Cononsolvency revisited: Solvent entrapment by N - ACS Publications

Dec 28, 2012 - Flipping the Pressure- and Temperature-Dependent Cloud-Point Behavior in the Cononsolvency System of Poly(N-isopropylacrylamide) in Wat...
0 downloads 11 Views 410KB Size
Article pubs.acs.org/Macromolecules

Cononsolvency Revisited: Solvent Entrapment by N‑Isopropylacrylamide and N,N‑Diethylacrylamide Microgels in Different Water/Methanol Mixtures Christian H. Hofmann,†,* Felix A. Plamper,† Christine Scherzinger,† Sami Hietala,‡ and Walter Richtering† †

Institute of Physical Chemistry, RWTH Aachen University, Landoltweg 2, D 52056 Aachen, Germany Laboratory of Polymer Chemistry, Department of Chemistry, University of Helsinki, PB 55, Helsinki, FIN 00014 Finland



S Supporting Information *

ABSTRACT: Aqueous dispersions of homo- and copolymer microgels of N-isopropylacrylamide (NiPAm) and N,Ndiethylacrylamide (DEAm) with different compositions are temperature-dependently studied by means of proton nuclear magnetic resonance spectroscopy (1H NMR) and differential scanning calorimetry (DSC). Furthermore, the effect of varying the solvent composition by adding methanol is investigated. Methanol addition leads to a broadening of the thermally induced volume phase transition in case of NiPAmcontaining samples, as confirmed by DSC. At the same time, the width of transition approaches the one of neat PDEAm. Two different solvent species, namely bulk-like and restricted solvent, are observed as separate lines in 1H NMR experiments when the gels deswell. The restricted nature of the second species is affirmed by pulsed field gradient (PFG) NMR self-diffusion experiments. The temperature Tsplit from which on the restricted species is found cannot be directly related to the volume phase transition temperature determined by DSC. The difference between Tsplit and the DSC peak temperature changes depending on the NiPAm-content of the microgel. An increase in the shift difference between the two solvent signals with temperature indicates a continuous change of the restricted solvent environment. At even higher temperature, the shift difference of restricted and bulk solvent approaches asymptotically a constant value. In general, the observed effects of methanol addition are consistent with an increasing complexation of the amide protons of the microgel (originating from the NiPAm units) with methanol. In contrast, poly(DEAm) does not show any anomaly concerning transition width and Tsplit upon methanol addition. This is attributed to the lack of amide protons. The results indicate that the presence of cononsolvency can be explained by the presence of the amide proton.

1. INTRODUCTION A number of differently N-substituted polyacrylamides show a change in water solubility upon temperature increase.1 The polymer becomes insoluble and precipitates when the temperature of an aqueous solution is increased to a certain value called lower critical solution temperature (LCST). The polymer becomes soluble again when the mixture is cooled down to temperatures below the LCST. Because of this, such polymers are often called thermoreversible. Suitable means for monitoring this transition are, e.g., turbidity measurements, differential scanning calorimetry (DSC),2−4 and proton nuclear magnetic resonance spectroscopy (1H NMR).5,6 Hydrogels that react to the temperature change by shrinking or swelling, respectively, can be obtained by adding a cross-linking agent during polymerization. Gels of small dimension with a radius in the range from tens of nanometers to about 1 μm often are colloidally stable even in the collapsed state and, compared to macroscopic gels, respond to the temperature change more quickly.7 The transition temperature for these types of gels is © XXXX American Chemical Society

termed volume phase transition temperature (VPTT, instead of LCST), because of a considerable change in size, while no macroscopic phase separation occurs (except the expelling of solvent). This makes such so-called microgels outstandingly well suitable for studying the LCST phenomenon, which might even give insight into the denaturation of proteins,8 but furthermore gives rise to possible applications as, e.g., triggered release. The by far best described polymer from this class is poly(Nisopropylacrylamide) (PNiPAm) with an LCST of 31−33 °C.9 Additionally to the LCST transition, PNiPAm shows a solvent dependent transition. Addition of a second solvent such as simple alcohols, acetone or tetrahydrofuran to an aqueous PNiPAm solution leads to precipitation or collapse already below the transition temperature observed in pure water, this Received: November 19, 2012 Revised: December 18, 2012

A

dx.doi.org/10.1021/ma302384v | Macromolecules XXXX, XXX, XXX−XXX

Macromolecules

Article

behavior is termed cononsolvency.10 Further addition of the cosolvent will finally lead to redissolving or reswelling since the mentioned solvents all are good solvents for PNiPAm. In the case of methanol, Walter et al. ascribe this phenomenon to hydrogen bonding of methanol to the polymer making the polymer less soluble in a water-dominated environment due to the exposed methyl groups of the hydrogen-bonded methanol (as seen by MD simulations).11 Tanaka et al. used the theoretic concept of cooperative hydration and competitive hydrogen bonding in order to describe the thermosensitive behavior of PNIPAM in water and methanol mixtures.12−15 An enrichment of methanol in vicinity of polymer chains with increasing temperature was found by theoretical analysis.16 Arndt et al. were able to describe the thermoreversible behavior by thermodynamic calculations where they introduced a stretchability contribution to apply the calculation to gels.17 A different approach of explaining the cononsolvency behavior is based on solvent molecule complexes that are formed in water/ methanol mixtures.18 The hydration of the polymers in presence of such aggregates then may be unfavored.19 Zhang and Wu concluded from experiments on extremely diluted chains that solvent complexes constitute a poor solvent for the polymer.20 Poly(N,N-diethylacrylamide) (PDEAm) has in water about the same LCST as PNiPAm but does not show cononsolvency upon addition of cosolvents.21,22 A satisfying explanation for the cononsolvency and its absence in the case of PDEAm is still lacking, especially when solvent clusters are discussed. Motivated by the remarkable difference between PDEAm and PNiPAM concerning cononsolvency, random copolymers of the two monomers DEAm and NiPAm have been investigated systematically.21,23−26 In water, an LCST of the copolymers was found which was always below the LCST of the respective homopolymers with the minimum value for the equimolar copolymer. As a possible explanation, binary interaction of one unit without and one with an amide proton as an H-bond donor is mentioned.23 For other copolymers of PNiPAm and other acrylamides, a partial loss of co-operativity is known. By combining NMR and DSC, Št’astná et al. found recently a decrease of co-operativity upon increasing the content of acrylamide (AA) in NiPAm/AA copolymers.27 Many studies have focused on the solvent to gain further insight by what processes the collapse is induced. It has been found that water molecules are present even inside globules formed of precipitated linear polymer above the transition. By a matter of course, this is even more pronounced in collapsed microgels. Ohta et al. studied aqueous PNiPAm microgels using water proton signals as probe in NMR relaxation experiments.28 They found an anomalous temperature course of the transversal relaxation time T2 when PNiPAm gels were dispersed. Below the transition, T2 showed typical behavior by slightly increasing with increasing temperatures. The phase transition, however, led to a pronounced decrease in T2. Above the transition T2 followed the typical temperature dependence again but at lower absolute values compared to temperatures below the transition. From that, they concluded a significantly reduced mobility of the water inside the gel above the transition. The same phenomenon can be ascertained by diffusion (pulsed field gradient, PFG) NMR.3 While the T2 times determined by Ohta et al. were averaged for bulk and possibly bound water, Dı ́ezPeña et al. as well as Wang et al. detected separated signals for a bulk and a “bound” solvent species above the VPTT by magic angle spinning (MAS) NMR in PNiPAm hydrogels.29,30 The

