Phosphonic Acid-Functionalized Diblock Copolymer Nano-Objects via Polymerization-Induced Self-Assembly: Synthesis, Characterization, and Occlusion into Calcite Crystals

: Dialkylphosphonate-functionalized and phosphonic acid-functionalized macromolecular chain transfer agents (macro-CTAs) are utilized for the reversible addition − fragmentation chain transfer (RAFT) dispersion polymerization of benzyl methacrylate (BzMA) at 20% w/w solids in methanol at 64 ° C. Spherical, worm-like, and vesicular nano-objects could each be generated through systematic variation of the mean degree of polymerization of the core-forming PBzMA block when using relatively short macro-CTAs. Construction of detailed phase diagrams is essential for the reproducible targeting of pure copolymer morphologies, which were characterized using transmission electron microscopy (TEM) and dynamic light scattering (DLS). For nano-objects prepared using the phosphonic acid-based macro-CTA, transfer from methanol to water leads to the development of anionic surface charge as a result of ionization of the stabilizer chains, but this does not adversely a ﬀ ect the copolymer morphology. Given the well-known strong a ﬃ nity of phosphonic acid for calcium ions, selected nano-objects were evaluated for their in situ occlusion within growing CaCO 3 crystals. Scanning electron microscopy (SEM) studies provide compelling evidence for the occlusion of both worm-like and vesicular phosphonic acid-based nano-objects and hence the production of a series of interesting new organic − inorganic nanocomposites.


■ INTRODUCTION
It is well-known that phosphorus-based polymers have a wide range of potential applications. They bind metals ions, 1,2 adsorb strongly onto metal oxide surfaces 3 and biominerals, 4,5 can serve as proton exchange membrane material, 6 allow the design of halogen-free flame-retardant materials, 7−9 and can exhibit excellent biocompatibility. 10,11 Literature examples include organophosphorus polymers based on methacrylate-type 12,13 or acrylamide-type 14 dialkylphosphonates. Anionic polyelectrolytes are also accessible directly either by polymerization of the free acid form of phosphate, 15 phosphonate, 6,13 or phosphinate 16 monomers or via hydrolysis of dialkylphosphonate polymers. 13,14 DNA is a particularly important example of a naturally occurring anionic polyphosphate. 17 However, phosphorus-based polymers can also exhibit cationic (e.g., phosphonium-based polymers 18 ) or zwitterionic character (e.g., poly(2-(methacryloyloxy)ethyl phosphorylcholine), or PMPC). 19,20 Both materials are of considerable interest in the field of biomaterials. 21,22 Recent progress in the development of controlled/living polymerization methods such as atom transfer radical polymerization (ATRP) 23 and reversible addition−fragmentation transfer (RAFT) polymerization 24 has facilitated the design of a wide range of well-defined controlled-structure polymers. Many examples of well-defined PMPC-based diblock copolymers have been prepared by ATRP 25,26 and RAFT. 27,28 Of particular relevance to the present study, (1-ethoxycarbonyl)vinyl dimethylphosphate was homopolymerized via ATRP and also copolymerized using either polystyrene or poly(methyl methacrylate) macroinitiators. 29 While control over the polymerization of this dialkylphosphate monomer was rather poorincomplete conversions or broad molecular weight distributions (M w /M n > 1.60) were obtained depending on the choice of catalystsubsequent hydrolysis produced the corresponding phosphoric acid-based diblock copolymers. In contrast, RAFT polymerization of either 2-(acryloyloxy)ethyl phosphate or 2-(methacryloyloxy)ethyl phosphate yielded welldefined diblock copolymers when employing 2-(acetoacetoxy)ethyl methacrylate as the second block. 30 Divinyl impurities present in both phosphate monomers limited the molecular weight of the first block to 20 000 g mol −1 , otherwise extensive cross-linking was observed.
As far as we are aware, vinylphosphonic acid (VPA) was the first example of a monomer bearing an unprotected phosphonic acid to be polymerized directly via RAFT polymerization. 31 As yet, however, only diblock copolymers with relatively low molecular weights have been obtained using this approach. Considerably higher molecular weights have been reported for the ATRP of 4-vinylbenzyl diethylphosphonate and relatively narrow molecular weight distributions were achieved in this case. 6 Both deprotection (to generate the free phosphonic acid groups) and chain extension (from a polystyrene macroinitiator) were demonstrated for this styrene-type monomer. 32 Similarly, well-defined homopolymers and diblock copolymers were obtained when using 2-(acrylamido)ethyl diethylphosphonate, with subsequent hydrolysis producing the corresponding free phosphonic acid-based polymers. 14 Another phosphorus-based monomer, methacryloyloxymethyl dimethylphosphonate (MP) has recently become commercially available, but attempts to polymerize this monomer via ATRP have only resulted in rather low conversions and relatively low molecular weights, presumably due to strong copper complexation. 33 Optimizing the ATRP formulation by appropriate selection of the initiator/catalyst system and polymerization temperature was required in order to improve control over the MP polymerization. 34 In this context, RAFT polymerization offers a potentially decisive advantage, since it involves an organosulfur-based chain transfer agent rather than a transition metal catalyst. Moreover, phosphonic acid-based homopolymers were recently prepared directly via RAFT polymerization. 13 In principle, a wide range of functional diblock copolymers incorporating either dialkylphosphonate or phosphonic acid repeat units should be accessible using this approach.
