Tailoring the amphiphilicity and self-assembly of thermosensitive polymers: end-capped PEG-PNIPAAM block copolymers.

In this work we report on the synthesis and self-assembly of a thermo-sensitive block copolymer system of n-octadecyl-poly(ethylene glycol)-block-poly(N-isopropylacrylamide), abbreviated as C18-PEGn-b-PNIPAAMm. We present a facile synthetic strategy for obtaining highly tunable thermo-responsive block copolymers starting from commercial PEG-based surfactants (Brij®) or a C18 precursor and conjugating with PNIPAAM via an Atom Transfer Radical Polymerization (ATRP) protocol. The self-assembly and detailed nanostructure were thoroughly investigated in aqueous solutions using both small-angle X-ray and neutron scattering (SAXS/SANS) combined with turbidity measurements. The results show that the system forms rather well defined classical micellar structures at room temperature that first undergo a collapse, followed by inter-micellar aggregation upon increasing the temperature. For the pure C18-PNIPAAM system, however, rather ill-defined micelles were formed, demonstrating the important role of PEG in regulating the nanostructure and the stability. It is found that the PEG content can be used as a convenient parameter to regulate the thermoresponse, i.e., the onset of collapse and aggregation. A detailed theoretical modeling analysis of the SAXS/SANS data shows that the system forms typical core-shell micellar structures. Interestingly, no evidence of back folding, where PEG allows PNIPAAM to form part of the C18 core, can be found upon crossing the lower critical solution temperature (LCST). This might be attributed to the entropic penalty of folding a polymer chain and/or enthalpic incompatibility between the blocks. The results show that by appropriately varying the balance between the hydrophobic and hydrophilic content, i.e. the amphiphilicity, tunable thermoresponsive micellar structures can be effectively designed. By means of SAXS/SANS we are able to follow the response on the nanoscale. These results thus give considerable insight into thermo-responsive micellar systems and provide guidelines as to how these systems can be tailor-made and designed. This is expected to be of considerable interest for potential applications such as in nanomedicine where an accurate and tunable thermoresponse is required.


Introduction
Stimuli-responsive polymers are intriguing materials that respond directly to small changes in physical or chemical conditions through changes in their conformation and/or solubility.2][3][4] These materials play an increasingly important part in a wide range of applications, such as in drug delivery, diagnostics, as well as in biosensors, micro-electromechanical systems, coatings etc. [4][5][6] Perhaps the most accessible external stimulus is the temperature, which can be used to trigger changes in solubility of thermoresponsive polymers upon either heating or cooling.][9][10][11][12] PNIPAAM contains a hydrophobic side group that together with the temperaturedependent conformation and hydrogen bonding with water determines the solubility of PNIPAAM in water.][9][10] However, it has been shown that for narrowly distributed polymer chains, the transition is molecular weight and concentration dependent and may vary between 25 and 45 C. [11][12][13] Upon heating to above the transition temperature, a coil-to-globule transition occurs that is followed by inter-molecular association if the solution is not too dilute and macroscopic phase separation, oen referred to as the cloud point.
In order to control the aggregation behavior of PNIPAAM, the polymer needs to be combined with another block that limits and controls the growth of the association complexes.A straightforward way to achieve this is by covalently adding a water-soluble polymer such as poly(ethylene glycol) (PEG) to PNIPAAM.This yields a double hydrophilic PNIPAAM-PEG block copolymer at room temperature, which again self-assembles into micelles consisting of dehydrated PNIPAAM cores and dissolved PEG in the corona at elevated temperatures. 14,15tudies have shown that for larger PNIPAAM blocks even polymeric vesicles can be formed using this strategy. 16However, to allow the nanostructures to form and also to load the system with, e.g., a hydrophobic drug, the solution needs to be kept at high temperatures.Alternatively, the PNIPAAM might be functionalized with hydrophobic residues such as an octadecyl (C 18 )-group, which promotes self-assembly at lower temperatures. 17This strategy also includes telechelic PNIPAAM with two C 18 groups at both ends (C 18 -PNIPAAM-C 18 ).9][20][21] Alternatively, PNIPAAM can be functionalized with hydrophobic blocks at both ends, e.g., polystyrene (PS)based PS-PNIPAAM-PS block copolymers. 22,23However, these systems form micelles that oen have a limited stability range and are prone to phase separation even at moderate temperatures.To obtain suitable nanostructures, the amphiphilicity of the block copolymers needs to be precisely tuned.
One possibility for achieving enhanced control of the selfassembly of PNIPAAM-based systems is to introduce a third polymer block, i.e., triblock terpolymer systems.5][26] These polymers exhibit a stepwise self-assembly mechanism forming "classical" micelles with PEP in the core and hydrophilic PEG/PNIPAAM coronas at low temperatures.Subsequently upon increasing the temperature above the LCST of PNIPAAM, the system undergoes a controlled aggregation into well-dened hydrogels where the strength of the network is given by the inter-chain association between PNIPAAM at the surface of the micelles.This was found to give hydrogels at a much lower concentration than commonly observed for B-A-B-type triblock copolymers. 25Interestingly, it was suggested that PNIPAAM could not fold back into the PEP core due to the limited miscibility and/or entropic penalty of loop formation.The former incompatibility between blocks represents one of the advantages of A-B-C type terpolymers and is found to result in lower sol-gel concentrations. 25n this work, we investigate a system that is designed along similar ideas to triblock terpolymers, but that can be prepared using a more facile synthetic scheme.Instead of using a hydrophobic polymer block, we base our system on PNIPAAM derivatives containing the commercial non-ionic surfactants PEG-octadecylether (BrijÒS10, S20 and S100).By utilizing either C 18 -OH or BrijÒ as the precursor, PNIPAAM could be graed at the end of PEG by using atom transfer radical polymerization (ATRP) of the corresponding NIPAAM monomer.Using this method we have successfully prepared n-octadecyl-poly(ethylene glycol)-block-poly(N-isopropylacrylamide), abbreviated as C 18 -PEG n -b-PNIPAAM m , where n varies from 0 to 100 and m is kept at a near constant value (m z 50), respectively.By employing turbidity measurements, combined with small-angle X-ray and neutron scattering techniques, we characterize the nanostructure and phase behavior in detail.Contrary to other techniques, SANS/SAXS provides high resolution structural data which, combined with data advanced modeling, provide very detailed in situ information on the internal structure and response of the nanostructures.We show that by systematically varying the amphiphilicity of the copolymer system the thermoresponsiveness, as well as the structure and aggregation behavior, can be accurately tuned.As far as we know, this is the rst report on this kind of stimulisensitive nonionic polymer surfactant system.

