Optimization of direct arylation polymerization (DArP) through the identification and control of defects in polymer structure


  • This manuscript is dedicated to Prof. Jean Fréchet on the occasion of his 70th birthday and to his extraordinary contributions to polymer science.


As a newly emerged protocol for the synthesis of conjugated polymers, direct arylation polymerization (DArP) is an environmentally friendly and cost-effective alternative to traditional methods of polymerization. DArP efficiently yields conjugated polymers with high yield and high molecular weight. However, DArP is also known to produce defects in polymer chemical structure. Together with molecular weight and polydispersity, these defects are considered to be important parameters of polymer structure and they have a strong impact on optical, electronic and thermal properties of conjugated polymers. The four major classes of conjugated polymer defects inherent for DArP have been identified: homocoupling regiodefects, branching defects, end group defects, and residual metal defects. To have a precise control over the polymer properties, it is important to understand what causes the defects to form during the polymerization process and be able to control their content. Here within the scope of current literature, we discuss in detail the definition and origin of all these defects, their influence on polymer properties and effective means to control the defects through fine tuning of the DArP reaction parameters. © 2014 Wiley Periodicals, Inc. J. Polym. Sci., Part A: Polym. Chem. 2015, 53, 135–147


Direct arylation polymerization (DArP) for the synthesis of conjugated polymers has emerged recently as an alternative to traditional methods of transition metal catalyzed polymerization, such as Stille,[1] Suzuki,[2, 3] Negishi,[4] and Kumada[5] polymerizations with Stille polymerization being especially popular since nearly all the top performing conjugated polymers for the organic photovoltaics (OPV) application are made by Stille polymerization. In contrast with these methods, which require functionalization of monomers with organotin (Stille), organoboron (Suzuki), organozinc (Negishi), or organomagnesium (Kumada) groups, DArP as a C–H activation method allows polymerization of unfunctionalized monomers as illustrated in Figure 1. For this reason, DArP offers such advantages as fewer synthetic steps, better atom economy and cost-effectiveness as well as elimination of toxic organotin waste in the case of Stille polymerization. Although C–H activation methods were known for over 50 years,[6] their application for the synthesis of conjugated polymers has not been explored until recently.[7-11]

Figure 1.

DArP compared to the classic methods for the synthesis of conjugated polymers.

The very first attempt to utilize C-H activation for preparation of poly(3-hexylthiophene) (P3HT) was reported in 1999 by Lemaire et al.[12] 2-Iodo-3-hexylthiophene was used as a monomer and only oligomeric products with number-average molecular weight of ∼3 kDa were obtained. In the next 10 years only two more articles were published also reporting oligomer synthesis.[13, 14] Then in 2010, a critical report of synthesis of P3HT with high molecular weight and high yield by Ozawa et al.[15] fueled a strong interest in DArP from many research groups, which resulted in a rapidly increasing number of publications in the following years (Fig. 2). While the majority of new publications to this day focus primarily on fine tuning the reaction conditions to yield polymers with desirable properties, there are already several reports, which simply employ the developed DArP conditions for the synthesis of target polymers demonstrating that DArP is a maturing and convenient method for the synthesis of conjugated polymers.[16-22]

Figure 2.

DArP publication timeline.

As such, DArP has been applied to synthesize a variety of conjugated polymers with high yields and high molecular weights that are comparable to those prepared via traditional methods. However, together with other polymerization methods, DArP does not yield polymers with perfectly ideal structure. The resulting polymers are known to contain a certain amount of structural imperfections, or defects. Such defects originate from side reactions during the cross-coupling process. These side reactions may have less impact on the synthesis of small molecules since it is usually possible to separate the by-products of cross-coupling. However, in the case of conjugated polymers these by-products are embedded into the polymer chains. Together with other elements of polymer structure, such as molecular weight and polydispersity index (PDI), the defects in polymer chains have a strong impact on polymer properties, which influence material's performance in many optoelectronic applications. Therefore, it is important to precisely control the nature and the amount of defects during polymerization. Through limiting or eliminating these defects, the structure and properties of DArP polymers can also be converged with those made by classic methods of polymerization, thereby yielding the target materials with target properties but via the environmentally friendly and cost-effective polymerization protocol. Since every polymerization method has certain types of polymer chain defects associated with it, we have identified the four major classes of conjugated polymer defects that are found in DArP polymers: homocoupling, branching, end-group defects and residual metal defects (Fig. 3). In this highlight, we discuss in detail the origin of these defects, the dependence of polymer properties on these defects and available means to control them. This highlight will benefit the conjugated polymer community as a whole and it will be especially useful for polymer chemists working on DArP as well as methods other than DArP since, to our knowledge, the subject of conjugated polymer main chain defects has not been reviewed to date.

Figure 3.

