Synthesis, Structure, Hydrodynamics and Thermoresponsiveness of Graft Copolymer with Aromatic Polyester Backbone at Poly(2-isopropyl-2-oxazoline) Side Chains

New thermoresponsive graft copolymers with an aromatic polyester backbone and poly(2-isopropyl-2-oxazoline) (PiPrOx) side chains are synthesized and characterized by NMR and GPC. The grafting density of side chains is 0.49. The molar masses of the graft-copolymer, its backbone, side chains, and the modeling poly-2-isopropyl-2-oxaziline are 74,000, 19,000, 4300, and 16,600 g·mol−1, respectively. Their conformational properties in nitropropane as well as thermoresponsiveness in aqueous solutions are studied and compared with that of free side chains, i.e., linear PiPrOx with a hydrophobic terminal group. In nitropropane, the graft-copolymer adopts conformation of a 13-arm star with a core of a collapsed main chain and a PiPrOx corona. Similarly, a linear PiPrOx chain protects its bulky terminal group by wrapping around it in a selective solvent. In aqueous solutions at low temperatures, graft copolymers form aggregates due to interaction of hydrophobic backbones, which contrasts to molecular solutions of the model linear PiPrOx. The lower critical solution temperature (LCST) for the graft copolymer is around 20 °C. The phase separation temperatures of the copolymer solution were lower than that of the linear chain counterpart, decreasing with concentration for both polymers.


Introduction
Physicochemical properties of molecular brushes (grafted copolymers) in solution and in the bulk strongly depend on architectural parameters, such as degree of polymerization of the main chain, length and grafting density of the side chains [1][2][3][4][5]. Excluded volume repulsion of densely grafted side chains determines the conformation of molecular brushes in good and Θ-solvents by enhancing backbone stiffness characterized by the Kuhn length A. The length scale of this effect depends primarily on the side chains length L s and their grafting density z [5][6][7][8]. In addition to architecture, solution properties of grafted copolymers depend on the interaction of chemically dissimilar side chains and backbone [9][10][11]. Of particular interest are water soluble grafted copolymers constructed from
The NMR spectra were recorded on the Bruker AC 400 spectrometer (400 MHz) for solutions in CDCl 3 . Dialysis was performed using dialysis bags (CellaSep, Orange Scientific, Wallonia, Belgium) with the MWCO of 3500 g·mol −1 .
Polymerization of 2-isopropyl-2-oxazoline on the polyester macroinitiator. The solution of the initiator and monomer in 1,1,2,2-tetrachloroethane (feed ratio monomer/functional groups of macroinitiator 30/1) was heated under argon during 72 h at 70 • C. The solvent was distilled off in the vacuum polymer was dissolved in ethanol, dialyzed against water during 48 h and freeze-dried.

Study of Behavior of the Synthesized Polymers in Organic Solvents
Molecular weight and hydrodynamic properties of APE-g-PiPrOx and PiPrOx-m were characterized in dilute solutions in nitropropane at 21 • C. The weight average molecular weight M w and hydrodynamic radii R h-D were measured by static and dynamic light scattering using a Photocor Complex instrument (Photocor Instrument Inc., Moscow, Russia), equipped with a Photocor-PC2 correlator with 288 channels and a Photocor-PD detector for measuring transmitted light intensity. A Photocor-DL semiconductor laser with a wavelength of λ 0 = 659.1 nm was used as a light source. Toluene was used as a calibration fluid with an absolute scattering intensity of R v = 1.38 × 10 −5 cm −1 . Prior to light scattering measurements, the solutions were filtered into dust-free cells using Chromafil polyamide filters (Macherey-Nagel GmbH & Co. KG, Düren, Germany) with the pore size of 0.45 µm. The values of the refractive index increment dn/dc were measured on the RA-620 refractometer (KEM, Kyoto, Japan).
At all concentrations, the scattered light intensity distribution was unimodal ( Figure 2). The hydrodynamic radii R h-D (c) determined at concentration c in nitropropane did not depend on c ( Figure S1), and the average value of R h-D (c) was taken as the hydrodynamic radius of the macromolecules R h-D .

Study of Behavior of the Synthesized Polymers in Organic Solvents
Molecular weight and hydrodynamic properties of APE-g-PiPrOx and PiPrOx-m were characterized in dilute solutions in nitropropane at 21 °C. The weight average molecular weight Mw and hydrodynamic radii Rh-D were measured by static and dynamic light scattering using a Photocor Complex instrument (Photocor Instrument Inc., Moscow, Russia), equipped with a Photocor-PC2 correlator with 288 channels and a Photocor-PD detector for measuring transmitted light intensity. A Photocor-DL semiconductor laser with a wavelength of λ0 = 659.1 nm was used as a light source. Toluene was used as a calibration fluid with an absolute scattering intensity of Rv = 1.38 × 10 −5 cm −1 . Prior to light scattering measurements, the solutions were filtered into dust-free cells using Chromafil polyamide filters (Macherey-Nagel GmbH & Co. KG, Düren, Germany) with the pore size of 0.45 μm. The values of the refractive index increment dn/dc were measured on the RA-620 refractometer (KEM, Kyoto, Japan).
At all concentrations, the scattered light intensity distribution was unimodal ( Figure 2). The hydrodynamic radii Rh-D(c) determined at concentration c in nitropropane did not depend on c ( Figure S1), and the average value of Rh-D(c) was taken as the hydrodynamic radius of the macromolecules Rh-D. The dependences of relative intensity I/Imax of scattered light on the hydrodynamic radius Rh of scattering species for solutions of APE-g-PiPrOx at c = 0.0073 g·cm −3 (1) and PiPrOx-m at c = 0.0357 g·cm −3 (2) in nitropropane; for solutions of PiPrOx-m at c = 0.0020 g·cm −3 (3) and APE-g-PiPrOx at c = 0.00010 g·cm −3 (4) in water. Imax is the maximum value of light scattering intensity I for given solution concentration; explanations for Rm, Rs, Sm, and Ss see below in Section 3.3. Figure 2. The dependences of relative intensity I/I max of scattered light on the hydrodynamic radius R h of scattering species for solutions of APE-g-PiPrOx at c = 0.0073 g·cm −3 (1) and PiPrOx-m at c = 0.0357 g·cm −3 (2) in nitropropane; for solutions of PiPrOx-m at c = 0.0020 g·cm −3 (3) and APE-g-PiPrOx at c = 0.00010 g·cm −3 (4) in water. I max is the maximum value of light scattering intensity I for given solution concentration; explanations for R m , R s , S m , and S s see below in Section 3.3.