Synthesis and materials
Fig. 1 shows the synthetic strategy of the C 18 -capped-PNIPAAM derivatives via an ATRP protocol.The chemical structures of PNIPAAM and its block copolymer derivatives are displayed in Fig. 2 and the selected 1 H NMR spectra are given in Fig. 3.As we can see from Fig. 4, these synthesized PNIPAAM derivatives have a fairly narrow molecular weight distribution.Analyzing the peaks we obtain values for the polydispersity index M w /M n between 1.1 and 1.2 (see Table 1).The sharp peaks centered at a retention time of 15-17 min were attributed to the response of the polymers, and the peaks appearing between 18 and 20 min arise from the solvent used in the synthesis, and are commonly observed in GPC. 27,28A very small shoulder of the C 18 -PEG-PNIPAAM polymers appearing between 17 and 18 min may be attributed to the small amount of the C 18 /C 18 -PEGmacroinitiator le in the sample aer purication.

Materials
Octadecanol, poly(ethylene glycol) octadecyl ether (BrijÒS10, M n value of 711, BrijÒS20, M n value of 1150 and BrijÒS100, M n value of 4670) and 2-bromoisobutyl bromide were purchased from Sigma-Aldrich and employed as received.N-Isopropylacrylamide (NIPAAM, Acros) was recrystallized from a toluene/n-hexane mixture and dried under vacuum before use.Triethylamine (TEA) was dried over anhydrous magnesium sulfate, ltered, distilled under N 2 and stored over 4 Å molecular sieves.Copper(I) chloride from Aldrich was washed with glacial acetic acid, followed by washing with methanol and diethyl ether and then dried under vacuum and kept under a N 2 atmosphere.N,N,N 0 ,N 000 ,N 000 ,N 0000 -(hexamethyl triethylene tetramine) (Me 6 TREN) was synthesized according to a procedure described in the literature. 29The homopolymers of poly(Nisopropylacrylamide) (PNIPAAM m , where m ¼ 47) used in this study were synthesized via an ATRP procedure, which has been described previously. 12The water used in this study was puri-ed with a Millipore Mill-Q system and the resistivity was ca.18 MU cm.The solutions were prepared by dissolving the polymer samples in D 2 O, which provides a better contrast and lower incoherent background for SANS.It should be stressed that the same solution was used for all types of measurements (turbidity, SANS, SAXS).
The repeating units of ethylene glycol (EG) in the PEG polymers were recalculated according to the 1 H NMR spectra (Fig. S4 and S6, ESI †) of the fully esteried products, based on a simple formula: n ¼ (3I a /2I b ), where I a is the corresponding integral area of the methenyl group of EG (-O-CH 2 CH 2 -) at 3.7 ppm and    I b is the integral area of the end-capped methyl group (-C(CH 3 ) 2 Br, 6H) at 1.9 ppm.The number of repeating units of EG were estimated to be 10 for BrijÒS10, 20 for BrijÒS20 and 100 for BrijÒS100, and they are designated as C 18 -PEG 10 , C 18 -PEG 20 and C 18 -PEG 100 , respectively. 31,32ynthesis of the C 18 -capped PNIPAAM and C 18 -PEG-b-PNI-PAAM.The C 18 -capped PNIPAAM and the C 18 -capped-PEG n -b-PNIPAAM diblock copolymers (n ¼ 10, 20 and 100) were prepared via a simple atom transfer radical polymerization (ATRP) procedure (Fig. 1).Briey, the polymerization was performed in a solvent mixture of water/DMF (40/60, v/v) at 25 C, and the initiator/catalyst system in the mixture contained the C 18 -initiator (C 18 -MI) or PEG-functional macroinitiator (C 18 -PEG n -MI), CuCl and Me 6 TREN (with a molar feed ratio ).][33][34] The chemical structure and composition of the PNIPAAM derivatives were also ascertained by their 1 H NMR spectra (Fig. 3).The number-average molecular weight and the unit numbers of n and m in C 18 -PEG n -b-P(NIPAAM) m were assessed by comparing the integral area of the methyne proton (8 in Fig. 3 1.

Gel permeation chromatography (GPC) measurement
The molecular weights and polydispersity indices (M w /M n ) of the synthesized PNIPAAM derivatives were determined by using a Perkin-Elmer 200 GPC instrument, operating at 40 C, which comprised of two PL gel 5 mm Mixed D columns (300 Â 7.5 mm) and a differential refractive index detector.Polystyrene standard samples were used for the calibration procedure, and the measurements were carried out by using tetrahydrofuran (THF) as the eluent with an elution rate of 1.0 mL min À1 .

Turbidity
The temperature dependences of the turbidity of the copolymer solutions were monitored at a heating rate of 0.2 C min À1 by employing an NK60-CPA cloud point analyzer from Phase Technology, Richmond, BC, Canada.A detailed description of the apparatus and the determination of turbidities have been given elsewhere. 35This apparatus makes use of a scanning diffusive technique to characterize phase changes of the samples with high sensitivity and accuracy.The light beam from a laser source, operating at 654 nm, was focused on the solution that was placed on a specially designed glass plate that is coated with a thin metallic layer of very high reectivity.Directly above the applied sample, an optical arrangement with a light scattering detector continuously monitors the scattered intensity signal (S) from the measured solutions as it is subjected to prescribed temperature alterations.The turbidity is dened as s ¼ (À1/d)ln(I/I 0 ), where I 0 and I are the transmitted beams of the sample and solvent, respectively, and d is the light path.

Small-angle X-ray scattering experiments (SAXS)
The synchrotron SAXS experiments were performed on the bioSAXS high-throughput P12 EMBL beamline located on the PETRA III storage ring at DESY, Hamburg.The instrument is equipped with a Pilatus 2M detector and the measurements were carried out in a Q-range of 0.0076-0.46ÅÀ1 .The data acquisition was executed by injecting a 10 mL amount of sample into quartz capillaries (2 mm) using 20 successive frames with 50 s exposures that were later added to improve the statistics.No sign of beam radiation damage was observed under these conditions.The data were averaged aer normalization to the intensity of the transmitted beam and calibrated on an absolute scale using Millipore water as a primary calibrating standard.

Small-angle neutron scattering experiments (SANS)
Small-angle neutron scattering (SANS) experiments were carried out with the SANS installation at the JEEP II reactor, Kjeller, Norway.The wavelength used was 5.1 and 10.2 Å, with a resolution (Dl/l) of 10%.The Q range employed in the experiments was 0.008-0.25 ÅÀ1 , where Q ¼ (4p/l) sin(q) and 2q is the scattering angle.The polymer solutions were lled in 2 mm Hellma quartz cuvettes (with stoppers), which were placed on a copper base for good thermal contact and mounted in the sample chamber.Standard reductions of the scattering data, including transmission corrections, were conducted by incorporating data collected from an empty cell, beam without the cell, and blocked-beam background.The data were nally transformed to an absolute scale (coherent differential scattering crosssection (dS/dU)) by calculating the normalized scattered intensity from direct beam measurements.

