Thermoresponsive Molecular Brushes with a Rigid-Chain Aromatic Polyester Backbone and Poly-2-alkyl-2-oxazoline Side Chains

A new polycondensation aromatic rigid-chain polyester macroinitiator was synthesized and used to graft linear poly-2-ethyl-2-oxazoline as well as poly-2-isopropyl-2-oxazoline by cationic polymerization. The prepared copolymers and the macroinitiator were characterized by NMR, GPC, AFM, turbidimetry, static, and dynamic light scattering. The molar masses of the polyester main chain and the grafted copolymers with poly-2-ethyl-2-oxazoline and poly-2-isopropyl-2-oxazoline side chains were 26,500, 208,000, and 67,900, respectively. The molar masses of the side chains of poly-2-ethyl-2-oxazoline and poly-2-isopropyl-2-oxazoline and their grafting densities were 7400 and 3400 and 0.53 and 0.27, respectively. In chloroform, the copolymers conformation can be considered as a cylinder wormlike chain, the diameter of which depends on the side chain length. In water at low temperatures, the macromolecules of the poly-2-ethyl-2-oxazoline copolymer assume a wormlike conformation because their backbones are well shielded by side chains, whereas the copolymer with short side chains and low grafting density strongly aggregates, which was visualized by AFM. The phase separation temperatures of the copolymers were lower than those of linear analogs of the side chains and decreased with the concentration for both samples. The LCST were estimated to be around 45 °C for the poly-2-ethyl-2-oxazoline graft copolymer, and below 20 °C for the poly-2-isopropyl-2-oxazoline graft copolymer.


Introduction
The synthesis of rigid-chain polymers in the middle of the 20th century was a breakthrough in polymer science and technology, which made it possible to obtain new ultrastrong materials. Rigid-chain polymers usually include polymers with a Kuhn segment length A of more than 10 nm. The high rigidity of macromolecules predetermines their elongated state in the direction of the main chain and often provides lyotropic mesomorphism [1][2][3][4][5]. Such features are useful for creating high-strength fibers and films and selfreinforcing plastics, including those with specific optical and conductive properties [6][7][8][9].
The establishment of the correlation of molecular characteristics of such polymers with their structure has been and is still being actively pursued by many scientific groups. Analysis of the various molecular structures showed that the high chain equilibrium and kinetic rigidity is the effect of a specific chemical structure of macromolecules [10,11]. The main reason for the increase in A of chain molecules is the presence in their repeating units of structural elements that prevent rotation around the chain direction, such as double thermosensitive PEtOx or PiPrOx are grafted to a rigid-chain aromatic polyester (APEr.ch.); (ii) determination of hydrodynamic and conformational characteristics of the synthesized APEr.ch.-graft-PAlOx; and (iii) the study of the thermal response of aqueous solutions of the synthesized molecular brushes and its dependence on the chemical structure of the side PAlOx chains. The structure of APEr.ch.-graft-PAlOx is shown in Figure 1.

Synthesis of the Polyester Multifunctional Macroinitiator
2-(4-(2-bromoethyl)phenylsulfonylhydroquinone was chosen as a polycondensation monomer containing functional groups suitable for initiating cationic polymerization of 2-alkyl-2-oxazolines. The choice of this compound was determined by the fact that previously described [72] soluble alkylene-aromatic polyesters based on phenylsulfonylhydroquinone can be synthesized using the method of non-acceptor polycondensation. It is necessary to keep in mind that the presence of a 2-bromoethyl group sensitive to tertiary amines and alkali in the target monomer makes it impossible to apply traditional acceptor methods for polyester synthesis. On the other hand, it can be assumed that the direct nucleophilic substitution of bromine in the 2-bromoethyl group as a result of the attack of a phenolic hydroxyl in the absence of bases is an unlikely process.
Synthesis of 2-(4-(2-bromoethyl)phenylsulfonylhydroquinone was carried out using the previously published procedure [73], which includes the reaction of 2-phenylethyl bromide with chlorosulfonic acid, reduction of the corresponding sulfonyl chloride to sulfinic acid, and addition of the latter, under the conditions of the Michael reaction, to 1,4-benzoquinone ( Figure 2).  2-(4-(2-bromoethyl)phenylsulfonylhydroquinone was chosen as a polycondensation monomer containing functional groups suitable for initiating cationic polymerization of 2alkyl-2-oxazolines. The choice of this compound was determined by the fact that previously described [72] soluble alkylene-aromatic polyesters based on phenylsulfonylhydroquinone can be synthesized using the method of non-acceptor polycondensation. It is necessary to keep in mind that the presence of a 2-bromoethyl group sensitive to tertiary amines and alkali in the target monomer makes it impossible to apply traditional acceptor methods for polyester synthesis. On the other hand, it can be assumed that the direct nucleophilic substitution of bromine in the 2-bromoethyl group as a result of the attack of a phenolic hydroxyl in the absence of bases is an unlikely process.
Synthesis of 2-(4-(2-bromoethyl)phenylsulfonylhydroquinone was carried out using the previously published procedure [73], which includes the reaction of 2-phenylethyl bromide with chlorosulfonic acid, reduction of the corresponding sulfonyl chloride to sulfinic acid, and addition of the latter, under the conditions of the Michael reaction, to 1,4-benzoquinone ( Figure 2).

Synthesis of the Polyester Multifunctional Macroinitiator
2-(4-(2-bromoethyl)phenylsulfonylhydroquinone was chosen as a polycondensation monomer containing functional groups suitable for initiating cationic polymerization of 2-alkyl-2-oxazolines. The choice of this compound was determined by the fact that previously described [72] soluble alkylene-aromatic polyesters based on phenylsulfonylhydroquinone can be synthesized using the method of non-acceptor polycondensation. It is necessary to keep in mind that the presence of a 2-bromoethyl group sensitive to tertiary amines and alkali in the target monomer makes it impossible to apply traditional acceptor methods for polyester synthesis. On the other hand, it can be assumed that the direct nucleophilic substitution of bromine in the 2-bromoethyl group as a result of the attack of a phenolic hydroxyl in the absence of bases is an unlikely process.
