Suzuki–Miyaura Catalyst-Transfer Polycondensation of Triolborate-Type Carbazole Monomers

Herein, we report the Suzuki–Miyaura catalyst-transfer polycondensation (SCTP) of triolborate-type carbazole monomers, i.e., potassium 3-(6-bromo-9-(2-octyldodecyl)-9H-carbazole-2-yl)triolborate (M1) and potassium 2-(7-bromo-9-(2-octyldodecyl)-9H-carbazole-2-yl) triolborate (M2), as an efficient and versatile approach for precisely synthesizing poly[9-(2-octyldodecyl)-3,6-carbazole] (3,6-PCz) and poly[9-(2-octyldodecyl)-2,7-carbazole] (2,7-PCz), respectively. The SCTP of triolborate-type carbazole monomers was performed in a mixture of THF/H2O using an initiating system consisted of 4-iodobenzyl alcohol, Pd2(dba)3•CHCl3, and t-Bu3P. In the SCTP of M1, cyclic by-product formation was confirmed, as reported for the corresponding pinacolboronate-type monomer. By optimizing the reaction temperature and reaction time, we successfully synthesized linear end-functionalized 3,6-PCz for the first time. The SCTP of M2 proceeded with almost no side reaction, yielding 2,7-PCz with a functional initiator residue at the α-chain end. Kinetic and block copolymerization experiments demonstrated that the SCTP of M2 proceeded in a chain-growth and controlled/living polymerization manner. This is a novel study on the synthesis of 2,7-PCz via SCTP. By taking advantage of the well-controlled nature of this polymerization system, we demonstrated the synthesis of high-molecular-weight 2,7-PCzs (Mn = 5–38 kg mol−1) with a relatively narrow ÐM (1.35–1.48). Furthermore, we successfully synthesized fluorene/carbazole copolymers as well as 2,7-PCz-containing diblock copolymers, demonstrating the versatility of the present polymerization system as a novel synthetic strategy for well-defined polycarbazole-based materials.


Introduction
π-Conjugated polymers have attracted immense attention owing to their potential as electroactive and photoactive materials for fabricating various organic electronic devices [1,2] such as organic light-emitting diodes (OLEDs) [3,4], organic field-effect transistors (OFETs) [5,6], organic photovoltaics (OPVs) [7,8], and organic memory devices [9][10][11]. Most π-conjugated polymers investigated thus far consisted of thiophene [12][13][14][15][16], phenylene [17][18][19], and fluorene [20][21][22][23] units, which have been accessed through the step-growth polycondensations using cross-coupling reactions, such as Suzuki-Miyaura and Kumada-Corriu coupling. However, precise control of the molecular weight, dispersity (Ð M ), and end group of the π-conjugated polymers has been challenging due to the step growth nature of the polymerization. On the other hand, Yokozawa and McCullough established chain-growth-type catalyst-transfer polycondensation based on Kumada coupling and Suzuki-Miyaura coupling in 2001 [24], 2005 [25], and 2006 [26], respectively. This novel polymerization mechanism opened the access to a range of π-conjugated polymers with a predictable molecular weight, low Ð M , and defined end groups. mobenzene derivative as the initiator [43]. However, a significant amount of macrocyclic byproduct was generated owing to the chain transfer reaction, in addition to the desired linear polymer obtained from the bromobenzene derivative. We hypothesized that the chain transfer reaction can be suppressed by using triolborate salt monomers because of its high reactivity even at low temperatures, resulting in suppressed cyclization side reactions (Scheme 1).