same phenomenon has also been reported for poly(vinyl methyl ether) (PVME) when kept for several hours at temperatures above the VPTT.31 In the case of PVME a second water signal has been even reported for extremely high concentrated (20 wt %) linear chains. There, a difference in the chemical shift of nearly 1 ppm has been observed between the two water signals.32 As a tool for directly detecting interactions between solvent and polymer, infrared spectroscopy (IR) has been established.33−35 It turned out that the phase transition is accompanied by a distinct change of the amide group environment from hydrophilic to rather hydrophobic, though polymer-bound, hydrogen-bonded water is still present far above the transition.36 Analogously, macroscopic gels retain approximately ∼5% solvent in the shrunken state, as indicated by experiments of Biswas et al.37,38 Here, we systematically investigate the influence of polymer/ solvent interactions and, particularly, the different effects on the VPTT by combining different methods. We choose microgels to avoid flocculation and enable detection of solvent inside the polymer network above the transition. PDEAm and PNiPAm homopolymers as well as differently composed copolymers thereof are taken to assess the role of the amide substitution pattern, especially the presence or absence of an amide proton. To include cononsolvency effects, we further vary the solvent composition from pure water up to a mixture containing 20 mol % of methanol. As methods we apply DSC, single pulse 1H NMR and 1H PFG NMR. First we present the direct results from the methods. In a subsequent section, we elaborate the greater context by comparing the different single results. From that we will conclude that three different solvent species, bulk, restricted and complexed solvent, can be distinguished. Finally, we round up the picture of what is happening with the solvent during the transition and in how far the solvent influences this macroscopic process.

2. EXPERIMENTAL SECTION Materials. Deuterated solvents have been used in all cases to enable clear separation of the different signals in NMR and comparability of the DSC results to NMR results. Deuterium oxide (D2O, 99.9% d) and perdeuterated methanol (MeOD-d4, 99% d) were purchased from Euriso-Top. N-Isopropylacrylamide (NIPAM), N,N′methylenebis(acrylamide) (BIS), sodium dodecyl sulfate (SDS), and potassium peroxodisulfate (KPS) were bought from Sigma-Aldrich and N,N-diethylacrylamide (DEAAM) from Polysciences Germany. From DEAAm the stabilizing agent has been removed by passing through basic aluminum oxide. All other chemicals were used without further purification. Twice distilled Milli-q-water was used during the synthesis and the cleaning process. Microgel Preparation. The microgels were synthesized by precipitation polymerization of the respective monomer with N,N′methylenebis(acrylamide) (BIS) as cross-linker and sodium dodecyl sulfate (SDS) as emulsifier (see Supporting Information on further experimental details and molecular characterization by light scattering analysis). A three-necked round-bottom flask was equipped with reflux condenser, KPG stirrer and a septum for gas inlet. 125 mL of twice distilled water was degassed with N2 for 30 min and then heated to 80 °C. The monomers (25 mmol in total), 5 mol % BIS, and 1.2 mol % SDS were dissolved under stirring at 350 rpm. The reaction was started with KPS (1 mol % in H2O) and then proceeded for 5 h. After that, the reaction mixture was cooled to room temperature under stirring and gas inlet, filtered over glass wool, and cleaned via three cycles of centrifugation and redispersion in twice distilled water. The product was freeze-dried. Five different microgels of similar size have been prepared this way: Two homopolymers, poly(N,N-diethylacrylamide) (PDE) and poly(N-isopropylacrylamide) (PNi), one copolyB

dx.doi.org/10.1021/ma302384v | Macromolecules XXXX, XXX, XXX−XXX

Macromolecules

Article

mer with approximately equimolar ratio (P11) and two copolymers with approximately three-to-one molar comonomer ratio (one DEAmrich, P31, and one NiPAm-rich, P13; for details, see Table 1). From all

composition in Figure 1. In contrast to turbidity measurements where often saturation effects allow only the detection of the beginning of the transition, DSC gives additional information. Two main effects can be extracted from the peak temperatures. One of them is best visible in pure D2O. Under that condition both homopolymer microgels (PNi, PDE, for composition see Table 1) show the highest values whereas P11 and the P31 copolymer show the lowest values. This is in agreement with what has been reported on non-cross-linked copolymers of DEAm and NiPAm before.23 The only deviation from what has been reported for non-cross-linked copolymers is that the nonequimolar copolymers (P31, P13) do not have the same peak temperature. One further and important difference between the samples can be seen when additionally the onset temperature is taken into account. The difference between onset and peak temperature differs significantly for different samples and different solvent compositions. That means that the width of the transition depends on the composition of both the microgel itself and the solvent. Figure 2 shows the peak width for the

