Thermoresponsive Triblock-Copolymers of Polyethylene Oxide and Polymethacrylates: Linking Chemistry, Nanoscale Morphology, and Rheological Properties

Thermoreversible gels switch from a free‐flowing liquid state to an elastic gel mesophase upon warming, displaying the reverse transition upon cooling. While this phenomenon makes these advanced materials highly attractive in numerous fields, the generation of optimal materials of tailored rheology and transition temperatures is stifled by the lack of design principles. To address this need, a library of ABA copolymers has been prepared with “A” blocks exhibiting thermoresponsive behavior and “B” blocks of poly(ethylene glycol). This library evaluates the effect of “A” chemistry, probing three polymer classes, and A/B block molecular weight on thermally‐induced phase changes in solutions of the polymers. An exploration by rheometry coupled to Small‐Angle Neutron Scattering (SANS) elucidates temperature‐dependent hierarchical self‐assembly processes occurring on the nanoscale as well as bulk rheology. This process deciphered links between rheology and supracolloidal assemblies (sphere, ellipses, and cylinders) within the gel state with interactions probed further via structure factors. Several design principles are identified to inform the genesis of next‐generation thermoreversible gels, alongside novel materials exhibited thermoresponsive behavior in the solution state for use in applied healthcare technologies.

When the PEG addition finished, the reaction was allowed to rise to room temperature and stirred for 18 h. The solution was filtered, and approximately half of the solvent removed in vacuo. The crude PEG initiator was then precipitated in cold diethyl ether (480 mL) and filtered. The solid was then recrystalised from absolute ethanol (300 mL) overnight. The recrystallised solid was then filtered and washed with cold diethyl ether, before drying in vacuo to yield pure PEG macroinitiator.

Synthesis of ABA triblock copolymers
Triblock copolymers were synthesised by atom transfer radical polymerisation from PEG macroinitiators using prediction of molecular weight based on pilot conversions to synthesise homopolymers (data not shown). A typical procedure is given, with the specific conditions in Table S1. The macroinitiator, ligand and monomer were dissolved in solvent and the flask sealed, followed by bubbling with nitrogen for 30 min to degas the mixture. The Cu(I)Br catalyst was placed in a separate flask and sealed before degassing with nitrogen bubbling for 30 min. The solvent solution was then transferred to the Cu(I)Br vessel via a degassed syringe and was allowed to react under constant stirring for 48 h at a constant temperature.
The reaction solvent was then evaporated in vacuo, the product redissolved in THF, and the mixture passed through Brockmann I neutral alumina to remove the copper-ligand complex.
The THF was removed in vacuo, and the crude product dissolved in DI H 2 O and dialysed for 48 h using a dialysis membrane with molecular weight cut off 3500 Da to remove any residual copper. After dialysis the solution was freeze dried yielding pure polymer (Table   S1).

Dynamic light scattering of triblock copolymers in aqueous solution
Dynamic light scattering (DLS) was performed using a Malvern Zetasizer Nano Series Nano -ZS with Zetasizer software at 1 mg/mL in aqueous solution as a function of temperature.
The sample was heated from 25 to 70 °C in 5 °C increments, and at each temperature three measurements were taken. The micellization temperature was taken as the temperature at which the derived count rate increased.

Rheology of triblock copolymers in 20 % w/v aqueous solution
Rheology was performed on a TA AR 1500 ex rheometer with a Peltier unit using rheology advantage software. 20 % w/v aqueous solutions of triblock copolymers were prepared in DI H 2 O and refrigerated overnight prior to analysis. Samples were analysed using a 40 mm parallel plate geometry with a 650 µm gap. Firstly, oscillatory stress sweeps were performed from 1 to 100 Pa with a frequency of 1 Hz in order to identify the linear viscoelastic region (LVR). Using an oscillatory stress from the LVR (1 Pa), temperature ramps were then performed from 15 to 70 °C at a heating rate of 2 °C per min and a frequency of 1 Hz. The data is presented as the storage modulus (G') and loss modulus (G'') as a function of temperature.

