Efficient Synthesis of Asymmetric Miktoarm Star Polymers

Abstract

Asymmetric miktoarm star polymers comprising an unequal number of chemically-distinct blocks connected at a common junction produce unique material properties, yet existing synthetic strategies are beleaguered by complicated reaction schemes that are restricted in both monomer scope and yield. Here, we introduce a new synthetic approach coined “μSTAR” — Miktoarm Synthesis by Termination After Ring-opening metathesis polymerization — that circumvents these traditional synthetic limitations by constructing the block–block junction in a scalable, one-pot process involving (1) grafting-through polymerization of a macromonomer followed by (2) in-situ enyne-mediated termination to install a single mikto-arm with exceptional efficiency. This modular μSTAR platform cleanly generates AB_n and A(BA′)_n miktoarm star polymers with unprecedented versatility in the selection of A and B chemistries as demonstrated using many common polymer building blocks: poly(siloxane), poly(acrylate), poly(methacrylate), poly(ether), poly(ester), and poly(styrene). The average number of B or BA′ arms (n) is easily controlled by the molar equivalents of macromonomer relative to Grubbs catalyst in the initial ring-opening metathesis polymerization step. While these materials are characterized by dispersity in n that arises from polymerization statistics, they self-assemble into mesophases that are identical to those predicted for precise miktoarm stars as evidenced by small-angle X-ray scattering experiments and self-consistent field theory simulations. In summary, the μSTAR technique provides a significant boost in design flexibility and synthetic simplicity while retaining the salient phase behavior of precise miktoarm star materials.

Graphical Abstract

Introduction

Block copolymers (BCPs) are important in a variety of emerging and established applications due to their self-assembly into well-ordered structures on the nanometer length scale.¹ The phase behavior of linear BCPs with two chemically-distinct blocks arrayed in simple sequences (AB, ABA, …) is now well-understood from both experiments^2–4 and theory^5–8 to depend on the Flory–Huggins interaction parameter (χ), volumetric degree of polymerization (N), block volume fractions (f_i, i = A, B), and conformational asymmetry (ε). These molecular design parameters dictate the self-assembly of two-component BCPs into a handful of classical phases⁴ (body-centered cubic spheres, hexagonally close-packed cylinders, a gyroid network, lamellae) and more exotic sphere packings^9–11 (e.g., σ, C14, C15, and A15). While the utility of many such mesophases is indisputable, linear chain connectivity imposes structure–property constraints that are not always desirable. For example, the number of known morphologies remains small,¹² domain periodicities are fairly restricted (typically within a range circa 5–100 nm),^13,14 and the coupling between volume fraction (f) and morphology favors the majority block on the convex side of curved block–block interfaces.^5,15 These (and other) limitations have motivated the search for new molecular design tools that broaden the utility of BCP self-assembly in contemporary applications.

An exciting opportunity that expands the confines of traditional polymer phase behavior¹⁶ lies in the controlled synthesis of BCPs with branched architectures.^17,18 The introduction of branching imparts useful thermodynamic,¹⁹ photonic,^20,21 and mechanical²² properties that are otherwise inaccessible with linear analogues. One archetypal example is miktoarm star polymers,^23–32 which are defined³³ as two or more chemically-distinct blocks connected to a common junction (e.g., A_mB_n, m + n > 2). The miktoarm star architecture is known or predicted to stabilize new phases,^9,15,34 reduce domain spacing,³⁵ and manipulate melt^25,26,36,37 or solution^38–40 properties, making them attractive for applications such as lithography and drug delivery. A further subset of miktoarm star polymers that accentuates the role of architecture in self-assembly involves asymmetry in arm number (m ≠ n); here, we focus on the limit m = 1, e.g., AB_n. As result of arm asymmetry, bulk phase boundaries are significantly deflected towards larger values of the A-block volume fraction (f_A).^5,15 This effect has been beautifully exploited by Lynd⁴¹ and Shi⁴² to design new thermoplastic elastomers (A(BA′)_n) that are stronger, stiffer, and tougher than commercial ABA linear triblock copolymers.