Synthesis of 2-(4-(2-bromoethyl)phenylsulfonylhydroquinone was carried out using the previously published procedure [73], which includes the reaction of 2-phenylethyl bromide with chlorosulfonic acid, reduction of the corresponding sulfonyl chloride to sulfinic acid, and addition of the latter, under the conditions of the Michael reaction, to 1,4-benzoquinone ( Figure 2).  [1,1'-biphenyl]-2,5-dicarbonyl dichloride was used as a comonomer. The polymer synthesis was carried out under the conditions of acceptorless high-temperature polycondensation (Figure 3), which has proven itself well in the synthesis of aromatic polyesters [74]. 1-Chloro-naphthalene was used as a solvent. It was found that the optimal conditions for polycondensation are as follows: temperature of 200 • C, monomer concentration of 25 wt%, and reaction time of 2 h. [1,1'-biphenyl]-2,5-dicarbonyl dichloride was used as a comonomer. The polymer synthesis was carried out under the conditions of acceptorless high-temperature polycondensation (Figure 3), which has proven itself well in the synthesis of aromatic polyesters [74]. 1-Chloro-naphthalene was used as a solvent. It was found that the optimal conditions for polycondensation are as follows: temperature of 200 °C, monomer concentration of 25 wt%, and reaction time of 2 h. It is well known that aromatic polyesters are soluble in a limited number of solvents (chlorinated hydrocarbons, CF3COOH, m-cresol), which significantly limits the choice of a medium for cationic polymerization. It is obvious that protonic acids are unsuitable for this purpose, while chlorinated hydrocarbons are close to theta-solvents for synthesized macroinitiators at room temperatures [75]. It is well known that carrying out polymer analogous transformations in poor solvents cannot provide a sufficient degree of grafting due to the unavailability of a significant part of the functional groups. Usually, the thermodynamic quality of the solvent improves with increasing temperature. In this regard, the grafting of polyoxazoline chains ( Figure 4) was carried out in tetrachloroethane at 150 °C. It is well known that aromatic polyesters are soluble in a limited number of solvents (chlorinated hydrocarbons, CF 3 COOH, m-cresol), which significantly limits the choice of a medium for cationic polymerization. It is obvious that protonic acids are unsuitable for this purpose, while chlorinated hydrocarbons are close to theta-solvents for synthesized macroinitiators at room temperatures [75]. It is well known that carrying out polymer analogous transformations in poor solvents cannot provide a sufficient degree of grafting due to the unavailability of a significant part of the functional groups. Usually, the thermodynamic quality of the solvent improves with increasing temperature. In this regard, the grafting of polyoxazoline chains ( Figure 4) was carried out in tetrachloroethane at 150 • C.  A comparison, for example, of the 1 H NMR spectra of the macroinitiator and the grafted copolymer APE r.ch. -graft-PEtOx shows ( Figure 5) that both signals of aromatic protons of the main chain and signals related to the PEtOx side chains are present in the spectrum of the copolymer. Together with the monomodality of the obtained polymer, these data allow us to assert that the sample under investigation is a graft copolymer. It should be noted, however, the asymmetric shape of the GPC trace of the graft copolymer ( Figure 6), which probably indicates some unevenness in the distribution of the side chains along the main one. In order to determine the molecular weight characteristics of the grafted chains and their grafting density, the main chains of copolymers were destroyed by alkaline hydrolysis under conditions that provide quantitative cleavage of ester groups [76] (Figure 7). A comparison, for example, of the 1 H NMR spectra of the macroinitiator an grafted copolymer APEr.ch.-graft-PEtOx shows ( Figure 5) that both signals of aro protons of the main chain and signals related to the PEtOx side chains are present i spectrum of the copolymer. Together with the monomodality of the obtained poly these data allow us to assert that the sample under investigation is a graft copolym should be noted, however, the asymmetric shape of the GPC trace of the graft copol ( Figure 6), which probably indicates some unevenness in the distribution of the chains along the main one. In order to determine the molecular weight characterist the grafted chains and their grafting density, the main chains of copolymers wer stroyed by alkaline hydrolysis under conditions that provide quantitative cleava ester groups [76] (Figure 7).      The molar masses and hydrodynamic characteristics of the macroinitiator and the graft copolymers are given in Table 1. First of all, we note that for the APEr.ch. in tetra-  The molar masses and hydrodynamic characteristics of the macroinitiator and the graft copolymers are given in Table 1. First of all, we note that for the APE r.ch. in tetrachloroethane, a very low value of the second virial coefficient A 2 was obtained. Consequently, tetrachloroethane is close in thermodynamic quality to the θ-solvent. This is also indicated by the absence of the concentration dependence of the hydrodynamic radius R h ( Figure S1), which is typical for θ-solvents in which thermodynamic interactions are absent. This fact makes it possible to roughly estimate the APE r.ch. equilibrium rigidity by the value of R h , using the relation valid for Gaussian chains: where f and R g are the translational friction coefficient and the radius of gyration of macromolecules, respectively, and the invariant P is P = 5.1 [11]. This relationship is similar to the Flory-Fox equation for intrinsic viscosity. According to the Stokes equation used to calculate the hydrodynamic radius: and therefore: where <h 2 > 1/2 is the root-mean-square distance between the ends of the polymer chain, which, according to the Kuhn equation, is related to the contour length of the macromolecule L and the Kuhn segment length A: The molar mass M 0-APE of the APE r.ch. monomer unit is 563 g·mol −1 . Therefore, the macroinitiator molecule consists of N APE = (n + m) = M w /M 0-APE = 26,500/563 = 47 monomer units, n and m being unsubstituted and substituted ones, respectively. The length of the macroinitiator monomer unit λ 0-APE can be estimated as the sum of the lengths of its bonds along the chain direction. Assuming that the length of all valence bonds of the main chain is close to 0.14 nm, and the valence angles are tetrahedral, in accordance with the APE r.ch. structural formula (Figure 1), we obtain λ 0-APE = 1.51 nm. Accordingly, the APE r.ch. contour length is L APE = 71 nm. Substitution of the R h and L APE values into Equations (3) and (4) results in the Kuhn segment length A = 28 nm.