Table 1. Relative Monomer Molar Fractions in the Investigated Microgels sample

NiPAm

DEAm

BIS

PDE P31 P11 P13 PNi

0.00 0.27 0.48 0.71 0.95

0.95 0.68 0.47 0.24 0.00

0.05 0.05 0.05 0.05 0.05

the microgels, stock solutions containing 5 wt % lyophilizate were prepared in both D2O and perdeuterated methanol. The final samples then were prepared by mixing the two stock solutions in the desired ratio. Samples containing 0, 5, 10, 15, and 20 mol % methanol were prepared. In the following, the solvent is named in a manner that the molar water/methanol content is given like, e.g., 85/15 for a dispersion containing 15 mol % methanol. For all experiments including DSC only deuterated solvents are used. 1 H NMR. NMR experiments were performed on a Bruker Avance III spectrometer at a proton frequency of 501 MHz using a 4 mm HRMAS probe. The samples were spinned at a frequency of 5 kHz. The sample temperature was kept constant using a nitrogen flow from a temperature control unit (Bruker). The actual temperature at the sample position was calculated from the shift difference of the two signals in neat methanol measured under identical conditions.39 Spectra were recorded with at least eight scans. The interscan delay was set to 5 s since a complete re-establishment of net magnetization after that time could be confirmed by inversion recovery measurements. Pulsed field gradient experiments were carried out using the longitudinal eddy current delay bipolar pulsed field gradient pulse sequence with 32 gradient steps up to 36 G/cm, 2 ms gradient duration and a diffusion time of 20 ms. The data were processed using the Bruker Topspin software. The split solvent peaks had additionally being modeled using the DMFit 2012 software40 applying one Gaussian and one hybrid Gaussian/Lorentzian component. DSC. Thermograms of the same samples as used for NMR experiments were recorded using a Mettler 822e differential scanning calorimeter at a scanning rate of 10 K/min in a range from −10 to +60 °C. By running three consecutive heating and cooling cycles, reproducibility was ascertained, and only the second heating cycle was representatively taken for further evaluation.

Figure 2. Peak width from DSC versus solvent composition (squares, PDE; pentagons, P31; diamonds, P11; triangles, P13; circles, PNi).

different microgels in dependence of the solvent composition. From the graph, the five samples can be divided into two groups. The one group, PDE and DEAm-rich P31, have a quite large and solvent-composition-independent transition width. The other group, P11, P13 and PNi, show a clear solventcomposition dependence of the peak width. In the case of pure D2O as solvent, PNi has the smallest transition width. The width increases then with increasing DEAm content, i.e., in the order P13, P11, P31, PDE. This difference between PDEAm and PNiPAm is well-known.41 Furthermore, the differences in transition width might now explain why the two nonequimolar copolymers did not show similar peak temperatures as could be expected since Maeda et al. reported similar cloud points for

3. RESULTS Differential Scanning Calorimetry. For a better understanding of the phase transition and cononsolvency effects, knowledge of the transition temperatures obtained by different methods is essential. The peak and onset temperatures as determined by DSC are plotted for all microgels versus solvent

Figure 1. Peak (filled symbols) and onset (open symbols) temperatures from DSC measurements at 10 K/min (squares, PDE; pentagons, P31; diamonds, P11; triangles, P13; circles, PNi). C

dx.doi.org/10.1021/ma302384v | Macromolecules XXXX, XXX, XXX−XXX

Macromolecules

Article

the corresponding non-cross-linked copolymers.23 Generally, there are differences between linear chains and cross-linked microgels when it comes to the polymers containing DEAm units. One explanation may be that the increase in transition width is more pronounced for cross-linked polymers than for non-cross-linked ones.41 This can be further confirmed by both peak and onset temperature of the PDE microgel being clearly below those of the PNi microgel, although no difference is reported for the respective linear polymers.23 The second effect or, more precisely, a superposition of some effects that can be found in Figure 1 and Figure 2 is the influence of the methanol content. In case of PDE and the DEAm-rich copolymer (P31) no significant dependence of the transition width on the methanol content is detected. Both peak and onset temperature do not change with the methanol concentrations investigated. This means that not only the transition temperature is constant but also the transition width. The other three samples show a completely different behavior. On the one hand, peak and onset temperature decrease with increasing methanol content which reveals the cononsolvency behavior of PNiPAm23 and leads to a crossover of the peak temperatures of PNi and of PDE and P31. On the other hand for P11, P13, and PNi, the transition width increases. This increase leads to similar values for the transition width for an 85/15 solvent composition (at 15 mol % methanol content) for all different polymers while the values in a methanol-free dispersion clearly deviate from each other. The integrated endothermic heat decreases with increasing methanol content for all samples (see Supporting Information). At 20 mol % methanol content no peak can be observed. This can be explained by an extrapolation of the transition heat that is more or less linearly decreasing with increasing methanol content. The extrapolated transition heat for 20 mol % methanol content is almost zero (see Supporting Information).42 For the transition temperature it holds that ΔG = 0 and the transition temperature is almost constant at least for PDE. As a consequence, in that case, the entropy ΔS must also be reduced. The same holds for the other samples in good approximation since the differences in the transition temperature are small on an absolute scale. The reduction in ΔS is more pronounced for PDEAm than for PNiPAm. Nuclear Magnetic Resonance. Figure 3 shows the 1H NMR spectra recorded for P11. All polymer signals can be identified at T < VPTT. The signals of the NiPAm repeating units are present at 3.9 ppm and from 2 to 0 ppm whereas the DEAm signals are from 2 to 0 ppm and around 3.5 ppm. The latter signals are overlapped by the methanol signal found at 3.3

ppm. Toward higher temperatures, the polymer signals more and more lose intensity until they are hardly visible at 37.0 °C. It has to be mentioned that for some of the other samples the decrease is not as clearly pronounced but still significant and with a similar integral-temperature course (see below, the dependence of the degree of signal loss on the methanol content is shown in the Supporting Information). Not only are the polymer signals affected by temperature but also there is an effect on the solvent signal as well. There, a change in shape can be observed at a certain temperature instead of a successive attenuation of the signals. At 18.1 °C as well as at 37.0 °C (and at the temperatures between which are not shown here for reasons of clarity), a second signal of much smaller intensity which does not occur at the two lower temperatures is present right to the main signal. This additional signal resembles a signal already observed by Dı ́ez-Peña et al. as well as Wang et al. for PNiPAm.29,30 This signal was assigned as so-called ‘bound water’ or more generally to so-called “bound solvent” (peak between 5.2 and 4.8 ppm is a common peak of the protonated hydroxyl groups of DOH and CD3OH traces). For a closer look to this phenomenon, the two solvent signals are shown in Figure 4. Spectra for PNi are taken where no

Figure 4. Solvent signals: mixed solvent peak and methanol methyl peak (inset) at different temperatures for aqueous PNi dispersion containing 15 mol % methanol.

overlapping of the methanol signal occurs in order to avoid complications in interpretation. One important difference between our measurements and those of Wang et al. is the use of perdeuterated methanol because of the larger methanol content used here. Thereby, the methanol methyl signal is split to a quintet caused by coupling between one methyl proton and two neighbored methyl deuterons. Unfortunately, this makes it rather difficult to identify a possible bound methanol signal as reported by Wang et al.29 In the case of ethanol where even the protonated species show multiplet splitting, they were only able to distinguish the “bound” solvent signal by modeling. In our case, additional superposition of polymer signals makes it even more complicated to distinguish a second methanol signal. The only exception is the PNi microgel with no DEAm units. There, a second but broad signal at chemical shifts slightly greater than those of the major methanol signal at higher temperatures can be identified (see Figure 4). Again, this finding is in agreement with Wang et al. For all other microgels than PNi, no conclusion on the existence of ‘bound’ methanol can be made from the spectra due to the above-mentioned complications. Details on the width and the ratio of the two mixed solvent peaks (at about 5 ppm) obtained by line fitting are shown in the Supporting Information. There is one further remarkable observation which, to our best knowledge, has not been reported yet: The shift difference

