that for all of the copolymers (P3-P6), the surface roughness remains relatively constant and low (generally ≤ 4 nm) at doping levels ranging from 0 to 17 mol%; however, at a doping level of 23 mol%, there is a significant increase in the topographical roughness, which likely indicates the prevalence of aggregates. The most substantial changes in roughness at high doping levels are seen for the expanded designs of P4 (14.7 nm), P5 (13.3 nm), and P6 (25.1 nm). This is attributed to increased aggregation, which is likely exacerbated by the lower solubility, as reflected in the HSPs of these copolymers. Additionally, the increased prevalence of aggregates that are observable in the topological images, especially for P5 and P6 (compare Figure 2c and 2i and Figure 2d and 2j, respectively), could be due to an increase in the amount or size of charged complexes formed during the solution doping process. This idea will be discussed later in more detail. The influence of copolymer design and doping level on the electrical performance of solution doped thin films was investigated by measuring the conductivity via the Van der Pauw method. 20,22 The sheet resistance, which was calculated from the measured I-V curves, was used along with the film thickness to calculate the conductivity. (Film thicknesses measured via AFM scratch test are summarized in Table S3.) It should be appreciated that the conductivity values measured here are lower than some values reported for thiophene-based polymers (particularly for P3HTs 42,43 ) due to the low film thicknesses (10-25 nm) and relatively large sample sizes (1 in × 1 in). As shown in Figure 3, there are drastic differences in thin film conductivity based on repeat unit structure. First, films made from P3HT (AA+BB) have very low conductivities at all doping levels. This is attributed to poor organization due to less efficient π-stacking that derives from the irregularity in regiochemistry, which reduces electron delocalization. On the other hand, P3HT

(AB) has an appreciably higher conductivity due to its high regioregularity, which enhances interchain organization and facilitates electron delocalization. The regio-irregularity of copolymers P3-P6 does not appear to impact the conductivity of their thin films, which again indicates that the spacing units allow the dopant to effectively access the copolymer backbone, which will be discussed in more detail later. For each copolymer (and P3HT (AB)) the conductivity increases by a few orders of magnitude as the doping level is increased. However, for P3 and P4 it appears that within experimental uncertainty, the conductivity plateaus after 10 mol%. This behavior is attributed to the formation of charged species and will be discussed later in more detail. Consistent with work by Li et al., who showed that decreasing the density of hexadecyl side chains through an alternating pattern of bithiophene and thienothiophene spacing units in the backbone enhanced the conductivity of polythiophene-based copolymers (relative to analogous copolymers without the spacing units), 12 an increase in conductivity is seen when a spacer unit is used. As the spacing length between alkyl chains in the repeat unit is increased from one (P3) to two (P5) to three (P6) thiophene units, the conductivity increases significantly, by orders-of-magnitude. The reason for this enhanced conductivity is inherent to the complex interplay between chain design, efficient doping, and structure at different length scales. To begin unraveling this behavior, it should be noted that the increase in conductivity for the copolymers follows the trend shown in

Table 2 of decreasing E gap opt as spacing length is increased from P3 (2.38 eV) to P5 (2.36 eV) to P6 (2.33 eV). with it having the highest optical band gap (2.40 eV). This suggests that the incorporation of thieno[3:2,b]thiophene in P4 results in a conformational change (i.e., non-zero deflection angle)