We emphasize once again that the approach used is a rough estimation of the APE r.ch. equilibrium rigidity. Indeed, the APE r.ch. macromolecule chain consists of only 71/28 = 2.6 Kuhn segments, and its behavior does not obey Gaussian statistics. However, the result obtained is important from the point of view, which allows us to reliably state that APE r.ch. is a typical rigid-chain polymer. Therefore, its conformation should be analyzed within the framework of the Porod model, that is, a worm-like chain with constant curvature. Several approaches are used to describe the hydrodynamic behavior of rigid-chain polymers (see [11]), for example, the model of a worm-shaped spherocylinder proposed by Yamakawa [77][78][79]. In any case, in order to quantitatively determine the equilibrium rigidity of a polymer from hydrodynamic data, it is necessary to study the homologous series.
The long Kuhn segment for the APE r.ch. was expected, since its chain is built of alternating planar ester groups and para-phenylene rings. Such structures are characterized by the so-called "crankshaft" conformation [11], which provides high equilibrium rigidity. For example, the Kuhn segment length of a para-aromatic polyester (pAPE) (Figure 9) is 26 nm according to hydrodynamic data [80] and 20 nm according to flow birefringence data [81]. As seen in Figures 1 and 9, the APE r.ch. and the pAPE differ very slightly (the structure of the main chains is identical); therefore, these polymers should have similar rigidity.
the result obtained is important from the point of view, which allows us to reliably state that APEr.ch. is a typical rigid-chain polymer. Therefore, its conformation should be analyzed within the framework of the Porod model, that is, a worm-like chain with constant curvature. Several approaches are used to describe the hydrodynamic behavior of rigid-chain polymers (see [11]), for example, the model of a worm-shaped spherocylinder proposed by Yamakawa [77][78][79]. In any case, in order to quantitatively determine the equilibrium rigidity of a polymer from hydrodynamic data, it is necessary to study the homologous series.
The long Kuhn segment for the APEr.ch. was expected, since its chain is built of alternating planar ester groups and para-phenylene rings. Such structures are characterized by the so-called "crankshaft" conformation [11], which provides high equilibrium rigidity. For example, the Kuhn segment length of a para-aromatic polyester (pAPE) ( Figure 9) is 26 nm according to hydrodynamic data [80] and 20 nm according to flow birefringence data [81]. As seen in Figures 1 and 9, the APEr.ch. and the pAPE differ very slightly (the structure of the main chains is identical); therefore, these polymers should have similar rigidity. Thus, it can be argued with high probability that the APEr.ch. Kuhn segment length is within the range from 20 to 26 nm. Accordingly, the macroinitiator molecules have a curved thin rod conformation ( Figure 10a). As a measure of the curving of a chain macromolecule, the ratio <h 2 > 1/2 /L can be considered, which varies from 1 for a straight rod to 0 for an infinitely long Gaussian chain. The value of <h 2 > can be estimated using the Porod equation for persistent chains [82,83]: Therefore, the ratio <h 2 > 1/2 /L is 0.49 at A = 20 nm and 0.56 at A = 28 nm, that is, the length of the APEr.ch. macromolecule about twice exceeds the distance between the ends Thus, it can be argued with high probability that the APE r.ch . Kuhn segment length is within the range from 20 to 26 nm. Accordingly, the macroinitiator molecules have a curved thin rod conformation ( Figure 10a). As a measure of the curving of a chain macromolecule, the ratio <h 2 > 1/2 /L can be considered, which varies from 1 for a straight rod to 0 for an infinitely long Gaussian chain. The value of <h 2 > can be estimated using the Porod equation for persistent chains [82,83]: To determine the conformation of the molecules of grafted copolymers, it is necessary to know a number of structural parameters, first of all, the length of the side chains Lsc (Figure 10b) and the grafting density z of the latter. It is easy to show that the value of z can be calculated from the relation: To determine the conformation of the molecules of grafted copolymers, it is necessary to know a number of structural parameters, first of all, the length of the side chains L sc (Figure 10b) and the grafting density z of the latter. It is easy to show that the value of z can be calculated from the relation: where M cop , M APE , M sc , M Br = 79.9 g·mol −1 are the molar masses of the graft copolymer, the macroinitiator, the side chains, and bromine, respectively. According to NMR spectroscopy, the degrees of polymerization of the side chains N sc were 75 and 30 for the APE r.ch. -graft-PEtOx and APE r.ch. -graft-PiPrOx (Table 2). Taking into account that the molar masses M 0-sc of the monomeric units of the side chains of PEtOx and PiPrOx are equal to 99 and 113 g·mol −1 , it is easy to obtain the values M sc = M 0-sc ·N sc for the studied copolymers ( Table 2). Calculations using Equation (6) show that the grafting density of the APE r.ch. -graft-PEtOx and the APE r.ch. -graft-PiPrOx side chains differs by a factor of two ( Table 2). Note that the z value for the APE r.ch. -graft-PEtOx is close to the corresponding characteristic for the previously investigated thermosensitive graft copolymers with PAlOx side chains and flexible-chain polyester backbones [69][70][71]. Thus, both copolymers under study are relatively loose brushes: in the APE r.ch. -graft-PEtOx macromolecules, only about a half of the backbone units contain side chains, while the APE r.ch. -graft-PiPrOx contains a quarter ones.