For both APE-g-PiPrOx and PiPrOx-m samples, scattered light asymmetry was absent, and the MM was determined by the Debye method using the following equation where n 0 = 1.3943 is the solvent refractive index and N A is the Avogadro's number. Figure S2 shows the dependence of the inverse light scattering intensity on solution concentration c. A positive slope the dependences of cH/I 90 on c indicates that nitropropane is a thermodynamically good solvent for APE-g-PiPrOx and PiPrOx-m; the values of the second virial coefficient being A 2 = 4.4 × 10 −4 and 5.4 × 10 −4 cm 3 ·mol·g −1 for the first and the second, respectively. The intrinsic viscosity was measured in nitropropane with the Ostwald-type Cannon-Manning capillary viscometer (Cannon Instrument Company Inc., State College, PA, USA) at 21 • C. The solvent efflux time was 48.5 s. The concentration range was (0.0026-0.0096) g/cm 3 for PiPrOx-m and 0.020-0.039 for APE-g-PiPrOx. Ratios of solution to solvent efflux times were always above 1.13.

Study of Behavior of APE-g-PiPrOx and PiPrOx-m in Aqueous Solutions upon Heating
The self-organization processes of APE-g-PiPrOx and PiPrOx-m in aqueous solutions were studied by light scattering and turbidimetry using the Photocor Complex setup described above. The solution temperature T was changed in increments ranging from 5 to 1 K at low temperatures and near the clouding point, respectively. The temperature was controlled with an accuracy of 0.1 • C.
Thermal variations of scattered and transmitted light intensities (I and I*) were used to obtain phase separation temperatures. In addition, hydrodynamic radii R h of scattering particles and their contribution S i to the total scattering intensity were measured as a function of temperature. The value of S i was estimated as the area under the corresponding distribution peak I(R h ). Particle sizes and their contributions were measured after reaching the equilibrium state of the solutions. Under these conditions, the angular dependences of the I, R h and S i values in the range of angles of light scattering from 45 • to 135 • were investigated in order to prove the diffusive nature of the modes, as well as to obtain extrapolated values of R h and S i , which were used in further discussion.
Compound 1 was synthesized using the Spinner method, which involves chlorosulfonation of 2-phenethyl bromide, reduction of the corresponding sulfochloride to sulfinic acid, and addition of the latter to 1,4-benzoquinone. For the macroinitiator synthesis we used high-temperature acceptorless polycondensation [51], which yielded regular copolyesters with polydispersity indexes at about 2.1-2.3. The targeted graft copolymer APE-g-PiPrOx was synthesized under conditions that are conventionally used for polyoxazolines ( Figure 4).
A scheme of the linear model polyester PiPrOx-m synthesis is shown in Figure 5. Keeping in mind the end group effect on thermoresponsiveness, compound 5, whose structure was the closest to that of the macroinitiator, was synthesized as an initiator for the model polymer synthesis. For this purpose, the hydroxyl groups 1 (given in Figure 3) were blocked by benzoyl groups under Einhorn reaction conditions allowing for high yield of the targeted compound.  Figure 3. Synthesis route of APE (3) ie chlorosulfonation of 2-bromoethylbenzene, reduction of arenesulfonylchloride to sulfinic acid, Michael addition the latter to quinone leading to 2,5dihydroxy-4′-(2-bromoethyl)-diphenylsulfone (1) and its polycondensation with 4,4′sebacyloyldioxydibenzoyl dichloride (2).
Compound 1 was synthesized using the Spinner method, which involves chlorosulfonation of 2-phenethyl bromide, reduction of the corresponding sulfochloride to sulfinic acid, and addition of the latter to 1,4-benzoquinone. For the macroinitiator synthesis we used high-temperature acceptorless polycondensation [51], which yielded regular copolyesters with polydispersity indexes at about 2.1-2.3. The targeted graft copolymer APE-g-PiPrOx was synthesized under conditions that are conventionally used for polyoxazolines ( Figure 4).   Compound 1 was synthesized using the Spinner method, which involves chlorosulfonation of 2-phenethyl bromide, reduction of the corresponding sulfochloride to sulfinic acid, and addition of the latter to 1,4-benzoquinone. For the macroinitiator synthesis we used high-temperature acceptorless polycondensation [51], which yielded regular copolyesters with polydispersity indexes at about 2.1-2.3. The targeted graft copolymer APE-g-PiPrOx was synthesized under conditions that are conventionally used for polyoxazolines ( Figure 4). A scheme of the linear model polyester PiPrOx-m synthesis is shown in Figure 5. Keeping in mind the end group effect on thermoresponsiveness, compound 5, whose structure was the closest to that of the macroinitiator, was synthesized as an initiator for the model polymer synthesis. For this purpose, the hydroxyl groups 1 (given in Figure 3) were blocked by benzoyl groups under Einhorn reaction conditions allowing for high yield of the targeted compound. Comparison of 1 H NMR spectra of the macroinitiator and graft copolymer shows that the copolymer spectrum exhibits both signals from aromatic protons in the backbone and from poly(2-isopropyl-2-oxazoline) side chains ( Figure S3). Given the unimodal molecular-mass distribution of the obtained polymer ( Figure S4), this suggested that the synthesized sample was a graft copolymer. Comparison of 1 Н NMR spectra of the macroinitiator and graft copolymer shows that the copolymer spectrum exhibits both signals from aromatic protons in the backbone and from poly(2isopropyl-2-oxazoline) side chains ( Figure S3). Given the unimodal molecular-mass distribution of the obtained polymer ( Figure S4), this suggested that the synthesized sample was a graft copolymer.