Recently, RAFT polymerization has also been exploited in the context of polymerization-induced self-assembly (PISA), which is a highly efficient and versatile method for directly generating diblock copolymer nano-objects at relatively high solids (up to 50% w/w). 35−38 In such formulations, a soluble macromolecular chain transfer agent is extended with a monomer to form a new block, which becomes increasingly insoluble in the reaction medium. The first soluble block confers steric stabilization, while the growing insoluble block drives in situ self-assembly. Varying the relative block volume fractions while conducting such syntheses at various copolymer concentrations allows predictive phase diagrams to be constructed, which are essential for the reproducible generation of pure phases comprising spherical, worm-like, or vesicular diblock copolymer nanoparticles. 39 This PISA approach has been shown to be generic. Various formulations have been reported in aqueous solution, 40−47 lower alcohols, 48−56 nalkanes, 57−59 toluene, 60 and solvent mixtures 61,62 that provide access to not only non-ionic (or zwitterionic) nanoparticles 43,63−66 but also anionic 67,68 or cationic nano-objects. 69 All of these aforementioned PISA formulations are based on dispersion polymerization, whereby the core-forming monomer is soluble in the reaction medium. There are also numerous examples of heterogeneous formulations based on aqueous emulsion polymerization. 70−77 There are very few literature examples of PISA syntheses involving phosphorus-based monomers. Such formulations combine the growing interest in phosphorus-based polymers with straightforward access to a range of diblock copolymer nano-objects without requiring post-polymerization processing. Potential applications include flame-retardant coatings, drug delivery formulations, and biocompatible worm gels for 3D cell growth. In 2011, our group reported aqueous formulations based on a zwitterionic PMPC macromolecular chain transfer agent (macro-CTA) for the preparation of diblock copolymer nano-objects. 43,66 Very recently, Zhang et al. employed 4diphenylphosphinostyrene as the core-forming monomer to prepare catalytically-active micellar nanoreactors. 78 However, we are not aware of any examples involving anionic phosphatebased monomers. As mentioned earlier, Monge and co-workers recently reported the RAFT solution polymerization of methacryloyloxymethyl dimethylphosphonate (MP) and methacryloyloxymethylphosphonic acid (MPA), 13 which are of potential interest for biomedical applications. In the present work, we describe the PISA synthesis of dialkylphosphonateand phosphonic acid-functionalized diblock copolymer nanoobjects using these two monomers. A range of nano-objects are prepared using a RAFT alcoholic dispersion polymerization formulation and characterized in terms of their size, morphology, and surface chemistry. Occlusion of selected worm-like and vesicular nanoparticles within growing CaCO 3 crystals is also briefly studied as a potential route to novel organic−inorganic nanocomposites.
Synthesis of Poly(methacryloyloxymethyl dimethylphosphonate) [PMP] Macro-CTA. The RAFT polymerization of MP was conducted as reported by Canniccioni et al. 13 using a 25% w/w MP solution in DMF and a CPDB/AIBN molar ratio of 3.0. In a typical protocol, the monomer (22.50 g, 108.1 mmol), CPDB (1.813 g, 4.9 mmol), and AIBN (266.3 mg, 1.6 mmol) were dissolved in 73 mL of DMF in a 250 mL round-bottomed flask. After purging with nitrogen for 20 min, the flask was sealed and placed in a preheated oil bath at 70°C for 3.5 h, before quenching the reaction by cooling with an ice bath. Samples were taken during the course of polymerization for 1 H NMR measurements, and the final conversion was determined to be 65%. The crude PMP was purified via precipitation into excess cold diethyl ether (twice) and finally freezedried from aqueous solution overnight. A mean degree of polymerization of 24 (or M n = 5.2 kg mol −1 ) was calculated via end-group analysis using 1 H NMR spectroscopy by comparing the integrated intensity of the two aromatic protons of the CTA at 7.42−7.56 ppm with the five protons assigned to the methacrylic backbone at 0. form, PMP was hydrolyzed with trimethylsilyl bromide (TMSBr). 13 In a typical reaction, PMP 24 (10.0 g) was dissolved in anhydrous DCM (100 mL), followed by addition of TMSBr (4 equiv per phosphonate group). The reaction mixture was stirred for 3 h at 20°C, and then the solvent was removed under reduced pressure. The free phosphonic acid homopolymer, PMPA, was generated by addition of methanol (500 mL) followed by stirring for 1 h at 20°C. After concentration under reduced pressure, the crude polymer was purified via dialysis against water, followed by freeze-drying overnight. Aqueous GPC analysis of the three PMPA homopolymers using 30% methanol cosolvent at pH 9 gave the following data: PMPA 24 M n = 10.1 kg mol −1 , M w /M n = 1.03; PMPA 32 M n = 11.1 kg mol −1 , M w /M n = 1.03; PMPA 42 M n = 12.6 kg mol −1 , M w /M n = 1.05.