Theoretical modeling of scattering data
The model tting of the scattering data was made on an absolute scale taking into account the molecular parameters of the system (see Synthesis and materials).The scattering length densities for both X-rays and neutrons were calculated based on the densities reported in the literature for PEG and C 18 . 35Based on these values we obtain for C 18 : r ¼ 7.54 Â 10 10 cm À2 and r ¼ À0.34 Â 10 10 cm À2 for X-rays and neutrons, respectively.For PEG we used r ¼ 11.1 Â 10 10 cm À2 (SAXS) and r ¼ 0.64 Â 10 10 cm À2 (SANS).For PNIPAAM the density was measured to be 1.135 g mL À1 at 20 C and consequently: r ¼ 0.85 Â 10 10 cm À2 and r ¼ 10.6 Â 10 10 cm À2 for SANS and SAXS, respectively.
In the data modeling we assumed that micelles formed by diblock copolymers can be described by a core-shell form factor, while for linear PNIPAAM homopolymers we used a general form factor for excluded volume polymer chains. 36ased on earlier work, [38][39][40][41] the core-shell model can be written in the following form assuming monodisperse star-like spherical entities: where P is the aggregation number (average number of chains per micelle), f is the volume fraction, and V BCP ¼ V cp + V sp is the total molar volume of the block copolymer.V cp is the volume of C 18 and V sp is given by: V sp ¼ V PNIPAAM + V PEG .Dr i ¼ r i À r 0 is the contrast determined by the scattering length density difference between the polymer block (shell-forming polymer (i ¼ sp) or core-forming polymer (i ¼ cp)) and the solvent (i ¼ 0).F(Q) is the form factor of a single polymer chain. 37t should be mentioned that optionally PNIPAAM can be considered to be in the core, which is easily included in the model by letting The scattering amplitude of the shell, A(Q) sh , was calculated using: Here s int is the width of the core-corona interface R c is the radius of the core.n(r) is a density prole for the corona for which we chose a exible power-law prole multiplied with a cut-off function: where R m and s m are the outer cut-off radius and smearing of the density prole, respectively, and x is a scaling exponent that takes a value of x ¼ 4/3 for star-like structures. 45,46For the micellar core the scattering amplitude is: To take into account nite inter-micellar interference effects, a structure factor was included.For simplicity we used the Percus-Yevick structure factor valid for hard spheres with an effective volume fraction h HS and radius R HS . 42In the case where attractive rather than repulsive interactions were observed, either the Baxter model for hard spheres with short-range attractive interactions 43 or the "Teixeira structure factor" 44 describing formation of particles arranged in "fractal clusters" of cut-off length x and fractal dimension d f was used.The Teixeira model can be written as where G(x) is the gamma function.
In addition a constant, B, was added to take into account a Q-independent background in the SANS data.Finally, the theoretical t functions were averaged over the experimental distribution in Q using a resolution function described previously. 47