that in turn decreases the persistence length compared to the other copolymers, which is known to decrease electron delocalization. 34 Therefore, a single characteristic property, such as E gap opt , cannot fully describe a macroscopic property such as electrical conductivity because it is also influenced by polymer structure and organization, efficient polymer-dopant interactions, and transport at multiple length scales. As previously discussed, doping relies on electron transfer, which requires the dopant and the conjugated backbone to be in close proximity. Because the anionic form of F4TCNQ, F4TCNQ -, is characteristic of efficient doping, it is useful to monitor this species to determine the relative extent of doping. To enable the identification of F4TCNQ and F4TCNQ -in a doped sample, a standard consisting of F4TCNQ -only was made via a LiI reduction which is described in detail in the Supporting Information and based on a procedure similar to that reported by Kiefer et al. 44 FTIR and solution-based UV-Vis measurements were used to confirm the complete conversion of F4TCNQ to give the F4TCNQ -standard (Figure S27-S28). Thin films of F4TCNQ and its anionic counterpart, F4TCNQ -, were made by drop-casting, and characteristic peak positions were assessed by UV-Vis-NIR measurements. In the solid-state, F4TCNQ (neutral form) has a characteristic absorbance peak in the range of 348 -444 nm while F4TCNQ -(anionic form) has one peak in the range of 475 -570 nm and a second that appears at 728 -955 nm, as shown in Figure S29. These peaks allow the amount of F4TCNQ -in doped thin films to be monitored as a function of doping level; however, because the F4TCNQ -peak from 475 -570 nm overlaps with the π-π* peak of the copolymers, the local peak maximum at 868 nm associated with the second mode was used.

As an example, the UV-Vis-NIR spectra of P6 doped with F4TCNQ at various doping levels is shown in Figure 4. It should be noted that because the spectra are normalized by film thickness, the normalized absorbance directly correlates with concentration. Therefore, and as expected, there is a significant increase in the presence of F4TCNQ -with increasing doping levels, as evidenced by the increase in absorbance band at 728 -955 nm. Additionally, with increasing doping levels, an increase in the vibrational mode assigned to the neutral form of F4TCNQ (observed from 348 to 444 nm) is also seen, which indicates that not all of the dopant is converted to its anionic form during the solution doping process. This saturation in doping is consistent with studies by Fontana et al. 14 and Duong et al. 39 on P3HT doped with F4TCNQ. The absorbance spectra measured at each doping level was normalized by film thickness, which ranged from 10-13 nm for these P6 films. Regions highlighted in blue correspond to the mode associated with neutral F4TCNQ while regions shaded in red correspond to the anionic form, F4TCNQ -.

Analogous UV-Vis-NIR measurements were performed on all thin films prepared by the solution doping method, and the spectra are shown in Figures S30-S34. From these results, the normalized absorbance (at 868 nm) of the anionic dopant is extracted and correlated with the measured conductivity of the doped thin films, as seen in Figure 5. As expected and shown in Figure 5a, there are low concentrations of F4TCNQ -in films made from P3HT (AA+BB) at all doping levels. Again, this is attributed to the irregular micro-organization which hinders effective doping, consistent with Hynynen et al. 8 The spectrum for P3HT (AB) (Figure 5b) and each of the copolymers (Figures 5c-5f) show that as the doping level increases, both the concentration of F4TCNQ -and the conductivity increase. Although the axes of each plot in Figure 5 were scaled consistently to facilitate comparison, the increase in normalized F4TCNQ -absorbance as a function of doping level is most apparent with the expanded designs of P5 and P6. Nevertheless, the results obtained from the sets of solution doped thin films show that there is a clear increase in the amount of F4TCNQ -as follows: P4 ≤ P3 < P5 < P6. This trend generally correlates with the increase in spacing length between the alkyl side chains on the comonomer used to create each copolymer. Specifically, as the spacing length increases from one (P3) to two (P5) to three (P6) thiophene units, the steric demand is decreased, which allows F4TCNQ to access the polymeric backbone, enabling electron transfer (i.e., doping). In turn, this results in the formation of F4TCNQ -, which enhances conductivity. It is also observed that the normalized absorbance of F4TCNQ -for P4 is very similar to that of P3, despite the conductivity of solution doped films of P4 being approximately one order-of-magnitude lower than solution doped films made from P3.

This suggests that the difference in conductivity is significantly impacted by repeat unit design in addition to doping efficiency. As previously described, the fused ring thieno[3:2,b]thiophene in P4 decreases the persistence length compared to P3, which is known to reduce electron delocalization which can reduce film conductivity. 34 This result again shows the importance of efficient doping and structural organization on film conductivity. Lastly, with P3 and P4, the absorbance of F4TCNQ -appears to remain relatively constant after 10 mol% doping, which indicates that the higher concentrations of dopant are not resulting in more doping events. This is consistent with the plateau in conductivity observed after 10 mol%. The F4TCNQ -absorbance is taken as the local maximum at 868 nm obtained from UV-Vis-NIR measurements and is normalized by film thickness. As a result, the normalized absorbance is proportional to F4TCNQ -concentration.