Structural and Molecular Characteristics of Grafted Copolymer APE-g-PiPrOx and Linear PiPrOx-m
Molecular and hydrodynamic characteristics obtained for APE-g-PiPrOx and PiPrOx-m are given in Table 1. As is known [52], the density of side chain grafting affects a number of physical characteristics of comb-like polymers, in particular, their behavior in solution. For determining side chains molar mass as well as their grafting density, polymer samples were subjected to alkali hydrolysis under conditions providing the quantitative scission of the polyester chain: 1 M KOH solution in ethyl 2-ethoxyethanol at 120 °С for 10 min [53]. The density of side chain grafting was found from the molar mass ratio of the backbone and side chains. With regular distribution of the initiating groups throughout the macroinitiator chain, the grafting density was determined by polymerization initiation efficiency. This parameter reflects the fraction of the macroinitiator initiating groups having reacted from the total content of the initiating groups, which was calculated as the ratio of the theoretical molar mass of side chains to the experimental value [54]. In the model experiment, under similar conditions, poly-2-ethyl-2-oxazoline was stable for at least 1 h. It was found that oligo-2-isopropyl-2-oxazoline side chains had Ms = (4300 ± 600) g·mol −1 and polydispersity Mw/Mn = 1.45 ± 0.19. Table 1. Molar mass, polydispersity index (PDI), degree of polymerization (DP) and hydrodynamic characteristics of grafted copolymer APE-g-PiPrOx, linear PiPrOx-m, APE macroinitiator, and abstracted side chains.

Structural and Molecular Characteristics of Grafted Copolymer APE-g-PiPrOx and Linear PiPrOx-m
Molecular and hydrodynamic characteristics obtained for APE-g-PiPrOx and PiPrOx-m are given in Table 1. As is known [52], the density of side chain grafting affects a number of physical characteristics of comb-like polymers, in particular, their behavior in solution. For determining side chains molar mass as well as their grafting density, polymer samples were subjected to alkali hydrolysis under conditions providing the quantitative scission of the polyester chain: 1 M KOH solution in ethyl 2-ethoxyethanol at 120 • C for 10 min [53]. The density of side chain grafting was found from the molar mass ratio of the backbone and side chains. With regular distribution of the initiating groups throughout the macroinitiator chain, the grafting density was determined by polymerization initiation efficiency. This parameter reflects the fraction of the macroinitiator initiating groups having reacted from the total content of the initiating groups, which was calculated as the ratio of the theoretical molar mass of side chains to the experimental value [54]. In the model experiment, under similar conditions, poly-2-ethyl-2-oxazoline was stable for at least 1 h. It was found that oligo-2-isopropyl-2-oxazoline side chains had M s = (4300 ± 600) g·mol −1 and polydispersity M w /M n = 1.45 ± 0.19. Table 1. Molar mass, polydispersity index (PDI), degree of polymerization (DP) and hydrodynamic characteristics of grafted copolymer APE-g-PiPrOx, linear PiPrOx-m, APE macroinitiator, and abstracted side chains.
19,000 27 -side chain 4300 1.45 37 --Grafting density z was calculated as where N 1 + N 2 = DP APE is the polymerization degree of APE macroinitiator, N 1 and N 2 are respectively the numbers of monomeric units of APE which contain or not contain PiPrOx side chain, M cop is molar mass of APE-g-PiPrOx copolymer (Figure 1), and M b = 19,000 g·mol −1 [55] is the backbone molar mass. According to calculations, z = 0.49, i.e., one side chain corresponded to two monomeric units of the copolymer. However, the APE monomers were very long, and in terms of the number of valence bonds, the grafting density was z* = 1/30 = 0.03, which suggested that APE-g-PiPrOx graft copolymer should be considered as a polymer comb [56], or as they are also called, loosely grafted molecular brushes [57]. The intrinsic viscosity [η] and hydrodynamic radius R h of the APE-g-PiPrOx graft copolymer and the linear PiPrOx were analyzed using the Mark-Kuhn-Houwink-Sakurada (MKHS) relationships obtained for some poly(2-alkyl-2-oxazolines) in [58,59] (see Figure 6). As is known, the intrinsic viscosity [η] is related to the molecular weight by the MKHS equation [η] = K x M a , where the parameter K x and the exponent a are constant values for a given polymer-solvent system. The value of [η] depends on both the size (MM) and the shape (conformation) of the macromolecule. The linear PiPrOx-m demonstrate a [η] value similar to that of poly(2-ethyl-and poly(2-methyl-2-oxazolines) given in literature [58,59], which is consistent with the similar chain stiffness. Therefore, it can be assumed that the conformational properties of PiPrOx-m with a molecular weight close to PAOx shown in the graph were similar. In contrast, the significantly lower [η] of the APE-g-PiPrOx graft copolymer suggests that their macromolecules were more compact than linear poly(2-alkyl-2-oxazoline)s. where N1 + N2 = DPAPE is the polymerization degree of APE macroinitiator, N1 and N2 are respectively the numbers of monomeric units of APE which contain or not contain PiPrOx side chain, Mcop is molar mass of APE-g-PiPrOx copolymer (Figure 1), and Mb = 19,000 g·mol −1 [55] is the backbone molar mass. According to calculations, z = 0.49, i.e., one side chain corresponded to two monomeric units of the copolymer. However, the APE monomers were very long, and in terms of the number of valence bonds, the grafting density was z* = 1/30 = 0.03, which suggested that APE-g-PiPrOx graft copolymer should be considered as a polymer comb [56], or as they are also called, loosely grafted molecular brushes [57]. The intrinsic viscosity [η] and hydrodynamic radius Rh of the APE-g-PiPrOx graft copolymer and the linear PiPrOx were analyzed using the Mark-Kuhn-Houwink-Sakurada (MKHS) relationships obtained for some poly(2-alkyl-2-oxazolines) in [58,59] (see Figure 6). As is known, the intrinsic viscosity [η] is related to the molecular weight by the MKHS equation [η] = KxM a , where the parameter Kx and the exponent a are constant values for a given polymer-solvent system. The value of [η] depends on both the size (MM) and the shape (conformation) of the macromolecule. The linear PiPrOx-m demonstrate a [η] value similar to that of poly(2-ethyl-and poly(2-methyl-2-oxazolines) given in literature [58,59], which is consistent with the similar chain stiffness. Therefore, it can be assumed that the conformational properties of PiPrOx-m with a molecular weight close to PAOx shown in the graph were similar. In contrast, the significantly lower [η] of the APE-g-PiPrOx graft copolymer suggests that their macromolecules were more compact than linear poly(2-alkyl-2oxazoline)s.  [58] and 2 [59]) and poly(2-methyl-2-oxazoline) (PMeOx) (line 3 [58]).