Figure 3. Proton spectra as measured in aqueous P11 solution containing 10 mol % methanol at different temperatures. D

dx.doi.org/10.1021/ma302384v | Macromolecules XXXX, XXX, XXX−XXX

Macromolecules

Article

between the additional and the main solvent signal increases with increasing temperature. This is clearly visible in Figure 4 and not only found for the PNi microgel in an 85/15 solvent mixture shown there but for all samples at all solvent compositions studied by NMR. The shift difference Δδ between the main and additional solvent is plotted versus temperature for one sample, namely the DEAm-rich copolymer P31 in an 85/15 solvent mixture, in Figure 5. At low

Figure 6. Integral from δ = 1.5 ppm to δ = 2.5 ppm calculated from spectra of PNi in a 90/10 solvent mixture. The line is a fit according to eq 3.

where I is the intensity integral, Imax is the plateau value at low temperatures, Imin is the plateau value at high temperatures and k is a parameter describing the transition width. The remaining parameter Ttrans is the temperature where the half loss of intensity has taken place. A table with the Ttrans values of the different polymers and solvent compositions is provided in the Supporting Information. This parameter will be discussed further below. Determination of the diffusion coefficient by means of pulsed field gradient experiments is very useful to further verify the nature of the additional mixed solvent andwhere present methanol signals. We find monoexponential decays for the individual solvent signals in nearly all cases. Exceptions are the methanol signals at lower temperatures in DEAm-containing samples due to overlap of signals. That only single exponential decays are found both below and above the phase transition is contrary to results on macroscopic gels where two diffusion coefficients were fitted to one signal at higher temperatures.45 The important difference to our study is the absence of a separated bound solvent signal. So, the one signal taken for evaluation by Ray et al. might have been caused by two species that in our case give different signals with different but single-exponential decays. Two or, depending on the temperature, four distinguishable solvent signals can be fitted in case of PNi, whereas in all other samples only two or three different signals can be fitted. Results for PNi in a solvent mixture containing 15 mol % methanol are shown in Figure 7. At first sight three important things can be recognized. First, the main water and methanol signals have similar values that are in the range of bulk water and methanol.

Figure 5. Chemical shift difference between main and additional mixed solvent signal measured at 85/15 solvent composition for P31. The solid line results from an exponential fit according to eq 1.

temperatures a steep increase of Δδ is seen which is more and more diminished implying an asymptotic course Δδ(T). Therefore, the data points are fitted by an exponential decay according to Δδ(T ) = Δδ∞ − A e−BT

(1)

where Δδ∞ is the value of the asymptote and A and B are constants whereof the latter describes the curvature of the fit. Additionally a parameter Tsplit can be introduced by calculating the T-axis intercept: Tsplit = −ln(Δδ∞/A)/B

(2)

Tsplit is that temperature from which on two different solvent signals are present in case of an infinitesimally narrow line width. In this manner, a series of Tsplit in dependence of the methanol content can be determined. In case of PDE some deviations appear insofar that the dependence of Δδ on temperature seems to be nearly linear leading to a curvature B close to zero. As can be seen from Figure 3, the collapse of the microgel is accompanied by a loss in signal intensity of the polymer resonances. Thus, the transition can be monitored complementary to the DSC measurements by plotting the polymer signal integral versus temperature. Such a plot is displayed in Figure 6. In contrast to measurements on linear PNiPAm chains in pure water where a nearly complete loss of intensity is observed,43 here about 40% of the intensity remain even at elevated temperature. The percentage of remaining signal at elevated temperature increases monotonically with increasing methanol content (details see Supporting Information). Still, a significant loss occurs in the characteristic temperature range. The course of the data can be fitted by a phenomenological function defined as44 I = Imin + (Imax − Imin)/(1 + exp((T − Ttrans)/k))

Figure 7. Diffusion coefficients from monoexponential fitting of PFG echo decays of PNi in a 85/15 solvent mixture; filled symbols represent the main solvent signal and open symbols the additional solvent signal (mixed solvent signal, squares; methanol methyl signal, triangles).

(3) E

dx.doi.org/10.1021/ma302384v | Macromolecules XXXX, XXX, XXX−XXX

Macromolecules

Article

Figure 8. Comparison of DSC peak temperatures (filled symbols) and Ttrans as derived by NMR (open symbols). Key: squares, PDE; pentagons, P31; diamonds, P11; triangles, P13; circles: PNi.

It might be the case that even at low temperatures solvent molecules inside the gel are moving slower than outside due to restriction. A nonetheless nonsplit solvent signal can be caused by a fast exchange leading to an averaged chemical shift and one-component diffusion coefficient. In such a case, the measured diffusion coefficient should differ from that of bulk solvent. However, at 23 °C all free solvent diffusion coefficients here are comparable to the diffusion coefficients reported for the neat solvent mixtures.46 The only exceptions are found in samples with a Tsplit close to this temperature, where, caused by an incomplete separation of the two solvent peaks, a nonmonoexponential decay is found that leads to large insecurities. This can be seen for instance in the case of PNi in a 15/85 solvent mixture, when comparing the diffusion coefficients of different samples at one temperature (see Supporting Information). By considering all this, we can state from the PFG results that the bulk water outside is not affected by the presence of the microgel in spite of the high microgel concentration of 5 wt %. Altogether, the diffusion results confirm the picture of two solvent molecule species, one free bulk species and a second species with restricted motion due to restricted surroundings caused by the collapse of the microgel. The following discussion will be based on this without further focusing on the diffusion data.