Sequential Doping of Thiophene-based Copolymer Films. Although it is a more complex process, sequential doping is viewed favorably because it circumvents issues of aggregation that are inherent to solution doping and typically leads to enhanced levels of electrical conductivity. 7,15 Efficient sequential doping requires a solvent that is a good solvent for the dopant but a marginal solvent for the polymer -the solvent must swell, but not dissolve, the polymer film to allow for effective integration of the dissolved dopant molecule. 16 Therefore, an array of solvents and solvent blends were studied to determine the optimum solvent for sequential doping of the thiophene-based copolymers. Six specific solvents were selected based on their solubility parameter distance between the solvent and the corresponding copolymer, Ra. The Ra value for each solvent-copolymer pair was calculated using Equation 1, where the lowercase subscripts on the three HSPs correspond to either the polymer, p, or to the solvent, s. 24,45 𝑅 𝑎 = (4(𝛿 𝐷 𝑝 -𝛿 𝐷 𝑠 ) 2 + (𝛿 𝑃 𝑝 -𝛿 𝑃 𝑠 ) 2 + (𝛿 𝐻 𝑝 -𝛿 𝐻 𝑠 ) 2 ) 1/2 (Eq. 1)

The corresponding Ra values for the six solvents used -THF, DCM, 50:50 Acetone:DCM, 75:25

Acetone:DCM, acetone, and DMF -and each copolymer are summarized in Table S4. With these solvents, Ra values range from approximately 5.45 to 14.00 MPa 1/2 , with lower numbers indicating greater solubility of the polymer and higher numbers indicating poorer solubility. This range of solvent character enables insights into how polymer-solvent interaction energy impacts the resultant thin film conductivity. For all sequential doping studies, the concentration of F4TCNQ was fixed at 1 mg/mL.

As was done for solution doped thin films, the surface roughness of sequentially doped films was measured to assess film topology and potential aggregation. As summarized in Table S5, the surface roughness values are rather high and display a large sample-to-sample variation, especially compared to solution doped thin films. These traits are indicative of aggregation or nonuniformity in the film. The AFM images presented in Figures S35-S40 show the presence of aggregates across the film surfaces. These aggregate structures are believed to be excess dopant that did not integrate into the polymer film. This contention is supported by results from imaging P3HT films that were sequentially doped with F4TCNQ using acetone as the dopant solvent and then rinsed with methanol, which dissolves F4TCNQ but does not dissolve or swell P3HT (Ra > 20 MPa 1/2 ). As shown in Figure S41, prior to rinsing, the surface was populated with large aggregates and had a measured roughness of 19.1 ± 7.2 nm. After rinsing with methanol, the surface was smooth and uniform with a roughness of 1.9 ± 1.4 nm. These results indicate that aggregates of excess F4TCNQ were the cause of the increased roughness and irregular topography observed in Figures S35-S40 for the sequentially doped thin films. This finding is consistent with results from Jacobs et al., who reported phase segregated F4TCNQ domains on thin films when sequentially doping P3HT with F4TCNQ in acetonitrile at 1 mg/mL. 7 Due to the surface roughness being largely dominated by the presence of excess dopant, the images and surface roughness values do not offer much insight into the uniformity/topology of the underlying doped film. However, because the films are thin (< 30 nm) and to prevent removal of dopant from the upper region or from the entirety of the thin film, the methanol washing was not implemented as part of the process used to generate sequentially doped thin films.