Despite the importance of miktoarm star polymers in contemporary polymer science, their synthesis still remains a major challenge. The standard approach to generate precise connectivity at a common junction uses some combination of “grafting-from” and “grafting-to” multi-step reaction schemes.⁴³ The need for orthogonal reactivity, high yields, and designer core molecules requires tedious synthetic routes that often include time-consuming coupling, polymerization, (de)protection, and purification steps such as fractional precipitation and high performance liquid chromatography.^36,44,45 For example, the materials studied by Shi and coworkers⁴² necessitated reaction times in excess of 30 days to push coupling to high conversion and still required purification via fractionation.^45–47 Moreover, changing the number of arms is non-trivial since a new core starting material must be selected each time.

Motivated by the difficulty of traditional miktoarm star polymer syntheses, we recently exploited the versatility, speed, and efficiency of Grubbs-type ring-opening metathesis polymerization (ROMP)^48–57 to synthesize miktoarm star polymers via the grafting-through copolymerization of two different macromonomers at low backbone degrees of polymerization (N_BB).⁵⁸ This type of statistical copolymerization is remarkably well controlled and the short backbone behaves physically like the core of a star polymer at N_BB ≲ 12 as evidenced by experiments and theory. However, simple copolymerization trades molecular precision for synthetic versatility since the reaction stoichiometry can only control the average number of arms and molecular composition. As a result, bulk phase behavior is dominated by dispersity effects that counteract phase boundary deflection, even in the case of nominally asymmetric architectures. Copolymerization therefore cannot generate the unique phase behavior that distinguishes asymmetric miktoarm stars from traditional block copolymers.

Here, we introduce a new synthetic method termed μSTAR (Table 1, top) — Miktoarm Synthesis by Termination After Ring-opening metathesis polymerization — that efficiently generates asymmetric miktoarm star polymers using ruthenium-catalyzed macromonomer polymerization (B → B_n) followed by in-situ enyne-mediated termination⁵⁹ to install the single A arm (B_n → AB_n). μSTAR sits at an optimal synthetic intersection, combining the versatility and speed of a macromonomer approach using ROMP with the precision of a highly efficient coupling step. Using a handful of macromonomers and macroterminators as building blocks, a diverse library of miktoarm stars can be easily prepared with different numbers of arms and block chemistries. We highlight this modularity by synthesizing AB_n and A(BA′)_n miktoarm star polymers comprising six different permutations of A and B block chemistry selected from poly(siloxane), poly(acrylate), poly(methacrylate), poly(ether), poly(ester), and poly(styrene). The average number of B arms (n) is easily controlled by the equivalents of Grubbs catalyst to macromonomer in the initial polymerization step. Importantly, the phase behavior of these polymers with disperse n exhibits significant phase boundary deflection, in agreement with self-consistent field theory (SCFT) simulations performed on precise (monodisperse n) analogues. A major implication of this finding is that the dispersity in n produced by μSTAR is advantageous from the perspective of significantly simplifying the synthesis of miktoarm star polymers while retaining the characteristic phase behavior that produces interesting bulk properties. The speed, efficiency, and broad scope of μSTAR establishes a compelling new synthetic platform for asymmetric miktoarm star polymers and supports the notion that low dispersity is not always better in block copolymer self-assembly.^60,61

Table 1.

(top) Generic μSTAR synthesis of miktoarm star polymers using norbornene-functionalized macromonomers and enyne macroterminators. (bottom) Macromonomers, macroterminators, and miktoarm star polymers synthesized in this work.

Results

Synthesis

Two types of simple linear precursors are needed in the μSTAR process to create miktoarm polymers: a macromonomer with a single polymerizable end-group and a macroterminator that will irreversibly couple exactly once to the active chain ends. We focus on using Grubbs-type ring-opening metathesis polymerization (ROMP) to construct the junction due to its well-established functional group tolerance, fast reaction rates, and high yields.⁶² Norbornene was therefore selected as the polymerizable group on the macromonomer because it undergoes efficient ROMP;⁶³ both homopolymer (B) and diblocks (BA′) will be discussed with norbornene installed on the B terminus. For the macroterminator, we exploit enyne-mediated termination chemistry recently developed by Gutekunst and coworkers⁵⁹ to perform macromolecular coupling of living metathesis polymers.^64,65 While enyne macroterminators were previously shown to efficiently prepare diblocks, the sterics involved in coupling to the core of a star polymer present a unique challenge.⁶⁶ Nevertheless, the high reactivity of enynes makes them suitable for macromolecular couplings that would otherwise not be possible with traditional ROMP termination methods employing substituted vinyl ethers or symmetrical cis-olefins.^67–71 The generic end-groups used in μSTAR are illustrated in Table 1 (top).