Chain Extension Experiments with PMPA Macro-CTAs. Macro-CTAs were chain-extended with glycerol monomethacrylate (GMA) to examine their blocking efficiencies. RAFT solution polymerizations were conducted at 20% w/w solids at 64°C for 16 h. In a typical protocol, the PMPA 24 macro-CTA (140.5 mg, 31.22 μmol), AIBN (1.0 mg, 6.24 μmol, CTA/AIBN molar ratio = 5.0), GMA (1.00 g, 6.24 mmol), and 1,3,5-trioxane (11.3 mg, 124.87 μmol, GMA/1,3,5-trioxane molar ratio = 50) were dissolved in methanol (4.0 mL). The solution was purged with nitrogen for 20 min, and then the sealed vial was placed in a preheated oil bath at 64°C for 16 h, before quenching the reaction by cooling with an ice bath. Samples were taken during the course of the polymerization for 1 H NMR measurements. For the three different blocking experiments, final GMA monomer conversions were determined to lie between 46 and 73%. The resulting PMPA−PGMA diblock copolymers were precipitated twice from cold diethyl ether and freeze-dried from aqueous solution overnight prior to analysis by aqueous GPC (with 30% methanol cosolvent at pH 9): PMPA 24  Diblock Copolymer Synthesis via Alcoholic Dispersion Polymerization. Alcoholic RAFT dispersion polymerizations were conducted with both the dialkylphosphonate (PMP) macro-CTAs as well as the phosphonic acid (PMPA) macro-CTAs at 20% w/w solids under identical conditions. In a typical synthesis targeting a diblock composition of PMPA 24 −PBzMA 300 , the protocol was as follows: PMPA macro-CTA (37.5 mg, 8.32 μmol), AIBN (0.30 mg, 1.67 μmol, CTA/AIBN molar ratio = 5.0), and BzMA (440 mg, 2.50 mmol) were weighed into a sample vial and dissolved in pre-degassed methanol (1.91 g). This vial was then sealed with a septum and purged using a gentle nitrogen flow for 10 min while cooling in ice to minimize evaporation. Polymerization was initiated by placing the vial in a preheated oil bath at 64°C. In order to guarantee a high BzMA conversion, the polymerization was conducted at this temperature for 24 h before quenching by exposure to air. In the case of the alcoholic dispersion polymerization syntheses conducted using the PMP macro-CTAs, the final PMP−PBzMA diblock copolymers were also analyzed using DMF GPC (see Tables 1 and 2

for details).
CaCO 3 Precipitation via the Ammonia Diffusion Method. 79 For the crystallization experiments, 50 μL of a 1.0% w/w aqueous copolymer dispersion was diluted with water (2.50 mL) and a 3 mM CaCl 2 solution (2.50 mL) in a 30 mL sample vial, giving a final 0.01% w/w copolymer dispersion in a 1.5 mM CaCl 2 aqueous solution. A glass slide that had been pre-cleaned with piranha solution was placed on the base of the vial, and then the samples were transferred to a desiccator previously charged with (NH 4 ) 2 CO 3 powder (5.0 g). Crystallization was allowed to proceed for 24 h at 20°C, and then the glass slides were removed from solution, washed with deionized water and ethanol, and finally air-dried. A second set of crystallization experiments was also conducted using a 0.005% w/w aqueous copolymer dispersion. Control experiments were conducted in the absence of added copolymer.
Characterization. NMR Spectroscopy. 1 H NMR spectra were recorded in d 6 -DMSO for PMP, PMPA and their related diblock copolymers using either a 400 or 250 MHz Bruker spectrometer. A spectrum of the PGMA homopolymer was also recorded in CD 3 OD. 31 P NMR spectra for the PMP and PMPA macro-CTAs were recorded in d 6 -DMSO using a 250 MHz Bruker spectrometer (400 scans).
Gel Permeation Chromatography (GPC). GPC analyses of PMP and PGMA homopolymers and PMP−PBzMA diblock copolymers were conducted using two Polymer Laboratories PL gel 5 μm mixed C columns and one PL polar gel 5 μm guard column arranged in series and maintained at 60°C, followed by a Varian 390 LC refractive index detector. The DMF eluent contained 10 mM LiBr, and the flow rate was 1.0 mL min −1 . Calibration was achieved using a series of nearmondisperse poly(methyl methacrylate) standards (ranging from 645 to 2.48 × 10 6 g mol −1 ). Molecular weight distributions for the PMPA and PMPA−PGMA diblock copolymers were assessed using aqueous GPC, which comprised an Agilent 1260 Infinity setup fitted with two Agilent PS Aquagel-OH 8 μm columns at 35°C and a refractive index detector. The eluent was a 10 mM NaH 2 PO 4 pH 9 buffer containing 200 mM NaNO 3 and 30% methanol at a flow rate of 1.0 mL min −1 . Calibration was achieved using a series of near-monodisperse poly(methacrylic acid) standards ranging from 1.27 to 4.83 × 10 5 g mol −1 .
Dynamic Light Scattering (DLS) and Zeta Potential Measurements. Intensity-average hydrodynamic diameters of the dispersions were obtained by DLS using a Malvern Zetasizer NanoZS instrument. This instrument detects scattered light at an angle of 173°and was equipped with a 4 mW He−Ne laser operating at 633 nm, an avalanche photodiode detector with high quantum efficiency, and an ALV/LSE-5003 multiple τ digital correlator electronics system. Copolymer dispersions of ∼0.20% w/w solids were analyzed using the cumulants method to obtain the hydrodynamic (z-average) diameter and polydispersity index (PDI). Aqueous electrophoresis studies were performed on 0.01% w/w copolymer dispersions using the same instrument equipped with DTS1070 cells. The solution pH was adjusted by the manual addition of 0.01−1.0 M HCl or NaOH, and the background electrolyte was 1 mM NaCl.