SCTP of M1 to Produce 3,6-PCz
We first investigated the synthesis of end-functionalized poly[9-(2-octyldodecyl)-3,6carbazole] (3,6-PCz) with a narrow dispersity (ĐM). Recently, Jager et al. demonstrated the synthesis of 3,6-PCzs by the SCTP of a pinacol boronate-type monomer, using bromobenzene derivative as the initiator [43]. However, a significant amount of macrocyclic byproduct was generated owing to the chain transfer reaction, in addition to the desired linear polymer obtained from the bromobenzene derivative. We hypothesized that the chain transfer reaction can be suppressed by using triolborate salt monomers because of its high reactivity even at low temperatures, resulting in suppressed cyclization side reactions (Scheme 1).  [49]. Although SCTP proceeded to afford 3,6-PCz (Mn,SEC = 5500 g mol −1 ), the size-exclusion chromatography (SEC) analysis revealed a bimodal distribution with a dispersity (ĐM) of 1.32 (run 1, Table 1, Figure 1). The structure of the 3,6-PCz was investigated using matrix-assisted laser desorption/ionization time-offlight mass spectroscopy (MALDI-TOF MS). As shown in Figure 2, the MALDI-TOF mass spectrum exhibited six series of peaks with a regular interval of 445. 47 Da, corresponding to 9-(2-octyldodecyl) carbazole repeating units. One of the observed series of peaks was assigned to HOCH2−3,6-PCz with a phenylmethanol residue at the α-chain end and a bromine atom at the ω-chain end (BnOH/Br); for example, the peak of 1969. 43   We first attempted the polymerization of potassium 3-(6-bromo-9-(2-octyldodecyl)-9H-carbazole-2-yl)triolborate (M1) using an initiating system that consisted of 4-iodobenzyl alcohol (FI−CH 2 OH), Pd 2 (dba) 3 [49]. Although SCTP proceeded to afford 3,6-PCz (M n,SEC = 5500 g mol −1 ), the size-exclusion chromatography (SEC) analysis revealed a bimodal distribution with a dispersity (Ð M ) of 1.32 (run 1, Table 1, Figure 1). The structure of the 3,6-PCz was investigated using matrix-assisted laser desorption/ionization time-of-flight mass spectroscopy (MALDI-TOF MS). As shown in Figure 2, the MALDI-TOF mass spectrum exhibited six series of peaks with a regular interval of 445.47 Da, corresponding to 9-(2-octyldodecyl) carbazole repeating units. One of the observed series of peaks was assigned to HOCH 2 −3,6-PCz with a phenylmethanol residue at the α-chain end and a bromine atom at the ω-chain end (BnOH/Br); for example, the peak of 1969.43 at m/z matched well with the theoretical mass for the 4-mer of HOCH 2 −3,6-PCz ([M + H] + = 1971.46 Da). However, peaks corresponding to 3,6-PCz with hydrogen and bromine ends (Br/H; e.g., m/z of 1864.42 Da), bromine at both ends (Br/Br; e.g., m/z of 1942.31 Da), and cyclic 3,6-PCz (e.g., m/z = 1783.50 Da) were observed, revealing the uncontrolled polymerization nature. According to a report by Jager et al. [43], the sharp SEC elution peaks observed in the lower-molecular-weight region and the relatively broad elution peak at high molecular weight in Figure S1 most likely arise from the cyclic by-product and linear polymer, respectively. The MALDI-TOF mass spectrum of the high-molecular-weight side, which was isolated by preparative SEC, predominantly showed peaks corresponding to 3,6-PCz, with a phenylmethanol residue at the end ( Figure S1). Nevertheless, these results suggest that the chain transfer reaction occurred significantly, even with the triolborate-type monomer under this polymerization condition.  Table 1).  Table 1).
According to a report by Jager et al. [43], the sharp SEC elution peaks observed in the lower-molecular-weight region and the relatively broad elution peak at high molecular weight in Figure S1 most likely arise from the cyclic by-product and linear polymer, respectively. The MALDI-TOF mass spectrum of the high-molecular-weight side, which was isolated by preparative SEC, predominantly showed peaks corresponding to 3,6-PCz, with a phenylmethanol residue at the end ( Figure S1). Nevertheless, these results suggest that the chain transfer reaction occurred significantly, even with the triolborate-type monomer under this polymerization condition.