Macromonomers with different B chemistry were synthesized by polymerization from functional norbornene initiators or coupling reactions between a norbornene acid and commercially available monotelechelic polymers. In summary, six different macromonomers were synthesized that span various classes of polymer chemistry: poly(lactide) (PLA), poly(dimethylsiloxane) (PDMS), poly(4-methylcaprolactone–block–lactide) (PMCL-PLA), poly(styrene) (PS), poly(2-trifluoroethyl acrylate) (PTFEA), and poly(methyl methacrylate) (PMMA). Similarly, six macroterminators (A) were prepared by coupling to or directly growing from the enyne terminator molecule. PDMS, PLA, poly(ethylene oxide) (PEO), poly(n-butyl acrylate) (PnBA), and poly(tert-butyl acrylate) (PtBA) were chosen as the A block (see Supporting Information). The methyl ester enyne small molecule can be prepared in four high yielding steps and is further derivatized into the terminator of choice by one or two additional reactions.⁵⁹ Table 1 (bottom) summarizes these materials; full characterization details are provided in the Supporting Information (Tables S1–S2, Figure S1).

The efficacy of μSTAR at synthesizing asymmetric miktoarm star polymers is evident in Figure 1, which summarizes size-exclusion chromatograms (SECs) of the macromonomers (dashed lines), macroterminators (dashed lines), and resultant miktoarm star polymers (solid lines) for the combinations described in Table 1. The general process involves two steps that occur in one pot. (1) Polymerization of the macromonomer creates a short bottlebrush (N_BB < 12) with star-like physical properties⁵⁸ (vide infra); after complete conversion, an aliquot of the poly(macromonomer) is extracted. (2) In situ termination by the addition of macroterminator efficiently couples a single A arm to the living star polymer, resulting in AB_n or A(BA′)_n chain connectivity. SEC traces of the poly(macromonomers) are omitted from Figure 1 for clarity but can be found in Figures S2–S10. Note that with the exception of Figure 1f, n = 4 was targeted in this initial set of examples. Kinetic experiments performed with a model 5 kDa PLA macroterminator indicate the coupling process is finished in about 2 hours at room temperature (Figure S11). After termination, the increase in poly(macromonomer) absolute molecular weight as measured with multi-angle light scattering (MALS) is consistent with the macroterminator size (Table S3–S4). A single precipitation into methanol, diethyl ether, or hexanes is sufficient to isolate the final miktoarm star polymers, which have low molar mass dispersities (Ð < 1.2, Table S4) and monomodal SEC traces (Figure 1). ¹H nuclear magnetic resonance (NMR) measurements further confirmed the stoichiometric coupling of macroterminator and poly(macromonomer) (Figure S12–S22) and were also used to calculate compositions as tabulated in Table S4. Diffusion-ordered spectroscopy (DOSY) analysis revealed that these miktoarm star polymers lack homopolymer contamination within measurement error (Tables S5–S6, Figure S23–S24)⁷² as attempts to determine the percent of homopolymer contamination with multi-component fits yielded inconsistent results and non-physical diffusion coefficients, which is evidence of data overfitting.⁷³

Figure 1.

Size-exclusion chromatograms (normalized differential refractive index signal, dRI) of the miktoarm star polymers (solid black lines) listed in Table 1. Macromonomers and macroterminators are depicted with dashed lines. See the Supporting Information (Figure S2–S10) for traces of the poly(macromonomers), which were omitted here for clarity. In (d), the macroterminator trace represents poly(tert-butyl acrylate) before deprotection, while the final miktoarm star curve comprises poly(acrylic acid) after deprotection. Also note that the small bump near 14 min is small molecule elution. In (e), the PTFEA macromonomer and PTFEA4-PnBA samples have negative dn/dc values in THF; the dRI data were multiplied by −1 for the purpose of consistent presentation.

