Upcycling aromatic polymers through C–H fluoroalkylation

This work provides a platform C–H functionalization method that introduces fluoroalkyl groups onto commercial aromatic polymers and post-consumer plastic waste that improve their material properties.


Introduction
Synthetic polymers that contain aromatic rings, including polystyrene, poly(ethylene terephthalate), polycarbonate, and high performance polymers, are omnipresent in modern materials. [1][2][3][4][5][6] The range of applications for these plastics is a testament to the unique properties imparted by planar and rigid aromatic moieties. A common approach to tune the bulk and/or surface properties of these materials is copolymerization, where a comonomer that contains the desired functionality is included in the polymerization. 6 Copolymerization of high-volume polymers, however, requires re-optimization of polymerization processes and can necessitate supplanting synthetic infrastructure on-scale, which is both cost-and laborintensive. 7 Furthermore, copolymerization strategies oen suffer from disparate monomer reactivity, resulting in anisotropic material properties.
Post-polymerization modication (PPM) of high-volume aromatic polymers is an appealing approach to modify material properties. 8 A general PPM strategy could diversify the properties of a wide-variety of commercial polymers while leveraging existing industrial infrastructure. Additionally, the development of PPM methods that increase the value of postindustrial and/or post-consumer aromatic polymers would provide a platform strategy for the upcycling of plastic waste. This upcycling approach is especially relevant for aromatic materials that are traditionally difficult and cost-prohibitive to mechanically recycle, such as expanded polystyrene (EPS) foam and high-performance polymers. [9][10][11][12] Exploiting the innate reactivity of aromatic polymers has the potential to introduce a wide array of chemical functionality while retaining the benecial properties of the parent polymer. Electrophilic radicals represent reactive intermediates that are well-known to undergo formal C-H functionalization with electron-rich or electron-neutral aromatic rings. 13 Despite a renewed interest in these species for the functionalization of small molecules, [14][15][16] general methods that enable functionalization of aromatic polymers using electrophilic radicals remains underexplored.
Electrophilic radical-mediated functionalization of aromatic polymers has historically focused on polymer halogenation. Radical bromination of polystyrene (PS) was rst reported by Staudinger and subsequently studied by a number of groups and developed into a commercial process. [17][18][19][20][21] Direct uorination of aromatic polymers using uorine gas was later reported by Lagow and coworkers to yield partially or peruorinated materials. 22 These and other halogen-centered radicals engage in a number of reactions with aromatic polymers, including hydrogen atom abstraction and aromatic substitution. The poor chemoselectivity of these approaches result in signicant polymer chain-cleavage and chain-coupling upon halogenation, thus compromising their otherwise attractive thermomechanical properties.
Functionalization of aromatic polymers by electrophilic peruoroalkyl radicals, however, has the potential to be a mild and chemoselective route for PPM. 23 The pyramidal ground state of the triuoromethyl radical and its low-lying singlyoccupied molecular orbital results in rapid addition to electron-rich alkenes or aromatic systems as opposed to participating in deleterious hydrogen-atom abstraction or radical recombination reactions. 13 In seminal work, Shuyama reported the uoroalkylation of poly(a-methylstyrene) using (peruoroethyl)phenyliodonium triate to achieve 24 per-uoroethyl groups per 100 repeat units, or 24 mol% functionalization. 24 Subsequently, Sawada and coworkers pioneered the use of peruoroacyl peroxides (PFAPs) as uoroalkyl radical precursors for the functionalization of a variety of aromatic substrates. 25 Notably, PFAP-mediated triuoromethylation of polystyrene resulted in 80 mol% functionalization. This approach, however, was accompanied by a coincident increase in the molecular weight distribution (MWD) of the polymer due to polymer chain coupling, with the dispersity (Đ) of polystyrene increasing from 1.06 to 2.41 at 80 mol% functionalization. 26 Furthermore, PFAPs must be formed and used immediately due to their rapid and sometimes violent decomposition at temperatures above 0 C. 27 A modular, tunable, and userfriendly method for the uoroalkylation of aromatic polymers that maintains the benecial properties of the parent material remains underdeveloped. 28 We envisioned an approach that utilizes electrophilic radicals not only to uorinate materials, but also to impart additional functionality onto commodity aromatic polymers to diversify their physical properties. The benets of uoroalkyl groups in medicinal chemistry have spurred the development of a variety of approaches for the generation of electrophilic uoroalkyl radical reactive intermediates. 13,[29][30][31][32][33][34][35][36][37] We identied a photocatalytic method developed by Stephenson and coworkers as an attractive system for polymer substrates. [38][39][40] This approach demonstrated mild reaction conditions, tolerance to a wide variety of functional groups, operationally simple reaction set-up, and low-cost reagents. Driven by the need for a chemoselective method to diversify the properties of commodity aromatic polymers, we sought to expand this photocatalytic strategy for uoroalkylation of macromolecules ( Fig. 1).