Also the temperature dependence is that of bulk solvents. The slightly slower diffusion of methanol can easily be attributed to the different molecular weight. Second, the additional solvent signals, i.e. the broader signal of small intensity next to the main solvent signals that do not occur at temperatures far below the VPTT, are connected to smaller diffusion coefficients compared to the bulk solvent signals (D(water, 85/15, 20 °C) = 1.43 × 10−9 m2/s; D(methanol, 85/15, 20 °C) = 1.06 × 10−9 m2/s).46 And third, there seems to be no temperature dependence of the additional solvent diffusion coefficient. Nevertheless, the diffusion coefficients found for the species represented by the additional solvent signal are about 2 orders of magnitude larger than the diffusion coefficient of the microgel as determined by DLS (cf. Supporting Information). For PNiPAm hydrogels in water, three different diffusion coefficients have been reported before but were not assigned to different signals.47 One of the three diffusion coefficients reported there can be explained by the slow motion of the polymer, a second by free solvent. Then the remaining diffusion coefficient between the two mentioned diffusion coefficients can be assigned to the species represented by the additional solvent signal. All these results clearly identify the additional signals from our NMR results as different from bulk water but also reveal that the solvent molecules belonging to this signal are not attached firmly to the polymer chains. Therefore, the term “bound solvent” as introduced for the additional solvent signal at elevated temperatures by Wang et al. is somewhat misleading. In fact the solvent molecules they assigned bound solvent must be still rather mobile but in a restricted environment. Additionally, the peak of “bound” solvent is still rather pronounced. This implies a rather large amount of solvent, which is hindered in motion by the microgel, not allowing possible hydrogen bonding of solvent and monomer unit for every entrapped solvent molecule. Thus, we will call this species restricted solvent henceforth. This does not exclude that there are solvent molecules fixed to the polymer that most reasonably should be named complexed solvent. Those would have diffusion coefficients that cannot be measured by the used experimental setup. Furthermore, an internal exchange of the complexed, hydrogen-bonded solvent and the domain-like, entrapped solvent cannot be excluded. One might speculate that the additional solvent signal could arise from possible solvent clusters with reduced dynamics as described in literature.18 However, there is no second signal at low temperatures where the clusters should be present as well. Furthermore, the signals are present even in pure D2O. Because of this, we assume that the additional solvent signal is due to a restricted species inside the polymer-rich phase and not caused by water/methanol clusters.

4. DISCUSSION In the following, we want to gain further insight in what is happening during the microgel collapse by comparing the results from the different methods. Furthermore, we are interested in possible differences between different samples and solvent mixtures. Thus, we correlate the DSC transition temperature (as obtained from the endothermic peak) with the NMR transition temperature TTrans, which is obtained by the integral change over the polymer signals. As a consequence, we can illustrate and discuss the differences between the DSC data and Tsplit, which is the extrapolated temperature from NMR experiments, where a split restricted signal in NMR appears for the first time upon heating. The comparison of the Ttrans values and peak temperatures from DSC is given in Figure 8. On the first sight, the very good agreement of both parameters is obvious like it has already has been reported for non-cross-linked PNiPAm.4 The decrease of the polymer signal intensity in NMR and the broadening of the lines is caused by a tremendous decrease of the transverse relaxation time T2 or a comparable increase in chemical shift anisotropy. A short T2 leads to a significant cut off of intensity since a larger fraction of the signal then will have decayed before the acquisition starts. Since T2 decreases monotonically with decreasing dynamics of the responding nuclei, a very direct F

dx.doi.org/10.1021/ma302384v | Macromolecules XXXX, XXX, XXX−XXX

Macromolecules

Article

Figure 9. Comparison of Tsplit (half filled symbols) with DSC data (taken from Figure 1; onset, open symbols; peak, filled symbols). Key: squares, PDE; pentagons, P31; diamonds, P11; circles, PNi.

As described above (cf. Figure 5), we find a temperature dependence of the chemical shift difference between bulk and restricted solvent signal. In general, even the chemical shift of free (bulk) solvent signal decreases with increasing temperature, like it is for all acidic protons in neat compounds. This means that this effect is more pronounced for the restricted solvent since its signal is found at smaller chemical shifts than the free water. The phenomenon of the temperature dependence of neat water NMR signal has been attributed to the breaking of hydrogen bonds induced by thermal energy.48 By considering this, it can be assumed that the second solvent signal arises from solvent molecules that are in less hydrophilic and thus less water-like surroundings than the common bulk water. The fact that the difference between the chemical shift of bulk and restricted solvent increases with increasing temperature indicates a gradual change in the structure by which the restricted solvent is surrounded. From this, it follows that solvent is entrapped above Tsplit and the nearest environment is successively becoming more and more different from bulk solvent. This can imply a direct relation to a continuous shrinking of the network, which leads to a second, restricted solvent species. One hint for this assumption being correct is the apparently asymptotic behavior of Tsplit toward high temperatures (see Figure 5). We find asymptote values of Δδ∞ = 0.10 ppm to Δδ∞ = 0.25. This is not the case for PDE where an almost linear dependence was detected but that can be explained by rather small amount of data for these fits. So in that case the reliability is too small for considering the almost linear dependence of Δδ on the temperature in case of PDE a real effect, not that much concerning Tsplit but definitively when it comes to Δδ∞. All in all, we may now consider Tsplit to be a measure for the temperature from which on the surface of microgels aggregates is no longer permeable for the solvent molecules. The entrapped solvent has a restricted mobility and is distinguishable from the bulk solvent. It is interesting to compare this Tsplit to onset and peak temperatures from DSC. The onset is characteristic for the starting point of the transition while the peak temperature marks the center of the transition, which has even been confirmed by the NMR transition temperature. Such a comparison is given in Figure 9. In contrast to the comparison of the NMR-determined Ttrans and the DSC peak temperature, the situation is more complex. An approximate accordance of Tsplit and the peak temperature is found only for the case of PNi. The completely opposite case, namely an approximate accordance of Tsplit and the onset temperature, is detected for PDE. The other two samples for which Tsplit values are determined show an intermediate behavior. The one-to-one copolymer P11 has Tsplit values that are below Tpeak but still