The four classes of defects in conjugated polymer chemical structure associated with DArP: (a) Homocoupling. (b) Branching. (c) End-group defects. (d) Residual metal defects.


DArP Reaction Conditions

The amount of defects in the primary polymer structure depends to a large extent on the reaction conditions that are used for polymerization. Currently, in the literature describing synthesis of conjugated polymers via DArP two distinct sets of reaction conditions could be identified: the so-called Fagnou conditions and the so-called Ozawa conditions (Table 1). The Ozawa conditions have been reported for the first time by Ozawa et al.[15] for the synthesis of P3HT in 2010 and since then these conditions have been adopted with or without modifications by many research groups for the synthesis of a variety of conjugated polymers.[23-32] The Ozawa conditions include palladium-based Herrmann–Beller catalyst,[33, 34] usually with trianisylphosphine as a ligand, THF or toluene as a solvent, Cs2CO3 as a base with or without pivalic acid as a proton shuttle. The reaction temperature is reported to be 110–120 °C, which is above the boiling points of both THF and toluene, which demands that polymerization is conducted in a pressurized vessel.

Table 1. The Two Sets of DArP Conditions and the Typically Used Reagents
  1. a

    Frequently, no phosphine ligand is used.

  2. b

    Sometimes, carboxylic acid is not used.

FagnouPd(OAc)2Phosphine ligandsaDMA, PhMeK2CO3, KOAc, KOPivCH3CO2H, PivOH, 1-AdCO2H
OzawaHerrmann–Beller, Pd2dba3*CHCl3, PdCl2(MeCN)2Phosphine ligandsTHF, PhMeCs2CO3CH3CO2H, PivOH, 1-AdCO2Hb

The Fagnou conditions have been developed by Fagnou et al. for the synthesis of small molecules via direct arylation in 2006[35] and then later adopted for the synthesis of conjugated polymers with little or no modifications.[36-58] The conditions include rather simple, inexpensive, and bench-stable reagents. As such, palladium acetate is used as a catalyst with or without[59] a phosphine ligand, N,N-dimethylacetamide (DMA) is used as a solvent, K2CO3 as a base and a carboxylic acid (usually, pivalic acid) as a carboxylate ligand/proton shuttle. The reaction temperature varies between ∼70 and 110 °C, which is well below the boiling point of the solvent (165 °C) allowing polymerization at ambient pressure. The presence of a carboxylate additive was demonstrated to be critical for successful C–H activation reaction due to a significant decrease in transition state energy.[60-62] Indeed, the conditions used by Lemaire et al. in 1999–2003[12-14] differ from the Fagnou conditions primarily by the lack of any carboxylate source (other than acetates from catalytic Pd(OAc)2), which may indeed have been the reason for low molecular weights of ∼3 kDa. Both Fagnou and Ozawa conditions have been successfully used to synthesize a wide variety of conjugated polymers with high yield and high molecular weight.

Homocoupling Regiodefects

Homocoupling regiodefects illustrated in Figure 3(a) are defects of the polymer primary structure that originate from the homocoupling side reaction in the cross-coupling catalytic cycle, as illustrated in Scheme 2(a). When a cross-coupling reaction is conducted for the synthesis of small molecules, the homocoupling side reactions are frequently encountered. In the case of small molecules, these side products can be separated from the target cross-coupling products, which could be the reason why so little attention is paid to these side reactions. The scientific community is indeed accustomed to less than quantitative reaction yields. For example, a crude product of a successful cross-coupling reaction could be purified by removing 10% homocoupling product giving the target cross-coupling product in an excellent 90% yield. In the case of polymers, however, the homocoupling side reactions create linkages that are covalently embedded into the polymer chain. They cannot be separated and should be viewed as defects in polymer structure. In case of polymers, 10% of homocoupling defects could mean the difference between a suitable electroactive polymer and a substantially underperforming material.


When a transition-metal-catalyzed cross-coupling reaction is conducted between an aryl nucleophile and an aryl electrophile, two manifolds for homocoupling side reaction are possible: the homocoupling of aryl nucleophiles and the homocoupling of aryl electrophiles [Scheme 1(a)]. In the case of aryl nucleophiles containing highly reactive organometallic species, such as arylmagnesium (Kumada), arylzinc (Negishi), and aryllithium (Murahashi), a metal–halogen exchange reaction may take place resulting in both aryl nucleophile homocoupling and aryl electrophile homocoupling [Scheme 1(b)]. Precise temperature control usually allows the elimination of this side reaction.