We report a platform approach for the C-H functionalization of aromatic polymers that enhances their physical properties through the use of electrophilic radical reactive intermediates. Photocatalytic generation of electrophilic uoroalkyl radicals under mild reaction conditions enables chemoselective polymer functionalization with no evidence of deleterious chain coupling or chain scission side reactions. The approach takes advantage of the innate reactivity of aromatic polymers, providing a platform for the functionalization of a variety of commercially valuable commodity polymers as well as post-industrial and postconsumer plastic waste. Electrophilic radicals generated from both short-and long-chain peruoroalkyl acid anhydrides enable a tunable density of uorine to be installed under operationally simple reaction conditions. Furthermore, installation of a bromodiuoromethyl group imparts functionality that serves as an initiator for atom transfer radical polymerization (ATRP), providing a facile approach for the chemical diversication of commodity polymers. 41 The versatile strategy described herein imparts valuable functionality onto otherwise recalcitrant commercial and waste polymers through C-H functionalization, providing opportunities to add value to these pervasive materials and upcycle post-consumer plastic waste.

Results & discussion
The photocatalytic method reported by Stephenson for the peruoroalkylation of small molecule substrates provided a conceptual framework to access uoroalkyl radicals. Translating the method to a successful polymer functionalization strategy, however, faced considerable hurdles. The modest yield for the triuoromethylation of non-functional substrates such as benzene and mesitylene indicates that not all uoroalkyl radicals are participating in productive reactivity. Additionally, the steric environment of a random coil polymer in solution could limit the extent of functionalization. To probe structurereactivity relationships for the uoroalkylation of aromatic polymers, we chose triuoroacetic anhydride (TFAA) due to its simple 19 F NMR signature and polystyrene as a model substrate due to its high-volume production and challenges in recycling streams.
PS substrate 1, designed and synthesized via ATRP to contain a triuoromethyl group at the chain end, was used for optimization studies. The chain end served as a polymerbound internal standard for quantication of tri-uoromethylation by 19 F NMR. The narrow dispersity of the model polymer (Đ ¼ 1.04) enabled the determination of subtle changes in the number-average molar mass (M n ) and Đ under the reaction conditions. Triuoromethylation of 1 was achieved by combining 1.0 equivalent of pyridine N-oxide and 1.1 equivalents of TFAA relative to repeat unit, and 1.0 mol% Ru(bpy) 3 Cl 2 relative to pyridine N-oxide in dichloromethane (DCM) at room temperature under 420 nm blue light irradiation. The reaction resulted in 32 AE 2 mol% triuoromethylation ( Fig. 2A). An analogous reaction performed in the dark resulted in no tri-uoromethylation of 1, and performing the reaction under irradiation without the Ru(bpy) 3 Cl 2 photocatalyst resulted in only 7 mol% triuoromethylation. The reaction performed equally well in the presence of oxygen or with rigorous exclusion of oxygen, which simplied reaction setup. A catalyst loading of 1.0 mol% proved to be optimal, as both decreasing and increasing the relative amount of Ru(bpy) 3 Cl 2 resulted in  19 F NMR spectra of the regiorandom trifluoromethylation of 1. Proposed peak assignments were determined by small molecule and copolymerization studies (Fig. S1 †). *The reaction was isolated from the experiment where X ¼ 5 and resubjected to the reaction conditions. decreased triuoromethylation. Alternative solvents and lower or higher concentrations resulted in inferior reaction performance. A complete reaction optimization table can be found in the ESI (Table S1 †).