Structural properties at room temperature
To investigate the nanostructure in solution, SAXS measurements were performed at the P12 bioSAXS beamline at EMBL/ DESY.The scattering curves showing the normalized absolute intensity (the macroscopic scattering cross-section dS/dU), plotted as a function of the module of the scattering vector, Q (Q ¼ 4p sin(q)/l, where 2q is the scattering angle and l is the wavelength), are shown in Fig. 5 for all considered polymers including the linear PNIPAAM.
As expected, the hydrophobic modication (C 18 endcapping) of PNIPAAM leads to micellar-like aggregate structures.This is particularly clear when comparing C 18 -PNIPAAM with linear PNIPAAM in Fig. 5.While PNIPAAM displays a typical scattering pattern of a dissolved polymer chain, the scattered intensity of C 18 -PNIPAAM is signicantly higher with a strong decay at intermediate Q.Similarly, C 18 -PEG 20 -PNIPAAM and C 18 -PEG 100 -PNIPAAM form typical spherical micelle-like structures.Interestingly, for C 18 -PEG 100 -PNIPAAM a slight depletion of the intensity at low-Q is observed, indicating repulsive inter-micellar interactions.This is different from C 18 -PNIPAAM where the intensity continuously increases at low Q and thus shows no evidence of repulsive interactions.Hence, introducing PEG into the shell induces an additional repulsion that stabilizes the micelles.
To gain further insight into the structure, the data were analyzed with the detailed tting models outlined above.For PNIPAAM, the data can be readily described using a simple form factor model for excluded volume chains 37 giving a radius of gyration, R g ¼ 19 AE 2 Å.The slope of the scattered curve in Fig. 5 at high-Q was compatible with a fractal dimension of about 1.7, valid for polymer chains exhibiting excluded volume monomer-monomer interactions.
For C 18 -PNIPAAM the data were tted using a spherical coreshell t model indicating an aggregation number of about 66 and an overall micellar radius of about R m ¼ 75 Å.However, as seen in Fig. 5 the t model does not provide a very good description of the data, in particular not at intermediate Q where the experimental data are more "smeared ", i.e. lacking pronounced oscillations.This can be attributed to a distribution of micellar sizes, i.e. polydispersity.Since the solution was observed to be slightly turbid already at room temperature, this can be a sign of incipient phase separation.Consequently, we did not attempt to rene the scattering model, which would require an assumption of the distribution function that is problematic under these conditions.We will return to the question of phase stability below.
For the diblock copolymers containing PEG, however, the scattering data can be rather well tted with the core-shell model outlined above.A very good description can be obtained assuming a simple classical micellar structure, where C 18 constitutes the core and PEG/PNIPAAM the corona.The ts gave P ¼ 32 and 19, for C 18 -PEG 20 -PNIPAAM and C 18 -PEG 100 -PNI-PAAM, respectively, while the cut-off radius of the corona was found to be 113 Å and 94 Å, respectively.Hence, the structure of the micelles follows a rather classical behavior where the aggregation tendency decreases upon an increase in the corona chain length due to an increased spontaneous curvature.From the theory of Daoud and Cotton 45 for star-like polymers, later adapted by Halperin 46 for star-like block copolymer micelles, one would expect a very weak dependence of P on the corona molecular weight, P $ M B 4/5 ln[R m /R c À 1] À6/5 where M B is the molecular weight of the core-forming block.By inserting the numbers, this would predict a reduction of P with a factor of P(PEG 20 )/P(PEG 100 ) $ 1.33, i.e. from P ¼ 32 to P ¼ 24 for C 18 -PEG 100 -PNIPAAM.This is fairly close to what is observed experimentally, P ¼ 19.For the micellar radius, we would expect the radius to be mainly determined by R m $ P 1/5 M A 3/5 , where M A is the molecular weight of the corona-forming chains.Consistent with the assumption of the ts, we assume that the corona chains can be treated as one entity and we calculate R m (PEG 100 )/ R m (PEG 20 ) ¼ 1.12, that is not far from the value observed experimentally: z1.2.
As previously mentioned, the slight depression of the intensity at low Q for C 18 -PEG 20 -PNIPAAM and C 18 -PEG 100 -PNIPAAM indicates repulsive interactions.From the data ts where a Percus-Yevick structure factor was included, this translates into a hard-core radius of 106 and 100 Å with an effective volume fraction of about 0.05 and 0.025 for the copolymers with PEG 100 and PEG 20 , respectively.The predominantly repulsive inter-micellar interaction potential was also conrmed in a preliminary study of C 18 -PEG 20 -PNIPAAM, where an increased depression of the forward scattering was observed at higher concentrations.However, as the LCST of PNIPAAM is expected to change with concentration, the aggregation behavior of the block copolymer might change in a non-trivial way.A full understanding of this point would require rather extensive systematic studies of the self-assembly, inter-micellar potential and phase behavior at elevated concentrations.These studies will be continued and addressed in a future publication.Nevertheless, the repulsive interactions provide additional evidence that the micelles behave rather classically at room temperature where the hydrophobic C 18 forms the core and PEG/PNIPAAM constitutes the shell, and the entity exhibits signicant excluded volume interactions.To gain further insight into the structure, a selected sample was also directly compared using both SAXS and SANS.

Comparison of SAXS and SANS results: simultaneous model ts
To establish further condence in the structure, a sample containing 1% C 18 -PEG100-PNIPAAM in D 2 O was investigated at room temperature using both SAXS and SANS.The results are given in Fig. 6.
Fig. 6 clearly demonstrates the signicantly different scattering contrast for X-rays and neutrons, where the SAXS data are about one order of magnitude lower in intensity than SANS.Nevertheless, the data can be described simultaneously in a joint t using the same spherical core-shell model without any further parameters.The results of the t analysis are shown in Fig. 6.It is clear that the tted lines describe the data relatively well although some slight deviations are observed.Nevertheless, the results can be seen to be consistent for both techniques, demonstrating the accuracy of the modeling and giving additional condence in the suggested structure.The data could be described with similar t parameters although a slightly smaller R m of 104 Å and larger s m ¼ 0.16 were observed.In the remainder of this work we will focus on the SANS results, which for practical reasons are more suitable for investigation of the structure at elevated temperatures because the samples can be equilibrated for longer times (typically hours) with the set-up we have for SANS.The self-assembly performance and structure at elevated temperatures are addressed in the next section.