Transmission Electron Microscopy (TEM). Copolymer particle morphologies were assessed using a Philips CM 100 instrument operating at 100 kV and equipped with a Gatan 1k CCD camera. For sample preparation, 8 μL of a diluted copolymer dispersion (∼0.20% w/w) was dropped onto a freshly glow-discharged grid, left for 30 s, and then blotted with filter paper to remove excess solution. Subsequently, staining was performed for 30 s using a 8 μL droplet of 0.75% w/v uranyl formate solution, followed by blotting the excess stain and drying with a vacuum hose. TEM grids were prepared by coating copper/palladium TEM grids (Agar Scientific, UK) with a thin film of amorphous carbon.
Optical Microscopy, Scanning Electron Microscopy (SEM), and Raman Microscopy. The dimensions and morphologies of the CaCO 3 (calcite) crystals were assessed using optical microscopy and scanning electron microscopy. Optical microscopy images of crystals grown on glass slides were recorded using a Motic DMBA 300 digital biological microscope with a built-in camera and analyzed using Motic Images 2.0 ML software. For SEM studies, the glass slides supporting the CaCO 3 crystals were mounted on SEM stubs using adhesive pads. Imaging of the uncoated samples was performed using a FEI Nova NanoSEM 450 operating at 2 kV. Cross sections of the CaCO 3 crystals were also imaged using SEM to investigate the extent of particle occlusion. Such samples were prepared by placing a clean glass slide on top of the glass slide coated with CaCO 3 crystals and pressing down while slightly twisting the upper glass slide. This protocol resulted in fracture of the crystals, hence revealing their interior structure. Characterization of the crystal polymorph was conducted via Raman microscopy studies of individual particles using a Renishaw Raman 200 System microscope operating at a laser wavelength of 785 nm.

■ RESULTS AND DISCUSSION
Synthesis of Poly(methacryloyloxymethyl dimethylphosphonate) and Poly(methacryloyloxymethyl phosphonic acid) Macro-CTAs. Methacryloyloxymethyl dimethylphosphonate (MP) was used for the synthesis of the macro-CTA in order to prepare phosphonic acid-stabilized nano-

Macromolecules
Article objects via RAFT dispersion polymerization. The RAFT solution polymerization of this commercially available monomer was recently studied using 2-cyano-2-propyl dithiobenzoate as chain transfer agent. 13 The resulting PMP macro-CTA can subsequently be converted into PMPA via ester hydrolysis (see Scheme 1). Three near-monodisperse PMP macro-CTAs with DPs of 24, 32, or 42 were synthesized in DMF at 70°C. The corresponding DMF GPC traces are shown in Figure 1a; M w /M n values were below 1.25 in each case. Subsequent hydrolysis in dichloromethane with trimethylsilyl bromide and methanol yields the corresponding PMPA in its free phosphonic acid form, as indicated by a shift from 21.6 to 14.0 ppm in the 31 P spectrum (Figure 1b and Figure S1). Because of unfavorable column interactions during attempted DMF GPC analysis, the PMPA macro CTAs had to be characterized using aqueous GPC (see Figure S2). These analyses indicated relatively low M w /M n values (less than 1.05).
Polymerization Induced Self-Assembly (PISA) in Aqueous and Alcoholic Media. First, the performance of the PMP macro-CTA as non-ionic stabilizer block for the aqueous dispersion polymerization of HPMA was investigated at pH 5. This facilitates comparison with later experiments conducted with the anionic PMPA macro-CTA, which is singly ionized at this pH. 13 Unfortunately, stable colloidal dispersions could not be obtained when targeting a PHPMA block with a degree of polymerization of 200 at 70°C using ACVA initiator. This behavior was rather unexpected, as each of the PMP macro-CTAs exhibited excellent water solubility and no LCSTlike behavior was observed. PISA syntheses were also attempted at 50°C using AIBA as a low temperature radical source, 64 but again precipitation was observed. Since we have recently developed various robust PISA formulations based on the RAFT alcoholic dispersion polymerization of BzMA, 49,50,55,56 we switched to using either ethanol or methanol as the continuous phase. However, coagulation was again obtained for the former solvent when targeting DPs of 200. It is perhaps noteworthy that complete dissolution of the steric stabilizer macro-CTA required stirring for around 1 h regardless of its molecular weight, which suggests that ethanol is perhaps a rather marginal solvent for PMP. Consequently, we selected methanol as a more polar solvent, which led to significantly faster PMP dissolution under the same conditions.
For kinetic studies of PISA syntheses conducted in methanol, an initial BzMA polymerization was conducted at 64°C for 24 h at 20% w/w solids (see Scheme 1) using a PMP 24 macro-CTA and targeting a final PBzMA block DP of 300. No precipitation was observed for this formulation, which produced a turbid dispersion. More than 99% conversion was attained after 24 h as judged by 1 H NMR studies, with almost complete monomer consumption (>97%) being achieved within 13 h (Figure 2a). Moreover, an enhanced rate of polymerization is only observed relatively late in the polymer-izationwell after micellar nucleation has occurred (see inset in Figure 2a). This differs markedly from previous observations for similar PISA formulations based on RAFT alcoholic dispersion polymerization, where the increase in the rate of propagation coincides with the onset of micellar nucleation. 50,80 A tentative explanation for this observation might be the relatively small size of the spherical nanoparticles initially formed during this PISA formulation (D h < 25 nm, see Table  S1). This suggests the presence of rather loose, hydrated aggregates and thus perhaps relatively inefficient initial solubilization of the BzMA monomer within the nascent nuclei. For PISA syntheses conducted under appropriate conditions, the copolymer morphology is known to evolve from spheres to vesicles via various intermediate morphologies, including worms and "jellyfish". 81 When the vesicle dimension (i.e., mean diameter and/or mean membrane thickness) reaches a certain critical value, these aggregates are able to solubilize BzMA. Thus, a significant increase in the rate of BzMA polymerization occurs at around 6 h, as seen in the third stage in the first-order kinetics plot.