These findings are consistent with our previous study of PI-g-PMMA and PI-g-PS graft copolymers with polyimide (PI) backbone and polymethylmethacrylate (PMMA) and polystyrene (PS) side chains in selective solvents [25]. The intrinsic viscosity of PI-g-PMMA and PI-g-PS did not depend on MM in the range from 270,000 to 2,000,000 g·mol −1 , and the [η] value for grafted copolymer solutions was 4-8 times lower than the intrinsic viscosity of linear PMMA (Mw = 490,000 g·mol −1 ) and PS (Mw = 370,000 g·mol −1 ). This behavior is explained by the fact that in selective solvents where the main chain does not dissolve, the macromolecules of loose cylindrical brushes resemble multiarm star-shaped molecules [24]. The core of the star is a collapsed main chain, and the arms are more or less folded side chains, which should be considered as a corona ensuring the solubility of the copolymer. These findings are consistent with our previous study of PI-g-PMMA and PI-g-PS graft copolymers with polyimide (PI) backbone and polymethylmethacrylate (PMMA) and polystyrene (PS) side chains in selective solvents [25]. The intrinsic viscosity of PI-g-PMMA and PI-g-PS did not depend on MM in the range from 270,000 to 2,000,000 g·mol −1 , and the [η] value for grafted copolymer solutions was 4-8 times lower than the intrinsic viscosity of linear PMMA (M w = 490,000 g·mol −1 ) and PS (M w = 370,000 g·mol −1 ). This behavior is explained by the fact that in selective solvents where the main chain does not dissolve, the macromolecules of loose cylindrical brushes resemble multiarm star-shaped molecules [24]. The core of the star is a collapsed main chain, and the arms are more or less folded side chains, which should be considered as a corona ensuring the solubility of the copolymer.
As shown by computer simulation [24], compact conformation requires (i) sufficiently high flexibility of the main chain and (ii) its access to the solvent. Both of these conditions were fulfilled for the studied APE-g-PiPrOx sample. The Kuhn segment length of APE did not exceed 2 nm, that is, the main chain was a typical flexible-chain polymer. On the other hand, the average distance ∆L between the grafted chains along the APE chain was not very different from the length L s of the latter (Figure 7). Indeed, where L b is the contour length of the main chain and λ b is the length of the APE 8 monomer unit.
Assuming that the length of all valence bonds of the main chain was close to 0.14 nm, and the valence angles were tetrahedral, we obtained λ b ≈ 3.5 nm and ∆L ≈ 7.7 nm. The average length of the PiPrOx side chains was L s = λ s N s = λ s M PiPrOx /M s0 ≈ ≈ 13.8 nm, where M PiPrOx and N s ≈ 37 are the MM and the polymerization degree of the grafted PiPrOx, M s0 = 112 g·mol −1 and λ s = 0.378 nm are the molar mass and the length of PiPrOx monomer unit, respectively. Therefore, ∆L and L s differed less than two times, and the length of the side chains was not sufficient to reliably shield the main chain from the solvent. This led to the collapse of the main chain, and the shape of the grafted copolymer molecule resembled a star-shaped molecule (Figure 7). Note that a similar conformation of comb-shaped macromolecules with a relatively short main chain was predicted by Tsvetkov over 30 years ago [60].
As shown by computer simulation [24], compact conformation requires (i) sufficiently high flexibility of the main chain and (ii) its access to the solvent. Both of these conditions were fulfilled for the studied APE-g-PiPrOx sample. The Kuhn segment length of APE did not exceed 2 nm, that is, the main chain was a typical flexible-chain polymer. On the other hand, the average distance ΔL between the grafted chains along the APE chain was not very different from the length Ls of the latter (Figure 7). Indeed, where Lb is the contour length of the main chain and λb is the length of the APE8 monomer unit.
Assuming that the length of all valence bonds of the main chain was close to 0.14 nm, and the valence angles were tetrahedral, we obtained λb ≈ 3.5 nm and ΔL ≈ 7.7 nm. The average length of the PiPrOx side chains was Ls = λsNs = λsMPiPrOx/Ms0 ≈ ≈ 13.8 nm, where MPiPrOx and Ns ≈ 37 are the ММ and the polymerization degree of the grafted PiPrOx, Ms0 = 112 g·mol −1 and λs = 0.378 nm are the molar mass and the length of PiPrOx monomer unit, respectively. Therefore, ΔL and Ls differed less than two times, and the length of the side chains was not sufficient to reliably shield the main chain from the solvent. This led to the collapse of the main chain, and the shape of the grafted copolymer molecule resembled a star-shaped molecule (Figure 7). Note that a similar conformation of comb-shaped macromolecules with a relatively short main chain was predicted by Tsvetkov over 30 years ago [60]. The behavior of APE-g-PiPrOx in nitropropane corresponded to that of a star-shaped molecule with a relatively massive core and the number of arms fa corresponding to the number of PiPrOx grafted chains, namely, fa = z(N1 + N2) ≈ 13. The length of arms Ls did not significantly differ from the hydrodynamic radius Rh-D of the APE-g-PiPrOx macromolecules. The ratio Ls/Rh-D is Ls/Rh-D = 1.5, which suggested that the grafted chains had the conformation of a slightly bent rod. This conclusion is in qualitative agreement with the predictions of Daoud and Cotton [61] theory describing the structure of multiarm polymer stars.