Temperature dependence: turbidity
Before focusing on the local structure, the samples were characterized on a more macroscopic scale using turbidimetry, where the transmission of light was monitored as a function of temperature.The turbidity is plotted as a function of temperature for the different block copolymers and PNIPAAM solutions of 1% in Fig. 7.As a comparison also a linear PNIPAAM is included.As expected, 9,11 PNIPAAM exhibits a sharp transition to a turbid solution at a temperature of about 34-35 C. Dening the cloud point, T cp , as the temperature where the turbidity rst starts to increase, we obtain a T cp ¼ 34 C for this concentration and molecular weight.2][13] In this work both variables were xed.
For the end-capped C 18 -PNIPAAM we observe intrinsic higher values of turbidity even at lower temperatures.This is accompanied by a shi in the cloud point towards lower temperatures to ca. 32 C.However, it should be mentioned that upon storing the sample for a longer time at room temperature, increased visual turbidity was noticed.The tendency for aggregation also increased signicantly with increasing polymer concentration.This suggests that the polymer may grow into rather large species following a more "open aggregation behaviour" approaching a macroscopic phase separation.The addition of a C 18 -group yields an increased hydrophobicity to the polymer, sufficient to destabilize the system.Interestingly, the added hydrophobic character does not seem to lead to wellcontrolled micelle formation, which might be due to some disruption of hydrogen bonds with water.
For the C 18 -PEG 20 -PNIPAAM and C 18 -PEG 100 -PNIPAAM polymers, however, a shi is observed towards higher cloud point temperatures located at approximately 38 and 41 C, respectively.This can be attributed to the increased solubility provided by PEG and probably a tendency to form more stable micelles.At elevated temperatures, the system undergoes macroscopic phase separation.Below we will investigate the detailed structure by SANS.
Temperature dependence: mesoscopic structure evolution by SANS In the following we will focus on the temperature dependence of the detailed structure of micelles formed by C 18 -PEG 20 -PNI-PAAM/C 18 -PEG 100 -PNIPAAM.In the case of PNIPAAM and C 18 -PNIPAAM, a trivial macroscopic phase separation was observed at augmented temperatures and this phenomenon was not further investigated.Let us rst consider the copolymer with the lowest PEG content, C 18 -PEG 20 -PNIPAAM, where rather small structural changes with temperature are observed up to about 40 C.However, a close inspection of the data at low Q reveals a signicant upturn of the intensity.Such a behavior suggests the start of cluster formation, i.e., an incipient aggregation.Further increase in temperature to 40 C leads to drastic changes in the shape of the scattering curves with a very low intensity at high Q and a strong Q À4 upturn at low Q.Such an appearance is characteristic of large irregular aggregates that undergo sedimentation and this behavior agrees with the turbidity data that show a strongly reduced transmittance of light at these temperatures.For temperatures below 40 C, the data can be described using a simple core-shell model including a structure factor for irregular fractal clusters.Since only the "wing" of the cluster scattering at low Q can be observed, the t analysis only reveals an apparent aggregate size, x, of about 1000 nm with a fractal dimension close to 3, i.e., reminiscent of a compact cluster.The range of values of x for good ts could be obtained was found to be typically AE30%.The size of the individual micelles is found to slightly decrease with rising temperature from ca. 94 to 77 Å in the temperature interval of 25 to 35 C, and this trend is accompanied by a weaker reduction of the aggregation number from 32 to 26.The important micellar structural parameters are given in Table 2.
The corresponding temperature dependence of the scattered intensity from C 18 -PEG 100 -PNIPAAM is given in Fig. 8(b) and the structural parameters deduced from the t analysis are given in Table 3.In this case a pronounced temperature dependence can be observed.Interestingly, the overall scattered intensity initially increases from 25 to 33 C, indicating an increase in the aggregation number.In addition, we observe a concomitant shi of the scattered intensity towards a higher Q, which suggests a reduction in the micellar dimension.Upon further increase in temperature a decrease in both aggregation number and micellar size can be detected.
To better visualize the structural changes, the radial density proles, n(r), deduced from the ts, are compared and depicted for two temperatures in the insets of Fig. 8.As seen for C 18 -PEG 20 -PNIPAAM in Fig. 8(a), the density distribution for the corona shows a small shi towards lower r-values with increasing temperature, i.e., we observe a compaction towards the core.A similar tendency is observed for the C 18 -PEG 100 -PNIPAAM sample, where an increasing amount of the corona chains is located closer to the core.Thus, since PEG is not expected to change its solubility drastically at this temperature, it is evident that the corona of PNIPAAM undergoes a partial collapse upon a temperature rise.However, the scattering data do not indicate that PNIPAAM folds back and constitutes a part of the core.This scenario was evaluated from the data modeling by assuming a dry PNIPAAM/C 18 core with folded PEG chains in the corona.Such a scenario is not compatible with the experimental data, as it would lead to signicant excess scattered intensity at high Q.This can be attributed to the entropic penalty of back folding and/or incompatibility of the blocks.However, it is interesting to point out that the density proles in the insets of Fig. 8 do indicate a compression of the density prole.But, possibly because of the incompatibility between the blocks and the entropic penalty of back folding, no intermixing in the core is observed even above the LCST.