Another advantage of μSTAR is the ability to easily vary the average number of B or BA′ arms by changing the equivalents of macromonomer to Grubbs initiator. A series of four A(BA′)_n asymmetric miktoarm star polymers (A = PLA, BA′ = PMCL–block–PLA) with n = 3, 5, 7, or 9 arms was prepared simultaneously in separate reaction vessels using the same macromonomer and macroterminator precursors (Figure 2). SEC traces smoothly decrease in elution time as n increases, and absolute molecular weight measurements are consistent with increasing the average number of poly(macromonomer) arms across the range n = 3 – 9 (Table S3). This ability to easily vary the number of arms stands in stark contrast to all previous synthetic strategies where a different initiator or core must be synthesized whenever the number of arms is varied.^29,45,74 These materials also highlight the tolerance of μSTAR chemistry to high molecular weights; for n = 9, a 54 kDa poly(macromonomer) cleanly couples to a 24 kDa PLA macroterminator using only 1.1 equivalents of the latter.

Figure 2.

μSTAR can easily vary the average number of arms n in an asymmetric miktoarm star polymer. (a) Chemical structure of (PLA-PMCL)n-PLA with n = 3 – 9. (b) Normalized differential refractive index signal from SEC analysis of the isolated miktoarm star polymers. See Table S4 for a summary of molecular weights and dispersities.

Self-Assembly

A key question that remains is whether asymmetric miktoarm star polymers synthesized via μSTAR (with dispersity in n) self-assemble as predicted by theory for precise analogues. We have opted to study in detail the phase behavior of the (PLA-PMCL)_n-PLA samples described in Figures 1 and 2 since the addition of a short A′ block flanking B is predicted to further accentuate the phase boundary deflections that are characteristic of asymmetric AB_n miktoarm star polymers.⁴¹ Figure 3a reports synchrotron small angle X-ray scattering (SAXS) patterns collected at room temperature after annealing (PLA-PMCL)_n-PLA with a varying number of PLA-PMCL diblock arms (n = 3 – 9) at 140 °C for 18 hours. Note that the volume fraction of PLA (f_PLA) changes with n such that these samples span f_PLA = 0.58 – 0.71. The SAXS traces for n = 3 (f_PLA = 0.71), n = 5 (f_PLA = 0.68), and n = 7 (f_PLA = 0.62) can be cleanly indexed as indicated by triangles that demarcate the expected location of scattering reflections for lamellar (LAM), gyroid (GYR), and hexagonally close-packed cylinders (HEX), respectively. The n = 9 material shows broader peaks that are less well-defined, but their intensity maxima roughly coincide with those expected for a spherical form factor and Percus–Yevick structure factor⁷⁵ (Figure S25); we tentatively ascribe this morphology as disordered spheres that possibly fail to order on a well-defined lattice due to kinetic limitations. Collectively, these data are consistent with a remarkable deflection of order–order phase boundaries towards larger f_A relative to linear AB diblock or ABA triblock copolymers. For example, the HEX–GYR transition occurs near f_A = 0.3 with linear diblocks versus in the vicinity of f_A = 0.62 – 0.68 that we measure for (PLA-PMCL)_n-PLA mikto polymers. We are confident that the PLA block resides in the interior of the cylinders since GYR (f_PLA = 0.68) and LAM (f_PLA = 0.71) occur at even larger volume fractions. Perhaps stronger direct proof is the HEX sample exhibits recoverable elasticity in cyclic tensile tests, the details of which will be described in a forthcoming report. These experimental data relating morphology and volume fraction are in agreement with SCFT simulations performed on A(BA′)_n asymmetric miktoarm star polymers using the literature-reported⁷⁶ value of χ_PLA–PMCL and the degrees of polymerization measured experimentally for A = PLA and BA′ = PMCL-PLA (Figure 3b, see Supporting Information for details). We conclude that asymmetric miktoarm star polymers synthesized via μSTAR — which necessarily have dispersity in n — can self-assemble into structures that mimic precise molecular analogues.

Figure 3.

