Uniform shape monodisperse single chain nanocrystals by living aqueous catalytic polymerization

The preparation of polymer nanoparticles with a uniform size and shape, beyond spheres, is an unresolved problem. Here we report a living aqueous catalytic polymerization, resulting in particles grown by a single active site and composed of a single ultra high molecular weight polyethylene (UHMWPE) chain. The control on a molecular level (Mw/Mn = 1.1–1.2) and at the same time on a particle level (PDI < 0.05) together with the immediate deposition of the growing chain on the growing nanocrystal results in a distinct evolution of the particle morphology over time. These uniform nanocrystals are obtained as concentrated aqueous dispersions of > 10 wt-% (N ≈ 1019 particles L−1) polymer content. Key to this robust procedure to single chain nanoparticles are long-lived water-stable Ni(II) catalysts that do not undergo any chain transfer. These findings are a relevant step towards polymer materials based on nanoparticle assembly.


S8
General procedure: The synthesis was conducted according to a modified literature procedure. 5 [Ir(COE)2Cl]2 (0.005 -0.01 equiv.) and 4,4'-di-tert-butyl-2,2'-dipyridyl (0.01 -0.02 equiv.) were stirred in 5 mL of heptane to give a dark blue catalyst suspension. This was transferred to a mixture of 1,3diperfluoroalkylbenzene (1a, 1 equiv.) and bis(bipinacolato)biboron (1 equiv.) in 150 mL of heptane and stirred for 12 hours at 90 °C to give an orange solution. The reaction mixture was allowed to cool to room temperature, extracted with brine (2 x 150 mL) and dried over sodium sulfate. After removing the solvent in vacuum, the product was obtained as yellow oil. This was kept under vacuum for several hours to remove residues of pinacolborane.

General procedure:
Terphenylamine (1c, 1 equiv.), 3,5-diiodosalicylaldehyde (2.0 -2.3 equiv.) and 50 mg ptoluene sulfonic acid hydrate (pTsOH) were added to 200 -250 mL of toluene. The flask was equipped with a Soxhlet apparatus filled with dried molecular sieves to allow an azeotropic water removal during the reaction. The reaction mixture was heated to intense reflux (heating bath temperature >160 °C) for 14 hours. Note, that a complete dissolution of the terphenylamine is essential for conversion to the desired product. The reaction mixture was allowed to cool down to room temperature, and the solvent removed was under reduced pressure. After purification via column chromatography on silica with pentane as eluent (yellow band), the product was obtained as a yellow sticky oil. Precipitation from a saturated pentane solution at -78 °C gave the pure product as a yellow powder. [2,6-bis (3,5-

3,5-Diiodo-N-
Compound 1d-C4F9 was obtained following the general procedure with 7.46 g 3,3',5,5'tetra(perfluorobutyl)terphenyl amine (1c-C4F9, 6.7 mmol, 1 equiv.), 5.0 g 3,5diiodosalicylaldehyde (13.4 mmol, 2 equiv.) and 50 mg pTsOH in 250 mL of toluene   General procedure: To [(tmeda)NiMe2] (1.1 equiv.) and the respective salicylaldimine (1d, 1 equiv.), a solution of pyridine (15 equiv.) in 4.5 -7 mL of benzene/hexafluorobenzene was added. Gas evolution (methane) was observed and the reaction mixture turned orange to red. The reactants were stirred for 2 hours at room temperature. Now, the formed Nickel black was removed via centrifugation. The red solution was frozen in liquid nitrogen und the solvent removed by freeze drying to give the desired product as a red powder.  General procedure: To solid [(tmeda)NiMe2] (1.2 equiv.) a solution of the respective salicylaldimine (1d, 1.1 equiv.) in a mixture of benzene/hexafluorobenzene or neat benzene was added, and stirred for 1 hour at room temperature. During the addition, gas evolution (methane) was observed and an orange to red solution (slightly turbid) was obtained. α-Methoxy-ω-amino poly(ethylene glycol) (1.0 equiv.; Mw = 1981 g mol -1 , unless noted otherwise) in 1 ml of benzene was added and the reaction was stirred for further 3 hours at room temperature (the solution cleared up). After filtration through a syringe filter to remove nickel black, the solvent was removed under vacuum. The orange residue was washed with portions of pentane (3-5 times, 7.5 mL each) until the filtrate remained almost colorless (slightly orange). After drying under vacuum, the desired product was obtained as an orange powder.

Process design for polymerization in aqueous media
Influence of the fluorocarbon chain length in the catalyst structure The length of the perfluorinated alkyl chain in the catalyst structure was found to have a strong impact on the catalytic properties of the active nickel center, as studied in detail in toluene as a reaction medium with lipophilic catalyst precursors (pyridine as labile ligand; cf. Figure 2).
These trends were also confirmed for hydrophilic analogues in aqueous system (Supplementary Table 1). Also in aqueous systems, yields and molecular weights generally increase with increasing length of the fluorocarbon group in the catalyst structure (cf. Supplementary Figure 9). An influence of the hydrophobic perfluoroalkyl on the dissociation rate of the labile amino-PEG ligand in aqueous system was also found. This is evident from low chains-per-nickel ratios for S28 catalyst 2/PEG (entry 1, Supplementary Table 1) as the labile ligand seems to bind strongly to the metal center and hinders a sufficient activation. This was not observed for catalysts 1/PEG with their hydrophobic perfluoroalkyl groups that force the highly hydrophilic labile ligand in close proximity to dissociate as revealed by chains-per-nickel rations close to unity (entry 2-4, Supplementary Table 1). The catalyst was further modified by changing the length of the PEGunit of the labile coordinated ligand molecule (1981 g mol -1 vs. 5516 g mol -1 ). Catalyst bearing longer PEG-units generally showed a slightly higher activity, presumably due to an increased water-solubility resulting in an enhanced dissociation rate (entries 1-3 vs. entries 4-6, Supplementary Table 4). Other than this, an essentially similar behavior was observed.