Small-angle neutron scattering (SANS) measurements
SANS measurements were performed on the D22 instrument at the Institut Laue-Langevin (Grenoble, France). [2] The neutron wavelength was set to 6 Å, the sample-detector distance at 2, 5.6, and 17.6 m, with collimator 2.8, 8, and 17.6, respectively. The detector offset was 300 mm. These settings resulted in a wave vector range 2.7 x 10 -3 ≤ q ≤ 0.45 Å -1 . Hellma cuvettes with a thickness of 1 mm were used for all samples. Measurements were performed at 25, 37, 40, and 50 °C with a minimum equilibration time of 15 min prior to sample run. Data reduction and stitching was performed on Igor Pro (Wavemetrics, USA) [3] and data fitting was conducted using SasView 4.2.2 (http://www.sasview.org/). The scattering length densities (SLDs) were calculated from the monomeric unit using the Neutron activation and scattering calculator website from NIST center for neutron research (Neutron activation and scattering calculator). [4] The scattering intensity I(q) can be written as follows: Where, A is a proportionality constant, BKG is the background, P(q) is the form factor of the scattering object,

S(q) A is the corresponding structure factor
If more than one scattering object is present or the object studied has a hierarchical structure that generates scattering at distinct length scales, the expression can be extended to include further terms.
For this work, the polymer constructs, in general, give rise to two scattering signals, one arising from its supramolecular structures and the other from the polymeric chains. Therefore, I(q) is expressed as: where A and B are proportionality constants, BKG is the background, P(q) A is the form factor for model A, S(q) A is the corresponding structure factor, P(q) PGC is the form factor for polydisperse polymer coils. [5] The model A varies depending on the ABA polymer studied as the different chemistries induce self-assembly into different shapes. More specifically, ellipsoids, [6] spheres, [7] coreshell spheres, [ref] cylinders, [8] elliptical cylinders, [9] flexible cylinder, [10], [11] and core-shell cylinders have been used. [12], [13] The models are described in detail elsewhere, but a brief description follows.

Cylinders
For cylinders: where R maj , is the major radius, ε, is the ellipticity of the cross ratio (ε=R min /R maj ), L, is the length of the cylinders, Is the contribution of the elliptical cross-section, Is the contribution from the cylinder's length, where ( ) ∫ J 1 , is a first order Bessel function and ζ represents the cross-section length polydispersity.
The form factor for cylinders is the same as for elliptical cylinders with ε=1 (spherical crosssection).
Flexible-cylinders form factor can be obtained by using the form factor the cylinders replacing P cylinder with:

( )
Where, L Kuhn is the Kuhn length of the flexible cylinder. [14] For core-shell cylinders: α, is the angle between the cylinder and the q vector.
Vs, is the cylinder's volume, Vc, is the volume of the cylinder's core, R, is the core's radius, T, is the shell's thickness, ρ c , ρ s , and ρ solv , are the scattering length densities of the core, shell, and solvent, respectively. J 1 , is a first order Bessel function.

Spheres
The form factor for sphere is as follows: Where, V, is the volume of the sphere, R, the radius of the sphere, Δρ, is the scattering length difference between sphere and solvent.
For a core-shell sphere, F is replaced with: Where, V s , is the sphere volume, V c , is the core's volume, r s , is the sphere radius, r c , is the core radius, ρ c , ρ s , and ρ solv , are the scattering length densities of the core, shell, and solvent, respectively.

Ellipsoids
For oriented ellipsoids: Rp, is the polar radius, Re, is the equatorial radius, α, is the angle between the ellipsoid and the q vector.
For randomly oriented ellipsoids, the particles orientation is averaged for all orientations.

Cytotoxicity testing of triblock copolymers on HaCat cells at 10 mg/mL
HaCat cells were seeded at 10,000 cells per well and grown for 4 days in an incubator at 37 °C with 5 % CO 2 . For each replicate there was a media blank, a positive control and a negative control. 50 µL of 20 mg/mL polymer solution was added to the cells in 50 µL of culture media to yield a tri-block copolymer concentration of 10 mg/mL. The dosed cells were stored in the incubator at 37 °C in 5 % CO 2 for 2 h until running the cytotoxicity assays.
For the LDH assay, 50 µL of cell supernatant was removed from each well and transferred to the wells on a black plate and 50 µL of assay solution was added to each well. The assay was

Data handling and statistical analysis
Data is presented as the mean ± standard deviation of a minimum of three experiments.
Statistical analysis was conducted on Prism (GraphPad, USA), with p < 0.05 considered statistically significant.

Effect of pH on thermoreversible gelation of PDMAEMA copolymers
PDMAEMA homopolymers exhibit pKas of approximately 7.5, depending on polymer molecular weight. [15] Copolymers of PDMAEMA, however, have been shown to exhibit pKas as low as 6.1 depending on both copolymer and molecular weight. [16] Thus, buffered solutions were used to maintain the pH above the pKa and hold the macromolecule in a predominantly unionised state. A single temperature ramp of the PDMAEMA triblock copolymers at 20 % w/v in pH 8.0 phosphate buffered solution was performed ( Figure S7), which demonstrated comparable rheology to the copolymers in water.

