Clean synthesis of linear and star amphiphilic poly( ϵ -caprolactone)-block-poly (ethyl ethylene phosphonate) block copolymers: assessing self-assembly and surface activity

Anionic ring-opening polymerization (AROP) of 2-ethyl-2-oxo-1,3,2-dioxaphospholane (EP) has been utilized to create alternative hydrophilic moieties for utilization in surfactants. The current “go-to” hydrophilic unit is poly(ethyleneglycol) (PEG), but this is non-biodegradable and there is some evidence that their use can lead to accumulation of toxic by-products. We have created new approaches leading to water-dispersible, fully degradable amphiphilic block copolymers. Our approach utilized poly( − caprolactone ) -based macroinitiators in a solvent free process with organocatalyst DBU. We have prepared a comprehensive set of novel amphiphilic PCL- b -PEP block copolymers including linear diblock, triblock and star diblock copolymers characterized by SEC, 1 H and 31 P NMR and MALDI-TOF analyses. Supercritical carbon dioxide (scCO 2 ) was exploited to efficiently extract the residual EP monomer leading to significantly increased gravimetric yields of purified product compared to conventional techniques such as dialysis. The self-assembly of the amphiphilic copolymers in water was investigated by DLS and cryo-TEM and their surface-activity as a function of concentration was investigated by tensiometry showing behavior that matches or exceeds commercially available surfactants.


Introduction
Stringent environmental regulations and political frameworks will require a shift towards utilisation of sustainable and biodegradable surfactants in the personal care industry and for this reason there is strong interest in developing new surfactants. 1 We present a unique approach to new surfactant architectures based upon linear and star polymers, developed from natural resources; ensuring they are both amphiphilic and degradable and utilizing eco-friendly processing systems.
Amphiphilic copolymers are ubiquitous in our lives; as laundry detergents, food additives, cosmetics or even medicines.2 In an aqueous medium or selective solvent, they spontaneously self-assemble into discrete aggregates consisting of a solvophobic core and solvophilic shell (corona).3,4 Such aggregates can develop into advanced morphologies such as micellar or vesicular4 conformations that are useful in a wide range of applications; for example as nanovehicles in the biomedical-pharmaceutical field.5,6 Conventional linear amphiphiles generally form aggregates such as micelles above their critical aggregation concentration (CAC) where the morphology obtained depends on the solvophilicity of the primary structure (i.e. the composition of blocks, chain length, and sequence)7 and on the concentration, temperature, solvent, and pH.8 However for pharmaceutical applications, the high dilution that occurs upon injection into the bloodstream can cause disassembly of such nanocarriers and premature drug release8-10 and hence there is a need for more robust systems.