Choice of lipophilic solvents in aqueous systems
To ensure a maximum degree of dispersion of the catalyst, followed by an undisturbed particle  The addition of small amounts organic lipophilic compounds (0.1 vol-%) significantly increased the polymer yield and the quality of the dispersion obtained after polymerization in terms of stability and particle size distribution, quantified by DLS measurements, in all cases (cf. Supplementary Figure 11). vs. entry 24). The overall data gives no indication that the lipophilic compound directly interacts with the active center and alters its catalytic properties with regard to chain growth and chain microstructure. This is evident from the fact that in all cases polyethylene with comparable thermal properties was formed in a living polymerization as suggested from chains per nickel ratios close to unity and narrow molecular weight distributions (molecular weights differ in accordance with yield). Rather the polymerization results appear to depend on the lipophilic solvents ability to dissolve/distribute the precatalyst/active center in solution, respectively, and on its ability to form stable and highly-dispersed emulsions in the water/SDS system. With this in mind, the organic compound does not influence the polymer formation mechanism itself, but rather the way the catalysts' polymer chain forms ordered single nanocrystals during chain growth and in particular the initial stages of the polymerization experiment. As alkanes and alkylbenzenes are known to form highly stable emulsions with SDS 6 , their superior role as lipophilic solvents compared to e.g. fluorinated solvents is plausible. The importance of a sufficient emulsion formation is further underlined by direct correlation of particle size distribution and yield. Reduced yields are usually found for dispersions with broad particle size distributions and vice versa. As particle/catalyst agglomeration disturbs chain growth and consequently influences yield, respectively, molecular weight, the found relation is expected.

S31
Supplementary We found mesitylene to be the solvent of choice in that it promotes high yields and molecular weights as well as narrow particle size distributions (entries 26 and 27, Supplementary Table   2). For the case of mesitylene even in the absence of ultrasonication similar favorable polymerization results were found, which underlines the suitability of this system to distribute the catalyst in the initial reaction mixture well. Studies with variable amounts of mesitylene showed the minimal concentration required to be 0.2 vol-% (entry 2, Supplementary Table 3).
Experiments performed with lower portions showed reduced yields and broadening in particle size distributions (entries 1-2, Supplementary Table 3). Above this concentration no significant impact of mesitylene loadings were found.

Surfactant
The surfactant is of major importance for a favorable particle formation process. Experiments conducted without surfactant lead to immediate polymer precipitation and polymerization activity ceasing within minutes. As the surfactant is absorbed during the reaction on the growing crystals surface, it is one major limiting factor of that reaction type and limits the polymerization progress. We further hypothesize that the employed hydrophobic catalyst systems with their perfluorinated alkyl chains require a certain amount of free surfactant in solution to be stabilized in the initial reaction mixture in order to perform undisturbed chain and particle growth. Usually polyethylene of 1.5-to 2.0-times the sodium dodecyl sulfate content (by mass) was found to be formed before the activity suddenly decreased. This is evident from the ethylene mass flow curves recorded during the polymerization experiment that allows for a precise monitoring of the catalytic activity (cf. Supplementary Figure 12). Results are summarized in Supplementary Table 4. The yields were found to increase for experiments, where more free surfactant per nickel center is available. On the other hand, also molecular weights and particle size were found to be higher. This shows that the surfactant content directly influences the amount of polymer that is formed by an active center. A high content enables the synthesis of particles with > 80 nm lateral size and molecular weights of Mn = 3 x 10 6 g mol -1 , while yield, molecular weight and particle size are reduced when half the amount of surfactant is used (entries 4 and 7, Table 2). In contrast, the molecular mechanism of chain growth itself is not influenced as in all cases polyethylene with comparable thermal properties is formed in a living polymerization.

S33
Supplementary  The average mass of one particle mpart was determined by multiplication of Vpart with the density ρPE (Supplementary Equation 3). The density was assumed to be ρPE = 0.94 g cm - 1 11 for these type of nanocrystals (bulk density of UHMWPE Supplementary Figure 13. Schematic segmentation of a lozenge-shaped particle into two triangles with surface areas S1 and S2. = 0.25 π = 4373 ± 1027 nm 2 (9)

Supplementary
Both methods gave similar results for an identical set of particles with only 10 % difference.
This shows that the ellipse-fit routine is not only suited for surface determination of hexagonal/round particles but can also describe lozenge-shaped exemplars with satisfactory accuracy. In addition, the statistical distribution is identically expressed in both cases (Supplementary Figure 15). Considering the propagation of systematic errors, we estimated the measurement error of the molecular weight determined by GPC ∆ W to be ± 10 %, the error of the particle volume determined by tem statistics ∆ part to be ± 20 % and the particle density error ∆ PE to be ± 10 %. Application of a Gaussian error propagation approach to Supplementary Equation 5 yields the error in determination of the chain per particle ratio by Supplementary Equation 10.
Results for Vpart, Mw and the chain-per-particle ratios are listed in Supplementary

Histograms from particle size TEM statistics
Supplementary Figure 18. Histograms of TEM statistical data. The class size was chosen depending on the determined statistical standard deviation exclusively for every data set. A distribution curve (red) is shown assuming Gaussian particle size distribution. Data from entries in Table 2

TEM images of PE nanocrystals
Exemplary TEM sections used for statistical calculations Supplementary Figure 19. TEM image of polyethylene dispersion (entry 3, Table 2); applied ellipse fit shown (right).