Additional comments on the fits of PNIPAM-b-PEG-b-PNIPAM series
The NIPAM ABA series, of the three studied, was the ABA which produced micelles with more distinctive core/shell segregation, as shown by the suitability of the core-shell model fittings. We observe a relatively "wet" (hydrated) core and shell. A SLD around 6 x10 -6 Å -2 corresponds to a shell comprising 95 wt% D 2 O and a SLD around 3 x10 -6 Å -2 corresponds to a core formed by 20 wt% D 2 O. The temperature dependence, when present, was weak but indicated a reduction of D 2 O penetration both in the shell and core, i.e., lower values of SLD, in line with a temperature-induced desolvation.
The radii of the micellar aggregates showed a weak sensitivity to temperature. For instance, the total radius for N10P10N10 (5 wt%) changed from 220 Å to 230 Å, from 37 to 50°C, respectively. At 20wt %, a decrease of 14 Å was observed, from 228 to 214 Å. The total values of the radius observed at 20 wt% are generally smaller than at 5 wt%.
The particle "stickiness", extracted from the S(q) when available, was the more sensitive parameter, with more sticky particles as temperature increases, while the correlated volume fraction showed low sensitivity.
In summary, the largest differences were observed below and above the transition temperature, i.e., from 25 to 37°C. Above 37°C, the micellar aggregates seem to have fully formed and do not undergo major morphology changes.

Additional comments on the fits of PDEGMEMA-b-PEG-b-PDEGMEMA series
The DEG ABA series showed less well-defined micellar aggregates, as the core and shell were not fully resolved in the DEG10-P10-DEG10 construct. For DEG20-P10-DEG20, a wetter core was observed, with SLD values suggesting a core formed of 87 wt% D 2 O at 5 wt% and 67 % at 20 wt% ABA. The data for DEG10-P10-DEG10 could not be fitted by a core-shell model, suggesting that difference in hydration between the core and the shell was too small to be resolved. This could either be due a small difference in polarity between PEG and PDEG or geometric constrains imposed to the ABA conformation due to differences between PEG and PDEG. The predicted log Ps for PEG and PDEG are 0.38 and -6.5 at 25°C, respectively, which shows that PDEG is more hydrophobic than PEG at 25°C.
Therefore, PDEG is also significantly more hydrophobic than PNIPAM, both would suggest that PDEG-PEG-PDEG micelles are even more segregated than PNIPAM-PEG-PNIPAM due to the larger differences in hydrophobicity. The fact that we do not observe segregation at all for DEG10-P10-DEG10 and some segregation with higher levels of hydration for DEG20-P10-DEG20, where larger blocks of DEG are present, suggests a geometric or steric reason for the wetter micelles. Curiously, DEG10-P10-DEG10 data are better fitted with ellipsoids, not spheres. Either the polymeric aggregates show a broader size polydispersity, which often results in either polydisperse spheres or ellipsoids being suitable models, or the DEG imposes such steric constraints that the micelles cannot form spherical objects.
At 25°C, the SANS data were fitted using ellipsoids, instead of polymeric Gaussian coils as was done for PNIPAM-PEG-PNIPAM. This shows that at 25°C supramolecular aggregates are already present, which could be due the higher hydrophobicity of DEG.

Additional comments on the fits of PDMAMA-b-PEG-b-PDMAMA series
The SANS data analysis for the PDMAMA ABA series showed no observable segregation between the core and shell of the micellar aggregates. For PNIPAM-PEG-PNIPAM, coreshell structures were observed for both A20-B10-A20 and A10-B10-A10 architectures. For PDEG-PEG-PDEG, core-shell structures were only observed for the A20-B1-A20 architecture. As discussed previously, this lack of observable segregation could be due to a reduced difference of hydrophobicity between the copolymers. The predicted Log P at 25°C for PDMAMA is 0.67, which lies between PNIPAM and PDEG, and is far larger than PEG.
This would suggest that PDMAMA should be the intermediate case. Therefore, this lack of segregation cannot be due only to differences in hydrophobicity of the copolymers. In a similar way to PDEG, the data from PDMAMA systems required a S(q) at 5 wt% and were fitted as discrete objects at 25°C, instead of polymeric Guassian coils. Unlike PDEG, PDMAMA systems form spheres instead of ellipsoids. While it does not explain the lack of core-shell structures, it seems that a high level of hydrophobicity, high enough to lead to the formation of supramolecular aggregates below the transition temperature is a common feature. A significant contributor to this could be the dependence of LCST on Mn in DMAMA, [17] such that higher Mn chains in the polydisperse sample transition even at the lowest temperatures studied. [17]