The difference in z determines the differences in the average distance ∆L = λ 0-APE /z along the chain between the adjacent grafted chains (Figure 10b) and in the number f sc = L APE /∆L = z·N APE of the latter for the studied copolymers (Table 2). It is interesting to compare the ∆L values with the contour length of the side chains L sc = λ 0-sc ·N sc , where λ 0-sc = 0.378 nm [84] is the projection length of the PAlOx monomer unit. As can be seen from Table 2, for the APE r.ch. -graft-PEtOx, the length L sc is almost 10 times greater than ∆L. For the APE r.ch. -graft-PiPrOx, the difference in the compared characteristics is much less, L sc /∆L ≈ 2. Consequently, in a selective solvent, the main chain in the APE r.ch. -graft-PEtOx macromolecules is sufficiently well shielded from the solvent, while in the case of the APE r.ch. -graft-PiPrOx, the APE r.ch. chain is available to solvent molecules.
As is known, PAlOx are flexible-chain polymers with a Kuhn segment length A = 1.7 nm. Accordingly, the side chains of the APE r.ch. -graft-PEtOx contain about 16 Kuhn segments, that is, they are in the Gaussian region in length. The Gaussian coil can be roughly modeled by an ellipsoid of revolution, the major axis of which is H = 1.4<h 2 > 1/2 , and the minor axis is Q = 0.7<h 2 > 1/2 (Figure 10c) [85]. For the APE r.ch. -graft-PEtOx copolymer, we have H = 1.4 × (L sc A) 1/2 = 1.4 × (28 × 1.7) 1/2 = 9.7 nm and Q = H/2 = 4.8 nm, that is, both H and Q is greater than the distance between the adjacent grafted chains ∆L. Consequently, steric repulsion occurs between the adjacent PEtOx chains, and their curvature decreases. It can be assumed that the PEtOx coil is "stretched" along the long axis, and the value of H increases. The interaction of the side chains in molecular brushes leads to the straightening of the main chain; however, if the main chain is a rigid-chain polymer, this effect is insignificant [11]. Then, the APE r.ch. -graft-PEtOx macromolecule can be modeled with a curved thick spherocylinder (Figure 10d). Its radius R sph is knowingly less than the length of the PEtOx side chains and greater than H, that is, 10 nm < R sph < 28 nm. The length of the axis of the curved spherocylinder L sph is the sum of the contour length of the macroinitiator and the contribution of two side chains, that is, L APE = 71 nm < L sph < L APE + 2H = 90 nm.
In the APE r.ch. -graft-PiPrOx copolymer, the side chains contain only six to seven Kuhn segments, and their conformation should be described within the framework of the Porod model. Then, in accordance with the ratio (5), the rms distance <h 2 > 1/2 between the ends of the PiPrOx chain is about 5 nm. Hence, the side chains of the APE r.ch. -graft-PiPrOx are markedly curled up. Since <h 2 > 1/2 and ∆L are comparable, steric interactions between PiPrOx chains are weak, and it can be assumed that the APE r.ch . chain does not change the conformation when passing from the macroinitiator to the graft copolymer. The APE r.ch.graft-PiPrOx macromolecules can also be modeled by a worm-shaped spherocylinder. Its radius is small, less than 5 nm (that is, <h 2 > 1/2 for the side PiPrOx chains), and the axis length is close to 80 nm ≈ L APE + 2<h 2 > 1/2 (Figure 10e). Probably, it is the difference in the thickness of the graft copolymer molecules that determines the difference in the values of the hydrodynamic radius for the APE r.ch. -graft-PEtOx and the APE r.ch. -graft-PiPrOx (Table 1). In the studied range of concentrations at a temperature T < 30 • C, aqueous solutions of the APE r.ch. -graft-PEtOx were molecularly dispersed. As can be seen in Figure 11, there is a tendency towards a decrease in the hydrodynamic radius R h with dilution. However, this change is small, and it fits within the experimental error. Aqueous solutions of the APEr.ch.-graft-PiPrOx at 21 °C were slightly cloudy. Even at the lowest temperatures of the experiment (7 °C), they were opalescent. However, hydrodynamic size distributions obtained for them by DLS were unimodal ( Figure 12). The hydrodynamic size Rh of scattering species was almost an order of magnitude higher than the Rh value obtained in chloroform (Table 1), which indicates the aggregation of the APEr.ch.-graft-PiPrOx macromolecules in water. Aggregation is caused by hydrophobic interactions of the APEr.ch. backbones, which are poorly screened from the solvent, since the grafting density of water-soluble PiPrOx chains is low (z = 0.26), and these chains themselves are short (Lsc/ΔL ≈ 2). As demonstrated by the example of molecular brushes with a flexible polyester backbone with short PiPrOx side chains, the mechanism of hydrophobic interactions is intramolecular: the backbone collapses, forming a core shielded from the solvent by a hydrophilic PiPrOx corona [71]. Unlike a brush with a flexible chain backbone, the rigid backbones of APEr.ch.-graft-PiPrOx cannot fold, and it can be assumed that in order to form a solubilizing shell consisting of hydrophilic short side chains, the PiPrOx chains must be packed tightly to each other, forming bundles or sheaves (Figure Aqueous solutions of the APE r.ch. -graft-PiPrOx at 21 • C were slightly cloudy. Even at the lowest temperatures of the experiment (7 • C), they were opalescent. However, hydrodynamic size distributions obtained for them by DLS were unimodal ( Figure 12). The hydrodynamic size R h of scattering species was almost an order of magnitude higher than the R h value obtained in chloroform (Table 1), which indicates the aggregation of the APE r.ch. -graft-PiPrOx macromolecules in water. Aggregation is caused by hydrophobic interactions of the APE r.ch. backbones, which are poorly screened from the solvent, since the grafting density of water-soluble PiPrOx chains is low (z = 0.26), and these chains themselves are short (L sc /∆L ≈ 2). As demonstrated by the example of molecular brushes with a flexible polyester backbone with short PiPrOx side chains, the mechanism of hydrophobic interactions is intramolecular: the backbone collapses, forming a core shielded from the solvent by a hydrophilic PiPrOx corona [71]. Unlike a brush with a flexible chain backbone, the rigid backbones of APE r.ch. -graft-PiPrOx cannot fold, and it can be assumed that in order to form a solubilizing shell consisting of hydrophilic short side chains, the PiPrOx chains must be packed tightly to each other, forming bundles or sheaves ( Figure 13).