5][26] The temperature dependence of the micellar structure is analyzed in more detail below.
The extracted aggregation number and the effective micellar radius dened as R tot ¼ R m $(1 + s m ) are depicted as a function of temperature in Fig. 9.
For typical charged surfactant micelles, the aggregation number is expected to decrease upon increasing the temperature 48 , whereas for ethylene oxide (EO) based non-ionic micelles, temperature-induced growth is expected. 49Similarly P is observed to increase 50 in systems of C 18 -PEO (BrijÒ), which corresponds to the precursor polymer.This behavior is ascribed to the inverse solubility appearance for PEG, which is reected in a LCST around 100 C. 51 Table 2 Structural parameters deduced from the model for C 18 -PEG 20 -PNIPAAM in D 2 O at various temperatures.P is the aggregation number, R m the overall micellar radius, s m outer roughness/profile smearing (in fraction of R m ), R c the core radius, R HS the hard core radius, SQ the type of structure factor (see below), x the cluster size, d f the fractal dimension and B the instrumental background a  Comparing the data, C 18 -PEG 20 -PNIPAAM exhibits a monotonous decrease in both the aggregation number and effective micellar radius as the temperature increases.As already mentioned, the system undergoes a complete phase separation at high temperatures.C 18 -PEG 100 -PNIPAAM seems to undergo a slight initial increase in P, followed by a reduction.The radius, however, decreases linearly up to 40-45 C.
The system generally exhibits an opposite trend compared to Brij that must be attributed to the inuence of PNIPAAM.In the case of C 18 -PEG 100 -PNIPAAM, which has the highest content of PEG, the micellization seems to follow a more intermediate appearance.To rationalize this, several (competing) effects must be considered.First, the reduced excluded volume interactions with increasing temperature for both PEG and PNIPAAM must be considered.This leads to shrinkage of the polymers, and in the case of PNIPAAM, even to collapse upon heating.This results in reduced inter-chain repulsion in the corona that gives rise to an increased preferential aggregation number.However, temperature induced collapse leads to, as revealed by the density proles in Fig. 8, an increased accumulation of chains closer to the core.This may result in a higher spontaneous curvature of the micelles, and thus a reduction in the aggregation number.In addition, hydrogen bonds, present below LCST, are likely to change the interactions within the corona and destabilize the micelles.Upon crossing the LCST, inter-micellar aggregation occurs where the micelles mainly maintain their integrity.This is particularly clear for C 18 -PEG 100 -PNIPAAM at 45 C in Fig. 8.Here we observe a well-dened correlation peak suggesting a preferred intermicellar distance, as well as an upturn at low Q which would indicate attractive interactions.Comparing with the ts, the scattering pattern can be described using a structure factor for hard spheres with short-range attractions (Baxter model).This model takes into account attractive interactions (responsible for the upturn at low Q) with a preferable inter-micellar distance that is not observed for C 18 -PEG 20 -PNIPAAM, which rather exhibited a direct formation of unstructured (random) fractal clusters at higher temperatures.It is tempting to interpret the difference as residual repulsive interactions due to the higher fraction of PEG.It should be mentioned however, that the tting approach using the Baxter model yields an unreasonably high effective volume fraction of about 0.3 indicating a drastic local densication of the micelles.This may be an artifact due to a phase separation that can be arrested within the SANS cells upon precipitation.Nevertheless, it is clear that both polymers undergo a transition from repulsive to attractive inter-micellar interactions at higher temperatures.This attractive potential eventually leads to aggregation and phase separation.The strength of the interactions and stability range of the micelles as well as the onset of the transition can be accurately tuned with the PEG content.
Finally, we comment on the temperature dependent interaction between micelles observed through the structure factor model ts.For the C 18 -PEG 20 -PNIPAAM sample we observe temperature-induced attractive interactions that lead to some aggregation in terms of fractal clusters.For the polymer with the longer PEG, C 18 -PEG 100 -PNIPAAM, a more gradual change from repulsive interactions at ambient temperatures to no, or even attractive, interactions at higher T occurs.This is reected in a decreasing effective volume fraction from about 0.05 to 0.03 followed by a vanishing inter-micellar repulsion (h HS ¼ 0).