The phase behavior of (PLA-PMCL)n-PLA miktoarm star polymers containing dispersity in n is consistent with simulations of precise analogues. (a) Small-angle X-ray scattering data with triangles demarcating the expected location of Bragg reflections for lamellar (LAM, n = 3), gyroid (GYR, n = 5), and hexagonally close-packed cylinder (HEX, n = 7) morphologies. (b) SCFT simulations at τ ≡ NA/(NA + NA′) = 0.91 relating morphology, PLA volume fraction (fPLA), and the number of PMCL-PLA (BA′) diblock arms (n) at χN = 36, which corresponds to the segregation strength at 298 K.76 Superposed symbols represent the four experimental samples from part (a).

Discussion

Historically, anionic polymerization has been the workhorse synthetic technique used to construct miktoarm star polymers, including AB_n^{32,66,77–79} and A(BA′)_n^42,45 asymmetric variants. While effective, rigorous purification requirements, a limited monomer scope, sequence constraints, sluggish coupling kinetics⁴⁷ (that can take months to reach full conversion), and the need for additional purification by fractional precipitation^32,45,78 are inconvenient from both practical and design perspectives. μSTAR overcomes all of these challenges, assuming that dispersity in n can be tolerated, by exploiting the well-established functional group compatibility and speed of ROMP. We note that a conceptually similar approach has been attempted with anionic polymerization in the past, namely the grafting-through polymerization of a polystyrene (polyisoprene) macromonomer to construct the B_n core, either preceded or followed by the polymerization of polyisoprene (polystyrene) to grow a single A block.⁸⁰ The result was rather broad and multimodal SEC traces, particularly for the poly(macromonomers). Despite improvement after repeated fractional precipitation, even a further optimized anionic methodology would lack the versatility of a ROMP-based approach.

The examples in Figures 1 and 2 were selected to accentuate different types of chemistry that are of contemporary importance and challenging to link together using traditional miktoarm star syntheses. For example, PLA_n-PDMS (Figure 1a) and PDMS_n-PLA (Figure 1b) may be useful as lithographic materials with higher resolution than linear analogues due to architecture effects while maintaining good etch contrast.^{13,19,35,37,81,82} In the field of electrochemical energy storage, miktoarm star polymers containing PEO blocks are of interest as safe battery electrolytes, yet their reported synthesis is involved.³⁶ We have demonstrated that PS_n-PEO miktoarm stars are straightforward to synthesize with μSTAR (Figure 1c). μSTAR also provides access to amphiphilic miktoarm star polymers, e.g., by combining a PMMA macromonomer and PtBA terminator (Figure 1d) followed by acid-catalyzed deprotection of the tert-butyl ester to poly(acrylic acid) (PAA). The sulfonamide–pyrroline linkage created during termination is robust enough withstand a concentrated solution of trifluoroacetic acid and yield the partially charged PMMA_n-PAA star polymer (Figure S26–S29). Figure 1e further showcases a combination of acrylates (PTFEA₄-PnBA) that would be especially difficult to access via a core-first approach since both monomers undergo polymerization with the same type of radical initiator; the incorporation of semi-fluorinated acrylates may also create opportunities in surface coatings and other advanced materials.^83–86

As introduced earlier, the A(BA′)_n architecture presents exciting opportunities for next-generation thermoplastic elastomers.⁴² To date, this concept has only been explored using A, A′ = poly(styrene) (PS) and B = poly(isoprene) (PI) blocks synthesized by anionic polymerization and silyl chloride coupling.⁴⁵ Inspired by the work of Hillmyer,⁷⁶ here we have shown that renewable types of glassy (PLA) and rubbery (PMCL) polyesters can form A(BAʹ)_n miktoarm stars with n = 3 – 9 using μSTAR (Figure 1f), which are inaccessible via the established anionic route. The phase behavior of (PLA-PMCL)_n-PLA asymmetric miktoarm star polymers synthesized with μSTAR is consistent with past experimental reports on precise (PS-PI)₃-PS⁴² and theory that anticipate significant deflection of order–order transitions towards larger volume fractions due to molecular architecture. We have not observed this effect in any simple ROMP copolymerizations involving A and B macromonomers,⁵⁸ even at unequal feed compositions, which suggests that efficient termination chemistry (or some other method of installing a single A arm) is key to unlocking the unique self-assembly of asymmetric miktoarm star polymers. This result bolsters our previous finding that short bottlebrushes actually behave like miktoarm star polymers despite the inherent dispersity in n.⁵⁸ Note that SCFT simulations reveal a large sensitivity to the relative lengths of A and A′ blocks as parameterized by τ = N_A/(N_A+N_A′) (Figure S30). Although our experimental calculation of τ is based on molar masses measured by NMR (τ = 0.896) and MALS (τ = 0.925) that are within reasonable experimental uncertainty, SCFT simulations match the data in Figure 3a best with an intermediate τ = 0.91 shown in Figure 3b. SCFT also accurately captures the temperature-dependent phase behavior of these materials. By measuring the order–disorder transition temperature (T_ODT) with variable temperature SAXS (Figure S31) and calculating χ(T_ODT) from the relationship reported by Watts,⁷⁶ (χN)_ODT was compared to SCFT predictions. Incredibly, for n = 3, the theoretical and experimental values differ by less than 1% (Figure S32). As n increases, the deviation grows, but it never exceeds 12%.