Self-Organization in Aqueous
interactions of the APEr.ch. backbones, which are poorly screened from the solvent, sin the grafting density of water-soluble PiPrOx chains is low (z = 0.26), and these cha themselves are short (Lsc/ΔL ≈ 2). As demonstrated by the example of molecular brush with a flexible polyester backbone with short PiPrOx side chains, the mechanism of h drophobic interactions is intramolecular: the backbone collapses, forming a core shield from the solvent by a hydrophilic PiPrOx corona [71]. Unlike a brush with a flexible cha backbone, the rigid backbones of APEr.ch.-graft-PiPrOx cannot fold, and it can be assum that in order to form a solubilizing shell consisting of hydrophilic short side chains, t PiPrOx chains must be packed tightly to each other, forming bundles or sheaves (Figu 13).   The hydrodynamic radius of the discussed particles Rh increased from 90 to 120 nm ( Figure 11) as the concentration increased, which is in agreement with the AFM data. Augmentation of the hydrodynamic radii of aggregates with the concentration was observed before for molecular brushes with a flexible aromatic polyester backbone and PEtOx [70] and PiPrOx [71] side chains. The increase is caused by an enhancement of the  The hydrodynamic radius of the discussed particles Rh increased from 90 to 120 nm ( Figure 11) as the concentration increased, which is in agreement with the AFM data. Augmentation of the hydrodynamic radii of aggregates with the concentration was observed before for molecular brushes with a flexible aromatic polyester backbone and PEtOx [70] and PiPrOx [71] side chains. The increase is caused by an enhancement of the hydrophobic interactions between the hydrophobic main chains of the grafted copoly- The hydrodynamic radius of the discussed particles R h increased from 90 to 120 nm ( Figure 11) as the concentration increased, which is in agreement with the AFM data. Augmentation of the hydrodynamic radii of aggregates with the concentration was observed before for molecular brushes with a flexible aromatic polyester backbone and PEtOx [70] and PiPrOx [71] side chains. The increase is caused by an enhancement of the hydrophobic interactions between the hydrophobic main chains of the grafted copolymers, especially between unsubstituted monomer units of the main chains, where they are less screened by hydrophilic side chains.
The temperature dependences of the scattered light intensity I and optical transmission I* for the APE r.ch. -graft-PEtOx are shown in Figure 15. Similar plots of I/I 15 (T) and I*/I* 15 (T) and I 15 and I* 15 , being the light scattering intensity and optical transmission at 15 • C, respectively, were obtained for other concentrations. The temperature of the onset of phase separation T 1 was determined as the temperature corresponding to the onset of a decrease in I*. Note that the termination of phase separation in the range of accessible temperatures was not observed at any of the studied concentrations, and for none of the studied APE r.ch.graft-PEtOx solutions was it possible to achieve zero optical transmission. Unlike I*, the value of which was constant up to T 1 , the intensity of the scattered light began to change at a temperature T s , that is, long before the start of phase separation ( Figure 15). The change in I was smooth, its rate increased with temperature, at least up to the temperature T 1 . Similar I(T) dependences were previously observed for thermosensitive polymer brushes, in particular, for graft copolymers with flexible-chain polyester backbones and side chains of poly-2-ethyl-2-oxazoline [69].  The observed change in the scattered light intensity is due to the aggregation of the APEr.ch.-graft-PEtOx macromolecules. As can be seen in Figure 16, at about the temperature Ts, the value of the hydrodynamic radius Rh of the scattering species begins to increase. This reflects aggregation due to dehydration of PEtOx units and, accordingly, a decrease in the solubility of graft copolymers with increasing T. Near T1, the rate of change in Rh increases. The maximum values of the hydrodynamic radii of the aggregates reach 100 nm.

R
h , nm The observed change in the scattered light intensity is due to the aggregation of the APE r.ch. -graft-PEtOx macromolecules. As can be seen in Figure 16, at about the temperature T s , the value of the hydrodynamic radius R h of the scattering species begins to increase. This reflects aggregation due to dehydration of PEtOx units and, accordingly, a decrease in the solubility of graft copolymers with increasing T. Near T 1 , the rate of change in R h increases. The maximum values of the hydrodynamic radii of the aggregates reach 100 nm.