To inhibit the chain transfer reaction, we optimized the polymerization conditions for M1. Hong et al. reported the synthesis of polyfluorene with a narrow Ð M by SCTP with the addition of extra t-Bu 3 P [50]. Therefore, we conducted the polymerization of M1 by adding 4.4 equiv. of t-Bu 3 P with respect to FI−CH 2 OH, without changing any other factors (run 2, Table 1). As a result, the SEC elution peak corresponding to the cyclic by-product was reduced compared to that in run 1, implying that the chain transfer reaction was suppressed to some extent ( Figure 1). We also examined the polymerization by changing the phosphine ligand (e.g., dppp, XPhos, and RuPhos), but no positive results were obtained. Next, polymerization was carried out by lowering the reaction temperature (−30 • C and −10 • C) to further reduce the unwanted chain transfer reaction. For polymerization at −30 • C (run 4), M1 precipitated during SCTP. In contrast, the polymer given at −10 • C (run 3) exhibited a nearly monomodal SEC elution peak with a  Figure 1). Consequently, lowering the reaction temperature was found to be an effective method to suppress the chain transfer reaction as well as cyclic by-product formation. For comparison, we examined the polymerization of a conventional pinacolboronate-type carbazole monomer, 3-(6-bromo-9-(2-octyldodecyl)-9H-carbazole-2-yl)4,4,5,5-tetramethyl-1,2,3dioxaborolane (M3), under the same conditions as for M1. However, no polymerization was observed at −10 • C (run 5, Table 1, Figure S3). This result demonstrates that the SCTP of the triolborate salt monomer at low temperatures is essential for the controlled synthesis of 3,6-PCz. To inhibit the chain transfer reaction, we optimized the polymerization conditions for M1. Hong et al. reported the synthesis of polyfluorene with a narrow ĐM by SCTP with the addition of extra t-Bu3P [50]. Therefore, we conducted the polymerization of M1 by adding 4.4 equiv. of t-Bu3P with respect to FI−CH2OH, without changing any other factors (run 2, Table 1). As a result, the SEC elution peak corresponding to the cyclic by-product was reduced compared to that in run 1, implying that the chain transfer reaction was suppressed to some extent ( Figure 1). We also examined the polymerization by changing the phosphine ligand (e.g., dppp, XPhos, and RuPhos), but no positive results were obtained. Next, polymerization was carried out by lowering the reaction temperature (−30 °C and −10 °C) to further reduce the unwanted chain transfer reaction. For polymerization at −30 °C (run 4), M1 precipitated during SCTP. In contrast, the polymer given at −10 °C (run 3) exhibited a nearly monomodal SEC elution peak with a narrow dispersity of 1.19 ( Figure  1). Consequently, lowering the reaction temperature was found to be an effective method to suppress the chain transfer reaction as well as cyclic by-product formation. For comparison, we examined the polymerization of a conventional pinacolboronate-type carbazole monomer, 3-(6-bromo-9-(2-octyldodecyl)-9H-carbazole-2-yl)4,4,5,5-tetramethyl-1,2,3dioxaborolane (M3), under the same conditions as for M1. However, no polymeriza-   The structure of 3,6-PCz obtained at −10 • C (run 3) was characterized in detail by NMR and mass spectral analyses. In the 1 H NMR spectrum (Figure 3a), the major signals were successfully assigned to the 3,6-PCz backbone (8.65-7.31 ppm: proton A) and 2-octyldodecyl side chain (2.11 ppm: proton B, 4.12 ppm: proton C, 1.82-0.75 ppm: proton D and E), which is consistent with the reported 1 H NMR data for poly[9-(2-octyl)-3,6carbazole] synthesized via SCTP [43]. More importantly, a minor signal due to the benzyl proton (proton F) in the α-chain end group was observed at 4.75 ppm, indicating the initiation from FI−CH 2 OH. Thus, the obtained product was identified as the expected poly[9-(2-octyldodecyl)-3,6-carbazole] possessing a phenylmethanol residue at the α-chain end (HOCH 2 −3,6-PCz). The number-average molecular weight (M n,NMR ) of the HOCH 2 −3,6-PCz (run 3), determined based on end group analysis of the 1 H NMR spectrum, was 6700 g mol −1 . In order to confirm the end-group fidelity, we performed MALDI-TOF MS analysis on this sample. The MALDI-TOF mass spectrum showed two series of peaks with a regular interval of 445.47 Da, corresponding to 9-(2-octyldodecyl) carbazole repeating units ( Figure 3b). The peaks indicated by filled circles (•) were assignable to HOCH 2 −3,6-PCz possessing a phenylmethanol residue at the α-chain end and a bromine atom at the ω-chain end (BnOH/Br); for example, the peak at m/z of 1970.40 showed good agreement with the theoretical mass for the 4-mer ([M + H] + = 1970.46 Da). The peaks indicated by filled triangles ( ) were assignable to HOCH 2 −3,6-PCz possessing a phenylmethanol residue at the α-chain end and a hydrogen atom ω-chain end (BnOH/H). Overall, the SCTP of M1 at low temperature enables better control, yielding narrowly dispersed 3,6-PCzs with sufficient end-group fidelity.