Despite this unusual kinetic behavior, the PISA formulation proved to be both efficient and reasonably well-controlled in terms of both copolymer morphology and molecular weight distribution. DLS studies of aliquots extracted during polymer- ization indicate that vesicles of around 160 nm are formed within 6 h, which suggests that the observed rate enhancement is associated with this morphology, rather than the initial spherical nuclei (see inset in Figure 2a and Table S1). These vesicles eventually attain a final hydrodynamic diameter of approximately 180 nm. A post mortem TEM image of the final copolymer morphology obtained after 24 h is shown in Figure  S3b; the vesicular morphology is confirmed, and the estimated mean vesicle diameter is in reasonably good agreement with the DLS studies. Given the non-ionic nature of the PMP macro-CTA used in this set of experiments, DMF-GPC analysis of the corresponding PMP 24 −PBzMA x diblock copolymers was possible. Selected chromatograms are plotted in Figure 2b. The unimodal GPC traces are symmetric and show no significant tailing due to prematurely terminated macro-CTA chains, indicating high blocking efficiencies. This is supported by the linear evolution of the number-average molecular weight, M n , while maintaining relatively narrow molecular weight distributions, with M w /M n values typically below 1.16 ( Figure  S3a and Table S1). These observations are consistent with the anticipated pseudo-living character of this RAFT dispersion polymerization. In addition, a high molecular weight shoulder gradually becomes more prominent above 30% conversion. Similar observations were also reported for another alcoholic PISA formulation utilizing BzMA as the core-forming block. 50 Using various PMP x macro-CTAs (x = 24, 32, and 42) and performing all BzMA polymerizations at 20% w/w solids in methanol, we were able to construct a detailed phase diagram by systematically varying the target DP for the core-forming PBzMA block from 50 to 300 (see Figure 3; details of the individual polymerizations are provided in Table 1). The corresponding GPC traces for all of these PMP−PBzMA syntheses are shown in Figure S4 and clearly indicate an evolution of the molecular weight with increasing target PBzMA DP. High blocking efficiencies and relatively low M w /M n values were obtained, where these are similar to the data discussed earlier for the kinetic study conducted when targeting PMP 24 −PBzMA 300 . However, a relatively low concentration of unreacted macro-CTA can be detected for the PMP 42 macro-CTA, which suggests that some degree of premature termination occurred in this particular case. Moreover, the final molecular weight distribution was significantly broader (M w /M n = 1.53) for PMP 42 −PBzMA 558 compared to all other polymerizations (M w /M n < 1.21). As   Table S1 for a more complete set of GPC data).

Macromolecules
Article discussed earlier, a high molecular weight shoulder becomes discernible when targeting PBzMA DPs greater than 100 for all PMP macro-CTAs. In previous studies involving a poly(2hydroxypropyl methacrylate) core-forming block, a similar shoulder was explained in terms of a relatively low level of dimethacrylate impurity in the monomer, which inevitably leads to light branching when targeting higher DPs. 42 However, this explanation seems to be unlikely for benzyl methacrylate, since there is no reason for such a monomer to contain a dimethacrylate impurity. An alternative explanation may be some degree of termination by combination, which is not unknown for methacrylic monomers. 82 For all three sets of PMP macro-CTAs the initially formed spheres were relatively small (D h < 25 nm), as previously observed for a PGMA−PHPMA PISA formulation. 42 Inspecting the phase diagram shown in Figure 3, it is clear that only spherical nanoparticles can be obtained when using the PMP 42 macro-CTA. Presumably, this longer block confers additional  steric stabilization during the BzMA polymerization and hence prevents sphere−sphere fusion, which is the key first step in the generation of worms and vesicles. Systematically increasing the target PBzMA DP from 50 to 558 leads to a monotonic increase in particle size, with hydrodynamic diameters ranging from 16 to 76 nm, as evidenced by both DLS and TEM (see Table 1 and Figure S5). The larger spheres are believed to be kinetically trapped morphologies. 39 We recently reported similar findings for a PHPMA−PBzMA RAFT alcoholic dispersion polymerization formulation, when employing a PHPMA macro-CTA with a DP of 63. 50 However, using shorter PMP stabilizer blocks (DP = 24 or 32) enables access to higher order morphologies as well as spheres. Such rich phase behavior was also observed by Zehm and co-workers for the PHPMA−PBzMA PISA formulation on reducing the mean DP of the stabilizer block. 50 In the present study, using the PMP 24 macro-CTA only enabled access to pure spheres, vesicles, or mixed phases (Figure 3a,d). However, pure worms occupying a very narrow phase region could be obtained when employing the PMP 32 macro-CTA (Figure 3b). Representative TEM images obtained for the entire PMP 32 −PBzMA y series are shown in Figure S6. Increasing y from 48 to 61 led to partial fusion of spherical nanoparticles to form a mixture of spheres and short worms, with the latter having a distinctive "pearl necklace" appearance. Increasing y by a further seven BzMA units produced longer worms, with only a few spheres present. When y = 79, a free-standing gel comprising well-defined worms was obtained. Longer core-forming blocks (89 < y < 135) produced mixed phases, with vesicular phases becoming increasingly common. Finally, a pure vesicular phase was attained for PMP 32 −PBzMA 146 (and higher y values). DLS studies suggest that increasing the DP of the PBzMA core induces some contraction in the mean vesicle dimensions (from D h = 158 nm for y = 146 to D h = 136 nm for y = 300). A similar size reduction was observed for the PMP 24 −PBzMA y series, where y = 100−300, with hydrodynamic diameters decreasing from 352 to 181 nm (see Table 1). This may be caused by thickening of the vesicle membranes exclusively via inward growth. TEM images recorded for the various PMP 32 −PBzMA y PISA formulations provide some evidence for this hypothesis. For mixed phases consisting of worms and vesicles, the vesicle membrane is clearly visible, and the vesicles are rather polydisperse (see Figure S6; y = 100, 120, and 135). On further increasing y, the vesicles become somewhat less polydisperse, and their associated lumen volumes are reduced (see Figure S6; y = 146, 200, and 300). Minimization of the vesicular interfacial area (and hence free energy) is the driving force for this phenomenon, which was recently described in detail for an aqueous PISA formulation. 83 On the other hand, it is not understood yet why vesicles of apparently increasing diameter were observed after approximately 6 h during the kinetic studies (inset in Figure 2a).