ture Ts, the value of the hydrodynamic radius Rh of the scattering species begin crease. This reflects aggregation due to dehydration of PEtOx units and, accord decrease in the solubility of graft copolymers with increasing T. Near T1, the change in Rh increases. The maximum values of the hydrodynamic radii of the agg reach 100 nm.  The temperature dependences of I, I*, and R h for solutions of the graft copolymer with PiPrOx side chains (Figures 17 and 18) are qualitatively similar to that observed for the APE r.ch. -graft-PEtOx. On the other hand, the changes in I, I*, and R h described above for the APE r.ch. -graft-PEtOx, in the case of APE r.ch. -graft-PiPrOx, occur at lower temperatures. This behavior is due to the higher hydrophobicity of the APE r.ch. -graft-PiPrOx. Indeed, the fraction ω of hydrophobic fragments in the APE r.ch. -graft-PiPrOx macromolecules is about 50 mol%, while for the APE r.ch. -graft-PEtOx, ω = 13 mol% (Table 3). In addition, as mentioned above, the hydrophobic backbone in the APE r.ch. -graft-PiPrOx is much more accessible to the solvent due to the low grafting density z of the side chains and the relatively short length L sc of the latter. Therefore, aqueous solutions of the APE r.ch. -graft-PiPrOx are not molecular, resulting in high R h values at low T. Heating them leads to a further increase in hydrophobicity due to the dehydration of PiPrOx units and, accordingly, to an increase in the size of aggregates (Figure 18), which causes an increase in I ( Figure 17). Note that in the studied concentration range, the change in I and R h begins at a very low temperature T = 7 • C (Figures 17 and 18). In contrast to the APE r.ch. -graft-PEtOx solutions, the dependences of I and R h for the APE r.ch. -graft-PiPrOx are more monotonic; they do not show an increase in the rate of I and R h change at a temperature around T 1 . Figure 19 shows the concentration dependences of the phase separation temperatures T 1 . For both studied copolymers, the T 1 values decrease with increasing c, which is typical for dilute solutions of thermosensitive polymers. For the APE r.ch. -graft-PEtOx, the T 1 (c) dependence flattens out in the region c > 0.0063 g·cm −3 , which makes it possible to reliably determine the LCST. For the APE r.ch. -graft-PiPrOx solutions, the temperature T 1 depends on the concentration over the entire studied range of c. Accordingly, it can be argued that for this graft copolymer, LCST is noticeably lower than the value T 1 = 20 • C for the solution with the highest concentration. crease in I (Figure 17). Note that in the studied concentration range, the change in begins at a very low temperature T = 7 °C (Figures 17 and 18). In contras APEr.ch.-graft-PEtOx solutions, the dependences of I and Rh for the APEr.ch.-graft are more monotonic; they do not show an increase in the rate of I and Rh cha temperature around T1.      The LCST = 45 °C for the APEr.ch.-graft-PEtOx copolymer with PEtOx chains is slightly lower than the LCST for linear PEtOx [66,86,87], which may be due to both the influence of the architecture and hydrophobicity of the APEr.ch.-graft-PEtOx macromolecules, and the molar mass, since its total value for PEtOx chains in the polymer brush is higher than the typical molar mass values for linear PEtOx. A similar situation takes place for the APEr.ch.-graft-PiPrOx, but in this case, the difference in LCST for the graft copolymer and the linear polymer is slightly larger [87][88][89], which can be explained by the large fraction of hydrophobic fragments.
It seems interesting to compare the obtained data with the LCST for graft copolymers with PEtOx and PiPrOx side chains and flexible polyester main chains (Table 3). For the copolymers APE6-graft-PEtOx with a spacer -(CH2)6-in the main chain, LCST is 6 and 11°C lower than the LCST for the APEr.ch.-graft-PEtOx [70]. This difference cannot be explained by the difference in the molar fraction of hydrophobic fragments, since the ω values are higher for the APE6-graft-PEtOx. The relative length Lsc of the side chains, more precisely the ratio Lsc/ΔL of this length Lsc to the distance between two adjacent side chains ΔL, for copolymers with a flexible main chain is 4-6 times less (Table 3). Therefore, in the APEr.ch.-graft-PEtOx, the APEr.ch. chain should be better shielded than the main chain in the APE6-graft-PEtOx. However, in reality, this is not the case, since the flexible backbone of the APE6-graft-PEtOx collapses, sharply decreasing the distance ΔL and increasing the density of the PEtOx corona. In addition, the molar mass can contribute to the decrease in the LCST upon passing from the APE6-graft-PEtOx to the APEr.ch.-graft-PEtOx, which differs for the compared graft copolymers by 2.8 and 3.5 times. The LCST = 45 • C for the APE r.ch. -graft-PEtOx copolymer with PEtOx chains is slightly lower than the LCST for linear PEtOx [66,86,87], which may be due to both the influence of the architecture and hydrophobicity of the APE r.ch. -graft-PEtOx macromolecules, and the molar mass, since its total value for PEtOx chains in the polymer brush is higher than the typical molar mass values for linear PEtOx. A similar situation takes place for the APE r.ch. -graft-PiPrOx, but in this case, the difference in LCST for the graft copolymer and the linear polymer is slightly larger [87][88][89], which can be explained by the large fraction of hydrophobic fragments.
It seems interesting to compare the obtained data with the LCST for graft copolymers with PEtOx and PiPrOx side chains and flexible polyester main chains (Table 3). For the copolymers APE 6 -graft-PEtOx with a spacer -(CH2) 6 -in the main chain, LCST is 6 and 11 • C lower than the LCST for the APE r.ch. -graft-PEtOx [70]. This difference cannot be explained by the difference in the molar fraction of hydrophobic fragments, since the ω values are higher for the APE 6 -graft-PEtOx. The relative length L sc of the side chains, more precisely the ratio L sc /∆L of this length L sc to the distance between two adjacent side chains ∆L, for copolymers with a flexible main chain is 4-6 times less (Table 3). Therefore, in the APE r.ch. -graft-PEtOx, the APE r.ch. chain should be better shielded than the main chain in the APE 6 -graft-PEtOx. However, in reality, this is not the case, since the flexible backbone of the APE 6 -graft-PEtOx collapses, sharply decreasing the distance ∆L and increasing the density of the PEtOx corona. In addition, the molar mass can contribute to the decrease in the LCST upon passing from the APE 6 -graft-PEtOx to the APE r.ch. -graft-PEtOx, which differs for the compared graft copolymers by 2.8 and 3.5 times.