To understand the polymerization properties of M1, we evaluated the time-dependent change of the reaction product via the SEC and MALDI-TOF MS analysis of the crude aliquot quenched by the addition of HCl (1 mol L −1 ). SEC did not confirm an increase in molecular weight over time, as expected for chain growth polycondensation ( Figure 4). This suggests that the polymerization of M1 does not proceed in a perfect living fashion, even with the best polymerization conditions. It is reasonable to deduce that various side reactions occur in  Figure S2). This complex polymerization behavior can be explained as shown in Scheme 2. The first step of polymerization involves the transmetallation between the initiator and monomer to yield the Pd complex (i) when the Pd(0) species, which are generated by reductive elimination, are intramolecularly shifted to the carbon-bromine bond at the end of the same molecule. Polymerization proceeds via chain growth from (ii), resulting in the formation of 3,6-PCz with a phenylmethanol residue at the initiation end (HOBn/Br or HOBn/H) (Scheme 2a). However, when the Pd(0) species does not shift to the bromine end in the same molecule but intramolecularly shifts to the new monomer, a new initiating species (iii) is generated [51][52][53][54]. The initiating species (iii) generates 3,6-PCz with a triolborate residue at the α-chain end and bromine atom at the ω-chain end, which results in the formation of the cyclic by-product though the intramolecular reaction. Meanwhile, when the Pd catalyst is oxidized by a trace amount of oxygen [55,56], the homocoupling of two monomer molecules (iv) produces another initiating species, which results in the formation of 3,6-PCz at the brominated α-chain end (Br/Br and Br/H) (Scheme 2b).
Overall, even the polymerization of the triolborate-type monomer M1 did not proceed in a perfect fashion. Nevertheless, we successfully obtained 3,6-PCz with the desired initiator residue virtually without contamination of the cyclic by-product, by optimizing the polymerization temperature and time.   Overall, even the polymerization of the triolborate-type monomer M1 did not proceed in a perfect fashion. Nevertheless, we successfully obtained 3,6-PCz with the desired initiator residue virtually without contamination of the cyclic by-product, by optimizing the polymerization temperature and time.    Overall, even the polymerization of the triolborate-type monomer M1 did not proceed in a perfect fashion. Nevertheless, we successfully obtained 3,6-PCz with the desired initiator residue virtually without contamination of the cyclic by-product, by optimizing the polymerization temperature and time.