We then examined the use of anionic PMPA macro-CTAs as putative steric stabilizers for the RAFT aqueous dispersion polymerization of HPMA. According to Semsarilar and coworkers, strong electrostatic repulsion between adjacent polyelectrolytic stabilizer blocks can impede in situ selfassembly during attempted aqueous PISA syntheses. 67−69 We investigated this hypothesis for our system by targeting a HPMA DP of 300; such an asymmetric diblock copolymer might be expected to form vesicles. However, it proved impossible to obtain pure vesicles for various PISA formulations at pH 5, even when utilizing binary mixtures of non-ionic poly(glycerol monomethacrylate) (PGMA 45 ) with an anionic PMPA x macro-CTA in the presence of added salt to screen the unfavorable electrostatics. Table S2 and Figure S7 summarize this series of experiments, for which a detailed protocol is given in the Supporting Information.
In view of these negative results, RAFT alcoholic dispersion polymerization of BzMA was performed in methanol at 20% w/ w solids using each of the PMPA macro-CTAs in turn (see Scheme 1). For this series of experiments, it was envisaged that the significantly lower dielectric constant of methanol (ε = 32.6 at 298 K) compared to water (ε = 78.5 at 298 K) 84 might reduce the unfavorable electrostatic repulsive forces and hence facilitate PISA. A PMPA macro-CTA with a mean DP of 24 was chosen for a kinetic study of the dispersion polymerization of BzMA at 64°C. When targeting a DP of 300 for the PBzMA block, greater than 99% conversion was achieved after 24 h, as judged by 1 H NMR (Figure 4a). A significant rate enhancement was observed, again not at the onset of micellar nucleation but instead when spherical micelles first begin to fuse to form higher order morphologies (see inset in Figure 4a). The

Macromolecules
Article apparent first-order rate constant, k app , for this second stage was a little higher when using the anionic PMPA 24 stabilizer (k app = 8.4 × 10 −5 s −1 ) compared to the non-ionic PMP 24 macro-CTA (k app = 5.5 × 10 −5 s −1 ). This suggests higher partitioning of the nonpolar BzMA monomer within the growing micelle cores in the former case. DLS data are shown for all samples in Figure  4a (see inset) and summarized in Table S3. A representative TEM image of the final vesicular morphology is depicted in Figure 4b: the estimated mean vesicle diameter of approximately 510 nm (based on analysis of 80 particles) is in reasonably good agreement with that indicated by DLS, bearing in mind that the latter technique reports a hydrodynamic diameter and is biased toward larger particles. Unfortunately, DMF GPC analysis of the various PMP 24 −PBzMA y diblock copolymers was not feasible because of unfavorable copolymer−column interactions. Moreover, the hydrophobic character of the PBzMA block precluded aqueous GPC analysis. In view of these problems, chain extension experiments were conducted for each of the three PMPA macro-CTAs in turn using GMA to produce a water-soluble PGMA block, simply in order to assess blocking efficiencies when polymerizing a second methacrylic monomer. Aqueous GPC traces obtained for these PMPA x −PGMA y diblock copolymers are compared to the respective PMPA macro-CTAs in Figure S2. In each case symmetric monomodal curves were obtained, with pronounced shifts to higher molecular weight relative to the original PMPA macro CTA. All three PMPA−PGMA diblock copolymers exhibited relatively narrow molecular weight distributions (M w / M n < 1.07). This indicates high blocking efficiencies for the solution polymerization of GMA. Given that the pseudo-living character of RAFT dispersion polymerization is enhanced compared to solution polymerization, 85 at least comparable (and most likely higher) blocking efficiencies are anticipated under the former conditions.