relation to the collapse is present. The identity of Ttrans and Tpeak documents that the very collapse is causing the heat consumption observed in DSC and DSC can be used for directly describing this collapse. The good correlation of the peak temperatures and Ttrans allows the discussion of another, more significant effect that can be observed in NMR but not in DSC: the detection of restricted solvent, which seems to be rather characteristic for cross-linked gels where it has been observed before.29−31 In fact, such restricted solvent has been reported for linear chains of other thermoresponsive polymers as well but only at very high concentrations that are four times larger than the concentration of microgels used here.32 Because of the rather high concentration of 5 wt % in our case, the microgels are in contact to each other in the swollen state. Thus, a partial aggregation can occur when the microgels begin to collapse with increasing temperature. Such aggregates will contain cavities and channels and the solvent molecules in these cavities could constitute the restricted species. Above Tsplit, the PFGNMR results (Figure 7) clearly identify the additional signals as restricted solvent signals, while an enrichment of one of the cononsolvents in a certain domain of the polymer-rich phase cannot be directly seen by this NMR data (see below). The mean square displacement (Drestricted solvent ≈ 2 × 10−10 m2/s at 20 ms diffusion time, leading to a length of about 2 μm) indicates that several microgels are connected to each other. If this is not the reason one has to consider at least partial aggregation. By such an aggregation interstices and channels populated by thus restricted solvent molecules would result outside thein the beginning partiallycollapsed microgels. The presence of only one solvent signal below Tsplit then means that the solvent molecules can diffuse freely through the sample. Then Tsplit is the temperature from which on a fast exchange of solvent molecules between microgel aggregates does no longer occur. The existence of a restricted environment can even be further verified by taking into account the width of the signals for restricted solvent. The signals of restricted solvent are much broader than the signals of free solvent (see Supporting Information) implying faster transverse relaxation and hence slower dynamics. The peak width of the restricted solvent signal increases with temperature in contrast to the temperature independent diffusion coefficient. That implies a more rigid environment canceling out the greater possible kinetic energy at high temperatures. This observation is also well in line with a possible domain-like picture of the restricted solvent environment. An explanation for the observed diffusion behavior can be the presence of solvent domains that shrink with increasing temperature. G

dx.doi.org/10.1021/ma302384v | Macromolecules XXXX, XXX, XXX−XXX

Macromolecules

Article

Figure 10. Different comparisons of Tsplit: difference of the values in solvent mixtures to the values in 5 mol % methanol (a); difference between Tsplit and Tonset at 15 mol % methanol content (b). For assignment of the microgels see Figure 9.

hydrophobic methyl group). Simultaneously, compartments, which entrap the solvent, can develop already at lower temperatures, seen as a separate peak in the NMR spectrum and giving a lowered T split with increasing methanol concentration. Regarding P31, this complexation leads already to solvent enclosure at lower temperatures, whichin turnis not reflected in a change of the macroscopic properties. Thus, cononsolvency can hardly be detected by other methods due to the rather low number of NiPAm units for this specific polymer. These explanations are in accordance with simulation findings.11 Summarizing, it can be stated that methanol addition makes the NiPAm units behaving more “DEAm-like” in the following aspects: The transition width and the dependence of Tsplit on the methanol content (Figure 10b). The methanol addition does not only alter the width of transition for the NiPAm-rich polymers (hydrogen-bonded methanol might obstruct the cooperative phase transition of the PNIPAM),12 but influences the phase-separation properties. Simultaneously, the drop in Tsplit translates into a reduction of the VPTT or LCST for the samples with higher NiPAm-content (cononsolvency). Again, all these findings can be explained by a preferred interaction of methanol with the particular amide proton of PNiPAm leading to cononsolvency. Indications from simulation results and from theoretical considerations have been reported recently for such an interaction.11,15 Our results agree with these findings, which state that the amide proton is essential for hydrogen bonding of methanol. That would explain why no cononsolvency is found for PDEAm. For this reason, we should shortly compare the LCST properties of PDEAm and PNiPAm to related polymers with and without an amide proton. The LCST of linear mono- or disubstituted poly(acrylamide) depends on both the length of the alkyl chain and on the number of substituents.1 Thus, it is not easy to compare a possible complexation of the NiPAm amide proton by methanol to the substitution of the proton by a methyl group. However, when one substituent is ethyl, depending on the second substituent, the LCST decreases from 72 °C for a proton to 56 °C for methyl and finally 32 °C for ethyl.1 That means that it is rather the number of carbons which counts and not the existence of the proton. Now when adding methanol to an aqueous PNiPAm solution, a stronger decrease in the transition temperature is induced than by substituting the amide proton by a methyl group (LCST at about 22 °C for poly(N-methyl-N-isopropylacrylamide) compared to an LCST below 0 °C of PNiPAm with 20 mol % of methanol).1 This implies that hydrogen bonding of methanol leads to a complex with enhanced hydrophobicity, which dehydrates even at low temperatures (Scheme 1). This

rather close to it; furthermore the difference between these two values remains more or less constant although the width of transition and thus the difference between onset and peak is increasing with increasing methanol content. Finally, the Tsplit of DEAm-rich P31 changes from peak-temperature-like to onset-temperature-like when the methanol content of the solvent is increased from 0 mol % to 15 mol %. This makes the P31 microgel unique. It is the only polymer in this study where cononsolvency on the one hand has no influence on the collapse temperature but on the other hand manifests itself by leading to a premature entrapment of solvent. This becomes noticeable by a decrease of the temperature from which on entrapped solvent is detected. In this respect, one can realize a transition from a PNi-type to a PDE-type behavior (see also Figure 10). Again, we can divide the samples into two groups like it was for the transition width and the existence of cononsolvency behavior: (i) The DEAm dominated ones, i.e., PDE and P31, and (ii) the other three, i.e., PNi, P13, and P11. The latter ones are of monomer compositions for which cononsolvency behavior was reported in case of linear, non-cross-linked chains.23 Linear polymers of compositions similar to that of P31 and PDE in turn showed (almost) no cononsolvency. The property of cononsolvency performance is brought to the system by the NiPAm units. Macroscopically, 50% of NiPAm turn out to be enough to keep this property even in copolymers. Beyond this, at least at about 33% of NiPAm, no minimum in transition temperature is found in dependence of the methanol content.23 It is more complicated to explain the observed behavior (especially for copolymer P31) regarding the specific dependencies of Tsplit with composition. Speculating, the transition width might play a role for the pure homopolymers in pure water when going from PNi to PDE. Further, the data imply that there is a critical degree of collapse, from which on restricted solvent is observed. This degree seems to be different for different polymers. For pure water, NiPAm-containing polymers might allow better permeability of solvent molecules by formation of kind of channels, which then are “decorated” by the rather hydrophilic amide protons. Since PDE lacks this amide proton, this is not possible for pure PDE, giving a potential explanation for the lowered Tsplit. Tsplit decreases in NiPAm-containing samples upon addition of methanol, meaning that a reduction of Tsplit is induced by the presence of the amide proton (Figure 10a). PDE does not exhibit this feature. Therefore, the presence of the NiPAm units allows the preferential complexation with methanol, making the whole polymer more hydrophobic (the hydrophilic amide proton is then shielded toward the bulk solvent by the exposed H