In case of Stille and Suzuki reactions, the aryl nucleophiles are not as reactive, for example aryltin and arylboron species are indeed much more stable and are not prone to metal–halogen exchange. Nevertheless, the homocoupling process still takes place. The oxidative impurities in the reaction mixture (e.g., residual oxygen) may oxidize Pd(0) to Pd(II) thereby inducing homocoupling of aryltin and arylboron nucleophiles via double transmetallation with Pd(II) and the subsequent reductive elimination as illustrated in Scheme 2(a). On the other hand, the presence of reducing impurities may reduce Pd(II) to Pd(0) inducing homocoupling of arylhalogen electrophiles [Scheme 2(a)].[63] However, even most carefully conducted reaction often gives products of homocoupling. It is commonly accepted in the organometallic community that arylpalladium[64-67] (and arylnickel)[68] species have a tendency to undergo aryl-group exchange (so-called disproportionation), as illustrated in Scheme 2(a). In other words, the arylpalladium species undergo transmetallation with other arylpalladium species instead of aryltin or aryboron, leading to products of electrophile homocoupling and producing Pd(II) thus opening a pathway to nucleophile homocoupling as well. Such disproportionation is indeed common in many arylmetal (and alkylmetal) species [Scheme 2(b)]. For example, in case of organomagnesium reagents (Grignard reagents) this process is well known and it is called “Schlenk equilibrium”.[69, 70] In fact the term “Schlenk-type equilibrium” has been applied to homoleptic–heteroleptic equilibria of many organometallic species, demonstrating the generality of this phenomenon.[71-73]


As such, various processes may lead to homocoupling defects in polymers. It is noteworthy that the properties of some polymers are insensitive towards homocoupling. For example, regiosymmetric AB-type monomers will give the same polymer structure regardless of the amount of homocoupling [Fig. 4(a)]. On the other hand, when it comes to regiounsymmetric repeat units, the homocoupling linkages within the polymer chains become critical. For example, the degree of homocoupling in the most studied conjugated polymer—P3HT[74]—has been the topic of a very active discussion. P3HT consists of regiounsymmetric 3-hexylthiophene (3HT) repeat units, which may be arranged in a head-to-tail (HT), head-to-head (HH), and tail-to-tail (TT) fashion within polymer chains [Fig. 4(b)]. The degree of regioregular HT arrangement is called regioregularity (RR). As illustrated in Figure 4(c), the RR of a particular P3HT sample can be easily quantified by 1H NMR because the α-methylene groups of the hexyl chains have different chemical shifts depending on their arrangement: 2.80 ppm for HT and 2.58 ppm for HH linkage.[4] It has been demonstrated that the presence of regiodefects in P3HT results in polymer backbone twisting, lesser chain ordering, decreased degree of π–π stacking of polymer chains and as a result significantly influences polymers crystallinity,[75] optical properties,[4, 76] thermal properties,[77] charge transport,[78] specific capacitance,[79] and performance in OPV.[77, 80] The homocoupling is of course not limited to AB-type monomers, it also can occur during the copolymerization of AA and BB-type monomers, which are used for the synthesis perfectly alternating copolymers. It is noteworthy that if homocoupling defects are present in alternating copolymers, the alternating monomer sequence will no longer be perfect.

Figure 4.

(a) Regiosymmetric AB-type monomers give polymers insensitive to homocoupling defects. (b) P3HT is an example of regiounsymmetric polymer. (c) Typical fragment of P3HT 1H NMR spectrum and calculation of RR via 1H NMR: integral intensity of α-CH2 signal of the homocoupled repeat units divided by the integral intensity of all α-CH2 peaks.


In the case of DArP P3HT, the polymerization conditions have a strong impact on the polymer regioregularity. Recently it was demonstrated that P3HT prepared using Fagnou conditions has different RR depending on the reaction temperature and catalyst loading.[81] Indeed, keeping other reaction parameters the same but lowering the reaction temperature from 120 to 20 °C results in a RR increase from 82.6 to 89.3% [Fig. 5(a)]. This can be attributed to the palladium catalytic center becoming more selective with decreasing temperature resulting in fewer acts of disproportionation and fewer homocoupling regiodefects. Additionally, lowering the catalyst loading from 2 to 0.25% also leads to fewer regiodefects and increased RR from 88.9 to 91.3% [Fig. 5(b)]. This is consistent with arylpalladium disproportionation mechanism of regiodefect generation. As such, lower concentration of palladium catalyst results in a decreased likelihood of arylpalladium species encountering one another. Instead, they are surrounded by a larger portion of CAr-H functionalities of the monomers and growing polymer chains, which is favorable for forming HT linkages in the P3HT chains.

Figure 5.

The influence of DArP reaction parameters on the regioregularity of P3HT: (a) Influence of reaction temperature (Fagnou conditions: 2 mol % Pd(OAc)2, 30% PivOH, 1.5 eq K2CO3, 72 h, 0.04 M in DMA); (b) Influence of Pd(OAc)2 loading (Fagnou conditions: 30% PivOH, 1.5 eq K2CO3, 45 °C, 72 h, 0.04 M in DMA); (c) Influence of ligand nature (Ozawa conditions: 2 mol % Herrmann–Beller catalyst, 4 mol % ligand, 1.0 eq of Cs2CO3, 120 °C, 48 h, 1 M in THF).