One promising route is to exploit branched polymer systems. In sharp contrast to the linear block copolymeric micelles, branched structures are often characterized by the formation of ''unimolecular micelles''6,11-16 which cannot disassemble, thereby forming a monodisperse and structurally stable dispersed phase that is by its very nature not affected by dilution. 15,[17][18][19][20] Examples of branched amphiphilic macromolecules include dendrimers,10,18 hyperbranched,10,21,22 and star polymers8,10,20,21,23 which can possess different moieties of contrasting polarity, bound together in a covalently tethered three-dimensional core-shell system. The improved stability upon dilution of these unimolecular micelles arises from the fact that they are multiple linear polymers, covalently connected, so that disassembly is no longer possible.24 Star polymers are branched polymers with three or more arms of often similar lengths, radiating from a central core where the core may be an atom, a small molecule, or a macromolecular structure. 25 They have attracted significant attention because their properties are similar to dendrimers but they are much easier to access.26-28 One prominent example is Tetronic® (commercialized by BASF) a four-arm star amphiphilic block copolymer [poly(ethylene oxide)-block-poly(propylene oxide) (star PEO-b-PPO)] used in the cosmetics, paper and petroleum industries.29 Among all the star-like polymers, polyesters constitute an attractive class for drug delivery systems because of their potential biodegradability and biocompatiblity. 30 The hydrophobic segment is most commonly poly(lactic acid) (PLA), poly(ε-caprolactone) (PCL) or poly(glycolic acid).6 However, a hydrophilic segment is also required31 and poly(ethylene glycol) (PEG) is today's gold standard in the biomedical field32 because it has high water solubility, low toxicity and low immunogenicity. However, PEG is non-biodegradable and there is some evidence of accumulation of toxic by-products from uncontrolled and undesirable oxidative degradation that have raised concerns. 33 The hydrolytic and enzymatic degradation of PPEs has been reported.52 Especially polyphosphonates have proven to undergo hydrolytic degradation at physiological pH or under basic conditions.53 Hydrolysis of phosphonate esters proceeds via a nucleophilic attack of water or hydroxyl ions to the phosphorous center54 or by a backbiting mechanism.55 The hydrolytic stability of polyphosphonates was found dependent of the nature of the alkyl side chain. Degradation times can be adjusted from hours to weeks at physiological pH for polymers with different side chains (e.g. methyl < ethyl < isopropyl).50 Could polyphosphonates provide a degradable alternative to PEG as the hydrophilic segment in unimolecular micelles? To the best of our knowledge, amphiphilic block copolymers containing poly(phosphonate) functional groups and hydrophobic polyesters (PCL or PLA) have not yet been reported, though previous studies have prepared gradient copolymers of PLA and PEP by direct copolymerization. 56 We present a new approach to controlled synthesis of poly(ethyl ethylene phosphonate) (PEP) leading to a range of amphiphilic linear, diblock, triblock and star diblock copolymers containing hydrophilic polyphosphonate groups and hydrophobic polyesters (PCL-b-PEP) that are fully degradable (Figure 1). Our synthetic approach also exploits supercritical carbon dioxide (scCO2) to ensure green synthesis and the removal of potentially toxic starting materials and by-products. The surface-activities of all the block copolymers are evaluated using the Wilhelmy plate method and dynamic light scattering (DLS) and transmission electron microscopy (TEM) are exploited to determine self-assembly and the ability to form micellar structures.