As for the model PiPrOx-m, its intrinsic viscosity and hydrodynamic radius were also lower than [η] and Rh for linear PAOx (Figure 6). Therefore, its molecules had a more compact structure compared to that of the PAOx molecules of the same MM. This is because the massive hydrophobic PiPrOx-m in selective solvent The behavior of APE-g-PiPrOx in nitropropane corresponded to that of a star-shaped molecule with a relatively massive core and the number of arms f a corresponding to the number of PiPrOx grafted chains, namely, f a = z(N 1 + N 2 ) ≈ 13. The length of arms L s did not significantly differ from the hydrodynamic radius R h-D of the APE-g-PiPrOx macromolecules. The ratio L s /R h-D is L s /R h-D = 1.5, which suggested that the grafted chains had the conformation of a slightly bent rod. This conclusion is in qualitative agreement with the predictions of Daoud and Cotton [61] theory describing the structure of multiarm polymer stars.
As for the model PiPrOx-m, its intrinsic viscosity and hydrodynamic radius were also lower than [η] and R h for linear PAOx (Figure 6). Therefore, its molecules had a more compact structure compared to that of the PAOx molecules of the same MM. This is because the massive hydrophobic end group was insoluble, and the linear PiPrOx chain shielded it from the solvent, providing an increase in intramolecular density and a decrease in the size of PiPrOx-m molecules (Figure 7). The same was true for solutions of diphilic linear dendritic block copolymers [62][63][64][65][66]. The analysis of the results obtained for them made it possible to propose a solubility model of such hybrid systems [65]. In particular, it was shown that their behavior in solutions was significantly affected by both the thermodynamic quality of the solvent with respect to the blocks and the size ratios of the soluble linear chain and the insoluble branched component. The decrease in the length of the linear block at fixed sizes of the second block resulted in the appearance of supramolecular micelle-like structures in solutions, where insoluble components of different macromolecules form the core, and the linear chains wrapping around the latter ensure the solubility of the aggregates. In the case of the studied PiPrOx-m, the size of the terminal group is approximately 1.5-2.0 nm, the length of the linear PiPrOx chain is close to 54 nm, that is, it can "wrap" around the hydrophobic fragment almost 10 times. In contrast, the length of PiPrOx in the isolated side chain is insufficient to screen the end group, and large aggregates are formed in their solutions in nitropropane and water. To conclude the conformational analysis, we note that the size of poly-2-ethyl-oxazoline molecules in tetrahydrofuran [67] is smaller than in other solvents [58,59], and their shape is closer to spherical in the authors' opinion. This is probably how the effect of the thermodynamic quality of the solvent on the solubility of PAOx is manifested.

Behavior of Grafted APE-g-PiPrOx Copolymer and Linear PiPrOx-m in Aqueous Solutions
In PiPrOx-m aqueous solutions at low temperatures, a single mode was observed by dynamic light scattering (Figure 2). The hydrodynamic radius R h of particles in solution did not depend on concentration ( Figure S5), while the average value of R h = (3.9 ± 0.7) nm matched the hydrodynamic sizes of PiPrOx-m molecules in nitropropane. Therefore, at low temperatures, aqueous PiPrOx-m solutions were molecularly dispersed. This conclusion is also supported by good agreement between the PiPrOx-m molar mass M w = (17,200 ± 900) g·mol −1 , determined in both water at 21 • C and nitropropane ( Figure S2). Thus, the presence of a massive hydrophobic group at one end of the model PiPrOx-m chain does not lead to formation of core-shell aggregates, including star-like and flower-like ones, which are often observed in solutions of linear amphiphilic PAOx [68][69][70][71][72][73]. Similar to nitropropane, hydrophobic fragments were completely shielded by long hydrophilic PiPrOx chains consisting of more than 140 monomer units.
In low temperature APE-g-PiPrOx solutions, two types of scattering objects with hydrodynamic radii R m (Figure 2, line 4) and R s (Figure 2, line 5) were found that significantly exceeded the size of the molecules of the grafted copolymer R h-D . This suggests particles aggregations due to attraction of hydrophobic APE main chains, as well as dehydrated PiPrOx units. The radii R m and R s decreased with dilution ( Figure S5), which could reflect both the concentration dependence of the diffusion coefficient and the gradual destruction of aggregates with decreasing solution concentration.
It should be noted that at all concentrations the weight fraction of large aggregates with radius R s was small. Indeed, the contribution of the i-th set of particles to the total light scattering intensity I is described by the relation where c i and R i are the weight concentration and radius of i-th particles, respectively [74]. The exponent x is determined by the shape of the particles: for hard rods x = 1, for coils x = 2 and for hard spheres x = 3. Using relation (4), we can roughly estimate the maximum value of the ratio c s /c m of the weight concentrations of aggregates of large c s and lower c m sizes: where I s and I m are contributions to the total light scattering of aggregates with radii R s and R m , respectively. For the studied APE-g-PiPrOx solutions at low temperatures, the ratio of the aggregates contribution I s /I m = S s /S m to the total light scattering I ranges from 0.1 to 0.3. Here S s and S m are the areas under the peaks corresponding to aggregates with radii R s and R m on dependences of relative intensity I/I max of scattered light on the hydrodynamic radius ( Figure 2). Therefore, the maximum value of S s /S m = 0.3, while the minimum value of the ratio of aggregate radii R s /R m = 7. Thus, assuming that the conformation of the aggregates is close to coil-like (x = 2), we obtain the maximum value c s /c m ≈ 0.007. Therefore, more than 99% of the polymer molecules formed aggregates with the radius R m , while large aggregates contained a maximum of 0.6% of the total number of molecules. Earlier, two types of particles were also found in solutions of the grafted APE-g-PEtOx copolymer with the polyester main chain and the side chains of poly(2-ethyl-2-oxazoline) at room temperature [46,47]. However, in the case of APE-g-PEtOx, these particles were grafted polymer molecules and aggregates similar in characteristics to smaller aggregates in APE-g-PiPrOx solutions. Considering that the studied APE-g-PiPrOx and APE-g-PEtOx samples had similar grafting densities z and side chain lengths L s , we concluded that the difference in the behavior of the compared polymers was due to the different hydrophilicity of PiPrOx and PEtOx. More hydrophobic PiPrOx chains contributed to more intense aggregation.