For the copolymers with PiPrOx side chains, the MM and L sc /∆L ratio are practically the same, and the decrease in the LCST for the APE r.ch. -graft-PiPrOx as compared to the APE 8 -graft-PEtOx with a flexible spacer -(CH 2 ) 8 -in the main chain [71] is probably primarily due to the greater hydrophobicity of the APE r.ch. -graft-PiPrOx macromolecules. Poor protection of the APE r.ch. backbones results in their hydrophobic interactions, to which, upon heating, interactions of dehydrated PiPrOx units are added. All the results discussed above refer to the "equilibrium" state of solutions, that is, conditions when their characteristics are constant over time. The times t eq and t* eq for the APE r.ch. -graft-PEtOx and APE r.ch. -graft-PiPrOx solutions to reach the equilibrium state after the temperature change were found by flattening of the I(T) or I*(T) dependences, which are shown in Figure S3.

Kinetics of Aggregation in Aqueous
For all solutions, t eq values depended on T. For the APE r.ch. -graft-PEtOx, they were minimal at low temperatures, increased with heating, taking the maximum value t eq max near the T 1 for a given concentration, and then t eq decreased ( Figure S4). Note that similar dependences were previously observed for thermosensitive star-shaped poly-2-alkyl-2oxazolines and PAlOx graft copolymers [70,71,90]. As seen in Figure 20, for the APE r.ch.graft-PEtOx solutions, t eq max decreases with dilution. The most important thing is that for both polymers, the obtained values are noticeably lower than t eq max determined earlier for solutions of the APE 6 -graft-PEtOx and APE 8 -graft-PiPrOx graft copolymers. In particular, for the APE 6 -graft-PEtOx, t eq max reached 12,000 s [70]. The acceleration of self-organization processes may be due to the fact that the APE r.ch. -graft-PEtOx sample proceeds mainly by the aggregation mechanism, while in the APE 6 -graft-PEtOx solutions, the main chain is also compacted, which leads to an increase in the density of the hydrophilic corona and hinders the contacts of hydrophobic nuclei [69,70]. All the results discussed above refer to the "equilibrium" state of solutions, that is, conditions when their characteristics are constant over time. The times teq and t*eq for the APEr.ch.-graft-PEtOx and APEr.ch.-graft-PiPrOx solutions to reach the equilibrium state after the temperature change were found by flattening of the I(T) or I*(T) dependences, which are shown in Figure S3.
For all solutions, teq values depended on T. For the APEr.ch.-graft-PEtOx, they were minimal at low temperatures, increased with heating, taking the maximum value teq max near the T1 for a given concentration, and then teq decreased ( Figure S4). Note that similar dependences were previously observed for thermosensitive star-shaped poly-2-alkyl-2-oxazolines and PAlOx graft copolymers [70,71,90]. As seen in Figure 20, for the APEr.ch.-graft-PEtOx solutions, teq max decreases with dilution. The most important thing is that for both polymers, the obtained values are noticeably lower than teq max determined earlier for solutions of the APE6-graft-PEtOx and APE8-graft-PiPrOx graft copolymers. In particular, for the APE6-graft-PEtOx, teq max reached 12,000 s [70]. The acceleration of self-organization processes may be due to the fact that the APEr.ch.-graft-PEtOx sample proceeds mainly by the aggregation mechanism, while in the APE6-graft-PEtOx solutions, the main chain is also compacted, which leads to an increase in the density of the hydrophilic corona and hinders the contacts of hydrophobic nuclei [69,70].  [91][92][93][94]. The maximum times teq max do not exceed 2000 s, which is 2.5-6 times less than the teq max for the APE8-graft-PiPrOx with a flexible main chain [71]. The high rate of aggregation in the APEr.ch.-graft-PiPrOx solutions upon heating can be explained by the fact that its macromolecules were already aggregated before heating, and the aggregates form faster by combining ready-made supramolecular structures. For the APE r.ch. -graft-PiPrOx solutions, the "settling" times have the smallest values: at low temperatures, the t eq is in the range from 200 to 400 s. Such values are typical for linear thermoresponsive polymers [91][92][93][94]. The maximum times t eq max do not exceed 2000 s, which is 2.5-6 times less than the t eq max for the APE 8 -graft-PiPrOx with a flexible main chain [71]. The high rate of aggregation in the APE r.ch. -graft-PiPrOx solutions upon heating can be explained by the fact that its macromolecules were already aggregated before heating, and the aggregates form faster by combining ready-made supramolecular structures. (1) [73] and 2-isopropyl-2-oxazoline [95] were synthesized according to the known procedures. 1-Chloronaphthalene and 1,1,2,2tetrachloroethane (Aldrich) as well as oxazolines were dried over calcium hydride and distilled.

Synthesis
NMR spectra were recorded on a Bruker AC 400 spectrometer (400 MHz) for solutions in CDCl 3 . Dialysis was performed with the use of dialysis bags (CellaSep, Orange Scientific Braine-l'Alleud, Belgium) with an MWCO of 3500 Da.

Hydrolysis of Graft-Copolymers
A solution of 0.1 g of a graft-copolymer in 5 mL of 1 M KOH in 2-methoxyethanol was heated under reflux for 10 min, after which it was neutralized and evaporated to dryness in vacuo. The residue was dissolved in 5 mL of ethanol, dialyzed against sodium bicarbonate (concentration 0.1 mol/L) using CellaSep dialysis bags with MWCO 1000 Da, and freeze-dried. The product was dissolved in 15 mL of propionic anhydride or isobutyric anhydride, heated at 50 • C for 30 min, and evaporated under reduced pressure.

Determination of Molar Mass and Hydrodynamic Characteristics of Polymers
Weight-average molar masses M w of the macroinitiator APE r.ch. and the APE r.ch.graft-PAlOx graft copolymers, and the hydrodynamic radii R h of their macromolecules were determined in dilute solutions in organic solvents by static (SLS) and dynamic (DLS) light scattering. Tetrachloroethane (dynamic viscosity η 0 = 1.74 × 10 −3 Pa·s, density ρ 0 = 1.595 g/cm 3 and refractive index n 0 = 1.49) was used as a solvent for the APE r.ch. , and it was chloroform for the molecular brushes (η 0 = 0.542 × 10 −3 Pa·s, ρ 0 = 1.483 g/cm 3 and n 0 = 1.4467), since associative phenomena were absent in them.