Synthesis of Linear or Star Polyol-PCL using Sn(Oct)2 in Supercritical CO2
The stainless steel high-pressure autoclave reactor (20 mL) used for polymer synthesis with scCO2 has been described previously.

Synthesis of PCL-b-PEP Diblock and Triblock Copolymer in the Bulk
A specific example of synthesis of poly(ε-caprolactone)-block-poly(ethyl phosphonate) (PCLnb-PEPm) diblock copolymers (linear and star) by bulk AROP using DBU is described ( In general, polymers were purified either by dialysis to give a gravimetric yield of 62 ± 5% or by supercritical extraction to give a gravimetric yield of 92 ± 5 %. The linear triblock poly(ethyl phosphonate)-blockpoly(caprolactone)-block-poly(ethyl phosphonate) (PEPm-b-PCLn-b-PEPm) copolymers and star diblock copolymers were synthesized following the same procedure and the specified molar ratio of feedstock (Table 1, Entry 6-8).

Results and Discussion
Our strategy was to prepare linear diblock and triblock copolymers and then to extend the technique to star block copolymers. Finally, we investigate the behavior of all the materials in aqueous solution.

Linear Diblock and Triblock copolymers 3.1.1. Synthesis
We first investigated whether the hydroxy end groups of the linear PCL macroinitiators were able to initiate the bulk AROP of EP using DBU catalyst. Three PCL macroinitiators of different MW were utilized; specifically, a monohydroxy (Bz-PCL19OH) and two dihydroxy telechelics (HO-PCL4-OH and HO-PCL9-OH). These were synthesized in scCO2 using benzyl alcohol and 1,6-hexanediol as initiator, respectively, with Sn(Oct)2 as catalyst (Table 1 and  Table S1).58 These first blocks were then extended through bulk polymerization of EP at a lowtemperature (30-60 °C). For example, from the monofunctional Bz-PCL19OH (MnNMR 2300 g/mol, Đ of 1.02) in the presence of DBU, we demonstrated high EP monomer conversions (~88-98% by 1H NMR) at a temperature of 60 °C. This seemed to be the most suitable temperature to balance the increased viscosity and to drive the reaction to the desired conversion. Linear Bz-PCL19-b-PEP30 diblock copolymer was obtained with 92% conversion in only 2 h under these experimental conditions (Table 1, Entry 5).
In the following discussion, the linear diblock copolymer and triblock copolymers are denoted as Bz-PCLn-b-PEPm and PEPm/2-b-PCLn-b-PEPm/2 respectively, where n and m correspond to the number average of CL and EP units ( Figure 2).   The SEC (RI, Double Angle LS) analysis confirmed the success of bulk copolymerization with a clear increase in the elution time and MW of the Bz-PCL19-b-PEP30 diblock compared to the Bz-PCL19OH macroinitiator (Figure 3 (a)). Additionally, the copolymer obtained showed narrow dispersity (Đ= 1.04) ( The composition and the molecular weight of the copolymers were assessed with 1H NMR and 31P NMR; with all the characteristic resonances of macroinitiator Bz-PCL19OH ( Figure S1) and diblock Bz-PCL19-b-PEP30 copolymer (Figure 3b) assigned. These data also confirm the successful growth of the PEP block from the macroinitiator. For information, 1H and 31P NMR spectra of EP monomer are presented ( Figure S2).    Thus, 1H NMR and SEC (RI, DALS) analyses confirm that there is indeed a covalent link between the PCL and PEP blocks and that the 'green' solvent-free bulk polymerization does produce copolymers with good control (Figure 3).
Having synthesized the block copolymers, the next task was to assess their surface-active properties and we began by looking at the solubility of the synthesized polymers in water. The Bz-PCL19-b-PEP30 copolymer gave a faint-cloudy suspension at 0.5 wt%. By contrast, homopolymer Bz-PCL19OH was found to be totally insoluble in water which confirms the amphiphilic nature of this Bz-PCL19-b-PEP30 diblock copolymer. To modulate the solubility of any copolymer in water, the ratio of PEP: PCL should be varied by tuning the block lengths.59 Indeed, when we increased the length of the PEP block, the linear block copolymer tended to solubilize more easily in water. Triblock copolymers, with a shorter central PCL hydrophobe (MnNMR (PCL) = 600 g/mol for HO-PCL4-OH or MnNMR (PCL) =1100 g/mol for HO-PCL9-OH) were explored. More precisely, two PEPm/2-b-PCLn-b-PEPm/2 triblock amphiphiles with increasing proportion of PEPm/2 blocks (for DPn targ (PCLn) = 4 then DPn targ (PEPm) ~5.3 and ~9.6) were synthesized (Table S1, Figure 4). Additionally, a third triblock copolymer was synthesized which nearly doubled both the hydrophilic PEPm and hydrophobic PCLn block (i.e., to increase the MW but maintained the same hydrophilic-lipophilic balance (HLB)). All the syntheses were conducted following protocols employed for diblock copolymer and analytical results and SEC plots are presented (Table 1, Table S1, and Figure S4).