Varying the stoichiometry of TFAA and pyridine N-oxide compared to repeat unit provided a tunable level of polymer C-H triuoromethylation. The addition of 0.5 equivalents of pyridine N-oxide and 0.55 equivalents of TFAA compared to repeat unit of 1 resulted in an average of 19 AE 2 mol% functionalization, while the addition of 5.0 and 5.5 equivalents, respectively, resulted in 71 AE 8 mol% functionalization ( Fig. 2A). Increasing reagent equivalents above 5.0 resulted in decreased mol% functionalization, likely due to separation of a TFAA phase from the rest of the reaction components. Resubmitting material with 71 mol% functionalization to the reaction conditions further increased triuoromethylation to an average of 108 CF 3 units per 100 styrene repeat units, which we attribute to the addition of two triuoromethyl groups to a single styrene repeat unit. This high density of functionalization is rarely observed in C-H functionalization reactions on commodity polymers 8 and speaks to the efficiency of the method.
Following triuoromethylation, gel permeation chromatography (GPC) demonstrated a shi to shorter retention times, indicating an increase in polymer molecular weight associated with a change in the hydrodynamic radius of the polymer upon functionalization (Fig. 2A). The Đ of the polymer changes very little even aer exposure to large excesses of the reagent, highlighting the chemoselective reaction conditions. This triuoromethylation approach is in contrast to previous work using PFAPs, which uniformly reported a signicant increase in Đ that scaled with the relative concentration of the acyl peroxide uorinating agent employed in the reaction. 19 F NMR provided quantitative evidence of polystyrene tri-uoromethylation. New peaks in the 19 F NMR spectrum aer the reaction demonstrated regiorandom polystyrene tri-uoromethylation (Fig. 2B). The synthesis of a number of well-dened copolymers and small molecules facilitated peak assignments (Fig. S1 †). Previous work has demonstrated that alkyl groups on an aromatic ring direct radical tri-uoromethylation ortho/para due to inductive effects. 37,38,42 We hypothesize that the broad singlet at À62 ppm encompasses both para and meta functionalization of a styrene repeat unit. The two broad peaks between À57 and À60 ppm represent ortho triuoromethylation, with the upeld-most peak being assigned to the ortho triuoromethyl group in a repeat unit that contains both ortho and para bistriuoromethylation. This hypothesis was conrmed through the synthesis and characterization of an electronically similar small-molecule analogue, 2 0 ,4 0 -bis(triuoromethyl)cumene.
As functionalization increases, the number of rings that include two triuoromethyl groups also increases (Fig. S2 †). Lastly, benzylic functionalization is also observed and corresponds to less than 4 mol% of the functionalized repeat units even at high reagent loadings (5.0 equivalents of pyridine N-oxide and 5.5 equivalents of TFAA) (Table S2 †).
Thermal properties of triuoromethylated polymers were investigated. Polystyrene 1 had a decomposition temperature (T D ), measured by thermal gravimetric analysis (TGA) where the polymer lost 10% of its initial mass, of 329 C. Tri-uoromethylation did not signicantly alter the T D of the materials, conrming the thermal stability of the functionalized polymers. The inuence of polystyrene triuoromethylation on the glass transition temperature (T g ) was analysed by differential scanning calorimetry (DSC), with data taken from the second heating cycle at a ramp rate of 10 C per minute. The T g of 1 lowered slightly from 84 C to 75 C at 45 mol% functionalization. Increasing the density of triuoromethylation to 63 mol% did not signicantly inuence thermal behaviour (T g ¼ 80 C). The application of Gordon-Taylor theory to previously measured values for the T g s of triuoromethylstyrene homopolymers ts qualitatively with the observed results (Table S3 †).
Aer successful polystyrene triuoromethylation, this methodology was expanded to other commercially valuable aromatic polymer substrates (Fig. 3). These reactions were run at 2.0 equivalents of pyridine N-oxide and 2.2 equivalents of TFAA compared to polymer repeat unit. For reference, the reaction achieved 45 mol% triuoromethylation of 1 under the same conditions. Functionalization reactions of poly(4methylstyrene) and poly(4-tert-butylstyrene) were successful, achieving 63 and 23 mol% triuoromethylation, respectively. Poly(4-tert-butylstyrene) demonstrated a lower mol% functionalization than PS under analogous conditions, presumably due to steric hindrance. Poly(4-methylstyrene), however, achieved a considerably higher degree of functionalization of 62 mol%, presumably due to the additional electron density of the aromatic ring. The introduction of electronegative functional groups to polymers made by step-growth polymerization is valuable due to the difficulty of incorporating electronically dissimilar monomers into a copolymerization. Polyesters and polycarbonates are the highest-volume polymers produced by stepgrowth polymerization; thus, representative substrates from these polymer classes were prioritized for tri-uoromethylation. 43,44 These reactions were run at 2.0 equivalents of pyridine N-oxide and 2.2 equivalents of TFAA compared to repeat unit. Exposure of bisphenol A-derived polycarbonate to the reaction conditions resulted in 10 mol% functionalization of the repeat units. Further, tri-uoromethylation of commercial polyesters including poly(ethylene terephthalate) (PET) and Eastman's Tritan® copolyester resulted in 14 and 20 mol% functionalization, respectively. For both polycarbonates and polyesters, we hypothesize that the lower mol% triuoromethylation compared to polystyrene is a result of the electron poor nature of the aromatic ring.