Ozawa et al. demonstrated that the RR of P3HT obtained via Ozawa conditions depends greatly on the nature of the ligand used. Employment of phosphine ligand with greater coordinating ability results in RR of P3HT increasing from 63 to 96% illustrating a strong ligand effect [Fig. 5(c)].[15] Interestingly, lowering the amount of catalyst from 2 to 1% resulted in RR increase from 96 to 98%, consistent with previous observation for a substantially different case of Fagnou conditions and suggesting the generality of this approach. It is worth pointing out that the RR of P3HT obtained via Ozawa conditions is noticeably higher than that of P3HT obtained via Fagnou conditions (98 vs. 93%). Such a difference could be attributed to the fact that while the Fagnou conditions used in this study did not employ any phosphine ligands, the Ozawa catalytic system contained bulky and strongly coordinating ligands. In a recent paper focused on the mechanistic aspects of small molecule coupling through direct arylation, Ozawa et al. described an interesting phenomenon of decreasing the degree of homocoupling with increasing the bulk of substituents at palladium center.[82] Although largely unexplored at this point, such control of homocoupling regiodefects through tuning the bulk of substituents at palladium catalytic center could be useful in the case of DArP polymers.

β-Defects (Branching)

As illustrated in Figure 3(b), branching defects are the points of bifurcation of the polymer chain. In the predominant class of heterole-based conjugated polymers (five-membered ring: furan-, thiophene-, selenophene-, pyrrole-based polymers), the heterole repeat units contain linkages in α,α-positions, leading to linear chains (Scheme 3). However, if there is an α,β-linkage and α,α-linkage on the same repeat unit, this fragment is called a branching point, or a β-defect. Therefore, the term β-defect is applicable only to heterole-based branching points. In the case of nonheterole-based bifurcation points (fluorene, carbazole, etc.) they are simply called branching defects. Although the influence of branching on conjugated polythiophenes has been studied previously, the monomers either did not contain alkyl chains[83, 84] or β,β-linkages instead of α,β-linkages were studied[85] making the direct comparison with DArP-produced β-defects difficult.

Scheme 1.

(a) Nucleophile and electrophile homocoupling manifolds. (b) Metal–halogen exchange on the example of Kumada cross-coupling leading to homocoupling.

Scheme 2.

(a) Cross-coupling cycle and homocoupling pathways for the example of the Stille reaction. (b) The general case of the Schlenk equilibrium.

Scheme 3.

α,α-linkage and α,β-linkage on P3HT chain. Numeral locants are provided as well for thiophene ring.


DArP as a C–H activation method is capable of activating several C–H bonds if they are present in the monomers and the growing polymer chain during polymerization possibly leading to decreased regioselectivity and formation of branching points [Scheme 4(a)]. The activation of β-protons on the growing polymer chain may be the predominant source of defects due to substantially larger number of β-protons relative to α-protons during the later stages of polymerization [Scheme 4(a)]. Such phenomenon is unique to DArP as opposed to other classic methods of polymerization such as Stille, Suzuki, Negishi, Kumada, etc. since all the corresponding monomers in those cases have only one reactive nucleophilic functionality, which leads to linear polymer primary structure [Scheme 4(b)]. However, the precedent for β-defect formation has been reported for the case of oxidatively polymerized polythiophenes (e.g., using FeCl3 as an oxidant) although in that case the mechanism of polymerization is entirely different.[86]

Scheme 4.

(a) α- and β-protons on the monomer on the example of 2-bromo-3-hexylthiophene and the example of a β-defect on a growing P3HT chain. As the polymerization continues the ratio Hβ/Hα increases. (b) Stille polymerization on the example of P3HT and a fragment of a linear growing chain. Only α,α-linkages are permitted by reactive functionalities.