The measurements were carried out in a Photocor Complex instrument (Photocor Instrument Inc., Russia), which was equipped with a Photocor-PC2 correlator with 288 channels, as well as a Photocor-PD detector for measuring the intensity of transmitted light. A Photocor-DL semiconductor laser with a wavelength λ 0 = 659.1 nm was used as a light source. Calibration was carried out using toluene, the absolute scattering intensity of which was R v = 1.38×10 −5 cm −1 . Before measurements, the solutions were filtered into dustfree cells using Chromafil polyamide filters (Macherey-Nagel GmbH & Co. KG, Dueren, Germany) with a pore size of 0.45 µm.
For all solutions, there was no scattered light asymmetry, and M w values were obtained by the usual Debye technique by measuring the light scattering intensity at an angle of 90 • . The plots for M w determination for the macroinitiator and the grafted copolymers are shown on Figure S5. The values of M w were calculated using the equation: where I 90 is the intensity of the light scattering at the 90 • angle, A 2 is the second virial coefficient, H is the optical constant, and the value of H is: where N A is the Avogadro's number. The scattered light intensity distributions were unimodal. The hydrodynamic radii of the macromolecules R h (c) collected at concentrations c did not depend on c ( Figure S1), and the R h (c) values were averaged over the concentration to obtain the R h .
The refractive index n was measured on a RA-620 refractometer (KEM, Kyoto, Japan). The refractive index increment dn/dc, which is a factor in (7), was calculated as a slope of the dependence (n − n 0 )/c on c, where n is the refractive index of the solution taken at a concentration c.

Study of the Solutions' Phase Separation upon Heating
The thermal sensitivity of the APE r.ch. -graft-PAlOx copolymers in aqueous solutions was studied by light scattering and turbidimetry using the Photocor Complex device described above. The temperature T was controlled with an accuracy of 0.1 • C, changing it discretely with a step from 5.0 • C at low temperatures to 1.0 • C near the phase separation interval.
The scattered light intensity I and the transmitted light intensity I* were measured as a function of T with increasing temperature. The temperatures T 1 of the beginning of phase separation were determined from the dependences I(T) and I*(T), taking as T 1 the temperatures at which a decrease in I* began. The hydrodynamic radii R h of scattering species were measured as a function of T The measurements of the particles' size were carried out after reaching the equilibrium state of the solutions, that is, after the values of I and I* reached constant values over time.

Microscopic Investigation
The surface morphology of the samples was investigated by the AFM method on the SPM-9700HT scanning probe microscope equipped with the SPM software v.4.76.1 (Shimadzu, Kyoto, Japan) using NSG30-SS Silicon probes with the radius of the tip curvature of 2 nm produced by "TipsNano" (Tallinn, Estonia). To take the sample, the tapping mode measurements were conducted in air using mica as a support.

Conclusions
New diphylic grafted copolymers were synthesized successfully. A polycondensation aromatic polyester served as a macroinitiator to graft poly(2-ethyl-2-oxazoline) and poly(2-isopropyl-2-oxazoline) side chains by cationic polymerization. A fundamentally different chemical class of the backbone and the side chains ensure diphilicity of the resulting polymer brushes. Analysis of their molecular and architectural characteristics made it possible to conclude that they can be considered as loose brushes, with a grafting degree of 0.53 and 0.27 for the APE r.ch. -graft-PEtOx and the APE r.ch. -graft-PiPrOx, respectively. In contrast to the previously studied molecular brushes APE-graft-PAlOx, in this case, the macroinitiator was a rigid-chain polymer. The Kuhn segment length for the APE r.ch. macroinitiator is estimated to be of the order of 23 nm, which results in specific properties of the polymer brushes as in organics as in a selective solvent. In chloroform, the APE r.ch. -graft-PEtOx and APE r.ch. -graft-PiPrOx macromolecules take a wormlike cylindrical conformation, the asymmetry of which depends on the PAlOx side chains' length.
It was shown that architectural parameters are essential for their conformational properties in selective solvents, self-organization, and thermoresponsiveness. Due to high equilibrium rigidity of the main chain, the nature of the self-assembly process of the graft copolymers of the aromatic polyester with PAlOx side chains differs significantly from that of the flexible-chain polymer brushes with similar side chains. Whereas a flexible APE main chain with short side chains is sufficiently labile and capable of changing conformation in various solvents, the rigid hydrophobic APE r.ch. backbone determines either the intra-or intermolecular organization of macromolecules, depending on the length of the side chains and the distance between the grafting points. A rigid-chain brush with a low side chain grafting density takes on a cylindrical wormlike conformation in the case of sufficiently long PAlOx side chains and aggregate in water in big structures if the side chains are short.
The range of thermosensitivity of the rigid-chain APE r.ch. copolymer, which is conditioned by the thermoresponsiveness of the PAlOx side chains, was studied. The LCST of the APE r.ch. -graft-PEtOx under consideration, at given molecular and architectural characteristics, is assumed to be around 45 • C, whereas for the APE r.ch. -graft-PiPrOx, the LCST is much lower than 20 • C. The phase separation temperatures are determined by both the structure and the length of the side chains, and the grafting density. Thus, architecture parameters play a prominent role as in conformational and aggregative properties as in the thermoresponsive behavior of hybrid graft copolymers with a hydrophobic aromatic polyester main chain and hydrophilic thermosensitive polyalkyloxazoline side chains.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.

Conflicts of Interest:
The authors declare no conflict of interest.