SCTP of M2 to Produce 2,7-PCz
Next, we investigated the polymerization of potassium 2-(7-bromo-9-(2-octyldodecyl)-9H-carbazole-2-yl)triolborate (M2). The initial attempt was carried out using an initiating system that consisted of FI−CH 2 OH, Pd 2 (dba) 3 Table 1), according to the best polymerization conditions for potassium 2-(7-bromo-9,9-dihexyl-9H-fluorene-2-yl)triolborate [49]. The SCTP of M2 successfully proceeded to afford a polymer with M n,SEC = 3700 g mol −1 and Ð M = 1.23. To study the effect of the polymerization temperature, we carried out SCTP at −30 • C and −10 • C while fixing the other reaction parameters. At −30 • C (run 7), the monomer and product precipitated during the SCTP. Meanwhile, at −10 • C (run 8), a low monomer conversion was observed even after polymerization for 24 h. Consequently, it became apparent that M2 should be performed at 30 • C. For comparison, we examined the SCTP of a conventional pinacolboronate-type monomer, i.e., 2-(7-bromo-9-(2-octyldodecyl)-9H-carbazole-2-yl)4,4,5,5-tetramethyl-1,2,3dioxaborolane (M4) under the comparable condition (run 9). As a result, no polymerization was observed even after 24 h. This result demonstrates the advantage of using a triolborate salt monomer for the synthesis of 2,7-PCz. Then, the structure of the obtained 2,7-PCz was characterized in detail by NMR and MALDI-TOF mass spectral analyses. Figure 6a shows the 1 H NMR spectrum of the product obtained from run 6. The major observed signals were reasonably assignable to the 2,7-PCz backbone (8.68-7.33 ppm: proton A) and 2-octyldodecyl side chain (2.12 ppm: proton B, 4.14 ppm: proton C, 1.82-0.73 ppm: proton D and E), which is consistent with the reported 1 H NMR data for poly[N-heptadecan-2,7-carbazole] synthesized via KCTP [42]. More importantly, a minor signal at 4.75 ppm was assigned to the benzyl proton at the α-chain end (proton E), which confirmed the successful installation of the functional end group. Therefore, the obtained product was ascribed to the expected poly[9-(2-octyldodecyl)-2,7-carbazole] possessing a phenylmethanol residue at the α-chain end (HOCH 2 −2,7-PCz). The M n,NMR of HOCH 2 −2,7-PCz (run 6) was calculated to be 5080 g mol −1 by end-group analysis of the 1 H NMR spectrum. In order to further confirm the end-group fidelity, we performed MALDI-TOF MS analysis on the product from run 6. The MALDI-TOF mass spectrum mainly exhibited two series of peaks showing a regular interval of 445.4 Da corresponding to 9-(2-octyldodecyl) carbazole repeating units (Figure 6b). The peaks denoted by filled triangles ( ) were assignable to HOCH 2 −2,7-PCz possessing a phenylmethanol residue at the α-chain end and a bromine atom at the ω-chain end. . This is strong evidence for the chain-growth mechanism of the SCTP. The kinetic plot indicates a distinct first-order kinetic behavior for this polymerization system. According to slope of the kinetic plot, the rate constant was estimated to be 5. 2,7-PCz-b-PF with Mn,NMR of 9400 g mol −1 and a ÐM of 1.38. The SEC traces of the products before and after block copolymerization are shown in Figure 7d, in which the SEC elution peak showed clear shift toward the high-molecular-weight side upon the addition of the second monomer. This result verified a truly controlled/living propagating end, which led to chain extension by the second monomer addition. Overall, the aforementioned results successfully support the chain-growth and controlled/living polymerization nature of the SCTP of M2.  Table 1, Mn,NMR; 5080-37,900 g mol −1 ) were successfully obtained (Figure 8). To the best of our knowledge, 2,7-PCz obtained from run 12 (Mn,SEC of 37,900 g mol −1 ) had the highest molecular weight among the poly[9-(2-octyldodecyl)-2,7-carbazoles] prepared by chain-growth polycondensation.  Table 1, M n,NMR ; 5080-37,900 g mol −1 ) were successfully obtained (Figure 8). To the best of our knowledge, 2,7-PCz obtained from run 12 (M n,SEC of 37,900 g mol −1 ) had the highest molecular weight among the poly[9-(2-octyldodecyl)-2,7-carbazoles] prepared by chaingrowth polycondensation.