To date there are only two thorough studies of the β-defects influence on DArP polymer properties and both were conducted on the example of P3HT.[54, 87] As shown in Scheme 4(a), in the 2-bromo-3-hexylthiophene monomer there are two aromatic protons that can be activated by the catalytic system—the more reactive α-proton and the less reactive β-proton.[61] Activation of α-proton leads to linear polymer chains, the configuration that is deemed desirable for optoelectronic application.[85] On the other hand, activation of β-proton leads to β-defects [Scheme 4(a)]. From a synthetic standpoint, formation of β-defects may lead to cross-linking of polymer chains, which in the case of P3HT will produce an insoluble polymer fraction. This in turn will result in lower polymer yields after Soxhlet extraction, and this is exactly what has been observed experimentally.[46, 54, 81] The impact of β-defects on the polymer properties was also found to be very strong. As such, DArP P3HT containing only 1.4% of β-defects had remarkably different optoelectronic and thermal properties from P3HT with perfectly linear chains. In the UV–vis spectra of thin films P3HT with β-defects had a lower absorption coefficient and less pronounced vibronic shoulder indicating a lower degree of chain ordering.[88] The lower degree of ordering was confirmed with GIXRD data, which corroborates a lower degree of crystallinity and also a somewhat larger lamellar spacing (100) (17.5 vs. 17 Å for linear chain P3HT), possibly suggesting the inclusion of a branched fragment within polymer crystallite. Similar effects of branching on optical properties of P3HT was reported by Luscombe et al.[54] Additionally, it was reported that the fluorescence intensity of the thin films increased with increasing content of branching and it was attributed to decreased π–π stacking preventing quenching.[54] The influence of branching on the conformation of dissolved P3HT chains was studied via viscosity analysis and it was found that branched P3HT chains display more globular chain conformation than the linear analogs.[54] The thermal properties of branched P3HT differ significantly as well. The DSC traces indicate that the melting point of DArP P3HT with 1.4% β-defects is suppressed by 50 °C. The magnitude of this impact is remarkable since in the case of homocoupling defects the 10% difference in RR of P3HT (86 vs. 96%) was shown to translate into only 14 °C difference in melting point.[77] All this data suggests the disruption of polymer chain packing and their decreased ability to self-assemble into ordered structures. Naturally, these differences translate into different OPV performance of branched P3HT, which is shown to be significantly inferior to that of linear chain analog. These OPV results are consistent with previous reports in the literature stating that branching of the conjugated polymer chains indeed decreases hole mobility and reduces solar cell performance.[85]


Several approaches to control β-defects were reported. For the synthesis of P3HT using Fagnou conditions it was demonstrated that lower reaction temperature and lower catalyst loading result in fewer β-defects (as well as fewer regiodefects).[81] However, tuning only these two reaction parameters could not completely eliminate the defect formation and 0.16% of β-defects could still be detected.[87] Interestingly, it was shown that a bulkier carboxylate ligand from neodecanoic acid (NDA)[89] instead of pivalic acid allowed the complete suppression of β-defects and yielded P3HT with linear chain configuration. NDA is an inexpensive commercial regent sold as a mixture of tertiary carboxylic acids with chemical composition of C9H19COOH [Scheme 5(a)], it is an environmentally benign reagent with low toxicity that has been developed for various industrial applications but has not been used as a ligand for catalysis previously.[89, 90] It is hypothesized that due to the bulk of the carboxylate ligand the palladium catalytic center is less likely to activate the hindered β-protons on the growing P3HT chain due to steric repulsion with hexyl side-chains, as illustrated in Scheme 5(b). As a result of that the optoelectronic, thermal and OPV properties of DArP P3HT nearly converged with those of Stille P3HT, which contains only linear chains as dictated by the monomer reactive functionalities and supported by 1H NMR.[81, 87] Additionally, these optimized DArP reaction parameters have been applied to the synthesis of random copolymers of 3-hexylthiophene and 3-cyanothiophene (P3HT-CNT copolymers). In that case it was also shown that no β-defects are formed in DArP P3HT-CNT copolymers, and the DArP polymer optoelectronic properties were found to be remarkably similar to the Stille analogs, suggesting a wider applicability and attractiveness of these optimized DArP reaction conditions.[91]

Scheme 5.

(a) Structures of pivalic acid and components of neodecanoic acid. (b) Hypothetical steric repulsion between the catalytic center and polymer side chains.

Interestingly, synthetic efforts have also been made in the opposite direction of increasing the amount of branching of P3HT for the purposes of preparing hyperbranched polymers.[54] To increase the degree of branching of P3HT the Fagnou reaction conditions have been modified in terms of increased reaction temperature, increased catalyst loading, complete elimination of carboxylate functionalities by using potassium fluoride as a base and palladium chloride as a catalyst. Such reaction conditions result in a catalytic system that is remarkably active and therefore unselective, which in turn results in a substantial degree of β-proton activation.

Formation of branching points has also been observed in the case of copolymerization of 3,6-dibromo-N-octadecylcarbazole with 1,2,4,5-tetrafluorobenzene by Kanbara et al.[40, 44] [Scheme 6(a)] and corroborated by a model small-molecule study, where up to 5% of side products originating from unselective C–H activation were observed.[40] In a different report, in order to prevent formation of branching defects, Kanbara et al. used thiophene monomers with blocked β-positions, such as 3,4-dimethylthiophene[37, 42] and ethylenedioxythiophene (EDOT) [Scheme 6(b)].[36] Such an approach does eliminate the possibility of β-proton activation, however, it may not be a general strategy since there is a need for effective synthesis of polymers that do contain β-protons.