Conclusions
In this work, we have demonstrated an efficient synthetic strategy to generate a family of PNIPAAM-based thermo-sensitive block copolymers.By systematically changing the amphiphilicity of the system, we have shown that control of the self-assembly and thermo-response can be obtained.The results show that the self-assembly of copolymers consisting of PNIPAAM can be tuned by a balance of a hydrophobic C 18 block and a hydrophilic PEG block.For PEGylated block copolymers, we detect well-dened micellar structures at low temperatures.Upon heating the system close to the LCST of PNIPAAM, we observe a two-step process: rst the micelles collapse into smaller micelles at moderate temperature, followed by intermicellar aggregation and nally macroscopic phase separation.Interestingly, the PEG content can effectively vary the thermoresponsive structure and the phase stability of the system.The micellar structure of the micelles has been analyzed using a detailed theoretical modeling analysis of the SAXS/SANS data.This analysis reveals a rather classical core-shell structure, at least at moderate temperatures.At higher temperatures, a signicant shrinkage of the micelles is observed that can be attributed to the collapse of PNIPAAM chains.Interestingly, the analysis does not provide any evidence of back folding where PNIPAAM forms part of the core upon crossing the LCST.This might be caused by prohibiting entropic penalty associated with the folding of polymer chains and/or enthalpic incompatibility between the blocks.The synthetic strategy and structural insight provided in this study might be of value for design of new thermoresponsive block copolymer systems for potential applications, such as in nanomedicine.In this respect, the system presented in this work constitutes a useful platform for the facile design of novel thermo-responsive nanostructures in the future.