Purification of Diblock and Triblock Copolymers using SCFE
It is well-known that purification of amphiphilic copolymers or surfactants is not straightforward and often multiple steps using volatile organic solvents are involved.44 For instance, Wolf et al. reported that the precipitation of the homo polyphosphonate by cold diethyl ether gave a low yield (65 ± 2%) with traces of DBU detected by 1H NMR.50 As an alternative, purification by dialysis (molecular weight cut-off 1000 g/mol) against Milli-Q water was performed. We found that most of the DBU was separated from the sample but the yield (60 ± 5%) was not significantly improved, probably because of the low MW of the surfactants (~600-3500 g/mol). Also, dialysis is time-consuming and always requires an Triblock PEP 1.6 -PCL 4 -PEP 1. 6 (220:600:220 g/mol) Triblock PEP 3.6 -PCL 4 -PEP 3.6 (500:600:500 g/mol) energy-intensive freeze-drying step. Thus, we decided to explore the use of scCO2 for purification, most notably for removal of solvent or monomer residues60 noting that supercritical fluid extraction (SCFE) is commercially viable on the large scale61,62 largely because CO2 is an abundant, renewable and inexpensive resource.
SCFE was employed to purify copolymers formed in bulk reactions, primarily targeting removal of the scCO2 soluble DBU organocatalyst and residual monomers. The scCO2 insoluble polymers remain in the autoclave and can be easily collected. The scCO2 extracted materials were collected via venting and analyzed by 1H NMR (SI SCFE purification) and showed that ~11.5 wt% of EP monomer was extracted from the PEP3.6-b-PCL4-b-PEP3.6 triblock amphiphile in just 25 min ( Figure S3). The clear viscous liquid polymer was retrieved from the autoclave with high gravimetric yield (~92 ± 2%) and showed only traces of EP (0.5 wt%).
These data highlight that the use of scCO2 dramatically improves the yield post purification (an increase of 42% was observed). In addition, comparing to other process, we can assume that this method is non toxic (no use of VOC), time saving (only 25 min to purify more than 2 g of polymer versus more than 6 h for less that 1 g by the dialysis method), energy saving (no drying or lyophilization step).
Additionally, the 1H NMR spectrum of the material extracted by scCO2 revealed that the cyclic EP ring remained intact after the purification process ( Figure S3 (B)). In the dialysispurification method this is not the case. Thus, scCO2 might allow for recovery and recycling of the monomer, which would be desirable on the commercial scale and for the circular economy.

MALDI-TOF of PEP-b-PCL-b-PEP Triblock Copolymer
The MALDI-TOF data obtained from a representative PEP7.2-b-PCL9-b-PEP7.2 triblock copolymer displays a monomodal Gaussian distribution ( Figure 5) implying the presence of only a single series of peaks assigned to the triblock copolymer. No residual HO-PCL-OH macroinitiator was detected, thus confirming efficient synthesis via chain extension in the bulk.  A more detailed analysis of the MALDI-TOF data in the region corresponding to the strongest peaks (1675 -1900 m/z; Figure 6 and Table 2 (Table 2). Hence, the MALDI-TOF spectrum confirms the formation of the copolymer PEPm-b-PCLn-b-PEPm synthesized in the bulk using DBU as a catalyst. To the best of our knowledge, this is the first example of structural elucidation of phosphonate copolymers by MALDI-TOF-MS analysis.

Synthesis of Star PCL-b-PEP Diblock Amphiphilic Copolymers
After demonstrating the clean synthesis of linear diblock and triblock copolymers and their purification by SCFE we then moved on to attempt the synthesis of more complex star diblock copolymers using a renewable material, D-sorbitol as the core.
In this section we developed star homopolymer and diblock copolymers; referred as star D- . Protons of D-sorbitol core were superseded by protons of the arms (Figure 7b).  These results highlight that high MW amphiphilic star diblock copolymers (MnSEC-DMF ~ 12500 g/mol) with narrow dispersity (Đ = 1.10) can be synthesized in the bulk in the absence of VOCs. The polymer end-groups were precisely identified by phosphitylation using a phospholane reagent58. This confirmed that all the end-groups of star D-sorbitol-PCL were utilized to radiate PEP arms, indicating formation of 6 arms with at least 5 amphiphilic arms (Figure 9, Figure  S5)  Linear diblock and triblock copolymers as well as star diblock copolymers were successfully synthesized by AROP of EP from linear and star PCL-based macroinitiators. Such amphiphilic copolymers could be surface active and able to self-assemble in aqueous solution, two key parameters for surfactant characterization.