We hasten to note that not all miktoarm star samples produced with μSTAR show scattering reflections that are as well-resolved as those in Figure 3. This may be the result of thermodynamic or kinetic factors that are influenced by architecture, dispersity, high molecular weight, or a combination thereof. For example, with n = 9 and f_A = 0.58 (Figure 3a), the thermodynamically stable phase might be A15,^15,87 which is likely kinetically inaccessible above a certain threshold molecular weight.⁸⁸ Another possibility is a complex free energy landscape; Grason and coworkers have previously argued that kinetic trapping could cause a similar glassy intermediate phase in AB₂ miktoarm stars due to the near degeneracy of BCC and A15.¹⁵ Thus, it is not surprising that complex sphere phase formation is suppressed.^10,88 Nevertheless, we find it remarkable that μSTAR can produce clean self-assembly given the dispersity in n.

Figure 4 illustrates the key differences in molecular composition and self-assembly that result from various miktoarm star synthesis techniques. Simple ROMP copolymerization with either a blocky or statistical sequence at low N_BB generates composition and arm-number dispersity that together tend to favor a flat block–block interface (Figure 4a).⁵⁸ At the same overall composition (i.e., f_A = 0.5), asymmetric miktoarm star polymers with a precise number of arms (for example, AB₃) bias interfacial curvature toward the A block (Figure 4b).^15,26 Samples synthesized using μSTAR sit somewhere in between — exactly one A arm and a distribution of B arms still results in self-assembly that favors interfacial curvature, the magnitude of which is evidently similar to precise analogues with the average μSTAR composition (Figure 4c). One benefit of incorporating such dispersity lies in relaxing the synthetic burden without drastically impacting self-assembly.

Figure 4.

Illustration of molecular composition and self-assembly resulting from different miktoarm star synthesis techniques. (a) Simple ROMP copolymerization of two macromonomers generates dispersity in composition and the number of A and B arms, which promotes flat block–block interfaces.58 (b) Asymmetric miktoarm stars (e.g., AB3) created by a precise synthesis favor interfacial curvature toward the A block.15 (c) μSTAR produces miktoarm stars with a distribution of B arms and exactly one A arm, resulting in interfacial curvature that is equivalent to precise analogues comprising the average molecular composition.

Conclusion

In summary, we have introduced a new synthetic technique termed μSTAR that generates AB_n and A(BA′)_n asymmetric miktoarm star polymers using grafting-through polymerization and efficient enyne-mediated polymer–polymer coupling chemistry. This modular approach is compatible with a wide variety of polymer chemistries and can accommodate high molecular weight arms. The average number of B or BA′ arms (n) is easily varied by the ratio of Grubbs catalyst to macromonomer in the initial polymerization step. Miktoarm star polymers made via μSTAR exhibit large deflections in the block copolymer phase diagram (relative to linear analogues) unlike stars produced by statistical grafting-through copolymerization. Despite the dispersity in n, experimental phase behavior matches SCFT calculations performed with precise molecular connectivity. μSTAR significantly simplifies the synthesis of asymmetric miktoarm star polymers when dispersity in arm number can be tolerated.