When heating PiPrOx-m aqueous solutions, phase separation was observed, which manifested itself in the sharp increase in the light scattering intensity I and the corresponding decrease in the optical transmittance I* (Figure 8). The profiles of the dependences I/I 21 (T) and I*/I* 21 (T) obtained for PiPrOx-m solutions of various concentrations were similar to those for thermosensitive polymers characterized by LCST behavior [35,71,[75][76][77][78][79]. The temperatures of the phase separation onset, T 1 and T 1 *, and termination, T 2 and T 2 *, obtained from both dependences, coincided pairwise. The change in I and I* in the interval from T 1 to T 2 was due to the aggregation of macromolecules. Large micron aggregates appeared in solutions as a result of the formation of hydrophobic intermolecular contacts due to dehydration of PiPrOx-m units upon heating. The size of aggregates increased with increasing T in the interval of phase separation and reached the maximum values near T 2 ( Figure 9). Above T 2 , the size of the aggregates decreased, which may reflect the compaction of macromolecules, but we should remember that under these conditions the solution was cloudy and light scattering was not classical. Figure 10 shows the dependences of light scattering intensity I for aqueous solutions of APE-g-PiPrOx. The value of I monotonically increased with temperature at an increasing rate dI/dT, which does not allow determining the temperature of the onset of phase separation T 1 using light scattering data. The intensity I reached its maximum value at temperature T 2 , which for all studied solutions was similar to T* 2 , corresponding to the termination of phase separation according to the turbidity data ( Figure 10). The temperatures of the phase separation onset T 1 were determined by the beginning of the I* decline. intensity I/Imax of scattered light on the hydrodynamic radius ( Figure 2). Therefore, the maximum value of Ss/Sm = 0.3, while the minimum value of the ratio of aggregate radii Rs/Rm = 7. Thus, assuming that the conformation of the aggregates is close to coil-like (x = 2), we obtain the maximum value сs/сm ≈ 0.007. Therefore, more than 99% of the polymer molecules formed aggregates with the radius Rm, while large aggregates contained a maximum of 0.6% of the total number of molecules. Earlier, two types of particles were also found in solutions of the grafted APE-g-PEtOx copolymer with the polyester main chain and the side chains of poly(2-ethyl-2-oxazoline) at room temperature [46,47]. However, in the case of APE-g-PEtOx, these particles were grafted polymer molecules and aggregates similar in characteristics to smaller aggregates in APE-g-PiPrOx solutions. Considering that the studied APE-g-PiPrOx and APE-g-PEtOx samples had similar grafting densities z and side chain lengths Ls, we concluded that the difference in the behavior of the compared polymers was due to the different hydrophilicity of PiPrOx and PEtOx. More hydrophobic PiPrOx chains contributed to more intense aggregation.
When heating PiPrOx-m aqueous solutions, phase separation was observed, which manifested itself in the sharp increase in the light scattering intensity I and the corresponding decrease in the optical transmittance I* (Figure 8). The profiles of the dependences I/I21(T) and I*/I*21(T) obtained for PiPrOx-m solutions of various concentrations were similar to those for thermosensitive polymers characterized by LCST behavior [35,71,[75][76][77][78][79]. The temperatures of the phase separation onset, T1 and T1*, and termination, T2 and T2*, obtained from both dependences, coincided pairwise. The change in I and I* in the interval from T1 to T2 was due to the aggregation of macromolecules. Large micron aggregates appeared in solutions as a result of the formation of hydrophobic intermolecular contacts due to dehydration of PiPrOx-m units upon heating. The size of aggregates increased with increasing T in the interval of phase separation and reached the maximum values near T2 (Figure 9). Above T2, the size of the aggregates decreased, which may reflect the compaction of macromolecules, but we should remember that under these conditions the solution was cloudy and light scattering was not classical.   Figure 10 shows the dependences of light scattering intensity I for aqueous solutions of APE-g-PiPrOx. The value of I monotonically increased with temperature at an increasing rate dI/dT, which does not allow determining the temperature of the onset of phase separation T1 using light scattering data. The intensity I reached its maximum value at temperature T2, which for all studied solutions was similar to T*2, corresponding to the termination of phase separation according to the turbidity data ( Figure 10). The temperatures of the phase separation onset T1 were determined by the beginning of the I* decline. As can be seen in 11, at T < T1, the hydrodynamic radii of the aggregates Rm and Rs slowly decreased upon heating, i.e., aggregates compacted in this temperature range due to dehydration of PiPrOx chains and the formation of hydrogen bonds, both on the intramolecular and intermolecular level. Note that the data obtained did not allow us to make a conclusion about which aggregates, large or small, changed their size more significantly. For example, at the concentration of c = 0.00020   Figure 10 shows the dependences of light scattering intensity I for aqueous solutions of APE-g-PiPrOx. The value of I monotonically increased with temperature at an increasing rate dI/dT, which does not allow determining the temperature of the onset of phase separation T1 using light scattering data. The intensity I reached its maximum value at temperature T2, which for all studied solutions was similar to T*2, corresponding to the termination of phase separation according to the turbidity data ( Figure 10). The temperatures of the phase separation onset T1 were determined by the beginning of the I* decline. As can be seen in 11, at T < T1, the hydrodynamic radii of the aggregates Rm and Rs slowly decreased upon heating, i.e., aggregates compacted in this temperature range due to dehydration of PiPrOx chains and the formation of hydrogen bonds, both on the intramolecular and intermolecular level. Note that the data obtained did not allow us to make a conclusion about which aggregates, large or small, changed their size more significantly.  Figure 10. The temperature dependences of relative light scattering intensity I/I 10 and relative transmittance I*/I* 10 for aqueous solutions of APE-g-PiPrOx. I 10 and I* 10 are intensities of light scattering and optical transmittance at T = 10 • C, respectively.