Using the three PMPA macro-CTAs (DP = 24, 32, or 42) to target a range of core-forming PBzMA blocks enables the construction of a predictive phase diagram ( Figure 5). A summary of the individual PISA syntheses conducted at 20% w/w solids in methanol at 64°C is shown in Table 2. In contrast to the phase diagram for the PMP−PBzMA formulation, nonspherical morphologies can be observed even for the PMPA 42 macro-CTA. At first sight, this might seem surprising given the anionic nature of the stabilizer chains, which might be expected to impede fusion of the initial spheres (and indeed hinder PISA itself). 68,69 However, electrostatic repulsive forces are substantially reduced in methanol   Table S2), which do not readily provide access to pure copolymer phases even when employing binary mixtures of anionic and non-ionic macro-CTAs. The phase diagram for the PMPA x −PBzMA y formulations shown in Figure 5 contains a relatively broad vesicle phase and a somewhat narrower mixed phase region. Spheres are accessible up to a target PBzMA DP of 35 (Figure 5b), while pure worms can only be obtained when using the PMPA 42 macro-CTA ( Figure 5a). Furthermore, the anionic PMPA 24 −PBzMA 300 vesicles shown in Figure 5d (D h from DLS = 529 nm) are significantly larger than the corresponding non-ionic PMP 24 −PBzMA 300 vesicles (D h from DLS = 181 nm). In principle, a higher packing parameter might be expected to favor the formation of larger vesicles. 86 This therefore suggests that the hydrodynamic volume occupied by an anionic PMPA 24 chain in methanol is less than that of a nonionic PMP 24 chain, indicating reduced solvation of the former stabilizer under these conditions. Similarly, increasing the PMPA stabilizer DP for a given core-forming PBzMA DP reduces the mean vesicle dimensions, presumably because of the larger hydrodynamic volume occupied by the PMPA chains (Figure 5c). The PMPA 42 −PBzMA y series provides access to all three copolymer morphologies, with representative TEM images being shown in Figure S8. As for the non-ionic PMP−PBzMA formulations, there is an apparent reduction in vesicle dimensions when targeting higher degrees of polymerization of the PBzMA block, and these observations are supported by DLS studies (see corresponding entries in Table  2).
Generation of Organic−Inorganic Hybrid Materials. It is well recognized that naturally occurring biopolymers are intimately involved in the formation of biominerals, 87 where these can not only direct features such as polymorph and morphology but also enhance mechanical properties through occlusion within the crystal lattice. 88,89 Importantly, this behavior is not restricted to biomacromolecules, and recent work has shown that a range of additives, including submicron latex particles, various anionic diblock copolymer nano-objects, and even small molecules, can become occluded within single crystals of calcite (CaCO 3 ). [90][91][92][93][94][95]98 In all cases, successful occlusion is dependent on the additives binding to the surfaces of the growing crystals but remaining dispersed in the crystal growth solution. 93 The PMPA−PBzMA nano-objects synthesized here provide interesting candidates for controlling CaCO 3 precipitation.
Both phosphate-and phosphonic acid-functionalized surfactants and polymers are well-known to interact strongly with CaCO 3 and are widely used as crystal growth inhibitors. 96,97 The efficiency of occlusion of these additives within calcite was therefore investigated, and the results compared with the corresponding non-ionic PMP−PBzMA nano-objects.
The nano-objects were first transferred from the methanol continuous phase to water via dialysis. TEM analysis of selected nano-objects before and after dialysis indicated no adverse effect on the copolymer morphology (Figure 6), where these findings were supported by DLS studies which showed no change in the mean particle diameter before and after dialysis. These observations are consistent with the relatively high T g of around 55°C for the core-forming PBzMA block, which suppresses exchange of copolymer chains between kinetically frozen nano-objects.
In addition to the successful transfer of vesicles into aqueous media (PMPA 32 −PBzMA 300 , PMPA 24 −PBzMA 300 , and PMP 24 −PBzMA 300 ), an aqueous worm dispersion was also obtained after dilution of a methanolic worm gel followed by dialysis (PMPA 42 −PBzMA 45 , see Figure 6). Aqueous electrophoresis studies of these anionic nano-objects confirmed highly anionic zeta-potentials of −35 to −45 mV across a wide pH Figure 6. (a) Representative TEM images recorded for non-ionic PMP 24 −PBzMA 300 vesicles and anionic PMPA x −PBzMA y worms and vesicles, obtained via RAFT dispersion polymerization of BzMA at 20% w/w solids in methanol at 64°C. (b) TEM images for the same copolymer nanoparticles after their transfer from methanol to water via exhaustive dialysis at 20°C. In all cases the copolymer concentration used to prepare the TEM grids was 0.2% w/w.

Macromolecules
Article range (see Figure 7). Only in relatively acidic media (pH 2) was a modest reduction in zeta potential observed, as expected for a steric stabilizer based on a strong polyacid such as PMPA. Concomitant DLS studies indicated no significant change in particle dimensions from pH 2 to 10, suggesting that colloidal stability was maintained (see Figure 7).
Occlusion experiments were conducted by precipitating CaCO 3 in the presence of non-ionic PMP 24 −PBzMA 300 vesicles and the various anionic PMPA x −PBzMA y nano-objects at pH ≈ 9 via a one-pot protocol using the ammonia diffusion method. 79 All experiments were conducted using a stock solution of 1.5 mM CaCl 2 and a copolymer concentration of either 0.010% w/ w (Figure 8a) or 0.005% w/w (Figure 8b). All dispersions remained colloidally stable in the presence of the CaCl 2 solution, which was added at an initial pH of less than 6. Rhombohedral calcite crystals of 10−15 μm were prepared in control experiments performed in the absence of diblock copolymer nano-objects. Figure S9 shows a typical Raman spectrum recorded for calcite crystals grown under these conditions, with the characteristic calcite bands being observed at 155 and 282 (lattice modes), 710 (ν 4 ), and 1085 cm −1 (ν 1 ). 93 Formation of calcite in the presence of non-ionic PMP 24 −PBzMA 300 vesicles at both additive concentrations produced crystals with comparable habits and dimensions to those formed in the control experiments. SEM studies confirmed that the crystals were perfect rhombohedra with well-defined edges, with just a few weakly interacting vesicles located on the crystal surface. These findings were not unexpected, since anionic character is believed to be a prerequisite for successful occlusion. Raman microscopy studies of such crystals formed when using 0.010% w/w copolymer confirmed that the polymorph was indeed calcite ( Figure S9).