Following the successful C-H uoroalkylation of structurally diverse polymer substrates, the functionalization of postindustrial and post-consumer plastic waste was tested. Polystyrene was selected as a substrate for these studies due to the difficulties in recycling EPS foam. [10][11][12] Post-industrial EPS foam waste with a M n of 71 kg mol À1 and Đ of 2.92 generated as a by-product during production of foam packaging was procured. Post-consumer polystyrene with a M n of 75 kg mol À1 and Đ of 2.85 was secured from an EPS foam cooler that would otherwise have been disposed. Both post-industrial and post-consumer polystyrene were exposed to the reaction conditions at 2.0 equivalents of pyridine N-oxide and 2.2 equivalents of TFAA. These materials demonstrated analogous reactivity and mol% functionalization to a sample of pristine polystyrene and had no signicant change in MWD aer functionalization (Fig. S6 and S7 †). These experiments prove that the formulation, processing, and use of postindustrial and post-consumer EPS foam does not have a substantial inuence on polymer C-H uoroalkylation. These results further display the chemoselectivity and efficiency of functionalization by electrophilic radicals and indicate the potential for C-H uoroalkylation to upcycle post-consumer plastic waste.
With photocatalytic C-H triuoromethylation established on a variety of polymer substrates, the generality of the method to incorporate longer peruoroalkyl groups was investigated. Commercially available pentauoropropionic anhydride and heptauorobutyric anhydride served as uoroalkylating agents in place of TFAA (Fig. 4A). Functionalization of 1 using 1.0 equivalent of pyridine N-oxide and 1.1 equivalent of the shortchain peruoro anhydride proceeded to produce polymers achieving 45 mol% peruoroethylation and 37 mol% per-uoropropylation. To covalently attach peruorochains of longer length, commercially available peruorooctanoyl chloride was employed (Fig. 4B). Use of this reagent successfully added peruoroheptyl groups to polystyrene, albeit at lower efficiency than for shorter-chain peruorocarbons, achieving 16 mol% at 1.0 equivalent of pyridine N-oxide and 1.1 equivalents of the acyl chloride.
The conceptual approach of C-H functionalization using electrophilic radicals also allows the incorporation of functional groups capable of further synthetic manipulations. Specically, we targeted the incorporation of chlorodiuoromethyl and bromodiuoromethyl groups onto aromatic polymers due to the versatile chemistry of the halomethyl groups (Fig. 4A). 40 The commercially available acetic anhydrides were used for the functionalization of 1, resulting in group transfer of the corresponding halodiuorocarbon onto polystyrene. Using 1.1 equivalent of the anhydride and 1.0 equivalent of pyridine Noxide to repeat unit, chlorodiuoromethylation (26 mol%) proceeded with similar efficiency to triuoromethylation (32 mol%), however the bromodiuoromethylation was less efficient (15 mol%) due to the reduced electrophilicity of the bromodiuoromethyl radical.
To demonstrate the potential of halodiuoromethyl groups as a functional handle, the bromodiuoromethylfunctionalized polystyrene was explored for further chemical diversication. We hypothesized that the lower bond dissociation free energy (BDFE) of the carbon-bromine bond would enable selective functionalization through radical chemistry. The diuoromethyl group is of emerging interest in medicinal chemistry due to its ability to act as a hydrophobic hydrogenbond donor, [45][46][47][48] but its potential has rarely been studied in polymer science. [49][50][51][52] Bromine-atom abstraction initiated by azobisisobutyronitrile (AIBN) in the presence of tributyltin hydride as a hydrogen atom transfer reagent led to quantitative conversion of the bromodiuoromethyl group to a diuoromethyl group. Characterization by 19 F NMR demonstrated an upeld shi of the peak representing the diuoromethyl group aer reaction, indicating 98% conversion to the desired functionality (Fig. S30 †).