As can be seen in 11, at T < T 1 , the hydrodynamic radii of the aggregates R m and R s slowly decreased upon heating, i.e., aggregates compacted in this temperature range due to dehydration of PiPrOx chains and the formation of hydrogen bonds, both on the intramolecular and intermolecular level. Note that the data obtained did not allow us to make a conclusion about which aggregates, large or small, changed their size more significantly. For example, at the concentration of c = 0.00020 g·cm −3 , the radius R s decreased noticeably faster, and at c = 0.0105 g·cm −3 , R m changed slightly more significantly (Figure 11). Compaction was accompanied by aggregation, as indicated by the increase in the ratio S s /S m of the contributions of smaller S m and large S s aggregates to integral light scattering ( Figure 11). As well as compaction, the formation of new aggregates was due to the occurrence of hydrogen bonds between the dehydrated PiPrOx units. Within the framework of the model of coil-shaped scattering objects, for a solution with c = 0.00020 g·cm −3 , the ratio of the weight concentrations of aggregates c s /c m varied from 0.006 at T = 10 • C to 0.4 at T 1 and up to 1.0 at T 2 .
Obviously, this estimation is approximate, since compaction may be accompanied by both shape and density variation of aggregates. Yet it correctly reflected the tendency towards phase separation of APE-g-PiPrOx solutions upon heating. g·cm −3 , the radius Rs decreased noticeably faster, and at c = 0.0105 g·cm −3 , Rm changed slightly more significantly (Figure 11). Compaction was accompanied by aggregation, as indicated by the increase in the ratio Ss/Sm of the contributions of smaller Sm and large Ss aggregates to integral light scattering ( Figure 11). As well as compaction, the formation of new aggregates was due to the occurrence of hydrogen bonds between the dehydrated PiPrOx units. Within the framework of the model of coilshaped scattering objects, for a solution with c = 0.00020 g·cm −3 , the ratio of the weight concentrations of aggregates сs/сm varied from 0.006 at T = 10 °C to 0.4 at T1 and up to 1.0 at T2. Obviously, this estimation is approximate, since compaction may be accompanied by both shape and density variation of aggregates. Yet it correctly reflected the tendency towards phase separation of APE-g-PiPrOx solutions upon heating. For both polymers investigated in this paper, the T1 and T2 temperatures as well as the phase separation range increase with dilution ( Figure 12). For the linear model PiPrOx-m, a strong concentration dependence of phase separation temperatures is retained in the entire studied concentration range. The minimum value of T1 = 35 °C is close to LCST of solutions of linear PiPrOx [34,35,80,81]. For example, Uyama and Kobayashi [81] showed that for a PiPrOx sample with molar mass M = 16,700 g·mol −1 , the phase transition temperature is ≈36 °С.  For both polymers investigated in this paper, the T 1 and T 2 temperatures as well as the phase separation range increase with dilution ( Figure 12). For the linear model PiPrOx-m, a strong concentration dependence of phase separation temperatures is retained in the entire studied concentration range. The minimum value of T 1 = 35 • C is close to LCST of solutions of linear PiPrOx [34,35,80,81]. For example, Uyama and Kobayashi [81] showed that for a PiPrOx sample with molar mass M = 16,700 g·mol −1 , the phase transition temperature is ≈36 • C. g·cm , the radius Rs decreased noticeably faster, and at c = 0.0105 g·cm , Rm changed slightly more significantly (Figure 11). Compaction was accompanied by aggregation, as indicated by the increase in the ratio Ss/Sm of the contributions of smaller Sm and large Ss aggregates to integral light scattering ( Figure 11). As well as compaction, the formation of new aggregates was due to the occurrence of hydrogen bonds between the dehydrated PiPrOx units. Within the framework of the model of coilshaped scattering objects, for a solution with c = 0.00020 g·cm −3 , the ratio of the weight concentrations of aggregates сs/сm varied from 0.006 at T = 10 °C to 0.4 at T1 and up to 1.0 at T2. Obviously, this estimation is approximate, since compaction may be accompanied by both shape and density variation of aggregates. Yet it correctly reflected the tendency towards phase separation of APE-g-PiPrOx solutions upon heating. For both polymers investigated in this paper, the T1 and T2 temperatures as well as the phase separation range increase with dilution ( Figure 12). For the linear model PiPrOx-m, a strong concentration dependence of phase separation temperatures is retained in the entire studied concentration range. The minimum value of T1 = 35 °C is close to LCST of solutions of linear PiPrOx [34,35,80,81]. For example, Uyama and Kobayashi [81] showed that for a PiPrOx sample with molar mass M = 16,700 g·mol −1 , the phase transition temperature is ≈36 °С.  In the case of APE-g-PiPrOx gradt copolymer, the T 1 and T 2 dependences leveled off at high concentrations, suggesting that the LCST of the studied molecular brush was close to 20 • C, which was noticeably lower than typical LCST values for linear PiPrOx, including the studied PiPrOx-m. In addition to the difference in architecture between APE-g-PiPrOx and PiPrOx-m molecules, the observed LCST variation may be caused stronger hydrophobicity of the molecular brush and almost five-fold difference in the MM of the compared samples.

Kinetics of Thermal Response of APE8-g-PiPrOx and PiPrOx-m Solutions
All the results discussed above relate to the "equilibrium" state of solutions, that is, to a state where the characteristics of the solution do not change over time. There are many reports that after exposure to a stimulus, the parameters of stimuli-responsive systems gradually change in time [70,72,[82][83][84][85], and for polymers with complex architecture, this process may last tens of hours [86].
It is convenient to use the time t eq , during which the light scattering intensity I and the optical transmission I* reach constant time values, as a characteristic of the duration of the "equilibrium" state of the system establishing ( Figure S6). The time t eq depends on the temperature, reaching a maximum value of t max eq near the temperature T 1 of the onset of phase separation ( Figure S7). For both polymers, at T > T 1 , the value of t eq rapidly decreased with heating. As mentioned above, for a molecular brush, the light scattering intensity I also depended on temperature at T < T 1 ; therefore, for this polymer, in the given temperature range, the values of t eq could be determined using the dependences of I on t. At low temperatures, t eq rapidly increased with increasing T. Similar temperature dependences of the settling time t eq were previously observed for polymers of various architectures, and are probably associated with intense aggregation at the beginning of phase separation [47,86].