In contrast, the anionic nano-objects had a stronger influence on the morphology and dimensions of the calcite crystals. Welldefined rhombohedral crystals of 15−20 μm were formed in the presence of 0.005% w/w PMPA 24 −PBzMA 300 and PMPA 32 −PBzMA 300 vesicles (DLS studies indicated mean diameters of approximately 520 and 210 nm, respectively).

Article
At the higher copolymer concentration, 20−25 μm crystals with increasingly rough faces and edge truncations were obtained. Again, Raman studies indicated calcite formation under these conditions ( Figure S9). Anionic PMPA 42 − PBzMA 45 worms had an even more pronounced effect, generating a bimodal distribution of small ∼7 μm calcite crystals and larger ∼20 μm polycrystalline vaterite particles at a copolymer concentration of 0.010% w/w, as confirmed by Raman microscopy (see Figure S9). At an additive concentration of 0.050% w/w, copolymer overgrown calcite rhombohedra with somewhat curved edges and roughened faces are produced. This indicates a strong interaction between the anionic nano-objects and the crystal lattice.
The crystals were also fractured to give a rough estimate of the relative degrees of occlusion of these contrasting nanoobjects. Figure 9 shows both intact and fractured crystals prepared at the two different additive concentrations. The nonionic PMP 24 −PBzMA 300 vesicles are not occluded at all, whereas a moderate degree of occlusion can be observed for both PMPA 32 −PBzMA 300 and PMPA 24 −PBzMA 300 anionic vesicles. This is particularly interesting for the latter vesicles, which have mean diameters of ∼520 nm and are thus significantly larger than the 200 nm latex particles occluded in earlier work. 95 These vesicles are of potential interest as carriers of functional additives. However, the degree of vesicle occlusion appears to be significantly higher at the periphery of the crystals for both additive concentrations. This may derive from an increase in incorporation efficiency at the lower supersaturation encountered toward the end of the crystallization experiments 79 or could suggest a mass-transport-limited crystallization mechanism in which vesicle diffusion to the surface of the growing calcite crystals limits occlusion. This would again promote occlusion during the later, slower stages of growth. In contrast, the ∼200 nm anionic PMPA 32 − PBzMA 300 vesicles are incorporated to much greater extents at both polymer concentrations. Furthermore, homogeneous occlusion throughout the crystal is observed. This suggests that the incorporation of larger particles may be more dependent on the reaction conditions, and in particular supersaturation, than that of smaller particles. Finally, examination of the distribution of the PMPA 42 −PBzMA 45 worms within the crystals precipitated at the lower polymer concentration showed that they were predominantly located in the outer 1 μm region of the crystal. In contrast, for the small calcite crystals precipitated at the higher additive concentration the worms appear to be occluded rather more homogeneously throughout the crystal.

■ CONCLUSIONS
A new PISA formulation has been explored for the generation of both phosphonate-and phosphonic acid-stabilized spheres, worms, and vesicles. More specifically, three poly(methacryloyloxymethyl dimethylphosphonate) [PMP] macro-CTAs with mean degrees of polymerization of 24, 32, and 42 were synthesized via RAFT solution polymerization and converted into the corresponding poly(methacryloxymethyl phosphonic acid) [PMPA] macro-CTAs via ester hydrolysis. Subsequent chain extension with BzMA via RAFT methanolic dispersion polymerization at 20% w/w solids enabled construction of detailed phase diagrams for both types of macro-CTAs. Systematic variation of the target degree of polymerization of the PBzMA block provides access to diblock copolymer spheres, worms, and vesicles, as well as mixed phases. Comparing the performance of the same macro-CTA in its non-ionic and anionic forms allowed the effect of stabilizer charge density on self-assembly in nonaqueous media (i.e., methanol) to be assessed for the first time. GPC studies indicate high blocking efficiencies and low final polydispersities for all non-ionic PMP x −PBzMA y diblock copolymers. However, for the anionic PMPA x macro-CTAs, blocking efficiencies could only be assessed indirectly via chain extension experiments conducted using a water-soluble monomer (glycerol monomethacrylate) as the second block. Transfer of selected diblock copolymer nano-objects from methanol to water via dialysis did not result in any discernible changes in copolymer morphology, as judged by TEM and DLS. Aqueous electrophoresis studies confirmed the highly anionic character of the phosphonic acid-stabilized nano-objects in aqueous media. Examination of both sets of copolymer nano-objects as crystal growth additives showed that anionic worms and vesicles were

Macromolecules
Article occluded within calcite crystals during the crsytal growth, while the corresponding non-ionic phosphonate-stabilized vesicles were not. These findings emphasize the importance of appropriate surface chemistry in promoting occlusion of organic nanoparticles within inorganic host crystals.

* S Supporting Information
The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.macromol.5b02212. 31 P NMR spectra and GPC data for the three PMPA macro-CTAs, GPC and DLS data for kinetics of dispersion polymerization experiments, GPC data for all PMP x −PBzMA y diblock copolymers, additional TEM images, DLS data for aqueous dispersion polymerizations conducted using binary mixtures of macro-CTAs, and Raman spectra for selected CaCO 3 occlusion experiments (PDF)