The low BDFE of the carbon-bromine bond and the previous use of bromotriuoromethane as an initiator for free radical polymerization led us to hypothesize that the bromodi-uoromethyl group could serve as a polymer-bound ATRP initiator. [53][54][55] To test this hypothesis, a well-dened polystyrene was made using reversible addition-fragmentation chain transfer (RAFT) polymerization. The trithiocarbonate end group was removed and the polymer was functionalized to contain 7.8 mol% bromodiuoromethyl groups (2, M n ¼ 13 kg mol À1 , Đ ¼ 1.19) (Fig. 5A).
The ATRP graing of methyl acrylate (MA), tert-butyl acrylate (tBuA), and poly(ethylene glycol) acrylate (PEGA) from polymer 2 was studied. Using 0.5 equivalents of CuBr relative to the bromodiuoromethyl initiator and 1.0 equivalent of PMDETA as a ligand at a 1 : 1 v/v ratio of monomer to anisole at 90 C, gra copolymers of poly(styrene-gra-tBuA), poly(styrene-gra-MA), and poly(styrene-gra-PEGA) (3) were produced (Fig. 5A). The kinetic data indicates well controlled polymerizations, demonstrating a linear increase in monomer conversion over time (Fig. 5C and S8-S10 †). Additionally, the 19 F NMR indicates full initiation of each CF 2 Br appended to polystyrene (Fig. 5B). TGA and DSC characterization further indicate gra copolymer products (Fig. S11 †). Poly(styrene-gra-tBuA), for example, showed the diagnostic thermal expulsion of isobutylene from poly(tBuA) beginning at 188 C by TGA, with polymer decomposition occurring at 406 C. The DSC data reveals that the T g decreases as the poly(tBuA) sidechains increase in molar mass, lowering from 79 C with 3.6 repeat units per side chain to 50 C with 24 repeat units per side chain (Fig. S11 †). These T g s are between that of polystyrene and poly(tBuA) and further conrm the covalent connection between the two otherwise immiscible polymers. To further demonstrate the versatility of these gra copolymers, the tert-butyl groups were cleaved through acid treatment to produce the amphiphilic poly(styrene-gra-acrylic acid) copolymers.
An advantage of polyaromatic C-H uoroalkylation is its potential to modify the surface and/or interfacial properties of commodity polymers. 33,56-62 Fluorinated polymers are well known to demonstrate high surface energy, and covalent bonding of peruoroalkyl chains to a material is an attractive approach to prevent the leaching of physically absorbed per-uoroalkyl coatings into the environment. 63,64 The static contact angle of water and n-hexadecane was used to determine qualitative changes in surface energy upon polymer C-H uoroalkylation. To measure contact angles a 20 wt% solution of functionalized polystyrene in toluene was spin coated onto glass slides, annealed at 140 C for 30 min, and the contact angle was measured by goniometry. For each peruoroalkyl group, the static water contact angle increased from that of the native polystyrene 1 (94 ) (Fig. 6A). We found that approximately 30 mol% functionalization of the polymers was sufficient to impart properties signicantly different from that of polystyrene, and further functionalization did not alter the surface properties. As the chain-length of the peruoroalkyl group increased, the static water contact angle increased in a commensurate fashion up to 111 for peruoroheptylated polystyrene (Fig. 6A). In addition to discovering water repellent polymers, uoroalkylated polystyrene are also more omniphobic. While n-hexadecane fully wets a PS surface, the contact angle of n-hexadecane on a 64 mol% peruoropropylated surface is 37 . Conversely, the ability to gra PEGA from bromodiuoromethyl-functionalized polystyrene enabled the realization of hydrophilic materials. For example, poly(styrene-gra-PEGA) copolymer becomes water soluble at high conversions of the PEG-containing gra polymer. At low conversion, the gra copolymer made lms that demonstrated hydrophilic surface properties, with a static water contact angle of 84 (Fig. 6B).

Conclusion
The work presented herein provides a platform C-H functionalization method that introduces uoroalkyl groups onto commercial aromatic polymers. The applicability of this method to a wide variety of commercially available uoroalkyl anhydrides and acyl chlorides, its broad substrate scope for both commodity and post-consumer aromatic polymers, and its utility for diversifying the properties of traditionally recalcitrant materials makes it attractive for applications in materials science and upcycling plastic waste.

Conflicts of interest
There are no conicts to declare.