Random Copolymerization of Triolborate-Type Carbazole and Fluorene Monomers
Recently, conjugated copolymers based on carbazole or fluorene groups with varying connectivity, such as at the 2,7 or 3,6 positions, were found to be useful for the extraction of single-walled carbon nanotubes [57]. Therefore, we are interested in testing our SCTP system for the synthesis of poly(carbazole-co-fluorene)s (Scheme 4). First, we examined the random copolymerization of M1 and M2 using the initiating system that con-  Table 2 run 13). SCTP proceeded to afford 2,7-PCz-co-3,6-PCz (Mn,SEC = 7800 g mol −1 ), which exhibited a unimodal SEC elution peak (Figure 9). The chemical structure of 2,7-PCz-co-3,6-PCz was identified by 1 H NMR analysis ( Figure 10). The major signals were assigned to the polycarbazole backbone (8.6-7.3 ppm: proton A) and 2-octyldodecyl side chain (2.1 ppm: proton B, 4.1 ppm: proton C, 1.8-0.7 ppm: proton D and E). More importantly, a minor signal at 4.7 ppm due to the α-chain end benzyl proton (proton F) was clearly observed, confirming the successful installation of the functional end group. Therefore, the product identified to be the expected copolymer possessed a phenylmethanol residue at the αchain end (2,7-PCz-co-3,6-PCz). The ratio of the 2,7-carbazole unit to the 3,6-carbazole unit in the obtained copolymer was 14:14, which was in good agreement with the monomer feed ratio. The Mn,NMR of 2,7-PCz-co-3,6-PCz (run 13) was calculated to be 11,500 g mol −1 by end-group analysis of the 1 H NMR spectrum.

Random Copolymerization of Triolborate-Type Carbazole and Fluorene Monomers
Recently, conjugated copolymers based on carbazole or fluorene groups with varying connectivity, such as at the 2,7 or 3,6 positions, were found to be useful for the extraction of single-walled carbon nanotubes [57]. Therefore, we are interested in testing our SCTP system for the synthesis of poly(carbazole-co-fluorene)s (Scheme 4). First, we examined the random copolymerization of M1 and M2 using the initiating system that consisted of FI−CH 2 OH, Pd 2 (dba) 3 Table 2 run 13). SCTP proceeded to afford 2,7-PCz-co-3,6-PCz (M n,SEC = 7800 g mol −1 ), which exhibited a unimodal SEC elution peak (Figure 9). The chemical structure of 2,7-PCz-co-3,6-PCz was identified by 1 H NMR analysis ( Figure 10). The major signals were assigned to the polycarbazole backbone (8.6-7.3 ppm: proton A) and 2-octyldodecyl side chain (2.1 ppm: proton B, 4.1 ppm: proton C, 1.8-0.7 ppm: proton D and E). More importantly, a minor signal at 4.7 ppm due to the α-chain end benzyl proton (proton F) was clearly observed, confirming the successful installation of the functional end group. Therefore, the product identified to be the expected copolymer possessed a phenylmethanol residue at the α-chain end (2,7-PCz-co-3,6-PCz). The ratio of the 2,7-carbazole unit to the 3,6-carbazole unit in the obtained copolymer was 14:14, which was in good agreement with the monomer feed ratio. The M n,NMR of 2,7-PCz-co-3,6-PCz (run 13) was calculated to be 11,500 g mol −1 by end-group analysis of the 1 H NMR spectrum.  In addition, random copolymerization was carried out for the combination of carbazole and fluorene monomers, i.e., M1/M5 and M2/M5 (Scheme 4). Both polymerizations homogeneously proceeded and afforded the corresponding random copolymers, i.e., PFco-3,6-PCz and PF-co-2,7-PCz, with ĐM values of 1.38 and 1.48, respectively. The 1 H NMR spectra of the resulted random copolymers showed characteristic signals from both the fluorene and carbazole units (Figure 10), which confirmed the successful synthesis of random copolymers.
In addition, random copolymerization was carried out for the combination of carbazole and fluorene monomers, i.e., M1/M5 and M2/M5 (Scheme 4). Both polymerizations homogeneously proceeded and afforded the corresponding random copolymers, i.e., PF-co-3,6-PCz and PF-co-2,7-PCz, with Ð M values of 1.38 and 1.48, respectively. The 1 H NMR spectra of the resulted random copolymers showed characteristic signals from both the fluorene and carbazole units (Figure 10), which confirmed the successful synthesis of random copolymers.

Synthesis of 2,7-PCz-Containing Diblock Copolymers Using Macroinitiators.
Finally, we employed iodobenzene-terminated macroinitiators for the SCTP of M2 in order to synthesize 2,7-PCz-containing block copolymers (Scheme 5). Based on our previous study, we reduced the water content as much as possible to prevent the precipitation of the macroinitiator [49]. We first attempted SCTP with iodobenzene-terminated poly-