dx.doi.org/10.1021/ma302384v | Macromolecules XXXX, XXX, XXX−XXX

Macromolecules

Article

between aggregated microgels. This aggregation arises from the high concentration of the microgels. Whether Tsplit is found at early or late stage of collapse depends on the NiPAm to DEAm ratio and the methanol content. These differences and the actual transition temperature upon methanol addition reveal the significance of the amide proton. In pure water, NiPAm repeat units have a different ability for entrapping some of the solvent molecules than DEAm units. For PDE Tsplit is found in the early state of collapse whereas in case of PNi Tsplit is found at later stage of collapse. In the copolymer microgels, Tsplit shifts to later stage of collapse with increasing NiPAm content. Upon addition of methanol, Tsplit moves to early stage of collapse for PNi and the NiPAm-rich copolymers. No such effect is observed for PDE, where Tsplit is always found at early transition stage independently of the methanol concentration. So, adding methanol makes the behavior of PNi similar to that of PDE. A similar trend is found when the transition width is discussed: In pure D2O PNi has a sharp VPT whereas PDEAm has a broad transition. Methanol addition broadens the transition of PNi as seen by both NMR and DSC. Comparing the results of homo- and copolymer microgels in pure water and water/methanol mixtures, respectively, shows the following trends: (i) Introducing DEAm units into PNiPAm microgel lowers the transition temperature and broadens the transition. (ii) Adding methanol to aqueous PNiPAm solution also lowers the transition temperature and broadens the transition. In other words, “replacing” some NiPAm units by DEAm units or solvating some NiPAm units by methanol (instead of water) has the same effect on the volume phase transition of PNiPAm. Hence, a methanolsolvated NiPAm unit behaves similar as a DEAm unit. All in all, interactions of solvent and polymer turn out to be crucial for the two effects of synergistic depression of the transition temperature in NiPAm/DEAm copolymers on the one hand and cononsolvency of PNiPAm on the other hand. The results suggest a favored interaction of methanol and the amide proton of the NiPAm repeat units to be a key in understanding the phenomenon of cononsolvency. The results imply that at first methanol is inhibiting water hydrogen bonds to the amide proton and at the same time leading to a polymer solvent complex resulting in the gel collapsing prematurely. In a second step, parts of the remaining solvent are then entrapped to cavities, as seen by NMR.

Scheme 1. Possible Hydrogen Bonding between Amide Function and Methanol

might be explained by six-membered associates of methanol and NiPAm units, which not only occupy the amide proton site, but also obstruct the carbonyl function for further solvation with the hydrophilic bulk mixture. This preferred complexation of methanol might be explained by the +I effect of the methyl group, which increases the electron density at the oxygen of the methanol molecule. From this it follows again that there are actually two different kinds of bound solvent. In a first step a number of methanol molecules is coordinating to the chain and making it more hydrophobic. These molecules are directly located at the polymer. Again, the reader might notice that these coordinated molecules are different from the species that has been mentioned as restricted solvent so far. As the coordinated molecules will be rather fixed to the polymer, these will diffuse with a similar diffusion coefficient as the polymer. Simultaneously, the chemical environment is expected to be changed that much that the NMR signal should be much more polymer-like, especially when regarding the peak width. Further, the molar amount complexed is limited by the binding sites and it cannot explain the pronounced restricted solvent peak (consider the deuterated solvent). Therefore, this complexed solvent is different to the so-called restricted solvent, which, in contrast, still has a rather free water-like diffusion coefficient. All in all, two consecutive processes have to be distinguished. In a first step, methanol forms hydrogen bonds to the amide proton and the other sites of the polymer. By this complexation, the polymer becomes increasingly hydrophobic and collapses prematurely compared to the case in pure water. The collapse of the microgel then leads to the formation of restricting cavitations or, as it were, kind of pockets. Inside these pockets or interstices, solvent molecules are entrapped that show properties that differ from the bulk solvent but nonetheless still behave rather bulk-solvent-like. The effect that the transition temperatures start to increase again above a certain methanol concentration, in turn, can be explained by an affinity of the polymer with hydrogen-bonded methanol to the bulk methanol which becomes dominant at such compositions. This can be verified by a minimum of the diffusion coefficients in neat water/methanol mixtures of just the same composition as that where the largest depression of the phase transition temperature occurs.46



ASSOCIATED CONTENT

S Supporting Information *

Detailed presentation of the DSC data, modeling of the NMR water signal, additional diffusion data, and DLS-determined microgel size. This material is available free of charge via the Internet at http://pubs.acs.org.

■ ■

AUTHOR INFORMATION

Notes

The authors declare no competing financial interest.

5. CONCLUSIONS We found further details on effects on the solvent during the volume phase transition (VPT) of thermosensitive microgels. A second solvent species is observed in 1H NMR from a certain point of collapse of the microgel with increasing temperature. Above Tsplit, the self-diffusion coefficient of the second solvent species revealed it to be restricted but not complexed to the polymer chain. Furthermore, the restricted solvent is not localized in one single microgel but in channels or cavities

ACKNOWLEDGMENTS We thank Tommi Virtanen for giving technical support with the NMR and Arjan Gelissen for synthesizing some of the microgels. C.H.H., F.A.P., and C.S. are thankful to the DAAD (German Academic Exchange Service) for an academic exchange program (PPP50741359). The financial support by the Deutsche Forschungsgemeinschaft (DFG), within the I

dx.doi.org/10.1021/ma302384v | Macromolecules XXXX, XXX, XXX−XXX

Macromolecules

Article

collaborative research center “Functional Microgels and Microgel Systems” (SFB 985), is gratefully acknowledged.



