The backbones of conjugated and heteroatom-linked aromatic polymers tend to possess fewer conformational degrees of freedom than polymers with more flexible aliphatic or partially aliphatic backbones. This reduced amount of conformational freedom can help enhance the folding of aromatic polymers, to advance a variety of useful properties such as selective supramolecular recognition (Goodman et al., 2007; Liu et al., 2015; Otte, 2016; Schneebeli et al., 2016; Adhikari et al., 2017; Xie et al., 2020), selective catalysis (Rajappan et al., 2020; Sharafi et al., 2020), and self-assembly (Cole et al., 2017; Greene et al., 2017; Bonduelle, 2018; Ong and Swager, 2018; Delawder et al., 2019; Zhao et al., 2019). However, while the precise chemical structures, lengths, and sequences of such macromolecules (Dobscha et al., 2019; Gerthoffer et al., 2020; Zhao et al., 2020) dictate their folding, and with it their functionalities and physical properties (Chen et al., 2015; Hanlon et al., 2017), it remains challenging to synthesize polyaromatic structures with precise lengths and/or sequences as unimolecular entities. One of the most efficient ways to precisely control the length and sequence of synthetic polymers is by iteratively coupling (Jones et al., 1997) polymer strands together in a convergent/divergent fashion (Hawker et al., 1997; Read et al., 2000; Grayson and Frechet, 2001; Li et al., 2005; Liess et al., 2006; Binauld et al., 2011). This methodology (Figure 1) is generally referred to as iterative exponential growth (IEG) (Barnes et al., 2015; Leibfarth et al., 2015).

IEG requires a dormant monomer that includes two distinct functionalities, which are orthogonally deprotected/activated for coupling, to afford two reactive species that can be coupled together selectively. These two reactive species are then combined in a chemo-selective manner, resulting in a dormant dimer, which contains, again, protected/masked functionalities identical to the original dormant monomer (Figure 1). Repeating this simple protocol results in the creation of architecturally defined structures with an exponential gain in polymer length. The IEG method has been applied extensively to synthesize flexible linear polymers (Barnes et al., 2015; Huang et al., 2017)—for example, via copper-catalyzed azide–alkyne click (CuAAC) reactions. However, it still remains a challenge to efficiently utilize arylation chemistry with an IEG-reaction framework, as is required in order to access conjugated or heteroatom-linked polyaromatic polymers with IEG methodology. In particular, while transition metal–catalyzed arylation reactions have been utilized previously to synthesize conjugated and/or heteroatom-linked polyaromatic macromolecules (Figure 1), (Zhang et al., 1992; Louie and Hartwig, 1998; Sadighi et al., 1998; Liess et al., 2006), there are (to the best of our knowledge) currently no transition metal-free reactions available to achieve exponential polymer growth for conjugated or heteroatom-linked polyaromatic structures. We are now able to meet this synthetic challenge by marrying iterative exponential polymer growth with nucleophilic aromatic substitution (S_NAr) reactions (Beugelmans et al., 1994; Blaziak et al., 2016; Landovsky et al., 2019) with a unique masking/unmasking technique. Unmasking was achieved by simply oxidizing aromatic sulfides to sulfones—a mild, simple, and scalable reaction well suited for IEG applications. At the same time, the IEG coupling reactions also proceeded under mild conditions (mild heating to ∼50°C), which represents an advantage over transition metal–catalyzed alternatives [which tend to require stronger heating (Sadighi et al., 1998)].

Our new synthetic technology for precision polymer growth helps resolve a number of concerns, which exist with traditional, transition metal–catalyzed cross-coupling methodology (Sun and Shi, 2014; Li et al., 2020). These issues include, but are not limited to 1) the inherent toxicity of many transition metal catalysts, which goes hand in hand with the need to accomplish/demonstrate complete removal of any residual transition metal, especially for healthcare-related applications (which can be difficult to achieve with polymers containing many Lewis basic groups like the ones presented in this article), 2) the relatively high cost of the required transition metal catalysts and complex ligands, which are often needed for high-yielding cross-coupling reactions suitable for polymer growth. Transition metal-free cross-coupling protocols like the one presented in this article have the potential to overcome these fundamental challenges associated with transition metal–catalyzed polymer growth.

To implement our sulfide oxidation–driven IEG coupling strategy, we first tested a variety of sulfone leaving groups for their ability to couple with phenolate anions in our IEG scheme (for related work, see: Guan et al., 2015). We discovered that simple methyl sulfones (Figure 1) are best suited for this purpose; as with sulfone substituents larger than methyl (e.g., phenyl), we tended to get lower IEG coupling yields. For our synthetic scheme, we decided to protect the phenols with methyl groups (which can readily be removed in high yield with BBr₃). However, it is very likely that instead of just methyl groups, alternative phenol protection groups (e.g., benzyl) can also be also be utilized in the future to further enhance the functional group tolerance of our IEG coupling scheme. To start the IEG couplings, we first condensed 3-methoxyphenol (1) with 4-chloro-6-(methylthio)pyrimidine (2) in quantitative yield under standard S_NAr conditions with K₂CO₃ as the base (Van Rossom et al., 2012). The resulting dimeric product 4-(3-methoxyphenoxy)-6-(methylthio)pyrimidine (3) was then split in half for the next IEG deprotection/activation/coupling cycle. The first aliquot was subjected to BBr₃ for methoxyl deprotection, while the second aliquot was activated via oxidization to the methyl sulfone with mCPBA. Following these deprotection/activation steps, we were able to perform an S_NAr-based IEG coupling, again in the presence of K₂CO₃ as the base. Ultimately, by iteratively implementing (Scheme 1) our new, transition metal-free IEG methodology, we were able to access macromolecules with up to 16 perfectly alternating AB units—where A and B represent resorcinol and pyrimidine rings—for the first time in a fully unimolecular fashion.