Surface Activity of Linear and Star PCL-b-PEP Block Copolymers
We investigated the surface activity of the building blocks and the synthesized compounds. A compound that reduces the surface tension (γ) of water from 72 mN/m to below ≤ 60 mN/m can be classed as a surfactant.60 The starting materials, D-sorbitol (core) and EP monomer do not reduce the surface tension and nor does the linear Bz-PCL19-b-PEP30. The first star compound D-sorbitol[PCL57-b-PEP36OH6] forms a cloudy solution under ambient condition at 5 wt% (Table 1, Entry 4 and Entry 7) but has no effect on surface tension. We speculate that this is a result of the longer length of hydrophobic ε-CL compared to only a short hydrophilic PEP unit.
In sharp contrast, the star counterpart D-sorbitol[PCL18-b-PEP27OH6] showed a significant effect upon surface tension lowering to 45 mN/m ( Figure S6, Table 1, Entry 5 vs. 8). These results are promising when compared with commercial oil-based and non-degradable surfactants (Tween 20 and 80) where the values ranged between 32-40 mN/m ( Figure S6).
The triblock copolymers were all soluble in water and showed surface tension in the range 38-40 mN/m ( Figure 10, Table S2). Interestingly, the linear PEP7.   To simplify the self-assembly and surface activity discussion, these copolymers are grouped based on MW with low, intermediate and high MW copolymer (Table S3).

CAC of triblock and star amphiphilic copolymers
The Critical aggregation concentration (CAC) of the triblock PEP7.2-b-PCL9-b-PEP7.2 (Triblock 3) was measured using the Wilhelmy plate method. Theoretically, the sequential addition of triblock surfactant into water will decrease surface tension as more surfactant accumulates at the air/water interface. Only when the interface becomes saturated does further addition of surfactant lead to aggregation/micellization. All the triblock copolymers behaved in a similar manner. A sharp change in the slope was identified from the point where the two regression lines intersect to give a concentration of 47.9 μM (0.014 wt%) for the CAC ( Figure  S7). Above the CAC, the surface tension remained ~34 mN/m. For the star copolymer, a similar analysis showed a concentration of 21.2 μM (0.012 wt%) for the CAC ( Figure S8).
We chose two points above the CAC (0.1 and 0.5 wt%) to study the self-assembly of aqueous solutions of all the amphiphiles.

Self-assembly of linear triblock copolymers
For the triblock copolymers, the auto-correlation functions of the low MW triblock copolymers showed two or three distinct populations of nanoparticles in the intensity-weighted distributions. (Figure 11, upper; Table 1 & Table S3). These samples were studied without filters to give a full picture of the range of populations present. As always with DLS it is important to note that scattering is much more sensitive to larger objects63 as shown in the intensity-weighted distributions (Figure 11, upper) whereas the number distributions ( Figure  11, lower) give a more realistic picture showing that these particles are insignificant in terms of the number-weighed distribution.

Intermediate MW Linear vs. Star Diblock Copolymers
The influence of the block copolymer architecture on self-assembly was further investigated by comparing the aqueous solutions of star  (Table S3).
The linear diblock Bz-PCL19-b-PEP30 gave a turbid solution at the higher concentration (0.5 wt%) and so only 0.1 wt% was investigated and this showed a broad monomodal distribution (DH-value = 145 nm in both intensity and number weighted plots (Figure 12)). By contrast, star D-sorbitol[PCL18-b-PEP27OH6] was fully water soluble at 0.5 wt%, reflecting the influence of topology on solubility and self-assembly. The star copolymer showed a bimodal size distribution at 0.5 wt% with a primary population of smaller nanoparticles (DH, DLS = 7 nm) coexisting with larger aggregates (DH, DLS = 111 nm) (Figure 12, top). Importantly, the size distribution of the objects remains unchanged at the lower conc. (0.1 wt%) (DH, DLS =6.3 and 107 nm). Water is a good solvent for the hydrophilic PEP blocks which will extend, and a poor solvent for the hydrophobic PCL block which will shrink. Thus, we would expect the theoretical size for D-sorbitol[PCL18-b-PEP27OH6] to be about ~5 nm, assuming a fully extended chain conformation with six PEP4.5 arms forming a corona around the condensed PCL18 core. Since the DLS experiments were performed at 0.1 and 0.5 wt%, values above the CAC (0.012 wt%) ( Figure S8), it is reasonable to conclude that the smallest observed diameter (DH ~7 nm) corresponds to star unimers (Figure 12). Bigger DH-values observed by IWD may reasonably be attributed to a small population of aggregates of a few star unimers (Figure 12, upper) since the number-weighted distributions showed only monomodal distributions around 4.85 nm (Figure 12, lower).