Fig. 4
Fig.4GPC measurements of the synthesized PNIPAAM derivatives (THF as the eluent, flow rate 1.0 mL min À1 , PS as the standard polymer).

Fig. 5
Fig. 5 Small-angle X-ray scattering data showing the scattered intensity as a function of Q for 1% PNIPAAM, C 18 -PNIPAAM, C 18 -PEG 20 -PNIPAAM and C 18 -PEG 100 -PNIPAAM in D 2 O at room temperature.Solid lines display fits for a spherical core-shell model or linear polymer chains.

Fig. 6
Fig. 6 Comparison of SAXS and SANS data for 1% C 18 -PEG 100 -PNIPAAM in D 2 O. Solid lines represent a simultaneous fit at an absolute scale using the same spherical core-shell scattering model.Note that no additional shift factors have been introduced.The only additional parameter is a flat instrumental background present for the SANS data.

Fig. 7
Fig. 7 Turbidity curves.The measured turbidity plotted as a function of temperature for the indicated polymers.

Fig. 8
Fig. 8 Small-angle neutron scattering data for (a) C 18 -PEG 20 -PNIPAAM and (b) C 18 -PEG 100 -PNIPAAM at different temperatures.The polymer concentration was held fixed to 1% in all cases.The solid lines display fits to the core-shell scattering models described in the text.The inset shows the extracted density profile for two selected temperatures.

Fig. 9
Fig. 9 Temperature dependence of (a) the effective micellar radius and (b) the aggregation number as deduced from the fit analysis of the scattering data.

Table 1
Chemical composition, number-average molecular weights, and polydispersity indices of the PNIPAAM-derivatives

Table 3
Structural parameters for C 18 -PEG100-PNIPAAM in D 2 O at various temperatures.P is the aggregation number, R m the overall micellar radius, s m the outer roughness (in fraction of R m ), R c the core radius, R HS the hard core radius, SQ the type of structure factor (see below), x the cluster size, d f the fractal dimension and B the instrumental background.Tau controls the depth of the short range attractive interactions in the Baxter model a SQstructure factor model: SHSsticky hard sphere (Baxter model), PY -Percus-Yevick hard sphere, fractalfractal cluster (Teixeira) model.b Apparent effective volume fraction.