Scheme 6.

(a) Formation of branching defects upon polymerization of 3,6-dibromocarbazole and 1,2,4,5-tetrafluorobenzene using Fagnou conditions. (b) Utilization of thiophenes with blocked β-positions to prevent β-defect formation.

Another strategy that was used to prevent branching is the installation of directing groups on the monomer that would ensure selective α-proton activation.[43, 92] In this work 1-(2-pyrimidinyl)pyrrole was copolymerized with 2,7-dibromo-9,9-dioctylfluorene using a ruthenium-based catalyst, carboxylic acid additive, K2CO3 as a base and m-xylene as a solvent (Scheme 7). These reaction conditions originate from the large body of work on ruthenium-catalyzed direct arylation by Ackerman et al.[93] It was demonstrated that after the polymerization, the directing pyrimidinyl functionality can be effectively removed. This work is very important for two reasons. First, it was demonstrated that in case of pyrrole monomers it is possible to effectively utilize a directing group to ensure selective α-proton activation. Second, this is so far the only example of DArP using a nonpalladium-based catalyst. At the same time, this strategy is rather specific to the case of pyrrole-based monomers and therefore cannot be generally applicable to all conjugated polymers. Furthermore, the installation and subsequent removal of the pyrimidinyl group significantly decreases atom economy, which is one of the main merits of DArP.

Scheme 7.

Using the directing group to avoid formation of β-defects in Ru-catalyzed DArP with subsequent removal of the directing group.

End Group Defects


Inconsistencies in the end groups of polymer chains represent a case of structural defects located not within the main fragment of polymer chain as is the case with regiodefects and β-defects but rather on the ends of polymer chains, as illustrated in Figure 3(c).


It has been demonstrated that the nature of polymer end groups has a substantial impact on the polymer optoelectronic properties[94] and performance in organic electronic devices. It has been shown that bromine and organotin functionalities—typical end groups of Stille polymers—are in fact detrimental for OPV performance due to charge trapping, changes in interchain packing and active layer morphology.[95-98] As a result of this, efforts have been made to remove these functionalities via end-capping the polymer chains with conjugated segments that are similar to main chain components. To do so a two-step treatment is required for Stille or Suzuki polymerizations: the prepared polymer chains are typically reacted with aryltin or arylboron reagents to end-cap halogen (usually, bromine) terminus and with arylbromides to end-cap tin or boron terminus.


Control over end groups has always been a point of interest for many classes of polymers including conjugated polymers. In the case of DArP polymers the expected end groups are H/Br in contrast with Stille and Suzuki polymers, where expected end groups are R3Sn/Br and R2B/Br, respectively. Therefore, the end-capping of DArP polymers is more straightforward because only the bromine terminus needs to be capped. As such, Kanbara et al. demonstrated effective end-capping of a copolymer obtained from EDOT and 2,7-dibromo-9,9-dioctylfluorene by simply adding more EDOT in the end of polymerization, as shown in Scheme 8(a).[36] Complete end-capping of Br terminus was evidenced by MALDI-TOF spectra and elemental analysis. In a similar way, Ozawa et al. reported end-capping of a copolymer obtained from thienopyrrolodione (TPD) and 5,5′-dibromo-4,4′-dioctyl-2,2′ bithiophene by adding more TPD monomer at the end of the polymerization, as illustrated in Scheme 8(b).[25] In this study, depletion of Br groups was confirmed by 1H NMR.

Scheme 8.

(a) End-capping the Br-terminus with EDOT using Fagnou conditions. (b) End-capping Br-terminus with TPD using Ozawa conditions.

Together with capping the Br terminus attempts were made to end-cap the H terminus with various aryl groups by using arylpalladium complexes as initiators. As such, Ozawa et al. reported initiation of P3HT chains with various aryl functionalities obtained from corresponding bromides and iodides, as illustrated in Scheme 9.[23] Such initiation is seemingly analogous to the widely reported initiation via Kumada Catalyst Transfer Polymerization (KCTP),[99] however, the authors point to the facts that make initiation via DArP significantly different from KCTP. First, unlike the case with KCTP, incompatible with Grignard reagents fragments that contain groups like carbonyl and ester can be used to end-cap P3HT in the case of DArP. Second, the authors demonstrate that unlike KCTP which has a living nature, after initiating a P3HT chain in case of DArP, the palladium atom may, in fact, detach from the original chain and initiate another chain. Along these lines, it is important to note that the end-capping ratio (percentage of P3HT chains end-capped) was high (86–98%) but it never reached 100% as is the case with KCTP. Eventually, after all the monomer is consumed, the smaller chains combine into larger chains at the end of the reaction thereby increasing both the molecular weight and end-capping ratio. Such a picture is consistent with a step-growth polymerization mechanism. Importantly, at this point this is the only report where authors attempt to elucidate the DArP mechanism.