High MW Star PCL-b-PEP Diblock Copolymers: Direct Dissolution vs. Nanoprecipitation Method
The large, high MW star copolymer star D-sorbitol[PCL57-b-PEP36OH6] (Table 1 entry 7) showed reasonable hydrophilicity of 8.4 (Table S3, Entry 6) and so its self-assembly was attempted by direct dissolution in water. However, an aqueous solution of it displayed cloudiness even at the lower conc. 0.1 wt%. DLS results revealed co-existence of two populations ( Figure S9, last vertical bars) with DH-values of 19 and 142 nm in the intensity weighed-distribution. These larger aggregates and the observed turbidity led us to examine nanoprecipitation to perhaps yield smaller objects.
Briefly, in nanoprecipitation method the star samples (high MW star D-sorbitol[PCL57-b-PEP36OH6], Table S3) were dissolved in acetone (a good solvent for both the PCL and PEP blocks), and self-assembly was induced by gradually adding the acetone solutions into Milli-Q water followed by vacuum removal of acetone. 64 The resultant aqueous solutions were transparent and DLS showed larger objects by intensity distribution DH, DLS (IWD) 240 nm in addition to smaller population (DH, DLS (IWD) 18 nm) but number distribution showed only single population with DH of 13 nm ( Figure S10). As a control, the self-assembly of hydrophobic star D-sorbitol [PCLnOHx]OH samples (at 0.1 wt%) were studied in parallel following the nanoprecipitation method. These were found to be cloudy suggesting the hydroxyl groups alone on the PCL chains are incapable of stabilizing the hydrophobic PCL and hence it precipitates.

TEM and Cryo-TEM images of Star PCL-b-PEP Diblock Copolymers
The morphology of the star D-sorbitol[PCL57-b-PEP36OH6] copolymer prepared by nanoprecipitation into the water was investigated by transmission electron microscopy (TEM). The TEM images show uniform size spheres ( Figure S11 and Figure S12). The average particle diameter measured by TEM (DTEM) was found to be smaller than the hydrodynamic diameter found by DLS (DTEM = 14.3 nm and DH, DLS (IWD) = 18 nm) (Figure S12, histogram and Figure  S10). This is exactly as one would expect since the TEM values are always slightly smaller because the aggregates are in their dry state whereas DLS measures the hydrated and swollen particles. Though in this case, the difference is actually relatively small suggesting limited shrinkage. The absence of the much larger aggregates on TEM images is not surprising as it reinforces the fact that these are overemphasized in the intensity weighted DLS measurements.
( Figure S11 and Figure S12).   Spherical morphologies have been frequently reported for unimers of star-like and dendrimer structures. 8,12,15,17 Looking closely at the cryo-TEM, the grey core clearly shows the condensed and water-free PCL core (Inset of Figure 13). Whilst the white outer rings show the PEP in the corona diluted by water (arrowed red). The unusual dark black rings at the interface (arrowed white) might relate to a high concentration of phosphorus that has collapsed into a PEP segment at the interface between the PCL core and the PEP corona. This could happen where the water concentration is low.65

Conclusion
We present the first synthesis of linear diblock, triblock and star diblock copolymers utilising renewable and degradable PCL cores with degradable, hydrophilic poly(phosphonates). We utilized only renewable building-blocks of 1,6-hexanediol, D-sorbitol and ε-CL monomer and we demonstrate effective use of solvent-free bulk and clean scCO2 based polymerisation systems to yield clean residue-free products. The triblocks show excellent activity as surfactants reducing the surface tension of water to a level that competes with the commercial Pluronic TM surfactants. The star diblock copolymers were shown to self-assemble in aqueous solvents yielding structures of ideal size for application in drug delivery or indeed a wide range of other possibilities. The demonstration of the utility of poly(phosphonates) as the hydrophile in these surfactants also opens up the possibility of replacing poly(ethylene glycol), the industry gold standard that is non-degradable and has recently been shown to bioaccumulate.