(33) Plamper, F. A.; Steinschulte, A. A.; Hofmann, C. H.; Drude, N.; Mergel, O.; Herbert, C.; Erberich, M.; Schulte, B.; Winter, R.; Richtering, W. Macromolecules 2012, 45 (19), 8021−8026. (34) Maeda, Y.; Higuchi, T.; Ikeda, I. Langmuir 2000, 16 (19), 7503−7509. (35) Sun, S.; Zhang, W.; Zhang, W.; Wu, P.; Zhu, X. Soft Matter 2012, 8 (14), 3980−3987. (36) Pühse, M.; Keerl, M.; Scherzinger, C.; Richtering, W.; Winter, R. Polymer 2010, 51 (16), 3653−3659. (37) Biswas, C. S.; Patel, V. K.; Vishwakarma, N. K.; Mishra, A. K.; Saha, S.; Ray, B. Langmuir 2010, 26 (9), 6775−6782. (38) Biswas, C. S.; Vishwakarma, N. K.; Patel, V. K.; Mishra, A. K.; Saha, S.; Ray, B. Langmuir 2012, 28 (17), 7014−7022. (39) Ammann, C.; Meier, P.; Merbach, A. E. J. Magn. Reson. 1982, 46 (2), 319−321. (40) Massiot, D.; Fayon, F.; Capron, M.; King, I.; Le Calve, S.; Alonso, B.; Durand, J. O.; Bujoli, B.; Gan, Z. H.; Hoatson, G. Magn. Reson. Chem. 2002, 40 (1), 70−76. (41) Panayiotou, M.; Freitag, R. Polymer 2005, 46 (3), 615−621. (42) Maeda, Y.; Nakamura, T.; Ikeda, I. Macromolecules 2002, 35 (27), 10172−10177. (43) Hofmann, C. H.; Schönhoff, M. Colloid Polym. Sci. 2012, 290 (8), 689−698. (44) Larsson, A.; Kuckling, D.; Schönhoff, M. Colloids Surf., A: Physicochem. Eng. Asp. 2001, 190 (1−2), 185−192. (45) Ray, S. S.; Rajamohanan, P. R.; Badiger, M. V.; Devotta, I.; Ganapathy, S.; Mashelkar, R. A. Chem. Eng. Sci. 1998, 53 (5), 869− 877. (46) Derlacki, Z. J.; Easteal, A. J.; Edge, A. V. J.; Woolf, L. A.; Roksandic, Z. J. Phys. Chem. 1985, 89 (24), 5318−5322. (47) Laszlo, K.; Guillermo, A.; Fluerasu, A.; Moussaid, A.; Geissler, E. Langmuir 2010, 26 (6), 4415−4420. (48) O’Reilly, D. E. J. Chem. Phys. 1974, 61 (4), 1592−1593.

REFERENCES

(1) Ito, S. Kobunshi Ronbunshu 1989, 46 (7), 437−443. (2) Schild, H. G.; Tirrell, D. A. J. Phys. Chem. 1990, 94 (10), 4352− 4356. (3) Woodward, N. C.; Chowdhry, B. Z.; Snowden, M. J.; Leharne, S. A.; Griffiths, P. C.; Winnington, A. L. Langmuir 2003, 19 (8), 3202− 3211. (4) Hofmann, C.; Schönhoff, M. Colloid Polym. Sci. 2009, 287 (12), 1369−1376. (5) Spěvácě k, J. Curr. Opin. Colloid Interface Sci. 2009, 14 (3), 184− 191. (6) Sierra-Martin, B.; Choi, Y.; Romero-Cano, M. S.; Cosgrove, T.; Vincent, B.; Fernandez-Barbero, A. Macromolecules 2005, 38 (26), 10782−10787. (7) Pelton, R. Adv. Colloid Interface Sci. 2000, 85 (1), 1−33. (8) Kunugi, S.; Tanaka, N. Biochim. Biophys. Acta: Protein Struct. Mol. Enzymol. 2002, 1595 (1−2), 329−344. (9) Schild, H. G. Prog. Polym. Sci. 1992, 17 (2), 163−249. (10) Winnik, F. M.; Ringsdorf, H.; Venzmer, J. Macromolecules 1990, 23 (8), 2415−2416. (11) Walter, J.; Sehrt, J.; Vrabec, J.; Hasse, H. J. Phys. Chem. B 2012, 116 (17), 5251−5259. (12) Tanaka, F.; Koga, T.; Winnik, F. M. Phys. Rev. Lett. 2008, 101, 2. (13) Tanaka, F.; Koga, T.; Kojima, H.; Winnik, F. A. Macromolecules 2009, 42 (4), 1321−1330. (14) Tanaka, F.; Koga, T.; Winnik, F. M. Competitive Hydrogen Bonds and Cononsolvency of Poly(N-isopropylacrylamide)s in Mixed Solvents of Water/Methanol. In Gels: Structures, Properties, and Functions: Fundamentals and Applications; Progress in Colloid and Polymer Science 136; Springer: Berlin, 2009; pp 1−7. (15) Kojima, H.; Tanaka, F. Soft Matter 2012, 8 (10), 3010−3020. (16) Kojima, H.; Tanaka, F.; Scherzinger, C.; Richtering, W. J. Polym. Sci., Part B: Polym. Phys. 2012, DOI: 10.1002/polb.23194. (17) Arndt, M. C.; Sadowski, G. Macromolecules 2012, 45 (16), 6686−6696. (18) Dixit, S.; Crain, J.; Poon, W. C. K.; Finney, J. L.; Soper, A. K. Nature 2002, 416 (6883), 829−832. (19) Sun, S.; Wu, P. Macromolecules 2010, 43 (22), 9501−9510. (20) Zhang, G. Z.; Wu, C. J. Am. Chem. Soc. 2001, 123 (7), 1376− 1380. (21) Scherzinger, C.; Lindner, P.; Keerl, M.; Richtering, W. Macromolecules 2010, 43 (16), 6829−6833. (22) Lu, Y. J.; Zhou, K. J.; Ding, Y. W.; Zhang, G. Z.; Wu, C. Phys. Chem. Chem. Phys. 2010, 12 (13), 3188−3194. (23) Maeda, Y.; Yamabe, M. Polymer 2009, 50 (2), 519−523. (24) Keerl, M.; Richtering, W. Colloid Polym. Sci. 2007, 285 (4), 471−474. (25) Keerl, M.; Smirnovas, V.; Winter, R.; Richtering, W. Angew. Chem., Int. Ed. 2008, 47 (2), 338−341. (26) Keerl, M.; Smirnovas, V.; Winter, R.; Richtering, W. Macromolecules 2008, 41 (18), 6830−6836. (27) Št’astná, J.; Hanyková, L.; Spěvácě k, J. Colloid Polym. Sci. 2012, 290 (17), 1811−1817. (28) Ohta, H.; Ando, I.; Fujishige, S.; Kubota, K. J. Polym. Sci., Part B: Polym. Phys. 1991, 29 (8), 963−968. (29) Wang, N.; Ru, G.; Wang, L.; Feng, J. Langmuir 2009, 25 (10), 5898−5902. (30) Díez-Peña, E.; Quijada-Garrido, I.; Barrales-Rienda, J. M.; Wilhelm, M.; Spiess, H. W. Macromol. Chem. Phys. 2002, 203 (3), 491−502. (31) Hanyková, L.; Spěvácě k, J.; Ilavsky, M. Polymer 2001, 42 (21), 8607−8612. (32) Spěvácě k, J.; Hanyková, L. Macromolecules 2005, 38 (22), 9187−9191. J

dx.doi.org/10.1021/ma302384v | Macromolecules XXXX, XXX, XXX−XXX