It is unusual that for both polymers studied, t max eq is highly concentration dependent, increasing with dilution ( Figure 13). It can be assumed that this was due to the large sizes of aggregates at low concentrations. However, this issue requires additional research. The maximum time to establish an "equilibrium" state for APE-g-PiPrOx solutions was 11,000 s, which is comparable with t max eq values for APE-g-PEtOx with a close MM [46,47] and for polymer brushes with polyimide backbone and poly (N,N-dimethylaminoethyl methacrylate) side chains [29], but noticeably less than t max eq for star-shaped PAOx [86]. For model PiPrOx-m, the t max eq times were noticeably lower, ranging from 2000 to 4000 s. The acceleration of self-organization processes for linear polymers compared to polymers of complex architecture was typical. However, in addition to this, an almost five-fold difference in the MM of the PiPrOx-m and APE-g-PiPrOx samples should be taken into account. Indeed, the molecular mass dependence of t max eq can be quite strong. For example, for high molecular weight samples of statistical copolymers of N-isopropylacrylamide with ionogenic comonomers, the t max eq values reached 9000 s [87].
was noticeably lower than typical LCST values for linear PiPrOx, including the studied PiPrOx-m. In addition to the difference in architecture between APE-g-PiPrOx and PiPrOx-m molecules, the observed LCST variation may be caused stronger hydrophobicity of the molecular brush and almost five-fold difference in the MM of the compared samples.

Kinetics of Thermal Response of APE8-g-PiPrOx and PiPrOx-m Solutions
All the results discussed above relate to the "equilibrium" state of solutions, that is, to a state where the characteristics of the solution do not change over time. There are many reports that after exposure to a stimulus, the parameters of stimuli-responsive systems gradually change in time [70,72,[82][83][84][85], and for polymers with complex architecture, this process may last tens of hours [86].
It is convenient to use the time teq, during which the light scattering intensity I and the optical transmission I* reach constant time values, as a characteristic of the duration of the "equilibrium" state of the system establishing ( Figure S6). The time teq depends on the temperature, reaching a maximum value of near the temperature T1 of the onset of phase separation ( Figure S7). For both polymers, at T > T1, the value of teq rapidly decreased with heating. As mentioned above, for a molecular brush, the light scattering intensity I also depended on temperature at T < T1; therefore, for this polymer, in the given temperature range, the values of teq could be determined using the dependences of I on t. At low temperatures, teq rapidly increased with increasing T. Similar temperature dependences of the settling time teq were previously observed for polymers of various architectures, and are probably associated with intense aggregation at the beginning of phase separation [47,86].
It is unusual that for both polymers studied, is highly concentration dependent, increasing with dilution ( Figure 13). It can be assumed that this was due to the large sizes of aggregates at low concentrations. However, this issue requires additional research. The maximum time to establish an "equilibrium" state for APE-g-PiPrOx solutions was 11,000 s, which is comparable with values for APE-g-PEtOx with a close MM [46,47] and for polymer brushes with polyimide backbone and poly (N,N-dimethylaminoethyl methacrylate) side chains [29], but noticeably less than for starshaped PAOx [86]. For model PiPrOx-m, the times were noticeably lower, ranging from 2000 to 4000 s. The acceleration of self-organization processes for linear polymers compared to polymers of complex architecture was typical. However, in addition to this, an almost five-fold difference in the MM of the PiPrOx-m and APE-g-PiPrOx samples should be taken into account. Indeed, the molecular mass dependence of can be quite strong. For example, for high molecular weight samples of statistical copolymers of N-isopropylacrylamide with ionogenic comonomers, the values reached 9000 s [87].

Conclusions
New thermosensitive grafted copolymer was synthesized successfully. The backbone was polycondensation polymer, namely, an aromatic polyester. Polymerization poly(2-isopropyl-2-oxazoline), i.e., the polymer belonging to a fundamentally different chemical class, was used as side chain. This structure of the components determines the diphilicity of the resulting polymer in most solvents. APE-g-PiPrOx should be considered as a loose molecular brush because of their rather low density of side chains grafting (z = 0.49) and the ratio of the length of the PiPrOx side chains to the distance between the points of their grafting is less than 2. For comparison the linear PiPrOx sample with a hydrophobic terminal group modeling the side chain and a fragment of backbone of the grafted copolymer was synthesized.
Analysis of molar mass and hydrodynamic characteristics of the obtained polymers in organic solvent made it possible to conclude that the structures of the APE-g-PiPrOx molecules are more compact as compared to that of the PAOx molecules of the same MM. The conformation of the grafted copolymer macromolecule resembles a star, where the collapsed APE backbone forms a relatively massive core, and PiPrOx side chains serve as arms being in a folded conformation. The massive terminal group of the linear PiPrOx-m molecule is also shielded from the solvent by a sufficiently long linear PiPrOx chain which is wrapping around it forming a shell. This provides an increase in intramolecular density and a decrease in the size of PiPrOx-m macromolecules.
The thermosensitivity of both APE-g-PiPrOx and PiPrOx-m was shown. The phase separation of the APE-g-PiPrOx is preceded by the compaction of dissolved particles, namely aggregates of two types, as a result of dehydration of PiPrOx chains and formation of intramolecular and intermolecular hydrogen bonds. Above the phase separation temperature, aggregates sizes increase rapidly. The linear modelling PiPrOx-m exhibits the behavior usual for thermosensitive linear polymers.
The phase separation temperatures of the both polymers decrease with concentration. The LCST of the APE-g-PiPrOx under consideration, at given molecular and architectural characteristics, is assumed to be not much different from 20 • C, that is noticeably lower than typical LCST values for linear PiPrOx, including the studied PiPrOx-m. The difference in the APE-g-PiPrOx and PiPrOx-m thermosensitivity characteristics upon heating is conditioned not only by the specific architecture and the great hydrophobicity of the molecular brush, but also the almost five-fold difference in the MM of the compared polymers.