We hypothesized that the electron-deficient pyrimidine units in our unimolecular oligomers might be able to promote folding (Scheme 1, inset) via hydrogen bonding of and/or π-stacking with the electron-rich resorcinol units. To investigate this hypothesis, we grew single crystals suitable for single-crystal X-ray diffraction analysis of compound 3 by slow vapor diffusion of hexanes into ethyl acetate solutions of 3. The packing observed (Figures 2A,B) upon analysis of the diffraction pattern of a single crystal of compound 3 clearly demonstrates the ability of the positively polarized -N=CH-N= hydrogens in the pyrimidine rings to form intermolecular hydrogen bonds with the methoxyl group of another molecule of 3 in the solid state. There are other hydrogen bonding interactions that are weaker, but still significant, that also contribute (Supplementary Table S1) to the supramolecular packing of this structure. To estimate the strength of the [pyrimidine–CH^…OCH₃] hydrogen bonds observed in the crystal structure of 3, we optimized a model dimer—1,3-dimethoxyphenol in complex with 4-methoxy-6-(methylthio)pyrimidine—with density functional theory at the B3LYP (Lee et al., 1988; Becke, 1993)/6-31G** level of theory. Next, we used the noncovalent interactions (NCI) code (Johnson et al., 2010) implemented in the Jaguar (Bochevarov et al., 2013) software package to find all major attractive supramolecular interactions present in our model dimer, characterized as critical points of the electron density. The NCI critical points of the DFT-optimized structure of the model dimer (Figure 2C) clearly show the presence of a [pyrimidine–CH^…OCH₃] hydrogen bond, with a strength of ∼5.8 kcal/mol.

Having established that the pyrimidine rings are attracted to the resorcinol units in the solid state via [CH^…O] hydrogen bonding, we next embarked on investigating the ability of the polymer 6—the longest unimolecular, alternating pyrimidine/resorcinol polymer synthesized to date—to fold, driven by [CH^…O] hydrogen bonding interactions. To demonstrate folding, we recorded a ¹H-¹H NOESY NMR spectrum of the hexadecamer 6 in CDCl₃ at 298 K. The ¹H-¹H NOESY NMR spectrum displays (Figure 3A) a NOE cross peak between the proton resonance x of the terminal methoxyl group and the -NCHN- proton resonances (b ¹–b ⁷, which all overlap at the same resonance frequency at 8.48 ppm). Notably, this NOE cross peak is absent in the ¹H-¹H NOESY NMR spectrum of the dimer 3 (Supplementary Figure S1), which demonstrates that folding and/or supramolecular aggregation (and not just simply the fact that the proton corresponding to resonance b ⁷ is present in a pyrimidine ring next to the terminal methoxyl group) is responsible for the observed NOE cross peak. Furthermore, a ¹H DOSY NMR spectrum of 6 (Supplementary Figure S10) confirmed that the compound does not aggregate significantly in solution, and thus the NOE cross peak observed must be stemming from intramolecular folding of the polymer chains. A folded model of the hexadecamer 6 (optimized with DFT at the B3LYP-D3/6-31G** level) is shown in Figure 3B. The structure shown represents the lowest energy conformation obtained from a MacroModel conformational search (carried out without constraints and the OPLS3e force field (Harder et al., 2016) of the hexadecamer 6. While other potential, folded structures exist, the observed NOE cross peak is consistent with the DFT-optimized structure shown in Figure 2B, which shows a [CH^…O] hydrogen bonding contact between a pyrimidine CH functionality and the terminal methoxyl group in 6. For property characterization, we also performed differential scanning calorimetry (DSC) of 6 (see Supplementary Figure S11 for the DSC thermogram). As 6 is still a relatively low molecular weight compound and fully unimolecular, we did not observe any phase transitions in the temperature range investigated (30–335°C).

Overall, our results demonstrate that methyl sulfones can be effective leaving groups for transition metal-free, iterative exponential growth processes. Furthermore, the electrophilic pyrimidine cores present in our polymers not only help further enhance the yields for the S_NAr-based IEG coupling reactions but also provide polarized C–H bonds, capable of directing the folding of the unimolecular alternating pyrimidine–resorcinol polymers presented in this work.

This work demonstrates a unique IEG technique, which employs nucleophilic aromatic substitution (S_NAr) reactions as a tool to fabricate architecturally defined oxygen-linked ABAB polymers with a metal-free coupling strategy. The monomers are designed to exhibit both nucleophilic and electrophilic character, either of which can be accessed selectively via deprotection of methoxyl groups and via oxidation of aromatic sulfides. Integrating S_NAr reactions with iterative exponential growth eliminates the need for expensive transition metal catalysts, which have been required so far in order to carry out iterative exponential growth of conjugated and/or unimolecular, heteroatom-linked aromatic polymers. Our transition metal-free IEG methodology allowed us to access heteroatom-linked, unimolecular aromatic polymers with up to 16 alternating resorcinol and pyrimidine units, enabling us to investigate, for the first time, the [CH^…O] hydrogen bond–driven folding of such polymers with single-crystal X-ray crystallography, ¹H-¹H NOESY NMR spectroscopy, and density functional theory (DFT). Overall, our new IEG methodology advances fundamental capacity to access new unimolecular polymers needed to investigate the diverse physical and electronic properties of heteroatom-linked polyaromatic systems. We are currently applying our new transition metal-free IEG methodology to longer precision polymers with extended solubilizing chains.