Scheme 9.

End-capping P3HT chains through initiation with arylpalladium catalysts.

An interesting example of end-capping DArP polymer chains through aryl transfer from phosphine ligand was observed by Ozawa et al.[24] As shown in Scheme 10(a), in this case, an o-methoxyphenyl substituent was introduced at the chain terminus of poly[(9,9-dioctylfluorene2,7-diyl)-alt-(2,3,5,6-tetrafluoro-1,4-phenylene)] due to aryl transfer from tri(o-methoxyphenyl)phosphine to the palladium center and subsequent reductive elimination. A small molecule modeling reaction was conducted to confirm this observation. Indeed, a direct arylation reaction between 2,7-dibromo-9,9-dioctylfluorene and pentafluorobenzene yielded 0.4% of aryl scrambling product [Scheme 10(b)]. Such phenomenon is not specific to DArP only and it has been reported for the Stille coupling[100] and Suzuki polymerization as well.[101] Naturally, ligandless Fagnou DArP conditions that do not employ phosphine ligands help to avoid such end group defects.

Scheme 10.

End-capping through aryl group transfer from phosphine ligand.

Residual Metal Defects


At this point, residual metal defects are the least studied class of defects and very little relevant information can be found in the literature regarding DArP. Originating from catalyst degradation during polymerization, the residual metal particles that are possibly bound to polymer chains represent a special kind of defect. Unlike the aforementioned homocoupling, branching, and end group defects, the residual metal defects are not covalently bound structural features. They are thought to be either physically entrapped between polymer chains or bound to polymer chains by the kind of interaction found in metal-ligand complexes.[102] They cannot be entirely removed by physical methods (extraction, filtration, ultracentrifugation, etc.)[102] and therefore here they are treated not as impurities but as defects in polymer structure.


These defects often go unnoticed when polymer is analyzed by common analytical techniques (NMR, UV-vis, FTIR, GPC etc.). However, the presence of residual metal defects have been clearly demonstrated to severely impact polymer's electroluminescent,[102] field-effect transistor,[103, 104] and OPV performance.[105] Since it is challenging to remove these defects using physical separation methods, attempts have been made to develop a process for effective removal of residual metal (palladium) via chemical treatment with palladium scavengers.[106] It is generally considered that the fewer metallic impurities the polymer contains, the better its performance in organic electronics. Therefore, when using DArP as a newly emerged method of polymerization, it is important to elucidate the levels of contamination with residual metals.


Towards this end, Reynolds et al. reported the synthesis of propylenedioxythiophene (ProDOT) containing copolymers via DArP as well as quantification of residual metal defects of the resulting polymers by ICP-MS (Scheme 11).[49] It has been shown that the element content for DArP polymer (Fagnou conditions) is less than 20 ppm for both palladium and phosphorus, whereas, the same polymer prepared via GRIM contained 1636 ppm of Mg and 926 ppm of Ni and the same polymer made by oxidative coupling with FeCl3 showed 1112 ppm of Fe. This illustrates that DArP is a promising polymerization method in terms of minimizing the residual metal defects.

Scheme 11.

Synthesis of ProDOT-based polymers via DArP, GRIM, and oxidative coupling polymerization.

In another recent report, Kanbara et al. compared the content of residual palladium and phosphorus in a copolymer of EDOT and 2,7-dibromo-9,9-dioctylfluorene prepared via by DArP with ligandless (phosphine-free) Fagnou conditions to the same polymer made by Suzuki polymerization, as illustrated in Scheme 12.[38] Depending on the DArP conditions, microwave heating or conventional heating, the content of residual Pd was 1590 or 2300 ppm, respectively, whereas residual phosphorus was not detected. This is in sharp contrast with the properties of the Suzuki polymer, which contained 4390 ppm of Pd and 470 ppm of phosphorus, thus indicating the attractiveness of ligandless DArP conditions. The lower content of residual Pd is thought to originate from lower loading of catalyst in the case of DArP, which uses 1% of catalyst as opposed to Suzuki polymerization, which uses 5%. In this regard, efforts have been directed to lower the amount of catalyst used during polymerization. As such, in an effort to improve upon previously reported DArP conditions[29] Ozawa et al. reported lowering the amount of palladium catalyst used for DArP by the factor of eight, using only 1% of Pd2dba3·CHCl3 (Pd(OAc)2 or PdCl2(MeCN)2 were also used) in combination with coordinating phosphine ligands and carboxylic acids, which actually results in high molecular weight polymer with high yields despite the lower catalyst loading [Scheme 8(b)].[25]

Scheme 12.