As with solution doped thin films, after measuring the I-V curve of each sequentially doped sample, film thicknesses were measured (Table S6) and used to calculate the resultant conductivity. Figure 6 shows the conductivity of each sequentially doped copolymer thin film as a function of the Ra value that characterizes the dopant solvent-copolymer pair. First and as seen in Figure 6a, P3HT (AA+BB) again shows low conductivity across the spectrum of solvents used for doping, which suggests that doping is inefficient in this regio-irregular polymer. In comparison, the conductivity of sequentially doped P3HT (AB) thin films increases by ~3 orders of magnitude, depending on the dopant solvent used. This conductivity dependence on dopant solvent is consistent with work done by Yoon et al. with P3HT. 16 At low Ra values (5-8 MPa 1/2 ) and high Ra values (>12 MPa 1/2 ), the conductivity is low, while at moderate Ra values (8-12 MPa 1/2 ), the conductivity is much higher. This Ra-dependent conductivity is also seen for each copolymer film (Figure 6b), though the conductivity values are significantly higher than those measured for the P3HTs. In addition, the pattern of solvent quality-dependence behavior that is based on the thermodynamic interaction between the dopant solvent and polymer appears to be universally shared by all of these thiophene-based copolymers. Specifically, at low Ra values, the solvent can partially dissolve the film, which leads to inconsistent dopant integration. At high Ra values, the solvent is a poor solvent for the copolymer, which results in low levels of dopant integration beyond the polymer-solvent interface, resulting in the dramatic fall-off in conductivity. Lastly, with moderate Ra values, the solvent swells the copolymer chains, which allows for effective integration of the dopant, leading to higher conductivities.

While the need for a marginal solvent to maximize conductivity has also been demonstrated for other systems 46,47 , HSPs provide a useful framework for selecting as well as creating a useful solvent system. In this work, solvent mixtures were used to tailor Ra values. Specifically, moderate Ra values were accessed by mixing DCM (low Ra) and acetone (near the high Ra border), resulting in high conductivities. This indicates that binary mixtures of poor or sub-optimal solvents can be used to tailor the Ra value, leading to an optimized conductivity. Additionally, for this series of thiophene-based copolymers, the trend of increasing conductivity with increasing spacing between alkyl side chains is also observed, but the effect is less prominent compared to the films made by solution doping. This indicates that altering the copolymer design by increasing the spacing length between alkyl side chains enhances conductivity by reducing steric demand along the chain backbone, which allows dopant molecules to access the backbone, perhaps without compromising the ability of the conjugated copolymers to organize through π-stacking interactions. UV-Vis-NIR measurements of sequentially doped thin films (Figure S42-S47) were used to gain insight into the efficiency of doping. However, unlike the solution doped thin films, there is a lack of correlation between conductivity and F4TCNQ -absorbance (concentration) (Figure S48). This indicates that in sequential doping, the amount of F4TCNQ -in the doped film is not the main factor that sets the conductivity. Rather, it is believed that micro-organization across multiple length scales plays a more important role, 8,48 which heightens the importance of using a marginal solvent to maintain the organization of the polymer film after doping, thereby enhancing conductivity. 15 The impact of solvent quality and dopant integration on thin film organization will be addressed in a follow-on study.

Comparison of Solution and Sequential Doping. Solution doping is generally considered to be inferior to sequential doping when comparing electrical performance, 7,15 however this heuristic was not observed for all of the thiophene-based copolymers. The conductivities of the sequentially doped thin films were very similar to that of the solution doped films. Therefore, it is insightful to compare the amount of F4TCNQ -generated by each doping method to the film conductivity, as shown in Figure 7. To facilitate comparisons between the different copolymer designs, a set of common axis scales was used. Figure 7a shows that both solution and sequential doping were ineffective for P3HT (AA+BB), as both methods resulted in films of low conductivity due to the regio-irregularity of the polymer. On the other hand, and despite also having overall low concentrations of F4TCNQ -, a trend seems to emerge for P3HT (AB): at low F4TCNQ - absorbance, a higher conductivity is seen with sequential doping. This indicates that reducing or eliminating aggregation that is prevalent in solution doping is a key element of promoting microstructural organization in the thin film that enhances electron transport. This contention is consistent with studies of doped P3HTs by Jacobs et al. 7 and Scholes et al. 15 , and a comprehensive analysis of a wide variety of conjugated polymers, including P3HTs, by Noriega et al. 10 The enhancement of conductivity with sequential doping becomes more apparent with the other copolymers: higher conductivities are reached at lower concentrations of F4TCNQ -with sequential doping. Thereby, expanded designs also offer enhanced conductivity at lower concentrations of dopant while also increasing the conductivity by orders-of-magnitude compared to P3HT.