Synthesis of EDOT and fluorine-based copolymer via DArP and Suzuki coupling.


To summarize, in this highlight we draw the attention of conjugated polymer community to an often overlooked feature of defects in polymer structure. Together with other elements of polymer structure—the molecular weight and PDI—structural defects determine polymer properties and therefore their suitability for certain applications. The presence of defects in polymer chains is, in fact, a feature that is specific to polymer chemistry when compared to small molecule chemistry, where the products of side reactions can usually be separated. Therefore, just as any other method of polymerization, DArP produces defects in the polymer chains. Some of these defects are inherent for all existing methods of conjugated polymer synthesis (homocoupling, end group defects, residual metal defects), some are specific to DArP only (branching). When considering defects in polymer structure, it is important to understand their origin, their influence on polymer properties and to know the means to control the amount of defects.

Importantly, despite the negative tone of the term, defects may not necessarily be an undesirable feature of polymer structure. Indeed, for some optoelectronic applications a specific amount of certain defects is, in fact, beneficial. For example, for application in OPV it was demonstrated that a certain amount of homocoupling regiodefects in P3HT help create more stable morphology and more efficient devices.[77, 107] While the amount of β-defects—the class of defects specific to DArP—has to be minimized or completely suppressed for OPV,[87] for hyperbranched polymer applications, such as light-emitting devices, it has to be increased.[54, 108] The end-group functionalities also have to be precisely controlled. For example, for light-emitting devices application, it is important to eliminate halogen end groups.[109] That is why, the capability to install specific end-group through initiation and end-capping is an important feature of DArP. Lastly, the residual metal defects are generally considered detrimental for all electronic applications of conjugated polymers, and so far there was no need to deliberately increase the amount of residual metal on the polymer. However, it is possible that certain applications, for example those that utilize polymer–nanoparticle composites, will indeed require a specific amount of residual metal defects. In fact, attempts to incorporate palladium nanoparticles into conjugated polymers have already been made.[110]

Ultimately it is desirable that a synthetic method of polymerization has the capability to control, limit or produce certain amount of defects on demand and deliver polymers with a specific type and content of defects via simple tuning of reaction conditions, such as reaction temperature, catalyst loading, nature of ligand, etc. and without change in monomer structure or significant complication of polymerization procedure. In this regard, DArP demonstrates its full potential by not only being able to produce a huge variety of conjugated polymers in an environmentally benign, inexpensive and straightforward way, but also to allow for fine tuning of the highly relevant polymer feature—the structural defects. The future development of DArP should be focused on further simplification of the reaction conditions in terms of elimination or minimization of loading of expensive reagents (palladium-based catalysts), expanding the substrate scope and, importantly, gaining a better control over the polymerization through the establishment of a living variant of DArP.


This material is based upon work supported as part of the Center for Energy Nanoscience, an Energy Frontier Research Center funded by the US Department of Energy, Office of Science, Office of Basic Energy Sciences, under Award DE-SC0001013, specifically for partial support of A.E.R and B.C.T.


  • Image of creator

    Andrey E. Rudenko was born in Crimea, Ukraine in 1987. After finishing high-school in 2004 he moved to Saint-Petersburg, Russia to study chemistry at Saint-Petersburg State University. From 2004 to 2008, his research was focused on the transformation of substituted aziridines. In 2008, he joined the research group of Prof. Victor V. Sokolov where he participated in developing a novel approach to the synthesis of aromatic cyclopentylamines. After receiving his Diploma in Chemistry Andrey started his Ph.D. studies at the University of Southern California and joined the research group of Prof. Barry C. Thompson in December 2009. Currently, his research focuses on the development of the direct arylation polymerization (DArP) as well as synthesis of novel conjugated polymers.

  • Image of creator

    Barry C. Thompson was born in Milwaukee, Wisconsin in 1977. Barry then attended the University of Rio Grande in Rio Grande, Ohio, where he majored in Chemistry and Physics. After completing his undergraduate studies, Barry moved to the University of Florida to pursue a Ph.D. in Chemistry with Prof. John R. Reynolds as an NSF Graduate Research Fellow, focusing on conjugated polymers for electrochromic and photovoltaic applications. Upon completion of his Ph.D. in 2005, Barry moved to Prof. Jean Fréchet's lab at UC Berkeley to further pursue his interests in polymer-based photovoltaics as an ACS-PRF Postdoctoral Fellow. After a 3-year stay at Berkeley, Barry moved to the University of Southern California, Department of Chemistry and Loker Hydrocarbon Research Institute in 2008 as an Assistant Professor of Chemistry. Barry serves on the editorial advisory board for three journals and is a member of the Center for Energy Nanoscience (a DOE Energy Frontier Research Center at USC). His current research interests are focused on the design, synthesis, and application of novel conjugated polymers for organic photovoltaics.